Does performance of evidence-based processes of care improve survival for patients with community-acquired pneumonia?

Lee JS, Primack BA, Mor MK, Stone RA, Obrosky DS, Yealy DM, Fine MJ. Does performance of evidence-based processes of care improve survival for patients with community-acquired pneumonia? Poster session presented at: Society of General Internal Medicine Annual Meeting; 2011 May 4; Phoenix, AZ.

Background: Although processes of care (POCs) are frequently used as proxy measures of health care quality, it is unclear whether their performance is associated with improved survival for patients with community-acquired pneumonia (CAP). Our objective was to assess the independent associations between 4 evidence-based POCs and mortality for patients hospitalized with CAP. Methods: We studied 2076 patients with clinical and radiographic evidence of CAP directly hospitalized from 32 emergency departments in PA and CT in calendar year 2001. All patients were enrolled in a cluster-randomized trial to assess the effectiveness of guideline implementation in increasing the performance of 4 POCs at presentation (i.e., assessing oxygenation, performing blood cultures, rapidly initiating and appropriately selecting antibiotic therapy). We followed 2062 (99.3%) patients for 30 days to assess all-cause mortality. We used hierarchical multiple logistic regression modeling to assess the independent associations between individual POCs and a categorical measure of the cumulative number performed (0-2, 3, or 4) and mortality, controlling for baseline severity of illness and other patient, provider, and site characteristics. Results: Overall, 2027 (97.6%) patients had oxygenation assessed, 1314 (63.3%) had blood cultures performed, and 1632 (78.6%) received timely ( < 4 hours) and 1308 (63.0%) received appropriate antibiotic therapy; cumulatively, 534 (25.7%) patients received 0 to 2, 837 (40.3%) received 3, and 705 (34.0%) received all 4 recommended POCs. Overall, 141 (6.8%) died. As shown in the Table, mortality was 1.1% to 1.7% lower for patients who had each of the individual POCs performed compared to those who did not (p > .15 for all comparisons); mortality was 7.5% and 7.2% for patients that had 0-2 and 3 POCs performed, respectively, compared to 5.8% for patients that had all 4 POCs performed (p = .43). Consistent with the bivariate analyses, the adjusted odds ratios demonstrated no statistically significant independent associations (p > .14 for all comparisons) between individual POCs or the cumulative number performed and mortality. Conclusions: Neither performance of individual POCs, nor the cumulative number performed is independently associated with short-term survival for patients hospitalized with CAP. Although this study was not originally designed to assess process-outcome associations, our findings challenge the use of POCs as proxy measures of health care quality for CAP.