COVID-19

The SARS-CoV-2 virus has resulted in > 778 million reported COVID-19 cases and > 7 million deaths worldwide. ¹ About 70% of the eligible US population has completed a primary COVID-19 vaccination series, yet only 17% have received an updated bivalent booster dose.² These immunization rates fall below the World Health Organization (WHO) target of 70%.³

Early in the pandemic, US Department of Veterans Affairs (VA) vaccination rates ranged from 46% to 71%.^4,5 Ensuring a high level of COVID-19 vaccination in the largest integrated US health care system aligns with the VA priority to provide high-quality, evidence-based care to a patient population that is older and has more comorbidities than the overall US population.^6-9

Vaccine hesitancy, defined as a “delay in acceptance or refusal of vaccination despite availability of vaccination service,” is a major contributor to suboptimal vaccination rates.^10-13 Previous studies used cluster analyses to identify the unique combinations of behavioral and social factors responsible for COVID-19 vaccine hesitancy.^10,11 Lack of perceived vaccine effectiveness and low perceived risk of the health consequences from COVID-19 infection were frequently identified in clusters where patients had the lowest intent for vaccination.^10,11 Similarly, low trust in health care practitioners (HCPs), government, and pharmaceutical companies diminished intent for vaccination in these clusters.¹⁰ These quantitative studies were limited by their exclusive focus on unvaccinated individuals, reliance on self-reported intent, and lack of assessment of a health care system with a COVID-19 vaccine delivery program designed to overcome barriers to health care access, such as the VA.

Prior qualitative studies of vaccine uptake in distinct veteran subgroups (ie, unhoused and in VA facilities with low vaccination rates) demonstrated that overriding medical priorities among the unhoused and vaccine safety concerns were associated with decreased vaccine uptake, and positive perceptions of HCPs and the health care system were associated with increased vaccine uptake.^11,12 However, these studies were conducted during periods of greater COVID-19 vaccine availability and acceptance, and prior to booster recommendations.^4,12,13

This mixed-methods quality improvement (QI) project assessed the barriers and facilitators of COVID-19 vaccination among veterans receiving primary care at a single VA health care facility. We assessed whether unique patient clusters could be identified based on COVID-19–related and vaccine-related thoughts and feelings and whether cluster membership was associated with COVID-19 vaccination. This analysis also explored how individuals’ beliefs and trust shaped motivations and hesitancies for vaccine uptake in quantitatively derived clusters with varying vaccination rates.

Methods

This QI project was conducted at the VA Pittsburgh Healthcare System (VAPHS), a tertiary care facility serving > 75,000 veterans in Pennsylvania, West Virginia, and Ohio. The VAPHS Institutional Review Board determined this QI study was exempt from review.^14-17 Participation was voluntary and had no bearing on VA health care or benefits. Financial support for the project, including key personnel and participant compensation, was provided by VAPHS. We followed the STROBE reporting guideline for cross-sectional studies and the COREQ checklist for qualitative research.^18,19

Quantitative Survey

The 32,271 veterans assigned to a VAPHS primary care HCP, effective April 1, 2020, were eligible. To ensure representation of subgroups underrecognized in research and/or QI projects, the sample included all 1980 female patients at VAPHS and a random sample of 500 White and 500 Hispanic and/or non-White men within 4 age categories (< 50, 50-64, 65-84, and > 84 years). For the < 50 years or > 84 years categories, all Hispanic and/or non-White men were included due to small sample sizes.^20-22 The nonrandom sampling frame comprised 1708 Hispanic and/or non-White men and 2000 White men. After assigning the 5688 potentially eligible individuals a unique identifier, 31 opted out, resulting in a final sample of 5657 individuals.

The 5657 individuals received a letter requesting their completion of a future questionnaire about COVID-19 infection and vaccines. An electronic Qualtrics questionnaire link was emailed to 3221 individuals; nonresponders received 2 follow-up email reminders. For the 2436 veterans without an email address on file, trained interviewers conducted phone surveys and entered responses. Those patients who completed the questionnaire could enter a drawing to win 1 of 100 cash prizes valued at $100. We collected questionnaire data from July to September 2021.

Questionnaire Items

We constructed a 60-item questionnaire based on prior research on COVID-19 vaccine hesitancy and the WHO Guidebook for Immunization Programs and Implementing Partners.^4,23-25 The WHO Guidebook comprises survey items organized within 4 domains reflecting the behavioral and social determinants of vaccination: thoughts and feelings; social processes; motivation and hesitancy; and practical factors.²³

Sociodemographic, clinical, and personal characteristics. The survey assessed respondent ethnicity and race and used these data to create a composite race and ethnicity variable. Highest educational level was also attained using 8 response options. The survey also assessed prior COVID-19 infection; prior receipt of vaccines for influenza, pneumonia, tetanus, or shingles; and presence of comorbidities that increase the risk of severe COVID-19 infection. We used administrative data from the VA Corporate Data Warehouse to determine respondent age, sex, geographic residence (urban, rural), and to fill in missing self-reported data on sex (n = 4) and ethnicity and race (n = 12). The survey assessed political views using a 5-point Likert scale (1, very liberal; 5, very conservative) and was collapsed into 3 categories (ie, very conservative or conservative, moderate, very liberal or liberal), with prefer not to answer reported separately

COVID-19 infection and vaccine. We asked veterans if they had ever been infected with COVID-19, whether they had been offered and/or received a COVID-19 vaccine, and type (Pfizer, Moderna, or Johnson & Johnson), and number of doses received. Positive vaccination status was defined as the receipt of ≥ 1 dose of a COVID-19 vaccine approved by the US Food and Drug Administration.

COVID-19 opinions. Respondents were asked about perceived risk of COVID-19 infection and related health outcomes, as well as beliefs about COVID-19 vaccines, using a 4-point Likert scale for all items: (1, not at all concerned; 4, very concerned). Respondents were asked about concerns related to COVID-19 infection and severe illness. They also were asked about vaccine-related short-term adverse effects (AEs) and long-term complications. Respondents were asked how effective they believed COVID-19 vaccines were at preventing infection, serious illness, or death. Unvaccinated and vaccinated veterans were asked similar items, with a qualifier of “before getting vaccinated…” for those who were vaccinated.

Social processes. Respondents were asked to rate their level of trust in various sources of COVID-19 vaccine information using a 4-point Likert scale (1, trust not at all; 4, trust very much). Respondents were asked whether community or religious leaders or close family or friends wanted them to get vaccinated (yes, no, or unsure).

Practical factors. Respondents were asked to rate the logistical difficulty of getting vaccinated or trying to get vaccinated using a 4-point Likert scale (1, not at all; 4, extremely).

Participants

Respondents were asked to participate in a follow-up qualitative interview. Among 293 participants who agreed, we sampled all 86 unvaccinated individuals regardless of cluster assignment, a random sample of 88 individuals in the cluster with the lowest vaccination rate, and all 33 vaccinated individuals in the cluster with the second-lowest vaccination rate. Forty-nine veterans completed qualitative interviews.

Two research staff trained in qualitative research completed telephone interviews, averaging 16.5 minutes (March to May 2022), using semistructured scripts to elicit vaccine-related motivations, hesitancies, or concerns. Interviews were recorded, transcribed, and deidentified. Participants provided written consent for recording and received $50 cash-equivalent compensation for interview completion.

Qualitative Interview Script

The interview script consisted of open-ended questions related to vaccine uptake across WHO domains.²³ Both unvaccinated and vaccinated respondents were asked similar questions and customized questions about boosters for the vaccinated subgroup. To assess motivations and hesitancies, respondents were asked how they made their decisions about vaccination and what they considered when deciding. Vaccinated participants were asked about motivations and overcoming concerns. Unvaccinated respondents were asked about reasons for concern. To assess social processes, the interviewers asked participants whose opinion or counsel they trusted when deciding whether to get vaccinated. Questions also focused on positive experiences and vaccination barriers. Vaccinated participants were asked what could have improved their vaccination experiences. Finally, the interviewers asked participants who received a complete primary vaccine series about their motivations and plans related to booster vaccines, and whether information about emerging COVID-19 variants influenced their decisions.

Data Analyses

This analysis used X² and Fisher exact tests to assess the associations among respondent characteristics, questionnaire responses, vaccination status, and cluster membership. Items phrased similarly were handled in a similar fashion for vaccinated and unvaccinated respondents.

Cluster analysis assessed the possible groupings in responses to the quantitative questionnaire items focused on thoughts and feelings about COVID-19 infection risk and severity, vaccine effectiveness, and vaccine safety. This analysis treated the items’ ordinal response categories as continuous. We performed factor analysis using principal component analysis to explore dimension reduction and account for covariance between items. Two principal components were calculated and applied k-means clustering, determining the number of clusters through agreement from the elbow, gap statistic, and silhouette methods.²⁶ Each cluster was named based on its unique pattern of responses to the items used to define them (eAppendix 1).

Multivariable logistic regression analyses assessed the independent association between cluster membership as the independent measure and vaccination status as the dependent measure, adjusting for respondent sociodemographic and personal characteristics and 2 measures of trust (ie, local VA HCP and the CDC). We selected these trust measures because they represent objective sources of medical information and were independently associated with COVID-19 vaccination status in a logistic regression model comprising all 6 trust items assessed.

This study defined statistical significance as a 2-tailed P value < .05. SAS 9.4 was used for all statistical analyses and Python 3.7.4 and the Scikit-learn package for cluster analyses.²⁷ For qualitative analyses, this study used an inductive thematic approach guided by conventional qualitative content analysis, NVivo 12 Plus for Windows to code and analyze interview transcripts.^28,29 We created an initial codebook based on 10 transcripts that were selected for high complexity and represented cluster membership and vaccination status.^30,31 After 2 qualitative staff developed the initial codebook, 11 of 49 (22%) transcripts were independently coded by a primary and secondary coder to ensure consistent code application. Both coders reviewed the cocoded transcripts and resolved all discrepancies through negotiated consensus.³² After the cocoding process was complete, the primary coder coded the remaining transcripts. The primary and secondary coder met as needed to review and discuss any questions that arose during the primary coder’s work.

Results

Of 5657 eligible participants, 1208 (21.4%) completed a questionnaire. Overall, 674 (55.8%) were aged < 65 years, 530 (43.9%) were women, 828 (68.5%) were non-Hispanic White, 303 (25.1%) were Black, and 47 (3.9%) were Hispanic, and 1034 (85.6%) were vaccinated (Table 1). Compared to the total sampled population, respondents were more often older, female, and White (eAppendix 2).

Cluster Membership

Four clusters were identified from 1183 (97.9%) participants who provided complete responses to 6 items assessing thoughts and feelings about COVID-19 infection and vaccines (Table 2). Of the 1183 respondents, 375 (31.7%) were Concerned Believers (cluster 1), 336 (28.4%) were Unconcerned Believers (cluster 2), 298 (25.2%) were Concerned Ambivalents (cluster 3), and 174 (14.7%) were Unconcerned Disbelievers (cluster 4). The Concerned Believers were moderately/ very concerned about COVID-19 infection (96.0%) and becoming very ill from infection (94.6%), believed the vaccine was moderately/very effective in preventing COVID-19 infection (100%) and severe illness or death from infection (98.7%), and had slight concern about short-term AEs (92.6%) or long-term complications (92.0%) from the vaccine. The Unconcerned Believers had no/slight concern about COVID-19 infection (76.5%) or becoming very ill (79.2%), believed the vaccine was effective in preventing infection (82.4%) and severe illness and death (83.6%), and had no/slight concern about short-term AEs (94.0%) or long-term complications (87.2%) from the vaccine. The Concerned Ambivalents were moderately/ very concerned about COVID-19 infection (94.3%) and becoming very ill (93.6%), believed the vaccine was moderately/very effective in preventing infection (86.6%) and severe illness or death (86.9%), and were moderately/very concerned about short-term AEs (81.9%) or long-term complications (89.3%) from the vaccine. The Unconcerned Disbelievers had no/slight concern about COVID-19 infection (90.8%) and becoming very ill (88.6%), believed the vaccine was not at all/slightly effective in preventing infection (90.3%) and severe illness or death (87.4%), and were moderately/very concerned about short-term AEs (52.8%) or long-term complications (75.9%) from the vaccine.

Cluster Membership

Respondent age, race and ethnicity, and political viewpoints differed significantly by cluster (P < .001). Compared with the other clusters, the Concerned Believer cluster was older (55.5% age ≥ 65 years vs 16.7%-48.0%) and more frequently reported liberal political views (28.8% vs 4.6%-15.1%). In contrast, the Unconcerned Disbeliever cluster was younger (83.4% age ≤ 64 years vs 44.5%-56.8%) and more frequently reported conservative political views (37.9% vs 17.1%-26.8%) than the other clusters. Whereas the Concerned Ambivalent cluster had the highest proportion of Black (37.7%) and the lowest proportion of White respondents (57.6%), the Unconcerned Disbelievers cluster had the lowest proportion of Black respondents (14.5%) and the highest proportion of White respondents (77.9%). The Unconcerned Disbelievers cluster were significantly less likely to trust COVID-19 vaccine information from any source and to believe those close to them wanted them to get vaccinated.

Association of Cluster Membership and COVID-19 Vaccination

COVID-19 vaccination rates varied more than 3-fold (P < .001) by cluster, with 29.9% of Unconcerned Disbelievers, 93.3% of Concerned Ambivalents, 93.5% of Unconcerned Believers, and 98.9% of Concerned Believers reporting being vaccinated. (Figure). Cluster membership was independently associated with vaccination, with adjusted odds ratios (AORs) of 12.0 (95% CI, 6.1-23.8) for the Concerned Ambivalent, 13.0 (95% CI, 6.9-24.5) for Unconcerned Believer, and 48.6 (95% CI, 15.5-152.1) for Concerned Believer clusters (Table 3). Respondent trust in COVID-19 vaccine information from their VA HCP (AOR 2.1; 95% CI, 1.6-2.8) and the CDC (AOR 1.6; 95% CI, 1.2-2.1) were independently associated with vaccination status, while the remaining respondent sociodemographic or personal characteristics were not.

Qualitative Interview Participants

A 49-participant convenience sample completed interviews, including 30 Concerned Ambivalent, 17 Unconcerned Disbeliever, and 2 Unconcerned Believer respondents cluster. The data were not calculated for Unconcerned Believers due to the small sample size. Interview participants were more likely to be younger, female, non-Hispanic, White, less educated, and more politically conservative than the questionnaire respondents as a whole (Appendix). The vaccination rate for the interview participants was 73.5%, ranging from 29.9% in the Unconcerned Disbeliever to 93.3% in the Concerned Ambivalent cluster. Qualitative themes and participant quotes for Concerned Ambivalent and Unconcerned Disbeliever respondents are in eAppendix 3.

Motivations. Wanting personal protection from becoming infected or severely ill from COVID-19 (63.8%), caregiver wanting to protect others (17.0%), and employment vaccine requirements (14.9%) were frequent motivations for vaccination. Whereas personal protection (90.0%) and protection of others (23.3%) were identified more frequently in the Concerned Ambivalents cluster, employment vaccine requirements (35.3%) were more frequently identified in the Unconcerned Disbelievers cluster.

Hesitancies or concerns. Lack of sufficient information related to rapid vaccine development (55.3%), vaccine AEs (38.3%), and low confidence in vaccine efficacy (23.4%) were frequent concerns or hesitancies about vaccination. Unconcerned Disbelievers expressed higher levels of concern about the vaccine’s rapid development (82.4%), low perceived vaccine efficacy (47.1%), and a lack of trust in governmental vaccine promotion (23.5%) than did the Concerned Ambivalents.

Overcoming concerns. Not wanting to get sick or die from infection coupled with an understanding that vaccine benefits exceed risks (23.4%) and receiving information from a trusted source (10.6%) were common ways of overcoming concerns for vaccination. Although the Unconcerned Disbelievers infrequently identified reasons for overcoming concerns, they identified employment requirements (17.6%) as a reason for vaccination despite concerns. They also identified seeing others with positive vaccine experiences and pressure from family or friends as ways of overcoming concerns (11.8% each).

Social influences. Family members or partners (38.3%), personal opinions (38.3%), and HCPs (23.4%) were frequent social influences for vaccination. Concerned Ambivalents mentioned family members and partners (46.7%), HCPs (26.7%), and friends (20.0%) as common influences, while Unconcerned Disbelievers more frequently relied on their opinion (41.2%) and quoted specific scientifically reputable data sources (17.6%) to guide vaccine decision-making, although it is unclear whether these sources were accessed directly or if this information was obtained indirectly through scientifically unvetted data platforms.

Practical factors. Most participants had positive vaccination experiences (68.1%), determined mainly by the Concerned Ambivalents (90.0%), who were more highly vaccinated. Barriers to vaccination were reported by 9 (19.1%) participants, driven by those in the Concerned Ambivalent cluster (26.7%). Eight (17.0%) participants suggested improvements for vaccination processes, with similar overall reporting frequencies across clusters.

COVID-19 boosters and variants. Wanting continued protection from COVID-19 (36.2%), recommendations from a doctor or trusted source (17.0%), and news about emerging variants (10.6%) were frequent motivations for receiving a vaccine booster (eAppendix 4). These motivations were largely driven by the Concerned Ambivalents, of whom 25 of 30 were booster eligible and 24 received a booster dose. Belief that boosters were unnecessary (8.5%), concerns about efficacy (6.4%), and concerns about AEs (6.4%) were frequently identified hesitancies. These concerns were expressed largely by the Unconcerned Disbelievers, of whom 7 of 17 were booster dose eligible, but only 1 received a dose.

Evolving knowledge about variants was not a major concern overall and did not change existing opinions about the vaccine (36.2%). Concerned Ambivalents believed vaccination provided extra protection against variants (36.7%) and the emergence of variants served as a reminder of the ongoing pandemic (30.0%). In contrast, Unconcerned Disbelievers believed that the threat of variants was overblown (35.3%) and mutations are to be expected (17.6%).

Discussion

This study used a complementary mixed-methods approach to understand the motivations, hesitancies, and social and practical drivers of COVID-19 vaccine uptake among VA beneficiaries. Our quantitative analyses identified 4 distinct clusters based on respondents’ opinions on COVID-19 infection severity and vaccine effectiveness and safety. Veterans in 3 clusters were 12 to 49 times more likely to be vaccinated than those in the remaining cluster, even when controlling for baseline respondent characteristics and level of trust in credible sources of COVID-19 information. The observed vaccination rate of nearly 86% was higher than the contemporaneous national average of 62% for vaccine-eligible individuals, likely reflecting the comprehensive VA vaccine promotion strategies tailored to a patient demographic with a high COVID-19 risk profile.^2,10

This cluster analyses demonstrated the importance of thoughts and feelings about COVID-19 infection and vaccination as influential social and behavioral drivers of vaccine uptake. These opinions help explain the strong association between cluster membership and vaccination status in this multivariable modeling. The cluster composition was consistent with findings from studies of nonveteran populations that identified perceived vulnerability to COVID-19 infection, beliefs in vaccine effectiveness, and adherence with protective behaviors during the pandemic as contributors to vaccine uptake.^13,33 Qualitative themes showed that personal protection, protecting others, and vaccine mandates were frequent motivators for vaccination. Whereas protection of self and others from COVID-19 infection were more often expressed by the highly vaccinated Concerned Ambivalents, employment and travel vaccine mandates were more often identified by Unconcerned Disbelievers, who had a lower vaccination rate. Among Unconcerned Disbelievers, an employer vaccine requirement was the most frequent qualitative theme for overcoming vaccination concerns.

In addition to cluster membership, our modeling showed that trust in local VA HCPs and the CDC were independently associated with COVID-19 vaccination, which has been found in prior research.²⁰ This qualitative analyses regarding vaccine hesitancy identified trust-related concerns that were more frequently expressed by Unconcerned Disbelievers than Concerned Ambivalents. Concerns included the rapid development of the vaccines potentially limiting the generation of scientifically sound effectiveness and safety data, and potential biases involving the entities promoting vaccine uptake.

Whereas the Concerned Believers, Unconcerned Believers, and Concerned Ambivalents all had high COVID-19 vaccination rates (≥ 93%), the decision-making pathways to vaccine uptake likely differ by their concerns about COVID-19 infection and perceptions of vaccine safety and effectiveness. For example, this mixed-methods analysis consistently showed that people in the Concerned Ambivalent cluster were positively motivated by concerns about COVID-19 infection and severity and beliefs about vaccine effectiveness that were tempered by concerns about vaccine AEs. For this cluster, their frequent thematic expression that the benefits of the vaccine exceed the risks, and the positive social influences of family, friends, and HCPs may explain their high vaccination rate.

