Experimental Design Details
Sample Size
The trial sample size was 213, calculated using Cochran's formula with adjustments for continuous outcomes, anticipated dropout rate, and clustering. Initially planned for 116 participants, the study expanded to 221 to enhance robustness using the design effect. The trial included criteria for potential cessation, including undue risk or adverse effects, significant negative impact on the control arm, and ineffectiveness of the IMB intervention. However, the trial proceeded without interruptions as none of these conditions were met.
Randomization
The study team generated a computer-based random allocation sequence. Participants were then randomized in a 1:1 ratio to either the IMB or MHE intervention groups. Allocation concealment was maintained by having participants select paper balls marked "I" or "C" from a container, with separate selections for male and female participants to ensure gender balance. The selection process was further refined by subdividing participants into smaller groups. This process ensured that the intervention assignment was concealed until the moment of selection, preserving the integrity of the randomization. Research assistants conducting baseline and follow-up assessments were blinded to the intervention assignments to prevent bias in outcome assessment.
Allocation Concealment Mechanism
The random allocation sequence was implemented through participant-driven selection from a container, ensuring concealment until intervention assignment. Participants selected paper balls marked "I" for IMB or "C" for MHE from a container. This procedure was carried out separately for male and female participants to ensure gender balance and further refined by subdividing them into smaller groups for optimal group dynamics.
Blinding
Research assistants conducting assessments at baseline and follow-up were blinded to the participants' assigned intervention groups.
Data collection and management
A parallel randomized controlled clinical trial was conducted at Butabika and Masaka hospitals from January 13th to 24th, 2020. The study utilized a mixed-methods approach for data collection and management, encompassing both quantitative and qualitative elements. It involved adult outpatients diagnosed with both Risky Sexual Behavior (RSB) and Severe Mental Illness (SMI) who were systematically randomized into either the Information-Motivation-Behavioral Skills (IMB) intervention group or the Mental Health Education (MHE) control group. Allocation concealment was ensured through a participant-driven selection process, and research assistants were blinded to group assignments. The IMB group received six comprehensive sessions incorporating educational videos, discussions, and demonstrations, while the MHE group received standard mental health education. Participants were followed up three months later from April 15th to 30th, 2020. Participants were subjected to a comprehensive 69-item reproductive health checklist, encompassing information, behavior, and motivation constructs, was administered to assess participants' reproductive health knowledge, attitudes, and practices. Data collection involved interviewer-administered face-to-face paper questionnaires, which were stored securely and reviewed for completeness and accuracy. A study database was designed in OpenClinica, with unique identifiers for each participant. Data underwent double entry and regular quality checks, with any discrepancies resolved promptly. Once validated, the data was exported to Stata for statistical analysis. The master database was maintained on a secure server with password-protected copies accessible to authorized personnel. A detailed data dictionary was prepared to facilitate data curation.
Data analysis
Data were summarized using descriptive statistics, including frequencies and proportions for categorical variables, and means and standard deviations for continuous variables. The primary analysis employed intention-to-treat (ITT) principles, utilizing random effects linear regression models to evaluate the intervention's effect on the continuous outcome variable while adjusting for baseline imbalances. Subgroup and sensitivity analyses were conducted to assess the robustness of the findings and explore potential effect modifiers.
The analysis accounted for potential biases, such as attrition and missing data, and explicitly stated model assumptions and limitations. All statistical analyses were performed using Stata version 17, with detailed documentation of syntax and commands to ensure reproducibility.