Original Research

Diabetes mellitus is one of the most common comorbid conditions among patients hospitalized for acute bacterial skin infections.[1, 2, 3, 4, 5, 6] Acute bacterial skin infections in diabetics represent a spectrum of conditions ranging from cellulitis or cutaneous abscess to more complicated infections such as infected ulcers or deep tissue infections. Although most skin infections in diabetics are caused by gram‐positive pathogens (Staphylococcus aureus and streptococci), the risk of gram‐negative pathogens is increased in certain complicated infections such as diabetic foot infections.[7] For such complicated infections, national guidelines therefore recommend broad‐spectrum empiric antibiotic therapy.[7]

The role of gram‐negative pathogens has not been clearly established in diabetics with cellulitis or cutaneous abscess not associated with an infected ulcer or diabetic foot infection. National guidelines for the treatment of cellulitis and abscess recommend antibiotic therapy targeted toward S aureus and streptococcal species irrespective of the presence of diabetes mellitus.[8, 9] However, in a recent multicenter study of patients hospitalized with acute bacterial skin infections in which cases involving infected ulcers or deep tissue infection were excluded, diabetes mellitus was an independent predictor of use of antibiotics with broad gram‐negative activity.[2] This suggests that either gram‐negative pathogens are more common or providers perceive gram‐negative pathogens to be more common among diabetics with otherwise uncomplicated cellulitis or abscess.

A better understanding of the relationship between the microbiology and antibiotic prescribing practices for diabetics with cellulitis or abscess is therefore necessary to promote the most appropriate spectrum of therapy for these patients. We evaluated a large cohort of patients hospitalized with acute bacterial skin infections in order to: (1) compare the microbiology of diabetics and nondiabetics with cellulitis or cutaneous abscess not associated with an ulcer or deep tissue infection; and (2) compare antibiotic prescribing practices among diabetics and nondiabetics. We hypothesized that diabetics would have a similar spectrum of microorganisms as nondiabetics but would be more frequently treated with antibiotics with broad gram‐negative activity.

METHODS

RESULTS

After excluding 102 pediatric cases and removing 5 duplicate cases, 770 total cases were included for analysis: 447 involved cellulitis and 323 involved cutaneous abscess (Figure 1). Overall, 167 (22%) patients had diabetes mellitus. Diabetics were significantly more likely than nondiabetics to have cellulitis as the presenting infection (67% of cases vs 56%, P=0.008) and to have lower extremity involvement (48% vs 33%, P<0.001) (Table 1). Diabetics were also older (median age 55 years vs 48 years, P<0.001), more likely to have cirrhosis or prior skin infection, and less likely to be injection‐drug users or human immunodeficiency virus (HIV) infected. Demographic and clinical characteristics among diabetics and nondiabetics stratified by the categorizations of cellulitis and cutaneous abscess are presented in the Supporting Information, Appendix Table 1, in the online version of this article.

Figure 1

The frequency of use of microbiological cultures was similar among diabetics and nondiabetics (Table 2). In cases of cellulitis, a microorganism was identified in 18% of diabetics and 12% of nondiabetics (P=0.09). In cases of cutaneous abscess, a microorganism was identified more commonly (69% and 74%, respectively, P=0.50). Among cases where a microorganism was identified, aerobic gram‐positive organisms were isolated in 90% of diabetics and 92% of nondiabetics (P=0.59). Aerobic gram‐negative organisms were isolated in 7% of diabetics and 12% of nondiabetics (P=0.28). Specific gram‐negative organisms isolated are shown in the Supporting Information, Appendix Table 2, in the online version of this article; no cases in diabetics involved Pseudomonas aeruginosa. The comparison of microbiological data among diabetics and nondiabetics was similar when stratified by cellulitis versus cutaneous abscess (Table 2).

Antibiotic utilization is summarized in Table 3. Among patients who were started on antibiotic therapy in the emergency department or urgent care, the initial regimen included an agent with broad gram‐negative activity in 31% of both diabetics and nondiabetics (P=0.97). During the entire hospital stay (including the emergency department or urgent care), diabetics were significantly more likely to be treated with ‐lactam/‐lactamase inhibitor combinations (42% vs 33%, P=0.04). At the time of hospital discharge, diabetics were more likely to be prescribed fluoroquinolones (11% vs 5%, P=0.01) (Table 3) particularly for cases of cellulitis (13% vs 6%, P=0.008) (see Supporting Information, Appendix Table 3, in the online version of this article). Diabetics were somewhat more likely to be prescribed parenteral antibiotics (10% vs 6%, P=0.07) after discharge. When considering both inpatient and discharge therapy, more diabetics than nondiabetics were exposed to at least 2 calendar days of broad gram‐negative therapy (54% vs 44%, P=0.02) and more were prescribed an antipseudomonal agent (38% vs 25%, P=0.002). In the group of patients who received at least 1 dose of an antibiotic with broad gram‐negative activity, broad gram‐negative agents accounted for 33% of the total days of therapy prescribed for diabetics and 32% for nondiabetics. Overall prescribing patterns were similar when stratified by cellulitis versus cutaneous abscess (see Supporting Information, Appendix Table 3, in the online version of this article).

After adjusting for covariates in the logistic regression model, diabetes mellitus was an independent predictor of exposure to broad gram‐negative therapy (see Supporting Information, Appendix Table 4, in the online version of this article). In addition to diabetes mellitus, culture of an aerobic gram‐negative microorganism, infectious diseases service consultation, presence of fever, and nonmedical admitting services were significantly associated with exposure to broad gram‐negative therapy. Prior methicillin‐resistant S aureus infection or colonization and HIV infection were inversely associated. Compared with nondiabetics, the total duration of antibiotic therapy in diabetics was somewhat longer (median 13 days vs 12 days, P=0.09) (Table 3). After adjusting for covariates in the linear regression model, there was a significant association between diabetes mellitus and treatment duration. On average, diabetics were treated 1 day (95% confidence interval: 0.2‐1.7 days) longer than nondiabetics.

Compared with nondiabetics, diabetics were more likely to have an outpatient follow‐up visit (73% vs 61%, P=0.002) and to be rehospitalized for any reason after discharge (16% vs 9%, P=0.02) (Table 4). Diabetics were overall more likely to be classified as clinical failure (15% vs 9%, P=0.02); this difference was driven by the cellulitis subgroup (19% vs 10%, P=0.01).

DISCUSSION

Diabetes mellitus is a common comorbidity in patients with acute bacterial skin infections. In this large cohort of patients hospitalized for cellulitis or cutaneous abscess, where those with infected ulcers or deep tissue infections were excluded, microbiological findings in cases associated with positive cultures were similar among diabetics and nondiabetics. Although aerobic gram‐negative microorganisms were not more likely to be identified in diabetics, diabetics were significantly more likely to be exposed to at least 2 calendar days of antibiotics with broad gram‐negative activity. After adjusting for covariates, diabetes mellitus was independently associated with exposure to broad gram‐negative therapy.

To our knowledge, this is the first study to compare the microbiology of cellulitis and cutaneous abscess among diabetics and nondiabetics. Lipsky and colleagues previously described the microbiology of a cohort of diabetic patients hospitalized with a broader range of skin infections including cellulitis, infected ulcers, and surgical site infections.[12] Similar to our findings, gram‐negative pathogens were uncommonly isolated in that study; however, in the absence of a comparator group, whether diabetics were at higher risk for gram‐negative involvement than nondiabetics was not known. Similar to the study by Lipsky and colleagues, most studies of skin infections in diabetics have included a relatively heterogeneous group of infections.[12, 13, 14, 15] The present study therefore contributes to the literature by providing a focused comparison of the microbiology of inpatient cellulitis and abscess in the absence of complicating factors such as an infected ulcer or deep tissue involvement. We found that among cases with a positive culture (13% of cases in the cellulitis group and 73% in the abscess group), the microbiology was similar among diabetics and nondiabetics. Although a microorganism was identified in only a minority of cases of cellulitis, our findings do not support the need for broad gram‐negative therapy in diabetics with cellulitis not associated with an ulcer or deep tissue infection. In diabetics with an abscess, antibiotics with broad gram‐negative activity do not appear to be indicated.

The present study also adds to the literature by providing a detailed comparison of antibiotic utilization patterns among diabetics and nondiabetics. We demonstrated that diabetics were more likely to have significant exposure to antibiotics with broad gram‐negative activity, particularly antipseudomonal agents (the broadest‐spectrum antibiotics). Because initiation of broad gram‐negative therapy in the emergency department or urgent care was not more common among diabetics, the increased use of these agents among diabetics appeared to be driven by inpatient providers. It is also notable that of patients who received any antibiotic with broad gram‐negative activity, these agents accounted for similar proportions of the total days of therapy in both diabetics and nondiabetics. In aggregate, our findings demonstrate that diabetics are more likely to be started on antibiotics with broad gram‐negative activity by inpatient providers, diabetics are not necessarily continued on longer durations of broad gram‐negative therapy once started, and the total amount of exposure to broad gram‐negative agents is substantial.

Overall, our findings suggest that inpatient providers perceive diabetics with cellulitis or abscess to be at increased risk for gram‐negative pathogens. This perhaps reflects an extrapolation of recommendations to use broad‐spectrum empiric therapy in diabetics with certain complicated skin infections.[7] However, for patients with cellulitis or cutaneous abscess, Infectious Diseases Society of America (IDSA) guidelines recommend antibiotic therapy targeted toward S aureus and streptococcal species; there is no suggestion to use a broader spectrum of therapy in diabetics.[8, 9] Our findings therefore highlight an important opportunity to improve antibiotic selection for all patients hospitalized with cellulitis and abscess, but particularly diabetics. It is also noteworthy that by linear regression, diabetes mellitus was independently associated with longer treatment durations. Although the average increase in treatment duration was small (1 day), this finding adds to the evidence that the presence of diabetes mellitus alters providers' treatment approach to cellulitis or abscess.

