In total, 165 patients with presumed cellulitis were initially screened. Of these, 116 patients (77 with cellulitis and 39 with pseudocellulitis) were enrolled into the study.
aSome patients with pseudocellulitis had more than 1 diagnosis.
eTable 1. Calculations for cost savings modeling
eTable 2. Antibiotic-days avoided per dermatology consult for presumed cellulitis
eTable 3. Cost savings per dermatology consult for presumed cellulitis
Customize your JAMA Network experience by selecting one or more topics from the list below.
Li DG, Xia FD, Khosravi H, et al. Outcomes of Early Dermatology Consultation for Inpatients Diagnosed With Cellulitis. JAMA Dermatol. 2018;154(5):537–543. doi:10.1001/jamadermatol.2017.6197
What are the clinical and economic implications of early dermatology consultation for patients with presumed cellulitis?
This cohort study found that early dermatology consultation for presumed cellulitis decreased rates of unnecessary antibiotic use by 74.4% and unnecessary hospitalizations by 85.0% in patients with pseudocellulitis. Nationally, dermatology-led screening could eliminate 97 000 to 256 000 unnecessary hospitalization days, protect between 34 000 and 91 000 patients from unnecessary antibiotic exposure, and save $80 million to $210 million annually.
Early dermatology consultation for presumed cellulitis represents an economical intervention that reduces hospitalization length and antibiotic use for patients with pseudocellulitis.
Many inflammatory skin dermatoses mimic cellulitis (pseudocellulitis) and are treated with antibiotics and/or hospitalization, leading to unnecessary patient morbidity and substantial health care spending.
To evaluate the impact of early dermatology consultation on clinical and economic outcomes associated with misdiagnosed cellulitis.
Design, Setting, and Participants
This prospective cohort study enrolled patients with presumed diagnosis of cellulitis in the emergency department, in the emergency department observation unit, or within 24 hours of admission to an inpatient unit of a large urban teaching hospital between February and September 2017. Patients were provided with telephone and clinic follow-up during the 30-day postdischarge period. We screened 165 patients with the primary concern of cellulitis. Of these, we excluded 44 who required antibiotics for cutaneous, soft-tissue, and deeper-tissue and/or bone infections irrespective of cellulitis status, and 5 who were scheduled to be discharged by the emergency department.
Early dermatology consultation for presumed cellulitis.
Main Outcomes and Measures
Primary outcomes were patient disposition and rates of antibiotic use.
Of 116 patients (63 [54.3%] women; 91 [78.4%] non-Hispanic white; mean [SD] age, 58.4 [19.1] years), 39 (33.6%) were diagnosed with pseudocellulitis by dermatologists. Of these, 34 (87.2%) had started using antibiotics for presumed cellulitis as prescribed by the primary team at the time of enrollment. The dermatology team recommended antibiotic discontinuation in 28 of 34 patients (82.4%), and antibiotics were stopped in 26 of 28 cases (92.9%). The dermatologists also recommended discharge from planned observation or inpatient admission in 20 of 39 patients with pseudocellulitis (51.3%), and the primary team acted on this recommendation in 17 of 20 cases (85.0%). No patients diagnosed with pseudocellulitis experienced worsening condition after discharge based on phone and clinic follow-up (30 of 39 [76.9%] follow-up rate). Extrapolating the impact of dermatology consultation for presumed cellulitis nationally, we estimate 97 000 to 256 000 avoided hospitalization days, 34 000 to 91 000 patients avoiding unnecessary antibiotic exposure, and $80 million to $210 million in net cost savings annually.
Conclusions and Relevance
Early consultation by dermatologists for patients with presumed cellulitis represents a cost-effective intervention to improve health-related outcomes through the reduction of inappropriate antibiotic use and hospitalization.
