Key Points
Question
Does early consultation with dermatology specialists improve outcomes for hospitalized patients with a presumed diagnosis of cellulitis?
Findings
In this randomized clinical trial of 175 hospitalized adults, the length of intravenous antibiotic use was significantly shorter and the 2-week improvement rate was significantly higher in patients in the intervention group compared with patients in the control group. Length of hospital stay was not significantly different between the 2 groups.
Meaning
Involvement of inpatient dermatology may enhance patient outcomes by improving diagnostic accuracy and facilitating antibiotic stewardship in hospitalized patients with suspected cellulitis.
This randomized clinical trial examines the association of dermatologic consultation with duration of hospital stay and intravenous antibiotic treatment in patients with suspected cellulitis.
Abstract
Importance
Each year, cellulitis leads to 650 000 hospital admissions and is estimated to cost $3.7 billion in the United States. Previous literature has demonstrated a high misdiagnosis rate for cellulitis, which results in unnecessary antibiotic use and health care cost.
Objective
To determine whether dermatologic consultation decreases duration of hospital stay or intravenous antibiotic treatment duration in patients with cellulitis.
Design, Setting, and Participants
This randomized clinical trial was conducted in a large urban tertiary care hospital between October 2012 and January 2017, with 1-month follow-up duration. Patients were randomized to the control group, which received the standard of care (ie, treatment by primary medicine team), or the intervention group, which received dermatology consultation. Medical chart review of demographic information and hospital courses was performed. Adult patients hospitalized with presumed diagnosis of cellulitis were eligible. A total of 1300 patients were screened, 1125 were excluded, and 175 were included. Statistical analysis was employed to identify significant outcome differences between the 2 groups.
Interventions
Dermatology consultation within 24 hours of hospitalization.
Main Outcomes and Measures
Length of hospital stay and duration of intravenous antibiotic treatment.
Results
Of 175 participants, 70 (40%) were women and 105 (60%) were men. The mean age was 58.8 years. Length of hospital stay was not statistically different between the 2 groups. The duration of intravenous antibiotic treatment (<4 days: 86.4% vs 72.5%; absolute difference, 13.9%; 95% CI, 1.9%-25.9%; P = .04) and duration of total antibiotic treatment was significantly lower in patients who had early dermatology consultation (<10 days: 50.6% vs 32.5%; absolute difference, 18.1%; 95% CI, 3.7%-32.5%; P = .01). Clinical improvement at 2 weeks was significantly higher for those in the intervention group (79 [89.3%] vs 59 [68.3%]; absolute difference, 21.0%; 95% CI, 9.3%-32.7%; P < .001). There was no significant difference in 1-month readmission rate between the groups (4 [4.5%] vs 6 [6.9%]; absolute difference, −2.4%; 95% CI, −9.3% to 4.5%; P = .54). In the intervention group, the rate of cellulitis misdiagnosis was 30.7% (27 of 88 participants). Among the entire cohort, 101 (57.7%) patients were treated with courses of antibiotics longer than what is recommended by guidelines.
Conclusions and Relevance
Early dermatologic consultation can improve outcomes in patients with suspected cellulitis by identifying alternate diagnoses, treating modifiable risk factors, and decreasing length of antibiotic treatment.
Trial Registration
clinicaltrials.gov Identifier: NCT01706913
Introduction
Cellulitis is a common skin infection accounting for 650 000 hospitalizations in the United States annually.1 As one of the most frequent indications for short stays in emergency department observation units (EDOU),2 cellulitis resulted in $4.5 billion of health care spending in 2009.3,4 The inappropriate diagnosis and treatment of cellulitis results in unnecessary antibiotic use and between $195 and $515 million in avoidable health care costs annually.5
Given the prevalence and associated expense of cellulitis, accurate diagnosis and appropriate treatment are imperative. Typically, suspected cellulitis is first evaluated by primary care, urgent care, or emergency department (ED) physicians and diagnosis is based on clinical features and risk factors. The characteristic finding of cellulitis is expanding, ill-defined redness, warmth, tenderness, and edema. Given its nonspecific features, cellulitis has many clinical mimics including venous stasis dermatitis, erythema migrans, and deep venous thrombosis (DVT), among others.6 These conditions are collectively known as pseudocellulitis. The inappropriate antibiotic treatment of pseudocellulitis has been estimated to cause approximately 130 000 unnecessary hospitalizations annually. With no accurate diagnostic laboratory or imaging test available, the proposed gold standard of cellulitis diagnosis is recognition by a dermatologist or infectious disease specialist because recent studies have suggested that dermatologic involvement improves diagnostic accuracy and treatment of cellulitis.5,7,8,9
This study was designed to determine whether early dermatology consultation can improve treatment of hospitalized patients with suspected cellulitis and to determine if and how outcomes are affected.7 Improved recognition and treatment of this common infection could have considerable impact on improved antibiotic stewardship, decreased widespread drug resistance, and most importantly, better patient outcomes.5,10
Methods
Setting and Study Population
This randomized clinical trial was approved by the Massachusetts General Hospital (MGH) institutional review board. All participants provided written informed consent. Participation was optional and renumeration was not offered. Patients (aged >18 years) were eligible for inclusion if they received a primary diagnosis of cellulitis from an ED physician and were sent to an inpatient ward or an EDOU. Between October 2012 and January 2017, 1300 patients were screened for eligibility and 175 patients were enrolled (Figure). Patients were excluded if they were pregnant, incarcerated, or were deemed to have impaired decision-making capacity. Patients with any of the following were excluded: surgical site infections (<4 weeks postoperation), hardware or line infections, abnormal vital signs (pulse >100 beats per minute, body temperature of >100.5° F), animal/human bites, or clinical or radiographic evidence of osteomyelitis or abscess. The trial protocol is available in Supplement 1.
