Test-Retest Reliability of a Timeline Follow-back Method to Assess Opioid Use and Treatment

Calendar-based timeline follow-back (TLFB) instruments have been used to assess alcohol use, smoking, and other behaviors. We assessed test-retest reliability of an adapted TLFB addressing opioid-related outcomes over 120 days among opioid overdose survivors using nonprescribed opioids.

The Repeated-dose Behavioral intervention to reduce Opioid Overdose Trial utilized a TLFB that collected data over the preceding 120 days. A subset of participants was administered a retest TLFB 3-21 days after their TLFB assessment. The test and retest assessed days of opioid and medication for opioid use disorder (MOUD) use, hospitalization, residential substance use disorder (SUD) treatment, incarceration, and overdose during overlapping time periods. For outcomes reported by ≥15% of the sample, intraclass correlation coefficients (ICC) were calculated between test and retest. ICC > 0.9 was considered "very high" reliability. For outcomes reported by <15%, frequencies were described; statistical tests were not conducted.

Seventy-seven participants completed a retest. On the test/retest, most participants reported opioid (87%/83%) and MOUD (58%/60%) use. Median (IQR) number of days of opioid and MOUD use on the test/retest was 71 (25-117)/86 (23-108) and 4 (0-72)/5 (0-79) days. ICC between test and retest was >0.9 for both opioid and MOUD use. On test/retest, few participants reported hospitalization (8%/9%), residential SUD treatment (3%/3%), incarceration (5%/7%), or overdose (4%/3%).

The adapted TLFB had very high reliability for self-reported opioid and MOUD use over 120 days. For less frequent outcomes, including overdose, a higher frequency or larger sample size is needed to assess reliability.