We found that despite more frequent treatment with broad gram‐negative therapy, diabetics were more likely than nondiabetics to be classified as clinical failures. It is important to point out that diabetics were also more likely than nondiabetics to have postdischarge outpatient follow‐up visits, raising the possibility of biased ascertainment of clinical failure events in this group. However, we also demonstrated that diabetics with cellulitis were more likely to be rehospitalized than nondiabetics. This is similar to a finding by Suaya and colleagues who showed that diabetics with skin infections were about twice as likely to be rehospitalized as nondiabetics.[13] One could hypothesize that the increased frequency of clinical failure events among diabetics was due to their older age, hyperglycemia, or vascular insufficiency; however, other factors may have contributed. For example, providers may have mistaken residual erythema for ongoing or recurrent cellulitis, or the diagnosis of cellulitis could have been incorrect to begin with. Additionally, there may have been uncertainty about the microbiology of cellulitis because the infecting pathogen was not usually identified. These factors may have led to alterations in treatment that would have resulted in a classification of clinical failure, and it is possible that providers had a lower threshold to alter treatment in diabetics. It is therefore not clear whether our findings represent a true difference in clinical outcomes between diabetics or nondiabetics. Regardless, in cases associated with a positive culture, our microbiological results do not support that the difference in clinical failure between diabetics and nondiabetics with cellulitis was related to a different spectrum of microorganisms.

In addition to the limitations outlined previously[2, 10] and above, the present study has at least 5 additional limitations. First, this was a secondary analysis of studies that were not designed to evaluate the effect of diabetes mellitus on the microbiology and treatment of skin infections. For example, hemoglobin A1C values were not collected; therefore, we could not examine whether the microbiology and antibiotic prescribing practices differed based on control of diabetes mellitus. Second, there were minor differences in inclusion and exclusion criteria between the 2 cohorts included in this study. Because the proportion of patients with diabetes mellitus was similar among both cohorts, and comparisons were not made between the cohorts, this should not have impacted our results. Third, the broad categorization of cellulitis used when combining the 2 cohorts raised the possibility of differences in infection characteristics between diabetics and nondiabetics (eg, presence of a wound) that could have confounded our findings regarding use of gram‐negative therapy. In the larger of the 2 cohorts from which the combined cohort was derived, only 17 (3%) of 533 patients had wound infections, whereas those with infected ulcers or suspected deep‐tissue infection were excluded from both cohorts. Furthermore, in the combined cohort, the increased frequency of broad gram‐negative therapy among diabetics was also observed in the cutaneous abscess group. It is therefore unlikely that the categorization of cellulitis had a significant impact on our results. Fourth, given the observational nature of the study, the microbiological data were subject to limitations. Importantly, because the infecting pathogen was identified in only 13% of cases of cellulitis, firm conclusions regarding the microbiology of cellulitis cannot be drawn. Finally, the small number of gram‐negative organisms isolated precluded comparisons of specific pathogens among diabetics and nondiabetics. In addition, because a number of gram‐negative organisms were isolated from wound cultures, it is not known whether they were clinically relevant or simply represented colonization.

In conclusion, in cases of cellulitis or abscess associated with a positive culture, gram‐negative microorganisms were not isolated more commonly among diabetics compared with nondiabetics. However, in general, diabetics were more likely to be treated with broad gram‐negative therapy suggesting that, particularly for cutaneous abscesses, this prescribing practice may not be warranted. These findings support current IDSA guidelines that recommend antibiotic therapy targeted toward gram‐positive pathogens for cellulitis or abscess, irrespective of the presence of diabetes mellitus.[8, 9] Because nearly one‐fourth of patients hospitalized with cellulitis or abscess are diabetic, these findings have relevance for national antimicrobial stewardship efforts aimed at curbing antimicrobial resistance through reducing use of antibiotics with broad gram‐negative activity in hospitals.[16]

Disclosures: This work was supported by the National Institute of Allergy and Infectious Diseases, National Institute of Health (TCJ: K23 AI099082). D.M.P. reports potential conflicts of interests with Optimer, Cubist, and Forest Pharmaceuticals. The authors report no other conflicts of interest.

Diabetes mellitus is one of the most common comorbid conditions among patients hospitalized for acute bacterial skin infections.[1, 2, 3, 4, 5, 6] Acute bacterial skin infections in diabetics represent a spectrum of conditions ranging from cellulitis or cutaneous abscess to more complicated infections such as infected ulcers or deep tissue infections. Although most skin infections in diabetics are caused by gram‐positive pathogens (Staphylococcus aureus and streptococci), the risk of gram‐negative pathogens is increased in certain complicated infections such as diabetic foot infections.[7] For such complicated infections, national guidelines therefore recommend broad‐spectrum empiric antibiotic therapy.[7]

The role of gram‐negative pathogens has not been clearly established in diabetics with cellulitis or cutaneous abscess not associated with an infected ulcer or diabetic foot infection. National guidelines for the treatment of cellulitis and abscess recommend antibiotic therapy targeted toward S aureus and streptococcal species irrespective of the presence of diabetes mellitus.[8, 9] However, in a recent multicenter study of patients hospitalized with acute bacterial skin infections in which cases involving infected ulcers or deep tissue infection were excluded, diabetes mellitus was an independent predictor of use of antibiotics with broad gram‐negative activity.[2] This suggests that either gram‐negative pathogens are more common or providers perceive gram‐negative pathogens to be more common among diabetics with otherwise uncomplicated cellulitis or abscess.

A better understanding of the relationship between the microbiology and antibiotic prescribing practices for diabetics with cellulitis or abscess is therefore necessary to promote the most appropriate spectrum of therapy for these patients. We evaluated a large cohort of patients hospitalized with acute bacterial skin infections in order to: (1) compare the microbiology of diabetics and nondiabetics with cellulitis or cutaneous abscess not associated with an ulcer or deep tissue infection; and (2) compare antibiotic prescribing practices among diabetics and nondiabetics. We hypothesized that diabetics would have a similar spectrum of microorganisms as nondiabetics but would be more frequently treated with antibiotics with broad gram‐negative activity.

METHODS

RESULTS

After excluding 102 pediatric cases and removing 5 duplicate cases, 770 total cases were included for analysis: 447 involved cellulitis and 323 involved cutaneous abscess (Figure 1). Overall, 167 (22%) patients had diabetes mellitus. Diabetics were significantly more likely than nondiabetics to have cellulitis as the presenting infection (67% of cases vs 56%, P=0.008) and to have lower extremity involvement (48% vs 33%, P<0.001) (Table 1). Diabetics were also older (median age 55 years vs 48 years, P<0.001), more likely to have cirrhosis or prior skin infection, and less likely to be injection‐drug users or human immunodeficiency virus (HIV) infected. Demographic and clinical characteristics among diabetics and nondiabetics stratified by the categorizations of cellulitis and cutaneous abscess are presented in the Supporting Information, Appendix Table 1, in the online version of this article.

Figure 1

The frequency of use of microbiological cultures was similar among diabetics and nondiabetics (Table 2). In cases of cellulitis, a microorganism was identified in 18% of diabetics and 12% of nondiabetics (P=0.09). In cases of cutaneous abscess, a microorganism was identified more commonly (69% and 74%, respectively, P=0.50). Among cases where a microorganism was identified, aerobic gram‐positive organisms were isolated in 90% of diabetics and 92% of nondiabetics (P=0.59). Aerobic gram‐negative organisms were isolated in 7% of diabetics and 12% of nondiabetics (P=0.28). Specific gram‐negative organisms isolated are shown in the Supporting Information, Appendix Table 2, in the online version of this article; no cases in diabetics involved Pseudomonas aeruginosa. The comparison of microbiological data among diabetics and nondiabetics was similar when stratified by cellulitis versus cutaneous abscess (Table 2).

Antibiotic utilization is summarized in Table 3. Among patients who were started on antibiotic therapy in the emergency department or urgent care, the initial regimen included an agent with broad gram‐negative activity in 31% of both diabetics and nondiabetics (P=0.97). During the entire hospital stay (including the emergency department or urgent care), diabetics were significantly more likely to be treated with ‐lactam/‐lactamase inhibitor combinations (42% vs 33%, P=0.04). At the time of hospital discharge, diabetics were more likely to be prescribed fluoroquinolones (11% vs 5%, P=0.01) (Table 3) particularly for cases of cellulitis (13% vs 6%, P=0.008) (see Supporting Information, Appendix Table 3, in the online version of this article). Diabetics were somewhat more likely to be prescribed parenteral antibiotics (10% vs 6%, P=0.07) after discharge. When considering both inpatient and discharge therapy, more diabetics than nondiabetics were exposed to at least 2 calendar days of broad gram‐negative therapy (54% vs 44%, P=0.02) and more were prescribed an antipseudomonal agent (38% vs 25%, P=0.002). In the group of patients who received at least 1 dose of an antibiotic with broad gram‐negative activity, broad gram‐negative agents accounted for 33% of the total days of therapy prescribed for diabetics and 32% for nondiabetics. Overall prescribing patterns were similar when stratified by cellulitis versus cutaneous abscess (see Supporting Information, Appendix Table 3, in the online version of this article).

After adjusting for covariates in the logistic regression model, diabetes mellitus was an independent predictor of exposure to broad gram‐negative therapy (see Supporting Information, Appendix Table 4, in the online version of this article). In addition to diabetes mellitus, culture of an aerobic gram‐negative microorganism, infectious diseases service consultation, presence of fever, and nonmedical admitting services were significantly associated with exposure to broad gram‐negative therapy. Prior methicillin‐resistant S aureus infection or colonization and HIV infection were inversely associated. Compared with nondiabetics, the total duration of antibiotic therapy in diabetics was somewhat longer (median 13 days vs 12 days, P=0.09) (Table 3). After adjusting for covariates in the linear regression model, there was a significant association between diabetes mellitus and treatment duration. On average, diabetics were treated 1 day (95% confidence interval: 0.2‐1.7 days) longer than nondiabetics.

Compared with nondiabetics, diabetics were more likely to have an outpatient follow‐up visit (73% vs 61%, P=0.002) and to be rehospitalized for any reason after discharge (16% vs 9%, P=0.02) (Table 4). Diabetics were overall more likely to be classified as clinical failure (15% vs 9%, P=0.02); this difference was driven by the cellulitis subgroup (19% vs 10%, P=0.01).

