Article

Wolff-Parkinson-White (WPW) syndrome is characterized by the presence of ≥ 1 accessory pathways and the development of both recurrent paroxysmal atrial fibrillation (AF) and supraventricular tachycardia that can lead to further malignant arrhythmias resulting in sudden cardiac death (SCD).^1-7 Historically, incidental, ventricular pre-excitation on electrocardiogram has conferred a relatively low SCD risk in adults; however, newer WPW syndrome data suggest the endpoint may not be as benign as previously thought.⁷ The current literature has defined atrioventricular reentrant tachycardia triggering AF, rather than symptoms, as an independent risk factor for malignant arrhythmias. Still, long-term data detailing the association of AF with serious cardiac events and death in patients with WPW syndrome are still limited.^1-7

While previous guidelines for the treatment of WPW syndrome only recommended routine electrophysiology testing (EPT) with liberal catheter ablation for symptomatic individuals, the 2015 American College of Cardiology/American Heart Association/Heart Rhythm Society guidelines now suggest its potential benefit for risk stratification in the asymptomatic population.^8-12 Given the limited existing data, more long-term studies are needed to corroborate the latest EPT recommendations before routinely applying them in practice. Furthermore, since concomitant AF can lead to adverse cardiac outcomes in patients with WPW syndrome, additional data evaluating this association are also necessary. In this study, we aimed to determine the impact of atrial fibrillation and/or flutter (AF/AFL) on adverse cardiac outcomes and mortality in patients with WPW syndrome.

METHODS

This study used data from the Military Health System (MHS) Database Repository. The MHS is one of the largest health care systems in the country and includes information on about 10 million active duty and retired military service members and their families (51% male; 49% female).^13,14 Data were fully anonymized and complied in accordance with federal and state laws, including the Health Insurance Portability and Accountability Act of 1996. The Naval Medical Center Portsmouth Institutional Review Board approved this study.

Study Design

This retrospective, observational cohort study identified MHS patients with WPW syndrome from January 1, 2014, to December 31, 2019. Patients were included if they had ≥ 2 International Classification of Diseases, Ninth Revision (ICD-9) or International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes for WPW syndrome (ICD-9, 426.7; ICD-10, I45.6) on separate dates; were aged ≥ 18 years at index date; and had ≥ 1 year of continuous eligibility prior to the index date (enrollment gaps ≤ 30 days were considered continuous). Patients were then divided into 2 subgroups by the presence or absence of AF/AFL using diagnostic codes. Patients were excluded if they had evidence of an implantable cardioverter-defibrillator, permanent pacemaker or were missing age or sex data. Patients were followed from index date until the first occurrence of the outcome of interest, MHS disenrollment, or the end of the study period.

Cardiac composite outcomes comprised of sudden cardiac arrest (SCA), ventricular fibrillation (VF), ventricular tachycardia and death, as well as death specifically, were the outcomes of interest and assessed after index date using ICD-9 and ICD-10 codes. Death was defined as all-cause mortality. Time to event was calculated based on the date of the initial component from the composite outcome and date of death specifically for mortality. Those not experiencing an outcome were followed until MHS disenrollment or the end of the study period.

Various patient characteristics were assessed at index including age, sex, military sponsor (the patient’s active or retired duty member through which their dependent receives TRICARE benefits) rank and branch, geographic region, type of US Department of Defense beneficiary, and index year. Clinical characteristics were assessed over a 1-year baseline period prior to index date and included the number of cardiologist and clinical visits for WPW syndrome, Charlson Comorbidity Index (CCI) scores calculated from diagnostic codes outlined in the Quan coding method, and preindex time.¹⁵ Comorbidities were assessed at baseline and defined as having ≥ 1 ICD-9 or ICD-10 code for a corresponding condition within 1 year prior to index.

Baseline characteristics were assessed and descriptive statistics for categorical and continuous variables were presented accordingly. To assess bivariate association with exposure, χ² tests were used to compare categorical variables, while t tests were used to compare continuous variables by exposure status. Incidence proportions and rates were reported for each outcome of interest. Kaplan-Meier curves were constructed to assess the bivariate association between exposure and study outcomes. Cox proportional hazard modeling was performed to estimate the association between AF/AFL and time to each of the outcomes. Multiple models were designed to assess cardiac and metabolic covariates, in addition to baseline characteristics. This included a base model adjusted for age, sex, military sponsor rank and branch, geographic region, and duty status.

Additional models adjusted for cardiac and metabolic confounders and CCI score. A comprehensive model included the base, cardiac, and metabolic covariates. Multicollinearity between covariates was assessed. Variables with a variance inflation factor > 4 or a tolerance level < 0.1 were added to the models. Cox proportional hazard models were used to estimate the unadjusted and adjusted hazard ratios (HRs) and 95% CIs of the association between AF/AFL and the study outcomes. Data were analyzed using SAS, version 9.4 for Windows.

RESULTS

From 2014 through 2019, 35,539 patients with WPW syndrome were identified in the MHS, 5291 had AF/AFL (14.9%); 19,961 were female (56.2%), the mean (SD) age was 62.9 (18.0) years, and 11,742 were aged ≥ 75 years (33.0%) (Table 1).

There were 4121 (11.6%), 322 (0.9%), and 848 (2.4%) patients with AF, AFL, and both arrhythmias, respectively. The mean (SD) number of cardiology visits was 3.9 (3.0). The mean (SD) baseline CCI score for the AF/AFL subgroup was 5.9 (3.5) vs 3.7 (2.2) for the non-AF/AFL subgroup (P < .001). The most prevalent comorbid conditions were hypertension, hyperlipidemia, chronic obstructive pulmonary disease, and diabetes (P < .001) (Figure 1).

Composite Outcomes

In the overall cohort, during a mean (SD) follow-up time of 3.4 (2.0) years comprising 119,682 total person-years, the components of the composite outcome occurred 6506 times with an incidence rate of 5.44 per 100 person-years. Ventricular tachycardia was the most common event, occurring 3281 times with an incidence rate of 2.74 per 100 person-years. SCA and VF occurred 841 and 135 times with incidence rates of 0.70 and 0.11 per 100 person-years, respectively. Death was the initial event 2249 times with an incidence rate of 1.88 per 100 person-years. Figure 2 shows the Kaplan-Meier curve of cardiac composite outcome by AF/AFL status.

The subgroup with AF/AFL comprised 17,412 total person-years and 1424 cardiac composite incidences compared with 102,270 person years and 5082 incidences in the no AF/AFL group (Table 2). Comparing AF/AFL vs no AF/AFL incidence rates were 8.18 vs 4.97 per 100 person-years, respectively (P < .001). SCA and VF occurred 233 and 38 times and respectively had incidence rates of 1.34 and 0.22 per 100 person-years in the AF/AFL group vs 0.59 and 0.09 per 100 person-years in the no AF/AFL group (P < .001). There were 549 deaths and a 3.15 per 100 person-years incidence rate in the AF/AFL group vs 1700 deaths and a 1.66 incidence rate in the no AF/AFL group (P < .001).

The HR for the composite outcome in the base model was 1.33 (95% CI, 1.26-1.42, P < .001) (Table 3). The association between AF/AFL and the composite outcome remained significant after adjusting for additional metabolic and cardiac covariates. The HRs for the metabolic and cardiac models were 1.30 (95% CI, 1.23-1.38, P < .001) and 1.11 (95% CI, 1.05-1.18, P < .001), respectively. After adjusting for the full model, the HR was 1.12 (95% CI, 1.05-1.19, P < .001).

Over the 6-year study period, there was a lower survival probability for patients with AF/AFL. In the overall cohort, during a mean (SD) follow-up time of 3.7 (1.9) years comprising 129,391 total person-years, there were 3130 (8.8%) deaths and an incidence rate of 2.42 per 100 person-years. Death occurred 786 times with a 4.09 incidence rate per 100 person-years in the AF/AFL vs 2344 deaths and a 2.13 incidence rate per 100 person-years in the no AF/AFL group (P < .001). In the non-AF/AFL subgroup, death occurred 2344 times during a mean (SD) follow-up of 3.7 (1.9) years comprising 110,151 total person-years. Figure 3 shows the Kaplan-Meier curve of mortality by AF/AFL status.

After adjusting for the base, metabolic and cardiac covariates, the HRs for mortality were 1.45 (95% CI, 1.33-1.57, P < .001), 1.40 (95% CI, 1.29-1.51, P < .001) and 1.15 (95% CI, 1.06-1.25, P = .001), respectively (Table 4). The HR after adjusting for the full model was 1.16 (95% CI, 1.07-1.26, P < .001).

DISCUSSION

In this large retrospective cohort study, patients with WPW syndrome and comorbid AF/AFL had a significantly higher association with the cardiac composite outcome and death during a 3-year follow-up period when compared with patients without AF/AFL. After adjusting for confounding variables, the AF/AFL subgroup maintained a 12% and 16% higher association with the composite outcome and mortality, respectively. There was minimal difference in confounding effects between demographic data and metabolic profiles, suggesting one may serve as a proxy for the other.

To our knowledge, this is the largest WPW syndrome cohort study evaluating cardiac outcomes and mortality to date. Although previous research has shown the relatively low and mostly anecdotal SCD incidence within this population,our results demonstrate a higher association of adverse cardiac outcomes and death in an AF/AFL subgroup.^16-18 Notably, in this study the AF/AFL cohort was older and had higher CCI scores than their counterparts (P < .001), thus inferring an inherently greater degree of morbidity and 10-year mortality risk. Our study is also unique in that the mean patient age was significantly older than previously reported (63 vs 27 years), which may suggest a longer living history of both ventricular pre-excitation and the comorbidities outlined in Figure 1.¹⁹ Given these age discrepancies, it is possible that our overall study population was still relatively low risk and that not all reported deaths were necessarily related to WPW syndrome. Despite these assumptions, when comparing the WPW syndrome subgroups, we still found the AF/AFL cohort maintained a statistically significant higher association with the 2 study outcomes, even after adjusting for the greater presence of comorbidities. This suggests that the presence of AF/AFL may still portend a worse prognosis in patients with WPW syndrome.

Although the association of AF and development of VF in patients with WPW syndrome—due to rapid conduction over the accessory pathway(s)—was first reported > 40 years ago, there has still been few large, long-term data studies exploring mortality in this cohort.^19-25 Furthermore, even though the current literature attributes the development of AF with the electrophysiologic properties of the accessory pathway, as well as intrinsic atrial architecture and muscle vulnerability, there is still equivocal consensus regarding EPT screening and ablation indications for asymptomatic patients with WPW syndrome.^26-28 Notably, Pappone and colleagues demonstrated the potential benefit of liberal ablation indications for asymptomatic patients, arguing that the intrinsic electrophysiologic properties of the accessory pathway—ie, short accessory-pathway antegrade effective refractory period, inducibility of atrioventricular reentrant tachycardia triggering AF, and multiple accessory pathway—rather than symptoms, are independent predictors of developing malignant arrhythmia.^1-5

These findings contradict those reported by Obeyesekere and colleagues, who concluded that the low SCD incidence rates in patients with WPW syndrome precluded routine invasive screening.^19,28 They argued that Pappone and colleagues used malignant arrhythmia as a surrogate marker for death, and that the positive predictive value of a short accessory-pathway antegrade effective refractory period for developing malignant arrhythmia was lower than reported (15% vs 82%, respectively) and that its negative predictive value was 100%.^1,19,28 Given these conflicting recommendations, we hope our data elucidates the higher association of adverse outcomes and support considerations for more intensive EPT indications in patients with WPW syndrome.

While our study does not report SCD incidence, it does provide robust and reliable mortality data that suggests a greater association of death within an AF/AFL subgroup. Our findings would support more liberal EPT recommendations in patients with WPW syndrome.^1-5,8,9 In this study, the SCA incidence rate was more than double the rate in the AF/AFL cohort (P < .001) and is commonly the initial presenting event in WPW syndrome.⁹ Even though the reported SCD incidence rate is low in WPW syndrome, our data demonstrated an increased association of death within the AF/AFL cohort. Physicians should consider early risk stratification and ablation to prevent potential recurrent malignant arrhythmia leading to death.^{1-5,8,9,12,19,20}

Limitations

As a retrospective study and without access to the National Death Index, we were unable to determine the exact cause or events leading to death and instead utilized all-cause mortality data. Subsequently, our observations may only demonstrate association, rather than causality, between AF/AFL and death in patients with WPW syndrome. Additionally, we could not distinguish between AF and AFL as the arrhythmia leading to death. However, since overall survivability was the outcome of interest, our adjusted HR models were still able to demonstrate the increased association of the composite outcome and death within an AF/AFL cohort.

Although a large cohort was analyzed, due to the constraints of utilizing diagnostic codes to determine study outcomes, we could not distinguish between symptomatic and asymptomatic patients, nor how they were managed prior to the outcome event. However, as recent literature demonstrates, updated predictors of malignant arrhythmia and decisions for early EPT are similar for both symptomatic and asymptomatic patients and should be driven by the intrinsic electrophysiologic properties of the accessory pathway, rather than symptomatology;thus, our inability to discern this should have negligible consequence in determining when to perform risk stratification and ablation.¹

MHS eligible patients have direct access to care; the generalizability of our data may not necessarily correspond to a community population with lower socioeconomic status (we did adjust for military sponsor rank which has been used as a proxy), reduced access to care, or uninsured individuals. However, the prevalence of WPW syndrome within our cohort was comparable to the general population, 0.4% vs 0.1%-0.3%, respectively.^13,14,19 Similarly, the incidence of AF within our population was comparable to the general population, 15% vs 16%-26%, respectively.²³ These similar data points suggest our results may apply beyond MHS patients.

CONCLUSIONS

Patients with WPW syndrome and AF/AFL have a higher association with adverse cardiac outcomes and death. Despite previously reported low SCD incidence rates in this population, our study demonstrates the increased association of mortality in an AF/AFL cohort. The limitations of utilizing all-cause mortality data necessitate further investigation into the etiology behind the deaths in our study population. Since ventricular pre-excitation can predispose patients to AF and potentially lead to malignant arrhythmia and SCD, understanding the cause of mortality will allow physicians to determine the appropriate monitoring and intervention strategies to improve outcomes in this population. Our results suggest consideration for more aggressive EPT screening and ablation recommendations in patients with WPW syndrome may be warranted.

Wolff-Parkinson-White (WPW) syndrome is characterized by the presence of ≥ 1 accessory pathways and the development of both recurrent paroxysmal atrial fibrillation (AF) and supraventricular tachycardia that can lead to further malignant arrhythmias resulting in sudden cardiac death (SCD).^1-7 Historically, incidental, ventricular pre-excitation on electrocardiogram has conferred a relatively low SCD risk in adults; however, newer WPW syndrome data suggest the endpoint may not be as benign as previously thought.⁷ The current literature has defined atrioventricular reentrant tachycardia triggering AF, rather than symptoms, as an independent risk factor for malignant arrhythmias. Still, long-term data detailing the association of AF with serious cardiac events and death in patients with WPW syndrome are still limited.^1-7

While previous guidelines for the treatment of WPW syndrome only recommended routine electrophysiology testing (EPT) with liberal catheter ablation for symptomatic individuals, the 2015 American College of Cardiology/American Heart Association/Heart Rhythm Society guidelines now suggest its potential benefit for risk stratification in the asymptomatic population.^8-12 Given the limited existing data, more long-term studies are needed to corroborate the latest EPT recommendations before routinely applying them in practice. Furthermore, since concomitant AF can lead to adverse cardiac outcomes in patients with WPW syndrome, additional data evaluating this association are also necessary. In this study, we aimed to determine the impact of atrial fibrillation and/or flutter (AF/AFL) on adverse cardiac outcomes and mortality in patients with WPW syndrome.

