Protocol for the effectiveness of multimonth refill of antituberculosis drugs (MORAD) on treatment success among people with drug-susceptible tuberculosis in rural eastern Uganda: a non-inferiority randomised trial

Study setting

The MORAD study will be implemented across 10 TB clinics in 4 districts in rural eastern Uganda (table 1). In this region, the majority of people with TB travel ≥5 km to access TB treatment leading to suboptimal treatment outcomes.5 6 The study will target health facilities with high patient load (50 or more people with TB registered and treated annually) in order to feasibly reach the desired sample size. The TB clinics are managed by experienced medical, clinical or nursing officers who are the focal persons for TB and all the clinics adhere to the National TB Control Programme guidelines for standardised care.

Trial design

The MORAD study is designed as a two-arm, non-inferiority, open-label, individually randomised trial. Findings will be reported per the Standard Protocol Items: Recommendations for Interventional Trials guideline.12 The protocol is registered with the Pan African Clinical Trials Registry (PACTR202403586718783). Participants will enrol in the study starting 17 June 2024 and ending 31 October 2024, with the anticipated date of the last follow-up as 31 April 2025.

The participants will include people with drug-susceptible TB aged ≥15 years being initiated on TB treatment. We will include those with clinically diagnosed pulmonary TB (PTB), bacteriologically confirmed PTB and extrapulmonary TB, receiving the standard 6-month regimen at the study sites. We will exclude those with TB meningitis and osteoarticular TB as their treatment lasts longer than 6 months (including potential for modification depending on the treatment response), those critically ill requiring close clinical and laboratory monitoring including hospitalisation and those likely to migrate during the trial. Participants will be consecutively sampled and distributed proportionately to the size of the TB clinic, guided by existing data on the number of people with TB who are treated annually.

Expert clients (people successfully treated for TB who assist at the TB clinic) will provide non-technical study information during health talks. Research assistants will screen and recruit eligible individuals using standardised screening and enrolment forms. No additional strategies for participation retention will be implemented rather than providing key TB messages under routine care.

Randomisation and blinding

We will perform a stratified block randomisation by study site to ensure balance in important covariates including the number of participants across the trial arms. We will stratify the participants by sex (males vs females) and within each stratum, a block size of 4 will be constructed, and the research assistants will randomise eligible individuals to the intervention or control arm. The randomisation sequence will be a priori-determined, sealed in a well-marked opaque envelope, and only opened whenever eligible individuals consent to participate in the study. Randomised individuals will thereafter be linked to the TB focal person for anti-TB treatment initiation. Blinding of the participants and TB focal persons will be impractical but the data analyst will be blinded to the study arms using unique codes during data analysis.

Intervention versus control group

Table 2 shows the scheduling of anti-TB refills. All participants will receive a standard 6-month anti-TB regimen comprising rifampicin, isoniazid, pyrazinamide and ethambutol for 2 months, followed by RH for 4 months (2RHZE/4RH) as a fixed-dose combination, same key TB messages according to national guidelines, and Medication Event Reminder Monitor (MERM) to track TB medication adherence. In the intervention group, the participants will receive multimonth refills of anti-TB drugs, with the treatment schedule as monthly for 2 months followed by a 2-month refill for 4 months (overall, 4 visits). People with bacteriologically confirmed PTB will receive sputum smear follow-up testing at 2 and 6 months, which is sufficient to establish whether the individual is cured or not. The timing and frequency of MORAD will be modified based on the findings from a formative qualitative study. In the control group, participants will receive routine care, with the anti-TB treatment refills scheduled biweekly for 2 months followed by monthly for 4 months (overall, eight visits). People with bacteriologically confirmed PTB receive sputum smear follow-up testing at 2, 5 and 6 months. Individuals who withdraw from the intervention group will automatically fit into the control group.

Study outcomes

TB treatment success will be the primary outcome, measured by treatment completion or cure being recorded as the treatment outcome in the TB treatment register. Individuals with treatment outcomes of failure, death or loss to follow-up will be considered unsuccessfully treated. The secondary outcomes will include monthly adherence to anti-TB treatment and sputum smear conversion at months 2 and 6 among those with bacteriologically confirmed PTB. The monthly anti-TB adherence rate will be calculated as the number of pills taken divided by the number expected to have been taken, expressed as a percentage. Adherence rates <85% will be considered poor, 85%–94% as good and ≥95% as excellent.

Sputum smear conversion will be measured as a change in sputum smear status from positive to negative among patients with bacteriologically confirmed PTB following sputum smear microscopy testing. Non-sputum smear converters will be managed according to the national TB treatment guidelines.

Data collection methods and tools

Data will be collected using a researcher-administered questionnaire and validated tools (table 3). Baseline sociodemographic data will include age, sex, level of education, employment, marital status and the estimated distance to the TB treatment unit in kilometres. Clinical data will include the type of person with TB (new vs previously treated), TB disease classification, HIV status, the form of directly observed therapy short course, treatment supporter availability and type, and pre-existing conditions (eg, diabetes mellitus, hypertension and asthma). Health facility-level data will include the type of ownership, level of care and location (rural vs urban).