Such insights into how the patterns of COVID-19–related thoughts and feelings vary across clusters can be used to design interventions to encourage initial and booster doses of COVID-19 vaccines. For example, messaging that highlights the infectivity and severity of COVID-19 and the potential for persistent negative health outcomes associated with long COVID could reinforce the beliefs of Concerned Believers and Concerned Ambivalents, and such messaging could also be used as a targeted intervention for Unconcerned Believers who expressed fewer concerns about the health consequences of COVID-19.23 Likewise, messaging about the safety profile of COVID-19 vaccines may reduce vaccine hesitancy for Concerned Ambivalents. Importantly, purposeful attention to health equity, community engagement, and involvement of racially diverse HCPs in patient discussions represent successful strategies to increase COVID-19 vaccine uptake among Black individuals, who were disproportionately represented in the Concerned Ambivalent cluster and may possess higher levels of mistrust due to racism experienced within the health care system.²⁴

Our findings suggest that the greatest challenge for overcoming vaccine hesitancy is for individuals in the suboptimally vaccinated (30%) Unconcerned Disbeliever cluster. These individuals had low levels of concern about COVID-19 infection and severity, high levels of concern about vaccine safety, low perceived vaccine effectiveness, and low levels of trust in all information sources about COVID-19. While the Unconcerned Disbelievers cited scientifically reputable data sources, we were unable to verify whether participants accessed these reputable sources of information directly or obtained such information indirectly through potentially biased online sources. Nearly half of this cluster trusted their VA HCP and believed their community or religious leaders would want them to get vaccinated. This qualitative analyses found that Unconcerned Disbelievers relied on personal beliefs for vaccine decision-making more than Concerned Ambivalents. While Unconcerned Disbelievers were less likely to be socially influenced by family, friends, or religious leaders, they still acknowledged some impact from these sources. These findings suggest that addressing vaccine hesitancy among Unconcerned Disbelievers may require a multifaceted approach that respects their reliance on personal research while also leveraging the potential social influences. This approach supports the promising, previously reported practices of harnessing the social influences of HCPs and other community and religious leaders to promote vaccine uptake among Unconcerned Disbelievers.^34,35 One evidence-based approach to effectively change patient health care behaviors is through motivational interviewing strategies that use open-ended questions, nonjudgmental interactions, and collaborative decision-making when discussing the risks and benefits of vaccination.^21,22

Limitations

This study was conducted at a single VA health care facility and our sampling technique was nonrandom, suggesting that these results may not be generalizable to all veterans or non-VA patient populations. The 21% questionnaire response rate could have introduced selection bias into the respondent sample. All questionnaire data were self-reported, including vaccination status. Finally, the qualitative interviews consisted of a small number of unvaccinated individuals in 2 clusters (ie, Concerned Ambivalents and Unconcerned Disbelievers) and may not have reached thematic saturation in these subgroups.

Conclusions

Quantitative analyses identified 4 clusters based on individual thoughts and feelings about COVID-19 infection and vaccines. Cluster membership and levels of trust in COVID-19 information sources were independently associated with vaccination. Understanding the quantitative patterns of thoughts and beliefs across clusters, enriched by common qualitative themes for vaccine hesitancy, help inform tailored interventions to augment COVID-19 vaccine uptake and highlight the importance of targeted, trust-based communication and culturally sensitive interventions to enhance vaccine uptake across diverse populations.

The SARS-CoV-2 virus has resulted in > 778 million reported COVID-19 cases and > 7 million deaths worldwide. ¹ About 70% of the eligible US population has completed a primary COVID-19 vaccination series, yet only 17% have received an updated bivalent booster dose.² These immunization rates fall below the World Health Organization (WHO) target of 70%.³

Early in the pandemic, US Department of Veterans Affairs (VA) vaccination rates ranged from 46% to 71%.^4,5 Ensuring a high level of COVID-19 vaccination in the largest integrated US health care system aligns with the VA priority to provide high-quality, evidence-based care to a patient population that is older and has more comorbidities than the overall US population.^6-9

Vaccine hesitancy, defined as a “delay in acceptance or refusal of vaccination despite availability of vaccination service,” is a major contributor to suboptimal vaccination rates.^10-13 Previous studies used cluster analyses to identify the unique combinations of behavioral and social factors responsible for COVID-19 vaccine hesitancy.^10,11 Lack of perceived vaccine effectiveness and low perceived risk of the health consequences from COVID-19 infection were frequently identified in clusters where patients had the lowest intent for vaccination.^10,11 Similarly, low trust in health care practitioners (HCPs), government, and pharmaceutical companies diminished intent for vaccination in these clusters.¹⁰ These quantitative studies were limited by their exclusive focus on unvaccinated individuals, reliance on self-reported intent, and lack of assessment of a health care system with a COVID-19 vaccine delivery program designed to overcome barriers to health care access, such as the VA.

Prior qualitative studies of vaccine uptake in distinct veteran subgroups (ie, unhoused and in VA facilities with low vaccination rates) demonstrated that overriding medical priorities among the unhoused and vaccine safety concerns were associated with decreased vaccine uptake, and positive perceptions of HCPs and the health care system were associated with increased vaccine uptake.^11,12 However, these studies were conducted during periods of greater COVID-19 vaccine availability and acceptance, and prior to booster recommendations.^4,12,13

This mixed-methods quality improvement (QI) project assessed the barriers and facilitators of COVID-19 vaccination among veterans receiving primary care at a single VA health care facility. We assessed whether unique patient clusters could be identified based on COVID-19–related and vaccine-related thoughts and feelings and whether cluster membership was associated with COVID-19 vaccination. This analysis also explored how individuals’ beliefs and trust shaped motivations and hesitancies for vaccine uptake in quantitatively derived clusters with varying vaccination rates.

Methods

This QI project was conducted at the VA Pittsburgh Healthcare System (VAPHS), a tertiary care facility serving > 75,000 veterans in Pennsylvania, West Virginia, and Ohio. The VAPHS Institutional Review Board determined this QI study was exempt from review.^14-17 Participation was voluntary and had no bearing on VA health care or benefits. Financial support for the project, including key personnel and participant compensation, was provided by VAPHS. We followed the STROBE reporting guideline for cross-sectional studies and the COREQ checklist for qualitative research.^18,19

Quantitative Survey

The 32,271 veterans assigned to a VAPHS primary care HCP, effective April 1, 2020, were eligible. To ensure representation of subgroups underrecognized in research and/or QI projects, the sample included all 1980 female patients at VAPHS and a random sample of 500 White and 500 Hispanic and/or non-White men within 4 age categories (< 50, 50-64, 65-84, and > 84 years). For the < 50 years or > 84 years categories, all Hispanic and/or non-White men were included due to small sample sizes.^20-22 The nonrandom sampling frame comprised 1708 Hispanic and/or non-White men and 2000 White men. After assigning the 5688 potentially eligible individuals a unique identifier, 31 opted out, resulting in a final sample of 5657 individuals.

The 5657 individuals received a letter requesting their completion of a future questionnaire about COVID-19 infection and vaccines. An electronic Qualtrics questionnaire link was emailed to 3221 individuals; nonresponders received 2 follow-up email reminders. For the 2436 veterans without an email address on file, trained interviewers conducted phone surveys and entered responses. Those patients who completed the questionnaire could enter a drawing to win 1 of 100 cash prizes valued at $100. We collected questionnaire data from July to September 2021.

Questionnaire Items

We constructed a 60-item questionnaire based on prior research on COVID-19 vaccine hesitancy and the WHO Guidebook for Immunization Programs and Implementing Partners.^4,23-25 The WHO Guidebook comprises survey items organized within 4 domains reflecting the behavioral and social determinants of vaccination: thoughts and feelings; social processes; motivation and hesitancy; and practical factors.²³

Sociodemographic, clinical, and personal characteristics. The survey assessed respondent ethnicity and race and used these data to create a composite race and ethnicity variable. Highest educational level was also attained using 8 response options. The survey also assessed prior COVID-19 infection; prior receipt of vaccines for influenza, pneumonia, tetanus, or shingles; and presence of comorbidities that increase the risk of severe COVID-19 infection. We used administrative data from the VA Corporate Data Warehouse to determine respondent age, sex, geographic residence (urban, rural), and to fill in missing self-reported data on sex (n = 4) and ethnicity and race (n = 12). The survey assessed political views using a 5-point Likert scale (1, very liberal; 5, very conservative) and was collapsed into 3 categories (ie, very conservative or conservative, moderate, very liberal or liberal), with prefer not to answer reported separately

COVID-19 infection and vaccine. We asked veterans if they had ever been infected with COVID-19, whether they had been offered and/or received a COVID-19 vaccine, and type (Pfizer, Moderna, or Johnson & Johnson), and number of doses received. Positive vaccination status was defined as the receipt of ≥ 1 dose of a COVID-19 vaccine approved by the US Food and Drug Administration.

COVID-19 opinions. Respondents were asked about perceived risk of COVID-19 infection and related health outcomes, as well as beliefs about COVID-19 vaccines, using a 4-point Likert scale for all items: (1, not at all concerned; 4, very concerned). Respondents were asked about concerns related to COVID-19 infection and severe illness. They also were asked about vaccine-related short-term adverse effects (AEs) and long-term complications. Respondents were asked how effective they believed COVID-19 vaccines were at preventing infection, serious illness, or death. Unvaccinated and vaccinated veterans were asked similar items, with a qualifier of “before getting vaccinated…” for those who were vaccinated.

Social processes. Respondents were asked to rate their level of trust in various sources of COVID-19 vaccine information using a 4-point Likert scale (1, trust not at all; 4, trust very much). Respondents were asked whether community or religious leaders or close family or friends wanted them to get vaccinated (yes, no, or unsure).

Practical factors. Respondents were asked to rate the logistical difficulty of getting vaccinated or trying to get vaccinated using a 4-point Likert scale (1, not at all; 4, extremely).

Participants

Respondents were asked to participate in a follow-up qualitative interview. Among 293 participants who agreed, we sampled all 86 unvaccinated individuals regardless of cluster assignment, a random sample of 88 individuals in the cluster with the lowest vaccination rate, and all 33 vaccinated individuals in the cluster with the second-lowest vaccination rate. Forty-nine veterans completed qualitative interviews.

Two research staff trained in qualitative research completed telephone interviews, averaging 16.5 minutes (March to May 2022), using semistructured scripts to elicit vaccine-related motivations, hesitancies, or concerns. Interviews were recorded, transcribed, and deidentified. Participants provided written consent for recording and received $50 cash-equivalent compensation for interview completion.

Qualitative Interview Script

The interview script consisted of open-ended questions related to vaccine uptake across WHO domains.²³ Both unvaccinated and vaccinated respondents were asked similar questions and customized questions about boosters for the vaccinated subgroup. To assess motivations and hesitancies, respondents were asked how they made their decisions about vaccination and what they considered when deciding. Vaccinated participants were asked about motivations and overcoming concerns. Unvaccinated respondents were asked about reasons for concern. To assess social processes, the interviewers asked participants whose opinion or counsel they trusted when deciding whether to get vaccinated. Questions also focused on positive experiences and vaccination barriers. Vaccinated participants were asked what could have improved their vaccination experiences. Finally, the interviewers asked participants who received a complete primary vaccine series about their motivations and plans related to booster vaccines, and whether information about emerging COVID-19 variants influenced their decisions.

Data Analyses

This analysis used X² and Fisher exact tests to assess the associations among respondent characteristics, questionnaire responses, vaccination status, and cluster membership. Items phrased similarly were handled in a similar fashion for vaccinated and unvaccinated respondents.

Cluster analysis assessed the possible groupings in responses to the quantitative questionnaire items focused on thoughts and feelings about COVID-19 infection risk and severity, vaccine effectiveness, and vaccine safety. This analysis treated the items’ ordinal response categories as continuous. We performed factor analysis using principal component analysis to explore dimension reduction and account for covariance between items. Two principal components were calculated and applied k-means clustering, determining the number of clusters through agreement from the elbow, gap statistic, and silhouette methods.²⁶ Each cluster was named based on its unique pattern of responses to the items used to define them (eAppendix 1).

Multivariable logistic regression analyses assessed the independent association between cluster membership as the independent measure and vaccination status as the dependent measure, adjusting for respondent sociodemographic and personal characteristics and 2 measures of trust (ie, local VA HCP and the CDC). We selected these trust measures because they represent objective sources of medical information and were independently associated with COVID-19 vaccination status in a logistic regression model comprising all 6 trust items assessed.

This study defined statistical significance as a 2-tailed P value < .05. SAS 9.4 was used for all statistical analyses and Python 3.7.4 and the Scikit-learn package for cluster analyses.²⁷ For qualitative analyses, this study used an inductive thematic approach guided by conventional qualitative content analysis, NVivo 12 Plus for Windows to code and analyze interview transcripts.^28,29 We created an initial codebook based on 10 transcripts that were selected for high complexity and represented cluster membership and vaccination status.^30,31 After 2 qualitative staff developed the initial codebook, 11 of 49 (22%) transcripts were independently coded by a primary and secondary coder to ensure consistent code application. Both coders reviewed the cocoded transcripts and resolved all discrepancies through negotiated consensus.³² After the cocoding process was complete, the primary coder coded the remaining transcripts. The primary and secondary coder met as needed to review and discuss any questions that arose during the primary coder’s work.

Results

Of 5657 eligible participants, 1208 (21.4%) completed a questionnaire. Overall, 674 (55.8%) were aged < 65 years, 530 (43.9%) were women, 828 (68.5%) were non-Hispanic White, 303 (25.1%) were Black, and 47 (3.9%) were Hispanic, and 1034 (85.6%) were vaccinated (Table 1). Compared to the total sampled population, respondents were more often older, female, and White (eAppendix 2).

Cluster Membership

Four clusters were identified from 1183 (97.9%) participants who provided complete responses to 6 items assessing thoughts and feelings about COVID-19 infection and vaccines (Table 2). Of the 1183 respondents, 375 (31.7%) were Concerned Believers (cluster 1), 336 (28.4%) were Unconcerned Believers (cluster 2), 298 (25.2%) were Concerned Ambivalents (cluster 3), and 174 (14.7%) were Unconcerned Disbelievers (cluster 4). The Concerned Believers were moderately/ very concerned about COVID-19 infection (96.0%) and becoming very ill from infection (94.6%), believed the vaccine was moderately/very effective in preventing COVID-19 infection (100%) and severe illness or death from infection (98.7%), and had slight concern about short-term AEs (92.6%) or long-term complications (92.0%) from the vaccine. The Unconcerned Believers had no/slight concern about COVID-19 infection (76.5%) or becoming very ill (79.2%), believed the vaccine was effective in preventing infection (82.4%) and severe illness and death (83.6%), and had no/slight concern about short-term AEs (94.0%) or long-term complications (87.2%) from the vaccine. The Concerned Ambivalents were moderately/ very concerned about COVID-19 infection (94.3%) and becoming very ill (93.6%), believed the vaccine was moderately/very effective in preventing infection (86.6%) and severe illness or death (86.9%), and were moderately/very concerned about short-term AEs (81.9%) or long-term complications (89.3%) from the vaccine. The Unconcerned Disbelievers had no/slight concern about COVID-19 infection (90.8%) and becoming very ill (88.6%), believed the vaccine was not at all/slightly effective in preventing infection (90.3%) and severe illness or death (87.4%), and were moderately/very concerned about short-term AEs (52.8%) or long-term complications (75.9%) from the vaccine.

Cluster Membership

Respondent age, race and ethnicity, and political viewpoints differed significantly by cluster (P < .001). Compared with the other clusters, the Concerned Believer cluster was older (55.5% age ≥ 65 years vs 16.7%-48.0%) and more frequently reported liberal political views (28.8% vs 4.6%-15.1%). In contrast, the Unconcerned Disbeliever cluster was younger (83.4% age ≤ 64 years vs 44.5%-56.8%) and more frequently reported conservative political views (37.9% vs 17.1%-26.8%) than the other clusters. Whereas the Concerned Ambivalent cluster had the highest proportion of Black (37.7%) and the lowest proportion of White respondents (57.6%), the Unconcerned Disbelievers cluster had the lowest proportion of Black respondents (14.5%) and the highest proportion of White respondents (77.9%). The Unconcerned Disbelievers cluster were significantly less likely to trust COVID-19 vaccine information from any source and to believe those close to them wanted them to get vaccinated.

Association of Cluster Membership and COVID-19 Vaccination

COVID-19 vaccination rates varied more than 3-fold (P < .001) by cluster, with 29.9% of Unconcerned Disbelievers, 93.3% of Concerned Ambivalents, 93.5% of Unconcerned Believers, and 98.9% of Concerned Believers reporting being vaccinated. (Figure). Cluster membership was independently associated with vaccination, with adjusted odds ratios (AORs) of 12.0 (95% CI, 6.1-23.8) for the Concerned Ambivalent, 13.0 (95% CI, 6.9-24.5) for Unconcerned Believer, and 48.6 (95% CI, 15.5-152.1) for Concerned Believer clusters (Table 3). Respondent trust in COVID-19 vaccine information from their VA HCP (AOR 2.1; 95% CI, 1.6-2.8) and the CDC (AOR 1.6; 95% CI, 1.2-2.1) were independently associated with vaccination status, while the remaining respondent sociodemographic or personal characteristics were not.

Qualitative Interview Participants

A 49-participant convenience sample completed interviews, including 30 Concerned Ambivalent, 17 Unconcerned Disbeliever, and 2 Unconcerned Believer respondents cluster. The data were not calculated for Unconcerned Believers due to the small sample size. Interview participants were more likely to be younger, female, non-Hispanic, White, less educated, and more politically conservative than the questionnaire respondents as a whole (Appendix). The vaccination rate for the interview participants was 73.5%, ranging from 29.9% in the Unconcerned Disbeliever to 93.3% in the Concerned Ambivalent cluster. Qualitative themes and participant quotes for Concerned Ambivalent and Unconcerned Disbeliever respondents are in eAppendix 3.

Motivations. Wanting personal protection from becoming infected or severely ill from COVID-19 (63.8%), caregiver wanting to protect others (17.0%), and employment vaccine requirements (14.9%) were frequent motivations for vaccination. Whereas personal protection (90.0%) and protection of others (23.3%) were identified more frequently in the Concerned Ambivalents cluster, employment vaccine requirements (35.3%) were more frequently identified in the Unconcerned Disbelievers cluster.

Hesitancies or concerns. Lack of sufficient information related to rapid vaccine development (55.3%), vaccine AEs (38.3%), and low confidence in vaccine efficacy (23.4%) were frequent concerns or hesitancies about vaccination. Unconcerned Disbelievers expressed higher levels of concern about the vaccine’s rapid development (82.4%), low perceived vaccine efficacy (47.1%), and a lack of trust in governmental vaccine promotion (23.5%) than did the Concerned Ambivalents.

Overcoming concerns. Not wanting to get sick or die from infection coupled with an understanding that vaccine benefits exceed risks (23.4%) and receiving information from a trusted source (10.6%) were common ways of overcoming concerns for vaccination. Although the Unconcerned Disbelievers infrequently identified reasons for overcoming concerns, they identified employment requirements (17.6%) as a reason for vaccination despite concerns. They also identified seeing others with positive vaccine experiences and pressure from family or friends as ways of overcoming concerns (11.8% each).

Social influences. Family members or partners (38.3%), personal opinions (38.3%), and HCPs (23.4%) were frequent social influences for vaccination. Concerned Ambivalents mentioned family members and partners (46.7%), HCPs (26.7%), and friends (20.0%) as common influences, while Unconcerned Disbelievers more frequently relied on their opinion (41.2%) and quoted specific scientifically reputable data sources (17.6%) to guide vaccine decision-making, although it is unclear whether these sources were accessed directly or if this information was obtained indirectly through scientifically unvetted data platforms.

Practical factors. Most participants had positive vaccination experiences (68.1%), determined mainly by the Concerned Ambivalents (90.0%), who were more highly vaccinated. Barriers to vaccination were reported by 9 (19.1%) participants, driven by those in the Concerned Ambivalent cluster (26.7%). Eight (17.0%) participants suggested improvements for vaccination processes, with similar overall reporting frequencies across clusters.

COVID-19 boosters and variants. Wanting continued protection from COVID-19 (36.2%), recommendations from a doctor or trusted source (17.0%), and news about emerging variants (10.6%) were frequent motivations for receiving a vaccine booster (eAppendix 4). These motivations were largely driven by the Concerned Ambivalents, of whom 25 of 30 were booster eligible and 24 received a booster dose. Belief that boosters were unnecessary (8.5%), concerns about efficacy (6.4%), and concerns about AEs (6.4%) were frequently identified hesitancies. These concerns were expressed largely by the Unconcerned Disbelievers, of whom 7 of 17 were booster dose eligible, but only 1 received a dose.

Evolving knowledge about variants was not a major concern overall and did not change existing opinions about the vaccine (36.2%). Concerned Ambivalents believed vaccination provided extra protection against variants (36.7%) and the emergence of variants served as a reminder of the ongoing pandemic (30.0%). In contrast, Unconcerned Disbelievers believed that the threat of variants was overblown (35.3%) and mutations are to be expected (17.6%).