METHODS

This study used data from the Military Health System (MHS) Database Repository. The MHS is one of the largest health care systems in the country and includes information on about 10 million active duty and retired military service members and their families (51% male; 49% female).^13,14 Data were fully anonymized and complied in accordance with federal and state laws, including the Health Insurance Portability and Accountability Act of 1996. The Naval Medical Center Portsmouth Institutional Review Board approved this study.

Study Design

This retrospective, observational cohort study identified MHS patients with WPW syndrome from January 1, 2014, to December 31, 2019. Patients were included if they had ≥ 2 International Classification of Diseases, Ninth Revision (ICD-9) or International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes for WPW syndrome (ICD-9, 426.7; ICD-10, I45.6) on separate dates; were aged ≥ 18 years at index date; and had ≥ 1 year of continuous eligibility prior to the index date (enrollment gaps ≤ 30 days were considered continuous). Patients were then divided into 2 subgroups by the presence or absence of AF/AFL using diagnostic codes. Patients were excluded if they had evidence of an implantable cardioverter-defibrillator, permanent pacemaker or were missing age or sex data. Patients were followed from index date until the first occurrence of the outcome of interest, MHS disenrollment, or the end of the study period.

Cardiac composite outcomes comprised of sudden cardiac arrest (SCA), ventricular fibrillation (VF), ventricular tachycardia and death, as well as death specifically, were the outcomes of interest and assessed after index date using ICD-9 and ICD-10 codes. Death was defined as all-cause mortality. Time to event was calculated based on the date of the initial component from the composite outcome and date of death specifically for mortality. Those not experiencing an outcome were followed until MHS disenrollment or the end of the study period.

Various patient characteristics were assessed at index including age, sex, military sponsor (the patient’s active or retired duty member through which their dependent receives TRICARE benefits) rank and branch, geographic region, type of US Department of Defense beneficiary, and index year. Clinical characteristics were assessed over a 1-year baseline period prior to index date and included the number of cardiologist and clinical visits for WPW syndrome, Charlson Comorbidity Index (CCI) scores calculated from diagnostic codes outlined in the Quan coding method, and preindex time.¹⁵ Comorbidities were assessed at baseline and defined as having ≥ 1 ICD-9 or ICD-10 code for a corresponding condition within 1 year prior to index.

Baseline characteristics were assessed and descriptive statistics for categorical and continuous variables were presented accordingly. To assess bivariate association with exposure, χ² tests were used to compare categorical variables, while t tests were used to compare continuous variables by exposure status. Incidence proportions and rates were reported for each outcome of interest. Kaplan-Meier curves were constructed to assess the bivariate association between exposure and study outcomes. Cox proportional hazard modeling was performed to estimate the association between AF/AFL and time to each of the outcomes. Multiple models were designed to assess cardiac and metabolic covariates, in addition to baseline characteristics. This included a base model adjusted for age, sex, military sponsor rank and branch, geographic region, and duty status.

Additional models adjusted for cardiac and metabolic confounders and CCI score. A comprehensive model included the base, cardiac, and metabolic covariates. Multicollinearity between covariates was assessed. Variables with a variance inflation factor > 4 or a tolerance level < 0.1 were added to the models. Cox proportional hazard models were used to estimate the unadjusted and adjusted hazard ratios (HRs) and 95% CIs of the association between AF/AFL and the study outcomes. Data were analyzed using SAS, version 9.4 for Windows.

RESULTS

From 2014 through 2019, 35,539 patients with WPW syndrome were identified in the MHS, 5291 had AF/AFL (14.9%); 19,961 were female (56.2%), the mean (SD) age was 62.9 (18.0) years, and 11,742 were aged ≥ 75 years (33.0%) (Table 1).

There were 4121 (11.6%), 322 (0.9%), and 848 (2.4%) patients with AF, AFL, and both arrhythmias, respectively. The mean (SD) number of cardiology visits was 3.9 (3.0). The mean (SD) baseline CCI score for the AF/AFL subgroup was 5.9 (3.5) vs 3.7 (2.2) for the non-AF/AFL subgroup (P < .001). The most prevalent comorbid conditions were hypertension, hyperlipidemia, chronic obstructive pulmonary disease, and diabetes (P < .001) (Figure 1).

Composite Outcomes

In the overall cohort, during a mean (SD) follow-up time of 3.4 (2.0) years comprising 119,682 total person-years, the components of the composite outcome occurred 6506 times with an incidence rate of 5.44 per 100 person-years. Ventricular tachycardia was the most common event, occurring 3281 times with an incidence rate of 2.74 per 100 person-years. SCA and VF occurred 841 and 135 times with incidence rates of 0.70 and 0.11 per 100 person-years, respectively. Death was the initial event 2249 times with an incidence rate of 1.88 per 100 person-years. Figure 2 shows the Kaplan-Meier curve of cardiac composite outcome by AF/AFL status.

The subgroup with AF/AFL comprised 17,412 total person-years and 1424 cardiac composite incidences compared with 102,270 person years and 5082 incidences in the no AF/AFL group (Table 2). Comparing AF/AFL vs no AF/AFL incidence rates were 8.18 vs 4.97 per 100 person-years, respectively (P < .001). SCA and VF occurred 233 and 38 times and respectively had incidence rates of 1.34 and 0.22 per 100 person-years in the AF/AFL group vs 0.59 and 0.09 per 100 person-years in the no AF/AFL group (P < .001). There were 549 deaths and a 3.15 per 100 person-years incidence rate in the AF/AFL group vs 1700 deaths and a 1.66 incidence rate in the no AF/AFL group (P < .001).

The HR for the composite outcome in the base model was 1.33 (95% CI, 1.26-1.42, P < .001) (Table 3). The association between AF/AFL and the composite outcome remained significant after adjusting for additional metabolic and cardiac covariates. The HRs for the metabolic and cardiac models were 1.30 (95% CI, 1.23-1.38, P < .001) and 1.11 (95% CI, 1.05-1.18, P < .001), respectively. After adjusting for the full model, the HR was 1.12 (95% CI, 1.05-1.19, P < .001).

Over the 6-year study period, there was a lower survival probability for patients with AF/AFL. In the overall cohort, during a mean (SD) follow-up time of 3.7 (1.9) years comprising 129,391 total person-years, there were 3130 (8.8%) deaths and an incidence rate of 2.42 per 100 person-years. Death occurred 786 times with a 4.09 incidence rate per 100 person-years in the AF/AFL vs 2344 deaths and a 2.13 incidence rate per 100 person-years in the no AF/AFL group (P < .001). In the non-AF/AFL subgroup, death occurred 2344 times during a mean (SD) follow-up of 3.7 (1.9) years comprising 110,151 total person-years. Figure 3 shows the Kaplan-Meier curve of mortality by AF/AFL status.

After adjusting for the base, metabolic and cardiac covariates, the HRs for mortality were 1.45 (95% CI, 1.33-1.57, P < .001), 1.40 (95% CI, 1.29-1.51, P < .001) and 1.15 (95% CI, 1.06-1.25, P = .001), respectively (Table 4). The HR after adjusting for the full model was 1.16 (95% CI, 1.07-1.26, P < .001).

DISCUSSION

In this large retrospective cohort study, patients with WPW syndrome and comorbid AF/AFL had a significantly higher association with the cardiac composite outcome and death during a 3-year follow-up period when compared with patients without AF/AFL. After adjusting for confounding variables, the AF/AFL subgroup maintained a 12% and 16% higher association with the composite outcome and mortality, respectively. There was minimal difference in confounding effects between demographic data and metabolic profiles, suggesting one may serve as a proxy for the other.

To our knowledge, this is the largest WPW syndrome cohort study evaluating cardiac outcomes and mortality to date. Although previous research has shown the relatively low and mostly anecdotal SCD incidence within this population,our results demonstrate a higher association of adverse cardiac outcomes and death in an AF/AFL subgroup.^16-18 Notably, in this study the AF/AFL cohort was older and had higher CCI scores than their counterparts (P < .001), thus inferring an inherently greater degree of morbidity and 10-year mortality risk. Our study is also unique in that the mean patient age was significantly older than previously reported (63 vs 27 years), which may suggest a longer living history of both ventricular pre-excitation and the comorbidities outlined in Figure 1.¹⁹ Given these age discrepancies, it is possible that our overall study population was still relatively low risk and that not all reported deaths were necessarily related to WPW syndrome. Despite these assumptions, when comparing the WPW syndrome subgroups, we still found the AF/AFL cohort maintained a statistically significant higher association with the 2 study outcomes, even after adjusting for the greater presence of comorbidities. This suggests that the presence of AF/AFL may still portend a worse prognosis in patients with WPW syndrome.

Although the association of AF and development of VF in patients with WPW syndrome—due to rapid conduction over the accessory pathway(s)—was first reported > 40 years ago, there has still been few large, long-term data studies exploring mortality in this cohort.^19-25 Furthermore, even though the current literature attributes the development of AF with the electrophysiologic properties of the accessory pathway, as well as intrinsic atrial architecture and muscle vulnerability, there is still equivocal consensus regarding EPT screening and ablation indications for asymptomatic patients with WPW syndrome.^26-28 Notably, Pappone and colleagues demonstrated the potential benefit of liberal ablation indications for asymptomatic patients, arguing that the intrinsic electrophysiologic properties of the accessory pathway—ie, short accessory-pathway antegrade effective refractory period, inducibility of atrioventricular reentrant tachycardia triggering AF, and multiple accessory pathway—rather than symptoms, are independent predictors of developing malignant arrhythmia.^1-5

These findings contradict those reported by Obeyesekere and colleagues, who concluded that the low SCD incidence rates in patients with WPW syndrome precluded routine invasive screening.^19,28 They argued that Pappone and colleagues used malignant arrhythmia as a surrogate marker for death, and that the positive predictive value of a short accessory-pathway antegrade effective refractory period for developing malignant arrhythmia was lower than reported (15% vs 82%, respectively) and that its negative predictive value was 100%.^1,19,28 Given these conflicting recommendations, we hope our data elucidates the higher association of adverse outcomes and support considerations for more intensive EPT indications in patients with WPW syndrome.

While our study does not report SCD incidence, it does provide robust and reliable mortality data that suggests a greater association of death within an AF/AFL subgroup. Our findings would support more liberal EPT recommendations in patients with WPW syndrome.^1-5,8,9 In this study, the SCA incidence rate was more than double the rate in the AF/AFL cohort (P < .001) and is commonly the initial presenting event in WPW syndrome.⁹ Even though the reported SCD incidence rate is low in WPW syndrome, our data demonstrated an increased association of death within the AF/AFL cohort. Physicians should consider early risk stratification and ablation to prevent potential recurrent malignant arrhythmia leading to death.^{1-5,8,9,12,19,20}

Limitations

As a retrospective study and without access to the National Death Index, we were unable to determine the exact cause or events leading to death and instead utilized all-cause mortality data. Subsequently, our observations may only demonstrate association, rather than causality, between AF/AFL and death in patients with WPW syndrome. Additionally, we could not distinguish between AF and AFL as the arrhythmia leading to death. However, since overall survivability was the outcome of interest, our adjusted HR models were still able to demonstrate the increased association of the composite outcome and death within an AF/AFL cohort.

Although a large cohort was analyzed, due to the constraints of utilizing diagnostic codes to determine study outcomes, we could not distinguish between symptomatic and asymptomatic patients, nor how they were managed prior to the outcome event. However, as recent literature demonstrates, updated predictors of malignant arrhythmia and decisions for early EPT are similar for both symptomatic and asymptomatic patients and should be driven by the intrinsic electrophysiologic properties of the accessory pathway, rather than symptomatology;thus, our inability to discern this should have negligible consequence in determining when to perform risk stratification and ablation.¹

MHS eligible patients have direct access to care; the generalizability of our data may not necessarily correspond to a community population with lower socioeconomic status (we did adjust for military sponsor rank which has been used as a proxy), reduced access to care, or uninsured individuals. However, the prevalence of WPW syndrome within our cohort was comparable to the general population, 0.4% vs 0.1%-0.3%, respectively.^13,14,19 Similarly, the incidence of AF within our population was comparable to the general population, 15% vs 16%-26%, respectively.²³ These similar data points suggest our results may apply beyond MHS patients.

CONCLUSIONS

Patients with WPW syndrome and AF/AFL have a higher association with adverse cardiac outcomes and death. Despite previously reported low SCD incidence rates in this population, our study demonstrates the increased association of mortality in an AF/AFL cohort. The limitations of utilizing all-cause mortality data necessitate further investigation into the etiology behind the deaths in our study population. Since ventricular pre-excitation can predispose patients to AF and potentially lead to malignant arrhythmia and SCD, understanding the cause of mortality will allow physicians to determine the appropriate monitoring and intervention strategies to improve outcomes in this population. Our results suggest consideration for more aggressive EPT screening and ablation recommendations in patients with WPW syndrome may be warranted.

Wolff-Parkinson-White (WPW) syndrome is characterized by the presence of ≥ 1 accessory pathways and the development of both recurrent paroxysmal atrial fibrillation (AF) and supraventricular tachycardia that can lead to further malignant arrhythmias resulting in sudden cardiac death (SCD).^1-7 Historically, incidental, ventricular pre-excitation on electrocardiogram has conferred a relatively low SCD risk in adults; however, newer WPW syndrome data suggest the endpoint may not be as benign as previously thought.⁷ The current literature has defined atrioventricular reentrant tachycardia triggering AF, rather than symptoms, as an independent risk factor for malignant arrhythmias. Still, long-term data detailing the association of AF with serious cardiac events and death in patients with WPW syndrome are still limited.^1-7

While previous guidelines for the treatment of WPW syndrome only recommended routine electrophysiology testing (EPT) with liberal catheter ablation for symptomatic individuals, the 2015 American College of Cardiology/American Heart Association/Heart Rhythm Society guidelines now suggest its potential benefit for risk stratification in the asymptomatic population.^8-12 Given the limited existing data, more long-term studies are needed to corroborate the latest EPT recommendations before routinely applying them in practice. Furthermore, since concomitant AF can lead to adverse cardiac outcomes in patients with WPW syndrome, additional data evaluating this association are also necessary. In this study, we aimed to determine the impact of atrial fibrillation and/or flutter (AF/AFL) on adverse cardiac outcomes and mortality in patients with WPW syndrome.

METHODS

This study used data from the Military Health System (MHS) Database Repository. The MHS is one of the largest health care systems in the country and includes information on about 10 million active duty and retired military service members and their families (51% male; 49% female).^13,14 Data were fully anonymized and complied in accordance with federal and state laws, including the Health Insurance Portability and Accountability Act of 1996. The Naval Medical Center Portsmouth Institutional Review Board approved this study.

