Opioid use is common among Social Security Disability Insurance (SSDI) beneficiaries, who account for a disproportionate share of opioid-related hospitalizations and mortality in the United States. However, little is known about the prevalence of opioid use prior to SSDI enrollment. Understanding when opioid use is established and how it correlates with individual characteristics and community prescribing practices would inform policy approaches to reducing opioid-related harm among SSDI beneficiaries. We estimated the prevalence of opioid use among SSDI applicants by applying a natural language processing algorithm to SSDI application data. We found the prevalence of opioid use among SSDI applicants declined from 33% in 2013 to 24% in 2018. In contrast, the share of applicants with musculoskeletal impairments, which are commonly associated with pain, was unchanged. The share of applications reporting opioid use declined across both sexes, all age groups and education levels, and all regions. There was substantial variation, however, in the magnitude of decline by geography, with the smallest declines in parts of the Midwest and Southeastern United States. SSDI application rates and applications reporting opioid use were more likely to come from communities with higher opioid prescribing rates. Our estimates suggest most SSDI beneficiaries began opioid use prior to entering the SSDI program.