DISCUSSION

Diabetes mellitus is a common comorbidity in patients with acute bacterial skin infections. In this large cohort of patients hospitalized for cellulitis or cutaneous abscess, where those with infected ulcers or deep tissue infections were excluded, microbiological findings in cases associated with positive cultures were similar among diabetics and nondiabetics. Although aerobic gram‐negative microorganisms were not more likely to be identified in diabetics, diabetics were significantly more likely to be exposed to at least 2 calendar days of antibiotics with broad gram‐negative activity. After adjusting for covariates, diabetes mellitus was independently associated with exposure to broad gram‐negative therapy.

To our knowledge, this is the first study to compare the microbiology of cellulitis and cutaneous abscess among diabetics and nondiabetics. Lipsky and colleagues previously described the microbiology of a cohort of diabetic patients hospitalized with a broader range of skin infections including cellulitis, infected ulcers, and surgical site infections.[12] Similar to our findings, gram‐negative pathogens were uncommonly isolated in that study; however, in the absence of a comparator group, whether diabetics were at higher risk for gram‐negative involvement than nondiabetics was not known. Similar to the study by Lipsky and colleagues, most studies of skin infections in diabetics have included a relatively heterogeneous group of infections.[12, 13, 14, 15] The present study therefore contributes to the literature by providing a focused comparison of the microbiology of inpatient cellulitis and abscess in the absence of complicating factors such as an infected ulcer or deep tissue involvement. We found that among cases with a positive culture (13% of cases in the cellulitis group and 73% in the abscess group), the microbiology was similar among diabetics and nondiabetics. Although a microorganism was identified in only a minority of cases of cellulitis, our findings do not support the need for broad gram‐negative therapy in diabetics with cellulitis not associated with an ulcer or deep tissue infection. In diabetics with an abscess, antibiotics with broad gram‐negative activity do not appear to be indicated.

The present study also adds to the literature by providing a detailed comparison of antibiotic utilization patterns among diabetics and nondiabetics. We demonstrated that diabetics were more likely to have significant exposure to antibiotics with broad gram‐negative activity, particularly antipseudomonal agents (the broadest‐spectrum antibiotics). Because initiation of broad gram‐negative therapy in the emergency department or urgent care was not more common among diabetics, the increased use of these agents among diabetics appeared to be driven by inpatient providers. It is also notable that of patients who received any antibiotic with broad gram‐negative activity, these agents accounted for similar proportions of the total days of therapy in both diabetics and nondiabetics. In aggregate, our findings demonstrate that diabetics are more likely to be started on antibiotics with broad gram‐negative activity by inpatient providers, diabetics are not necessarily continued on longer durations of broad gram‐negative therapy once started, and the total amount of exposure to broad gram‐negative agents is substantial.

Overall, our findings suggest that inpatient providers perceive diabetics with cellulitis or abscess to be at increased risk for gram‐negative pathogens. This perhaps reflects an extrapolation of recommendations to use broad‐spectrum empiric therapy in diabetics with certain complicated skin infections.[7] However, for patients with cellulitis or cutaneous abscess, Infectious Diseases Society of America (IDSA) guidelines recommend antibiotic therapy targeted toward S aureus and streptococcal species; there is no suggestion to use a broader spectrum of therapy in diabetics.[8, 9] Our findings therefore highlight an important opportunity to improve antibiotic selection for all patients hospitalized with cellulitis and abscess, but particularly diabetics. It is also noteworthy that by linear regression, diabetes mellitus was independently associated with longer treatment durations. Although the average increase in treatment duration was small (1 day), this finding adds to the evidence that the presence of diabetes mellitus alters providers' treatment approach to cellulitis or abscess.

We found that despite more frequent treatment with broad gram‐negative therapy, diabetics were more likely than nondiabetics to be classified as clinical failures. It is important to point out that diabetics were also more likely than nondiabetics to have postdischarge outpatient follow‐up visits, raising the possibility of biased ascertainment of clinical failure events in this group. However, we also demonstrated that diabetics with cellulitis were more likely to be rehospitalized than nondiabetics. This is similar to a finding by Suaya and colleagues who showed that diabetics with skin infections were about twice as likely to be rehospitalized as nondiabetics.[13] One could hypothesize that the increased frequency of clinical failure events among diabetics was due to their older age, hyperglycemia, or vascular insufficiency; however, other factors may have contributed. For example, providers may have mistaken residual erythema for ongoing or recurrent cellulitis, or the diagnosis of cellulitis could have been incorrect to begin with. Additionally, there may have been uncertainty about the microbiology of cellulitis because the infecting pathogen was not usually identified. These factors may have led to alterations in treatment that would have resulted in a classification of clinical failure, and it is possible that providers had a lower threshold to alter treatment in diabetics. It is therefore not clear whether our findings represent a true difference in clinical outcomes between diabetics or nondiabetics. Regardless, in cases associated with a positive culture, our microbiological results do not support that the difference in clinical failure between diabetics and nondiabetics with cellulitis was related to a different spectrum of microorganisms.

In addition to the limitations outlined previously[2, 10] and above, the present study has at least 5 additional limitations. First, this was a secondary analysis of studies that were not designed to evaluate the effect of diabetes mellitus on the microbiology and treatment of skin infections. For example, hemoglobin A1C values were not collected; therefore, we could not examine whether the microbiology and antibiotic prescribing practices differed based on control of diabetes mellitus. Second, there were minor differences in inclusion and exclusion criteria between the 2 cohorts included in this study. Because the proportion of patients with diabetes mellitus was similar among both cohorts, and comparisons were not made between the cohorts, this should not have impacted our results. Third, the broad categorization of cellulitis used when combining the 2 cohorts raised the possibility of differences in infection characteristics between diabetics and nondiabetics (eg, presence of a wound) that could have confounded our findings regarding use of gram‐negative therapy. In the larger of the 2 cohorts from which the combined cohort was derived, only 17 (3%) of 533 patients had wound infections, whereas those with infected ulcers or suspected deep‐tissue infection were excluded from both cohorts. Furthermore, in the combined cohort, the increased frequency of broad gram‐negative therapy among diabetics was also observed in the cutaneous abscess group. It is therefore unlikely that the categorization of cellulitis had a significant impact on our results. Fourth, given the observational nature of the study, the microbiological data were subject to limitations. Importantly, because the infecting pathogen was identified in only 13% of cases of cellulitis, firm conclusions regarding the microbiology of cellulitis cannot be drawn. Finally, the small number of gram‐negative organisms isolated precluded comparisons of specific pathogens among diabetics and nondiabetics. In addition, because a number of gram‐negative organisms were isolated from wound cultures, it is not known whether they were clinically relevant or simply represented colonization.

In conclusion, in cases of cellulitis or abscess associated with a positive culture, gram‐negative microorganisms were not isolated more commonly among diabetics compared with nondiabetics. However, in general, diabetics were more likely to be treated with broad gram‐negative therapy suggesting that, particularly for cutaneous abscesses, this prescribing practice may not be warranted. These findings support current IDSA guidelines that recommend antibiotic therapy targeted toward gram‐positive pathogens for cellulitis or abscess, irrespective of the presence of diabetes mellitus.[8, 9] Because nearly one‐fourth of patients hospitalized with cellulitis or abscess are diabetic, these findings have relevance for national antimicrobial stewardship efforts aimed at curbing antimicrobial resistance through reducing use of antibiotics with broad gram‐negative activity in hospitals.[16]

Disclosures: This work was supported by the National Institute of Allergy and Infectious Diseases, National Institute of Health (TCJ: K23 AI099082). D.M.P. reports potential conflicts of interests with Optimer, Cubist, and Forest Pharmaceuticals. The authors report no other conflicts of interest.

Diabetes mellitus is one of the most common comorbid conditions among patients hospitalized for acute bacterial skin infections.[1, 2, 3, 4, 5, 6] Acute bacterial skin infections in diabetics represent a spectrum of conditions ranging from cellulitis or cutaneous abscess to more complicated infections such as infected ulcers or deep tissue infections. Although most skin infections in diabetics are caused by gram‐positive pathogens (Staphylococcus aureus and streptococci), the risk of gram‐negative pathogens is increased in certain complicated infections such as diabetic foot infections.[7] For such complicated infections, national guidelines therefore recommend broad‐spectrum empiric antibiotic therapy.[7]

The role of gram‐negative pathogens has not been clearly established in diabetics with cellulitis or cutaneous abscess not associated with an infected ulcer or diabetic foot infection. National guidelines for the treatment of cellulitis and abscess recommend antibiotic therapy targeted toward S aureus and streptococcal species irrespective of the presence of diabetes mellitus.[8, 9] However, in a recent multicenter study of patients hospitalized with acute bacterial skin infections in which cases involving infected ulcers or deep tissue infection were excluded, diabetes mellitus was an independent predictor of use of antibiotics with broad gram‐negative activity.[2] This suggests that either gram‐negative pathogens are more common or providers perceive gram‐negative pathogens to be more common among diabetics with otherwise uncomplicated cellulitis or abscess.

A better understanding of the relationship between the microbiology and antibiotic prescribing practices for diabetics with cellulitis or abscess is therefore necessary to promote the most appropriate spectrum of therapy for these patients. We evaluated a large cohort of patients hospitalized with acute bacterial skin infections in order to: (1) compare the microbiology of diabetics and nondiabetics with cellulitis or cutaneous abscess not associated with an ulcer or deep tissue infection; and (2) compare antibiotic prescribing practices among diabetics and nondiabetics. We hypothesized that diabetics would have a similar spectrum of microorganisms as nondiabetics but would be more frequently treated with antibiotics with broad gram‐negative activity.

METHODS

RESULTS

After excluding 102 pediatric cases and removing 5 duplicate cases, 770 total cases were included for analysis: 447 involved cellulitis and 323 involved cutaneous abscess (Figure 1). Overall, 167 (22%) patients had diabetes mellitus. Diabetics were significantly more likely than nondiabetics to have cellulitis as the presenting infection (67% of cases vs 56%, P=0.008) and to have lower extremity involvement (48% vs 33%, P<0.001) (Table 1). Diabetics were also older (median age 55 years vs 48 years, P<0.001), more likely to have cirrhosis or prior skin infection, and less likely to be injection‐drug users or human immunodeficiency virus (HIV) infected. Demographic and clinical characteristics among diabetics and nondiabetics stratified by the categorizations of cellulitis and cutaneous abscess are presented in the Supporting Information, Appendix Table 1, in the online version of this article.