Figure. Study Screening and Enrollment.
Study Design
Patients were randomized to either the intervention or control group via simple randomization on Excel (version 15.19.1, Microsoft). A random allocation sequence was implemented through use of sequentially numbered folders. A statistician generated the random allocation sequence, and 4 research assistants were responsible for enrolling patients. All patients in the intervention group were seen by a consulting dermatologist within 24 hours of hospitalization. During consultation, 1 dermatologist (D.K.) performed a medical history screening and physical examination, determined a diagnosis, and provided recommendations to the patient’s primary care team. Patients in the control group received the standard of care (treatment by the medicine/ED team). In some cases, standard of care incorporated a dermatology consultation called by the primary team during the patients’ hospitalization. All patients were offered an outpatient dermatology follow-up or a phone call if they were unable to attend a visit.
The primary outcomes measured were length of hospitalization and duration of intravenous (IV) antibiotic use. To better characterize the benefits of dermatology consultation, secondary outcomes were analyzed and included: number of antibiotics used, clinical improvement at 2 weeks, and hospital readmission rate at 1 month. Moreover, the rate of pseudocellulitis was quantified in the intervention group because dermatologic diagnosis was considered to be the gold standard diagnosis.7,8,9,10,11 Total antibiotic treatment length for all study patients was recorded.
Medical chart review of both cohorts was completed and baseline demographics, medical history, physical examination findings, and cellulitis risk factors (chronic lymphedema, tinea pedis, onychomycosis, type 2 diabetes, history of trauma, end-stage renal disease, active cancer, HIV/AIDS)12,13 were recorded. Details of patients’ hospital courses including laboratory workup, imaging studies, antibiotic treatment, length of stay, discharge medications, and final discharge diagnosis were also recorded. For patients in the intervention group, dermatology’s assessment and plan as well as the primary team’s adherence to the consultation recommendations was logged.
Two weeks after discharge, follow up status was defined as “improved” if there was a visible improvement of affected area on physical examination (office visit), or if the patient endorsed resolution of erythema, swelling, and other symptoms (phone call) if unable to attend the office visit. During each follow-up phone call, medical chart reviewers asked a standardized set of questions in the same order to assess patient progress, and answers were recorded in real time. One month following discharge, medical chart review was performed to identify patients readmitted to the hospital for recurrent disease. Study data were securely collected and managed by REDCap Electronic Data Capture tools (version 7.4.19, REDCap Consortium).14
Statistical Analysis
Sample size was based on a power calculation anticipating a 30% decrease in hospital stay or 20% IV antibiotic duration reduction in the intervention group, assuming a power level of 80% and an α of 0.05. Data was analyzed using SAS statistical software (version 9.4, SAS Inc). Summary statistics (means, standard deviations, medians, proportions) were obtained for all variables. As appropriate, 2-sided χ2 or Fisher exact tests and t tests (or Kruskal-Wallis tests) were used to test whether variables were distributed differently between the 2 groups. A 2-sided P value <.05 was considered significant.
Results
Demographics
Among 175 enrolled and randomized participants, 87 were randomized to the control group and 88 to the intervention group (Figure). A total of 70 (40%) participants were women and 105 (60%) were men, with a mean (SD) age of 58.8 (19.2) years. No statistical differences were observed in baseline demographics or cellulitis risk factors (Table 1).
Table 1. Baseline Patient Demographics, Cellulitis Risk Factors, and Select Medical History of Study Patients.