Study Design

This retrospective, observational cohort study identified MHS patients with WPW syndrome from January 1, 2014, to December 31, 2019. Patients were included if they had ≥ 2 International Classification of Diseases, Ninth Revision (ICD-9) or International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes for WPW syndrome (ICD-9, 426.7; ICD-10, I45.6) on separate dates; were aged ≥ 18 years at index date; and had ≥ 1 year of continuous eligibility prior to the index date (enrollment gaps ≤ 30 days were considered continuous). Patients were then divided into 2 subgroups by the presence or absence of AF/AFL using diagnostic codes. Patients were excluded if they had evidence of an implantable cardioverter-defibrillator, permanent pacemaker or were missing age or sex data. Patients were followed from index date until the first occurrence of the outcome of interest, MHS disenrollment, or the end of the study period.

Cardiac composite outcomes comprised of sudden cardiac arrest (SCA), ventricular fibrillation (VF), ventricular tachycardia and death, as well as death specifically, were the outcomes of interest and assessed after index date using ICD-9 and ICD-10 codes. Death was defined as all-cause mortality. Time to event was calculated based on the date of the initial component from the composite outcome and date of death specifically for mortality. Those not experiencing an outcome were followed until MHS disenrollment or the end of the study period.

Various patient characteristics were assessed at index including age, sex, military sponsor (the patient’s active or retired duty member through which their dependent receives TRICARE benefits) rank and branch, geographic region, type of US Department of Defense beneficiary, and index year. Clinical characteristics were assessed over a 1-year baseline period prior to index date and included the number of cardiologist and clinical visits for WPW syndrome, Charlson Comorbidity Index (CCI) scores calculated from diagnostic codes outlined in the Quan coding method, and preindex time.¹⁵ Comorbidities were assessed at baseline and defined as having ≥ 1 ICD-9 or ICD-10 code for a corresponding condition within 1 year prior to index.

Baseline characteristics were assessed and descriptive statistics for categorical and continuous variables were presented accordingly. To assess bivariate association with exposure, χ² tests were used to compare categorical variables, while t tests were used to compare continuous variables by exposure status. Incidence proportions and rates were reported for each outcome of interest. Kaplan-Meier curves were constructed to assess the bivariate association between exposure and study outcomes. Cox proportional hazard modeling was performed to estimate the association between AF/AFL and time to each of the outcomes. Multiple models were designed to assess cardiac and metabolic covariates, in addition to baseline characteristics. This included a base model adjusted for age, sex, military sponsor rank and branch, geographic region, and duty status.

Additional models adjusted for cardiac and metabolic confounders and CCI score. A comprehensive model included the base, cardiac, and metabolic covariates. Multicollinearity between covariates was assessed. Variables with a variance inflation factor > 4 or a tolerance level < 0.1 were added to the models. Cox proportional hazard models were used to estimate the unadjusted and adjusted hazard ratios (HRs) and 95% CIs of the association between AF/AFL and the study outcomes. Data were analyzed using SAS, version 9.4 for Windows.

RESULTS

From 2014 through 2019, 35,539 patients with WPW syndrome were identified in the MHS, 5291 had AF/AFL (14.9%); 19,961 were female (56.2%), the mean (SD) age was 62.9 (18.0) years, and 11,742 were aged ≥ 75 years (33.0%) (Table 1).

There were 4121 (11.6%), 322 (0.9%), and 848 (2.4%) patients with AF, AFL, and both arrhythmias, respectively. The mean (SD) number of cardiology visits was 3.9 (3.0). The mean (SD) baseline CCI score for the AF/AFL subgroup was 5.9 (3.5) vs 3.7 (2.2) for the non-AF/AFL subgroup (P < .001). The most prevalent comorbid conditions were hypertension, hyperlipidemia, chronic obstructive pulmonary disease, and diabetes (P < .001) (Figure 1).

Composite Outcomes

In the overall cohort, during a mean (SD) follow-up time of 3.4 (2.0) years comprising 119,682 total person-years, the components of the composite outcome occurred 6506 times with an incidence rate of 5.44 per 100 person-years. Ventricular tachycardia was the most common event, occurring 3281 times with an incidence rate of 2.74 per 100 person-years. SCA and VF occurred 841 and 135 times with incidence rates of 0.70 and 0.11 per 100 person-years, respectively. Death was the initial event 2249 times with an incidence rate of 1.88 per 100 person-years. Figure 2 shows the Kaplan-Meier curve of cardiac composite outcome by AF/AFL status.

The subgroup with AF/AFL comprised 17,412 total person-years and 1424 cardiac composite incidences compared with 102,270 person years and 5082 incidences in the no AF/AFL group (Table 2). Comparing AF/AFL vs no AF/AFL incidence rates were 8.18 vs 4.97 per 100 person-years, respectively (P < .001). SCA and VF occurred 233 and 38 times and respectively had incidence rates of 1.34 and 0.22 per 100 person-years in the AF/AFL group vs 0.59 and 0.09 per 100 person-years in the no AF/AFL group (P < .001). There were 549 deaths and a 3.15 per 100 person-years incidence rate in the AF/AFL group vs 1700 deaths and a 1.66 incidence rate in the no AF/AFL group (P < .001).

The HR for the composite outcome in the base model was 1.33 (95% CI, 1.26-1.42, P < .001) (Table 3). The association between AF/AFL and the composite outcome remained significant after adjusting for additional metabolic and cardiac covariates. The HRs for the metabolic and cardiac models were 1.30 (95% CI, 1.23-1.38, P < .001) and 1.11 (95% CI, 1.05-1.18, P < .001), respectively. After adjusting for the full model, the HR was 1.12 (95% CI, 1.05-1.19, P < .001).

Over the 6-year study period, there was a lower survival probability for patients with AF/AFL. In the overall cohort, during a mean (SD) follow-up time of 3.7 (1.9) years comprising 129,391 total person-years, there were 3130 (8.8%) deaths and an incidence rate of 2.42 per 100 person-years. Death occurred 786 times with a 4.09 incidence rate per 100 person-years in the AF/AFL vs 2344 deaths and a 2.13 incidence rate per 100 person-years in the no AF/AFL group (P < .001). In the non-AF/AFL subgroup, death occurred 2344 times during a mean (SD) follow-up of 3.7 (1.9) years comprising 110,151 total person-years. Figure 3 shows the Kaplan-Meier curve of mortality by AF/AFL status.

After adjusting for the base, metabolic and cardiac covariates, the HRs for mortality were 1.45 (95% CI, 1.33-1.57, P < .001), 1.40 (95% CI, 1.29-1.51, P < .001) and 1.15 (95% CI, 1.06-1.25, P = .001), respectively (Table 4). The HR after adjusting for the full model was 1.16 (95% CI, 1.07-1.26, P < .001).

DISCUSSION

In this large retrospective cohort study, patients with WPW syndrome and comorbid AF/AFL had a significantly higher association with the cardiac composite outcome and death during a 3-year follow-up period when compared with patients without AF/AFL. After adjusting for confounding variables, the AF/AFL subgroup maintained a 12% and 16% higher association with the composite outcome and mortality, respectively. There was minimal difference in confounding effects between demographic data and metabolic profiles, suggesting one may serve as a proxy for the other.

To our knowledge, this is the largest WPW syndrome cohort study evaluating cardiac outcomes and mortality to date. Although previous research has shown the relatively low and mostly anecdotal SCD incidence within this population,our results demonstrate a higher association of adverse cardiac outcomes and death in an AF/AFL subgroup.^16-18 Notably, in this study the AF/AFL cohort was older and had higher CCI scores than their counterparts (P < .001), thus inferring an inherently greater degree of morbidity and 10-year mortality risk. Our study is also unique in that the mean patient age was significantly older than previously reported (63 vs 27 years), which may suggest a longer living history of both ventricular pre-excitation and the comorbidities outlined in Figure 1.¹⁹ Given these age discrepancies, it is possible that our overall study population was still relatively low risk and that not all reported deaths were necessarily related to WPW syndrome. Despite these assumptions, when comparing the WPW syndrome subgroups, we still found the AF/AFL cohort maintained a statistically significant higher association with the 2 study outcomes, even after adjusting for the greater presence of comorbidities. This suggests that the presence of AF/AFL may still portend a worse prognosis in patients with WPW syndrome.

Although the association of AF and development of VF in patients with WPW syndrome—due to rapid conduction over the accessory pathway(s)—was first reported > 40 years ago, there has still been few large, long-term data studies exploring mortality in this cohort.^19-25 Furthermore, even though the current literature attributes the development of AF with the electrophysiologic properties of the accessory pathway, as well as intrinsic atrial architecture and muscle vulnerability, there is still equivocal consensus regarding EPT screening and ablation indications for asymptomatic patients with WPW syndrome.^26-28 Notably, Pappone and colleagues demonstrated the potential benefit of liberal ablation indications for asymptomatic patients, arguing that the intrinsic electrophysiologic properties of the accessory pathway—ie, short accessory-pathway antegrade effective refractory period, inducibility of atrioventricular reentrant tachycardia triggering AF, and multiple accessory pathway—rather than symptoms, are independent predictors of developing malignant arrhythmia.^1-5

These findings contradict those reported by Obeyesekere and colleagues, who concluded that the low SCD incidence rates in patients with WPW syndrome precluded routine invasive screening.^19,28 They argued that Pappone and colleagues used malignant arrhythmia as a surrogate marker for death, and that the positive predictive value of a short accessory-pathway antegrade effective refractory period for developing malignant arrhythmia was lower than reported (15% vs 82%, respectively) and that its negative predictive value was 100%.^1,19,28 Given these conflicting recommendations, we hope our data elucidates the higher association of adverse outcomes and support considerations for more intensive EPT indications in patients with WPW syndrome.

While our study does not report SCD incidence, it does provide robust and reliable mortality data that suggests a greater association of death within an AF/AFL subgroup. Our findings would support more liberal EPT recommendations in patients with WPW syndrome.^1-5,8,9 In this study, the SCA incidence rate was more than double the rate in the AF/AFL cohort (P < .001) and is commonly the initial presenting event in WPW syndrome.⁹ Even though the reported SCD incidence rate is low in WPW syndrome, our data demonstrated an increased association of death within the AF/AFL cohort. Physicians should consider early risk stratification and ablation to prevent potential recurrent malignant arrhythmia leading to death.^{1-5,8,9,12,19,20}

Limitations

As a retrospective study and without access to the National Death Index, we were unable to determine the exact cause or events leading to death and instead utilized all-cause mortality data. Subsequently, our observations may only demonstrate association, rather than causality, between AF/AFL and death in patients with WPW syndrome. Additionally, we could not distinguish between AF and AFL as the arrhythmia leading to death. However, since overall survivability was the outcome of interest, our adjusted HR models were still able to demonstrate the increased association of the composite outcome and death within an AF/AFL cohort.

Although a large cohort was analyzed, due to the constraints of utilizing diagnostic codes to determine study outcomes, we could not distinguish between symptomatic and asymptomatic patients, nor how they were managed prior to the outcome event. However, as recent literature demonstrates, updated predictors of malignant arrhythmia and decisions for early EPT are similar for both symptomatic and asymptomatic patients and should be driven by the intrinsic electrophysiologic properties of the accessory pathway, rather than symptomatology;thus, our inability to discern this should have negligible consequence in determining when to perform risk stratification and ablation.¹

MHS eligible patients have direct access to care; the generalizability of our data may not necessarily correspond to a community population with lower socioeconomic status (we did adjust for military sponsor rank which has been used as a proxy), reduced access to care, or uninsured individuals. However, the prevalence of WPW syndrome within our cohort was comparable to the general population, 0.4% vs 0.1%-0.3%, respectively.^13,14,19 Similarly, the incidence of AF within our population was comparable to the general population, 15% vs 16%-26%, respectively.²³ These similar data points suggest our results may apply beyond MHS patients.

CONCLUSIONS

Patients with WPW syndrome and AF/AFL have a higher association with adverse cardiac outcomes and death. Despite previously reported low SCD incidence rates in this population, our study demonstrates the increased association of mortality in an AF/AFL cohort. The limitations of utilizing all-cause mortality data necessitate further investigation into the etiology behind the deaths in our study population. Since ventricular pre-excitation can predispose patients to AF and potentially lead to malignant arrhythmia and SCD, understanding the cause of mortality will allow physicians to determine the appropriate monitoring and intervention strategies to improve outcomes in this population. Our results suggest consideration for more aggressive EPT screening and ablation recommendations in patients with WPW syndrome may be warranted.

The prevalence of chronic obstructive pulmonary disease (COPD) among male US veterans is higher than in the general population.¹ Veterans with COPD have higher rates of comorbidities and increased respiratory-related and all-cause health care use, including the use of long-term oxygen therapy (LTOT).^2-5 It has been well established that LTOT reduces all-cause mortality in patients with COPD and resting hypoxemic chronic respiratory failure (CRF) when used for ≥ 15 hours per day.^6-8

Delivery of domiciliary LTOT entails placing a nasal cannula into both nostrils and loosely securing it around both ears throughout the wake-sleep cycle. Several veterans with hypoxemic CRF due to COPD at the Jesse Brown Veterans Affairs Medical Center (JBVAMC) in Chicago, Illinois, who were receiving LTOT reported nasal cannula dislodgement (NCD) while they slept. However, the clinical significance and impact of these repeated episodes on respiratory-related health care utilization, such as frequent COPD exacerbations with hospitalization, were not recognized. Moreover, we found no published reports or clinical practice guidelines alluding to similar events reported by patients with hypoxemic CRF due to COPD receiving LTOT either at home or in an acute care setting.^8,9 Nonetheless, frequent COPD exacerbations are associated with increased hospital admissions and account for a large portion of health care costs attributed to COPD.^10-13

The purpose of this study was to determine whether veterans with hypoxemic CRF due to COPD and receiving 24-hour LTOT at JBVAMC were experiencing NCD during sleep and, if so, its impact on their hospitalizations for COPD exacerbations.

METHODS

We reviewed electronic health records (EHRs) of veterans with hypoxemic CRF from COPD who received 24-hour LTOT administered through nasal cannula and were followed in the JBVAMC pulmonary outpatient clinic between February 1, 2022, and December 31, 2022. In each case, LTOT was prescribed by a board-certified pulmonologist based on Veterans Health Administration clinical practice guidelines.¹⁴ A licensed durable medical equipment company contracted by the JBVAMC delivered and established home oxygen equipment at each veteran’s residence.

Pertinent patient demographics, clinical and physiologic variables, and hospitalizations with length of JBVAMC stay for each physician-diagnosed COPD exacerbation in the preceding year from the date last seen in the clinic were abstracted from EHRs. Overall hospital cost, defined as a veteran overnight stay in either the medical intensive care unit (MICU) or a general acute medicine bed in a US Department of Veterans Affairs (VA) facility, was calculated for each hospitalization for physician-diagnosed COPD exacerbation using VA Managerial Cost Accounting System National Cost Extracts for inpatient encounters.¹⁵ We then contacted each veteran by telephone and asked whether they had experienced NCD and, if so, its weekly frequency ranging from once to nightly.