Discussion

This study used a complementary mixed-methods approach to understand the motivations, hesitancies, and social and practical drivers of COVID-19 vaccine uptake among VA beneficiaries. Our quantitative analyses identified 4 distinct clusters based on respondents’ opinions on COVID-19 infection severity and vaccine effectiveness and safety. Veterans in 3 clusters were 12 to 49 times more likely to be vaccinated than those in the remaining cluster, even when controlling for baseline respondent characteristics and level of trust in credible sources of COVID-19 information. The observed vaccination rate of nearly 86% was higher than the contemporaneous national average of 62% for vaccine-eligible individuals, likely reflecting the comprehensive VA vaccine promotion strategies tailored to a patient demographic with a high COVID-19 risk profile.^2,10

This cluster analyses demonstrated the importance of thoughts and feelings about COVID-19 infection and vaccination as influential social and behavioral drivers of vaccine uptake. These opinions help explain the strong association between cluster membership and vaccination status in this multivariable modeling. The cluster composition was consistent with findings from studies of nonveteran populations that identified perceived vulnerability to COVID-19 infection, beliefs in vaccine effectiveness, and adherence with protective behaviors during the pandemic as contributors to vaccine uptake.^13,33 Qualitative themes showed that personal protection, protecting others, and vaccine mandates were frequent motivators for vaccination. Whereas protection of self and others from COVID-19 infection were more often expressed by the highly vaccinated Concerned Ambivalents, employment and travel vaccine mandates were more often identified by Unconcerned Disbelievers, who had a lower vaccination rate. Among Unconcerned Disbelievers, an employer vaccine requirement was the most frequent qualitative theme for overcoming vaccination concerns.

In addition to cluster membership, our modeling showed that trust in local VA HCPs and the CDC were independently associated with COVID-19 vaccination, which has been found in prior research.²⁰ This qualitative analyses regarding vaccine hesitancy identified trust-related concerns that were more frequently expressed by Unconcerned Disbelievers than Concerned Ambivalents. Concerns included the rapid development of the vaccines potentially limiting the generation of scientifically sound effectiveness and safety data, and potential biases involving the entities promoting vaccine uptake.

Whereas the Concerned Believers, Unconcerned Believers, and Concerned Ambivalents all had high COVID-19 vaccination rates (≥ 93%), the decision-making pathways to vaccine uptake likely differ by their concerns about COVID-19 infection and perceptions of vaccine safety and effectiveness. For example, this mixed-methods analysis consistently showed that people in the Concerned Ambivalent cluster were positively motivated by concerns about COVID-19 infection and severity and beliefs about vaccine effectiveness that were tempered by concerns about vaccine AEs. For this cluster, their frequent thematic expression that the benefits of the vaccine exceed the risks, and the positive social influences of family, friends, and HCPs may explain their high vaccination rate.

Such insights into how the patterns of COVID-19–related thoughts and feelings vary across clusters can be used to design interventions to encourage initial and booster doses of COVID-19 vaccines. For example, messaging that highlights the infectivity and severity of COVID-19 and the potential for persistent negative health outcomes associated with long COVID could reinforce the beliefs of Concerned Believers and Concerned Ambivalents, and such messaging could also be used as a targeted intervention for Unconcerned Believers who expressed fewer concerns about the health consequences of COVID-19.23 Likewise, messaging about the safety profile of COVID-19 vaccines may reduce vaccine hesitancy for Concerned Ambivalents. Importantly, purposeful attention to health equity, community engagement, and involvement of racially diverse HCPs in patient discussions represent successful strategies to increase COVID-19 vaccine uptake among Black individuals, who were disproportionately represented in the Concerned Ambivalent cluster and may possess higher levels of mistrust due to racism experienced within the health care system.²⁴

Our findings suggest that the greatest challenge for overcoming vaccine hesitancy is for individuals in the suboptimally vaccinated (30%) Unconcerned Disbeliever cluster. These individuals had low levels of concern about COVID-19 infection and severity, high levels of concern about vaccine safety, low perceived vaccine effectiveness, and low levels of trust in all information sources about COVID-19. While the Unconcerned Disbelievers cited scientifically reputable data sources, we were unable to verify whether participants accessed these reputable sources of information directly or obtained such information indirectly through potentially biased online sources. Nearly half of this cluster trusted their VA HCP and believed their community or religious leaders would want them to get vaccinated. This qualitative analyses found that Unconcerned Disbelievers relied on personal beliefs for vaccine decision-making more than Concerned Ambivalents. While Unconcerned Disbelievers were less likely to be socially influenced by family, friends, or religious leaders, they still acknowledged some impact from these sources. These findings suggest that addressing vaccine hesitancy among Unconcerned Disbelievers may require a multifaceted approach that respects their reliance on personal research while also leveraging the potential social influences. This approach supports the promising, previously reported practices of harnessing the social influences of HCPs and other community and religious leaders to promote vaccine uptake among Unconcerned Disbelievers.^34,35 One evidence-based approach to effectively change patient health care behaviors is through motivational interviewing strategies that use open-ended questions, nonjudgmental interactions, and collaborative decision-making when discussing the risks and benefits of vaccination.^21,22

Limitations

This study was conducted at a single VA health care facility and our sampling technique was nonrandom, suggesting that these results may not be generalizable to all veterans or non-VA patient populations. The 21% questionnaire response rate could have introduced selection bias into the respondent sample. All questionnaire data were self-reported, including vaccination status. Finally, the qualitative interviews consisted of a small number of unvaccinated individuals in 2 clusters (ie, Concerned Ambivalents and Unconcerned Disbelievers) and may not have reached thematic saturation in these subgroups.

Conclusions

Quantitative analyses identified 4 clusters based on individual thoughts and feelings about COVID-19 infection and vaccines. Cluster membership and levels of trust in COVID-19 information sources were independently associated with vaccination. Understanding the quantitative patterns of thoughts and beliefs across clusters, enriched by common qualitative themes for vaccine hesitancy, help inform tailored interventions to augment COVID-19 vaccine uptake and highlight the importance of targeted, trust-based communication and culturally sensitive interventions to enhance vaccine uptake across diverse populations.

The SARS-CoV-2 virus has resulted in > 778 million reported COVID-19 cases and > 7 million deaths worldwide. ¹ About 70% of the eligible US population has completed a primary COVID-19 vaccination series, yet only 17% have received an updated bivalent booster dose.² These immunization rates fall below the World Health Organization (WHO) target of 70%.³

Early in the pandemic, US Department of Veterans Affairs (VA) vaccination rates ranged from 46% to 71%.^4,5 Ensuring a high level of COVID-19 vaccination in the largest integrated US health care system aligns with the VA priority to provide high-quality, evidence-based care to a patient population that is older and has more comorbidities than the overall US population.^6-9

Vaccine hesitancy, defined as a “delay in acceptance or refusal of vaccination despite availability of vaccination service,” is a major contributor to suboptimal vaccination rates.^10-13 Previous studies used cluster analyses to identify the unique combinations of behavioral and social factors responsible for COVID-19 vaccine hesitancy.^10,11 Lack of perceived vaccine effectiveness and low perceived risk of the health consequences from COVID-19 infection were frequently identified in clusters where patients had the lowest intent for vaccination.^10,11 Similarly, low trust in health care practitioners (HCPs), government, and pharmaceutical companies diminished intent for vaccination in these clusters.¹⁰ These quantitative studies were limited by their exclusive focus on unvaccinated individuals, reliance on self-reported intent, and lack of assessment of a health care system with a COVID-19 vaccine delivery program designed to overcome barriers to health care access, such as the VA.

Prior qualitative studies of vaccine uptake in distinct veteran subgroups (ie, unhoused and in VA facilities with low vaccination rates) demonstrated that overriding medical priorities among the unhoused and vaccine safety concerns were associated with decreased vaccine uptake, and positive perceptions of HCPs and the health care system were associated with increased vaccine uptake.^11,12 However, these studies were conducted during periods of greater COVID-19 vaccine availability and acceptance, and prior to booster recommendations.^4,12,13

This mixed-methods quality improvement (QI) project assessed the barriers and facilitators of COVID-19 vaccination among veterans receiving primary care at a single VA health care facility. We assessed whether unique patient clusters could be identified based on COVID-19–related and vaccine-related thoughts and feelings and whether cluster membership was associated with COVID-19 vaccination. This analysis also explored how individuals’ beliefs and trust shaped motivations and hesitancies for vaccine uptake in quantitatively derived clusters with varying vaccination rates.

Methods

This QI project was conducted at the VA Pittsburgh Healthcare System (VAPHS), a tertiary care facility serving > 75,000 veterans in Pennsylvania, West Virginia, and Ohio. The VAPHS Institutional Review Board determined this QI study was exempt from review.^14-17 Participation was voluntary and had no bearing on VA health care or benefits. Financial support for the project, including key personnel and participant compensation, was provided by VAPHS. We followed the STROBE reporting guideline for cross-sectional studies and the COREQ checklist for qualitative research.^18,19

Quantitative Survey

The 32,271 veterans assigned to a VAPHS primary care HCP, effective April 1, 2020, were eligible. To ensure representation of subgroups underrecognized in research and/or QI projects, the sample included all 1980 female patients at VAPHS and a random sample of 500 White and 500 Hispanic and/or non-White men within 4 age categories (< 50, 50-64, 65-84, and > 84 years). For the < 50 years or > 84 years categories, all Hispanic and/or non-White men were included due to small sample sizes.^20-22 The nonrandom sampling frame comprised 1708 Hispanic and/or non-White men and 2000 White men. After assigning the 5688 potentially eligible individuals a unique identifier, 31 opted out, resulting in a final sample of 5657 individuals.

The 5657 individuals received a letter requesting their completion of a future questionnaire about COVID-19 infection and vaccines. An electronic Qualtrics questionnaire link was emailed to 3221 individuals; nonresponders received 2 follow-up email reminders. For the 2436 veterans without an email address on file, trained interviewers conducted phone surveys and entered responses. Those patients who completed the questionnaire could enter a drawing to win 1 of 100 cash prizes valued at $100. We collected questionnaire data from July to September 2021.

Questionnaire Items

We constructed a 60-item questionnaire based on prior research on COVID-19 vaccine hesitancy and the WHO Guidebook for Immunization Programs and Implementing Partners.^4,23-25 The WHO Guidebook comprises survey items organized within 4 domains reflecting the behavioral and social determinants of vaccination: thoughts and feelings; social processes; motivation and hesitancy; and practical factors.²³

Sociodemographic, clinical, and personal characteristics. The survey assessed respondent ethnicity and race and used these data to create a composite race and ethnicity variable. Highest educational level was also attained using 8 response options. The survey also assessed prior COVID-19 infection; prior receipt of vaccines for influenza, pneumonia, tetanus, or shingles; and presence of comorbidities that increase the risk of severe COVID-19 infection. We used administrative data from the VA Corporate Data Warehouse to determine respondent age, sex, geographic residence (urban, rural), and to fill in missing self-reported data on sex (n = 4) and ethnicity and race (n = 12). The survey assessed political views using a 5-point Likert scale (1, very liberal; 5, very conservative) and was collapsed into 3 categories (ie, very conservative or conservative, moderate, very liberal or liberal), with prefer not to answer reported separately

COVID-19 infection and vaccine. We asked veterans if they had ever been infected with COVID-19, whether they had been offered and/or received a COVID-19 vaccine, and type (Pfizer, Moderna, or Johnson & Johnson), and number of doses received. Positive vaccination status was defined as the receipt of ≥ 1 dose of a COVID-19 vaccine approved by the US Food and Drug Administration.

COVID-19 opinions. Respondents were asked about perceived risk of COVID-19 infection and related health outcomes, as well as beliefs about COVID-19 vaccines, using a 4-point Likert scale for all items: (1, not at all concerned; 4, very concerned). Respondents were asked about concerns related to COVID-19 infection and severe illness. They also were asked about vaccine-related short-term adverse effects (AEs) and long-term complications. Respondents were asked how effective they believed COVID-19 vaccines were at preventing infection, serious illness, or death. Unvaccinated and vaccinated veterans were asked similar items, with a qualifier of “before getting vaccinated…” for those who were vaccinated.

Social processes. Respondents were asked to rate their level of trust in various sources of COVID-19 vaccine information using a 4-point Likert scale (1, trust not at all; 4, trust very much). Respondents were asked whether community or religious leaders or close family or friends wanted them to get vaccinated (yes, no, or unsure).

Practical factors. Respondents were asked to rate the logistical difficulty of getting vaccinated or trying to get vaccinated using a 4-point Likert scale (1, not at all; 4, extremely).

Participants

Respondents were asked to participate in a follow-up qualitative interview. Among 293 participants who agreed, we sampled all 86 unvaccinated individuals regardless of cluster assignment, a random sample of 88 individuals in the cluster with the lowest vaccination rate, and all 33 vaccinated individuals in the cluster with the second-lowest vaccination rate. Forty-nine veterans completed qualitative interviews.

Two research staff trained in qualitative research completed telephone interviews, averaging 16.5 minutes (March to May 2022), using semistructured scripts to elicit vaccine-related motivations, hesitancies, or concerns. Interviews were recorded, transcribed, and deidentified. Participants provided written consent for recording and received $50 cash-equivalent compensation for interview completion.

Qualitative Interview Script

The interview script consisted of open-ended questions related to vaccine uptake across WHO domains.²³ Both unvaccinated and vaccinated respondents were asked similar questions and customized questions about boosters for the vaccinated subgroup. To assess motivations and hesitancies, respondents were asked how they made their decisions about vaccination and what they considered when deciding. Vaccinated participants were asked about motivations and overcoming concerns. Unvaccinated respondents were asked about reasons for concern. To assess social processes, the interviewers asked participants whose opinion or counsel they trusted when deciding whether to get vaccinated. Questions also focused on positive experiences and vaccination barriers. Vaccinated participants were asked what could have improved their vaccination experiences. Finally, the interviewers asked participants who received a complete primary vaccine series about their motivations and plans related to booster vaccines, and whether information about emerging COVID-19 variants influenced their decisions.

Data Analyses

This analysis used X² and Fisher exact tests to assess the associations among respondent characteristics, questionnaire responses, vaccination status, and cluster membership. Items phrased similarly were handled in a similar fashion for vaccinated and unvaccinated respondents.

Cluster analysis assessed the possible groupings in responses to the quantitative questionnaire items focused on thoughts and feelings about COVID-19 infection risk and severity, vaccine effectiveness, and vaccine safety. This analysis treated the items’ ordinal response categories as continuous. We performed factor analysis using principal component analysis to explore dimension reduction and account for covariance between items. Two principal components were calculated and applied k-means clustering, determining the number of clusters through agreement from the elbow, gap statistic, and silhouette methods.²⁶ Each cluster was named based on its unique pattern of responses to the items used to define them (eAppendix 1).

Multivariable logistic regression analyses assessed the independent association between cluster membership as the independent measure and vaccination status as the dependent measure, adjusting for respondent sociodemographic and personal characteristics and 2 measures of trust (ie, local VA HCP and the CDC). We selected these trust measures because they represent objective sources of medical information and were independently associated with COVID-19 vaccination status in a logistic regression model comprising all 6 trust items assessed.

This study defined statistical significance as a 2-tailed P value < .05. SAS 9.4 was used for all statistical analyses and Python 3.7.4 and the Scikit-learn package for cluster analyses.²⁷ For qualitative analyses, this study used an inductive thematic approach guided by conventional qualitative content analysis, NVivo 12 Plus for Windows to code and analyze interview transcripts.^28,29 We created an initial codebook based on 10 transcripts that were selected for high complexity and represented cluster membership and vaccination status.^30,31 After 2 qualitative staff developed the initial codebook, 11 of 49 (22%) transcripts were independently coded by a primary and secondary coder to ensure consistent code application. Both coders reviewed the cocoded transcripts and resolved all discrepancies through negotiated consensus.³² After the cocoding process was complete, the primary coder coded the remaining transcripts. The primary and secondary coder met as needed to review and discuss any questions that arose during the primary coder’s work.

Results

Of 5657 eligible participants, 1208 (21.4%) completed a questionnaire. Overall, 674 (55.8%) were aged < 65 years, 530 (43.9%) were women, 828 (68.5%) were non-Hispanic White, 303 (25.1%) were Black, and 47 (3.9%) were Hispanic, and 1034 (85.6%) were vaccinated (Table 1). Compared to the total sampled population, respondents were more often older, female, and White (eAppendix 2).

Cluster Membership

Four clusters were identified from 1183 (97.9%) participants who provided complete responses to 6 items assessing thoughts and feelings about COVID-19 infection and vaccines (Table 2). Of the 1183 respondents, 375 (31.7%) were Concerned Believers (cluster 1), 336 (28.4%) were Unconcerned Believers (cluster 2), 298 (25.2%) were Concerned Ambivalents (cluster 3), and 174 (14.7%) were Unconcerned Disbelievers (cluster 4). The Concerned Believers were moderately/ very concerned about COVID-19 infection (96.0%) and becoming very ill from infection (94.6%), believed the vaccine was moderately/very effective in preventing COVID-19 infection (100%) and severe illness or death from infection (98.7%), and had slight concern about short-term AEs (92.6%) or long-term complications (92.0%) from the vaccine. The Unconcerned Believers had no/slight concern about COVID-19 infection (76.5%) or becoming very ill (79.2%), believed the vaccine was effective in preventing infection (82.4%) and severe illness and death (83.6%), and had no/slight concern about short-term AEs (94.0%) or long-term complications (87.2%) from the vaccine. The Concerned Ambivalents were moderately/ very concerned about COVID-19 infection (94.3%) and becoming very ill (93.6%), believed the vaccine was moderately/very effective in preventing infection (86.6%) and severe illness or death (86.9%), and were moderately/very concerned about short-term AEs (81.9%) or long-term complications (89.3%) from the vaccine. The Unconcerned Disbelievers had no/slight concern about COVID-19 infection (90.8%) and becoming very ill (88.6%), believed the vaccine was not at all/slightly effective in preventing infection (90.3%) and severe illness or death (87.4%), and were moderately/very concerned about short-term AEs (52.8%) or long-term complications (75.9%) from the vaccine.

Cluster Membership

Respondent age, race and ethnicity, and political viewpoints differed significantly by cluster (P < .001). Compared with the other clusters, the Concerned Believer cluster was older (55.5% age ≥ 65 years vs 16.7%-48.0%) and more frequently reported liberal political views (28.8% vs 4.6%-15.1%). In contrast, the Unconcerned Disbeliever cluster was younger (83.4% age ≤ 64 years vs 44.5%-56.8%) and more frequently reported conservative political views (37.9% vs 17.1%-26.8%) than the other clusters. Whereas the Concerned Ambivalent cluster had the highest proportion of Black (37.7%) and the lowest proportion of White respondents (57.6%), the Unconcerned Disbelievers cluster had the lowest proportion of Black respondents (14.5%) and the highest proportion of White respondents (77.9%). The Unconcerned Disbelievers cluster were significantly less likely to trust COVID-19 vaccine information from any source and to believe those close to them wanted them to get vaccinated.

Association of Cluster Membership and COVID-19 Vaccination

COVID-19 vaccination rates varied more than 3-fold (P < .001) by cluster, with 29.9% of Unconcerned Disbelievers, 93.3% of Concerned Ambivalents, 93.5% of Unconcerned Believers, and 98.9% of Concerned Believers reporting being vaccinated. (Figure). Cluster membership was independently associated with vaccination, with adjusted odds ratios (AORs) of 12.0 (95% CI, 6.1-23.8) for the Concerned Ambivalent, 13.0 (95% CI, 6.9-24.5) for Unconcerned Believer, and 48.6 (95% CI, 15.5-152.1) for Concerned Believer clusters (Table 3). Respondent trust in COVID-19 vaccine information from their VA HCP (AOR 2.1; 95% CI, 1.6-2.8) and the CDC (AOR 1.6; 95% CI, 1.2-2.1) were independently associated with vaccination status, while the remaining respondent sociodemographic or personal characteristics were not.

Qualitative Interview Participants

A 49-participant convenience sample completed interviews, including 30 Concerned Ambivalent, 17 Unconcerned Disbeliever, and 2 Unconcerned Believer respondents cluster. The data were not calculated for Unconcerned Believers due to the small sample size. Interview participants were more likely to be younger, female, non-Hispanic, White, less educated, and more politically conservative than the questionnaire respondents as a whole (Appendix). The vaccination rate for the interview participants was 73.5%, ranging from 29.9% in the Unconcerned Disbeliever to 93.3% in the Concerned Ambivalent cluster. Qualitative themes and participant quotes for Concerned Ambivalent and Unconcerned Disbeliever respondents are in eAppendix 3.