Cellulitis is a common infection of the dermis and subcutaneous tissue that leads to 2.3 million emergency department (ED) visits and 650 000 hospital admissions per year in the United States.1,2Quiz Ref ID Clinical similarities between cellulitis and its mimickers (pseudocellulitis) lead to misdiagnosis rates of 30% to 90%3-5 and result in unnecessary interventions.6-9 Our prior research has estimated that inappropriate treatment and hospitalization for patients with misdiagnosed lower extremity cellulitis may result in more than 9000 nosocomial infections, 1000 to 5000 Clostridium difficile infections, and 2 to 6 cases of anaphylaxis yearly, in addition to annual hospitalization costs of $195 million to $515 million.10,11
Considering the clinical and economic consequences of misdiagnosed cellulitis, potential solutions to improve diagnostic accuracy are needed. While prior investigations have posited that dermatology consultation may lead to improved outcomes in patients with presumed cellulitis, these efforts are primarily centered on outpatient care.4,12 In this prospective study, we evaluate the impact of early dermatology consultation on the rates of unnecessary antibiotic use and hospitalization for patients with presumed cellulitis and model the potential public health implications and cost savings of this intervention.
We performed a prospective cohort study evaluating the impact of dermatology consultation on patients presenting to the ED, to the ED observation unit, or within 24 hours of admission to an inpatient unit with presumed cellulitis at Brigham and Women’s Hospital, a tertiary care center in Boston, Massachusetts, between February and September of 2017. This project was granted a quality improvement exemption by the Partners Healthcare Institutional Review Board, waiving the requirement for patient consent.
Patients were screened using the electronic medical record system by a trained research fellow from 8 am to 5 pm on weekdays. Those who were given a primary diagnosis of cellulitis by the ED team and designated for admission or observation status were enrolled. For patients who had physician documentation concerning for cellulitis (eg, using descriptors such as skin erythema, skin redness, spreading erythema, redness and swelling, and/or painful erythema) but without a coded diagnosis of cellulitis at the time of screening, we manually reviewed the entire patient record and contacted the responding clinician to confirm the proposed diagnosis of cellulitis. In addition, we performed daily searches of all admitted patients to identify inpatients with a primary or secondary billed diagnosis of cellulitis who were admitted within the past 24 hours to capture overnight admissions.
We excluded patients who required antibiotics for cutaneous, soft-tissue, and deeper-tissue and/or bone infections irrespective of cellulitis status, such as patients with known soft-tissue abscess, deep penetrating wound, animal bite, osteomyelitis, diabetic ulcer infection, or indwelling hardware at the site of infection. We also excluded patients who were independently scheduled to be discharged by the ED (Figure). All enrolled patients were evaluated in person by a clinical research fellow (D.G.L., F.D.X., H.K.) and a board-certified dermatologist (A.M., A.K.D.), who provided the primary treatment team with recommendations on diagnosis, management, and disposition.
For cases in which a dermatologist diagnosed the patient with pseudocellulitis, the dermatologist relayed recommendations regarding antibiotic discontinuation, further evaluation, and disposition to the primary team. In addition, a disease-specific clinic visit or phone follow-up was scheduled for all patients shortly after discharge.
We reviewed the medical records of eligible patients to retrieve demographic information, relevant medical history, and pertinent details of the present illness. We also collected information on proposed treatment and planned disposition status (observation, inpatient admission, discharge) by the primary team at the time of patient enrollment, and subsequent changes after the dermatologist’s recommendations. Each patient’s length of stay was calculated as the difference between discharge time and the time of admission into an inpatient or ED observation unit. Same-day discharges were defined as a hospital or observation stay of less than 12 hours. Data were recorded using Research Electronic Data Capture version 6 (REDCap).13
The postdischarge course over 30 days was evaluated. Patients with a scheduled cellulitis- or pseudocellulitis-related clinic visit were seen in person by the study staff. Clinical notes for patients seen by nondermatologists were manually reviewed. For patients without scheduled visits, a clinical research fellow contacted patients via phone at 2, 7, and 30 days postdischarge. In both follow-up methods, we collected information regarding clinical improvement, medication adherence, and complications stemming from relevant treatment (eg, rash, anaphylaxis, gastrointestinal distress, diarrhea, C difficile infection).