| Variable | No. (%) | P Valuea | ||
|---|---|---|---|---|
| Total (N = 175) | Control (N = 87) | Intervention (N = 88) | ||
| Age, mean (SD), y | 58.8 (19.2) | 60.9 (17.8) | 56.7 (20.4) | .15 |
| Female | 70 (40.0) | 35 (40.2) | 35 (39.8) | .95 |
| Race | ||||
| White | 151 (86.3) | 76 (87.4) | 75 (85.2) | .68 |
| Asian or Pacific Islander | 8 (4.6) | 3 (3.4) | 5 (5.7) | |
| American Indian | 1 (0.6) | 1 (1.1) | 0 (0.0) | |
| Black | 3 (1.7) | 2 (2.3) | 1 (1.1) | |
| Hispanic or Latino | 12 (6.9) | 5 (5.7) | 7 (8.0) | |
| Cellulitis risk factors | ||||
| Chronic lymphedema | 42 (24.0) | 23 (26.4) | 19 (21.6) | .45 |
| Tinea pedis | 36 (20.6) | 20 (23.0) | 16 (18.2) | .43 |
| Known trauma | 59 (33.7) | 32 (36.8) | 27 (30.7) | .39 |
| History of skin disease | 40 (22.9) | 19 (21.8) | 21 (23.9) | .75 |
| Onychomycosis | 50 (28.6) | 28 (32.2) | 22 (25.0) | .29 |
| Type II diabetes | 47 (26.9) | 26 (29.9) | 21 (23.9) | .37 |
| ESRD | 7 (4.0) | 3 (3.4) | 4 (4.5) | .99 |
| Low-dose immunosuppressionb | 11 (6.3) | 8 (9.2) | 3 (3.4) | .13 |
| High-dose immunosuppressionc | 1 (0.6) | 0 (0.0) | 1 (1.1) | .99 |
| Active cancer | 5 (2.9) | 2 (2.3) | 3 (3.4) | .99 |
| Medical history | ||||
| Hypertension | 101 (57.7) | 51 (58.6) | 50 (56.8) | .81 |
| Hyperlipidemia | 45 (25.7) | 26 (29.9) | 19 (21.6) | .21 |
| Coronary artery disease | 28 (16.0) | 15 (17.2) | 13 (14.8) | .66 |
| Congestive heart failure | 24 (13.7) | 12 (13.8) | 12 (13.6) | .98 |
| Chronic obstructive pulmonary disease | 5 (2.9) | 3 (3.4) | 2 (2.3) | .68 |
| Obesity | 68 (38.9) | 34 (39.1) | 34 (38.6) | .95 |
| Peripheral vascular disease | 22 (12.6) | 12 (13.8) | 10 (11.4) | .63 |
Abbreviation: ESRD, end-stage renal disease.
Based on t test for age and 2-sided χ2 or Fisher exact test for other variables.
Low-dose immunosuppression is defined as less than 20 mg prednisone daily or equivalent regimen.
High-dose immunosuppression is defined as 20 mg or more prednisone daily or equivalent regime.
Primary and Secondary Outcomes
Both length of hospital stay (<4 days: 76.2% vs 64.4%; absolute difference, 11.8%; 95% CI, −1.6% to 25.2%; P = .14) and length of IV antibiotic use were lower (<4 days: 85.2% vs 74.7%; absolute difference, 10.5%; 95% CI, 0.5%-20.5%; P = .04) in the intervention group compared with the control group, though only length of IV antibiotic treatment reached statistical significance. The mean number of antibiotics used in the control group was higher than in the intervention (2.8 vs 2.5, absolute difference, 0.3; 95% CI, 0.02-0.58; P = .04). The total duration of antibiotic treatment (hospital and discharge) in the control and intervention groups was less than 5 days in 10 (11.5%) and 16 (18.2%), 5 to 9 days in 20 (23%) and 28 (31.8%), 10 to 12 days in 36 (41.4%), and 20 (22.7%), and 13 days or more in 21 (24.1%) and 24 (27.3%), respectively (Table 2). The total duration of antibiotic treatment (hospital and discharge) did not differ significantly between the control and intervention groups. However, total length of antibiotic treatment became statistically significant when sensitivity analysis of all patients seen by dermatology was assessed. When comparing the patients with cellulitis in the control and intervention groups, there was no significant difference in the length of IV antibiotic use (<4 days: 72.7% vs 78.7%).
Table 2. Primary and Secondary Outcomes in Control and Intervention Groups.
| Variable | No. (%) | P Valuea | ||
|---|---|---|---|---|
| Total (N = 175) | Control (N = 87) | Intervention (N = 88) | ||
| Total No. of antibiotics (hospital/home), mean (SD) | 2.8 (1.3) | 2.8 (0.9) | 2.5 (1.0) | .04 |
| Discharged with antibiotics | ||||
| Yes | 142 (81.1) | 73 (83.9) | 69 (78.4) | .35 |
| No | 33 (18.9) | 14 (16.1) | 19 (21.6) | |
| Length of hospital stay, d | ||||
| 1 | 64 (36.6) | 32 (36.8) | 32 (36.4) | .14 |
| 2-3 | 59 (33.7) | 24 (27.6) | 35 (39.8) | |
| ≥4 | 52 (29.7) | 31 (35.6) | 21 (23.9) | |
| Duration of IV antibiotics, d | ||||
| 0 | 13 (7.4) | 2 (2.3) | 11 (12.5) | .04 |
| 1 | 74 (42.3) | 36 (41.4) | 38 (43.2) | |
| 2 | 24 (13.7) | 11 (12.6) | 13 (14.8) | |
| 3 | 29 (16.6) | 16 (18.4) | 13 (14.8) | |
| ≥4 | 35 (20) | 22 (25.3) | 13 (14.8) | |
| Total duration of antibiotic treatment (hospital/home), d | ||||
| <5 | 26 (14.9) | 10 (11.5) | 16 (18.2) | .06 |
| 5-9 | 48 (27.4) | 20 (23.0) | 28 (31.8) | |
| 10-12 | 56 (32.0) | 36 (41.4) | 20 (22.7) | |
| ≥13 | 45 (25.7) | 21 (24.1) | 24 (27.3) | |
| Patient’s condition at 2-week follow-up | ||||
| Improved | 131 (78.9) | 56 (68.3) | 75 (89.3) | .003 |
| Stable | 25 (15.1) | 19 (23.2) | 6 (7.1) | |
| Worse | 10 (6.0) | 7 (8.5) | 3 (3.6) | |
| Readmission with diagnosis of cellulitis 1 month after discharge | ||||
| Yes | 10 (5.7) | 6 (6.9) | 4 (4.5) | .54 |
| No | 165 (94.3) | 81 (93.1) | 84 (95.5) | |
Based on Kruskal-Wallis test for sequential interval variables and 2-sided χ2 or Fisher exact test for other variables.