Figure 1

The frequency of use of microbiological cultures was similar among diabetics and nondiabetics (Table 2). In cases of cellulitis, a microorganism was identified in 18% of diabetics and 12% of nondiabetics (P=0.09). In cases of cutaneous abscess, a microorganism was identified more commonly (69% and 74%, respectively, P=0.50). Among cases where a microorganism was identified, aerobic gram‐positive organisms were isolated in 90% of diabetics and 92% of nondiabetics (P=0.59). Aerobic gram‐negative organisms were isolated in 7% of diabetics and 12% of nondiabetics (P=0.28). Specific gram‐negative organisms isolated are shown in the Supporting Information, Appendix Table 2, in the online version of this article; no cases in diabetics involved Pseudomonas aeruginosa. The comparison of microbiological data among diabetics and nondiabetics was similar when stratified by cellulitis versus cutaneous abscess (Table 2).

Antibiotic utilization is summarized in Table 3. Among patients who were started on antibiotic therapy in the emergency department or urgent care, the initial regimen included an agent with broad gram‐negative activity in 31% of both diabetics and nondiabetics (P=0.97). During the entire hospital stay (including the emergency department or urgent care), diabetics were significantly more likely to be treated with ‐lactam/‐lactamase inhibitor combinations (42% vs 33%, P=0.04). At the time of hospital discharge, diabetics were more likely to be prescribed fluoroquinolones (11% vs 5%, P=0.01) (Table 3) particularly for cases of cellulitis (13% vs 6%, P=0.008) (see Supporting Information, Appendix Table 3, in the online version of this article). Diabetics were somewhat more likely to be prescribed parenteral antibiotics (10% vs 6%, P=0.07) after discharge. When considering both inpatient and discharge therapy, more diabetics than nondiabetics were exposed to at least 2 calendar days of broad gram‐negative therapy (54% vs 44%, P=0.02) and more were prescribed an antipseudomonal agent (38% vs 25%, P=0.002). In the group of patients who received at least 1 dose of an antibiotic with broad gram‐negative activity, broad gram‐negative agents accounted for 33% of the total days of therapy prescribed for diabetics and 32% for nondiabetics. Overall prescribing patterns were similar when stratified by cellulitis versus cutaneous abscess (see Supporting Information, Appendix Table 3, in the online version of this article).

After adjusting for covariates in the logistic regression model, diabetes mellitus was an independent predictor of exposure to broad gram‐negative therapy (see Supporting Information, Appendix Table 4, in the online version of this article). In addition to diabetes mellitus, culture of an aerobic gram‐negative microorganism, infectious diseases service consultation, presence of fever, and nonmedical admitting services were significantly associated with exposure to broad gram‐negative therapy. Prior methicillin‐resistant S aureus infection or colonization and HIV infection were inversely associated. Compared with nondiabetics, the total duration of antibiotic therapy in diabetics was somewhat longer (median 13 days vs 12 days, P=0.09) (Table 3). After adjusting for covariates in the linear regression model, there was a significant association between diabetes mellitus and treatment duration. On average, diabetics were treated 1 day (95% confidence interval: 0.2‐1.7 days) longer than nondiabetics.

Compared with nondiabetics, diabetics were more likely to have an outpatient follow‐up visit (73% vs 61%, P=0.002) and to be rehospitalized for any reason after discharge (16% vs 9%, P=0.02) (Table 4). Diabetics were overall more likely to be classified as clinical failure (15% vs 9%, P=0.02); this difference was driven by the cellulitis subgroup (19% vs 10%, P=0.01).

DISCUSSION

Diabetes mellitus is a common comorbidity in patients with acute bacterial skin infections. In this large cohort of patients hospitalized for cellulitis or cutaneous abscess, where those with infected ulcers or deep tissue infections were excluded, microbiological findings in cases associated with positive cultures were similar among diabetics and nondiabetics. Although aerobic gram‐negative microorganisms were not more likely to be identified in diabetics, diabetics were significantly more likely to be exposed to at least 2 calendar days of antibiotics with broad gram‐negative activity. After adjusting for covariates, diabetes mellitus was independently associated with exposure to broad gram‐negative therapy.

To our knowledge, this is the first study to compare the microbiology of cellulitis and cutaneous abscess among diabetics and nondiabetics. Lipsky and colleagues previously described the microbiology of a cohort of diabetic patients hospitalized with a broader range of skin infections including cellulitis, infected ulcers, and surgical site infections.[12] Similar to our findings, gram‐negative pathogens were uncommonly isolated in that study; however, in the absence of a comparator group, whether diabetics were at higher risk for gram‐negative involvement than nondiabetics was not known. Similar to the study by Lipsky and colleagues, most studies of skin infections in diabetics have included a relatively heterogeneous group of infections.[12, 13, 14, 15] The present study therefore contributes to the literature by providing a focused comparison of the microbiology of inpatient cellulitis and abscess in the absence of complicating factors such as an infected ulcer or deep tissue involvement. We found that among cases with a positive culture (13% of cases in the cellulitis group and 73% in the abscess group), the microbiology was similar among diabetics and nondiabetics. Although a microorganism was identified in only a minority of cases of cellulitis, our findings do not support the need for broad gram‐negative therapy in diabetics with cellulitis not associated with an ulcer or deep tissue infection. In diabetics with an abscess, antibiotics with broad gram‐negative activity do not appear to be indicated.

The present study also adds to the literature by providing a detailed comparison of antibiotic utilization patterns among diabetics and nondiabetics. We demonstrated that diabetics were more likely to have significant exposure to antibiotics with broad gram‐negative activity, particularly antipseudomonal agents (the broadest‐spectrum antibiotics). Because initiation of broad gram‐negative therapy in the emergency department or urgent care was not more common among diabetics, the increased use of these agents among diabetics appeared to be driven by inpatient providers. It is also notable that of patients who received any antibiotic with broad gram‐negative activity, these agents accounted for similar proportions of the total days of therapy in both diabetics and nondiabetics. In aggregate, our findings demonstrate that diabetics are more likely to be started on antibiotics with broad gram‐negative activity by inpatient providers, diabetics are not necessarily continued on longer durations of broad gram‐negative therapy once started, and the total amount of exposure to broad gram‐negative agents is substantial.

Overall, our findings suggest that inpatient providers perceive diabetics with cellulitis or abscess to be at increased risk for gram‐negative pathogens. This perhaps reflects an extrapolation of recommendations to use broad‐spectrum empiric therapy in diabetics with certain complicated skin infections.[7] However, for patients with cellulitis or cutaneous abscess, Infectious Diseases Society of America (IDSA) guidelines recommend antibiotic therapy targeted toward S aureus and streptococcal species; there is no suggestion to use a broader spectrum of therapy in diabetics.[8, 9] Our findings therefore highlight an important opportunity to improve antibiotic selection for all patients hospitalized with cellulitis and abscess, but particularly diabetics. It is also noteworthy that by linear regression, diabetes mellitus was independently associated with longer treatment durations. Although the average increase in treatment duration was small (1 day), this finding adds to the evidence that the presence of diabetes mellitus alters providers' treatment approach to cellulitis or abscess.

We found that despite more frequent treatment with broad gram‐negative therapy, diabetics were more likely than nondiabetics to be classified as clinical failures. It is important to point out that diabetics were also more likely than nondiabetics to have postdischarge outpatient follow‐up visits, raising the possibility of biased ascertainment of clinical failure events in this group. However, we also demonstrated that diabetics with cellulitis were more likely to be rehospitalized than nondiabetics. This is similar to a finding by Suaya and colleagues who showed that diabetics with skin infections were about twice as likely to be rehospitalized as nondiabetics.[13] One could hypothesize that the increased frequency of clinical failure events among diabetics was due to their older age, hyperglycemia, or vascular insufficiency; however, other factors may have contributed. For example, providers may have mistaken residual erythema for ongoing or recurrent cellulitis, or the diagnosis of cellulitis could have been incorrect to begin with. Additionally, there may have been uncertainty about the microbiology of cellulitis because the infecting pathogen was not usually identified. These factors may have led to alterations in treatment that would have resulted in a classification of clinical failure, and it is possible that providers had a lower threshold to alter treatment in diabetics. It is therefore not clear whether our findings represent a true difference in clinical outcomes between diabetics or nondiabetics. Regardless, in cases associated with a positive culture, our microbiological results do not support that the difference in clinical failure between diabetics and nondiabetics with cellulitis was related to a different spectrum of microorganisms.

In addition to the limitations outlined previously[2, 10] and above, the present study has at least 5 additional limitations. First, this was a secondary analysis of studies that were not designed to evaluate the effect of diabetes mellitus on the microbiology and treatment of skin infections. For example, hemoglobin A1C values were not collected; therefore, we could not examine whether the microbiology and antibiotic prescribing practices differed based on control of diabetes mellitus. Second, there were minor differences in inclusion and exclusion criteria between the 2 cohorts included in this study. Because the proportion of patients with diabetes mellitus was similar among both cohorts, and comparisons were not made between the cohorts, this should not have impacted our results. Third, the broad categorization of cellulitis used when combining the 2 cohorts raised the possibility of differences in infection characteristics between diabetics and nondiabetics (eg, presence of a wound) that could have confounded our findings regarding use of gram‐negative therapy. In the larger of the 2 cohorts from which the combined cohort was derived, only 17 (3%) of 533 patients had wound infections, whereas those with infected ulcers or suspected deep‐tissue infection were excluded from both cohorts. Furthermore, in the combined cohort, the increased frequency of broad gram‐negative therapy among diabetics was also observed in the cutaneous abscess group. It is therefore unlikely that the categorization of cellulitis had a significant impact on our results. Fourth, given the observational nature of the study, the microbiological data were subject to limitations. Importantly, because the infecting pathogen was identified in only 13% of cases of cellulitis, firm conclusions regarding the microbiology of cellulitis cannot be drawn. Finally, the small number of gram‐negative organisms isolated precluded comparisons of specific pathogens among diabetics and nondiabetics. In addition, because a number of gram‐negative organisms were isolated from wound cultures, it is not known whether they were clinically relevant or simply represented colonization.

