Study Design and Data Collection
This chart review was approved by the Research Ethics Board at Sunnybrook Health Sciences Centre in December 2005. Sunnybrook is a 1275-bed university- affiliated tertiary care hospital in Toronto, Ontario, with approximately 26 605 patient discharges per year and an average length of stay of 7 days. All patients with a chart-documented diagnosis of community-acquired pneumonia who had been patients at the hospital between January 1, 2002 (when the guidelines for treatment of community-acquired pneumonia were implemented), and December 31, 2005, and who had been discharged or had died during this period were eligible for inclusion. Because International Classification of Diseases (ICD) codes were revised during the patient eligibility period, patients admitted between January 1 and March 31, 2002, were identified by ICD-9 (ninth revision) codes for pneumonia (481 to 483.8), and patients admitted between April 1, 2002, and December 31, 2005, were identified by ICD-10 codes for pneumonia (J13, J14, J15-16.0, A48.1, and J17.0) and type M diagnosis (i.e., community-acquired pneumonia as most responsible diagnosis influencing the length of stay).
The following information was collected from each chart: patient demographic characteristics, past medical history (including comorbid conditions and past utilization of antimicrobials), patient factors that directed the requirement of hospital admission (i.e., patient factors necessary to determine pneumonia severity index; see Appendix 1), choice of empiric antibiotic therapy, results of culture and sensitivity testing, time to defervescence (defined as time to the first day that the patient remained afebrile for more than 24 h), duration of therapy, change in treatment regimen, length of hospital stay, total antimicrobial acquisition cost, and outcome.
The medical ward or critical care setting to which the patient had been admitted was documented to note concordance or discordance with the patient triage location indicated by the pneumonia severity index. The empiric therapy selected for each patient was assessed to determine whether it was appropriate (i.e., concordant with the institution’s guidelines). Empiric regimens that did not strictly fit the institution’s guidelines were considered either appropriate or inappropriate non-guideline-based therapies on the basis of an evaluation of the patient’s past medical history, initial pneumonia severity, and previous antibiotic utilization (within 6 months before admission). As examples, empiric antimicrobial therapy with levofloxacin and vancomycin for a patient admitted to hospital for community-acquired pneumonia and requiring S. aureus coverage secondary to transfer to the intensive care unit was identified as appropriate non-guideline-based therapy, whereas empiric treatment with ampicillin and ceftriaxone was considered inappropriate non-guideline-based therapy. The total antimicrobial acquisition cost for the treatment of community-acquired pneumonia was calculated for each patient. In addition, the cost of antimicrobials for guideline-based and non-guideline-based initial empiric management was determined. Costs for additional antimicrobial therapy for concurrent infections (e.g., urinary tract infection, Clostridium difficile infection) were not included. Resolution of community-acquired pneumonia was defined as any of the following clinical outcomes:
• Cure: Chart-documented resolution of infection or resolution of or improvement in signs and symptoms of community-acquired pneumonia, such as fever, crackles on auscultation, cough, sputum
purulence and volume, and leukocytosis, and discontinuation of antimicrobials.
• Death: Death due to infection or death due to any cause.
• Microbiological cure: Culture and sensitivity results indicating eradication of the causative organism(s).
Descriptive statistics, including measures of central tendency (mean, median, mode) and measures of variation (standard deviation), were calculated with Microsoft Excel 2000. Nominal data were compared with either the x2 or Fisher exact tests. Interval data were compared with one-way analysis of variance (ANOVA) with a Tukey multiple-comparison post-test analysis or, when assumptions regarding Gaussian distribution or equal variance were not valid, the Kruskal-Wallis test (nonparametric one-way ANOVA) with a Dunn multiple-comparison post-test analysis) or, for comparisons of 2 groups of interval data in this setting, the Mann-Whitney test. A p value less than 0.05 was considered statistically significant for all comparisons. The appropriate multiple-comparison post-test was completed only when p was less than 0.05. eriacta tablets