Motivations. Wanting personal protection from becoming infected or severely ill from COVID-19 (63.8%), caregiver wanting to protect others (17.0%), and employment vaccine requirements (14.9%) were frequent motivations for vaccination. Whereas personal protection (90.0%) and protection of others (23.3%) were identified more frequently in the Concerned Ambivalents cluster, employment vaccine requirements (35.3%) were more frequently identified in the Unconcerned Disbelievers cluster.

Hesitancies or concerns. Lack of sufficient information related to rapid vaccine development (55.3%), vaccine AEs (38.3%), and low confidence in vaccine efficacy (23.4%) were frequent concerns or hesitancies about vaccination. Unconcerned Disbelievers expressed higher levels of concern about the vaccine’s rapid development (82.4%), low perceived vaccine efficacy (47.1%), and a lack of trust in governmental vaccine promotion (23.5%) than did the Concerned Ambivalents.

Overcoming concerns. Not wanting to get sick or die from infection coupled with an understanding that vaccine benefits exceed risks (23.4%) and receiving information from a trusted source (10.6%) were common ways of overcoming concerns for vaccination. Although the Unconcerned Disbelievers infrequently identified reasons for overcoming concerns, they identified employment requirements (17.6%) as a reason for vaccination despite concerns. They also identified seeing others with positive vaccine experiences and pressure from family or friends as ways of overcoming concerns (11.8% each).

Social influences. Family members or partners (38.3%), personal opinions (38.3%), and HCPs (23.4%) were frequent social influences for vaccination. Concerned Ambivalents mentioned family members and partners (46.7%), HCPs (26.7%), and friends (20.0%) as common influences, while Unconcerned Disbelievers more frequently relied on their opinion (41.2%) and quoted specific scientifically reputable data sources (17.6%) to guide vaccine decision-making, although it is unclear whether these sources were accessed directly or if this information was obtained indirectly through scientifically unvetted data platforms.

Practical factors. Most participants had positive vaccination experiences (68.1%), determined mainly by the Concerned Ambivalents (90.0%), who were more highly vaccinated. Barriers to vaccination were reported by 9 (19.1%) participants, driven by those in the Concerned Ambivalent cluster (26.7%). Eight (17.0%) participants suggested improvements for vaccination processes, with similar overall reporting frequencies across clusters.

COVID-19 boosters and variants. Wanting continued protection from COVID-19 (36.2%), recommendations from a doctor or trusted source (17.0%), and news about emerging variants (10.6%) were frequent motivations for receiving a vaccine booster (eAppendix 4). These motivations were largely driven by the Concerned Ambivalents, of whom 25 of 30 were booster eligible and 24 received a booster dose. Belief that boosters were unnecessary (8.5%), concerns about efficacy (6.4%), and concerns about AEs (6.4%) were frequently identified hesitancies. These concerns were expressed largely by the Unconcerned Disbelievers, of whom 7 of 17 were booster dose eligible, but only 1 received a dose.

Evolving knowledge about variants was not a major concern overall and did not change existing opinions about the vaccine (36.2%). Concerned Ambivalents believed vaccination provided extra protection against variants (36.7%) and the emergence of variants served as a reminder of the ongoing pandemic (30.0%). In contrast, Unconcerned Disbelievers believed that the threat of variants was overblown (35.3%) and mutations are to be expected (17.6%).

Discussion

This study used a complementary mixed-methods approach to understand the motivations, hesitancies, and social and practical drivers of COVID-19 vaccine uptake among VA beneficiaries. Our quantitative analyses identified 4 distinct clusters based on respondents’ opinions on COVID-19 infection severity and vaccine effectiveness and safety. Veterans in 3 clusters were 12 to 49 times more likely to be vaccinated than those in the remaining cluster, even when controlling for baseline respondent characteristics and level of trust in credible sources of COVID-19 information. The observed vaccination rate of nearly 86% was higher than the contemporaneous national average of 62% for vaccine-eligible individuals, likely reflecting the comprehensive VA vaccine promotion strategies tailored to a patient demographic with a high COVID-19 risk profile.^2,10

This cluster analyses demonstrated the importance of thoughts and feelings about COVID-19 infection and vaccination as influential social and behavioral drivers of vaccine uptake. These opinions help explain the strong association between cluster membership and vaccination status in this multivariable modeling. The cluster composition was consistent with findings from studies of nonveteran populations that identified perceived vulnerability to COVID-19 infection, beliefs in vaccine effectiveness, and adherence with protective behaviors during the pandemic as contributors to vaccine uptake.^13,33 Qualitative themes showed that personal protection, protecting others, and vaccine mandates were frequent motivators for vaccination. Whereas protection of self and others from COVID-19 infection were more often expressed by the highly vaccinated Concerned Ambivalents, employment and travel vaccine mandates were more often identified by Unconcerned Disbelievers, who had a lower vaccination rate. Among Unconcerned Disbelievers, an employer vaccine requirement was the most frequent qualitative theme for overcoming vaccination concerns.

In addition to cluster membership, our modeling showed that trust in local VA HCPs and the CDC were independently associated with COVID-19 vaccination, which has been found in prior research.²⁰ This qualitative analyses regarding vaccine hesitancy identified trust-related concerns that were more frequently expressed by Unconcerned Disbelievers than Concerned Ambivalents. Concerns included the rapid development of the vaccines potentially limiting the generation of scientifically sound effectiveness and safety data, and potential biases involving the entities promoting vaccine uptake.

Whereas the Concerned Believers, Unconcerned Believers, and Concerned Ambivalents all had high COVID-19 vaccination rates (≥ 93%), the decision-making pathways to vaccine uptake likely differ by their concerns about COVID-19 infection and perceptions of vaccine safety and effectiveness. For example, this mixed-methods analysis consistently showed that people in the Concerned Ambivalent cluster were positively motivated by concerns about COVID-19 infection and severity and beliefs about vaccine effectiveness that were tempered by concerns about vaccine AEs. For this cluster, their frequent thematic expression that the benefits of the vaccine exceed the risks, and the positive social influences of family, friends, and HCPs may explain their high vaccination rate.

Such insights into how the patterns of COVID-19–related thoughts and feelings vary across clusters can be used to design interventions to encourage initial and booster doses of COVID-19 vaccines. For example, messaging that highlights the infectivity and severity of COVID-19 and the potential for persistent negative health outcomes associated with long COVID could reinforce the beliefs of Concerned Believers and Concerned Ambivalents, and such messaging could also be used as a targeted intervention for Unconcerned Believers who expressed fewer concerns about the health consequences of COVID-19.23 Likewise, messaging about the safety profile of COVID-19 vaccines may reduce vaccine hesitancy for Concerned Ambivalents. Importantly, purposeful attention to health equity, community engagement, and involvement of racially diverse HCPs in patient discussions represent successful strategies to increase COVID-19 vaccine uptake among Black individuals, who were disproportionately represented in the Concerned Ambivalent cluster and may possess higher levels of mistrust due to racism experienced within the health care system.²⁴

Our findings suggest that the greatest challenge for overcoming vaccine hesitancy is for individuals in the suboptimally vaccinated (30%) Unconcerned Disbeliever cluster. These individuals had low levels of concern about COVID-19 infection and severity, high levels of concern about vaccine safety, low perceived vaccine effectiveness, and low levels of trust in all information sources about COVID-19. While the Unconcerned Disbelievers cited scientifically reputable data sources, we were unable to verify whether participants accessed these reputable sources of information directly or obtained such information indirectly through potentially biased online sources. Nearly half of this cluster trusted their VA HCP and believed their community or religious leaders would want them to get vaccinated. This qualitative analyses found that Unconcerned Disbelievers relied on personal beliefs for vaccine decision-making more than Concerned Ambivalents. While Unconcerned Disbelievers were less likely to be socially influenced by family, friends, or religious leaders, they still acknowledged some impact from these sources. These findings suggest that addressing vaccine hesitancy among Unconcerned Disbelievers may require a multifaceted approach that respects their reliance on personal research while also leveraging the potential social influences. This approach supports the promising, previously reported practices of harnessing the social influences of HCPs and other community and religious leaders to promote vaccine uptake among Unconcerned Disbelievers.^34,35 One evidence-based approach to effectively change patient health care behaviors is through motivational interviewing strategies that use open-ended questions, nonjudgmental interactions, and collaborative decision-making when discussing the risks and benefits of vaccination.^21,22

Limitations

This study was conducted at a single VA health care facility and our sampling technique was nonrandom, suggesting that these results may not be generalizable to all veterans or non-VA patient populations. The 21% questionnaire response rate could have introduced selection bias into the respondent sample. All questionnaire data were self-reported, including vaccination status. Finally, the qualitative interviews consisted of a small number of unvaccinated individuals in 2 clusters (ie, Concerned Ambivalents and Unconcerned Disbelievers) and may not have reached thematic saturation in these subgroups.

Conclusions

Quantitative analyses identified 4 clusters based on individual thoughts and feelings about COVID-19 infection and vaccines. Cluster membership and levels of trust in COVID-19 information sources were independently associated with vaccination. Understanding the quantitative patterns of thoughts and beliefs across clusters, enriched by common qualitative themes for vaccine hesitancy, help inform tailored interventions to augment COVID-19 vaccine uptake and highlight the importance of targeted, trust-based communication and culturally sensitive interventions to enhance vaccine uptake across diverse populations.

The COVID-19 pandemic changed the education and training experiences of health care students and those set to comprise the future workforce. Apart from general training disruptions or delays due to the pandemic, behavioral health trainees such as psychologists and social workers faced limited opportunities to provide in-person services.^1-5 Trainees also experienced fewer referrals to mental health services from primary care and more disrupted, no-show, or cancelled appointments.^4-6 Behavioral health trainees experienced a limited ability to establish rapport and more difficulty providing effective services because of the limited in-person interaction presented by telehealth.⁶ The pandemic also resulted in feelings of increased isolation and decreased teamwork.^1,7 The virtual or remote setting made it more difficult for trainees to feel as if they were a member of a team or community of behavioral health professionals.^1,7

Behavioral health trainees had to adapt to conducting patient visits and educational didactics through virtual platforms.^1,3-7 Challenges included access or technological problems with online platforms and a lack of telehealth training use.^3,4,6 One study found that while both behavioral health trainees and licensed practitioners reported similar rates of telehealth use for mental health services by early April 2020, trainees had more difficulties implementing telehealth compared with licensed practitioners. This study found that US Department of Veteran Affairs (VA) facilities reported higher use of telehealth in February 2020.⁵

A mission of the VA is to provide education and training to health care professionals through partnerships with affiliated academic institutions. The VA is the largest education and training supplier for health care professions in the US. As many as 50% of psychologists in the US received some training at the VA.⁸ Additionally, more graduate-level social work students are trained at the VA than at any other organization.⁹ The VA is a major contributor to not only its own behavioral health workforce, but that of the entire country.

The VA is also the largest employer of psychologists and social workers in the US.^10,11 The VA Office of Academic Affiliations (OAA) oversees health care profession education and training at all VA facilities. In 2012, OAA began the Mental Health Education Expansion program to increase training for behavioral health professionals, including psychologists and social workers. ¹² The OAA initiative was aligned with VA training and workforce priorities.^8,12 To gauge the effectiveness of VA education and training, OAA encourages VA trainees to complete the Trainee Satisfaction Survey (TSS), which measures trainee satisfaction and the likelihood of a trainee to consider the VA for future employment.

Researchers at the Veterans Emergency Management Evaluation Center sought to understand the impact the COVID-19 pandemic had on behavioral health trainees’ experiences by examining TSS data from before and after the onset of the pandemic. This study expands on prior research among physician residents and fellows which found associations between VA training experiences and the COVID- 19 pandemic. The previous study found declines in trainee satisfaction and a decreased likelihood to consider the VA for future employment.¹³

It is important to understand the effects the pandemic had on the professional development and wellness for both physician and behavioral health professional trainees. Identifying how the pandemic impacted trainee satisfaction may help improve education programs and mitigate the impact of future public health emergencies. This is particularly important due to the shortage of behavioral health professionals in the VA and the US.^12,14

METHODS

This study used TSS data collected from August 2018 to July 2021 from 153 VA facilities. A behavioral health trainee was defined as any psychology or social work trainee who completed 1 rotation at a VA facility. Psychiatric trainees were excluded because as physicians their training programs differ markedly from those for psychology and social work. Excluding psychiatry, psychology and social work comprise the 2 largest mental health care training groups.

This study was reviewed and approved as a quality improvement project by the VA Greater Los Angeles Healthcare System (VAGLAHS) Institutional Review Board, which waived informed consent requirements. The OAA granted access to data using a process open to all VA researchers. At the time of data collection, respondents were assured their anonymity; participation was voluntary.

Measures

Any response provided before February 29, 2020, was defined as the prepandemic period. The pandemic period included any response from April 1, 2020, to July 31, 2021. Responses collected in March 2020 were excluded as it would be unclear from the survey whether the training period occurred before or after the onset of the pandemic.

To measure overall trainee satisfaction with the VA training experience, responses were grouped as satisfied (satisfied/ very satisfied) and dissatisfied (dissatisfied/ very dissatisfied). To measure a trainee’s likelihood to consider the VA for future employment as a result of their training experience, responses were grouped as likely (likely/very likely) and unlikely (unlikely/very unlikely).

Other components of satisfaction were also collected including onboarding, clinical faculty/preceptors, clinical learning environment, physical environment, working environment, and respect felt at work. If a respondent chose very dissatisfied or dissatisfied, they were subsequently asked to specify the reason for their dissatisfaction with an open-ended response. Open-ended responses were not permitted if a respondent indicated a satisfied or very satisfied response.

Statistical Analyses

Stata SE 17 was used for statistical analyses. To test the relationship between the pandemic group and the 2 separate outcome variables, logistic regressions were conducted to measure overall satisfaction and likelihood of future VA employment. Margin commands were used to calculate the difference in the probability of reporting satisfied/very satisfied and likely/very likely for the prepandemic and pandemic groups. The association of the COVID-19 group with each outcome variable was expressed as the difference in the percentage of the outcome between the prepandemic and pandemic groups. Preliminary analyses demonstrated similar effects of the pandemic on psychology and social work trainees; therefore, the groups were combined.

Rapid Coding and Thematic Analyses

Qualitative data were based on open-ended responses from behavioral health trainees when they were asked to specify the cause of dissatisfaction in the aforementioned areas of satisfaction. Methods for qualitative data included rapid coding and thematic content analyses.^15,16 Additional general information regarding the qualitative data analyses is described elsewhere.¹³ A keyword search was completed to identify all open-ended responses related to COVID-19 pandemic causes of dissatisfaction. Keywords included: virus, COVID, corona, pandemic, PPE, N95, mask, social distance, and safety. All open-ended responses were reviewed to ensure keywords were appropriately identifying pandemic-related causes of dissatisfaction and did not overlook other references to the pandemic, and to identify initial themes and corresponding definitions based on survey questions. After review, additional keywords were included in the content analyses that were related to providing mental health services using remote or telehealth options. This included the following keywords: remote, video, VVC (VA Video Connect), and tele. The research team completed a review of the initial themes and definitions and created a final coding construct with definitions before completing an independent coding of all identified pandemic-related responses. Frequency counts of each code were provided to identify which pandemic-related causes of dissatisfaction were mentioned most.

RESULTS

A total of 3950 behavioral health trainees responded to the TSS, including 2715 psychology trainees and 1235 social work trainees who indicated they received training at the VA in academic years 2018/2019, 2019/2020, or 2020/2021. The academic year 2018/2019 was considered in an effort to provide a larger sample of prepandemic trainees.

The percentage of trainees reporting satisfaction with their training decreased across prepandemic to pandemic groups. In the pandemic group, 2166 of 2324 respondents (93.2%) reported satisfaction compared to 1474 of 1555 (94.8%) in the prepandemic trainee group (P = .04; 95% CI, -3.10 to -0.08). There was no association between the pandemic group and behavioral health trainees’ reported willingness to consider the VA for future employment (Table 1). Preliminary analyses demonstrated similar effects of the pandemic on psychology and social work trainees, therefore the groups were combined, and overall effects were reported.

Pandemic-Related Dissatisfaction

Of the 3950 psychology and social work trainees who responded to the survey, 75 (1.9%) indicated dissatisfaction with their VA training experience using pandemic-related keywords. Open-ended responses were generally short (range, 1-32 words; median, 19 words). Qualitative analyses revealed 7 themes (Table 2).

The most frequently identified theme was challenges with onboarding. One respondent indicated the modified onboarding procedures in place due to the pandemic were difficult to understand and resulted in delays. Another frequently mentioned cause of dissatisfaction was limited work or office space and insufficient computer availability. This was often noted to relate to a lack of private space to conduct telehealth visits or computers that were not equipped to provide telehealth. Several respondents also noted technological issues when attempting to use VVC to provide telehealth.

Another common theme was that the pandemic diminished teamwork, generated feelings of isolation, and created unsupportive environments for trainees. For instance, some trainees indicated that COVID-19 decreased the inclusion of trainees as part of the regular staff groups and accordingly resulted in limited networking opportunities. Other causes of dissatisfaction included the pandemic’s impacts on the learning environment, such as decreases in patient volume, decreased diversity of patient cases, and a limited presence of faculty mentors. Several respondents indicated that the pandemic limited their caseloads and indicated that most patients were seen virtually. Open-ended responses from a few respondents indicated their training environments were noncompliant with social distancing, personal protective equipment requirements, or other safety guidelines.

DISCUSSION

This study illustrates the impact of the COVID-19 pandemic on the behavioral health trainee experience, which was expressed through decreased satisfaction with their clinical training at the VA. The narrative data indicated that the observed pandemic-related dissatisfaction was linked specifically to onboarding, a lack of safe and private workspaces and computers, as well as a lack of a supportive work environment.

Although the reported decrease in satisfaction was statistically significant, the effect size was not large. Additionally, while satisfaction did decrease, the trainees’ reported likelihood to consider the VA for future employment was not impacted. This may suggest psychologist and social work trainees’ perseverance and dedication to their chosen profession persisted despite the challenges presented by the pandemic. Furthermore, the qualitative data suggest potential ways to mitigate health care profession trainee challenges that can follow a crisis like the COVID-19 pandemic, although further study is warranted.

While narrative responses with pandemic-related keywords did indicate challenges specific to COVID-19 (ie, limited access to workspaces and/or computers equipped for telehealth), the overall frequency of pandemic-related responses was low. This may indicate these are institutional challenges trainees face independent of the pandemic. These findings warrant longterm attention to the satisfaction of psychology and social worker trainees’ during the pandemic. For example, additional training for the use of telehealth could be provided. One study indicated that < 61% of psychology postdoctoral fellows received telepsychology training during the pandemic, and of those who did receive training, less than half were satisfied with it.³

Similarly, strategies could be developed to ensure a more supportive learning and work environment, and provide additional networking opportunities for trainees, despite social distancing. Education specific to disaster response should be incorporated into behavioral health care professionals’ training, especially because behavioral health care professionals provided major contributions during the pandemic due to reported increases in mental health concerns (eg, anxiety and depression) during the period.^7,17,18 As the pandemic progressed, policies and procedures were established or modified to address some of these concerns because they were not necessarily limited to trainees. For example, additional training resources were developed to support the use of various telehealth technologies, virtual resources were used more often for meetings, and supervisors developed more comfort and familiarity with how to manage in a virtual or hybrid environment.

Limitations

Although the TSS data provide a large national sample of behavioral health care trainees, it only includes VA trainees, and therefore may not be completely generalizable across health care. However, because many psychologists and social workers throughout the US train at the VA, and because the VA is the largest employer of practicing psychologists and social workers, understanding the impacts felt at the VA informs institutions nationally.^8-11 The TSS has limited demographic data (eg, age, race, ethnicity, and sex), so it is unclear whether the respondent groups before and during the pandemic differed in ways that could relate to outcomes. The data also do not specify exact training dates; however, anecdotal evidence suggests respondents generally complete the survey close to the end of their training.

Additionally, open-ended narrative responses were only asked for replies that indicated dissatisfaction, precluding a more nuanced understanding of potential positive outcomes. Furthermore, the TSS is limited to questions about the trainees’ clinical experiences, but because the pandemic created many stressors, there may have been personal issues that affected their work. It is possible that changes in overall satisfaction may have been rooted in something outside of their clinical experience. Finally, the response rate for the TSS is consistently low both before and during the pandemic and includes a limited number of narrative responses.

CONCLUSIONS

The VA is an important contributor to the education, training, and composition of the behavioral health care workforce. A deeper understanding of the VA trainee experience is important to identify how to improve behavioral health care professional education and training. This is especially true as behavioral health care faces shortages within the VA and nationwide.^8,12,19

This study reinforces research that found health care trainees experienced decreased learning opportunities and telehealth-related challenges during the COVID-19 pandemic. ^13,20 Despite the observed decline in trainee satisfaction, the lack of a corresponding change in likelihood to seek employment with the VA is encouraging for VA efforts to maintain and grow its behavioral health care workforce and for similar efforts outside VA. This resilience may relate to the substantial prepandemic time invested in their professional development. Future studies should examine long term impacts of the pandemic on trainee’s clinical experience and whether the pipeline of behavioral health care workers declines over time as students that are earlier in their career paths instead chose other professions. Future research should also explore ways to improve professional development and wellness of behavioral health care trainees during disasters (eg, telehealth training, additional networking, and social support).