In conclusion, in cases of cellulitis or abscess associated with a positive culture, gram‐negative microorganisms were not isolated more commonly among diabetics compared with nondiabetics. However, in general, diabetics were more likely to be treated with broad gram‐negative therapy suggesting that, particularly for cutaneous abscesses, this prescribing practice may not be warranted. These findings support current IDSA guidelines that recommend antibiotic therapy targeted toward gram‐positive pathogens for cellulitis or abscess, irrespective of the presence of diabetes mellitus.[8, 9] Because nearly one‐fourth of patients hospitalized with cellulitis or abscess are diabetic, these findings have relevance for national antimicrobial stewardship efforts aimed at curbing antimicrobial resistance through reducing use of antibiotics with broad gram‐negative activity in hospitals.[16]

Disclosures: This work was supported by the National Institute of Allergy and Infectious Diseases, National Institute of Health (TCJ: K23 AI099082). D.M.P. reports potential conflicts of interests with Optimer, Cubist, and Forest Pharmaceuticals. The authors report no other conflicts of interest.

Community‐acquired pneumonia (CAP) is a common and serious infection in children. With more than 150,000 children requiring hospitalization annually, CAP is the fifth most prevalent and the second most costly diagnosis of all pediatric hospitalizations in the United States.[1, 2, 3]

In August 2011, the Pediatric Infectious Diseases Society (PIDS) and the Infectious Diseases Society of America (IDSA) published an evidence‐based guideline for the management of CAP in children. This guideline recommended that fully immunized children without underlying complications who require hospitalization receive an aminopenicillin as first‐line antibiotic therapy.[4] Additionally, the guideline recommends empirically adding a macrolide to an aminopenicillin when atypical pneumonia is a diagnostic consideration.

This recommendation was a substantial departure from practice for hospitals nationwide, as a multicenter study of children's hospitals (20052010) demonstrated that <10% of patients diagnosed with CAP received aminopenicillins as empiric therapy.[5] Since publication of the PIDS/IDSA guidelines, the use of aminopenicillins has increased significantly across institutions, but the majority of hospitalized patients still receive broad‐spectrum cephalosporin therapy for CAP.[6]

At baseline, 30% of patients hospitalized with CAP received guideline‐recommended antibiotic therapy at our institution. Through the use of quality‐improvement methods, the proportion of patients receiving guideline‐recommended therapy increased to 100%.[7] The objective of this study was to ensure that there were not unintended negative consequences to guideline implementation. Specifically, we sought to identify changes in length of stay (LOS), hospital costs, and treatment failures associated with use of guideline‐recommended antibiotic therapy for children hospitalized with uncomplicated CAP.

METHODS

RESULTS

Of the 220 unique patients included, 122 (55%) were male. The median age was 2.9 years (IQR: 1.36.3 years). Empiric guideline‐recommended therapy was prescribed to 168 (76%) patients (Table 1). Aminopenicillins were the most common guideline‐recommended therapy, accounting for 84% of guideline‐recommended antibiotics. An additional 10% of patients received the guideline‐recommended combination therapy with an aminopenicillin and a macrolide. Nonguideline‐recommended therapy included third‐generation cephalosporin antibiotics (54%) and macrolide monotherapy (33%).

Those who received empiric guideline‐recommended antibiotic therapy were similar to those who received nonguideline‐recommended therapy with respect to sex, Emergency Severity Index, physical exam findings on presentation, oxygen requirement in the first 24 hours, abnormal laboratory findings, presence of effusion on chest radiograph, and need for additional imaging (Table 1). However, patients in the guideline‐recommended therapy group were significantly younger (median 2.5 years vs 5.6 years, P0.01), more likely to have elevated respiratory rate on presentation (60.2% vs 44.4%, P=0.04), and more likely to have an infiltrate on chest radiograph (86.3% vs 73.6%, P=0.03) (Table 1). Patients who received nonguideline‐recommended macrolide monotherapy had a median age of 7.4 years (IQR: 5.89.8 years).

Median hospital LOS for the total cohort was 1.3 days (IQR: 0.91.9 days) (Table 2). There were no differences in LOS between patients who received and did not receive guideline‐recommended therapy in the unadjusted or the adjusted model (Table 3).

Median total costs of the index hospitalization for the total cohort were $4097 (IQR: $2657$6054), with median inpatient pharmacy costs of $92 (IQR: $40$183) (Table 2). There were no differences in total or inpatient pharmacy costs for patients who received guideline‐recommended therapy compared with those who did not in unadjusted or adjusted analyses. Fourteen patients (6.4%) had antibiotic therapy broadened during hospitalization, 10 were initially prescribed guideline‐recommended therapy, and 4 were initially prescribed nonguideline‐recommended therapy (Table 4).

Of the 9 pneumonia‐related ED revisits within 30 days of discharge, 7 occurred in patients prescribed empiric guideline‐recommended therapy (Table 5). No ED revisit resulted in hospital readmission or antibiotic change related to pneumonia. Two ED revisits resulted in new antibiotic prescriptions for diagnoses other than pneumonia.

Two patients were readmitted for a pneumonia‐related illness within 30 days of discharge; 1 had received guideline‐recommended therapy (Table 5). Both patients were directly admitted to the inpatient ward without an associated ED visit. Antibiotic class was not changed for either patient upon readmission, despite the decision to convert to intravenous form.

DISCUSSION

In this retrospective cohort study, patients who received empiric guideline‐recommended antibiotic therapy on admission for CAP had no difference in LOS, total cost of hospitalization, or inpatient pharmacy costs compared with those who received therapy that varied from guideline recommendations. Our study suggests that prescribing narrow‐spectrum therapy and, in some circumstances, combination therapy, as recommended by the 2011 PIDS/IDSA pneumonia guideline, did not result in negative unintended consequences.

In our study, children receiving guideline‐recommended therapy were younger, more likely to have elevated respiratory rate on presentation, and more likely to have an infiltrate on chest radiograph. We hypothesize the age difference is a reflection of common use of nonguideline macrolide monotherapy in the older, school‐age child, as macrolides are commonly used for coverage of Mycoplasma pneumonia in older children with CAP.[15] Children receiving macrolide monotherapy were older than those receiving guideline‐recommended therapy (median age of 7.4 years and 2.5 years, respectively). We also hypothesize that some providers may prescribe macrolide monotherapy to children deemed less ill than expected for CAP (eg, normal percutaneous oxygen saturation). This hypothesis is supported by the finding that 60% of patients who had a normal respiratory rate and received nonguideline therapy were prescribed macrolide monotherapy. We did control for the characteristics that varied between the two treatment groups in our models to eliminate potential confounding.

One prior study evaluated the effects of guideline implementation in CAP. In evaluation of a clinical practice guideline that recommended ampicillin as first‐line therapy for CAP, no significant difference was found following guideline introduction in the number of treatment failures (defined as the need to broaden therapy or development of complicated pneumonia within 48 hours or 30‐day inpatient readmission).[16] Our study builds on these findings by directly comparing outcomes between recipients of guideline and nonguideline therapy, which was not done in the prestudy or poststudy design of prior work.[16] We believe that classifying patients based on empiric therapy received rather than timing of guideline introduction thoroughly examines the effect of following guideline recommendations. Additionally, outcomes other than treatment failures were examined in our study including LOS, costs, and ED revisits.

Our results are similar to other observational studies that compared narrow‐ and broad‐spectrum antibiotics for children hospitalized with CAP. Using administrative data, Williams et al. found no significant difference in LOS or cost between narrow‐ and broad‐spectrum intravenous antibiotic recipients ages 6 months to 18 years at 43 children's hospitals.[17] Queen et al. compared narrow‐ and broad‐spectrum antibiotic therapy for children ages 2 months to 18 years at 4 children's hospitals.[18] Differences in average daily cost were not significant, but children who received narrow‐spectrum antibiotics had a significantly shorter LOS. Finally, an observational study of 319 Israeli children <2 years of age found no significant difference in duration of oxygen requirement, LOS, or need for change in antibiotic therapy between the 66 children prescribed aminopenicillins and the 253 children prescribed cefuroxime.[19] These studies suggest that prescribing narrow‐spectrum antimicrobial therapy, as per the guideline recommendations, results in similar outcomes to broad‐spectrum antimicrobial therapy.

Our study adds to prior studies that compared narrow‐ and broad‐spectrum therapy by comparing outcomes associated with guideline‐recommended therapy to nonguideline therapy. We considered antibiotic therapy as guideline recommended if appropriately chosen per the guideline, not just by simple classification of antibiotic. For example, the use of a cephalosporin in a patient with aminopenicillin allergy was considered guideline‐recommended therapy. We chose to classify the exposure of empiric antibiotic therapy in this manner to reflect true clinical application of the guideline, as not all children are able to receive aminopenicillins. Additionally, as our study stemmed from our prior improvement work aimed at increasing guideline adherent therapy,[7] we have a higher frequency of narrow‐spectrum antibiotic use than prior studies. In our study, almost two‐thirds of patients received narrow‐spectrum therapy, whereas narrow‐spectrum use in prior studies ranged from 10% to 33%.[17, 18, 19] Finally, we were able to confirm the diagnosis of pneumonia via medical record review and to adjust for severity of illness using clinical variables including vital signs, physical exam findings, and laboratory and radiologic study results.

This study must be interpreted in the context of several limitations. First, our study population was defined through discharge diagnosis codes, and therefore dependent on the accuracy of coding. However, we minimized potential for misclassification through use of a previously validated approach to identify patients with CAP and through medical record review to confirm the diagnosis. Second, we may be unable to detect very small differences in outcomes given limited power, specifically for outcomes of LOS, ED revisits, hospital readmissions, and need to broaden antibiotic therapy. Third, residual confounding may be present. Although we controlled for many clinical variables in our analyses, antibiotic prescribing practices may be influenced by unmeasured factors. The potential of system level confounding is mitigated by standardized care for patients with CAP at our institution. Prior system‐level changes using quality‐improvement science have resulted in a high level of adherence with guideline‐recommended antimicrobials as well as standardized medical discharge criteria.[7, 20] Additionally, nonmedical factors may influence LOS, limiting its use as an outcome measure. This limitation was minimized in our study by standardizing medical discharge criteria. Prior work at our institution demonstrated that the majority of patients, including those with CAP, were discharged within 2 hours of meeting medical discharge criteria.[20] Fourth, discharged patients who experienced adverse outcomes may have received care at a nearby adult emergency department or at their pediatrician's office. Although these events would not have been captured in our electronic health record, serious complications due to treatment failure (eg, empyema) would require hospitalization. As our hospital is the only children's hospital in the Greater Cincinnati metropolitan area, these patents would receive care at our institution. Therefore, any misclassification of revisits or readmissions is likely to be minimal. Finally, study patients were admitted to a large tertiary academic children's hospital, warranting further investigation to determine if these findings can be translated to smaller community settings.

