Opioid Overdose Risk in Low-Wage Online Workers

Dear Editor, We would like to commend Dunn et al. for their innovative use of Amazon’s Mechanical Turk (MTurk) to investigate risk factors for a history of opioid overdose [1]. Their large data set enabled the derivation of detailed receiver operating characteristics curves, clearly illustrating one of MTurk’s strengths as a survey platform. MTurk’s online interface may have had the additional advantage of easing respondents into disclosing sensitive personal information and reaching a new population of chronic pain (CP) patients. Such features likely contributed to the authors’ discovery that 19.3% of CP patients had a lifetime history of opioid overdose, a startling result that is almost 20-fold larger than previous estimates [2]. To the dismay of the research community, MTurk imposes a worker payment structure that incentivizes rapid task turnover (i.e., survey completion) rather than conscientious responses unless investigators employ safeguards such as knowledge assessment questions, comprehension assessment questions, and limiting survey access to MTurk workers with high task approval ratings to ensure the accuracy and validity of their data. The paucity of such safeguards in this study causes us to wonder if the authors’ results primarily reflect the erroneous responses of a frequently surveyed population rather than actual CP phenomena. Support for this hypothesis may be found in Table 3, “Knowledge of opioid overdose and risk factors,” wherein respondents fare no better than chance in a series of true or false questions. Previous research into MTurk has established the importance of including questions to assess recall and comprehension in order to identify and invalidate the responses of participants who are negligent [3]. We are also curious as to how the authors decided on limiting the survey to MTurk workers with a previous task approval rate greater than 80%. Many recent clinical surveys use a cutoff of 90% and above in order to attract more reliable participants [4,5]. Despite these shortcomings, the sheer magnitude of the overdose rate observed in CP patients on MTurk coupled with the realistic rate of opioid use disorder in those receiving treatment makes the results difficult to dismiss. The authors correctly acknowledge that their study population is much younger, with a male and Caucasian predominance compared with typical chronic pain populations [2], and a 10% rate of lifetime intravenous drug use. Moreover, 15.9% of the 3,157 US-based MTurk participants who completed the screening survey indicated ongoing opioid use for CP. Rather than seeking to generalize these findings, perhaps research should be focused on studying and educating this large, potentially high-risk population of online workers with CP.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.