At 2-week follow-up, study patients were reevaluated by phone call (88, 50%), office visit (45, 26%), or medical chart review (33, 19%) if unable to be reached. Nine (5%) patients were lost to follow-up. In the intervention group, 79 (89.3%) reported improvement of the affected area compared with 59 (68.3%) of control patients (89.3% vs. 68.3%; absolute difference, 21.0%; 95% CI, 9.3%-32.7%; P < .001). One month after discharge, 4 (4.5%) patients in the intervention group and 6 (6.9%) patients in the control group were readmitted for recurrent disease (4.5% vs 6.9%; absolute difference, −2.4%; 95% CI, −9.3 to 4.5%; P = .54) (Table 2). There were no adverse events in either group.
Sensitivity Analysis of Patients Receiving Dermatology Consultation
There were 11 patients in the control group whose primary teams sought dermatologic consultation independent of the study, 7 of whom were seen by dermatology within 24 hours. A sensitivity analysis was performed categorizing these 95 patients into the intervention group (Table 3). Dermatology consultation led to a significantly decreased length of IV antibiotic use (<4 days: 86.4% vs 72.5%; absolute difference, 13.9%; 95% CI, 1.9%-25.9%; P = .04), as well as a decreased total number of antibiotic treatment days (<10 days: 50.6% vs 32.5%; absolute difference, 18.1%; 95% CI, 3.7%-32.5%; P = .01) (eFigure 1 in Supplement 2). In addition, we completed a subgroup analysis of the 11 patients in the control group (n = 87) whose primary team called a dermatology consult via retrospective medical chart review. In comparison to the overall control group, this cohort did not exhibit any significant differences in race or sex. Patients were older, but not significantly so (mean age, 69.9 vs 60.9 years; P = .12). In terms of cellulitis risk factors and comorbidities, these 11 patients had higher rates of peripheral vascular disease (54.4% vs 13.8%; absolute difference, 40.6%; 95% CI, 10.3%-70.9%; P = .001) compared with the rest of the control group. All patients were on antibiotic therapy prior to being seen by dermatologists, and most (8 of 11, 72.7%) of the consultations were ordered owing to lack of response to initial antibiotic therapy. Of the remaining 3 consultations, 1 was ordered to perform a skin biopsy owing to uncertain diagnosis, whereas the remaining 2 were ordered to evaluate possible superinfections.
Table 3. Sensitivity Analysis Comparing Outcomes Among Patients Who Did and Did Not Have an Early Dermatology Consultation.
| Variable | No. (%) | P Valuea | ||
|---|---|---|---|---|
| Total (N = 175) | No Early Dermatology Consultation (N = 80) | Early Dermatology Consultation (N = 95) | ||
| Discharged with antibiotics | ||||
| Yes | 142 (81.1) | 70 (87.5) | 72 (75.8) | .05 |
| No | 33 (18.9) | 10 (12.5) | 23 (24.2) | |
| Length of hospital stay, d | ||||
| 1 | 64 (36.6) | 29 (36.3) | 35 (36.8) | .15 |
| 2-3 | 59 (33.7) | 22 (27.5) | 37 (38.9) | |
| ≥4 | 52 (29.7) | 29 (36.3) | 23 (24.2) | |
| Duration of IV antibiotics, d | ||||
| 0 | 13 (7.4) | 2 (2.5) | 11 (11.6) | .04 |
| 1 | 74 (42.3) | 31 (38.8) | 43 (45.3) | |
| 2 | 24 (13.7) | 10 (12.5) | 14 (14.7) | |
| 3 | 29 (16.6) | 15 (18.8) | 14 (14.7) | |
| ≥4 | 35 (20) | 22 (27.5) | 13 (13.6) | |
| No. of discharge antibiotics | ||||
| 0 | 33 (18.9) | 10 (12.5) | 23 (24.2) | .06 |
| 1 | 97 (55.4) | 44 (55.0) | 53 (55.8) | |
| ≥2 | 45 (25.7) | 26 (32.5) | 19 (20.0) | |
| Total duration of antibiotic treatment (hospital/home), d | ||||
| <5 | 26 (14.9) | 6 (7.5) | 20 (21.1) | .006 |
| 5-9 | 48 (27.4) | 20 (25.0) | 28 (29.5) | |
| 10-12 | 56 (32.0) | 35 (43.8) | 21 (22.1) | |
| ≥13 | 45 (25.7) | 19 (23.8) | 26 (27.4) | |
| Patient’s condition at 2 week follow-up | ||||
| Improved | 131 (78.9) | 50 (66.7) | 81 (89.0) | .002 |
| Stable | 25 (15.1) | 18 (24.0) | 7 (7.7) | |
| Worse | 10 (6.0) | 7 (9.3) | 3 (3.3) | |
| Readmission with dx of cellulitis 1 mo after discharge | ||||
| Yes | 10 (5.7) | 6 (7.5) | 4 (4.2) | .52 |
| No | 165 (94.3) | 74 (92.5) | 91 (95.8) | |
Abbreviations: dx, diagnosis; IV, intravenous.
