AM Arozullah, VA Chicago Health Care System, Westside Division, Chicago, IL; WG Henderson, Hines Cooperative Studies Program Coordinating Center, Hines, IL; S Khuri, Boston VA Health Care System, West Roxbury, MA; J Daley, Partners Health Care/Massachusetts General Hospital, Boston, MA

Objectives: To determine the correlation between risk-adjusted hospital rates and rankings for postoperative mortality, postoperative pneumonia (POP), and respiratory failure (RF).

Methods: Cases were selected from those who underwent major noncardiac surgery at 44 VA hospitals participating in Phase I and II of the National VA Surgical Quality Improvement Program. Postoperative mortality was defined as death within 30 days postoperatively. POP was defined as a positive sputum culture with antibiotic treatment or an infiltrate on chest x-ray diagnosed as pneumonia or pneumonitis following surgery. RF was defined as mechanical ventilation greater than 48 hours postoperatively and/or reintubation and mechanical ventilation subsequent to postoperative extubation.

Separate logistic regression models predicting mortality, POP, and RF were developed using Phase I cases (10/91-12/93) and validated using Phase II cases (1/94-8/95). These models estimated the probability of mortality, POP, and RF for each case and the sum within each hospital represented the 'expected' number of deaths or complications. Hospital mortality, POP and RF observed to expected (O/E) ratios were calculated. Hospitals were ranked by O/E ratio resulting in six hospital outcome measures: mortality O/E ratio, mortality rank, POP O/E ratio, POP rank, RF O/E ratio, and RF rank. O/E ratio correlations were assessed using Pearson correlation coefficients. Rank correlations were assessed using Spearman rank-correlation coefficients.

Results: POP vs. RF vs. Mortality O/E ratio: There was significant correlation between POP and RF O/E ratios in Phase I (r=0.33, p=0.03) and Phase II (r=0.52, p<0.01). There was no significant correlation between POP and mortality O/E ratios in Phase I (r=0.11, p=0.49), but there was significant correlation in Phase II (r=0.37, p=0.01). There was no significant correlation between RF and hospital mortality O/E ratios in Phase I (r=0.29, p=0.054) or Phase II (r=0.21, p=0.17).

RF vs. POP vs. Mortality rank: There was a significant correlation between RF and POP rank in Phase I (r=0.36, p=.02) and Phase II (r=0.54, p<0.01). There was also significant correlation between RF and mortality rank in Phase I (r=0.32, p=0.03) and Phase II (r=0.33, p=0.03). There was no correlation between POP and mortality rank in Phase I (r=-0.02, p=0.90), but there was a significant correlation in Phase II (r=0.30, p=0.04).

In comparing Phase I O/E ratios and rankings with their Phase II counterparts, mortality O/E ratios (r=0.39, p<0.01), mortality rankings (r=0.47, p<0.01), POP O/E ratios (r=0.77, p<0.01), POP rankings (r=0.74, p<0.01), RF O/E ratios (r=0.57, p<0.01), and RF rankings (r=0.63, p<0.01) were all highly correlated across Phases.

Conclusions: There is significant correlation between risk-adjusted hospital O/E ratios and hospital rankings of POP and RF. There is significant correlation between hospital mortality ranking and RF ranking, but no correlation between mortality and RF O/E ratios. There is inconsistent correlation between hospital mortality and POP O/E ratios and rankings.

Impact: Reporting complementary measures, such as risk-adjusted hospital POP and RF rates and rankings, along with mortality rates may improve the detection of high quality and low quality hospitals by measuring dimensions of surgical care not captured by mortality rates alone.