All Purulence Is Local

Epidemiology and Management of Skin and Soft Tissue Infections in Three Urban Emergency Departments

Chris Merritt; John P Haran; Jacob Mintzer; Joseph Stricker; Roland C Merchant

Disclosures

BMC Emerg Med. 2013;13(26) 

In This Article

Methods

Study Design and Setting

A retrospective analysis of patient visits for suspected community-acquired SSTIs to three urban, academic EDs located in one New England city was performed for the first quarter of 2010 (January 1 through March 31). The EDs included an urban adult ED in a large academic hospital, a pediatric ED in the affiliated academic children's hospital, and an academically-affiliated community hospital. Together, the three EDs care for approximately 200,000 patients per year. This study was approved by the institutional review board of the Lifespan Corporation and was performed in accordance with the appropriate guidelines for protection of human subjects and protected health information.

Study Population

Two billing databases containing data for all patient visits to each of the study EDs – one from the hospitals' billing system and one from the physician practice that staffs the EDs – were combined to maximize catchment. ED visits for all patients diagnosed with SSTIs were identified from the combined database using International Classification of Diseases, 9th Edition (ICD-9) diagnosis codes 680–686 (Infection of Skin and Subcutaneous Tissue). Duplicate records due to the combination of datasets were eliminated. Repeat visits to the ED for the same SSTI by the same patient also were eliminated from the study, as were patients deemed to have likely healthcare-associated infection by virtue of having been hospitalized or having surgery within the previous 3 months, or currently residing in a skilled nursing facility. The latter were identified by record review from the study hospitals, mention in the physician chart, or identifying the patient's address at a skilled nursing facility.

Study Protocol

The protocol adhered to recommendations on the optimal conduct of retrospective studies for emergency medicine.[22] A research assistant (RA), blinded to the study objectives, reviewed and abstracted data from the electronic medical record into a data collection form, recording demographic, historical and clinical data, and the clinician's diagnosis as recorded by the treating clinician (irrespective of ICD-9 code assigned by billers). A second RA, blinded to clinical and historical data and using the hospital's microbiology laboratory reporting record, recorded whether or not a culture was ordered in the ED and recorded the resulting isolates' antimicrobial susceptibilities. RAs were trained by the primary investigator (PI), who met regularly with RAs for monitoring of case selection and data management.

ED visits identified by ICD-9 code that were in fact not for SSTIs (i.e. coding errors) were excluded following verification by the PI. The PI reviewed ten percent of records for data quality and to assess interrater reliability on three key variables. The kappa statistic for performance of culture in the ED was 0.81, for performance of I&D was 0.79, and for infection type was 0.90.

Outcome Measures

Descriptive measures included MRSA prevalence and antibiotic prescribing patterns among cultured SSTIs in the study ED populations. Primary outcomes measured were (a) the frequency of in vitro activity of ED clinicians' empiric antibiotic therapy against the cultured isolates among ED patients with cultured SSTIs, (b) factors associated with use of discordant antibiotic therapy or multi-drug antibiotic therapy, and (c) antibiotic resistance patterns among the most common pathogens identified.

Data Analysis

We estimated that 25% of all SSTI patients evaluated in the ED would undergo culture and that 90% of these patients would receive antibiotics, with 50% concordance for MRSA treatment when MRSA was isolated. Given these assumptions, between 674 and 1199 patient records would need to be abstracted to arrive at an estimate of antibiotic/culture discordance with 95% confidence intervals encompassing a range of 15 to 20 percentage points.

Antibiotic usage was stratified by culture results, and age groups were compared using Pearson's chi-square and 2-sample tests of binomial proportions.

Antibiotics were categorized based on their spectrum of activity. Anti-staphylococcal antibiotics typically active against CA-MRSA include trimethoprim-sulfamethoxazole, tetracycline, doxycycline, clindamycin, rifampin, linezolid or vancomycin.[1,2,13] Antibiotics with anti-staphylococcal properties but typically ineffective against CA-MRSA were categorized as "MSSA antibiotics": penicillins, first-generation cephalosporins, macrolides, and fluoroquinolones. "Double coverage" describes treatment with two or more antibiotics with gram-positive coverage. Three univariable logistic models were created to identify demographic or clinical variables associated with (1) in vitro coverage of any organism isolated by the empiric ED antibiotic therapy, (2) use of double antibiotic coverage, and (3) discordance between treatment and culture. In the third regression model, concordance was defined as presence of MRSA in culture when any anti-MRSA treatment was prescribed or presence of MSSA in culture when only anti-MSSA treatment was prescribed. Discordance is the converse. Data analysis was performed using Stata (StataCorp, College Station, TX).

processing....