Data Analysis

Data were reported as mean (SD) where appropriate. The t test and Fisher exact test were used as indicated. P < .05 was considered statistically significant. The study protocol was determined to be exempt by the JBVAMC Institutional Review Board (Protocol #1725748).

During the study period, 75 patients with hypoxemic CRF from COPD received LTOT and were followed at the JBVAMC. No patients were hospitalized outside the JBVAMC for COPD exacerbation during this time frame. We also found no documentation in the EHRs indicating that the clinicians managing these patients at the JBVAMC inquired about NCD during sleep.

Of the 75 patients, 66 (88%) responded to the telephone survey and 22 patients (33%) reported weekly episodes of NCD while they slept (median, 4 dislodgments per week). (Table 1). Eight patients (36%) reported nightly NCDs (Figure). All 66 respondents were male and 14 of 22 in the NCD group as well as 21 of 44 in the no NCD group were Black veterans. The mean age was similar in both groups: 71 years in the NCD group and 72 years in the no NCD group. There were no statistically significant differences in demographics, including prevalence of obstructive sleep apnea (OSA), supplemental oxygen flow rate, and duration of LTOT, or in pulmonary function test results between patients who did and did not experience NCD while sleeping (Table 2).

Ten of 22 patients (45%) with NCD and 9 of 44 patients (20%) without NCD were hospitalized at the JBVAMC for ≥ 1 COPD exacerbation in the preceding year that was diagnosed by a physician (P = .045). Three of 22 patients (14%) with NCD and no patients in the no NCD group were admitted to the MICU. No patients required intubation and mechanical ventilation during hospitalization, and no patients died. Overall hospital costs were 25% ($64,342) higher in NCD group compared with the no NCD group and were attributed to the MICU admissions in the NCD group (Table 3). Nine veterans did not respond to repeated telephone calls. One physician-diagnosed COPD exacerbation requiring hospitalization was documented in the nonresponder group; the patient was hospitalized for 2 days. One veteran died before being contacted.

DISCUSSION

There are 3 new findings in this study. First, health care practitioners at JBVAMC did not document the presence of NCD during sleep in patients with hypoxemic CRF due to COPD and receiving LTOT. Second, one-third of these patients reported frequent NCD during sleep when interviewed. Third, the nocturnal events were associated with a higher hospitalization rate for physician-diagnosed COPD exacerbation and higher overall hospital costs. These findings are unlikely to be explained by differences in COPD severity and/or known triggers that lead to COPD exacerbation and require hospitalization because baseline physiologic and LTOT parameters were similar in both groups. Conceivably, patients with untreated OSA could be restless while asleep, leading to NCD. However, this explanation seems unlikely because the frequency of OSA was similar in both groups.

Nocturnal arterial oxygen desaturation in patients with COPD without evidence of OSA may contribute to the frequency of exacerbations.¹⁶ Although the mechanism(s) underlying this phenomenon is uncertain, we posit that prolonged nocturnal airway wall hypoxia could amplify underlying chronic inflammation through local generation of reactive oxygen species, thereby predisposing patients to exacerbations. Frequent COPD exacerbations promote disease progression and health status decline and are associated with increased mortality.^11,13 Moreover, hospitalization of patients with COPD is the largest contributor to the annual direct cost of COPD per patient.^10,12 The higher hospitalization rate observed in the NCD group in our study suggests that interruption of supplemental oxygen delivery while asleep may be a risk factor for COPD exacerbation. Alternatively, an independent factor or factors may have contributed to both NCD during sleep and COPD exacerbation in these patients or an impending exacerbation resulted in sleep disturbances that led to NCD. Additional research is warranted on veterans with hypoxemic CRF from COPD who are receiving LTOT and report frequent NCD during sleep that may support or refute these hypotheses.

To the best of our knowledge, NCD during sleep has not been previously reported in patients with hypoxemic CRF due to COPD who are receiving LTOT at home or in an acute care setting.^17-20 Several layperson proposals to secure nasal cannulas to the face while sleeping are posted online. These include wearing a commercially available headband with 2 Velcro loops that fix the cannula tube, using fabric medical tape on both cheeks, and wearing a sleep mask. Conceivably, the efficacy and safety of these inexpensive interventions to mitigate NCD during sleep in patients receiving LTOT with hypoxemic CRF from COPD could be tested in clinical trials.

Limitations

This was a small, single-site study, comprised entirely of male patients who are predominantly Black veterans. The telephone interviews with veterans self-reporting NCD during their sleep are prone to recall bias. In addition, the validity and reproducibility of NCD during sleep were not addressed in this study. Missing data from 9 nonresponders may have introduced a nonresponse bias in data analysis and interpretation. The overall hospital cost for a COPD exacerbation at JBVAMC was derived from VA data; US Centers for Medicare & Medicaid Services or commercial carrier data may be different.^15,21 Lastly, access to LTOT for veterans with hypoxemic CRF from COPD is regulated and supervised at VA medical facilities.¹⁴ This process may be different for patients outside the VA. Taken together, it is difficult to generalize our initial observations to non-VA patients with hypoxemic CRF from COPD who are receiving LTOT. We suggest a large, prospective study of veterans be conducted to determine the prevalence of NCD during sleep and its relationship with COPD exacerbations in veterans receiving LTOT with hypoxemic CRF due to COPD.

CONCLUSIONS

Clinicians at the JBVAMC did not document the presence of NCD during sleep in patients with hypoxemic CRF from COPD who received LTOT. However, self-reported, weekly nocturnal NCD episodes were associated with a higher hospitalization rate for COPD exacerbation and higher hospital costs. Accordingly, user-friendly devices to mitigate NCD during sleep should be developed.

Acknowledgments

We thank Yolanda Davis, RRT, and George Adam for their assistance with this project.

The prevalence of chronic obstructive pulmonary disease (COPD) among male US veterans is higher than in the general population.¹ Veterans with COPD have higher rates of comorbidities and increased respiratory-related and all-cause health care use, including the use of long-term oxygen therapy (LTOT).^2-5 It has been well established that LTOT reduces all-cause mortality in patients with COPD and resting hypoxemic chronic respiratory failure (CRF) when used for ≥ 15 hours per day.^6-8

Delivery of domiciliary LTOT entails placing a nasal cannula into both nostrils and loosely securing it around both ears throughout the wake-sleep cycle. Several veterans with hypoxemic CRF due to COPD at the Jesse Brown Veterans Affairs Medical Center (JBVAMC) in Chicago, Illinois, who were receiving LTOT reported nasal cannula dislodgement (NCD) while they slept. However, the clinical significance and impact of these repeated episodes on respiratory-related health care utilization, such as frequent COPD exacerbations with hospitalization, were not recognized. Moreover, we found no published reports or clinical practice guidelines alluding to similar events reported by patients with hypoxemic CRF due to COPD receiving LTOT either at home or in an acute care setting.^8,9 Nonetheless, frequent COPD exacerbations are associated with increased hospital admissions and account for a large portion of health care costs attributed to COPD.^10-13

The purpose of this study was to determine whether veterans with hypoxemic CRF due to COPD and receiving 24-hour LTOT at JBVAMC were experiencing NCD during sleep and, if so, its impact on their hospitalizations for COPD exacerbations.

METHODS

We reviewed electronic health records (EHRs) of veterans with hypoxemic CRF from COPD who received 24-hour LTOT administered through nasal cannula and were followed in the JBVAMC pulmonary outpatient clinic between February 1, 2022, and December 31, 2022. In each case, LTOT was prescribed by a board-certified pulmonologist based on Veterans Health Administration clinical practice guidelines.¹⁴ A licensed durable medical equipment company contracted by the JBVAMC delivered and established home oxygen equipment at each veteran’s residence.

Pertinent patient demographics, clinical and physiologic variables, and hospitalizations with length of JBVAMC stay for each physician-diagnosed COPD exacerbation in the preceding year from the date last seen in the clinic were abstracted from EHRs. Overall hospital cost, defined as a veteran overnight stay in either the medical intensive care unit (MICU) or a general acute medicine bed in a US Department of Veterans Affairs (VA) facility, was calculated for each hospitalization for physician-diagnosed COPD exacerbation using VA Managerial Cost Accounting System National Cost Extracts for inpatient encounters.¹⁵ We then contacted each veteran by telephone and asked whether they had experienced NCD and, if so, its weekly frequency ranging from once to nightly.

Data Analysis

Data were reported as mean (SD) where appropriate. The t test and Fisher exact test were used as indicated. P < .05 was considered statistically significant. The study protocol was determined to be exempt by the JBVAMC Institutional Review Board (Protocol #1725748).

During the study period, 75 patients with hypoxemic CRF from COPD received LTOT and were followed at the JBVAMC. No patients were hospitalized outside the JBVAMC for COPD exacerbation during this time frame. We also found no documentation in the EHRs indicating that the clinicians managing these patients at the JBVAMC inquired about NCD during sleep.

Of the 75 patients, 66 (88%) responded to the telephone survey and 22 patients (33%) reported weekly episodes of NCD while they slept (median, 4 dislodgments per week). (Table 1). Eight patients (36%) reported nightly NCDs (Figure). All 66 respondents were male and 14 of 22 in the NCD group as well as 21 of 44 in the no NCD group were Black veterans. The mean age was similar in both groups: 71 years in the NCD group and 72 years in the no NCD group. There were no statistically significant differences in demographics, including prevalence of obstructive sleep apnea (OSA), supplemental oxygen flow rate, and duration of LTOT, or in pulmonary function test results between patients who did and did not experience NCD while sleeping (Table 2).

Ten of 22 patients (45%) with NCD and 9 of 44 patients (20%) without NCD were hospitalized at the JBVAMC for ≥ 1 COPD exacerbation in the preceding year that was diagnosed by a physician (P = .045). Three of 22 patients (14%) with NCD and no patients in the no NCD group were admitted to the MICU. No patients required intubation and mechanical ventilation during hospitalization, and no patients died. Overall hospital costs were 25% ($64,342) higher in NCD group compared with the no NCD group and were attributed to the MICU admissions in the NCD group (Table 3). Nine veterans did not respond to repeated telephone calls. One physician-diagnosed COPD exacerbation requiring hospitalization was documented in the nonresponder group; the patient was hospitalized for 2 days. One veteran died before being contacted.

DISCUSSION

There are 3 new findings in this study. First, health care practitioners at JBVAMC did not document the presence of NCD during sleep in patients with hypoxemic CRF due to COPD and receiving LTOT. Second, one-third of these patients reported frequent NCD during sleep when interviewed. Third, the nocturnal events were associated with a higher hospitalization rate for physician-diagnosed COPD exacerbation and higher overall hospital costs. These findings are unlikely to be explained by differences in COPD severity and/or known triggers that lead to COPD exacerbation and require hospitalization because baseline physiologic and LTOT parameters were similar in both groups. Conceivably, patients with untreated OSA could be restless while asleep, leading to NCD. However, this explanation seems unlikely because the frequency of OSA was similar in both groups.

Nocturnal arterial oxygen desaturation in patients with COPD without evidence of OSA may contribute to the frequency of exacerbations.¹⁶ Although the mechanism(s) underlying this phenomenon is uncertain, we posit that prolonged nocturnal airway wall hypoxia could amplify underlying chronic inflammation through local generation of reactive oxygen species, thereby predisposing patients to exacerbations. Frequent COPD exacerbations promote disease progression and health status decline and are associated with increased mortality.^11,13 Moreover, hospitalization of patients with COPD is the largest contributor to the annual direct cost of COPD per patient.^10,12 The higher hospitalization rate observed in the NCD group in our study suggests that interruption of supplemental oxygen delivery while asleep may be a risk factor for COPD exacerbation. Alternatively, an independent factor or factors may have contributed to both NCD during sleep and COPD exacerbation in these patients or an impending exacerbation resulted in sleep disturbances that led to NCD. Additional research is warranted on veterans with hypoxemic CRF from COPD who are receiving LTOT and report frequent NCD during sleep that may support or refute these hypotheses.

To the best of our knowledge, NCD during sleep has not been previously reported in patients with hypoxemic CRF due to COPD who are receiving LTOT at home or in an acute care setting.^17-20 Several layperson proposals to secure nasal cannulas to the face while sleeping are posted online. These include wearing a commercially available headband with 2 Velcro loops that fix the cannula tube, using fabric medical tape on both cheeks, and wearing a sleep mask. Conceivably, the efficacy and safety of these inexpensive interventions to mitigate NCD during sleep in patients receiving LTOT with hypoxemic CRF from COPD could be tested in clinical trials.

Limitations

This was a small, single-site study, comprised entirely of male patients who are predominantly Black veterans. The telephone interviews with veterans self-reporting NCD during their sleep are prone to recall bias. In addition, the validity and reproducibility of NCD during sleep were not addressed in this study. Missing data from 9 nonresponders may have introduced a nonresponse bias in data analysis and interpretation. The overall hospital cost for a COPD exacerbation at JBVAMC was derived from VA data; US Centers for Medicare & Medicaid Services or commercial carrier data may be different.^15,21 Lastly, access to LTOT for veterans with hypoxemic CRF from COPD is regulated and supervised at VA medical facilities.¹⁴ This process may be different for patients outside the VA. Taken together, it is difficult to generalize our initial observations to non-VA patients with hypoxemic CRF from COPD who are receiving LTOT. We suggest a large, prospective study of veterans be conducted to determine the prevalence of NCD during sleep and its relationship with COPD exacerbations in veterans receiving LTOT with hypoxemic CRF due to COPD.

CONCLUSIONS

Clinicians at the JBVAMC did not document the presence of NCD during sleep in patients with hypoxemic CRF from COPD who received LTOT. However, self-reported, weekly nocturnal NCD episodes were associated with a higher hospitalization rate for COPD exacerbation and higher hospital costs. Accordingly, user-friendly devices to mitigate NCD during sleep should be developed.

Acknowledgments

We thank Yolanda Davis, RRT, and George Adam for their assistance with this project.

The prevalence of chronic obstructive pulmonary disease (COPD) among male US veterans is higher than in the general population.¹ Veterans with COPD have higher rates of comorbidities and increased respiratory-related and all-cause health care use, including the use of long-term oxygen therapy (LTOT).^2-5 It has been well established that LTOT reduces all-cause mortality in patients with COPD and resting hypoxemic chronic respiratory failure (CRF) when used for ≥ 15 hours per day.^6-8

Delivery of domiciliary LTOT entails placing a nasal cannula into both nostrils and loosely securing it around both ears throughout the wake-sleep cycle. Several veterans with hypoxemic CRF due to COPD at the Jesse Brown Veterans Affairs Medical Center (JBVAMC) in Chicago, Illinois, who were receiving LTOT reported nasal cannula dislodgement (NCD) while they slept. However, the clinical significance and impact of these repeated episodes on respiratory-related health care utilization, such as frequent COPD exacerbations with hospitalization, were not recognized. Moreover, we found no published reports or clinical practice guidelines alluding to similar events reported by patients with hypoxemic CRF due to COPD receiving LTOT either at home or in an acute care setting.^8,9 Nonetheless, frequent COPD exacerbations are associated with increased hospital admissions and account for a large portion of health care costs attributed to COPD.^10-13

The purpose of this study was to determine whether veterans with hypoxemic CRF due to COPD and receiving 24-hour LTOT at JBVAMC were experiencing NCD during sleep and, if so, its impact on their hospitalizations for COPD exacerbations.

