Study design and sample selection
This study was a retrospective chart review conducted at nursing stations in First Nations communities in five Canadian provinces (Alberta, Saskatchewan, Manitoba, Ontario, Québec). As part of a quality improvement project at Health Canada, communities were invited to participate on a voluntary basis and were enrolled in consultation with the Regional Nursing Officers (RNO). At the request of the communities, the facility names and communities are not identified to ensure confidentiality. A simple random sample of patient paper charts was selected by choosing a random starting point in the charts drawer and pulling out the charts to be reviewed in an evenly-spaced manner until the required number of charts to be review was reached. The review was conducted by research team (DJ, HN, YS) independent of the nursing stations. Patients charts were included in the review if they had accessed care in the previous 12 months and were ≥18 years of age. Any time an antibiotic was prescribed during this one-year period, it was captured in the Antibiotic Tracking Case Report Form (CRF) for analysis.
At the time of the study design, there were a total of 85 nursing stations with the First Nations and Inuit Health Branch (FNIHB) at Health Canada. The sample size calculation was based on 12 nursing stations (14.1%) being able to participate in the study and calculated the estimated proportion of antibiotic use. -proportional allocation strategy produced numbers of charts that were needed in each province to capture the minimum number of cases (i.e. infections treated with antibiotic(s)): 45 in AB, 55 in MB, 30 in ON, 20 in BC, 35 in QC and 35 in SK). The number of charts needed was calculated based on unpublished internal audit conducted by Health Canada in 2011 on antibiotic use in the First Nations communities, it was conservatively estimated that 25% of patient charts included antibiotic prescriptions for treatment. In order to estimate the proportion of correct antibiotic treatment within a margin of error of 10%, the minimum number of antibiotic cases required to sample was calculated as n = 96 rounded to 100. Based on the estimate from the internal audit, we assumed 1 in 4 charts would contain an antibiotic case, hence the number of charts required was calculated as n = 400 to find 100 charts containing an antibiotic case. Charts were randomly selected at a rate of approximately 20 per 1000 population in the nursing station catchment area. Upon reviewing 372 charts, we found 224 charts that documented at least one antibiotic case in the previous 12 months, we ceased collecting additional data.
Measures
SSTIs were measured as both a patient-based estimate (e.g. the number of patients with at least one SSTI in the previous 12 months) and case-based estimate (e.g. the total number of unique SSTI cases in the previous 12 months, to account for repeat infections).
Demographic variables included gender and age. Clinical variables included pre-existing conditions such as diabetes, cardiac disease, alcohol use disorder, renal disease, pregnancy, liver disease, Chronic Obstructive Pulmonary Disease (COPD), immune-compromise, and/or asplenia, and presence of a medical device (e.g. urinary catheter, central line, or other).
Information on etiologic agent was collected for each unique SSTI case. CA-MRSA was defined based on the standardized epidemiological case definition proposed by US Centers for Disease Control and Prevention (CDC) (15), and the diagnosis of CA-MRSA was either confirmed by wound culture associated with the SSTI episode or presumed by existing MRSA colonization, based on previous wound culture results or an indication of MRSA colonization in the patient’s chart. Other etiologies included Methicillin-Sensitive Staphylococcus aureus (MSSA), other organism, or unknown.
Primary SSTI symptoms were categorized as purulent (abscesses, boils, folliculitis, impetigo, open wounds, sores with pus or purulent discharge), non-purulent (cellulitis, any SSTI with swelling or redness but no pus or discharge), skin breakdown (lacerations, rashes, scabs, warts and abrasions), wounds (open sores, ulcers, bite wounds and surgical site infections), or unknown. All cases were considered SSTI in the present analysis as they were treated as infections with antibiotic(s) prescription at the nursing stations. Treatment options included wound care, incision and drainage, prescribed antibiotic, type of antibiotic (topical only, topical and oral, oral only, and intravenous), antimicrobial combination therapy, and adjunctive therapy. Clinical outcomes were categorized as cure, failure, improvement and unknown. Data was extracted on diagnostic tests including wound culture, blood culture, complete blood count (CBC), no diagnostic test, and missing.
An evaluation of appropriateness of antibiotic use for SSTI was conducted by subject experts (YS, HN) using two guidelines - the Clinical Practice Guidelines (CPG) administered by the FNIHB (16) and the Infectious Diseases Society of America (IDSA) guidelines (17). Evaluations were based on the patients’ pre-existing medical conditions, clinical symptoms, and availability of diagnostic test results including culture and susceptibility as applicable. In cases where multiple antibiotics were used in combination, the appropriateness for the combination therapy was determined. Appropriateness was assessed based on choice of antibiotic (e.g. use of narrow or broad-spectrum agent), dosing of antibiotic (e.g. higher or lower dose than indicated), duration of prescription (e.g. longer or shorter duration) and use of combination therapy. Overall appropriateness was assessed based on the appropriateness of choice, dosing, duration of antibiotics, use of combination therapy, as well as the patient’s predisposing medical conditions.
Statistical analyses
Patient-based analyses calculated the prevalence of antibiotic use by dividing the number of patients with any antibiotic prescription by the total patient-charts sampled. Case-based analyses used the total number of unique SSTIs to summarize etiologic agents, diagnostic testing, clinical symptoms, outcomes, and appropriateness of the prescribed antibiotics. All data were analyzed using SAS statistical software, version 9.3 (18).