Background and Objectives The effect of postoperative epidural analgesia (vs. systemic analgesia) on patient outcomes is unclear. Available randomized controlled trials (RCTs) have focused on the intraoperative period and not properly examined the effect of postoperative epidural analgesia (EA) on outcomes.

Methods A 5% nationally random sample of Medicare beneficiaries from 1994 to 1999 was analyzed to identify patients undergoing total hip arthroplasty (Common Procedural Terminology [CPT] code 27130, 27132, 27134, 27137, 27138). Patients were divided into 2 groups depending on the presence or absence of postoperative EA based on the CPT coding (01996). The rate of major morbidity (acute myocardial infarction, deep venous thrombosis, pulmonary embolism, angina, respiratory failure, heart failure, cardiac dysrhythmias, pneumonia, pulmonary edema, sepsis, acute renal failure, paralytic ileus, acute cerebrovascular event) and death at 7 and 30 days after the procedure were compared. Multivariate regression analysis was performed to determine if the presence of postoperative (EA) had an independent effect on mortality or major morbidity. Data were reported as an odds ratio with 95% confidence intervals (CI) when appropriate.

Results The unadjusted 7- and 30-day death rate was significantly lower for EA versus no EA (1.9/1000 [95% CI: 0.2-3.6] vs. 3.9/1000 [95% CI: 3.0-6.2] at 7 days [P = .04] and 5.8/1000 [95% CI: 2.9-8.7] vs. 9.9/1000 [95% CI: 8.6-11.3] at 30 days [P = 0.01]). However, multivariate regression analysis revealed that there was no difference between the groups with regard to mortality or major morbidity with the exception of an increase in deep venous thrombosis in patients who received EA.

Conclusions The use of postoperative EA was not associated a lower incidence of mortality and major morbidity in Medicare patients undergoing total hip arthroplasty. However, the results should be interpreted with caution because of limitations in using the Medicare claims data for analysis. Further trials using other properly conducted and designed studies (e.g., RCTs) would be ideal to validate these results.