METHODS

We reviewed electronic health records (EHRs) of veterans with hypoxemic CRF from COPD who received 24-hour LTOT administered through nasal cannula and were followed in the JBVAMC pulmonary outpatient clinic between February 1, 2022, and December 31, 2022. In each case, LTOT was prescribed by a board-certified pulmonologist based on Veterans Health Administration clinical practice guidelines.¹⁴ A licensed durable medical equipment company contracted by the JBVAMC delivered and established home oxygen equipment at each veteran’s residence.

Pertinent patient demographics, clinical and physiologic variables, and hospitalizations with length of JBVAMC stay for each physician-diagnosed COPD exacerbation in the preceding year from the date last seen in the clinic were abstracted from EHRs. Overall hospital cost, defined as a veteran overnight stay in either the medical intensive care unit (MICU) or a general acute medicine bed in a US Department of Veterans Affairs (VA) facility, was calculated for each hospitalization for physician-diagnosed COPD exacerbation using VA Managerial Cost Accounting System National Cost Extracts for inpatient encounters.¹⁵ We then contacted each veteran by telephone and asked whether they had experienced NCD and, if so, its weekly frequency ranging from once to nightly.

Data Analysis

Data were reported as mean (SD) where appropriate. The t test and Fisher exact test were used as indicated. P < .05 was considered statistically significant. The study protocol was determined to be exempt by the JBVAMC Institutional Review Board (Protocol #1725748).

During the study period, 75 patients with hypoxemic CRF from COPD received LTOT and were followed at the JBVAMC. No patients were hospitalized outside the JBVAMC for COPD exacerbation during this time frame. We also found no documentation in the EHRs indicating that the clinicians managing these patients at the JBVAMC inquired about NCD during sleep.

Of the 75 patients, 66 (88%) responded to the telephone survey and 22 patients (33%) reported weekly episodes of NCD while they slept (median, 4 dislodgments per week). (Table 1). Eight patients (36%) reported nightly NCDs (Figure). All 66 respondents were male and 14 of 22 in the NCD group as well as 21 of 44 in the no NCD group were Black veterans. The mean age was similar in both groups: 71 years in the NCD group and 72 years in the no NCD group. There were no statistically significant differences in demographics, including prevalence of obstructive sleep apnea (OSA), supplemental oxygen flow rate, and duration of LTOT, or in pulmonary function test results between patients who did and did not experience NCD while sleeping (Table 2).

Ten of 22 patients (45%) with NCD and 9 of 44 patients (20%) without NCD were hospitalized at the JBVAMC for ≥ 1 COPD exacerbation in the preceding year that was diagnosed by a physician (P = .045). Three of 22 patients (14%) with NCD and no patients in the no NCD group were admitted to the MICU. No patients required intubation and mechanical ventilation during hospitalization, and no patients died. Overall hospital costs were 25% ($64,342) higher in NCD group compared with the no NCD group and were attributed to the MICU admissions in the NCD group (Table 3). Nine veterans did not respond to repeated telephone calls. One physician-diagnosed COPD exacerbation requiring hospitalization was documented in the nonresponder group; the patient was hospitalized for 2 days. One veteran died before being contacted.

DISCUSSION

There are 3 new findings in this study. First, health care practitioners at JBVAMC did not document the presence of NCD during sleep in patients with hypoxemic CRF due to COPD and receiving LTOT. Second, one-third of these patients reported frequent NCD during sleep when interviewed. Third, the nocturnal events were associated with a higher hospitalization rate for physician-diagnosed COPD exacerbation and higher overall hospital costs. These findings are unlikely to be explained by differences in COPD severity and/or known triggers that lead to COPD exacerbation and require hospitalization because baseline physiologic and LTOT parameters were similar in both groups. Conceivably, patients with untreated OSA could be restless while asleep, leading to NCD. However, this explanation seems unlikely because the frequency of OSA was similar in both groups.

Nocturnal arterial oxygen desaturation in patients with COPD without evidence of OSA may contribute to the frequency of exacerbations.¹⁶ Although the mechanism(s) underlying this phenomenon is uncertain, we posit that prolonged nocturnal airway wall hypoxia could amplify underlying chronic inflammation through local generation of reactive oxygen species, thereby predisposing patients to exacerbations. Frequent COPD exacerbations promote disease progression and health status decline and are associated with increased mortality.^11,13 Moreover, hospitalization of patients with COPD is the largest contributor to the annual direct cost of COPD per patient.^10,12 The higher hospitalization rate observed in the NCD group in our study suggests that interruption of supplemental oxygen delivery while asleep may be a risk factor for COPD exacerbation. Alternatively, an independent factor or factors may have contributed to both NCD during sleep and COPD exacerbation in these patients or an impending exacerbation resulted in sleep disturbances that led to NCD. Additional research is warranted on veterans with hypoxemic CRF from COPD who are receiving LTOT and report frequent NCD during sleep that may support or refute these hypotheses.

To the best of our knowledge, NCD during sleep has not been previously reported in patients with hypoxemic CRF due to COPD who are receiving LTOT at home or in an acute care setting.^17-20 Several layperson proposals to secure nasal cannulas to the face while sleeping are posted online. These include wearing a commercially available headband with 2 Velcro loops that fix the cannula tube, using fabric medical tape on both cheeks, and wearing a sleep mask. Conceivably, the efficacy and safety of these inexpensive interventions to mitigate NCD during sleep in patients receiving LTOT with hypoxemic CRF from COPD could be tested in clinical trials.

Limitations

This was a small, single-site study, comprised entirely of male patients who are predominantly Black veterans. The telephone interviews with veterans self-reporting NCD during their sleep are prone to recall bias. In addition, the validity and reproducibility of NCD during sleep were not addressed in this study. Missing data from 9 nonresponders may have introduced a nonresponse bias in data analysis and interpretation. The overall hospital cost for a COPD exacerbation at JBVAMC was derived from VA data; US Centers for Medicare & Medicaid Services or commercial carrier data may be different.^15,21 Lastly, access to LTOT for veterans with hypoxemic CRF from COPD is regulated and supervised at VA medical facilities.¹⁴ This process may be different for patients outside the VA. Taken together, it is difficult to generalize our initial observations to non-VA patients with hypoxemic CRF from COPD who are receiving LTOT. We suggest a large, prospective study of veterans be conducted to determine the prevalence of NCD during sleep and its relationship with COPD exacerbations in veterans receiving LTOT with hypoxemic CRF due to COPD.

CONCLUSIONS

Clinicians at the JBVAMC did not document the presence of NCD during sleep in patients with hypoxemic CRF from COPD who received LTOT. However, self-reported, weekly nocturnal NCD episodes were associated with a higher hospitalization rate for COPD exacerbation and higher hospital costs. Accordingly, user-friendly devices to mitigate NCD during sleep should be developed.

Acknowledgments

We thank Yolanda Davis, RRT, and George Adam for their assistance with this project.

Longitudinal melanonychia (LM) is a pigmented linear band—brown, black, or gray—spanning the length of the nail plate due to the presence of excess melanin, which may be attributed to a benign or malignant process and may warrant further investigation.^1,2 The majority of patients who present with LM are diagnosed with melanocytic activation of the nail matrix due to their inherent darker skin tone or various triggers including trauma, infection, and medications. Longitudinal melanonychia secondary to melanocytic activation often occurs spontaneously in patients with skin of color.³ Less commonly, LM is caused by a nail matrix nevus or lentigo; however, LM may arise secondary to subungual melanoma, a more dangerous cause.

A thorough clinical history including duration, recent changes in LM manifestation, nail trauma, or infection is helpful in evaluating patients with LM; however, a history of nail trauma can be misleading, as nail changes attributed to the trauma may in fact be melanoma. Irregularly spaced vertical lines of pigmentation ranging from brown to black with variations in spacing and width are characteristic of subungual melanoma.⁴ Nail dystrophy, granular hyperpigmentation, and Hutchinson sign (extension of pigmentation to the nail folds) also are worrisome features.⁵ In recent years, dermoscopy has become an important tool in the clinical examination of LM, with the development of criteria based on color and pattern recognition.^5,6 Dermoscopy can be useful in screening potential candidates for biopsy. Although clinical examination and dermoscopy are essential to evaluating LM, the gold-standard diagnostic test when malignancy is suspected is a nail matrix biopsy.^1,2,6,7

Epidemiology

It is not unusual for patients with darker skin tones to develop LM due to melanocytic activation of multiple nails with age. This finding can be seen in approximately 80% of African American individuals, 30% of Japanese individuals, and 50% of Hispanic individuals.² It has even been reported that approximately 100% of Black patients older than 50 years will have evidence of LM.³

In a retrospective analysis, children presenting with LM tend to have a higher prevalence of nail matrix nevi compared to adults (56.1% [60/106] vs 34.3% [23/66]; P =.005).⁸ Involvement of a single digit in children is most likely indicative of a nevus; however, when an adult presents with LM in a single digit, suspicion for subungual melanoma should be raised.^2,3,9

Two separate single-center retrospective studies showed the prevalence of subungual melanoma in patients presenting with melanonychia in Asia. Jin et al¹⁰ reported subungual melanoma in 6.2% (17/275) of Korean patients presenting with melanonychia at a general dermatology clinic from 2002 to 2014. Lyu et al⁸ studied LM in 172 Chinese patients in a dermatology clinic from 2018 to 2021 and reported 9% (6/66) of adults (aged ≥ 18 years) with subungual melanoma, with no reported cases in childhood (aged < 18 years).

Although the prevalence of subungual melanoma in patients with LM is low, it is an important diagnosis that should not be missed. In confirmed cases of subungual melanoma, two-thirds of lesions manifested as LM.^3,10,11 Thus, LM arising in an adult in a single digit is more concerning for malignancy.^2,3,7,9

Individuals of African and Asian descent as well as American Indian individuals are at highest risk for subungual melanoma with a poor prognosis compared to other types of melanoma, largely due to diagnosis at an advanced stage of disease.^3,9 In a retrospective study of 25 patients with surgically treated subungual melanoma, the mean recurrence-free survival was 33.6 months. The recurrence-free survival was 66% at 1 year and 40% at 3 years, and the overall survival rate was 37% at 3 years.¹²

Key clinical features in individuals with darker skin tones

• In patients with darker skin tones, LM tends to occur on multiple nails as a result of melanocytic activation.^2,13

• Several longitudinal bands may be noted on the same nail and the pigmentation of the bands may vary. With age, these longitudinal bands typically increase in number and width.¹³

• Pseudo-Hutchinson sign may be present due to ethnic melanosis of the proximal nail fold.^13,14

• Dermoscopic findings of LM in patients with skin of color include wider bands (P = .0125), lower band brightness (P < .032), and higher frequency of changing appearance of bands (P = .0071).¹⁵

Worth noting

When patients present with LM, thorough examination of the nail plate, periungual skin, and distal pulp of all digits on all extremities with adequate lighting is important.² Dermoscopy is useful, and a gel interface helps for examining the nail plates.⁷

Clinicians should be encouraged to biopsy or immediately refer patients with concerning nail unit lesions. Cases of LM most likely are benign, but if some doubt exists, the lesions should be biopsied or tracked closely with clinical and dermoscopic images, with a biopsy if changes occur.¹⁶ In conjunction with evaluation by a qualified clinician, patients also should be encouraged to take photographs, as the evolution of nail changes is a critical part of clinical decision-making on the need for a biopsy or referral.

Health disparity highlight

Despite the disproportionately high mortality rates from subungual melanoma in Black and Hispanic populations,^3,9 studies often do not adequately represent these populations. Although subungual melanoma is rare, a delay in the diagnosis contributes to high morbidity and mortality rates.

Longitudinal melanonychia (LM) is a pigmented linear band—brown, black, or gray—spanning the length of the nail plate due to the presence of excess melanin, which may be attributed to a benign or malignant process and may warrant further investigation.^1,2 The majority of patients who present with LM are diagnosed with melanocytic activation of the nail matrix due to their inherent darker skin tone or various triggers including trauma, infection, and medications. Longitudinal melanonychia secondary to melanocytic activation often occurs spontaneously in patients with skin of color.³ Less commonly, LM is caused by a nail matrix nevus or lentigo; however, LM may arise secondary to subungual melanoma, a more dangerous cause.

A thorough clinical history including duration, recent changes in LM manifestation, nail trauma, or infection is helpful in evaluating patients with LM; however, a history of nail trauma can be misleading, as nail changes attributed to the trauma may in fact be melanoma. Irregularly spaced vertical lines of pigmentation ranging from brown to black with variations in spacing and width are characteristic of subungual melanoma.⁴ Nail dystrophy, granular hyperpigmentation, and Hutchinson sign (extension of pigmentation to the nail folds) also are worrisome features.⁵ In recent years, dermoscopy has become an important tool in the clinical examination of LM, with the development of criteria based on color and pattern recognition.^5,6 Dermoscopy can be useful in screening potential candidates for biopsy. Although clinical examination and dermoscopy are essential to evaluating LM, the gold-standard diagnostic test when malignancy is suspected is a nail matrix biopsy.^1,2,6,7

Epidemiology

It is not unusual for patients with darker skin tones to develop LM due to melanocytic activation of multiple nails with age. This finding can be seen in approximately 80% of African American individuals, 30% of Japanese individuals, and 50% of Hispanic individuals.² It has even been reported that approximately 100% of Black patients older than 50 years will have evidence of LM.³

In a retrospective analysis, children presenting with LM tend to have a higher prevalence of nail matrix nevi compared to adults (56.1% [60/106] vs 34.3% [23/66]; P =.005).⁸ Involvement of a single digit in children is most likely indicative of a nevus; however, when an adult presents with LM in a single digit, suspicion for subungual melanoma should be raised.^2,3,9

Two separate single-center retrospective studies showed the prevalence of subungual melanoma in patients presenting with melanonychia in Asia. Jin et al¹⁰ reported subungual melanoma in 6.2% (17/275) of Korean patients presenting with melanonychia at a general dermatology clinic from 2002 to 2014. Lyu et al⁸ studied LM in 172 Chinese patients in a dermatology clinic from 2018 to 2021 and reported 9% (6/66) of adults (aged ≥ 18 years) with subungual melanoma, with no reported cases in childhood (aged < 18 years).

