Intraoperative Hydromorphone Decreases Post-Operative Pain. Who Would Have Thought? An Instrumental Variable Analysis

BackgroundA growing body of literature suggests that intraoperative opioid administration can lead to both increased post-operative pain and opioid requirements. However, there has been minimal data regarding the effects of the intraoperative administration of intermediate duration opioids such as hydromorphone on post-operative outcomes. Causal inference using observational studies is often hampered by unmeasured confounding, where classical adjustment techniques, such as multivariable regression, are insufficient. Instrumental variable analysis is able to generate unbiased causal effect estimates in the presence of unmeasured confounding, assuming a valid instrumental variable can be found. We previously demonstrated, using a natural experiment, how hydromorphone presentation dose, i.e. the unit dose provided to the clinician, affects intraoperative administration dose, with the switch from a 2-mg to a 1-mg vial associated with decreased administration. As the change in hydromorphone presentation dose was unrelated to any external factors, presentation dose could serve as an instrumental variable to estimate the effect of intraoperative hydromorphone administration dose on post-operative outcomes. MethodsIn this observational study with 6,751 patients, an instrumental variable analysis was employed to estimate the causal effect of an increased intraoperative administration dose of hydromorphone on post-operative pain and opioid administration. The study population included patients who received intraoperative hydromorphone as part of an anesthetic at the University of California, Los Angeles, from October 2016 to November 2018. Before July 2017, hydromorphone was available as a 2-mg unit dose. From July 1, 2017 to November 20, 2017, hydromorphone was only available in a 1-mg unit dose. A two-stage least squares regression analysis was performed to estimate the effect of intraoperative hydromorphone administration dose on post-operative pain scores and opioid administration. ResultsAn increase in hydromorphone administration caused a statistically significant decrease in Post-Anesthesia Care Unit pain scores as well as maximum and mean pain scores on post-operative days one and two, without a statistically significant effect on post-operative opioid administration. Various sensitivity analyses support the validity of the instrumental variable assumptions and suggest that the results are robust against violations of these assumptions. ConclusionsThe results of this study suggests that the intraoperative administration of intermediate duration opioids do not cause the same effects as short acting opioids with respect to post-operative pain. Instrumental variables, when identified, can be invaluable in estimating causal effects using observation data whereby unmeasured confounding is likely present.