Based on 2-sided χ2 or Fisher exact test for other variables.
Consultation Benefits
In the intervention group, the rate of misdiagnosis was 30.7% (27 of 88) (ie, pseudocellulitis), significantly higher than in the control group (30.7 vs. 5.7; absolute difference, 25.0%; 95% CI, 14.2%-35.8%; P < .001). Among those patients, 26% (7 of 27) had venous stasis dermatitis, whereas 5 (19%) had erythema migrans, 2 (7%) had contact dermatitis, 3 (11%) had eczema, and 2 (7%) had erythema nodosum or other more uncommon entities including calciphylaxis, vestibulitis, and lymphangitis (eFigures 2 and 3 in Supplement 2). Patients in the intervention group with pseudocellulitis had significantly shorter durations of IV antibiotic treatment (<4 days: 100% vs 79%; absolute difference, 21.0%; 95% CI, 10.8%-31.2%; P <.001) compared with those with cellulitis (eTable in Supplement 2).
The dermatology department’s recommended treatment corresponded to the final treatment by the primary team in 61 (69.3%) of 88 cases. Overall, 65 (73.8%) of 88 dermatology consultations affected treatment in some way. For example, wound care of 41 (46.6%) patients was improved following consultation, significantly more than in the control group (46.6% vs 6.9%, absolute difference, 39.7%; 95% CI, 28.0%-51.4%; P < .001). Modifications included dressing options such as addition of collagenase for enhanced wound healing, compression to address concurrent venous stasis or lymphedema, or agents such as mupirocin and hydrocolloid wound paste (Table 4).
Table 4. Comparison of Treatment Interventions Between Intervention and Control Groups.
| Variable | No. (%) | P Value | |
|---|---|---|---|
| Intervention (N = 88)a | Control (N = 87) | ||
| Pseudocellulitis rate | 27 (30.7) | 5 (5.7) | <.001 |
| Dermatology recommendations align with treatment team’s final treatment plan | 61 (69.3) | NA | NA |
| Dermatology recommendations changed treatment | 65 (73.9) | NA | NA |
| Primary treatment changed | 21 (23.9) | 5 (5.7) | |
| Antibiosis for different infection (ie, lyme, varicella) | 15 (71.4) | 1 (20) | <.001 |
| New regimen for noninfectious etiology (ie, sodium thiosulfate) | 6 (28.6) | 4 (80) | |
| Antibiotic discontinuation | 16 (18.2) | 3 (3.4) | .002 |
| Wound care | 41 (46.6) | 6 (6.9) | <.001 |
| IV Antibiotics narrowed to PO | 12 (13.6) | 7 (8.0) | .23 |
| Topical steroid cream added | 31 (35.2) | 3 (3.4) | <.001 |
| Wound swab obtained | 6 (6.8) | 11 (12.6) | .19 |
| Tinea pedis medication | 4 (4.5) | 5 (5.7) | .72 |
| Onychomycosis medication | 3 (3.4) | 1 (1.1) | .31 |
| Skin biopsy | 4 (4.5) | 2 (2.3) | .42 |
| Other laboratory tests ordered (ie, lyme titers, HIV, HCV antibodies) | 4 (4.5) | 5 (5.7) | .72 |
| Consultation from specialty (ie, rheumatology, orthopedics, surgery) | 7 (8.0) | 25 (28.7) | <.001 |
| Antibiotic coverage broadened | 3 (3.4) | 15 (17.2) | .003 |
Abbreviations: HCV, hepatitis C virus; HIV, human immunodeficiency virus; IV, intravenous; NA, not applicable; PO, per oral.
The percentages in this category do not add up to 100% because dermatology consultation may have affected patients in more than 1 way.
Among the intervention group, the treatment plans of 23.9% patients (21 of 88) were altered by dermatology consultation to reflect their true diagnosis; these patients would not have improved on their current treatment for presumed cellulitis. Of those 21 patients, 15 (71.4%) were diagnosed with noninfectious etiologies and their antibiotics were discontinued in lieu of other regimens (ie, sodium thiosulfate for calciphylaxis, topical steroids for eczematous dermatitis). The remaining 6 (28.6%) were diagnosed with either erythema migrans or abscess, and their antibiotics were changed to an appropriate regimen. In the control group, the rate at which primary treatment was altered was significantly less (23.9% vs 5.7%, absolute difference, 18.2%; 95% CI, 8.0%-28.4%; P = .007). Treatment of tinea pedis and onychomycosis was added in 4 (4.5%) and 3 (3.4%) patients, respectively, though not at a rate significantly higher than in the control group. Three (3.4%) patients’ antibiotic coverage was broadened based on dermatologic examination, or following results of wound swab, which was ordered by the dermatology department in 6 (6.8%) cases (Table 4).
Antibiotic Use
Of 175 participants, 142 (81.1%) were discharged with antibiotics; 97 (55.4%) with one, and 45 (25.7%) with 2 antibiotics. The total length of antibiotic treatment (inpatient and discharge antibiotic) was less than 5 days for 26 (14.9%) participants, 5 to 9 days for 48 (27.4%), 10 to 12 days for 56 (32%), and 13 or more days for 45 (25.7%) patients (eFigure 1 in Supplement 2).