The COVID-19 pandemic changed the education and training experiences of health care students and those set to comprise the future workforce. Apart from general training disruptions or delays due to the pandemic, behavioral health trainees such as psychologists and social workers faced limited opportunities to provide in-person services.^1-5 Trainees also experienced fewer referrals to mental health services from primary care and more disrupted, no-show, or cancelled appointments.^4-6 Behavioral health trainees experienced a limited ability to establish rapport and more difficulty providing effective services because of the limited in-person interaction presented by telehealth.⁶ The pandemic also resulted in feelings of increased isolation and decreased teamwork.^1,7 The virtual or remote setting made it more difficult for trainees to feel as if they were a member of a team or community of behavioral health professionals.^1,7

Behavioral health trainees had to adapt to conducting patient visits and educational didactics through virtual platforms.^1,3-7 Challenges included access or technological problems with online platforms and a lack of telehealth training use.^3,4,6 One study found that while both behavioral health trainees and licensed practitioners reported similar rates of telehealth use for mental health services by early April 2020, trainees had more difficulties implementing telehealth compared with licensed practitioners. This study found that US Department of Veteran Affairs (VA) facilities reported higher use of telehealth in February 2020.⁵

A mission of the VA is to provide education and training to health care professionals through partnerships with affiliated academic institutions. The VA is the largest education and training supplier for health care professions in the US. As many as 50% of psychologists in the US received some training at the VA.⁸ Additionally, more graduate-level social work students are trained at the VA than at any other organization.⁹ The VA is a major contributor to not only its own behavioral health workforce, but that of the entire country.

The VA is also the largest employer of psychologists and social workers in the US.^10,11 The VA Office of Academic Affiliations (OAA) oversees health care profession education and training at all VA facilities. In 2012, OAA began the Mental Health Education Expansion program to increase training for behavioral health professionals, including psychologists and social workers. ¹² The OAA initiative was aligned with VA training and workforce priorities.^8,12 To gauge the effectiveness of VA education and training, OAA encourages VA trainees to complete the Trainee Satisfaction Survey (TSS), which measures trainee satisfaction and the likelihood of a trainee to consider the VA for future employment.

Researchers at the Veterans Emergency Management Evaluation Center sought to understand the impact the COVID-19 pandemic had on behavioral health trainees’ experiences by examining TSS data from before and after the onset of the pandemic. This study expands on prior research among physician residents and fellows which found associations between VA training experiences and the COVID- 19 pandemic. The previous study found declines in trainee satisfaction and a decreased likelihood to consider the VA for future employment.¹³

It is important to understand the effects the pandemic had on the professional development and wellness for both physician and behavioral health professional trainees. Identifying how the pandemic impacted trainee satisfaction may help improve education programs and mitigate the impact of future public health emergencies. This is particularly important due to the shortage of behavioral health professionals in the VA and the US.^12,14

METHODS

This study used TSS data collected from August 2018 to July 2021 from 153 VA facilities. A behavioral health trainee was defined as any psychology or social work trainee who completed 1 rotation at a VA facility. Psychiatric trainees were excluded because as physicians their training programs differ markedly from those for psychology and social work. Excluding psychiatry, psychology and social work comprise the 2 largest mental health care training groups.

This study was reviewed and approved as a quality improvement project by the VA Greater Los Angeles Healthcare System (VAGLAHS) Institutional Review Board, which waived informed consent requirements. The OAA granted access to data using a process open to all VA researchers. At the time of data collection, respondents were assured their anonymity; participation was voluntary.

Measures

Any response provided before February 29, 2020, was defined as the prepandemic period. The pandemic period included any response from April 1, 2020, to July 31, 2021. Responses collected in March 2020 were excluded as it would be unclear from the survey whether the training period occurred before or after the onset of the pandemic.

To measure overall trainee satisfaction with the VA training experience, responses were grouped as satisfied (satisfied/ very satisfied) and dissatisfied (dissatisfied/ very dissatisfied). To measure a trainee’s likelihood to consider the VA for future employment as a result of their training experience, responses were grouped as likely (likely/very likely) and unlikely (unlikely/very unlikely).

Other components of satisfaction were also collected including onboarding, clinical faculty/preceptors, clinical learning environment, physical environment, working environment, and respect felt at work. If a respondent chose very dissatisfied or dissatisfied, they were subsequently asked to specify the reason for their dissatisfaction with an open-ended response. Open-ended responses were not permitted if a respondent indicated a satisfied or very satisfied response.

Statistical Analyses

Stata SE 17 was used for statistical analyses. To test the relationship between the pandemic group and the 2 separate outcome variables, logistic regressions were conducted to measure overall satisfaction and likelihood of future VA employment. Margin commands were used to calculate the difference in the probability of reporting satisfied/very satisfied and likely/very likely for the prepandemic and pandemic groups. The association of the COVID-19 group with each outcome variable was expressed as the difference in the percentage of the outcome between the prepandemic and pandemic groups. Preliminary analyses demonstrated similar effects of the pandemic on psychology and social work trainees; therefore, the groups were combined.

Rapid Coding and Thematic Analyses

Qualitative data were based on open-ended responses from behavioral health trainees when they were asked to specify the cause of dissatisfaction in the aforementioned areas of satisfaction. Methods for qualitative data included rapid coding and thematic content analyses.^15,16 Additional general information regarding the qualitative data analyses is described elsewhere.¹³ A keyword search was completed to identify all open-ended responses related to COVID-19 pandemic causes of dissatisfaction. Keywords included: virus, COVID, corona, pandemic, PPE, N95, mask, social distance, and safety. All open-ended responses were reviewed to ensure keywords were appropriately identifying pandemic-related causes of dissatisfaction and did not overlook other references to the pandemic, and to identify initial themes and corresponding definitions based on survey questions. After review, additional keywords were included in the content analyses that were related to providing mental health services using remote or telehealth options. This included the following keywords: remote, video, VVC (VA Video Connect), and tele. The research team completed a review of the initial themes and definitions and created a final coding construct with definitions before completing an independent coding of all identified pandemic-related responses. Frequency counts of each code were provided to identify which pandemic-related causes of dissatisfaction were mentioned most.

RESULTS

A total of 3950 behavioral health trainees responded to the TSS, including 2715 psychology trainees and 1235 social work trainees who indicated they received training at the VA in academic years 2018/2019, 2019/2020, or 2020/2021. The academic year 2018/2019 was considered in an effort to provide a larger sample of prepandemic trainees.

The percentage of trainees reporting satisfaction with their training decreased across prepandemic to pandemic groups. In the pandemic group, 2166 of 2324 respondents (93.2%) reported satisfaction compared to 1474 of 1555 (94.8%) in the prepandemic trainee group (P = .04; 95% CI, -3.10 to -0.08). There was no association between the pandemic group and behavioral health trainees’ reported willingness to consider the VA for future employment (Table 1). Preliminary analyses demonstrated similar effects of the pandemic on psychology and social work trainees, therefore the groups were combined, and overall effects were reported.

Pandemic-Related Dissatisfaction

Of the 3950 psychology and social work trainees who responded to the survey, 75 (1.9%) indicated dissatisfaction with their VA training experience using pandemic-related keywords. Open-ended responses were generally short (range, 1-32 words; median, 19 words). Qualitative analyses revealed 7 themes (Table 2).

The most frequently identified theme was challenges with onboarding. One respondent indicated the modified onboarding procedures in place due to the pandemic were difficult to understand and resulted in delays. Another frequently mentioned cause of dissatisfaction was limited work or office space and insufficient computer availability. This was often noted to relate to a lack of private space to conduct telehealth visits or computers that were not equipped to provide telehealth. Several respondents also noted technological issues when attempting to use VVC to provide telehealth.

Another common theme was that the pandemic diminished teamwork, generated feelings of isolation, and created unsupportive environments for trainees. For instance, some trainees indicated that COVID-19 decreased the inclusion of trainees as part of the regular staff groups and accordingly resulted in limited networking opportunities. Other causes of dissatisfaction included the pandemic’s impacts on the learning environment, such as decreases in patient volume, decreased diversity of patient cases, and a limited presence of faculty mentors. Several respondents indicated that the pandemic limited their caseloads and indicated that most patients were seen virtually. Open-ended responses from a few respondents indicated their training environments were noncompliant with social distancing, personal protective equipment requirements, or other safety guidelines.

DISCUSSION

This study illustrates the impact of the COVID-19 pandemic on the behavioral health trainee experience, which was expressed through decreased satisfaction with their clinical training at the VA. The narrative data indicated that the observed pandemic-related dissatisfaction was linked specifically to onboarding, a lack of safe and private workspaces and computers, as well as a lack of a supportive work environment.

Although the reported decrease in satisfaction was statistically significant, the effect size was not large. Additionally, while satisfaction did decrease, the trainees’ reported likelihood to consider the VA for future employment was not impacted. This may suggest psychologist and social work trainees’ perseverance and dedication to their chosen profession persisted despite the challenges presented by the pandemic. Furthermore, the qualitative data suggest potential ways to mitigate health care profession trainee challenges that can follow a crisis like the COVID-19 pandemic, although further study is warranted.

While narrative responses with pandemic-related keywords did indicate challenges specific to COVID-19 (ie, limited access to workspaces and/or computers equipped for telehealth), the overall frequency of pandemic-related responses was low. This may indicate these are institutional challenges trainees face independent of the pandemic. These findings warrant longterm attention to the satisfaction of psychology and social worker trainees’ during the pandemic. For example, additional training for the use of telehealth could be provided. One study indicated that < 61% of psychology postdoctoral fellows received telepsychology training during the pandemic, and of those who did receive training, less than half were satisfied with it.³

Similarly, strategies could be developed to ensure a more supportive learning and work environment, and provide additional networking opportunities for trainees, despite social distancing. Education specific to disaster response should be incorporated into behavioral health care professionals’ training, especially because behavioral health care professionals provided major contributions during the pandemic due to reported increases in mental health concerns (eg, anxiety and depression) during the period.^7,17,18 As the pandemic progressed, policies and procedures were established or modified to address some of these concerns because they were not necessarily limited to trainees. For example, additional training resources were developed to support the use of various telehealth technologies, virtual resources were used more often for meetings, and supervisors developed more comfort and familiarity with how to manage in a virtual or hybrid environment.

Limitations

Although the TSS data provide a large national sample of behavioral health care trainees, it only includes VA trainees, and therefore may not be completely generalizable across health care. However, because many psychologists and social workers throughout the US train at the VA, and because the VA is the largest employer of practicing psychologists and social workers, understanding the impacts felt at the VA informs institutions nationally.^8-11 The TSS has limited demographic data (eg, age, race, ethnicity, and sex), so it is unclear whether the respondent groups before and during the pandemic differed in ways that could relate to outcomes. The data also do not specify exact training dates; however, anecdotal evidence suggests respondents generally complete the survey close to the end of their training.

Additionally, open-ended narrative responses were only asked for replies that indicated dissatisfaction, precluding a more nuanced understanding of potential positive outcomes. Furthermore, the TSS is limited to questions about the trainees’ clinical experiences, but because the pandemic created many stressors, there may have been personal issues that affected their work. It is possible that changes in overall satisfaction may have been rooted in something outside of their clinical experience. Finally, the response rate for the TSS is consistently low both before and during the pandemic and includes a limited number of narrative responses.

CONCLUSIONS

The VA is an important contributor to the education, training, and composition of the behavioral health care workforce. A deeper understanding of the VA trainee experience is important to identify how to improve behavioral health care professional education and training. This is especially true as behavioral health care faces shortages within the VA and nationwide.^8,12,19

This study reinforces research that found health care trainees experienced decreased learning opportunities and telehealth-related challenges during the COVID-19 pandemic. ^13,20 Despite the observed decline in trainee satisfaction, the lack of a corresponding change in likelihood to seek employment with the VA is encouraging for VA efforts to maintain and grow its behavioral health care workforce and for similar efforts outside VA. This resilience may relate to the substantial prepandemic time invested in their professional development. Future studies should examine long term impacts of the pandemic on trainee’s clinical experience and whether the pipeline of behavioral health care workers declines over time as students that are earlier in their career paths instead chose other professions. Future research should also explore ways to improve professional development and wellness of behavioral health care trainees during disasters (eg, telehealth training, additional networking, and social support).

The COVID-19 pandemic changed the education and training experiences of health care students and those set to comprise the future workforce. Apart from general training disruptions or delays due to the pandemic, behavioral health trainees such as psychologists and social workers faced limited opportunities to provide in-person services.^1-5 Trainees also experienced fewer referrals to mental health services from primary care and more disrupted, no-show, or cancelled appointments.^4-6 Behavioral health trainees experienced a limited ability to establish rapport and more difficulty providing effective services because of the limited in-person interaction presented by telehealth.⁶ The pandemic also resulted in feelings of increased isolation and decreased teamwork.^1,7 The virtual or remote setting made it more difficult for trainees to feel as if they were a member of a team or community of behavioral health professionals.^1,7

Behavioral health trainees had to adapt to conducting patient visits and educational didactics through virtual platforms.^1,3-7 Challenges included access or technological problems with online platforms and a lack of telehealth training use.^3,4,6 One study found that while both behavioral health trainees and licensed practitioners reported similar rates of telehealth use for mental health services by early April 2020, trainees had more difficulties implementing telehealth compared with licensed practitioners. This study found that US Department of Veteran Affairs (VA) facilities reported higher use of telehealth in February 2020.⁵

A mission of the VA is to provide education and training to health care professionals through partnerships with affiliated academic institutions. The VA is the largest education and training supplier for health care professions in the US. As many as 50% of psychologists in the US received some training at the VA.⁸ Additionally, more graduate-level social work students are trained at the VA than at any other organization.⁹ The VA is a major contributor to not only its own behavioral health workforce, but that of the entire country.

The VA is also the largest employer of psychologists and social workers in the US.^10,11 The VA Office of Academic Affiliations (OAA) oversees health care profession education and training at all VA facilities. In 2012, OAA began the Mental Health Education Expansion program to increase training for behavioral health professionals, including psychologists and social workers. ¹² The OAA initiative was aligned with VA training and workforce priorities.^8,12 To gauge the effectiveness of VA education and training, OAA encourages VA trainees to complete the Trainee Satisfaction Survey (TSS), which measures trainee satisfaction and the likelihood of a trainee to consider the VA for future employment.

Researchers at the Veterans Emergency Management Evaluation Center sought to understand the impact the COVID-19 pandemic had on behavioral health trainees’ experiences by examining TSS data from before and after the onset of the pandemic. This study expands on prior research among physician residents and fellows which found associations between VA training experiences and the COVID- 19 pandemic. The previous study found declines in trainee satisfaction and a decreased likelihood to consider the VA for future employment.¹³

It is important to understand the effects the pandemic had on the professional development and wellness for both physician and behavioral health professional trainees. Identifying how the pandemic impacted trainee satisfaction may help improve education programs and mitigate the impact of future public health emergencies. This is particularly important due to the shortage of behavioral health professionals in the VA and the US.^12,14

METHODS

This study used TSS data collected from August 2018 to July 2021 from 153 VA facilities. A behavioral health trainee was defined as any psychology or social work trainee who completed 1 rotation at a VA facility. Psychiatric trainees were excluded because as physicians their training programs differ markedly from those for psychology and social work. Excluding psychiatry, psychology and social work comprise the 2 largest mental health care training groups.

This study was reviewed and approved as a quality improvement project by the VA Greater Los Angeles Healthcare System (VAGLAHS) Institutional Review Board, which waived informed consent requirements. The OAA granted access to data using a process open to all VA researchers. At the time of data collection, respondents were assured their anonymity; participation was voluntary.

Measures

Any response provided before February 29, 2020, was defined as the prepandemic period. The pandemic period included any response from April 1, 2020, to July 31, 2021. Responses collected in March 2020 were excluded as it would be unclear from the survey whether the training period occurred before or after the onset of the pandemic.

To measure overall trainee satisfaction with the VA training experience, responses were grouped as satisfied (satisfied/ very satisfied) and dissatisfied (dissatisfied/ very dissatisfied). To measure a trainee’s likelihood to consider the VA for future employment as a result of their training experience, responses were grouped as likely (likely/very likely) and unlikely (unlikely/very unlikely).

Other components of satisfaction were also collected including onboarding, clinical faculty/preceptors, clinical learning environment, physical environment, working environment, and respect felt at work. If a respondent chose very dissatisfied or dissatisfied, they were subsequently asked to specify the reason for their dissatisfaction with an open-ended response. Open-ended responses were not permitted if a respondent indicated a satisfied or very satisfied response.

Statistical Analyses

Stata SE 17 was used for statistical analyses. To test the relationship between the pandemic group and the 2 separate outcome variables, logistic regressions were conducted to measure overall satisfaction and likelihood of future VA employment. Margin commands were used to calculate the difference in the probability of reporting satisfied/very satisfied and likely/very likely for the prepandemic and pandemic groups. The association of the COVID-19 group with each outcome variable was expressed as the difference in the percentage of the outcome between the prepandemic and pandemic groups. Preliminary analyses demonstrated similar effects of the pandemic on psychology and social work trainees; therefore, the groups were combined.

Rapid Coding and Thematic Analyses

Qualitative data were based on open-ended responses from behavioral health trainees when they were asked to specify the cause of dissatisfaction in the aforementioned areas of satisfaction. Methods for qualitative data included rapid coding and thematic content analyses.^15,16 Additional general information regarding the qualitative data analyses is described elsewhere.¹³ A keyword search was completed to identify all open-ended responses related to COVID-19 pandemic causes of dissatisfaction. Keywords included: virus, COVID, corona, pandemic, PPE, N95, mask, social distance, and safety. All open-ended responses were reviewed to ensure keywords were appropriately identifying pandemic-related causes of dissatisfaction and did not overlook other references to the pandemic, and to identify initial themes and corresponding definitions based on survey questions. After review, additional keywords were included in the content analyses that were related to providing mental health services using remote or telehealth options. This included the following keywords: remote, video, VVC (VA Video Connect), and tele. The research team completed a review of the initial themes and definitions and created a final coding construct with definitions before completing an independent coding of all identified pandemic-related responses. Frequency counts of each code were provided to identify which pandemic-related causes of dissatisfaction were mentioned most.

RESULTS

A total of 3950 behavioral health trainees responded to the TSS, including 2715 psychology trainees and 1235 social work trainees who indicated they received training at the VA in academic years 2018/2019, 2019/2020, or 2020/2021. The academic year 2018/2019 was considered in an effort to provide a larger sample of prepandemic trainees.

The percentage of trainees reporting satisfaction with their training decreased across prepandemic to pandemic groups. In the pandemic group, 2166 of 2324 respondents (93.2%) reported satisfaction compared to 1474 of 1555 (94.8%) in the prepandemic trainee group (P = .04; 95% CI, -3.10 to -0.08). There was no association between the pandemic group and behavioral health trainees’ reported willingness to consider the VA for future employment (Table 1). Preliminary analyses demonstrated similar effects of the pandemic on psychology and social work trainees, therefore the groups were combined, and overall effects were reported.

Pandemic-Related Dissatisfaction

Of the 3950 psychology and social work trainees who responded to the survey, 75 (1.9%) indicated dissatisfaction with their VA training experience using pandemic-related keywords. Open-ended responses were generally short (range, 1-32 words; median, 19 words). Qualitative analyses revealed 7 themes (Table 2).

The most frequently identified theme was challenges with onboarding. One respondent indicated the modified onboarding procedures in place due to the pandemic were difficult to understand and resulted in delays. Another frequently mentioned cause of dissatisfaction was limited work or office space and insufficient computer availability. This was often noted to relate to a lack of private space to conduct telehealth visits or computers that were not equipped to provide telehealth. Several respondents also noted technological issues when attempting to use VVC to provide telehealth.

Another common theme was that the pandemic diminished teamwork, generated feelings of isolation, and created unsupportive environments for trainees. For instance, some trainees indicated that COVID-19 decreased the inclusion of trainees as part of the regular staff groups and accordingly resulted in limited networking opportunities. Other causes of dissatisfaction included the pandemic’s impacts on the learning environment, such as decreases in patient volume, decreased diversity of patient cases, and a limited presence of faculty mentors. Several respondents indicated that the pandemic limited their caseloads and indicated that most patients were seen virtually. Open-ended responses from a few respondents indicated their training environments were noncompliant with social distancing, personal protective equipment requirements, or other safety guidelines.

DISCUSSION

This study illustrates the impact of the COVID-19 pandemic on the behavioral health trainee experience, which was expressed through decreased satisfaction with their clinical training at the VA. The narrative data indicated that the observed pandemic-related dissatisfaction was linked specifically to onboarding, a lack of safe and private workspaces and computers, as well as a lack of a supportive work environment.

