To validate the National Nosocomial Infections Surveillance system risk index through administrative data to predict surgical site infections.

Retrospective cohort study.

Population-based analysis in Ontario, Canada.

All elderly patients who underwent elective surgery from April 1, 1992, through March 31, 2006 (n = 469,349).

Data on procedural and patient outcomes were gathered from linked population-wide hospital discharge records and physician claims. The 75th percentile of surgical duration was estimated through anesthesiologist billing fees recorded in 15-minute increments; the American Society of Anesthesiology score of at least 3 out of 5 was estimated by diagnostic codes for severe systemic illness; and all surgeries were classified as clean or clean-contaminated because of their elective nature (thus, the maximum score on the modified index was 2).

A total of 147,216 surgeries (31%) had a score of 0;246,592 (53%) had a score of 1; and 75,541 (16%) had a score of 2 on the modified index. The 30-day risk of surgical site infection increased with each increment in the modified index (score of 0, 5.4%; score of 1, 8.0%; score of 2, 14.3%; P < .001). The association was evident for surgical site infection diagnosed during the index admission (score of 0, 2.0%; score of 1, 3.7%; score of 2, 8.9%; P < .001), as well as that associated with reoperation or death (score of 0, 0.04%; score of 1, 0.23%; score of 2, 0.73%; P < .001). The modified index predicted increases in surgical site infection risk within each of 11 surgical subgroups. In accord with past research, the modified index had modest discrimination (C statistic, 0.59), and the majority of surgical site infections (72%) occurred within lower risk strata.

The modified index predicts surgical site infection in population-based analyses and is associated with incremental increases in risk.

To develop an operational definition and a parsimonious list of postulated determinants for urban emergency department (ED) overcrowding.

A panel was formed from clinical and administrative experts in pre-hospital, ED and hospital domains. Key studies and reports were reviewed in advance by panel members, an experienced health services researcher facilitated the panel’s discussions, and a formal content analysis of audiotaped recordings was conducted.

The panel considered community, patient, ED and hospital determinants of overcrowding. Of 46 factors postulated in the literature, 21 were not retained by the experts as potentially important determinants of overcrowding. Factors not retained included access to primary care services and seasonal influenza outbreaks. Key determinants retained included admitted patients awaiting beds and patient characteristics. Ambulance diversion was considered to be an appropriate operational definition and proxy measure of ED overcrowding.

These results help to clarify the conceptual framework around ED overcrowding, and may provide a guide for future research. The relative importance of the determinants must be assessed by prospective studies.