RESUMEN
BACKGROUND: Exploring factors associated with retention in randomised trials provides insight into potential threats to internal and external study validity, and may inform the development of interventions to increase retention in future trials. Given a paucity of existing research in the field, a study was conducted to explore factors associated with retention in a smoking intervention trial involving persons with a mental illness, considering demographic and smoking characteristics, treatment condition and engagement in prior follow-up assessments. METHOD: A descriptive study was undertaken using data derived from a RCT of a smoking cessation intervention initiated in four adult psychiatric inpatient units in New South Wales (NSW), Australia. Retention assessment was undertaken at 1, 6 and 12-months post-discharge. A Generalised Linear Mixed Model was adopted to explore associations between retention at any follow up time point and demographic and smoking characteristics. Chi square analyses explored the association between retention at all follow up time points and treatment condition, and binary logistic regression analyses assessed for relationships between retention at 12-month follow up and engagement in prior follow up assessments. RESULTS: Retention rates were 63, 56 and 60% at the 1, 6 and 12-month assessments, respectively. No association was found between retention at any follow-up time point and 13 of 15 demographic and smoking characteristics. Younger participants and those who identified to be Aboriginal and/or Torres Strait Islander were more likely to be retained (both ps > 0.05). Retention rates did not vary according to treatment condition at any follow-up time point. Participants who completed a prior assessment were more likely to complete the 12 month assessment (both prior assessments: OR 10.7, p < 0.001; 6 month assessment: OR 6.01, p < 0.001; and 1 month assessment: OR 1.8, p = 0.002). CONCLUSION: The underrepresentation of younger participants and those identifying to be Aboriginal and/or Torres Strait Islander may limit the generalisability of findings. Findings suggest that inclusion of multiple contacts during a trial follow up period may increase retention at the final assessment. Interventions to improve retention, overall and for those sub-groups less likely to be retained, in smoking trials involving persons with a mental illness are needed. Further assessment of sample characteristics, and also trial design factors, associated with retention in this field is warranted.