In conclusion, receipt of guideline‐recommended antibiotic therapy for patients hospitalized with CAP was not associated with increases in LOS, total costs of hospitalization, or inpatient pharmacy costs. Our findings highlight the importance of changing antibiotic prescribing practices to reflect guideline recommendations, as there was no evidence of negative unintended consequences with our local practice change.

Community‐acquired pneumonia (CAP) is a common and serious infection in children. With more than 150,000 children requiring hospitalization annually, CAP is the fifth most prevalent and the second most costly diagnosis of all pediatric hospitalizations in the United States.[1, 2, 3]

In August 2011, the Pediatric Infectious Diseases Society (PIDS) and the Infectious Diseases Society of America (IDSA) published an evidence‐based guideline for the management of CAP in children. This guideline recommended that fully immunized children without underlying complications who require hospitalization receive an aminopenicillin as first‐line antibiotic therapy.[4] Additionally, the guideline recommends empirically adding a macrolide to an aminopenicillin when atypical pneumonia is a diagnostic consideration.

This recommendation was a substantial departure from practice for hospitals nationwide, as a multicenter study of children's hospitals (20052010) demonstrated that <10% of patients diagnosed with CAP received aminopenicillins as empiric therapy.[5] Since publication of the PIDS/IDSA guidelines, the use of aminopenicillins has increased significantly across institutions, but the majority of hospitalized patients still receive broad‐spectrum cephalosporin therapy for CAP.[6]

At baseline, 30% of patients hospitalized with CAP received guideline‐recommended antibiotic therapy at our institution. Through the use of quality‐improvement methods, the proportion of patients receiving guideline‐recommended therapy increased to 100%.[7] The objective of this study was to ensure that there were not unintended negative consequences to guideline implementation. Specifically, we sought to identify changes in length of stay (LOS), hospital costs, and treatment failures associated with use of guideline‐recommended antibiotic therapy for children hospitalized with uncomplicated CAP.

METHODS

RESULTS

Of the 220 unique patients included, 122 (55%) were male. The median age was 2.9 years (IQR: 1.36.3 years). Empiric guideline‐recommended therapy was prescribed to 168 (76%) patients (Table 1). Aminopenicillins were the most common guideline‐recommended therapy, accounting for 84% of guideline‐recommended antibiotics. An additional 10% of patients received the guideline‐recommended combination therapy with an aminopenicillin and a macrolide. Nonguideline‐recommended therapy included third‐generation cephalosporin antibiotics (54%) and macrolide monotherapy (33%).

Those who received empiric guideline‐recommended antibiotic therapy were similar to those who received nonguideline‐recommended therapy with respect to sex, Emergency Severity Index, physical exam findings on presentation, oxygen requirement in the first 24 hours, abnormal laboratory findings, presence of effusion on chest radiograph, and need for additional imaging (Table 1). However, patients in the guideline‐recommended therapy group were significantly younger (median 2.5 years vs 5.6 years, P0.01), more likely to have elevated respiratory rate on presentation (60.2% vs 44.4%, P=0.04), and more likely to have an infiltrate on chest radiograph (86.3% vs 73.6%, P=0.03) (Table 1). Patients who received nonguideline‐recommended macrolide monotherapy had a median age of 7.4 years (IQR: 5.89.8 years).

Median hospital LOS for the total cohort was 1.3 days (IQR: 0.91.9 days) (Table 2). There were no differences in LOS between patients who received and did not receive guideline‐recommended therapy in the unadjusted or the adjusted model (Table 3).

Median total costs of the index hospitalization for the total cohort were $4097 (IQR: $2657$6054), with median inpatient pharmacy costs of $92 (IQR: $40$183) (Table 2). There were no differences in total or inpatient pharmacy costs for patients who received guideline‐recommended therapy compared with those who did not in unadjusted or adjusted analyses. Fourteen patients (6.4%) had antibiotic therapy broadened during hospitalization, 10 were initially prescribed guideline‐recommended therapy, and 4 were initially prescribed nonguideline‐recommended therapy (Table 4).

Of the 9 pneumonia‐related ED revisits within 30 days of discharge, 7 occurred in patients prescribed empiric guideline‐recommended therapy (Table 5). No ED revisit resulted in hospital readmission or antibiotic change related to pneumonia. Two ED revisits resulted in new antibiotic prescriptions for diagnoses other than pneumonia.

Two patients were readmitted for a pneumonia‐related illness within 30 days of discharge; 1 had received guideline‐recommended therapy (Table 5). Both patients were directly admitted to the inpatient ward without an associated ED visit. Antibiotic class was not changed for either patient upon readmission, despite the decision to convert to intravenous form.

DISCUSSION

In this retrospective cohort study, patients who received empiric guideline‐recommended antibiotic therapy on admission for CAP had no difference in LOS, total cost of hospitalization, or inpatient pharmacy costs compared with those who received therapy that varied from guideline recommendations. Our study suggests that prescribing narrow‐spectrum therapy and, in some circumstances, combination therapy, as recommended by the 2011 PIDS/IDSA pneumonia guideline, did not result in negative unintended consequences.

In our study, children receiving guideline‐recommended therapy were younger, more likely to have elevated respiratory rate on presentation, and more likely to have an infiltrate on chest radiograph. We hypothesize the age difference is a reflection of common use of nonguideline macrolide monotherapy in the older, school‐age child, as macrolides are commonly used for coverage of Mycoplasma pneumonia in older children with CAP.[15] Children receiving macrolide monotherapy were older than those receiving guideline‐recommended therapy (median age of 7.4 years and 2.5 years, respectively). We also hypothesize that some providers may prescribe macrolide monotherapy to children deemed less ill than expected for CAP (eg, normal percutaneous oxygen saturation). This hypothesis is supported by the finding that 60% of patients who had a normal respiratory rate and received nonguideline therapy were prescribed macrolide monotherapy. We did control for the characteristics that varied between the two treatment groups in our models to eliminate potential confounding.

One prior study evaluated the effects of guideline implementation in CAP. In evaluation of a clinical practice guideline that recommended ampicillin as first‐line therapy for CAP, no significant difference was found following guideline introduction in the number of treatment failures (defined as the need to broaden therapy or development of complicated pneumonia within 48 hours or 30‐day inpatient readmission).[16] Our study builds on these findings by directly comparing outcomes between recipients of guideline and nonguideline therapy, which was not done in the prestudy or poststudy design of prior work.[16] We believe that classifying patients based on empiric therapy received rather than timing of guideline introduction thoroughly examines the effect of following guideline recommendations. Additionally, outcomes other than treatment failures were examined in our study including LOS, costs, and ED revisits.

Our results are similar to other observational studies that compared narrow‐ and broad‐spectrum antibiotics for children hospitalized with CAP. Using administrative data, Williams et al. found no significant difference in LOS or cost between narrow‐ and broad‐spectrum intravenous antibiotic recipients ages 6 months to 18 years at 43 children's hospitals.[17] Queen et al. compared narrow‐ and broad‐spectrum antibiotic therapy for children ages 2 months to 18 years at 4 children's hospitals.[18] Differences in average daily cost were not significant, but children who received narrow‐spectrum antibiotics had a significantly shorter LOS. Finally, an observational study of 319 Israeli children <2 years of age found no significant difference in duration of oxygen requirement, LOS, or need for change in antibiotic therapy between the 66 children prescribed aminopenicillins and the 253 children prescribed cefuroxime.[19] These studies suggest that prescribing narrow‐spectrum antimicrobial therapy, as per the guideline recommendations, results in similar outcomes to broad‐spectrum antimicrobial therapy.

Our study adds to prior studies that compared narrow‐ and broad‐spectrum therapy by comparing outcomes associated with guideline‐recommended therapy to nonguideline therapy. We considered antibiotic therapy as guideline recommended if appropriately chosen per the guideline, not just by simple classification of antibiotic. For example, the use of a cephalosporin in a patient with aminopenicillin allergy was considered guideline‐recommended therapy. We chose to classify the exposure of empiric antibiotic therapy in this manner to reflect true clinical application of the guideline, as not all children are able to receive aminopenicillins. Additionally, as our study stemmed from our prior improvement work aimed at increasing guideline adherent therapy,[7] we have a higher frequency of narrow‐spectrum antibiotic use than prior studies. In our study, almost two‐thirds of patients received narrow‐spectrum therapy, whereas narrow‐spectrum use in prior studies ranged from 10% to 33%.[17, 18, 19] Finally, we were able to confirm the diagnosis of pneumonia via medical record review and to adjust for severity of illness using clinical variables including vital signs, physical exam findings, and laboratory and radiologic study results.

This study must be interpreted in the context of several limitations. First, our study population was defined through discharge diagnosis codes, and therefore dependent on the accuracy of coding. However, we minimized potential for misclassification through use of a previously validated approach to identify patients with CAP and through medical record review to confirm the diagnosis. Second, we may be unable to detect very small differences in outcomes given limited power, specifically for outcomes of LOS, ED revisits, hospital readmissions, and need to broaden antibiotic therapy. Third, residual confounding may be present. Although we controlled for many clinical variables in our analyses, antibiotic prescribing practices may be influenced by unmeasured factors. The potential of system level confounding is mitigated by standardized care for patients with CAP at our institution. Prior system‐level changes using quality‐improvement science have resulted in a high level of adherence with guideline‐recommended antimicrobials as well as standardized medical discharge criteria.[7, 20] Additionally, nonmedical factors may influence LOS, limiting its use as an outcome measure. This limitation was minimized in our study by standardizing medical discharge criteria. Prior work at our institution demonstrated that the majority of patients, including those with CAP, were discharged within 2 hours of meeting medical discharge criteria.[20] Fourth, discharged patients who experienced adverse outcomes may have received care at a nearby adult emergency department or at their pediatrician's office. Although these events would not have been captured in our electronic health record, serious complications due to treatment failure (eg, empyema) would require hospitalization. As our hospital is the only children's hospital in the Greater Cincinnati metropolitan area, these patents would receive care at our institution. Therefore, any misclassification of revisits or readmissions is likely to be minimal. Finally, study patients were admitted to a large tertiary academic children's hospital, warranting further investigation to determine if these findings can be translated to smaller community settings.