Although the reported decrease in satisfaction was statistically significant, the effect size was not large. Additionally, while satisfaction did decrease, the trainees’ reported likelihood to consider the VA for future employment was not impacted. This may suggest psychologist and social work trainees’ perseverance and dedication to their chosen profession persisted despite the challenges presented by the pandemic. Furthermore, the qualitative data suggest potential ways to mitigate health care profession trainee challenges that can follow a crisis like the COVID-19 pandemic, although further study is warranted.

While narrative responses with pandemic-related keywords did indicate challenges specific to COVID-19 (ie, limited access to workspaces and/or computers equipped for telehealth), the overall frequency of pandemic-related responses was low. This may indicate these are institutional challenges trainees face independent of the pandemic. These findings warrant longterm attention to the satisfaction of psychology and social worker trainees’ during the pandemic. For example, additional training for the use of telehealth could be provided. One study indicated that < 61% of psychology postdoctoral fellows received telepsychology training during the pandemic, and of those who did receive training, less than half were satisfied with it.³

Similarly, strategies could be developed to ensure a more supportive learning and work environment, and provide additional networking opportunities for trainees, despite social distancing. Education specific to disaster response should be incorporated into behavioral health care professionals’ training, especially because behavioral health care professionals provided major contributions during the pandemic due to reported increases in mental health concerns (eg, anxiety and depression) during the period.^7,17,18 As the pandemic progressed, policies and procedures were established or modified to address some of these concerns because they were not necessarily limited to trainees. For example, additional training resources were developed to support the use of various telehealth technologies, virtual resources were used more often for meetings, and supervisors developed more comfort and familiarity with how to manage in a virtual or hybrid environment.

Limitations

Although the TSS data provide a large national sample of behavioral health care trainees, it only includes VA trainees, and therefore may not be completely generalizable across health care. However, because many psychologists and social workers throughout the US train at the VA, and because the VA is the largest employer of practicing psychologists and social workers, understanding the impacts felt at the VA informs institutions nationally.^8-11 The TSS has limited demographic data (eg, age, race, ethnicity, and sex), so it is unclear whether the respondent groups before and during the pandemic differed in ways that could relate to outcomes. The data also do not specify exact training dates; however, anecdotal evidence suggests respondents generally complete the survey close to the end of their training.

Additionally, open-ended narrative responses were only asked for replies that indicated dissatisfaction, precluding a more nuanced understanding of potential positive outcomes. Furthermore, the TSS is limited to questions about the trainees’ clinical experiences, but because the pandemic created many stressors, there may have been personal issues that affected their work. It is possible that changes in overall satisfaction may have been rooted in something outside of their clinical experience. Finally, the response rate for the TSS is consistently low both before and during the pandemic and includes a limited number of narrative responses.

CONCLUSIONS

The VA is an important contributor to the education, training, and composition of the behavioral health care workforce. A deeper understanding of the VA trainee experience is important to identify how to improve behavioral health care professional education and training. This is especially true as behavioral health care faces shortages within the VA and nationwide.^8,12,19

This study reinforces research that found health care trainees experienced decreased learning opportunities and telehealth-related challenges during the COVID-19 pandemic. ^13,20 Despite the observed decline in trainee satisfaction, the lack of a corresponding change in likelihood to seek employment with the VA is encouraging for VA efforts to maintain and grow its behavioral health care workforce and for similar efforts outside VA. This resilience may relate to the substantial prepandemic time invested in their professional development. Future studies should examine long term impacts of the pandemic on trainee’s clinical experience and whether the pipeline of behavioral health care workers declines over time as students that are earlier in their career paths instead chose other professions. Future research should also explore ways to improve professional development and wellness of behavioral health care trainees during disasters (eg, telehealth training, additional networking, and social support).

People with vasculitis may need at least three or four vaccinations for COVID-19 before they start to show an immune response against SARS-CoV-2 infection, new research has suggested.

In a longitudinal retrospective study, serum antibody neutralization against the Omicron variant of the virus and its descendants was found to be “largely absent” after the first two doses of COVID-19 vaccine had been given to patients. But increasing neutralizing antibody titers were seen after both the third and fourth vaccine boosters had been administered.

Results also showed that the more recently people had been treated with the B cell–depleting therapy rituximab, the lower the levels of immunogenicity that were achieved, and thus protection against SARS-CoV-2.

“Our results have significant implications for individuals treated with rituximab in the post-Omicron era, highlighting the value of additive boosters in affirming increasing protection in clinically vulnerable populations,” the team behind the work at the University of Cambridge in England, has reported in Science Advances.

Moreover, because the use of rituximab reduced the neutralization of not just wild-type (WT) Omicron but also the Omicron-descendant variants BA.1, BA.2, BA.4, and XBB, this highlights “the urgent need for additional adjunctive strategies to enhance vaccine-induced immunity as well as preferential access for such patients to updated vaccines using spike from now circulating Omicron lineages,” the team added.

 

Studying Humoral Responses to SARS-CoV-2 Vaccines

Corresponding author Ravindra K. Gupta, BMBCh, MA, MPH, PhD, told this news organization that studying humoral responses to SARS-CoV-2 vaccines in immunocompromised individuals such as those with vasculitis was important for two main reasons.

“It is really important at individual level for their own health, of course, but also because we know that variants of concern have often evolved and developed within patients and can then spread in wider populations,” he said.

Gupta, who is professor of clinical microbiology at the Cambridge Institute for Therapeutic Immunology & Infectious Disease added: “We believe that the variants of concern that we’re having to deal with right now, including Omicron, have come from such [immunocompromised] individuals.”

Omicron “was a big shift,” Gupta noted. “It had a lot of new mutations on it, so it was almost like a new strain of the virus.” Few studies have looked at the longitudinal immunogenicity proffered by COVID vaccines in the post-Omicron era, particularly in those with vasculitis who are often treated with immunosuppressive drugs, including rituximab.

 

Two-Pronged Study Approach

For the study, a population of immunocompromised individuals diagnosed with vasculitis who had been treated with rituximab in the past 5 years was identified. Just over half (58%) had received adenovirus-based AZD1222/ChAdOx1 nCoV-19 (AstraZeneca-Oxford; AZN) and 37% BNT162b2 (Pfizer-BioNTech; mRNA) as their primary vaccines. Patients with antineutrophil cytoplasmic antibody–associated vasculitis comprised the majority of those who received rituximab (83%), compared with less than half of those who did not take rituximab (48%).

A two-pronged approach was taken with the researchers first measuring neutralizing antibody titers before and 30 days after four successive COVID vaccinations in a group of 32 individuals with available samples. They then performed a cross-sectional, case-control study in 95 individuals to look at neutralizing antibody titers and antibody-dependent cell-mediated cytotoxicity (ADCC) in individuals who had (n = 64) and had not (n = 31) been treated with rituximab in the past 5 years and had samples available after their third and fourth COVID vaccinations.

The first analysis was done to see how people were responding to vaccination over time. “That told us that there was a problem with the first two doses and that we got some response after doses three and four, but the response was uniformly quite poor against the new variants of concern,” Gupta said.

A human embryonic kidney cell model had been used to determine individuals’ neutralizing antibody titers in response to WT, BA.1, BA.2, BA.4, and XBB pseudotyped viruses. After the first and second COVID vaccinations, the geometric mean titer (GMT) against each variant barely increased from a baseline of 40.0. The greatest increases in GMT was seen with the WT virus, at 43.7, 90.7, 256.3, and 394.2, after the first, second, third and fourth doses, respectively. The lowest increases in GMT were seen with the XBB variant, with respective values of 40.0, 40.8, 45.7, and 53.9.

 

Incremental Benefit Offers Some ‘Reassurance’

Vasculitis specialist Rona Smith, MA, MB BChir, MD, who was one of the authors of the paper, told this news organization separately that the results showed there was “an incremental benefit of having COVID vaccinations,” which “offers a little bit of reassurance” that there can be an immune response in people with vasculitis.

Although results of the cross-sectional study showed that there was a significant dampening effect of rituximab treatment on the immune response, “I don’t think it’s an isolated effect in our [vasculitis] patients,” Smith suggested, adding the results were “probably still relevant to patients who receive routine dosing of rituximab for other conditions.”

Neutralizing antibody titers were consistently lower among individuals who had been treated with rituximab vs those who had not, with treatment in the past 18 months found to significantly impair immunogenicity.

The ADCC response was better preserved than the neutralizing antibody response, Gupta said, although it was still significantly lower in the rituximab-treated than in the non–rituximab-treated patients.

 

When to Vaccinate in Vasculitis?

Regarding when to give vaccines to people with vasculitis, Smith said: “Current recommendations are that patients should receive any vaccines that they’re offered routinely, whether that be COVID vaccines, flu vaccines, pneumococcal vaccines.”

As for the timing of those vaccinations, she observed that the current thinking was that vaccinations should “ideally be at least 1 month before a rituximab treatment, and ideally 3-4 [months] after their last dose. However, as many patients are on a 6-month dosing cycle, it can be difficult for some of them to find a suitable time window to have the COVID vaccine when it is offered.”

Additional precautions, such as wearing masks in crowded places and avoiding visits to acutely unwell friends or relatives, may still be prudent, Smith acknowledged, but he was clear that people should not be locking themselves away as they did during the COVID-19 pandemic.

When advising patients, “our general recommendation is that it is better to have a vaccine than not, but we can’t guarantee how well you will respond to it, but some response is better than none,” Smith said.

The study was independently supported. Gupta had no relevant financial relationships to disclose. Smith was a coauthor of the paper and has received research grant funding from Union Therapeutics, GlaxoSmithKline/Vir Biotechnology, Addenbrooke’s Charitable Trust, and Vasculitis UK. Another coauthor reported receiving research grants from CSL Vifor, Roche, and GlaxoSmithKline and advisory board, consultancy, and lecture fees from Roche and CSL Vifor.

A version of this article first appeared on Medscape.com.

People with vasculitis may need at least three or four vaccinations for COVID-19 before they start to show an immune response against SARS-CoV-2 infection, new research has suggested.

In a longitudinal retrospective study, serum antibody neutralization against the Omicron variant of the virus and its descendants was found to be “largely absent” after the first two doses of COVID-19 vaccine had been given to patients. But increasing neutralizing antibody titers were seen after both the third and fourth vaccine boosters had been administered.

Results also showed that the more recently people had been treated with the B cell–depleting therapy rituximab, the lower the levels of immunogenicity that were achieved, and thus protection against SARS-CoV-2.

“Our results have significant implications for individuals treated with rituximab in the post-Omicron era, highlighting the value of additive boosters in affirming increasing protection in clinically vulnerable populations,” the team behind the work at the University of Cambridge in England, has reported in Science Advances.

Moreover, because the use of rituximab reduced the neutralization of not just wild-type (WT) Omicron but also the Omicron-descendant variants BA.1, BA.2, BA.4, and XBB, this highlights “the urgent need for additional adjunctive strategies to enhance vaccine-induced immunity as well as preferential access for such patients to updated vaccines using spike from now circulating Omicron lineages,” the team added.

 

Studying Humoral Responses to SARS-CoV-2 Vaccines

Corresponding author Ravindra K. Gupta, BMBCh, MA, MPH, PhD, told this news organization that studying humoral responses to SARS-CoV-2 vaccines in immunocompromised individuals such as those with vasculitis was important for two main reasons.

“It is really important at individual level for their own health, of course, but also because we know that variants of concern have often evolved and developed within patients and can then spread in wider populations,” he said.

Gupta, who is professor of clinical microbiology at the Cambridge Institute for Therapeutic Immunology & Infectious Disease added: “We believe that the variants of concern that we’re having to deal with right now, including Omicron, have come from such [immunocompromised] individuals.”

Omicron “was a big shift,” Gupta noted. “It had a lot of new mutations on it, so it was almost like a new strain of the virus.” Few studies have looked at the longitudinal immunogenicity proffered by COVID vaccines in the post-Omicron era, particularly in those with vasculitis who are often treated with immunosuppressive drugs, including rituximab.

 

Two-Pronged Study Approach

For the study, a population of immunocompromised individuals diagnosed with vasculitis who had been treated with rituximab in the past 5 years was identified. Just over half (58%) had received adenovirus-based AZD1222/ChAdOx1 nCoV-19 (AstraZeneca-Oxford; AZN) and 37% BNT162b2 (Pfizer-BioNTech; mRNA) as their primary vaccines. Patients with antineutrophil cytoplasmic antibody–associated vasculitis comprised the majority of those who received rituximab (83%), compared with less than half of those who did not take rituximab (48%).

A two-pronged approach was taken with the researchers first measuring neutralizing antibody titers before and 30 days after four successive COVID vaccinations in a group of 32 individuals with available samples. They then performed a cross-sectional, case-control study in 95 individuals to look at neutralizing antibody titers and antibody-dependent cell-mediated cytotoxicity (ADCC) in individuals who had (n = 64) and had not (n = 31) been treated with rituximab in the past 5 years and had samples available after their third and fourth COVID vaccinations.

The first analysis was done to see how people were responding to vaccination over time. “That told us that there was a problem with the first two doses and that we got some response after doses three and four, but the response was uniformly quite poor against the new variants of concern,” Gupta said.

A human embryonic kidney cell model had been used to determine individuals’ neutralizing antibody titers in response to WT, BA.1, BA.2, BA.4, and XBB pseudotyped viruses. After the first and second COVID vaccinations, the geometric mean titer (GMT) against each variant barely increased from a baseline of 40.0. The greatest increases in GMT was seen with the WT virus, at 43.7, 90.7, 256.3, and 394.2, after the first, second, third and fourth doses, respectively. The lowest increases in GMT were seen with the XBB variant, with respective values of 40.0, 40.8, 45.7, and 53.9.

 

Incremental Benefit Offers Some ‘Reassurance’

Vasculitis specialist Rona Smith, MA, MB BChir, MD, who was one of the authors of the paper, told this news organization separately that the results showed there was “an incremental benefit of having COVID vaccinations,” which “offers a little bit of reassurance” that there can be an immune response in people with vasculitis.

Although results of the cross-sectional study showed that there was a significant dampening effect of rituximab treatment on the immune response, “I don’t think it’s an isolated effect in our [vasculitis] patients,” Smith suggested, adding the results were “probably still relevant to patients who receive routine dosing of rituximab for other conditions.”

Neutralizing antibody titers were consistently lower among individuals who had been treated with rituximab vs those who had not, with treatment in the past 18 months found to significantly impair immunogenicity.

The ADCC response was better preserved than the neutralizing antibody response, Gupta said, although it was still significantly lower in the rituximab-treated than in the non–rituximab-treated patients.

 

When to Vaccinate in Vasculitis?

Regarding when to give vaccines to people with vasculitis, Smith said: “Current recommendations are that patients should receive any vaccines that they’re offered routinely, whether that be COVID vaccines, flu vaccines, pneumococcal vaccines.”

As for the timing of those vaccinations, she observed that the current thinking was that vaccinations should “ideally be at least 1 month before a rituximab treatment, and ideally 3-4 [months] after their last dose. However, as many patients are on a 6-month dosing cycle, it can be difficult for some of them to find a suitable time window to have the COVID vaccine when it is offered.”

Additional precautions, such as wearing masks in crowded places and avoiding visits to acutely unwell friends or relatives, may still be prudent, Smith acknowledged, but he was clear that people should not be locking themselves away as they did during the COVID-19 pandemic.

When advising patients, “our general recommendation is that it is better to have a vaccine than not, but we can’t guarantee how well you will respond to it, but some response is better than none,” Smith said.

The study was independently supported. Gupta had no relevant financial relationships to disclose. Smith was a coauthor of the paper and has received research grant funding from Union Therapeutics, GlaxoSmithKline/Vir Biotechnology, Addenbrooke’s Charitable Trust, and Vasculitis UK. Another coauthor reported receiving research grants from CSL Vifor, Roche, and GlaxoSmithKline and advisory board, consultancy, and lecture fees from Roche and CSL Vifor.

A version of this article first appeared on Medscape.com.

People with vasculitis may need at least three or four vaccinations for COVID-19 before they start to show an immune response against SARS-CoV-2 infection, new research has suggested.

In a longitudinal retrospective study, serum antibody neutralization against the Omicron variant of the virus and its descendants was found to be “largely absent” after the first two doses of COVID-19 vaccine had been given to patients. But increasing neutralizing antibody titers were seen after both the third and fourth vaccine boosters had been administered.

Results also showed that the more recently people had been treated with the B cell–depleting therapy rituximab, the lower the levels of immunogenicity that were achieved, and thus protection against SARS-CoV-2.

“Our results have significant implications for individuals treated with rituximab in the post-Omicron era, highlighting the value of additive boosters in affirming increasing protection in clinically vulnerable populations,” the team behind the work at the University of Cambridge in England, has reported in Science Advances.

Moreover, because the use of rituximab reduced the neutralization of not just wild-type (WT) Omicron but also the Omicron-descendant variants BA.1, BA.2, BA.4, and XBB, this highlights “the urgent need for additional adjunctive strategies to enhance vaccine-induced immunity as well as preferential access for such patients to updated vaccines using spike from now circulating Omicron lineages,” the team added.

 

Studying Humoral Responses to SARS-CoV-2 Vaccines

Corresponding author Ravindra K. Gupta, BMBCh, MA, MPH, PhD, told this news organization that studying humoral responses to SARS-CoV-2 vaccines in immunocompromised individuals such as those with vasculitis was important for two main reasons.

“It is really important at individual level for their own health, of course, but also because we know that variants of concern have often evolved and developed within patients and can then spread in wider populations,” he said.

Gupta, who is professor of clinical microbiology at the Cambridge Institute for Therapeutic Immunology & Infectious Disease added: “We believe that the variants of concern that we’re having to deal with right now, including Omicron, have come from such [immunocompromised] individuals.”

Omicron “was a big shift,” Gupta noted. “It had a lot of new mutations on it, so it was almost like a new strain of the virus.” Few studies have looked at the longitudinal immunogenicity proffered by COVID vaccines in the post-Omicron era, particularly in those with vasculitis who are often treated with immunosuppressive drugs, including rituximab.

 

Two-Pronged Study Approach

For the study, a population of immunocompromised individuals diagnosed with vasculitis who had been treated with rituximab in the past 5 years was identified. Just over half (58%) had received adenovirus-based AZD1222/ChAdOx1 nCoV-19 (AstraZeneca-Oxford; AZN) and 37% BNT162b2 (Pfizer-BioNTech; mRNA) as their primary vaccines. Patients with antineutrophil cytoplasmic antibody–associated vasculitis comprised the majority of those who received rituximab (83%), compared with less than half of those who did not take rituximab (48%).

A two-pronged approach was taken with the researchers first measuring neutralizing antibody titers before and 30 days after four successive COVID vaccinations in a group of 32 individuals with available samples. They then performed a cross-sectional, case-control study in 95 individuals to look at neutralizing antibody titers and antibody-dependent cell-mediated cytotoxicity (ADCC) in individuals who had (n = 64) and had not (n = 31) been treated with rituximab in the past 5 years and had samples available after their third and fourth COVID vaccinations.

The first analysis was done to see how people were responding to vaccination over time. “That told us that there was a problem with the first two doses and that we got some response after doses three and four, but the response was uniformly quite poor against the new variants of concern,” Gupta said.

A human embryonic kidney cell model had been used to determine individuals’ neutralizing antibody titers in response to WT, BA.1, BA.2, BA.4, and XBB pseudotyped viruses. After the first and second COVID vaccinations, the geometric mean titer (GMT) against each variant barely increased from a baseline of 40.0. The greatest increases in GMT was seen with the WT virus, at 43.7, 90.7, 256.3, and 394.2, after the first, second, third and fourth doses, respectively. The lowest increases in GMT were seen with the XBB variant, with respective values of 40.0, 40.8, 45.7, and 53.9.

 

Incremental Benefit Offers Some ‘Reassurance’

Vasculitis specialist Rona Smith, MA, MB BChir, MD, who was one of the authors of the paper, told this news organization separately that the results showed there was “an incremental benefit of having COVID vaccinations,” which “offers a little bit of reassurance” that there can be an immune response in people with vasculitis.

Although results of the cross-sectional study showed that there was a significant dampening effect of rituximab treatment on the immune response, “I don’t think it’s an isolated effect in our [vasculitis] patients,” Smith suggested, adding the results were “probably still relevant to patients who receive routine dosing of rituximab for other conditions.”

Neutralizing antibody titers were consistently lower among individuals who had been treated with rituximab vs those who had not, with treatment in the past 18 months found to significantly impair immunogenicity.

The ADCC response was better preserved than the neutralizing antibody response, Gupta said, although it was still significantly lower in the rituximab-treated than in the non–rituximab-treated patients.

 

When to Vaccinate in Vasculitis?

