Volume 8, Number 12—December 2002
Research
Using Automated Health Plan Data to Assess Infection Risk from Coronary Artery Bypass Surgery
Richard Platt*†‡
, Ken Kleinman*†, Kristin Thompson*†, Rachel S. Dokholyan*†, James M. Livingston*‡, Andrew Bergman*§, John H. Mason*#, Teresa C. Horan**, Robert P. Gaynes#, Steven L. Solomon#, and Kenneth E. Sands*††
, Ken Kleinman*†, Kristin Thompson*†, Rachel S. Dokholyan*†, James M. Livingston*‡, Andrew Bergman*§, John H. Mason*#, Teresa C. Horan**, Robert P. Gaynes#, Steven L. Solomon#, and Kenneth E. Sands*††Table 4
Adjusted hospital specific risksa
| Patients with at least one indicator code for infection | Hospital B vs. A | Hospital C vs. A | Hospital D vs. A | Other hospitals vs. A | p value | |
|---|---|---|---|---|---|---|
| Including pharmacy data | 536/1,953 | 0.68 (0.49–0.94)b | 1.03 (0.68–1.55) | 1.57 (1.16–2.13) | 0.91 (0.67–1.24) | <0.0001 |
| Excluding pharmacy data | 363/1,953 | 0.84 (0.58–1.20) | 0.92 (0.56–1.50) | 1.62 (1.15–2.28) | 1.05 (0.74–1.50) | 0.03 |
aAdjusted for health plan, age, and sex. The interaction between health plan and hospital was not significant in any of these models. Similar results were obtained in models adjusting for chronic disease score (a composite of age, sex, and pharmacy information), instead of age and sex.
bAdjusted odds ratios and (95% confidence intervals).