Discussion
This study assessed the effect of dermatologic consultation on the outcomes of hospitalized patients with presumed cellulitis. Patients in the intervention group, all of whom received an early dermatologic consultation, were compared with patients in the control group who received standard of care. Dermatology specialists identified pseudocellulitis at a rate of 30.7% (27 of 88), corroborating studies finding misdiagnosis and unnecessary antibiotic use in 30% of patients. For 21 (23.9%) of 88 patients in the intervention group, dermatologists made an alternative diagnosis that would not have responded to the patients’ current treatment. In these cases, regimens were switched to more appropriate therapies, such as doxycycline for lyme disease, sodium thiosulfate for calciphylaxis, topical steroids for stasis dermatitis. Intravenous antibiotics were discontinued in 16 (18.2%) of 88 patients after dermatologic consultation resulted in an alternate diagnosis, compared with only 3 (3.4%) in patients in the control group (18.2% vs. 3.4%; absolute difference, 14.8%; 95% CI, 5.9%-23.7%; P = .002). Antibiotics were narrowed from IV to per oral in another 12 (13.6%) patients in the intervention group. In the intervention group, patients with pseudocellulitis received significantly fewer days of IV antibiotic use compared with patients with cellulitis (<4 days: 100% vs 79%; absolute difference, 21.0%; 95% CI, 10.8%-31.2%; P < .001), suggesting that dermatologic involvement can foster antibiotic stewardship through improved diagnostic accuracy.
Treatment of 65 (73.8%) of the patients in this arm was affected by dermatologic consultation, regardless of whether primary outcomes were affected. In addition, 69.3% of dermatologic recommendations aligned entirely with the primary team’s ultimate plan (Table 4). Recommendations often addressed modifiable risk factors for cellulitis, with 7 (8%) patients treated for lower extremity fungal infections (tinea pedis, onychomycosis) whereas wound care of ulcers and erosions was optimized in 41 (46.6%) patients.12,15 These results suggest that consultation can have prophylactic benefit in addressing concurrent skin conditions, many of which exacerbate or lead to infection.
In comparing the control and intervention groups, length of hospital stay was shorter for patients in the intervention group (<4 days: 76.2% vs 64.4%; absolute difference, 11.8%; 95% CI, −1.6% to 25.2%; P = .14). This approached but did not reach statistical significance, likely for several reasons. First, the study may have lacked adequate power because we initially sought to detect a 30% decrease in length of stay. Second, EDOU patients are, by definition, in the hospital for very short periods and thus consultation may have done little to change length of stay. The intervention’s effect on hospitalization length may also have been blunted given that many patients were admitted with comorbidities requiring lengthy inpatient treatment unrelated to cellulitis (Table 1). Dermatologic consultation diagnosed alternate pathologies in approximately one-third of patients, some of which, like calciphylaxis, required longer stays than uncomplicated cellulitis. Thus, though hospitalization length did not decrease with intervention, the improvement in diagnostic accuracy and addressing of modifiable risk factors and concurrent skin conditions benefitted patients.
Intravenous antibiotic treatment duration was significantly shorter for patients in the intervention group compared with those in the control group (<4 days: 85.2 vs 74.7%; absolute difference, 10.5%; 95% CI, 0.5%-20.5%; P = .04). Further sensitivity analysis of all patients who received dermatologic consultation revealed that these patients had shorter durations of IV antibiotics (<4 days: 86.4 vs 72.5; absolute difference, 13.9%; 95% CI, 1.9%-25.9%; P = .04) and total antibiotic treatment which was statistically significant, suggesting that dermatologists can facilitate targeted antibiotic use (eFigure 1 in Supplement 2). This may be because approximately one-fifth of patients (18.2%) in the intervention group were deemed to have an alternate noninfectious diagnosis and antibiotics were discontinued in lieu of more appropriate therapy. Furthermore, dermatologic consultation successfully narrowed antibiotics from IV to per oral in 12 (13.6%) patients.
There was a statistically higher percentage of patients in the intervention arm with reported improvement at 2-week follow up (89.3% vs 68.3%; absolute difference, 21.0%; 95% CI, 9.3%-32.7%; P < .001). The rationale for higher improvement rates may be multifold in addition to actual clinical improvement. Dermatologic consultation addressed the primary skin issue as well as any concurrent skin and wound conditions (many of which exacerbate infection), potentially creating a psychological benefit of seeing a skin expert. Moreover, dermatology consultation addressed other subjective symptoms such as itch and pain. Lastly, patient education on causes of infection, and the differences between cellulitis and pseudocellulitis was often discussed during consultation.
There was no significant difference in the 1-month readmission rate between groups, likely given the low overall readmission rate (10 of 175). Thirty days may be insufficient to capture the longitudinal benefits of dermatologic consultation, especially among those patients with recurrent cellulitis, defined as 2 or more episodes in 1 year. In 1 such example, a patient in the intervention group with recurrent cellulitis (10 episodes over the previous 3 years) was noted to have peripheral sensory loss during his consultation. Podiatry and neurology consults were arranged, revealing considerable L5 nerve damage that likely contributed to his medical history of recurrent infection. This patient now has regular podiatric foot checks, the benefits of which are not captured by this study. Another patient in the intervention group diagnosed with erythema nodosum had visited the ED 4 times in 1 year and was treated with vancomycin each time with no improvement. Work-up by dermatology specialists revealed latent tuberculosis as the underlying trigger. Thus, dermatology input may be helpful not only in addressing risk factors that lead to recurrent episodes of infection, but also in identifying clinical mimics, and ensuring appropriate work-up and treatment of alternative diagnoses.7,10 There were no adverse events associated with earlier hospital discharge or discontinuation of antibiotics in the intervention group.
