Background
The many risks associated with opioid therapy for chronic non‐cancer pain (CNCP) have led to questions about use. This is particularly relevant for risk of increased mortality. However, underlying medical conditions of those using opioids may influence mortality findings due to confounding by indication. Similarly, non‐opioid analgesics are also associated with an increased risk of mortality, too.
Methods
We have conducted a systematic review of propensity score matched observational studies comparing mortality associated with opioid use compared to non‐opioid analgesics. Clinicaltrials.gov, Google Scholar, MEDLINE and Scopus were searched from inception to July 2020. Propensity score matched observational studies comparing opioids to non‐opioid analgesics in real‐world settings were analysed. Primary outcome was pooled adjusted hazard ratio (aHR) of all‐cause death. Effects were summarized by a random effects model.
Results
Four studies with seven study arms and 120,186 patients were analysed. Pooled aHR for all‐cause death was 1.69 (95% confidence interval [CI] 1.47, 1.95). When mortality risk was confined to out‐of‐hospital deaths, the pooled aHR was 2.12 (95% CI 1.46, 3.09). The most frequent cause of death was cardiovascular death. Before matching, patients with opioids were older and had more somatic diseases than patients with non‐opioids. Despite extensive propensity score matchings and sensitivity analyses, all studies could not fully exclude confounding by indication.
Conclusions
Possibly, opioids are associated with an increased all‐cause mortality risk compared to non‐opioid analgesics. When considering treatment options for patients with CNCP, the possible risk of increased all‐cause mortality with opioids should be discussed.
Significance
An increased all‐cause mortality associated with opioid use compared to non‐opioid analgesics for CNCP was identified by a systematic review of four propensity score matched cohort studies in real‐world settings. The number needed to harm for an additional excess death per 10,000 person‐years was 116. Despite extensive propensity score matchings and sensitivity analyses, all studies could not fully exclude confounding by indication. The potential risk of increased all‐cause mortality with opioids should be discussed with patients when considering opioid treatment.