The primary outcome measure was the rate of antibiotic use and disposition change associated with dermatology consultation among patients who were determined to have pseudocellulitis. Secondary outcome measures included the rate of adverse events or worsening cellulitis or pseudocellulitis in the 30 days following discharge. All data were analyzed and reported descriptively with means (standard deviation) for continuous measures and counts (percentage) for categorical variables. Means were compared with 2-sample independent t tests and nominal variables were compared with χ2 or Fisher exact tests as appropriate. Analyses were performed in SAS 9.4 (SAS Institute Inc).
We estimated the potential clinical and economic impact of early dermatology consultation for US patients being admitted from the ED with a concern for cellulitis. When determining the number of patients admitted annually through the ED for cellulitis, we excluded cellulitis cases complicated by abscess from our cost calculations, using abscess prevalence rates of 12.9% to 67.0% per literature review.10,14-17 The Household Component of the Medical Expenditure Panel Survey for hospital inpatient stays in 2010 was used to estimate the cost savings associated with variable lengths of stay. The median cost of dermatology services was calculated using the Medicare Physician Fee Schedule, Healthcare Common Procedure Coding System.18
One hundred sixty-five patients with presumed cellulitis were identified, of whom 116 (70.3%) were ultimately enrolled (Figure). Quiz Ref IDOf these 116 patients, 39 (33.6%) were diagnosed with pseudocellulitis by dermatology consultation during the hospital encounter. The remaining 77 patients (66.4%) were treated per usual care.
Among the patients, the mean (SD) age was 58.4 (19.1) years, 63 (54.3%) were women, and 91 (78.4%) were non-Hispanic white; patient demographics did not differ significantly by diagnosis (P > .05 for all comparisons) (Table 1). The most commonly affected location in both cohorts was the lower extremity, which was affected in 24 of 39 patients with pseudocellulitis (61.5%) and 54 of 77 with cellulitis (70.1%). Of 39 patients diagnosed with pseudocellulitis, the most common alternative diagnoses were stasis dermatitis (23.1%), contact dermatitis (15.4%), and other variants of dermatitis (10.3%) (Figure).
Among 39 patients with pseudocellulitis, 34 (87.2%) had begun receiving oral or intravenous antibiotics at the time of enrollment. Quiz Ref IDDiscontinuation of oral and intravenous antibiotics was recommended in 28 of 34 cases (82.4%), and the primary treatment team acted on these recommendations in 26 of 28 cases (92.9%) (Table 2). The 6 remaining cases had secondary indications for antibiotic use. Additionally, dermatologists recommended initiation of topical or systemic steroids and conservative management (eg, application of compression stockings, elastic bandages, and hydration) for 26 of 39 (66.7%) and 8 of 39 (20.5%) patients with pseudocellulitis, respectively. The primary team acted on this recommendation in 24 of 26 cases (92.3%) and 7 of 8 cases (87.5%), respectively.
Quiz Ref IDDischarge from planned in-hospital observation or inpatient admission was recommended in 20 of 39 patients with pseudocellulitis (51.3%), and the primary team acted on this recommendation in 17 of 20 cases (85.0%) (Table 2). The remaining 19 (48.7%) had another medical condition requiring inpatient observation or admission.
All enrolled patients received in-clinic or telephone follow-up during the 30 days following discharge, with an overall response rate of 75.0% (87 of 116). Thirty of 39 patients with pseudocellulitis (76.9%) and 57 of 77 with cellulitis (74.0%) were reached through a clinic or telephone encounter during the first month after discharge. Quiz Ref IDAmong patients with pseudocellulitis, all patients with follow-up stated that the condition of the affected region had resolved, improved, or stayed the same. Findings were similar for the cellulitis cohort, apart from 1 patient who indicated worsening pain at the site, despite the physician note documenting resolution of cellulitis.
Among 39 patients with pseudocellulitis, 10 (25.6%) received antibiotics after early dermatology consultation. Additionally, 8 of 39 patients with pseudocellulitis (20.5%) were discharged with oral antibiotics. The mean (SD) length of hospital stay for patients with pseudocellulitis was 2.7 (4.8) days, with 9 patients (23.1%) discharged on the same day (Table 3). During the 30 days following discharge, 2 patients with pseudocellulitis (5.1%) experienced a drug rash and 1 (2.6%) had diarrhea, with none requiring readmission related to cellulitis.
