Mean analgesic consumption is inappropriate for testing analgesic efficacy in post-operative pain: analysis and alternative suggestion

Abstract

Background and objective: Post-operative analgesic consumption is often used as a surrogate measure for pain; analyses of mean data assume a Gaussian distribution and use parametric statistics to assess statistical differences, often in small samples. We used a large individual patient dataset to examine the distribution of analgesic consumption, the validity of such analyses and alternative dichotomous outcomes.

Methods: Analysis of individual patient data from 913 patients over 48 post-operative hours in five randomised trials. Patients had either epidural injection of placebo or morphine (as sulphate and extended release epidural morphine) and use of patient-controlled analgesia. Post-operative fentanyl consumption was calculated over 0-24, 24-48 and 0-48 h.

Results: The distribution of analgesic consumption for all patients over the periods 0-24, 24-48 and 0-48 h was exponential. Most patients used less than 750 μg fentanyl over 48 h; 34% used over 1000 μg fentanyl (100 mg morphine), 13% over 2000 μg and 5% over 3000 μg. Mean, median and mode were very different; 20% of patients consumed almost 60% of post-operative analgesic, and standard deviations were generally larger than means. A useful dichotomous outcome was less than 750 μg fentanyl consumed over 48 h, a level associated with very good or excellent patient pain rating. Use of very good or excellent patient pain rating differentiated between different doses of epidural morphine.

Conclusion: Because of a highly skewed distribution, post-operative analgesic consumption is an uncertain method of measuring analgesic efficacy of an intervention designed to limit pain during and after surgery.