In conclusion, receipt of guideline‐recommended antibiotic therapy for patients hospitalized with CAP was not associated with increases in LOS, total costs of hospitalization, or inpatient pharmacy costs. Our findings highlight the importance of changing antibiotic prescribing practices to reflect guideline recommendations, as there was no evidence of negative unintended consequences with our local practice change.

Community‐acquired pneumonia (CAP) is a common and serious infection in children. With more than 150,000 children requiring hospitalization annually, CAP is the fifth most prevalent and the second most costly diagnosis of all pediatric hospitalizations in the United States.[1, 2, 3]

In August 2011, the Pediatric Infectious Diseases Society (PIDS) and the Infectious Diseases Society of America (IDSA) published an evidence‐based guideline for the management of CAP in children. This guideline recommended that fully immunized children without underlying complications who require hospitalization receive an aminopenicillin as first‐line antibiotic therapy.[4] Additionally, the guideline recommends empirically adding a macrolide to an aminopenicillin when atypical pneumonia is a diagnostic consideration.

This recommendation was a substantial departure from practice for hospitals nationwide, as a multicenter study of children's hospitals (20052010) demonstrated that <10% of patients diagnosed with CAP received aminopenicillins as empiric therapy.[5] Since publication of the PIDS/IDSA guidelines, the use of aminopenicillins has increased significantly across institutions, but the majority of hospitalized patients still receive broad‐spectrum cephalosporin therapy for CAP.[6]

At baseline, 30% of patients hospitalized with CAP received guideline‐recommended antibiotic therapy at our institution. Through the use of quality‐improvement methods, the proportion of patients receiving guideline‐recommended therapy increased to 100%.[7] The objective of this study was to ensure that there were not unintended negative consequences to guideline implementation. Specifically, we sought to identify changes in length of stay (LOS), hospital costs, and treatment failures associated with use of guideline‐recommended antibiotic therapy for children hospitalized with uncomplicated CAP.

METHODS

RESULTS

Of the 220 unique patients included, 122 (55%) were male. The median age was 2.9 years (IQR: 1.36.3 years). Empiric guideline‐recommended therapy was prescribed to 168 (76%) patients (Table 1). Aminopenicillins were the most common guideline‐recommended therapy, accounting for 84% of guideline‐recommended antibiotics. An additional 10% of patients received the guideline‐recommended combination therapy with an aminopenicillin and a macrolide. Nonguideline‐recommended therapy included third‐generation cephalosporin antibiotics (54%) and macrolide monotherapy (33%).

Those who received empiric guideline‐recommended antibiotic therapy were similar to those who received nonguideline‐recommended therapy with respect to sex, Emergency Severity Index, physical exam findings on presentation, oxygen requirement in the first 24 hours, abnormal laboratory findings, presence of effusion on chest radiograph, and need for additional imaging (Table 1). However, patients in the guideline‐recommended therapy group were significantly younger (median 2.5 years vs 5.6 years, P0.01), more likely to have elevated respiratory rate on presentation (60.2% vs 44.4%, P=0.04), and more likely to have an infiltrate on chest radiograph (86.3% vs 73.6%, P=0.03) (Table 1). Patients who received nonguideline‐recommended macrolide monotherapy had a median age of 7.4 years (IQR: 5.89.8 years).

Median hospital LOS for the total cohort was 1.3 days (IQR: 0.91.9 days) (Table 2). There were no differences in LOS between patients who received and did not receive guideline‐recommended therapy in the unadjusted or the adjusted model (Table 3).

Median total costs of the index hospitalization for the total cohort were $4097 (IQR: $2657$6054), with median inpatient pharmacy costs of $92 (IQR: $40$183) (Table 2). There were no differences in total or inpatient pharmacy costs for patients who received guideline‐recommended therapy compared with those who did not in unadjusted or adjusted analyses. Fourteen patients (6.4%) had antibiotic therapy broadened during hospitalization, 10 were initially prescribed guideline‐recommended therapy, and 4 were initially prescribed nonguideline‐recommended therapy (Table 4).

Of the 9 pneumonia‐related ED revisits within 30 days of discharge, 7 occurred in patients prescribed empiric guideline‐recommended therapy (Table 5). No ED revisit resulted in hospital readmission or antibiotic change related to pneumonia. Two ED revisits resulted in new antibiotic prescriptions for diagnoses other than pneumonia.

Two patients were readmitted for a pneumonia‐related illness within 30 days of discharge; 1 had received guideline‐recommended therapy (Table 5). Both patients were directly admitted to the inpatient ward without an associated ED visit. Antibiotic class was not changed for either patient upon readmission, despite the decision to convert to intravenous form.

DISCUSSION

In this retrospective cohort study, patients who received empiric guideline‐recommended antibiotic therapy on admission for CAP had no difference in LOS, total cost of hospitalization, or inpatient pharmacy costs compared with those who received therapy that varied from guideline recommendations. Our study suggests that prescribing narrow‐spectrum therapy and, in some circumstances, combination therapy, as recommended by the 2011 PIDS/IDSA pneumonia guideline, did not result in negative unintended consequences.

In our study, children receiving guideline‐recommended therapy were younger, more likely to have elevated respiratory rate on presentation, and more likely to have an infiltrate on chest radiograph. We hypothesize the age difference is a reflection of common use of nonguideline macrolide monotherapy in the older, school‐age child, as macrolides are commonly used for coverage of Mycoplasma pneumonia in older children with CAP.[15] Children receiving macrolide monotherapy were older than those receiving guideline‐recommended therapy (median age of 7.4 years and 2.5 years, respectively). We also hypothesize that some providers may prescribe macrolide monotherapy to children deemed less ill than expected for CAP (eg, normal percutaneous oxygen saturation). This hypothesis is supported by the finding that 60% of patients who had a normal respiratory rate and received nonguideline therapy were prescribed macrolide monotherapy. We did control for the characteristics that varied between the two treatment groups in our models to eliminate potential confounding.

One prior study evaluated the effects of guideline implementation in CAP. In evaluation of a clinical practice guideline that recommended ampicillin as first‐line therapy for CAP, no significant difference was found following guideline introduction in the number of treatment failures (defined as the need to broaden therapy or development of complicated pneumonia within 48 hours or 30‐day inpatient readmission).[16] Our study builds on these findings by directly comparing outcomes between recipients of guideline and nonguideline therapy, which was not done in the prestudy or poststudy design of prior work.[16] We believe that classifying patients based on empiric therapy received rather than timing of guideline introduction thoroughly examines the effect of following guideline recommendations. Additionally, outcomes other than treatment failures were examined in our study including LOS, costs, and ED revisits.

Our results are similar to other observational studies that compared narrow‐ and broad‐spectrum antibiotics for children hospitalized with CAP. Using administrative data, Williams et al. found no significant difference in LOS or cost between narrow‐ and broad‐spectrum intravenous antibiotic recipients ages 6 months to 18 years at 43 children's hospitals.[17] Queen et al. compared narrow‐ and broad‐spectrum antibiotic therapy for children ages 2 months to 18 years at 4 children's hospitals.[18] Differences in average daily cost were not significant, but children who received narrow‐spectrum antibiotics had a significantly shorter LOS. Finally, an observational study of 319 Israeli children <2 years of age found no significant difference in duration of oxygen requirement, LOS, or need for change in antibiotic therapy between the 66 children prescribed aminopenicillins and the 253 children prescribed cefuroxime.[19] These studies suggest that prescribing narrow‐spectrum antimicrobial therapy, as per the guideline recommendations, results in similar outcomes to broad‐spectrum antimicrobial therapy.

Our study adds to prior studies that compared narrow‐ and broad‐spectrum therapy by comparing outcomes associated with guideline‐recommended therapy to nonguideline therapy. We considered antibiotic therapy as guideline recommended if appropriately chosen per the guideline, not just by simple classification of antibiotic. For example, the use of a cephalosporin in a patient with aminopenicillin allergy was considered guideline‐recommended therapy. We chose to classify the exposure of empiric antibiotic therapy in this manner to reflect true clinical application of the guideline, as not all children are able to receive aminopenicillins. Additionally, as our study stemmed from our prior improvement work aimed at increasing guideline adherent therapy,[7] we have a higher frequency of narrow‐spectrum antibiotic use than prior studies. In our study, almost two‐thirds of patients received narrow‐spectrum therapy, whereas narrow‐spectrum use in prior studies ranged from 10% to 33%.[17, 18, 19] Finally, we were able to confirm the diagnosis of pneumonia via medical record review and to adjust for severity of illness using clinical variables including vital signs, physical exam findings, and laboratory and radiologic study results.

This study must be interpreted in the context of several limitations. First, our study population was defined through discharge diagnosis codes, and therefore dependent on the accuracy of coding. However, we minimized potential for misclassification through use of a previously validated approach to identify patients with CAP and through medical record review to confirm the diagnosis. Second, we may be unable to detect very small differences in outcomes given limited power, specifically for outcomes of LOS, ED revisits, hospital readmissions, and need to broaden antibiotic therapy. Third, residual confounding may be present. Although we controlled for many clinical variables in our analyses, antibiotic prescribing practices may be influenced by unmeasured factors. The potential of system level confounding is mitigated by standardized care for patients with CAP at our institution. Prior system‐level changes using quality‐improvement science have resulted in a high level of adherence with guideline‐recommended antimicrobials as well as standardized medical discharge criteria.[7, 20] Additionally, nonmedical factors may influence LOS, limiting its use as an outcome measure. This limitation was minimized in our study by standardizing medical discharge criteria. Prior work at our institution demonstrated that the majority of patients, including those with CAP, were discharged within 2 hours of meeting medical discharge criteria.[20] Fourth, discharged patients who experienced adverse outcomes may have received care at a nearby adult emergency department or at their pediatrician's office. Although these events would not have been captured in our electronic health record, serious complications due to treatment failure (eg, empyema) would require hospitalization. As our hospital is the only children's hospital in the Greater Cincinnati metropolitan area, these patents would receive care at our institution. Therefore, any misclassification of revisits or readmissions is likely to be minimal. Finally, study patients were admitted to a large tertiary academic children's hospital, warranting further investigation to determine if these findings can be translated to smaller community settings.