Although the prevalence of subungual melanoma in patients with LM is low, it is an important diagnosis that should not be missed. In confirmed cases of subungual melanoma, two-thirds of lesions manifested as LM.^3,10,11 Thus, LM arising in an adult in a single digit is more concerning for malignancy.^2,3,7,9

Individuals of African and Asian descent as well as American Indian individuals are at highest risk for subungual melanoma with a poor prognosis compared to other types of melanoma, largely due to diagnosis at an advanced stage of disease.^3,9 In a retrospective study of 25 patients with surgically treated subungual melanoma, the mean recurrence-free survival was 33.6 months. The recurrence-free survival was 66% at 1 year and 40% at 3 years, and the overall survival rate was 37% at 3 years.¹²

Key clinical features in individuals with darker skin tones

• In patients with darker skin tones, LM tends to occur on multiple nails as a result of melanocytic activation.^2,13

• Several longitudinal bands may be noted on the same nail and the pigmentation of the bands may vary. With age, these longitudinal bands typically increase in number and width.¹³

• Pseudo-Hutchinson sign may be present due to ethnic melanosis of the proximal nail fold.^13,14

• Dermoscopic findings of LM in patients with skin of color include wider bands (P = .0125), lower band brightness (P < .032), and higher frequency of changing appearance of bands (P = .0071).¹⁵

Worth noting

When patients present with LM, thorough examination of the nail plate, periungual skin, and distal pulp of all digits on all extremities with adequate lighting is important.² Dermoscopy is useful, and a gel interface helps for examining the nail plates.⁷

Clinicians should be encouraged to biopsy or immediately refer patients with concerning nail unit lesions. Cases of LM most likely are benign, but if some doubt exists, the lesions should be biopsied or tracked closely with clinical and dermoscopic images, with a biopsy if changes occur.¹⁶ In conjunction with evaluation by a qualified clinician, patients also should be encouraged to take photographs, as the evolution of nail changes is a critical part of clinical decision-making on the need for a biopsy or referral.

Health disparity highlight

Despite the disproportionately high mortality rates from subungual melanoma in Black and Hispanic populations,^3,9 studies often do not adequately represent these populations. Although subungual melanoma is rare, a delay in the diagnosis contributes to high morbidity and mortality rates.

Longitudinal melanonychia (LM) is a pigmented linear band—brown, black, or gray—spanning the length of the nail plate due to the presence of excess melanin, which may be attributed to a benign or malignant process and may warrant further investigation.^1,2 The majority of patients who present with LM are diagnosed with melanocytic activation of the nail matrix due to their inherent darker skin tone or various triggers including trauma, infection, and medications. Longitudinal melanonychia secondary to melanocytic activation often occurs spontaneously in patients with skin of color.³ Less commonly, LM is caused by a nail matrix nevus or lentigo; however, LM may arise secondary to subungual melanoma, a more dangerous cause.

A thorough clinical history including duration, recent changes in LM manifestation, nail trauma, or infection is helpful in evaluating patients with LM; however, a history of nail trauma can be misleading, as nail changes attributed to the trauma may in fact be melanoma. Irregularly spaced vertical lines of pigmentation ranging from brown to black with variations in spacing and width are characteristic of subungual melanoma.⁴ Nail dystrophy, granular hyperpigmentation, and Hutchinson sign (extension of pigmentation to the nail folds) also are worrisome features.⁵ In recent years, dermoscopy has become an important tool in the clinical examination of LM, with the development of criteria based on color and pattern recognition.^5,6 Dermoscopy can be useful in screening potential candidates for biopsy. Although clinical examination and dermoscopy are essential to evaluating LM, the gold-standard diagnostic test when malignancy is suspected is a nail matrix biopsy.^1,2,6,7

Epidemiology

It is not unusual for patients with darker skin tones to develop LM due to melanocytic activation of multiple nails with age. This finding can be seen in approximately 80% of African American individuals, 30% of Japanese individuals, and 50% of Hispanic individuals.² It has even been reported that approximately 100% of Black patients older than 50 years will have evidence of LM.³

In a retrospective analysis, children presenting with LM tend to have a higher prevalence of nail matrix nevi compared to adults (56.1% [60/106] vs 34.3% [23/66]; P =.005).⁸ Involvement of a single digit in children is most likely indicative of a nevus; however, when an adult presents with LM in a single digit, suspicion for subungual melanoma should be raised.^2,3,9

Two separate single-center retrospective studies showed the prevalence of subungual melanoma in patients presenting with melanonychia in Asia. Jin et al¹⁰ reported subungual melanoma in 6.2% (17/275) of Korean patients presenting with melanonychia at a general dermatology clinic from 2002 to 2014. Lyu et al⁸ studied LM in 172 Chinese patients in a dermatology clinic from 2018 to 2021 and reported 9% (6/66) of adults (aged ≥ 18 years) with subungual melanoma, with no reported cases in childhood (aged < 18 years).

Although the prevalence of subungual melanoma in patients with LM is low, it is an important diagnosis that should not be missed. In confirmed cases of subungual melanoma, two-thirds of lesions manifested as LM.^3,10,11 Thus, LM arising in an adult in a single digit is more concerning for malignancy.^2,3,7,9

Individuals of African and Asian descent as well as American Indian individuals are at highest risk for subungual melanoma with a poor prognosis compared to other types of melanoma, largely due to diagnosis at an advanced stage of disease.^3,9 In a retrospective study of 25 patients with surgically treated subungual melanoma, the mean recurrence-free survival was 33.6 months. The recurrence-free survival was 66% at 1 year and 40% at 3 years, and the overall survival rate was 37% at 3 years.¹²

Key clinical features in individuals with darker skin tones

• In patients with darker skin tones, LM tends to occur on multiple nails as a result of melanocytic activation.^2,13

• Several longitudinal bands may be noted on the same nail and the pigmentation of the bands may vary. With age, these longitudinal bands typically increase in number and width.¹³

• Pseudo-Hutchinson sign may be present due to ethnic melanosis of the proximal nail fold.^13,14

• Dermoscopic findings of LM in patients with skin of color include wider bands (P = .0125), lower band brightness (P < .032), and higher frequency of changing appearance of bands (P = .0071).¹⁵

Worth noting

When patients present with LM, thorough examination of the nail plate, periungual skin, and distal pulp of all digits on all extremities with adequate lighting is important.² Dermoscopy is useful, and a gel interface helps for examining the nail plates.⁷

Clinicians should be encouraged to biopsy or immediately refer patients with concerning nail unit lesions. Cases of LM most likely are benign, but if some doubt exists, the lesions should be biopsied or tracked closely with clinical and dermoscopic images, with a biopsy if changes occur.¹⁶ In conjunction with evaluation by a qualified clinician, patients also should be encouraged to take photographs, as the evolution of nail changes is a critical part of clinical decision-making on the need for a biopsy or referral.

Health disparity highlight

Despite the disproportionately high mortality rates from subungual melanoma in Black and Hispanic populations,^3,9 studies often do not adequately represent these populations. Although subungual melanoma is rare, a delay in the diagnosis contributes to high morbidity and mortality rates.

The US Department of Veterans Affairs (VA) has partnered with academic medical centers and programs since 1946 to provide clinical training for physician residents. Ranking second in federal graduate medical education (GME) funding to the Centers for Medicare and Medicaid Services (CMS), the $850 million VA GME budget annually reimburses > 250 GME-sponsoring institutions (affiliates) of 8000 GME programs for the clinical training of 49,000 individual residents rotating through > 11,000 full-time equivalent (FTE) positions.¹ The VA also distributes $1.6 billion to VA facilities to offset the costs of conducting health professions education (HPE) (eg, facility infrastructure, salary support for VA instructors and preceptors, education office administration, and instructional equipment).² The VA financial and educational contributions account for payment of 11% of resident positions nationally and allow academic medical centers to be less reliant on CMS GME funding.^3,4 The VA contributions also provide opportunities for GME expansion,^1,5,6 educational innovations,^5,7 interprofessional and team-based care,^8,9 and quality and safety training.^10,11 The Table provides a comparison of CMS and VA GME reimbursability based on activity.

GME financing is complex, particularly the formulaic approach used by CMS, the details of which are often obscured in federal regulations. Due to this complexity and the $16 billion CMS GME budget, academic publications have focused on CMS GME financing while not fully explaining the VA GME policies and processes.^4,12-14 By comparison, the VA GME financing model is relatively straightforward and governed by different statues and VA regulations, yet sharing some of the same principles as CMS regulations. Given the challenges in CMS reimbursement to fully support the cost of resident education, as well as the educational opportunities at the VA, the VA designs its reimbursement model to assure that affiliates receive appropriate payments.^4,12,15 To ensure the continued success of VA GME partnerships, knowledge of VA GME financing has become increasingly important for designated institutional officers (DIOs) and residency program directors, particularly in light of recent investigations into oversight of the VA’s reimbursement to academic affiliates.^16-18 This report describes VA GME reimbursement and, where applicable, VA and CMS reimbursement policies are compared to highlight similarities, differences, and common principles.

VA AUTHORITY

While the VA’s primary mission is “to provide a complete hospital medical service for the medical care and treatment of veterans,”early VA leaders recognized the importance of affiliating with the nation’s academic institutions.¹⁹ In 1946, the VA Policy Memorandum Number 2 established a partnership between the VA and the academic medical community.²⁰ Additional legislation authorized specific agreements with academic affiliates for the central administration of salary and benefits for residents rotating at VA facilities. This process, known as disbursement, is an alternative payroll mechanism whereby the VA reimburses the academic affiliate for resident salary and benefits and the affiliate acts as the disbursing agent, issuing paychecks to residents.^21,22

Resident FUNDING

By policy, with rare exceptions, the VA does not sponsor residency programs due to the challenges of providing an appropriate patient mix of age, sex, and medical conditions to meet accreditation standards.⁴ Nearly all VA reimbursements are for residents in affiliate-sponsored programs, while just 1% pays for residents in legacy, VA-sponsored residency programs at 2 VA facilities. The VA budget for resident (including fellows) salary and benefits is managed by the VA Office of Academic Affiliations (OAA), the national VA office responsible for oversight, policy, and funding of VA HPE programs.

Resident Salaries and Benefits

VA funding of resident salary and benefits are analogous with CMS direct GME (DGME), which is designed to cover resident salary and benefits costs.^4,14,23 CMS DGME payments depend on a hospital’s volume of CMS inpatients and are based on a statutory formula, which uses the hospital’s resident FTE positions, the per-resident amount, and Medicare’s share of inpatient beds (Medicare patient load) to determine payments.¹² The per-resident amount is set by statute, varies geographically, and is calculated by dividing the hospital’s allowable costs of GME (percentage of CMS inpatient days) divided by the number of residents.^12,24

By comparison, the VA GME payment reimburses for each FTE based on the salary and benefits rate set by the academic affiliate. Reimbursement is calculated based on resident time spent at the VA multiplied by a daily salary rate. The daily salary rate is determined by dividing the resident’s total compensation (salary and benefits) by the number of calendar days in an academic year. Resident time spent at the VA facility is determined by obtaining rotation schedules provided by the academic affiliate and verifying resident clinical and educational activity during scheduled rotations.

Indirect Medical Education Funding

In addition to resident salary and benefits, funds to offset the cost of conducting HPE are provided to VA facilities. These funds are intended to improve and maintain necessary infrastructure for all HPE programs not just GME, including education office administration needs, teaching costs (ie, a portion of VA preceptors salary), and instructional equipment.

The Veterans Equitable Resource Allocation (VERA) is a national budgeting process for VA medical facilities that funds facility operational needs such as staff salary and benefits, infrastructure, and equipment.² The education portion of the VERA, the VERA Education Support Component (VESC), is not managed by the OAA, but rather is distributed through the VERA model to the general budget of VA facilities hosting HPE (Figure). VESC funding in the VA budget is based on labor mapping of physician time spent in education; other labor mapping categories include clinical care, research, and administration. VA facility VESC funding is calculated based on the number of paid health profession trainees (HPTs) from all professions, apportioned according to the number of FTEs for physician residents and VA-paid HPTs in other disciplines. In fiscal year 2024, VA facilities received $115,812 for each physician resident FTE position and $84,906 for each VA-paid, non-GME FTE position.

The VESC is like CMS's indirect GME funding, termed Indirect Medical Education (IME), an additional payment for each Medicare patient discharged reflecting teaching hospitals’ higher patient care costs relative to nonteaching hospitals. Described elsewhere, IME is calculated using a resident-to-bed ratio and a multiplier, which is set by statute.^4,25 While IME can be used for reimbursement for some resident clinical and educational activities(eg, research), VA VESC funds cannot be used for such activities and are part of the general facility budget and appropriated per the discretion of the medical facility director.

ESTABLISHING GME PARTNERSHIPS

An affiliation agreement establishes the administrative and legal requirements for educational relationships with academic affiliates and includes standards for conducting HPE, responsibilities for accreditation standards, program leadership, faculty, resources, supervision, academic policies, and procedures. The VA uses standardized affiliation agreement templates that have been vetted with accrediting bodies and the VA Office of General Counsel.

A disbursement agreement authorizes the VA to reimburse affiliates for resident salary and benefits for VA clinical and educational activities. The disbursement agreement details the fiscal arrangements (eg, payment in advance vs arrears, salary, and benefit rates, leave) for the reimbursement payments. Veterans Health Administration (VHA) Directive 1400.05 provides the policy and procedures for calculating reimbursement for HPT educational activities.²⁶

The VA facility designated education officer (DEO) oversees all HPE programs and coordinates the affiliation and disbursement agreement processes.²⁷ The DEO, affiliate DIO, residency program director, and VA residency site director determine the physician resident FTE positions assigned to a VA facility based on educational objectives and availability of educational resources at the VA facility, such as patient care opportunities, faculty supervisors, space, and equipment. The VA facility requests for resident FTE positions are submitted to the OAA by the facility DEO.

Once GME FTE positions are approved by the OAA, VA facilities work with their academic affiliate to submit the physician resident salary and benefit rate. Affiliate DIOs attest to the accuracy of the salary rate schedule and the local DEO submits the budget request to the OAA. Upon approval, the funds are transferred to the VA facility each fiscal year, which begins October 1. DEOs report quarterly to the OAA both budget needs and excesses based on variations in the approved FTEs due to additional VA rotations, physician resident attrition, or reassignment.