Knowledge of current antibiotic prescribing practices is limited across the United States. This study illustrates the wide practice variability in antibiotic choice and duration. Many patients were treated with longer courses than recommended by expert guidelines.16 Large-scale studies suggest that hospitalized patients with uncomplicated SSTIs should not receive 10 or more days of antibiotics.16 Other studies posit that 5 to 7-day courses are sufficient and no less effective than longer treatment courses.17,18,19 By these standards, 101 (57.7%) of the 175 study patients received longer treatment courses (≥10 days) than what is recommended by expert guidelines (eFigure 1 in Supplement 2), suggesting that additional measures are needed to overcome perceptions that excessive courses of antibiotics cannot be harmful. Sensitivity analysis showed that all patients with early dermatologic consultation received shorter periods of treatment compared with those who did not see dermatologists (total antibiotic treatment length <10 days: 50.6% vs. 32.5%; absolute difference, 18.1%; 95% CI, 3.7%-32.5%; P = .01). Fastidious antibiotic stewardship cannot only limit adverse events and toxic effects at an individual level, but also drug resistance at a community level.17 These results indicate that dermatologists may in the future have a role in reassuring clinicians on the use of shorter, more limited antimicrobial treatment courses.
Limitations
There were a few limitations in this study. The study inclusion and exclusion criteria were developed to focus on uncomplicated cellulitis, so additional study is needed to determine if these models can be applied more broadly for SSTIs in general. Another limitation is that 30.2% of primary teams did not follow dermatologic recommendations to narrow or discontinue antibiotic use, suggesting that true outcome differences between groups may be masked. Nevertheless, this is a common problem in all subspecialty consultations, and ensuring 100% adherence to recommendations was outside the scope of this study. In addition, another limitation is that duration of antibiotic use in pseudocellulitis patients may be overestimated in this study because primary teams had often already started treatment by the time of intervention. Another limitation was that research assistants were not blinded to final diagnosis by the time of follow-up communication. However, to avoid bias, all study patients were asked a standardized set of questions to assess their improvement.
The study’s generalizability may be limited because inpatient dermatology consultation is not present at all hospitals; however, inpatient services are available at most academic medical centers20 and the practice of teledermatology has been proven to be effective,21,22 providing a potential future bridge to community hospitals. Finally, the findings should be interpreted in the context of the study design. First, in this study, cellulitis was diagnosed based on a dermatologist’s clinical impression, however in reality there is no gold standard for diagnosis. Second, this investigation was performed at a single academic medical center; therefore, additional studies are needed to determine the generalizability of these findings.
Conclusions
Both the number of ambulatory visits for and the hospitalization rate of cellulitis patients have increased markedly over the past 2 decades.3,4,23,24 It is critical that the diagnosis and treatment of cellulitis continue to be streamlined. In this study, early dermatologic consultation was shown to increase patient improvement rates at 2 week follow-up and to significantly decrease length of IV antibiotic use. The results echo others that collaboration with dermatology specialists can improve outcomes in patients with presumed cellulitis by treating modifiable risk factors in true cellulitis patients, and avoiding unnecessary antibiotics in patients with mimicking diagnoses. The results of this study also suggest that improved adherence to standardized cellulitis treatment guidelines may promote antibiotic stewardship and improved patient outcomes.
