Workplace injuries, early opioid prescribing, and disability in Washington State
Abstract
INTRODUCTION: From the late 1990s through early 2010s, opioid prescribing increased dramatically in workers’ compensation (WC) systems, following national trends within the United States. The increase in opioid prescribing coincided with an increase in overdose deaths and morbidity linked to prescription opioids. While subsequent policy interventions including Prescription Drug Monitoring Programs (PDMP) and updated clinical guidelines have been associated with a decrease in opioid prescribing rates, high-risk opioid prescribing patterns, including higher duration and higher dose, persist as an area of concern for injured workers in Washington State. Prior research suggests that early opioid prescribing patterns are associated with extended work disability, although these studies have faced challenges including confounding by indication concerns and limited clinical data availability within workers’ compensation settings. In this dissertation, we (1) assessed and compared opioid exposure measures from administrative data and survey sources, (2) used administrative and survey sources to measure the association between high-risk opioid prescribing and duration of workplace disability among injured workers, and (3) conducted an interrupted time series analysis to measure the effects of recent Washington State guidelines and rules on opioid prescribing among injured workers in a workers’ compensation setting.
METHODS: In Chapter 2, we used data from Washington State’s WC system and PDMP along with surveys from injured workers to examine the degree to which opioid measures (e.g., longer duration, higher dose, and concurrent opioid/sedative prescribing) aligned between these data sources across varying time points after injury. In Chapter 3, we used a target trial framework to measure the association between high-risk opioid prescribing and the duration of disability, among (1) workers that were temporarily disabled because of their workplace injury, (2) were prescribed opioids shortly following their injury, and (3) were not prescribed opioids in the 90 days prior to their workplace injury. We defined high-risk opioid prescribing as longer duration (>7 days), high dose (≥50 mg morphine equivalent daily dose), or concurrent opioid/sedative prescribing. Low-risk opioid prescribing was defined as having at least one opioid prescription but none of the high-risk indicators shortly after injury. Using inverse probability weighting, we accounted for non-response bias and assigned opioid treatment weights while controlling for injury severity, injury characteristics, comorbidities and sociodemographic characteristics. Cox proportional hazards models were used to measure the difference in the duration of workplace disability between workers prescribed high-risk versus low-risk opioids. In Chapter 4, we conducted an interrupted time series analysis to measure the impact of state-level opioid guidelines and rules in Washington State among injured workers that required surgery for their workplace injury. Analyses were stratified by both surgical severity and whether a worker had opioids in the 90 days prior to their surgery.
RESULTS: In Chapter 2, the analysis of linked data from 10,604 injured workers revealed that PDMP data captured substantially more opioid prescriptions than workers’ compensation records. The WC system had large gaps in identifying high-risk opioid prescribing patterns, including extended duration, high-dose, and concurrent opioid/sedative prescriptions, particularly after six-weeks post-injury. The gaps between prescriptions captured in WC versus PDMP records varied by worker characteristics, including preferred language, claim severity, industry, and occupation. In Chapter 3, we found that in adjusted analyses, workers exposed to high risk opioids were 41% less likely to end temporary disability at any point in time than those prescribed low-risk opioids, on average, after accounting for important confounding variables. In Chapter 4, we found that prior to policy implementation, longer duration opioid prescribing was declining at a rate of 0.52 percentage points per month, on average, among all surgeries. Following the guidelines and rules, and contrary to expectations, this decline slowed significantly to 0.14 percentage points monthly (slope change: +0.38 percentage points; 95% CI: 0.11, 0.65). Similar decelerations were observed across surgical severity types.
CONCLUSIONS: Relying on single administrative data sources may lead to an underestimation of opioid exposure following workplace injury, with the degree of underestimation varying by worker characteristics and time since injury. In support of prior studies in workers’ compensation settings, early high-risk opioid prescribing patterns were found to be significantly associated with a higher risk of extended duration of work disability compared to early low-risk opioid prescribing patterns among temporarily disabled workers. Finally, while longer duration opioid prescribing is generally declining among injured workers requiring surgeries, this effect is waning, and recent policy interventions, contrary to our expectations, resulted in a modest slowing of pre-existing downward trends in longer duration opioid prescribing. Improvements to data quality and data collection should be a priority to further improve workers’ health outcomes in Washington State.