Resident Position Allocation

VA GME financing provides flexibility through periodic needs assessments and expansion initiatives. In August and December, DEOs collaborate with an academic affiliate to submit reports to the OAA confirming their projected GME needs for the next academic year. Additional positions requests are reviewed by the OAA; funding depends on budget and the educational justification. The OAA periodically issues GME expansion requests for proposal, which typically arise from legislation to address specific VA workforce needs. The VA facility DEO and affiliate GME leaders collaborate to apply for additional positions. For example, a VA GME expansion under the Veterans Access, Choice, and Accountability Act of 2014 added 1500 GME positions in 8 years for critically needed specialties and in rural and underserved areas.⁵ The Maintaining Internal Systems and Strengthening Outside Networks (MISSION) Act of 2018 authorized a pilot program for VA to fund residents at non-VA facilities with priority for Indian Health Services, Tribes and Tribal Organizations, Federally Qualified Health Centers, and US Department of Defense facilities to provide access to veterans in underserved areas.⁶

The VA GME financing system has flexibility to meet local needs for additional resident positions and to address broader VA workforce gaps through targeted expansion. Generally, CMS does not fund positions to address workforce needs, place residents in specific geographic areas, or require the training of certain types of residents.⁴ However, the Consolidated Appropriations Act of 2021 has provided the opportunity to address rural workforce needs.²⁸

Reimbursement

The VA provides reimbursement for clinical and educational activities performed in VA facilities for the benefit of veterans as well as research, didactics, meetings and conferences, annual and sick leave, and orientation. The VA also may provide reimbursement for educational activities that occur off VA grounds (eg, the VA proportional share of a residency program’s didactic sessions). The VA does not reimburse for affiliate clinical duties or administrative costs, although a national policy allows VA facilities to reimburse affiliates for some GME overhead costs.²⁹

CMS similarly reimburses for residency training time spent in patient care activities as well as orientation activities, didactics, leave, and, in some cases, research.^4,30,31 CMS makes payments to hospitals, which may include sponsoring institutions and Medicare-eligible participating training sites.^4,30,31 For both the VA and CMS, residents may not be counted twice for reimbursement by 2 federal agencies; in other words, a resident may not count for > 1 FTE.^4,30-32

GME Oversight

VA GME funding came under significant scrutiny. At a 2016 House Veterans Affairs Committee hearing, Representative Phil Roe, MD (R-Tennessee), noted that no process existed at many VA facilities for “determining trainee presence” and that many VA medical centers had “difficulty tracking resident rotations”¹⁶ A VA Office of the Inspector General investigation recommended that the VA implement policies and procedures to improve oversight to “ensure residents are fully participating in educational activities” and that the VA is “paying the correct amount” to the affiliate.¹⁷ A 2020 General Accountability Office report outlined unclear policy guidance, incomplete tracking of resident activities, and improper fiscal processes for reimbursement and reconciliation of affiliate invoices.¹⁸

In response, the OAA created an oversight and compliance unit, revised VHA Directive 1400.05 (the policy for disbursement), and improved resident tracking procedures.²⁶ The standard operating procedure that accompanied VHA Directive 1400.05 provides detailed information for the DEO and VA facility staff for tracking resident clinical and educational activities. FTE counts are essential to both VA and CMS for accurate reimbursement. The eAppendix and the Table provide a guide to reimbursable activities in the VA for the calculation of reimbursement, with a comparison to CMS.^33,34 The OAA in cooperation with other VA staff and officers periodically conducts audits to assess compliance with disbursement policy and affiliate reimbursement accuracy.

In the VA, resident activities are captured on the VA Educational Activity Record, a standardized spreadsheet to track activities and calculate reimbursement. Each VA facility hosting resident physicians manually records resident activity by the half-day. This process is labor intensive, involving both VA and affiliate staff to accurately reconcile payments. To address the workload demands, the OAA is developing an online tool that will automate aspects of the tracking process. Also, to ensure adequate staffing, the OAA is in the process of implementing an office optimization project, providing standardized position descriptions, an organizational chart, and staffing levels for DEO offices in VA facilities.

This report describes the key policies and principles of VA GME financing, highlighting the essential similarities and differences between VA and CMS. Neither the VA nor CMS regulations allow for reimbursement for > 1 FTE position per resident, a principle that underpins the assignment of resident rotations and federal funding for GME and are similar with respect to reimbursement for patient care activities, didactics, research, orientation, and scholarly activity. While reimbursable activities in the VA require physical presence and care of veteran patients, CMS also limits reimbursement to resident activities in the hospital and approved other settings if the hospital is paying for resident salary and benefits in these settings. The VA provides some flexibility for offsite activities including didactics and, in specific circumstances, remote care of veteran patients (eg, teleradiology).

The VA and CMS use different GME financing models. For example, the CMS calculations for resident FTEs are complex, whereas VA calculations reimburse the salary and benefits as set by the academic affiliate. The VA process accounts for local variation in salary rates, whereas the per-resident amount set by CMS varies regionally and does not fully account for differences in the cost of living.²⁴ Because all patients in VA facilities are veterans, VA calculations for reimbursement do not involve ratios of beds like the CMS calculations to determine a proportional share of reimbursement. The VA GME expansion tends to be more directed to VA health workforce needs than CMS, specifying the types of programs and geographic locations to address these needs.

The VA regularly reevaluates how affiliates are reimbursed for VA resident activity, balancing compliance with VA policies and the workload for VA and its affiliates. The VA obtains input from key stakeholders including DEOs, DIOs, and professional organizations such as the Association of American Medical Colleges and the Accreditation Council for Graduate Medical Education.^35,36

Looking ahead, the VA is developing an online tool to improve the accuracy of affiliate reimbursement. The VA will also implement a standardized staffing model, organizational structure, and position descriptions for DEO offices. These initiatives will help reduce the burden of tracking and verifying resident activity and continue to support the 77-year partnership between VA and its affiliated institutions.

The US Department of Veterans Affairs (VA) has partnered with academic medical centers and programs since 1946 to provide clinical training for physician residents. Ranking second in federal graduate medical education (GME) funding to the Centers for Medicare and Medicaid Services (CMS), the $850 million VA GME budget annually reimburses > 250 GME-sponsoring institutions (affiliates) of 8000 GME programs for the clinical training of 49,000 individual residents rotating through > 11,000 full-time equivalent (FTE) positions.¹ The VA also distributes $1.6 billion to VA facilities to offset the costs of conducting health professions education (HPE) (eg, facility infrastructure, salary support for VA instructors and preceptors, education office administration, and instructional equipment).² The VA financial and educational contributions account for payment of 11% of resident positions nationally and allow academic medical centers to be less reliant on CMS GME funding.^3,4 The VA contributions also provide opportunities for GME expansion,^1,5,6 educational innovations,^5,7 interprofessional and team-based care,^8,9 and quality and safety training.^10,11 The Table provides a comparison of CMS and VA GME reimbursability based on activity.

GME financing is complex, particularly the formulaic approach used by CMS, the details of which are often obscured in federal regulations. Due to this complexity and the $16 billion CMS GME budget, academic publications have focused on CMS GME financing while not fully explaining the VA GME policies and processes.^4,12-14 By comparison, the VA GME financing model is relatively straightforward and governed by different statues and VA regulations, yet sharing some of the same principles as CMS regulations. Given the challenges in CMS reimbursement to fully support the cost of resident education, as well as the educational opportunities at the VA, the VA designs its reimbursement model to assure that affiliates receive appropriate payments.^4,12,15 To ensure the continued success of VA GME partnerships, knowledge of VA GME financing has become increasingly important for designated institutional officers (DIOs) and residency program directors, particularly in light of recent investigations into oversight of the VA’s reimbursement to academic affiliates.^16-18 This report describes VA GME reimbursement and, where applicable, VA and CMS reimbursement policies are compared to highlight similarities, differences, and common principles.

VA AUTHORITY

While the VA’s primary mission is “to provide a complete hospital medical service for the medical care and treatment of veterans,”early VA leaders recognized the importance of affiliating with the nation’s academic institutions.¹⁹ In 1946, the VA Policy Memorandum Number 2 established a partnership between the VA and the academic medical community.²⁰ Additional legislation authorized specific agreements with academic affiliates for the central administration of salary and benefits for residents rotating at VA facilities. This process, known as disbursement, is an alternative payroll mechanism whereby the VA reimburses the academic affiliate for resident salary and benefits and the affiliate acts as the disbursing agent, issuing paychecks to residents.^21,22

Resident FUNDING

By policy, with rare exceptions, the VA does not sponsor residency programs due to the challenges of providing an appropriate patient mix of age, sex, and medical conditions to meet accreditation standards.⁴ Nearly all VA reimbursements are for residents in affiliate-sponsored programs, while just 1% pays for residents in legacy, VA-sponsored residency programs at 2 VA facilities. The VA budget for resident (including fellows) salary and benefits is managed by the VA Office of Academic Affiliations (OAA), the national VA office responsible for oversight, policy, and funding of VA HPE programs.

Resident Salaries and Benefits

VA funding of resident salary and benefits are analogous with CMS direct GME (DGME), which is designed to cover resident salary and benefits costs.^4,14,23 CMS DGME payments depend on a hospital’s volume of CMS inpatients and are based on a statutory formula, which uses the hospital’s resident FTE positions, the per-resident amount, and Medicare’s share of inpatient beds (Medicare patient load) to determine payments.¹² The per-resident amount is set by statute, varies geographically, and is calculated by dividing the hospital’s allowable costs of GME (percentage of CMS inpatient days) divided by the number of residents.^12,24

By comparison, the VA GME payment reimburses for each FTE based on the salary and benefits rate set by the academic affiliate. Reimbursement is calculated based on resident time spent at the VA multiplied by a daily salary rate. The daily salary rate is determined by dividing the resident’s total compensation (salary and benefits) by the number of calendar days in an academic year. Resident time spent at the VA facility is determined by obtaining rotation schedules provided by the academic affiliate and verifying resident clinical and educational activity during scheduled rotations.

Indirect Medical Education Funding

In addition to resident salary and benefits, funds to offset the cost of conducting HPE are provided to VA facilities. These funds are intended to improve and maintain necessary infrastructure for all HPE programs not just GME, including education office administration needs, teaching costs (ie, a portion of VA preceptors salary), and instructional equipment.

The Veterans Equitable Resource Allocation (VERA) is a national budgeting process for VA medical facilities that funds facility operational needs such as staff salary and benefits, infrastructure, and equipment.² The education portion of the VERA, the VERA Education Support Component (VESC), is not managed by the OAA, but rather is distributed through the VERA model to the general budget of VA facilities hosting HPE (Figure). VESC funding in the VA budget is based on labor mapping of physician time spent in education; other labor mapping categories include clinical care, research, and administration. VA facility VESC funding is calculated based on the number of paid health profession trainees (HPTs) from all professions, apportioned according to the number of FTEs for physician residents and VA-paid HPTs in other disciplines. In fiscal year 2024, VA facilities received $115,812 for each physician resident FTE position and $84,906 for each VA-paid, non-GME FTE position.

The VESC is like CMS's indirect GME funding, termed Indirect Medical Education (IME), an additional payment for each Medicare patient discharged reflecting teaching hospitals’ higher patient care costs relative to nonteaching hospitals. Described elsewhere, IME is calculated using a resident-to-bed ratio and a multiplier, which is set by statute.^4,25 While IME can be used for reimbursement for some resident clinical and educational activities(eg, research), VA VESC funds cannot be used for such activities and are part of the general facility budget and appropriated per the discretion of the medical facility director.

ESTABLISHING GME PARTNERSHIPS

An affiliation agreement establishes the administrative and legal requirements for educational relationships with academic affiliates and includes standards for conducting HPE, responsibilities for accreditation standards, program leadership, faculty, resources, supervision, academic policies, and procedures. The VA uses standardized affiliation agreement templates that have been vetted with accrediting bodies and the VA Office of General Counsel.

A disbursement agreement authorizes the VA to reimburse affiliates for resident salary and benefits for VA clinical and educational activities. The disbursement agreement details the fiscal arrangements (eg, payment in advance vs arrears, salary, and benefit rates, leave) for the reimbursement payments. Veterans Health Administration (VHA) Directive 1400.05 provides the policy and procedures for calculating reimbursement for HPT educational activities.²⁶

The VA facility designated education officer (DEO) oversees all HPE programs and coordinates the affiliation and disbursement agreement processes.²⁷ The DEO, affiliate DIO, residency program director, and VA residency site director determine the physician resident FTE positions assigned to a VA facility based on educational objectives and availability of educational resources at the VA facility, such as patient care opportunities, faculty supervisors, space, and equipment. The VA facility requests for resident FTE positions are submitted to the OAA by the facility DEO.

Once GME FTE positions are approved by the OAA, VA facilities work with their academic affiliate to submit the physician resident salary and benefit rate. Affiliate DIOs attest to the accuracy of the salary rate schedule and the local DEO submits the budget request to the OAA. Upon approval, the funds are transferred to the VA facility each fiscal year, which begins October 1. DEOs report quarterly to the OAA both budget needs and excesses based on variations in the approved FTEs due to additional VA rotations, physician resident attrition, or reassignment.

Resident Position Allocation

VA GME financing provides flexibility through periodic needs assessments and expansion initiatives. In August and December, DEOs collaborate with an academic affiliate to submit reports to the OAA confirming their projected GME needs for the next academic year. Additional positions requests are reviewed by the OAA; funding depends on budget and the educational justification. The OAA periodically issues GME expansion requests for proposal, which typically arise from legislation to address specific VA workforce needs. The VA facility DEO and affiliate GME leaders collaborate to apply for additional positions. For example, a VA GME expansion under the Veterans Access, Choice, and Accountability Act of 2014 added 1500 GME positions in 8 years for critically needed specialties and in rural and underserved areas.⁵ The Maintaining Internal Systems and Strengthening Outside Networks (MISSION) Act of 2018 authorized a pilot program for VA to fund residents at non-VA facilities with priority for Indian Health Services, Tribes and Tribal Organizations, Federally Qualified Health Centers, and US Department of Defense facilities to provide access to veterans in underserved areas.⁶

The VA GME financing system has flexibility to meet local needs for additional resident positions and to address broader VA workforce gaps through targeted expansion. Generally, CMS does not fund positions to address workforce needs, place residents in specific geographic areas, or require the training of certain types of residents.⁴ However, the Consolidated Appropriations Act of 2021 has provided the opportunity to address rural workforce needs.²⁸

Reimbursement

The VA provides reimbursement for clinical and educational activities performed in VA facilities for the benefit of veterans as well as research, didactics, meetings and conferences, annual and sick leave, and orientation. The VA also may provide reimbursement for educational activities that occur off VA grounds (eg, the VA proportional share of a residency program’s didactic sessions). The VA does not reimburse for affiliate clinical duties or administrative costs, although a national policy allows VA facilities to reimburse affiliates for some GME overhead costs.²⁹

CMS similarly reimburses for residency training time spent in patient care activities as well as orientation activities, didactics, leave, and, in some cases, research.^4,30,31 CMS makes payments to hospitals, which may include sponsoring institutions and Medicare-eligible participating training sites.^4,30,31 For both the VA and CMS, residents may not be counted twice for reimbursement by 2 federal agencies; in other words, a resident may not count for > 1 FTE.^4,30-32

GME Oversight

VA GME funding came under significant scrutiny. At a 2016 House Veterans Affairs Committee hearing, Representative Phil Roe, MD (R-Tennessee), noted that no process existed at many VA facilities for “determining trainee presence” and that many VA medical centers had “difficulty tracking resident rotations”¹⁶ A VA Office of the Inspector General investigation recommended that the VA implement policies and procedures to improve oversight to “ensure residents are fully participating in educational activities” and that the VA is “paying the correct amount” to the affiliate.¹⁷ A 2020 General Accountability Office report outlined unclear policy guidance, incomplete tracking of resident activities, and improper fiscal processes for reimbursement and reconciliation of affiliate invoices.¹⁸

In response, the OAA created an oversight and compliance unit, revised VHA Directive 1400.05 (the policy for disbursement), and improved resident tracking procedures.²⁶ The standard operating procedure that accompanied VHA Directive 1400.05 provides detailed information for the DEO and VA facility staff for tracking resident clinical and educational activities. FTE counts are essential to both VA and CMS for accurate reimbursement. The eAppendix and the Table provide a guide to reimbursable activities in the VA for the calculation of reimbursement, with a comparison to CMS.^33,34 The OAA in cooperation with other VA staff and officers periodically conducts audits to assess compliance with disbursement policy and affiliate reimbursement accuracy.

In the VA, resident activities are captured on the VA Educational Activity Record, a standardized spreadsheet to track activities and calculate reimbursement. Each VA facility hosting resident physicians manually records resident activity by the half-day. This process is labor intensive, involving both VA and affiliate staff to accurately reconcile payments. To address the workload demands, the OAA is developing an online tool that will automate aspects of the tracking process. Also, to ensure adequate staffing, the OAA is in the process of implementing an office optimization project, providing standardized position descriptions, an organizational chart, and staffing levels for DEO offices in VA facilities.

