Risk Factors and Pooled Rate of Prolonged Opioid Use Following Trauma or Surgery: A Systematic Review and Meta-(Regression) Analysis.

BACKGROUND

Prolonged use of opioids initiated for surgical or trauma-related pain management has become a global problem. While several factors have been reported to increase the risk of prolonged opioid use, there is considerable inconsistency regarding their significance or effect size. Therefore, we aimed to pool the effects of risk factors for prolonged opioid use following trauma or surgery and to assess the rate and temporal trend of prolonged opioid use in different settings.

METHODS

Following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, we searched Embase, PubMed, Web of Science, EBM (Evidence-Based Medicine) Reviews - Cochrane Database of Systematic Reviews, and ClinicalTrials.gov from inception to August 28, 2017, without language restriction. Observational studies reporting risk factors for, or the rate of, prolonged opioid use among adult patients following surgery or trauma with a minimum of 1 month of follow-up were included. Study and patient characteristics, risk factors, and the rate of prolonged opioid use were synthesized.

RESULTS

Thirty-seven studies with 1,969,953 patients were included; 4.3% (95% confidence interval [CI] = 2.3% to 8.2%) of patients continued opioid use after trauma or surgery. Prior opioid use (number needed to harm [NNH] = 3, odds ratio [OR] = 11.04 [95% CI = 9.39 to 12.97]), history of back pain (NNH = 23, OR = 2.10 [95% CI = 2.00 to 2.20]), longer hospital stay (NNH = 25, OR = 2.03 [95% CI = 1.03 to 4.02]), and depression (NNH = 40, OR = 1.62 [95% CI = 1.49 to 1.77]) showed some of the largest effects on prolonged opioid use (p < 0.001 for all but hospital stay [p = 0.042]). The rate of prolonged opioid use was higher in trauma (16.3% [95% CI = 13.6% to 22.5%]; p < 0.001) and in the Workers' Compensation setting (24.6% [95% CI = 2.0% to 84.5%]; p = 0.003) than in other subject enrollment settings. The temporal trend was not significant for studies performed in the U.S. (p = 0.07) while a significant temporal trend was observed for studies performed outside of the U.S. (p = 0.014).

CONCLUSIONS

To our knowledge, this is the first meta-analysis reporting the pooled effect of risk factors that place patients at an increased chance for prolonged opioid use. Understanding the pooled effect of risk factors and their respective NNH values can aid patients and physicians in developing effective and individualized pain-management strategies with a lower risk of prolonged opioid use.

LEVEL OF EVIDENCE: Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.