We compared health care utilization rates of patients with lower-extremity pseudocellulitis in our study with previously published historical data from usual care at our institution (Table 3).10 Patients with lower-extremity pseudocellulitis in this study were more likely to have a same-day discharge, had a decrease in length of hospital stay of 2.0 days, were 75.0% less likely to receive unnecessary antibiotics, were 71.1% less likely to be discharged on antibiotics, and were 86.9% less likely to have complications than historical controls during the 30 days after discharge. These characteristic findings were similar when comparing historical controls with all patients with pseudocellulitis in this study (2.1-day reduction in length of stay).
Reductions in length of stay and antibiotic use from broad evaluation of patients with suspected cellulitis in the United States may result in 97 000 to 256 000 avoided inpatient hospitalization days and in 34 000 to 91 000 patients avoiding unnecessary antibiotics annually. Notably, each dermatology consultation for presumed cellulitis could, on average, safeguard a patient from 1.6 to 2.2 antibiotic days (eTable 1 and eTable 2 in the Supplement).
Less antibiotic use leads to fewer antibiotic-associated adverse effects. Every decile of patients with presumed cellulitis screened by a dermatologist may prevent 90 to 1135 cases of C difficile colitis, 0 to 1 cases of antibiotic-induced anaphylaxis, and 90 to 238 nosocomial infections annually (Table 4).
Based on the Household Component of the Medical Expenditure Panel Survey for hospital inpatient stays in 2010, the cost savings associated with a 2.1-day reduction in hospitalization was $2146.68.19 Across the United States, this reduction leads to an estimated annual cost reduction of $100 million to $260 million stemming from avoidance of unnecessary hospitalizations for pseudocellulitis. After considering consultation-related expenditures of between $20 million and $50 million, we estimate net savings of between $80 million and $210 million annually (eTable 1 in the Supplement). These estimates translate to an average of $582.39 in net cost savings (after accounting for consultation fee) per dermatology consultation for presumed cellulitis (eTable 3 in the Supplement).
Our data demonstrate the clinical and economic impact of early dermatology consultation in patients admitted with presumed cellulitis. Of 116 patients with presumed cellulitis, dermatology consultation determined that 39 (33.6%) were incorrectly diagnosed. Dermatology consultation also resulted in antibiotic discontinuation and discharge from planned observation or inpatient admission in 92.9% and 85.0% of recommended cases, respectively. Compared with historical controls, patients with pseudocellulitis in our study had a reduction in length of stay by 2.1 days.
Our findings support and expand on prior studies4,5 that have demonstrated substantial pseudocellulitis misdiagnosis rates. When extrapolated across the United States, universal dermatology consultation for cellulitis may lead to a reduction of 97 000 to 256 000 inpatient days. Additionally, this intervention is estimated to save 34 000 to 91 000 patients with presumed cellulitis from unnecessary antibiotic exposure each year. These results reveal a major opportunity to reduce mortality and morbidity associated with overhospitalization and unnecessary antibiotic exposure.20-22
Although these results are promising, broad implementation of dermatology consultation for presumed cellulitis will be challenging. The dermatologic workforce is currently ill-equipped to deal with the influx of inpatient consultations that a broad screening program would entail. Timely access is a challenge nationally for dermatology, and it is unclear whether reallocation of dermatologic resources to the inpatient service en masse would be ideal from a societal perspective.23,24 Additionally, the economic models to support seeing patients in the ED or inpatient settings are nascent and at present insufficient to support dermatologists seeking to change practice patterns.
Preventing unnecessary admission to the hospital and reducing antibiotic use, while beneficial to the patient and correct from an ethical perspective, leads to lost revenue in the current fee-for-service environment. To enable dermatology consultation more broadly, we need to align reimbursement systems with what is best for patients and create an environment that rewards dermatologists (and other clinicians) for saving money by reducing unnecessary use.
