Menu Close

Retention rate in a longitudinal cannabis survey: lessons for future studies

Juan Perez
University of Florida

Co-Authors: Sophie A Maloney, Hanzhi Gao, Ruba Sajdeya, Gabriel A Spandau, Yan Wang, Robert L Cook, Catalina Lopez-Quintero
University of Florida

Background: Longitudinal studies have found attrition to surveys up to 30%. The overall retention rate will typically continue decreasing over time and change according to the survey modality (mail, phone, online, social media). Some studies report that incentives, such as $10 gift cards increase retention to follow-up. Additionally, cannabis cessation, reported in 16.3% of current or former cannabis users, can be associated with attrition. However, associations between participant’s characteristics with loss to follow-up status remain underreported and poorly understood.

Objectives: This study aims to describe retention rates in a cannabis longitudinal study and characterize the population lost to follow-up in terms of sociodemographics, reasons for cannabis use, and cannabis use patterns.

Methods: The Medical Marijuana and Me study (M3) study included a longitudinal survey to characterize a population of new medical marijuana users in Florida. Follow-up methods included mail, email reminders, and phone calls (3 maximum attempts per participant). The study offered a $20 card plus a $10-$20 bonus for completing the follow-up survey on time. We described retention rates among participants according to the top three reasons for use. Using bivariate analysis we compared sociodemographics, standardized self-reported measures (Generalized Anxiety Disorder-7 scale; Patient Health Questionnaire Depression scale, and Cannabis Use Disorder Test-Revised: CUDIT-R), and cannabis use patterns between participants lost to follow-up at 3-month and those completing the 3-month follow-up visit.

Results: Among 602 participants at baseline, 239 (39.7%) were lost to follow-up in the 3-month visit (median age: 37 years; ICR: 26 years). Depression was the reason for use with higher loss to follow-up (41.9%), followed by anxiety (40.4%), and chronic pain (37.6%). Between the lost to follow-up group and the group completing the 3-month visit, being female (55.2% vs. 65.6%, respectively), having a college degree or higher (65.4% vs. 78.2%), having private health insurance (44.8% vs. 53.7%) were different (p<0.05). Anxiety and depression status, main reason for cannabis use, type of product used, and cannabis experience were not associated with loss to follow-up. Conversely, having a normal CUDIT-R and using cannabis for medical purposes were associated with a complete follow-up (p<0.005).

Conclusions: In the M3 study, some characteristics informed the likelihood of being non-adherent to the study visits. This information will help researchers adjust sample size calculation and target specific subpopulations to increase study participation and reduce selection bias. Further research will help in understanding adherence to study procedures in cannabis research across the United States.