Abstract The central goal of this study is to determine the impact opioid treatment regimens have on patient reported pain scores and adverse or unintended clinical outcomes among patients with chronic, non-cancer pain (CNCP). This study will enhance our understanding of the long-term effectiveness of opioid prescribing patterns, namely dose escalations, reductions, and discontinuations. Using data from the Veterans Health Administration, this study proposal will first (Specific Aim 1) evaluate the effect of opioid dosage escalation on pain scores and adverse opioid outcomes (opioid and other drug-related accidents, overdoses, and injuries) associated with prescription opioid use. For this analysis, Veterans who have a 20% increase in their opioid dose over a 180 day period will be identified. Next, these dose escalators will be matched to dose maintainers (less than a 20% change in opioid dose). Adverse opioid outcomes and follow-up pain scores will be evaluated over the 12 month period following the first day of the qualifying dose escalation period (for maintainers, this will be ? 30 days to the maintainer's matched escalator) using Poisson/negative binomial and linear regression modeling respectively. Secondly (Specific Aim 2), this study proposal will assess the difference in pain scores and unintended clinical outcomes among Veterans who discontinue chronic opioid use. To achieve this aim, chronic opioid users with CNCP who discontinue opioids will be used. Discontinuation will be defined as a 180 day period without opioid use. First, opioid tapering regimens in the 6 months before discontinuation will be described. Second, changes in pain scores using 3 month averages will be assessed using mixed longitudinal models with tapering regimen groups as the primary independent variable. Third, the effect of opioid tapering rates on overdoses and accidents will be tested using Poisson or negative binomial models. Next, to test differences in follow-up pain scores, overdoses and accidents, service utilization, and non-opioid medication use between continuers and discontinuers of chronic opioid use, propensity scores for discontinuation will be derived, and Veterans who discontinued chronic opioid use will be matched to Veterans who continue chronic opioid use based on the propensity score. Using regression modeling among these matched pairs, follow-up pain scores will be evaluated. Overdoses/accidents and service utilization will be assessed in relation to chronic opioid use discontinuation using Poisson/negative binomial models, and non-opioid medication use will be assessed using logistic regression models. Finally, heterogeneous treatment effects (HTE) will be estimated based on race, sex, and pain conditions for differences in discontinuation, pain scores, service utilization, and non-opioid medication use. In efforts to overcome temporal confounding, marginal structural modeling (MSM) will be used as sensitivity analyses for both specific aims to adjust for the time dependent confounding between pain scores and opioid dosage changes, as pain scores will influence opioid doses which will in turn likely affect subsequent pain scores, opioid-related adverse events, and chronic opioid use discontinuation.

R36DA046717

2018-09-30

2020-08-31