Regarding when to give vaccines to people with vasculitis, Smith said: “Current recommendations are that patients should receive any vaccines that they’re offered routinely, whether that be COVID vaccines, flu vaccines, pneumococcal vaccines.”

As for the timing of those vaccinations, she observed that the current thinking was that vaccinations should “ideally be at least 1 month before a rituximab treatment, and ideally 3-4 [months] after their last dose. However, as many patients are on a 6-month dosing cycle, it can be difficult for some of them to find a suitable time window to have the COVID vaccine when it is offered.”

Additional precautions, such as wearing masks in crowded places and avoiding visits to acutely unwell friends or relatives, may still be prudent, Smith acknowledged, but he was clear that people should not be locking themselves away as they did during the COVID-19 pandemic.

When advising patients, “our general recommendation is that it is better to have a vaccine than not, but we can’t guarantee how well you will respond to it, but some response is better than none,” Smith said.

The study was independently supported. Gupta had no relevant financial relationships to disclose. Smith was a coauthor of the paper and has received research grant funding from Union Therapeutics, GlaxoSmithKline/Vir Biotechnology, Addenbrooke’s Charitable Trust, and Vasculitis UK. Another coauthor reported receiving research grants from CSL Vifor, Roche, and GlaxoSmithKline and advisory board, consultancy, and lecture fees from Roche and CSL Vifor.

A version of this article first appeared on Medscape.com.

A new analysis of long-COVID patients has identified five distinct subtypes that researchers say will help doctors diagnose the condition.

The new five-type index, developed by federal researchers with the National Institutes of Health’s RECOVER COVID Initiative, identified the most common symptoms in 14,000 people with long COVID, with data from an additional 4000 people added to the updated 2024 index.

By using the index, physicians and researchers can better understand the condition, which is difficult to treat and diagnose because no standard definitions or therapies have been developed. Doctors can use the index to offer more targeted care and help patients manage their symptoms more effectively.

The index may also help researchers find more treatments for long COVID. Because long COVID can affect so many different parts of the body, it will take time to fully understand how to treat it, but studies like this are making progress in the right direction, experts said.

This new index uses an updated point system, where points are allotted to each symptom in a list of the 44 most reported symptoms in people with likely long COVID based on how often they occur. Among people in the study with prior COVID infection, 2213 (18%) met the threshold for long COVID.

The 44 most common symptoms were then distributed among 5 subtypes, with each representing a difference in impact on quality of life and overall health. The most common symptoms were fatigue (85.8%), postexertional malaise (87.4%), and postexertional soreness (75.0%) — where persistent fatigue and discomfort occur after physical or mental exertion — dizziness (65.8%), brain fog (63.8%), gastrointestinal symptoms (59.3%), and palpitations (58%).

For those with prior COVID infection, symptoms were more prevalent in all cases.

 

Subtype 1

Those grouped into subtype 1 did not report a high incidence of impact on quality of life, physical health, or daily function. Only 21% of people in subtype 1 reported a “poor or fair quality of life.”

A change in smell or taste — usually a symptom that’s bothersome but doesn’t seriously impact overall health — was most present in subtype 1, with 100% of people in subtype 1 reporting it.

The only other symptoms in over 50% of people with subtype 1— which were 490 of the 2213 with prior COVID infection — were fatigue (66%), postexertional malaise (53%), and postexertional soreness (55%).

Though these two symptoms can certainly impact quality of life, they became much more prevalent in other subtypes.

 

Subtype 2

The prevalence of possibly debilitating symptoms like postexertional malaise (94%), fatigue (81%), and chronic cough (100%) rose dramatically in people grouped into subtype 2. 

Plus, 25% of people in subtype 2 reported a “poor or fair quality of life. Postexertional malaise, I think, is probably one of the most debilitating of the symptoms. When somebody comes in and tells me that they’re tired and I think they might have long COVID, the first thing I try to do is see if it is postexertional malaise vs just postinfectious fatigue,” said Lisa Sanders, MD, medical director of Yale’s Long Covid Multidisciplinary Care Center in New Haven, Connecticut.

Postinfectious fatigue usually resolves much more quickly than postexertional malaise. The latter accounts for several symptoms as also associated with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). ME/CFS is a chronic illness that causes severe fatigue and makes it difficult for sufferers to perform routine, daily activities.

“Postexertional malaise is an additive symptom of ME/CFS, and that can take a long time to resolve,” Sanders added.

The similarity between these two symptoms highlights the importance that physicians must place in scrutinizing symptoms to a high degree when they suspect a patient of having long COVID, experts said. By doing so, clinicians can unveil the mask of overlapping symptoms between long COVID symptoms and symptoms of other illnesses.

 

Subtype 3

About 37% of people grouped in subtype 3 reported a poor or fair quality of life, a significant rise from subtypes 1 and 2.

Fatigue symptoms were reported by 92%, whereas 82% reported postexertional soreness, and 70% reported dizziness. Additionally, 100% of people in subtype 3 reported brain fog as a symptom.

Sanders said these symptoms are also common in people with postural orthostatic tachycardia syndrome. This condition results from a reduced volume of blood returning to the heart after standing up, which leads to an abnormally fast heart rate. Palpitations and fainting can then occur.

Brain fog can be especially debilitating in people who are used to multitasking. With brain fog, people accustomed to easily alternating between tasks or doing multiple tasks at once can only do one thing at a time. This can cause stress and an overload of thoughts, even precipitating a change in careers if severe enough.

Though brain fog tends to resolve within 6-9 months after infection, it can last up to 18 months or more. Experts say doctors should always be on the lookout if a patient complains they have trouble concentrating or multitasking in the months after a COVID infection. A neurological exam and cognitive testing can identify abnormalities in brain function.

 

Subtype 4

About 40% of people in the study grouped into subtype 4 reported a poor or fair quality of life, a modest increase from those with subtype 3. About 65% reported symptoms of brain fog and 92% reported palpitations.

Dizziness was also prevalent at 71%, whereas 60% reported gastrointestinal issues, and 36% said they experienced fever, sweats, and chills.

Nearly 700 of the 2213 people fell into this subtype group, by far the highest number.

 

Subtype 5

A whopping 66% of people in subtype 5 reported a poor to fair quality of life. These people usually reported multisystem symptoms.

In terms of prevalence rises across the spectrum of 44 common long-COVID symptoms, 99% reported shortness of breath; 98%, postexertional soreness; 94%, dizziness; 92%, postexertional malaise; 80%, GI problems; 78%, weakness; and 69%, chest pain.

A higher proportion of Hispanic and multiracial participants were classified as having subtype 5. Also, according to the study, “higher proportions of unvaccinated participants and those with SARS-CoV-2 infection before circulation of the Omicron variant were in subtype 5.”

This suggests the severity of the Delta variant of COVID-19 be linked to some of the worst long COVID symptoms, but further study would have to be done to conclusively determine may be just a correlation.

 

When Do Symptoms Resolve?

According to Sanders, around 17 million Americans are thought to have long COVID. Although 90%-100% of people typically recover within 3 years, that still leaves possibly around 5% of those who don’t recover.

“What people usually say is, ‘I got COVID, and I never quite recovered,” Sanders said.

“Five percent of 17 million turns out to be a lot. It’s a lot of suffering,” she added. “I would say that the most common symptoms are fatigue, brain fog, anosmia or dysgeusia, and sleep disorders,” as evidenced by the high percentage of people in certain subtypes of the study reporting a poor quality of life.

A version of this article first appeared on Medscape.com.

A new analysis of long-COVID patients has identified five distinct subtypes that researchers say will help doctors diagnose the condition.

The new five-type index, developed by federal researchers with the National Institutes of Health’s RECOVER COVID Initiative, identified the most common symptoms in 14,000 people with long COVID, with data from an additional 4000 people added to the updated 2024 index.

By using the index, physicians and researchers can better understand the condition, which is difficult to treat and diagnose because no standard definitions or therapies have been developed. Doctors can use the index to offer more targeted care and help patients manage their symptoms more effectively.

The index may also help researchers find more treatments for long COVID. Because long COVID can affect so many different parts of the body, it will take time to fully understand how to treat it, but studies like this are making progress in the right direction, experts said.

This new index uses an updated point system, where points are allotted to each symptom in a list of the 44 most reported symptoms in people with likely long COVID based on how often they occur. Among people in the study with prior COVID infection, 2213 (18%) met the threshold for long COVID.

The 44 most common symptoms were then distributed among 5 subtypes, with each representing a difference in impact on quality of life and overall health. The most common symptoms were fatigue (85.8%), postexertional malaise (87.4%), and postexertional soreness (75.0%) — where persistent fatigue and discomfort occur after physical or mental exertion — dizziness (65.8%), brain fog (63.8%), gastrointestinal symptoms (59.3%), and palpitations (58%).

For those with prior COVID infection, symptoms were more prevalent in all cases.

 

Subtype 1

Those grouped into subtype 1 did not report a high incidence of impact on quality of life, physical health, or daily function. Only 21% of people in subtype 1 reported a “poor or fair quality of life.”

A change in smell or taste — usually a symptom that’s bothersome but doesn’t seriously impact overall health — was most present in subtype 1, with 100% of people in subtype 1 reporting it.

The only other symptoms in over 50% of people with subtype 1— which were 490 of the 2213 with prior COVID infection — were fatigue (66%), postexertional malaise (53%), and postexertional soreness (55%).

Though these two symptoms can certainly impact quality of life, they became much more prevalent in other subtypes.

 

Subtype 2

The prevalence of possibly debilitating symptoms like postexertional malaise (94%), fatigue (81%), and chronic cough (100%) rose dramatically in people grouped into subtype 2. 

Plus, 25% of people in subtype 2 reported a “poor or fair quality of life. Postexertional malaise, I think, is probably one of the most debilitating of the symptoms. When somebody comes in and tells me that they’re tired and I think they might have long COVID, the first thing I try to do is see if it is postexertional malaise vs just postinfectious fatigue,” said Lisa Sanders, MD, medical director of Yale’s Long Covid Multidisciplinary Care Center in New Haven, Connecticut.

Postinfectious fatigue usually resolves much more quickly than postexertional malaise. The latter accounts for several symptoms as also associated with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). ME/CFS is a chronic illness that causes severe fatigue and makes it difficult for sufferers to perform routine, daily activities.

“Postexertional malaise is an additive symptom of ME/CFS, and that can take a long time to resolve,” Sanders added.

The similarity between these two symptoms highlights the importance that physicians must place in scrutinizing symptoms to a high degree when they suspect a patient of having long COVID, experts said. By doing so, clinicians can unveil the mask of overlapping symptoms between long COVID symptoms and symptoms of other illnesses.

 

Subtype 3

About 37% of people grouped in subtype 3 reported a poor or fair quality of life, a significant rise from subtypes 1 and 2.

Fatigue symptoms were reported by 92%, whereas 82% reported postexertional soreness, and 70% reported dizziness. Additionally, 100% of people in subtype 3 reported brain fog as a symptom.

Sanders said these symptoms are also common in people with postural orthostatic tachycardia syndrome. This condition results from a reduced volume of blood returning to the heart after standing up, which leads to an abnormally fast heart rate. Palpitations and fainting can then occur.

Brain fog can be especially debilitating in people who are used to multitasking. With brain fog, people accustomed to easily alternating between tasks or doing multiple tasks at once can only do one thing at a time. This can cause stress and an overload of thoughts, even precipitating a change in careers if severe enough.

Though brain fog tends to resolve within 6-9 months after infection, it can last up to 18 months or more. Experts say doctors should always be on the lookout if a patient complains they have trouble concentrating or multitasking in the months after a COVID infection. A neurological exam and cognitive testing can identify abnormalities in brain function.

 

Subtype 4

About 40% of people in the study grouped into subtype 4 reported a poor or fair quality of life, a modest increase from those with subtype 3. About 65% reported symptoms of brain fog and 92% reported palpitations.

Dizziness was also prevalent at 71%, whereas 60% reported gastrointestinal issues, and 36% said they experienced fever, sweats, and chills.

Nearly 700 of the 2213 people fell into this subtype group, by far the highest number.

 

Subtype 5

A whopping 66% of people in subtype 5 reported a poor to fair quality of life. These people usually reported multisystem symptoms.

In terms of prevalence rises across the spectrum of 44 common long-COVID symptoms, 99% reported shortness of breath; 98%, postexertional soreness; 94%, dizziness; 92%, postexertional malaise; 80%, GI problems; 78%, weakness; and 69%, chest pain.

A higher proportion of Hispanic and multiracial participants were classified as having subtype 5. Also, according to the study, “higher proportions of unvaccinated participants and those with SARS-CoV-2 infection before circulation of the Omicron variant were in subtype 5.”

This suggests the severity of the Delta variant of COVID-19 be linked to some of the worst long COVID symptoms, but further study would have to be done to conclusively determine may be just a correlation.

 

When Do Symptoms Resolve?

According to Sanders, around 17 million Americans are thought to have long COVID. Although 90%-100% of people typically recover within 3 years, that still leaves possibly around 5% of those who don’t recover.

“What people usually say is, ‘I got COVID, and I never quite recovered,” Sanders said.

“Five percent of 17 million turns out to be a lot. It’s a lot of suffering,” she added. “I would say that the most common symptoms are fatigue, brain fog, anosmia or dysgeusia, and sleep disorders,” as evidenced by the high percentage of people in certain subtypes of the study reporting a poor quality of life.

A version of this article first appeared on Medscape.com.

A new analysis of long-COVID patients has identified five distinct subtypes that researchers say will help doctors diagnose the condition.

The new five-type index, developed by federal researchers with the National Institutes of Health’s RECOVER COVID Initiative, identified the most common symptoms in 14,000 people with long COVID, with data from an additional 4000 people added to the updated 2024 index.

By using the index, physicians and researchers can better understand the condition, which is difficult to treat and diagnose because no standard definitions or therapies have been developed. Doctors can use the index to offer more targeted care and help patients manage their symptoms more effectively.

The index may also help researchers find more treatments for long COVID. Because long COVID can affect so many different parts of the body, it will take time to fully understand how to treat it, but studies like this are making progress in the right direction, experts said.

This new index uses an updated point system, where points are allotted to each symptom in a list of the 44 most reported symptoms in people with likely long COVID based on how often they occur. Among people in the study with prior COVID infection, 2213 (18%) met the threshold for long COVID.

The 44 most common symptoms were then distributed among 5 subtypes, with each representing a difference in impact on quality of life and overall health. The most common symptoms were fatigue (85.8%), postexertional malaise (87.4%), and postexertional soreness (75.0%) — where persistent fatigue and discomfort occur after physical or mental exertion — dizziness (65.8%), brain fog (63.8%), gastrointestinal symptoms (59.3%), and palpitations (58%).

For those with prior COVID infection, symptoms were more prevalent in all cases.

 

Subtype 1

Those grouped into subtype 1 did not report a high incidence of impact on quality of life, physical health, or daily function. Only 21% of people in subtype 1 reported a “poor or fair quality of life.”

A change in smell or taste — usually a symptom that’s bothersome but doesn’t seriously impact overall health — was most present in subtype 1, with 100% of people in subtype 1 reporting it.

The only other symptoms in over 50% of people with subtype 1— which were 490 of the 2213 with prior COVID infection — were fatigue (66%), postexertional malaise (53%), and postexertional soreness (55%).

Though these two symptoms can certainly impact quality of life, they became much more prevalent in other subtypes.

 

Subtype 2

The prevalence of possibly debilitating symptoms like postexertional malaise (94%), fatigue (81%), and chronic cough (100%) rose dramatically in people grouped into subtype 2. 

Plus, 25% of people in subtype 2 reported a “poor or fair quality of life. Postexertional malaise, I think, is probably one of the most debilitating of the symptoms. When somebody comes in and tells me that they’re tired and I think they might have long COVID, the first thing I try to do is see if it is postexertional malaise vs just postinfectious fatigue,” said Lisa Sanders, MD, medical director of Yale’s Long Covid Multidisciplinary Care Center in New Haven, Connecticut.

Postinfectious fatigue usually resolves much more quickly than postexertional malaise. The latter accounts for several symptoms as also associated with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). ME/CFS is a chronic illness that causes severe fatigue and makes it difficult for sufferers to perform routine, daily activities.

“Postexertional malaise is an additive symptom of ME/CFS, and that can take a long time to resolve,” Sanders added.

The similarity between these two symptoms highlights the importance that physicians must place in scrutinizing symptoms to a high degree when they suspect a patient of having long COVID, experts said. By doing so, clinicians can unveil the mask of overlapping symptoms between long COVID symptoms and symptoms of other illnesses.

 

Subtype 3

About 37% of people grouped in subtype 3 reported a poor or fair quality of life, a significant rise from subtypes 1 and 2.

Fatigue symptoms were reported by 92%, whereas 82% reported postexertional soreness, and 70% reported dizziness. Additionally, 100% of people in subtype 3 reported brain fog as a symptom.

Sanders said these symptoms are also common in people with postural orthostatic tachycardia syndrome. This condition results from a reduced volume of blood returning to the heart after standing up, which leads to an abnormally fast heart rate. Palpitations and fainting can then occur.

Brain fog can be especially debilitating in people who are used to multitasking. With brain fog, people accustomed to easily alternating between tasks or doing multiple tasks at once can only do one thing at a time. This can cause stress and an overload of thoughts, even precipitating a change in careers if severe enough.

Though brain fog tends to resolve within 6-9 months after infection, it can last up to 18 months or more. Experts say doctors should always be on the lookout if a patient complains they have trouble concentrating or multitasking in the months after a COVID infection. A neurological exam and cognitive testing can identify abnormalities in brain function.

 

Subtype 4

About 40% of people in the study grouped into subtype 4 reported a poor or fair quality of life, a modest increase from those with subtype 3. About 65% reported symptoms of brain fog and 92% reported palpitations.

Dizziness was also prevalent at 71%, whereas 60% reported gastrointestinal issues, and 36% said they experienced fever, sweats, and chills.

Nearly 700 of the 2213 people fell into this subtype group, by far the highest number.

 

Subtype 5

A whopping 66% of people in subtype 5 reported a poor to fair quality of life. These people usually reported multisystem symptoms.

In terms of prevalence rises across the spectrum of 44 common long-COVID symptoms, 99% reported shortness of breath; 98%, postexertional soreness; 94%, dizziness; 92%, postexertional malaise; 80%, GI problems; 78%, weakness; and 69%, chest pain.

A higher proportion of Hispanic and multiracial participants were classified as having subtype 5. Also, according to the study, “higher proportions of unvaccinated participants and those with SARS-CoV-2 infection before circulation of the Omicron variant were in subtype 5.”

This suggests the severity of the Delta variant of COVID-19 be linked to some of the worst long COVID symptoms, but further study would have to be done to conclusively determine may be just a correlation.

 

When Do Symptoms Resolve?

According to Sanders, around 17 million Americans are thought to have long COVID. Although 90%-100% of people typically recover within 3 years, that still leaves possibly around 5% of those who don’t recover.

“What people usually say is, ‘I got COVID, and I never quite recovered,” Sanders said.

“Five percent of 17 million turns out to be a lot. It’s a lot of suffering,” she added. “I would say that the most common symptoms are fatigue, brain fog, anosmia or dysgeusia, and sleep disorders,” as evidenced by the high percentage of people in certain subtypes of the study reporting a poor quality of life.

A version of this article first appeared on Medscape.com.

Chronic cough remains a common reason for consultation in pulmonology post–COVID-19. But what do we really know about this condition, now 5 years after the pandemic’s onset? This topic was discussed at the recent French-Speaking Pneumology Congress held in Marseille, France, from January 24-26, 2025.

Before discussing post-COVID cough, it is crucial to differentiate between an acute cough, often viral in origin (including those associated with SARS-CoV-2), a subacute cough (lasting 3-8 weeks), and a chronic cough (persisting over 8 weeks).

“This distinction allows us to tailor treatment and prescribe the appropriate investigations, according to the duration and the probability of symptom resolution,” explained Laurent Guilleminault, MD, PhD, pulmonologist at Toulouse University Hospital Centre, Toulouse, France.

In the case of an acute cough, for instance, after a viral infection, the probability of spontaneous resolution is very high. It is often unnecessary to carry out additional examinations or initiate specific treatments because none has proven its effectiveness in shortening this type of cough. On the other hand, when a cough persists beyond 8 weeks, the chance of spontaneous resolution decreases considerably. “This is when an assessment is necessary to identify a possible underlying cause,” Guilleminault noted.

“The absence of coughing during the consultation should not lead to ruling out a diagnosis,” he added.

Neurological Link

A large-scale French study of 70,000 patients examined the demographic profiles of patients with COVID-19. It revealed a lower frequency of coughing among children and older individuals, with a notable prevalence among adults aged 30-60 years.

Furthermore, during the acute phase of COVID, coughing did not appear to indicate severity. A comparison between survivors and nonsurvivors revealed no significant differences in the frequency and severity of coughing. Another study concluded that, contrary to expectations, COVID-related pneumonia, although potentially severe, does not necessarily involve severe cough.

