Impact of Urine Drug Screening on No Shows and Dropouts among Chronic Pain Patients: A Propensity-Matched Cohort Study.

BACKGROUND The last 2 decades have seen a substantial increase in both the prescription of opioids for managing chronic pain, and an increase in opioid-related deaths in the US. Urine drug screening (UDS) is the de facto monitoring tool aimed at detecting and deterring opioid misuse. OBJECTIVE We study whether administering UDS on pain patients influences post-screening behavior of no-shows and dropouts. STUDY DESIGN Observational cohort study of electronic medical records. SETTING Single urban academic pain-clinic. METHODS A retrospective cohort comparison of patients receiving UDS versus those not receiving UDS was conducted on the entire sample as well as in the propensity score-matched samples in which matching was based on age, gender, pain-score, procedure-scheduled, systolic and diastolic blood pressure (BP), pulse, temperature, physician ID, year of visit, psychology referral, and opioid prescription in the first visit. In addition, we conducted within-subjects logistic-regression to study no-shows and non-proportional hazards survival modeling to study dropout. RESULTS Analyses of 4,448 clinic visits by 723 pain patients indicated that UDS exposure in the first visit is associated with increased risk of no-show in the second visit (OR = 2.73, P < .0001); no-show rate was 10.24% for those without UDS compared to 23.75% for those with a UDS. Among those tested, the no-show rate was higher for those testing positive for illicit substances (34.57%) than for those testing negative (21.74%). These findings were replicated in 8 different propensity-score matched subsamples aimed at addressing potential non-random selection, as well as in within-subject analysis accounting for individual-level no-show propensity. Non-proportional hazards survival analysis shows that risk of dropout increased by 100.3% with every additional UDS (HR 95% CI: 1.54 to 2.61). LIMITATIONS Retrospective design, non-randomized sample, single-setting. CONCLUSIONS The results indicate that UDS is associated with increased no-shows and dropout from clinic subject to limitations of observational studies such as selection bias and confound by unobserved variables. These results serve as a call for additional prospective randomized studies to understand the impact of UDS, and where the patients might go when they dropout from the clinic.