Trial Protocol
eFigure 1. Duration of Total Antibiotic Treatment
eFigure 2. Alternate Diagnoses in the Intervention Group
eFigure 3. Examples of Patients with Pseudocellulitis
eTable. Comparison of Primary Outcomes Among Intervention Patients Based on their Cellulitis Status
References
- 1.Dong SL, Kelly KD, Oland RC, Holroyd BR, Rowe BH. ED management of cellulitis: a review of five urban centers. Am J Emerg Med. 2001;19(7):535-540. [DOI] [PubMed] [Google Scholar]
- 2.Volz KA, Canham L, Kaplan E, Sanchez LD, Shapiro NI, Grossman SA. Identifying patients with cellulitis who are likely to require inpatient admission after a stay in an ED observation unit. Am J Emerg Med. 2013;31(2):360-364. [DOI] [PubMed] [Google Scholar]
- 3.Kaye KS, Patel DA, Stephens JM, Khachatryan A, Patel A, Johnson K. Rising United states hospital admissions for acute bacterial skin and skin structure infections: recent trends and economic impact. PLoS One. 2015;10(11):e0143276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hersh AL, Chambers HF, Maselli JH, Gonzales R. National trends in ambulatory visits and antibiotic prescribing for skin and soft-tissue infections. Arch Intern Med. 2008;168(14):1585-1591. [DOI] [PubMed] [Google Scholar]
- 5.Weng QY, Raff AB, Cohen JM, et al. Costs and consequences associated with misdiagnosed lower extremity cellulitis. JAMA Dermatol. 2016. [DOI] [PubMed] [Google Scholar]
- 6.Raff AB, Kroshinsky D. Cellulitis: a review. JAMA. 2016;316(3):325-337. [DOI] [PubMed] [Google Scholar]
- 7.Arakaki RY, Strazzula L, Woo E, Kroshinsky D. The impact of dermatology consultation on diagnostic accuracy and antibiotic use among patients with suspected cellulitis seen at outpatient internal medicine offices: a randomized clinical trial. JAMA Dermatol. 2014;150(10):1056-1061. [DOI] [PubMed] [Google Scholar]
- 8.Strazzula L, Cotliar J, Fox LP, et al. Inpatient dermatology consultation aids diagnosis of cellulitis among hospitalized patients: A multi-institutional analysis. J Am Acad Dermatol. 2015;73(1):70-75. [DOI] [PubMed] [Google Scholar]
- 9.Jain SR, Hosseini-Moghaddam SM, Dwek P, et al. Infectious diseases specialist management improves outcomes for outpatients diagnosed with cellulitis in the emergency department: a double cohort study. Diagn Microbiol Infect Dis. 2017;87(4):371-375. [DOI] [PubMed] [Google Scholar]
- 10.Levell NJ, Wingfield CG, Garioch JJ. Severe lower limb cellulitis is best diagnosed by dermatologists and managed with shared care between primary and secondary care. Br J Dermatol. 2011;164(6):1326-1328. [DOI] [PubMed] [Google Scholar]
- 11.David CV, Chira S, Eells SJ, et al. Diagnostic accuracy in patients admitted to hospitals with cellulitis. Dermatol Online J. 2011;17(3):1. [PubMed] [Google Scholar]
- 12.Björnsdóttir S, Gottfredsson M, Thórisdóttir AS, et al. Risk factors for acute cellulitis of the lower limb: a prospective case-control study. Clin Infect Dis. 2005;41(10):1416-1422. [DOI] [PubMed] [Google Scholar]
- 13.Quirke M, Ayoub F, McCabe A, et al. Risk factors for nonpurulent leg cellulitis: a systematic review and meta-analysis. Br J Dermatol. 2017;177(2):382-394. [DOI] [PubMed] [Google Scholar]
- 14.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377-381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Bristow IR, Spruce MC. Fungal foot infection, cellulitis and diabetes: a review. Diabet Med. 2009;26(5):548-551. [DOI] [PubMed] [Google Scholar]
- 16.Hurley HJ, Knepper BC, Price CS, Mehler PS, Burman WJ, Jenkins TC. Avoidable antibiotic exposure for uncomplicated skin and soft tissue infections in the ambulatory care setting. Am J Med. 2013;126(12):1099-1106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hepburn MJ, Dooley DP, Skidmore PJ, Ellis MW, Starnes WF, Hasewinkle WC. Comparison of short-course (5 days) and standard (10 days) treatment for uncomplicated cellulitis. Arch Intern Med. 2004;164(15):1669-1674. [DOI] [PubMed] [Google Scholar]
- 18.Ruhe JJ, Smith N, Bradsher RW, Menon A. Community-onset methicillin-resistant Staphylococcus aureus skin and soft-tissue infections: impact of antimicrobial therapy on outcome. Clin Infect Dis. 2007;44(6):777-784. [DOI] [PubMed] [Google Scholar]
- 19.Rajendran PM, Young D, Maurer T, et al. Randomized, double-blind, placebo-controlled trial of cephalexin for treatment of uncomplicated skin abscesses in a population at risk for community-acquired methicillin-resistant Staphylococcus aureus infection. Antimicrob Agents Chemother. 2007;51(11):4044-4048. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ko L, Kroshinsky D.. Hospitalist dermatology: a cross-sectional multicenter survey study characterizing the infrastructure of consultative dermatology in United States hospitals. Int J Dermatol. [DOI] [PubMed] [Google Scholar]
- 21.Barbieri JS, Nelson CA, James WD, et al. The reliability of teledermatology to triage inpatient dermatology consultations. JAMA Dermatol. 2014;150(4):419-424. [DOI] [PubMed] [Google Scholar]
- 22.Sharma P, Kovarik CL, Lipoff JB. Teledermatology as a means to improve access to inpatient dermatology care. J Telemed Telecare. 2016;22(5):304-310. [DOI] [PubMed] [Google Scholar]
- 23.Christensen KLY, Holman RC, Steiner CA, Sejvar JJ, Stoll BJ, Schonberger LB. Infectious disease hospitalizations in the United States. Clin Infect Dis. 2009;49(7):1025-1035. [DOI] [PubMed] [Google Scholar]
- 24.Pallin DJ, Egan DJ, Pelletier AJ, Espinola JA, Hooper DC, Camargo CA Jr. Increased US emergency department visits for skin and soft tissue infections, and changes in antibiotic choices, during the emergence of community-associated methicillin-resistant Staphylococcus aureus. Ann Emerg Med. 2008;51(3):291-298. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Trial Protocol
eFigure 1. Duration of Total Antibiotic Treatment
eFigure 2. Alternate Diagnoses in the Intervention Group
eFigure 3. Examples of Patients with Pseudocellulitis
eTable. Comparison of Primary Outcomes Among Intervention Patients Based on their Cellulitis Status