These findings highlight the absence of a direct link between coughing and pulmonary involvement in patients with COVID-19.

“Coughing appears to be more closely linked to neurological dysfunction than to classic respiratory involvement. A distinction that is essential for better understanding the pathophysiology of the disease and guiding therapeutic strategies,” Guilleminault noted.

Cough Mechanism

“The analysis of cough in the context of phylogenetic evolution is fascinating,” explained Guilleminault. “It illustrates how this reflex has provided an advantage to the virus for its propagation.” Studies on the transmission of SARS-CoV-2 have confirmed that coughing plays a key role in the spread of viral particles. However, this mechanism does not involve severe pulmonary damage. The primary goal of the virus is to induce neurological dysfunction in the host by triggering a cough reflex. This neurological activation enables the virus to trigger a cough reflex for dissemination even without significant pulmonary damage. This mechanism provides an evolutionary advantage by enhancing the ability of the virus to spread and colonize new hosts.

The cough mechanism remains partially understood and involves cough hypersensitivity, characterized by increased neural responsivity to a range of stimuli that affect the airways, lungs, and other tissues innervated by common nerve supplies. The cough reflex begins with the activation of sensitive peripheral receptors located mainly in the respiratory tract that detect irritants or abnormalities.

These receptors, such as P2X2, P2X3, and others, transmit information to the brainstem, which coordinates the reflex response. This process is modulated by cortical controls that normally inhibit spontaneous coughing, explaining why we do not cough constantly even in the presence of moderate stimuli.

However, when there is an imbalance in this inhibition mechanism, coughing can be triggered either excessively or uncontrollably. SARS-CoV-2 appears to interact directly with these peripheral receptors, stimulating the cough reflex. The widespread presence and density of these receptors make this mechanism highly effective for the virus’s transmission.

Additionally, the vagus nerve likely plays a central role in triggering cough, particularly in viral infections. Studies of influenza have shown the involvement of sensory cells associated with the vagus nerve.

The virus stimulates the vagus nerve, which activates the cough reflex. Research suggests that neurotropism, neuroinflammation, and neuroimmunomodulation via the vagal sensory nerves, which are involved in SARS-CoV-2 infection, lead to cough hypersensitivity.

One question remains: Could vagus nerve involvement prolong coughing beyond the active phase of viral infection? The data indicate that viral infection significantly increases the sensitivity of the cough reflex, regardless of the level of irritation. The brain areas involved in inhibiting this reflex appear less effective during viral infection, resulting in reduced inhibitory control and easier triggering of cough. This phenomenon reflects temporary dysfunction of the neurological modulation system, which gradually recovers after recovery.

Long-Term Effects

The epidemiology of post-COVID cough and its integration into the framework of the long COVID framework remain subjects of ongoing debate. Early studies have revealed that cough could be either an isolated symptom or associated with other manifestations of long COVID. These studies were often conducted over relatively short periods (14-110 days) and estimated that approximately 19% of patients with long COVID experienced persistent cough. Another study found that 14% of patients reported cough between 3 weeks and 3 months after hospital discharge for COVID-19.

Longer follow-up periods showed a significant decrease in the prevalence of cough over time. For instance, a 1-year study reported that only 2.5% of patients had episodes of chronic cough.

However, a 2023 study published in JAMA found that the prevalence of post-COVID chronic cough exceeded 30% in some groups of patients.

“It is not relevant to wait so long before acting,” Guilleminault said. A reasonable threshold for evaluation and treatment is 8-12 weeks postinfection to begin investigations and consider appropriate treatment. What should be done when a patient presents with “Doctor, I had COVID, I have a cough, and it hasn’t stopped?” These situations are common in clinical practice. In terms of severity, quality of life, and overall impact, patients with chronic post-COVID cough are not significantly different from those with other chronic coughs. Moreover, both conditions involve a real neurological dysfunction.

Same Diagnostic Steps

Management should follow existing guidelines, including the recent French recommendations for chronic cough.

A visual analog scale can be used, and possible complications should be assessed. A chest x-ray is recommended to identify any warning signs, such as cough, although linked to COVID — may coincide with other conditions, such as bronchial cancer. In smokers, chest CT should be considered to rule out neoplastic pathology. The presence of interstitial lesions, particularly fibrosing lesions, suggests that fibrosing interstitial pneumonia requires specialized management.

Smoking, which is an aggravating factor, should be discontinued. Discontinuing angiotensin-converting enzyme inhibitors for 4 weeks can help determine if they contribute to cough.

The three most common causes of chronic cough — rhinosinusitis, asthma, and gastroesophageal reflux disease — should be ruled out. Diagnosis is based on history, physical examination, and specific tests: Nasofibroscopy for rhinosinusitis, spirometry, fractional exhaled nitric oxide for asthma and clinical history of gastroesophageal reflux disease. Studies have indicated that asthma may develop after a COVID infection.

Laryngeal abnormalities are also common in chronic post-COVID cough. One study found that a quarter of patients had increased laryngeal sensitivity or voice changes. “The larynx, a highly cough-producing organ, causes more coughing than the lungs,” Guilleminault explained.

Laryngeal abnormalities are frequently observed. A study found that 63% of patients experienced dysphonia, 56% had a sensation of a foreign body in the larynx, and 10% experienced laryngospasms.

These issues are common in patients with post-COVID cough and are often associated with neurological dysfunction. Innervation of the larynx is complex and can be affected by viruses, leading to hypersensitivity, paresthesia, and other sensory disturbances, which may explain the laryngeal symptoms observed in these patients.

Next Steps

If common causes such as asthma, abnormal imaging findings, or laryngeal pathology are ruled out, the condition may be classified as a chronic refractory or unexplained cough. In these cases, the neurological origin is likely due to nervous system dysfunction. Neuromodulatory treatments including amitriptyline, pregabalin, and gabapentin may be considered in some cases. Corticosteroids are generally ineffective against chronic coughs.

This story was translated from Medscape’s French edition using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.

Chronic cough remains a common reason for consultation in pulmonology post–COVID-19. But what do we really know about this condition, now 5 years after the pandemic’s onset? This topic was discussed at the recent French-Speaking Pneumology Congress held in Marseille, France, from January 24-26, 2025.

Before discussing post-COVID cough, it is crucial to differentiate between an acute cough, often viral in origin (including those associated with SARS-CoV-2), a subacute cough (lasting 3-8 weeks), and a chronic cough (persisting over 8 weeks).

“This distinction allows us to tailor treatment and prescribe the appropriate investigations, according to the duration and the probability of symptom resolution,” explained Laurent Guilleminault, MD, PhD, pulmonologist at Toulouse University Hospital Centre, Toulouse, France.

In the case of an acute cough, for instance, after a viral infection, the probability of spontaneous resolution is very high. It is often unnecessary to carry out additional examinations or initiate specific treatments because none has proven its effectiveness in shortening this type of cough. On the other hand, when a cough persists beyond 8 weeks, the chance of spontaneous resolution decreases considerably. “This is when an assessment is necessary to identify a possible underlying cause,” Guilleminault noted.

“The absence of coughing during the consultation should not lead to ruling out a diagnosis,” he added.

Neurological Link

A large-scale French study of 70,000 patients examined the demographic profiles of patients with COVID-19. It revealed a lower frequency of coughing among children and older individuals, with a notable prevalence among adults aged 30-60 years.

Furthermore, during the acute phase of COVID, coughing did not appear to indicate severity. A comparison between survivors and nonsurvivors revealed no significant differences in the frequency and severity of coughing. Another study concluded that, contrary to expectations, COVID-related pneumonia, although potentially severe, does not necessarily involve severe cough.

These findings highlight the absence of a direct link between coughing and pulmonary involvement in patients with COVID-19.

“Coughing appears to be more closely linked to neurological dysfunction than to classic respiratory involvement. A distinction that is essential for better understanding the pathophysiology of the disease and guiding therapeutic strategies,” Guilleminault noted.

Cough Mechanism

“The analysis of cough in the context of phylogenetic evolution is fascinating,” explained Guilleminault. “It illustrates how this reflex has provided an advantage to the virus for its propagation.” Studies on the transmission of SARS-CoV-2 have confirmed that coughing plays a key role in the spread of viral particles. However, this mechanism does not involve severe pulmonary damage. The primary goal of the virus is to induce neurological dysfunction in the host by triggering a cough reflex. This neurological activation enables the virus to trigger a cough reflex for dissemination even without significant pulmonary damage. This mechanism provides an evolutionary advantage by enhancing the ability of the virus to spread and colonize new hosts.

The cough mechanism remains partially understood and involves cough hypersensitivity, characterized by increased neural responsivity to a range of stimuli that affect the airways, lungs, and other tissues innervated by common nerve supplies. The cough reflex begins with the activation of sensitive peripheral receptors located mainly in the respiratory tract that detect irritants or abnormalities.

These receptors, such as P2X2, P2X3, and others, transmit information to the brainstem, which coordinates the reflex response. This process is modulated by cortical controls that normally inhibit spontaneous coughing, explaining why we do not cough constantly even in the presence of moderate stimuli.

However, when there is an imbalance in this inhibition mechanism, coughing can be triggered either excessively or uncontrollably. SARS-CoV-2 appears to interact directly with these peripheral receptors, stimulating the cough reflex. The widespread presence and density of these receptors make this mechanism highly effective for the virus’s transmission.

Additionally, the vagus nerve likely plays a central role in triggering cough, particularly in viral infections. Studies of influenza have shown the involvement of sensory cells associated with the vagus nerve.

The virus stimulates the vagus nerve, which activates the cough reflex. Research suggests that neurotropism, neuroinflammation, and neuroimmunomodulation via the vagal sensory nerves, which are involved in SARS-CoV-2 infection, lead to cough hypersensitivity.

One question remains: Could vagus nerve involvement prolong coughing beyond the active phase of viral infection? The data indicate that viral infection significantly increases the sensitivity of the cough reflex, regardless of the level of irritation. The brain areas involved in inhibiting this reflex appear less effective during viral infection, resulting in reduced inhibitory control and easier triggering of cough. This phenomenon reflects temporary dysfunction of the neurological modulation system, which gradually recovers after recovery.

Long-Term Effects

The epidemiology of post-COVID cough and its integration into the framework of the long COVID framework remain subjects of ongoing debate. Early studies have revealed that cough could be either an isolated symptom or associated with other manifestations of long COVID. These studies were often conducted over relatively short periods (14-110 days) and estimated that approximately 19% of patients with long COVID experienced persistent cough. Another study found that 14% of patients reported cough between 3 weeks and 3 months after hospital discharge for COVID-19.

Longer follow-up periods showed a significant decrease in the prevalence of cough over time. For instance, a 1-year study reported that only 2.5% of patients had episodes of chronic cough.

However, a 2023 study published in JAMA found that the prevalence of post-COVID chronic cough exceeded 30% in some groups of patients.

“It is not relevant to wait so long before acting,” Guilleminault said. A reasonable threshold for evaluation and treatment is 8-12 weeks postinfection to begin investigations and consider appropriate treatment. What should be done when a patient presents with “Doctor, I had COVID, I have a cough, and it hasn’t stopped?” These situations are common in clinical practice. In terms of severity, quality of life, and overall impact, patients with chronic post-COVID cough are not significantly different from those with other chronic coughs. Moreover, both conditions involve a real neurological dysfunction.

Same Diagnostic Steps

Management should follow existing guidelines, including the recent French recommendations for chronic cough.

A visual analog scale can be used, and possible complications should be assessed. A chest x-ray is recommended to identify any warning signs, such as cough, although linked to COVID — may coincide with other conditions, such as bronchial cancer. In smokers, chest CT should be considered to rule out neoplastic pathology. The presence of interstitial lesions, particularly fibrosing lesions, suggests that fibrosing interstitial pneumonia requires specialized management.

Smoking, which is an aggravating factor, should be discontinued. Discontinuing angiotensin-converting enzyme inhibitors for 4 weeks can help determine if they contribute to cough.

The three most common causes of chronic cough — rhinosinusitis, asthma, and gastroesophageal reflux disease — should be ruled out. Diagnosis is based on history, physical examination, and specific tests: Nasofibroscopy for rhinosinusitis, spirometry, fractional exhaled nitric oxide for asthma and clinical history of gastroesophageal reflux disease. Studies have indicated that asthma may develop after a COVID infection.

Laryngeal abnormalities are also common in chronic post-COVID cough. One study found that a quarter of patients had increased laryngeal sensitivity or voice changes. “The larynx, a highly cough-producing organ, causes more coughing than the lungs,” Guilleminault explained.

Laryngeal abnormalities are frequently observed. A study found that 63% of patients experienced dysphonia, 56% had a sensation of a foreign body in the larynx, and 10% experienced laryngospasms.

These issues are common in patients with post-COVID cough and are often associated with neurological dysfunction. Innervation of the larynx is complex and can be affected by viruses, leading to hypersensitivity, paresthesia, and other sensory disturbances, which may explain the laryngeal symptoms observed in these patients.

Next Steps

If common causes such as asthma, abnormal imaging findings, or laryngeal pathology are ruled out, the condition may be classified as a chronic refractory or unexplained cough. In these cases, the neurological origin is likely due to nervous system dysfunction. Neuromodulatory treatments including amitriptyline, pregabalin, and gabapentin may be considered in some cases. Corticosteroids are generally ineffective against chronic coughs.

This story was translated from Medscape’s French edition using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.

Chronic cough remains a common reason for consultation in pulmonology post–COVID-19. But what do we really know about this condition, now 5 years after the pandemic’s onset? This topic was discussed at the recent French-Speaking Pneumology Congress held in Marseille, France, from January 24-26, 2025.

Before discussing post-COVID cough, it is crucial to differentiate between an acute cough, often viral in origin (including those associated with SARS-CoV-2), a subacute cough (lasting 3-8 weeks), and a chronic cough (persisting over 8 weeks).

“This distinction allows us to tailor treatment and prescribe the appropriate investigations, according to the duration and the probability of symptom resolution,” explained Laurent Guilleminault, MD, PhD, pulmonologist at Toulouse University Hospital Centre, Toulouse, France.

In the case of an acute cough, for instance, after a viral infection, the probability of spontaneous resolution is very high. It is often unnecessary to carry out additional examinations or initiate specific treatments because none has proven its effectiveness in shortening this type of cough. On the other hand, when a cough persists beyond 8 weeks, the chance of spontaneous resolution decreases considerably. “This is when an assessment is necessary to identify a possible underlying cause,” Guilleminault noted.

“The absence of coughing during the consultation should not lead to ruling out a diagnosis,” he added.

Neurological Link

A large-scale French study of 70,000 patients examined the demographic profiles of patients with COVID-19. It revealed a lower frequency of coughing among children and older individuals, with a notable prevalence among adults aged 30-60 years.

Furthermore, during the acute phase of COVID, coughing did not appear to indicate severity. A comparison between survivors and nonsurvivors revealed no significant differences in the frequency and severity of coughing. Another study concluded that, contrary to expectations, COVID-related pneumonia, although potentially severe, does not necessarily involve severe cough.

These findings highlight the absence of a direct link between coughing and pulmonary involvement in patients with COVID-19.

“Coughing appears to be more closely linked to neurological dysfunction than to classic respiratory involvement. A distinction that is essential for better understanding the pathophysiology of the disease and guiding therapeutic strategies,” Guilleminault noted.

Cough Mechanism

“The analysis of cough in the context of phylogenetic evolution is fascinating,” explained Guilleminault. “It illustrates how this reflex has provided an advantage to the virus for its propagation.” Studies on the transmission of SARS-CoV-2 have confirmed that coughing plays a key role in the spread of viral particles. However, this mechanism does not involve severe pulmonary damage. The primary goal of the virus is to induce neurological dysfunction in the host by triggering a cough reflex. This neurological activation enables the virus to trigger a cough reflex for dissemination even without significant pulmonary damage. This mechanism provides an evolutionary advantage by enhancing the ability of the virus to spread and colonize new hosts.

The cough mechanism remains partially understood and involves cough hypersensitivity, characterized by increased neural responsivity to a range of stimuli that affect the airways, lungs, and other tissues innervated by common nerve supplies. The cough reflex begins with the activation of sensitive peripheral receptors located mainly in the respiratory tract that detect irritants or abnormalities.

These receptors, such as P2X2, P2X3, and others, transmit information to the brainstem, which coordinates the reflex response. This process is modulated by cortical controls that normally inhibit spontaneous coughing, explaining why we do not cough constantly even in the presence of moderate stimuli.

However, when there is an imbalance in this inhibition mechanism, coughing can be triggered either excessively or uncontrollably. SARS-CoV-2 appears to interact directly with these peripheral receptors, stimulating the cough reflex. The widespread presence and density of these receptors make this mechanism highly effective for the virus’s transmission.

Additionally, the vagus nerve likely plays a central role in triggering cough, particularly in viral infections. Studies of influenza have shown the involvement of sensory cells associated with the vagus nerve.

The virus stimulates the vagus nerve, which activates the cough reflex. Research suggests that neurotropism, neuroinflammation, and neuroimmunomodulation via the vagal sensory nerves, which are involved in SARS-CoV-2 infection, lead to cough hypersensitivity.

One question remains: Could vagus nerve involvement prolong coughing beyond the active phase of viral infection? The data indicate that viral infection significantly increases the sensitivity of the cough reflex, regardless of the level of irritation. The brain areas involved in inhibiting this reflex appear less effective during viral infection, resulting in reduced inhibitory control and easier triggering of cough. This phenomenon reflects temporary dysfunction of the neurological modulation system, which gradually recovers after recovery.

Long-Term Effects

The epidemiology of post-COVID cough and its integration into the framework of the long COVID framework remain subjects of ongoing debate. Early studies have revealed that cough could be either an isolated symptom or associated with other manifestations of long COVID. These studies were often conducted over relatively short periods (14-110 days) and estimated that approximately 19% of patients with long COVID experienced persistent cough. Another study found that 14% of patients reported cough between 3 weeks and 3 months after hospital discharge for COVID-19.

Longer follow-up periods showed a significant decrease in the prevalence of cough over time. For instance, a 1-year study reported that only 2.5% of patients had episodes of chronic cough.

However, a 2023 study published in JAMA found that the prevalence of post-COVID chronic cough exceeded 30% in some groups of patients.

“It is not relevant to wait so long before acting,” Guilleminault said. A reasonable threshold for evaluation and treatment is 8-12 weeks postinfection to begin investigations and consider appropriate treatment. What should be done when a patient presents with “Doctor, I had COVID, I have a cough, and it hasn’t stopped?” These situations are common in clinical practice. In terms of severity, quality of life, and overall impact, patients with chronic post-COVID cough are not significantly different from those with other chronic coughs. Moreover, both conditions involve a real neurological dysfunction.

Same Diagnostic Steps

Management should follow existing guidelines, including the recent French recommendations for chronic cough.

A visual analog scale can be used, and possible complications should be assessed. A chest x-ray is recommended to identify any warning signs, such as cough, although linked to COVID — may coincide with other conditions, such as bronchial cancer. In smokers, chest CT should be considered to rule out neoplastic pathology. The presence of interstitial lesions, particularly fibrosing lesions, suggests that fibrosing interstitial pneumonia requires specialized management.

Smoking, which is an aggravating factor, should be discontinued. Discontinuing angiotensin-converting enzyme inhibitors for 4 weeks can help determine if they contribute to cough.

The three most common causes of chronic cough — rhinosinusitis, asthma, and gastroesophageal reflux disease — should be ruled out. Diagnosis is based on history, physical examination, and specific tests: Nasofibroscopy for rhinosinusitis, spirometry, fractional exhaled nitric oxide for asthma and clinical history of gastroesophageal reflux disease. Studies have indicated that asthma may develop after a COVID infection.

Laryngeal abnormalities are also common in chronic post-COVID cough. One study found that a quarter of patients had increased laryngeal sensitivity or voice changes. “The larynx, a highly cough-producing organ, causes more coughing than the lungs,” Guilleminault explained.

Laryngeal abnormalities are frequently observed. A study found that 63% of patients experienced dysphonia, 56% had a sensation of a foreign body in the larynx, and 10% experienced laryngospasms.

These issues are common in patients with post-COVID cough and are often associated with neurological dysfunction. Innervation of the larynx is complex and can be affected by viruses, leading to hypersensitivity, paresthesia, and other sensory disturbances, which may explain the laryngeal symptoms observed in these patients.

Next Steps

If common causes such as asthma, abnormal imaging findings, or laryngeal pathology are ruled out, the condition may be classified as a chronic refractory or unexplained cough. In these cases, the neurological origin is likely due to nervous system dysfunction. Neuromodulatory treatments including amitriptyline, pregabalin, and gabapentin may be considered in some cases. Corticosteroids are generally ineffective against chronic coughs.

This story was translated from Medscape’s French edition using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.