While advocating these changes, we must simultaneously experiment with alternative delivery models that allow increased access to timely inpatient dermatologic consultations while minimizing unintended consequences on clinical practice schedules. The use of teledermatology for inpatient triage has been evaluated, but the accuracy and reliability of teledermatology in patients with presumed cellulitis have not been determined.25 The number of dermatology consultations may be reduced by using diagnostic tools such as the ALT-70 cellulitis score or bedside diagnostics such as thermal imaging.26,27 Although these efforts may incrementally improve diagnostic accuracy for cellulitis, the ultimate goal for investigators should be the identification of a sensitive and specific point-of-care biomarker for cellulitis.
Despite challenges to the implementation of dermatology consultation for presumed cellulitis, we believe our findings serve as a call to action for dermatologists nationally to become more proactively involved with the treatment and evaluation of this disease. Although interventions to educate nondermatological specialties on recognition of pseudocellulitis may theoretically decrease misdiagnosis rates, current efforts have been limited in their efficacy.28 While broad-based change may be difficult, there is a distinct value in incrementalism, and optimizing services where possible will play a large role in improving patient outcomes while reinforcing the role of dermatology within the house of medicine.
Our findings must be interpreted in the context of the study design. This investigation was carried out at a single institution, and although patients span a wide range of age and racial demographics, further evaluation may be needed to assess the generalizability of our findings. We compared the results of our intervention with recent historical data at our institution. It is possible that other changes in hospital triage and flow are affecting patient stay and antibiotic use, although no other major efforts have been undertaken for these conditions. We estimated complication rates, and additional large-scale prospective studies may be needed to evaluate the spectrum of potentially avoidable complications from unnecessary antibiotic exposure. In this study, dermatology evaluation was considered the gold standard for evaluation of cellulitis. Among all patients with pseudocellulitis for whom dermatologists recommended treatment discontinuation and discharge, none had worsening outcomes during follow-up, suggesting robust diagnostic accuracy. However, it is possible that some cases were misclassified as cellulitis and that pseudocellulitis cases were in fact underdiagnosed.
The results of our prospective study support the findings of our previous retrospective investigation on the costs and consequences associated with misdiagnosis of lower-extremity cellulitis.10 Our findings suggest that dermatology consultation confers substantial clinical and economic benefit for cases of presumed cellulitis. Further studies are needed to confirm our estimates regarding the costs and benefits of our intervention. Moving forward, collective efforts between EDs and dermatologists to improve diagnostic accuracy of cellulitis may allow cost-effective optimization of health outcomes for patients through the reduction of unnecessary antibiotic use and hospitalizations.
Accepted for Publication: November 16, 2017.
Corresponding Author: Arash Mostaghimi, MD, MPH, MPA, Department of Dermatology, Brigham and Women’s Hospital, 221 Longwood Ave, Boston, MA 02115 (firstname.lastname@example.org).
Published Online: February 16, 2018. doi:10.1001/jamadermatol.2017.6197
Author Contributions: Drs Li and Mostaghimi had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Li, Xia, Pallin, Baugh, Laskowski, Mostaghimi.
Acquisition, analysis, or interpretation of data: Li, Xia, Khosravi, Dewan, Pallin, Laskowski, Joyce, Mostaghimi.
Drafting of the manuscript: Li, Khosravi, Joyce, Mostaghimi.
Critical revision of the manuscript for important intellectual content: Li, Xia, Khosravi, Dewan, Pallin, Baugh, Laskowski, Mostaghimi.
Statistical analysis: Li, Joyce, Mostaghimi.
Obtained funding: Li, Laskowski, Mostaghimi.
Administrative, technical, or material support: Khosravi, Dewan, Pallin, Baugh, Mostaghimi.
Study supervision: Khosravi, Mostaghimi.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by the National Center for Advancing Translational Sciences, National Institutes of Health (grant TL1TR001062) and by the Brigham and Women’s Physicians Organization Brigham Care Redesign Incubator and Startup Program (BCRISP).
Role of the Funder/Sponsor: The funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This study was presented at the 2018 American Academy of Dermatology Annual Meeting; February 16, 2018; San Diego, California.