Background and Objectives
Despite limited analgesic benefits, long-term opioid therapy (L-TOT) is common among older adults with chronic pain. Extended opioid use poses a threat to older adults as aging metabolisms retain opioids for longer, increasing the risk of injury, overdose, and other negative health outcomes. In contrast to predictors of general opioid use, predictors of L-TOT in older adults are not well-documented. We aimed to identify such predictors using all available data on self-reported opioid use in the Health and Retirement Study.
Research Design and Methods
Using five waves of data, respondents (N=10,713) aged 51 and older were identified as reporting no opioid use (n= 8,621), a single wave of use (n=1,410), or multiple waves of use (n=682). We conducted a multinomial logistic regression to predict both single- and multi-wave opioid use relative to no use. Demographic, socioeconomic, geographic, health, and healthcare-related factors were included in our model.
Results
Multivariable findings show that, relative to non-users, both single- and multi-wave users were significantly more likely to be younger (RRR=1.33;RRR=2.88); report lower household wealth (RRR=1.47;RRR=2.88); live in the US Midwest (RRR=1.29;RRR=1.56), South (RRR=1.34;RRR=1.58), or West (RRR=1.46;RRR=2.34); experience interfering pain (RRR=1.59; RRR=3.39), back pain (RRR=1.35;RRR=1.53), or arthritic pain (RRR=1.46;RRR=2.32); and see the doctor frequently (RRR=1.50;RRR=2.02). Multi-wave users were less likely to be Black (RRR=0.69) or Hispanic (RRR=0.45), and less likely to be never married (RRR=0.52).
Discussion and Implications
We identified demographic, socioeconomic, geographic, and healthcare-related predictors of chronic multi-year opioid use. Our focus on individuals taking opioids for this extended duration is novel. Differences in opioid use by geographic region and frequency of doctor visits particularly warrant attention from policymakers and researchers. We make additional recommendations based on a sensitivity analysis limited to 2016-2020 data.