In conclusion, receipt of guideline‐recommended antibiotic therapy for patients hospitalized with CAP was not associated with increases in LOS, total costs of hospitalization, or inpatient pharmacy costs. Our findings highlight the importance of changing antibiotic prescribing practices to reflect guideline recommendations, as there was no evidence of negative unintended consequences with our local practice change.

There are as many as 3 million cases of severe sepsis and 750,000 resulting deaths in the United States annually.[1] Interventions such as goal‐directed resuscitation and antibiotics can reduce sepsis mortality, but their effectiveness depends on early administration. Thus, timely recognition is critical.[2, 3, 4, 5]

Despite this, early recognition in hospitalized patients can be challenging. Using chart documentation as a surrogate for provider recognition, we recently found only 20% of patients with severe sepsis admitted to our hospital from the emergency department were recognized.[6] Given these challenges, there has been increasing interest in developing automated systems to improve the timeliness of sepsis detection.[7, 8, 9, 10] Systems described in the literature have varied considerably in triggering criteria, effector responses, and study settings. Of those examining the impact of automated surveillance and response in the nonintensive care unit (ICU) acute inpatient setting, results suggest an increase in the timeliness of diagnostic and therapeutic interventions,[10] but less impact on patient outcomes.[7] Whether these results reflect inadequacies in the criteria used to identify patients (parameters or their thresholds) or an ineffective response to the alert (magnitude or timeliness) is unclear.

Given the consequences of severe sepsis in hospitalized patients, as well as the introduction of vital sign (VS) and provider data in our electronic health record (EHR), we sought to develop and implement an electronic sepsis detection and response system to improve patient outcomes. This study describes the development, validation, and impact of that system.

METHODS

The Postimplementation (Live) Period and Impact Analysis

The EWRS went live September 12, 2012, upon which new admissions triggering the alert would result in a notification and response. Unadjusted analyses using the [2] test for dichotomous variables and the Wilcoxon rank sum test for continuous variables compared demographics and the proportion of clinical process and outcome measures for those admitted during the silent period (June 6, 2012September 4, 2012) and a similar timeframe 1 year later when the intervention was live (June 6, 2013September 4, 2013). To be included in either of the time periods, patients had to trigger the alert during the period and be discharged within 45 days of the end of the period. The pre‐ and post‐sepsis mortality index was also examined (see the Supporting Information in the online version of this article for a detailed description of study measures). Multivariable regression models estimated the impact of the EWRS on the process and outcome measures, adjusted for differences between the patients in the preimplementation and postimplementation periods with respect to age, gender, Charlson index on admission, admitting service, hospital, and admission month. Logistic regression models examined dichotomous variables. Continuous variables were log transformed and examined using linear regression models. Cox regression models explored time to ICU transfer from trigger. Among patients with sepsis, a logistic regression model was used to compare the odds of mortality between the silent and live periods, adjusted for expected mortality, both within each hospital and across all hospitals.

Because there is a risk of providers becoming overly reliant on automated systems and overlooking those not triggering the system, we also examined the discharge disposition and mortality outcomes of those in both study periods not identified by the EWRS.

The primary analysis examined the impact of the EWRS across UPHS; we also examined the EWRS impact at each of our hospitals. Last, we performed subgroup analyses examining the EWRS impact in those assigned an International Classification of Diseases, 9th Revision code for sepsis at discharge or death. All analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC).

RESULTS

In the derivation cohort, 4575 patients met the inclusion criteria. The proportion of those in each category (06) achieving our outcomes of interest are described in Supporting Table 1 in the online version of this article. We defined a positive trigger as a score 4, as this threshold identified a limited number of patients (3.9% [180/4575]) with a high proportion experiencing our composite outcome (25.6% [46/180]). The proportion of patients with an EWRS score 4 and their time to event by hospital and health system is described in Supporting Table 2 in the online version of this article. Those with a score 4 were almost 4 times as likely to be transferred to the ICU, almost 7 times as likely to experience an RRT, and almost 10 times as likely to die. The screen positive, sensitivity, specificity, and positive and negative predictive values and likelihood ratios using this threshold and our composite outcome in the derivation cohort were 6%, 16%, 97%, 26%, 94%, 5.3, and 0.9, respectively, and in our validation cohort were 6%, 17%, 97%, 28%, 95%, 5.7, and 0.9, respectively.

In the preimplementation period, 3.8% of admissions (595/15,567) triggered the alert, as compared to 3.5% (545/15,526) in the postimplementation period. Demographics were similar across periods, except that in the postimplementation period patients were slightly younger and had a lower Charlson Comorbidity Index at admission (Table 1). The distribution of alerts across medicine and surgery services were similar (Table 1).

In our postimplementation period, 99% of coordinator pages and over three‐fourths of provider notifications were sent successfully. Almost three‐fourths of nurses reviewed the initial alert notification, and over 99% completed the electronic data verification and adverse trend review, with over half documenting adverse trends. Ninety‐five percent of the time the coordinators completed the follow‐up assessment. Over 90% of the time, the entire team evaluated the patient at bedside within 30 minutes. Almost half of the time, the team thought the patient had no critical illness. Over a third of the time, they thought the patient had sepsis, but reported over 90% of the time that they were aware of the diagnosis prior to the alert. (Supporting Table 3 in the online version of this article includes more details about the responses to the electronic notifications and follow‐up assessments.)

In unadjusted and adjusted analyses, ordering of antibiotics, intravenous fluid boluses, and lactate and blood cultures within 3 hours of the trigger increased significantly, as did ordering of blood products, chest radiographs, and cardiac monitoring within 6 hours of the trigger (Tables 2 and 3).

Hospital and ICU length of stay were similar in the preimplementation and postimplementation periods. There was no difference in the proportion of patients transferred to the ICU following the alert; however, the proportion transferred within 6 hours of the alert increased, and the time to ICU transfer was halved (see Supporting Figure 4 in the online version of this article), but neither change was statistically significant in unadjusted analyses. Transfer to the ICU within 6 hours became statistically significant after adjustment. All mortality measures were lower in the postimplementation period, but none reached statistical significance. Discharge to home and sepsis documentation were both statistically higher in the postimplementation period, but discharge to home lost statistical significance after adjustment (Tables 4 and 5) (see Supporting Table 4 in the online version of this article).

In a subanalysis of EWRS impact on patients documented with sepsis at discharge, unadjusted and adjusted changes in clinical process and outcome measures across the time periods were similar to that of the total population (see Supporting Tables 5 and 6 and Supporting Figure 5 in the online version of this article). The unadjusted composite outcome of mortality or inpatient hospice was statistically lower in the postimplementation period, but lost statistical significance after adjustment.

The disposition and mortality outcomes of those not triggering the alert were unchanged across the 2 periods (see Supporting Tables 7, 8, and 9 in the online version of this article).

DISCUSSION

This study demonstrated that a predictive tool can accurately identify non‐ICU inpatients at increased risk for deterioration and death. In addition, we demonstrated the feasibility of deploying our EHR to screen patients in real time for deterioration and to trigger electronically a timely, robust, multidisciplinary bedside clinical evaluation. Compared to the control (silent) period, the EWRS resulted in a marked increase in early sepsis care, transfer to the ICU, and sepsis documentation, and an indication of a decreased sepsis mortality index and mortality, and increased discharge to home, although none of these latter 3 findings reached statistical significance.

Our study is unique in that it was implemented across a multihospital health system, which has identical EHRs, but diverse cultures, populations, staffing, and practice models. In addition, our study includes a preimplementation population similar to the postimplementation population (in terms of setting, month of admission, and adjustment for potential confounders).

Interestingly, patients identified by the EWRS who were subsequently transferred to an ICU had higher mortality rates (30% and 26% in the preimplementation and postimplementation periods, respectively, across UPHS) than those transferred to an ICU who were not identified by the EWRS (7% and 6% in the preimplementation and postimplementation periods, respectively, across UPHS) (Table 4) (see Supporting Table 7 in the online version of this article). This finding was robust to the study period, so is likely not related to the bedside evaluation prompted by the EWRS. It suggests the EWRS could help triage patients for appropriateness of ICU transfer, a particularly valuable role that should be explored further given the typical strains on ICU capacity,[13] and the mortality resulting from delays in patient transfers into ICUs.[14, 15]

Although we did not find a statistically significant mortality reduction, our study may have been underpowered to detect this outcome. Our study has other limitations. First, our preimplementation/postimplementation design may not fully account for secular changes in sepsis mortality. However, our comparison of similar time periods and our adjustment for observed demographic differences allow us to estimate with more certainty the change in sepsis care and mortality attributable to the intervention. Second, our study did not examine the effect of the EWRS on mortality after hospital discharge, where many such events occur. However, our capture of at least 45 hospital days on all study patients, as well as our inclusion of only those who died or were discharged during our study period, and our assessment of discharge disposition such as hospice, increase the chance that mortality reductions directly attributable to the EWRS were captured. Third, although the EWRS changed patient management, we did not assess the appropriateness of management changes. However, the impact of care changes was captured crudely by examining mortality rates and discharge disposition. Fourth, our study was limited to a single academic healthcare system, and our experience may not be generalizable to other healthcare systems with different EHRs and staff. However, the integration of our automated alert into a commercial EHR serving a diverse array of patient populations, clinical services, and service models throughout our healthcare system may improve the generalizability of our experience to other settings.

CONCLUSION

By leveraging readily available electronic data, an automated prediction tool identified at‐risk patients and mobilized care teams, resulting in more timely sepsis care, improved sepsis documentation, and a suggestion of reduced mortality. This alert may be scalable to other healthcare systems.