This report describes the key policies and principles of VA GME financing, highlighting the essential similarities and differences between VA and CMS. Neither the VA nor CMS regulations allow for reimbursement for > 1 FTE position per resident, a principle that underpins the assignment of resident rotations and federal funding for GME and are similar with respect to reimbursement for patient care activities, didactics, research, orientation, and scholarly activity. While reimbursable activities in the VA require physical presence and care of veteran patients, CMS also limits reimbursement to resident activities in the hospital and approved other settings if the hospital is paying for resident salary and benefits in these settings. The VA provides some flexibility for offsite activities including didactics and, in specific circumstances, remote care of veteran patients (eg, teleradiology).

The VA and CMS use different GME financing models. For example, the CMS calculations for resident FTEs are complex, whereas VA calculations reimburse the salary and benefits as set by the academic affiliate. The VA process accounts for local variation in salary rates, whereas the per-resident amount set by CMS varies regionally and does not fully account for differences in the cost of living.²⁴ Because all patients in VA facilities are veterans, VA calculations for reimbursement do not involve ratios of beds like the CMS calculations to determine a proportional share of reimbursement. The VA GME expansion tends to be more directed to VA health workforce needs than CMS, specifying the types of programs and geographic locations to address these needs.

The VA regularly reevaluates how affiliates are reimbursed for VA resident activity, balancing compliance with VA policies and the workload for VA and its affiliates. The VA obtains input from key stakeholders including DEOs, DIOs, and professional organizations such as the Association of American Medical Colleges and the Accreditation Council for Graduate Medical Education.^35,36

Looking ahead, the VA is developing an online tool to improve the accuracy of affiliate reimbursement. The VA will also implement a standardized staffing model, organizational structure, and position descriptions for DEO offices. These initiatives will help reduce the burden of tracking and verifying resident activity and continue to support the 77-year partnership between VA and its affiliated institutions.

The US Department of Veterans Affairs (VA) has partnered with academic medical centers and programs since 1946 to provide clinical training for physician residents. Ranking second in federal graduate medical education (GME) funding to the Centers for Medicare and Medicaid Services (CMS), the $850 million VA GME budget annually reimburses > 250 GME-sponsoring institutions (affiliates) of 8000 GME programs for the clinical training of 49,000 individual residents rotating through > 11,000 full-time equivalent (FTE) positions.¹ The VA also distributes $1.6 billion to VA facilities to offset the costs of conducting health professions education (HPE) (eg, facility infrastructure, salary support for VA instructors and preceptors, education office administration, and instructional equipment).² The VA financial and educational contributions account for payment of 11% of resident positions nationally and allow academic medical centers to be less reliant on CMS GME funding.^3,4 The VA contributions also provide opportunities for GME expansion,^1,5,6 educational innovations,^5,7 interprofessional and team-based care,^8,9 and quality and safety training.^10,11 The Table provides a comparison of CMS and VA GME reimbursability based on activity.

GME financing is complex, particularly the formulaic approach used by CMS, the details of which are often obscured in federal regulations. Due to this complexity and the $16 billion CMS GME budget, academic publications have focused on CMS GME financing while not fully explaining the VA GME policies and processes.^4,12-14 By comparison, the VA GME financing model is relatively straightforward and governed by different statues and VA regulations, yet sharing some of the same principles as CMS regulations. Given the challenges in CMS reimbursement to fully support the cost of resident education, as well as the educational opportunities at the VA, the VA designs its reimbursement model to assure that affiliates receive appropriate payments.^4,12,15 To ensure the continued success of VA GME partnerships, knowledge of VA GME financing has become increasingly important for designated institutional officers (DIOs) and residency program directors, particularly in light of recent investigations into oversight of the VA’s reimbursement to academic affiliates.^16-18 This report describes VA GME reimbursement and, where applicable, VA and CMS reimbursement policies are compared to highlight similarities, differences, and common principles.

VA AUTHORITY

While the VA’s primary mission is “to provide a complete hospital medical service for the medical care and treatment of veterans,”early VA leaders recognized the importance of affiliating with the nation’s academic institutions.¹⁹ In 1946, the VA Policy Memorandum Number 2 established a partnership between the VA and the academic medical community.²⁰ Additional legislation authorized specific agreements with academic affiliates for the central administration of salary and benefits for residents rotating at VA facilities. This process, known as disbursement, is an alternative payroll mechanism whereby the VA reimburses the academic affiliate for resident salary and benefits and the affiliate acts as the disbursing agent, issuing paychecks to residents.^21,22

Resident FUNDING

By policy, with rare exceptions, the VA does not sponsor residency programs due to the challenges of providing an appropriate patient mix of age, sex, and medical conditions to meet accreditation standards.⁴ Nearly all VA reimbursements are for residents in affiliate-sponsored programs, while just 1% pays for residents in legacy, VA-sponsored residency programs at 2 VA facilities. The VA budget for resident (including fellows) salary and benefits is managed by the VA Office of Academic Affiliations (OAA), the national VA office responsible for oversight, policy, and funding of VA HPE programs.

Resident Salaries and Benefits

VA funding of resident salary and benefits are analogous with CMS direct GME (DGME), which is designed to cover resident salary and benefits costs.^4,14,23 CMS DGME payments depend on a hospital’s volume of CMS inpatients and are based on a statutory formula, which uses the hospital’s resident FTE positions, the per-resident amount, and Medicare’s share of inpatient beds (Medicare patient load) to determine payments.¹² The per-resident amount is set by statute, varies geographically, and is calculated by dividing the hospital’s allowable costs of GME (percentage of CMS inpatient days) divided by the number of residents.^12,24

By comparison, the VA GME payment reimburses for each FTE based on the salary and benefits rate set by the academic affiliate. Reimbursement is calculated based on resident time spent at the VA multiplied by a daily salary rate. The daily salary rate is determined by dividing the resident’s total compensation (salary and benefits) by the number of calendar days in an academic year. Resident time spent at the VA facility is determined by obtaining rotation schedules provided by the academic affiliate and verifying resident clinical and educational activity during scheduled rotations.

Indirect Medical Education Funding

In addition to resident salary and benefits, funds to offset the cost of conducting HPE are provided to VA facilities. These funds are intended to improve and maintain necessary infrastructure for all HPE programs not just GME, including education office administration needs, teaching costs (ie, a portion of VA preceptors salary), and instructional equipment.

The Veterans Equitable Resource Allocation (VERA) is a national budgeting process for VA medical facilities that funds facility operational needs such as staff salary and benefits, infrastructure, and equipment.² The education portion of the VERA, the VERA Education Support Component (VESC), is not managed by the OAA, but rather is distributed through the VERA model to the general budget of VA facilities hosting HPE (Figure). VESC funding in the VA budget is based on labor mapping of physician time spent in education; other labor mapping categories include clinical care, research, and administration. VA facility VESC funding is calculated based on the number of paid health profession trainees (HPTs) from all professions, apportioned according to the number of FTEs for physician residents and VA-paid HPTs in other disciplines. In fiscal year 2024, VA facilities received $115,812 for each physician resident FTE position and $84,906 for each VA-paid, non-GME FTE position.

The VESC is like CMS's indirect GME funding, termed Indirect Medical Education (IME), an additional payment for each Medicare patient discharged reflecting teaching hospitals’ higher patient care costs relative to nonteaching hospitals. Described elsewhere, IME is calculated using a resident-to-bed ratio and a multiplier, which is set by statute.^4,25 While IME can be used for reimbursement for some resident clinical and educational activities(eg, research), VA VESC funds cannot be used for such activities and are part of the general facility budget and appropriated per the discretion of the medical facility director.

ESTABLISHING GME PARTNERSHIPS

An affiliation agreement establishes the administrative and legal requirements for educational relationships with academic affiliates and includes standards for conducting HPE, responsibilities for accreditation standards, program leadership, faculty, resources, supervision, academic policies, and procedures. The VA uses standardized affiliation agreement templates that have been vetted with accrediting bodies and the VA Office of General Counsel.

A disbursement agreement authorizes the VA to reimburse affiliates for resident salary and benefits for VA clinical and educational activities. The disbursement agreement details the fiscal arrangements (eg, payment in advance vs arrears, salary, and benefit rates, leave) for the reimbursement payments. Veterans Health Administration (VHA) Directive 1400.05 provides the policy and procedures for calculating reimbursement for HPT educational activities.²⁶

The VA facility designated education officer (DEO) oversees all HPE programs and coordinates the affiliation and disbursement agreement processes.²⁷ The DEO, affiliate DIO, residency program director, and VA residency site director determine the physician resident FTE positions assigned to a VA facility based on educational objectives and availability of educational resources at the VA facility, such as patient care opportunities, faculty supervisors, space, and equipment. The VA facility requests for resident FTE positions are submitted to the OAA by the facility DEO.

Once GME FTE positions are approved by the OAA, VA facilities work with their academic affiliate to submit the physician resident salary and benefit rate. Affiliate DIOs attest to the accuracy of the salary rate schedule and the local DEO submits the budget request to the OAA. Upon approval, the funds are transferred to the VA facility each fiscal year, which begins October 1. DEOs report quarterly to the OAA both budget needs and excesses based on variations in the approved FTEs due to additional VA rotations, physician resident attrition, or reassignment.

Resident Position Allocation

VA GME financing provides flexibility through periodic needs assessments and expansion initiatives. In August and December, DEOs collaborate with an academic affiliate to submit reports to the OAA confirming their projected GME needs for the next academic year. Additional positions requests are reviewed by the OAA; funding depends on budget and the educational justification. The OAA periodically issues GME expansion requests for proposal, which typically arise from legislation to address specific VA workforce needs. The VA facility DEO and affiliate GME leaders collaborate to apply for additional positions. For example, a VA GME expansion under the Veterans Access, Choice, and Accountability Act of 2014 added 1500 GME positions in 8 years for critically needed specialties and in rural and underserved areas.⁵ The Maintaining Internal Systems and Strengthening Outside Networks (MISSION) Act of 2018 authorized a pilot program for VA to fund residents at non-VA facilities with priority for Indian Health Services, Tribes and Tribal Organizations, Federally Qualified Health Centers, and US Department of Defense facilities to provide access to veterans in underserved areas.⁶

The VA GME financing system has flexibility to meet local needs for additional resident positions and to address broader VA workforce gaps through targeted expansion. Generally, CMS does not fund positions to address workforce needs, place residents in specific geographic areas, or require the training of certain types of residents.⁴ However, the Consolidated Appropriations Act of 2021 has provided the opportunity to address rural workforce needs.²⁸

Reimbursement

The VA provides reimbursement for clinical and educational activities performed in VA facilities for the benefit of veterans as well as research, didactics, meetings and conferences, annual and sick leave, and orientation. The VA also may provide reimbursement for educational activities that occur off VA grounds (eg, the VA proportional share of a residency program’s didactic sessions). The VA does not reimburse for affiliate clinical duties or administrative costs, although a national policy allows VA facilities to reimburse affiliates for some GME overhead costs.²⁹

CMS similarly reimburses for residency training time spent in patient care activities as well as orientation activities, didactics, leave, and, in some cases, research.^4,30,31 CMS makes payments to hospitals, which may include sponsoring institutions and Medicare-eligible participating training sites.^4,30,31 For both the VA and CMS, residents may not be counted twice for reimbursement by 2 federal agencies; in other words, a resident may not count for > 1 FTE.^4,30-32

GME Oversight

VA GME funding came under significant scrutiny. At a 2016 House Veterans Affairs Committee hearing, Representative Phil Roe, MD (R-Tennessee), noted that no process existed at many VA facilities for “determining trainee presence” and that many VA medical centers had “difficulty tracking resident rotations”¹⁶ A VA Office of the Inspector General investigation recommended that the VA implement policies and procedures to improve oversight to “ensure residents are fully participating in educational activities” and that the VA is “paying the correct amount” to the affiliate.¹⁷ A 2020 General Accountability Office report outlined unclear policy guidance, incomplete tracking of resident activities, and improper fiscal processes for reimbursement and reconciliation of affiliate invoices.¹⁸

In response, the OAA created an oversight and compliance unit, revised VHA Directive 1400.05 (the policy for disbursement), and improved resident tracking procedures.²⁶ The standard operating procedure that accompanied VHA Directive 1400.05 provides detailed information for the DEO and VA facility staff for tracking resident clinical and educational activities. FTE counts are essential to both VA and CMS for accurate reimbursement. The eAppendix and the Table provide a guide to reimbursable activities in the VA for the calculation of reimbursement, with a comparison to CMS.^33,34 The OAA in cooperation with other VA staff and officers periodically conducts audits to assess compliance with disbursement policy and affiliate reimbursement accuracy.

In the VA, resident activities are captured on the VA Educational Activity Record, a standardized spreadsheet to track activities and calculate reimbursement. Each VA facility hosting resident physicians manually records resident activity by the half-day. This process is labor intensive, involving both VA and affiliate staff to accurately reconcile payments. To address the workload demands, the OAA is developing an online tool that will automate aspects of the tracking process. Also, to ensure adequate staffing, the OAA is in the process of implementing an office optimization project, providing standardized position descriptions, an organizational chart, and staffing levels for DEO offices in VA facilities.

This report describes the key policies and principles of VA GME financing, highlighting the essential similarities and differences between VA and CMS. Neither the VA nor CMS regulations allow for reimbursement for > 1 FTE position per resident, a principle that underpins the assignment of resident rotations and federal funding for GME and are similar with respect to reimbursement for patient care activities, didactics, research, orientation, and scholarly activity. While reimbursable activities in the VA require physical presence and care of veteran patients, CMS also limits reimbursement to resident activities in the hospital and approved other settings if the hospital is paying for resident salary and benefits in these settings. The VA provides some flexibility for offsite activities including didactics and, in specific circumstances, remote care of veteran patients (eg, teleradiology).

The VA and CMS use different GME financing models. For example, the CMS calculations for resident FTEs are complex, whereas VA calculations reimburse the salary and benefits as set by the academic affiliate. The VA process accounts for local variation in salary rates, whereas the per-resident amount set by CMS varies regionally and does not fully account for differences in the cost of living.²⁴ Because all patients in VA facilities are veterans, VA calculations for reimbursement do not involve ratios of beds like the CMS calculations to determine a proportional share of reimbursement. The VA GME expansion tends to be more directed to VA health workforce needs than CMS, specifying the types of programs and geographic locations to address these needs.

The VA regularly reevaluates how affiliates are reimbursed for VA resident activity, balancing compliance with VA policies and the workload for VA and its affiliates. The VA obtains input from key stakeholders including DEOs, DIOs, and professional organizations such as the Association of American Medical Colleges and the Accreditation Council for Graduate Medical Education.^35,36

Looking ahead, the VA is developing an online tool to improve the accuracy of affiliate reimbursement. The VA will also implement a standardized staffing model, organizational structure, and position descriptions for DEO offices. These initiatives will help reduce the burden of tracking and verifying resident activity and continue to support the 77-year partnership between VA and its affiliated institutions.