Social and lifestyle data will include alcohol consumption and cigarette smoking. Mental health data will include psychological distress and depression. Alcohol consumption will be measured using the Alcohol Use Disorders Identification Test Consumption (AUDIT-C) tool,13 a 10-item screening tool although the initial measurement will use 3 questions each measured on a 0–4 scale (the overall score will be 12). Individuals with AUDIT-C scores ≥5 will receive a full assessment using a set of 7 extra questions to evaluate for harmful alcohol consumption. The risk of harmful alcohol consumption will be graded based on total scores as 0–7, 8–15, 16–19 and ≥20 to suggest low, increasing, high and possible dependence, respectively. The Kessler Psychological Distress Scale, a 10-item checklist with 5 value responses ranging from 10 to 50, will be used to evaluate psychological distress14 with total scores of 10–15, 16–21, 22–29 and 30–50 to suggest low, moderate, high and very high psychological distress, respectively.

Depression will be assessed using the Patient Health Questionnaire-9, a 9-item checklist that screens for depression in the past 2 weeks with each question scored on a 0–3 scale,15 overall 0–27. Individuals with total scores of 0–4, 5–9, 10–14, 15–19 and 20–27 will be considered to have minimal, mild, moderate, moderately severe and severe depression, respectively. All individuals with severe depression will be linked to a mental health unit for further assessment and management.

Adherence will be measured using MERM each month, computed and reported at each clinic visit by the TB focal person. MERM is a digital pillbox that automatically reminds patients to take their medications, enabling real-time monitoring of adherence.16 For this study, the MERM notifications will be turned off in order to prevent sending automatic messages and impacting adherence and study outcomes. The MERM will have multiple blister packs for anti-TB drugs to support the daily dosing and refills. The MERM will collect data on dosing histories based on pillbox opening, and transmit the data to a central server which will then be downloaded by TB focal persons at refill visits to document the monthly average adherence rates.16 The treatment outcomes data, categorised into the six WHO standard outcomes, namely cure, completed, failure, loss to follow-up, transfer out and not evaluated will be obtained from the TB register.

Sample size estimation

We hypothesise that the intervention will improve treatment success among people with drug-susceptible TB. The rationale for this hypothesis is focused on the intervention’s potential to reduce the need for frequent TB clinic visits thereby enhancing TB treatment adherence as missed TB clinic appointments for medication pick-ups will be minimised. We anticipate a 6% difference in treatment success (effect size) between the intervention and control groups. In our recent study in rural eastern Uganda, we found 81.1% treatment success among people with drug-susceptible TB.2 Assuming a 6% non-inferiority margin, 80% statistical power, and a 1.11 design effect, the study will need 260 participants (130 intervention vs 130 control) in a 1:1 ratio. This sample size estimation was computed in R V.4.2.1 using the ‘SampleSize4ClinicalTrials’ package.17

Data analysis

We will summarise the baseline data based on the study arms. Numerical data like age will be summarised using the mean and the SD when normally distributed, or the median and the IQR when skewed. Categorical data such as sex will be summarised as frequencies and percentages. We will assess the distribution of baseline categorical data between the study arms using the χ² test for larger cell frequencies and Fisher’s exact test for smaller cell frequencies. Mean differences in numerical data will be assessed using the Student’s t-test if normally distributed or the Wilcoxon-rank sum test if skewed. The level of statistical significance will be set at 5%.

The effectiveness analysis will use the intention-to-treat (ITT) and per-protocol (PP) principles. The ITT analysis will analyse data for all participants randomised to the intervention or control arm, regardless of their adherence while the PP analysis will only analyse data for those who adhered to the randomisation arm. We will use a multivariable generalised linear mixed-model with identity link and binomial distribution, adjusting for fixed and random effects to estimate the effect of the intervention on the study outcomes and report the adjusted risk differences with a 95% CI. Variables such as age groups, sex, TB disease classification, HIV status, treatment support availability and pre-existing conditions will be considered as fixed effects while random effects will include the district and the health facility ownership type as well as the level and location. To assess if the intervention is non-inferior to routine care, we will consider a 95% CI for the risk difference being less than 6% (the non-inferiority margin). We anticipate that the intervention effect will vary based on factors such as sex, HIV status, TB disease classification and TB type. Therefore, a subgroup analysis using an interaction term between the intervention and these variables will be fitted.

Measurements and data collection

The formative qualitative study will use the Consolidated Framework for Implementation Research (CFIR) to thoroughly analyse the implementation contexts and identify both facilitators and barriers to the intervention,18 19 including exploring the intervention’s practicability, relevance and implementation strategy. Table 5 describes the CFIR domains. We will explore treatment adherence challenges among people with TB and their preferences for early follow-up, either at 2 weeks or every 2 months.

We will explore the timing and approach for involving TB treatment supporters, the type of people with TB who might benefit from the intervention, the expected benefits of the intervention for people with TB and the healthcare system and whether the intervention would lead to improvements in TB treatment success. Overall, we will ask whether the stakeholders would recommend the intervention including the reasons. IDIs targeting people with TB and their treatment supporters will be conducted by trained qualitative research assistants in the local language (‘Ateso’) within the health facility premises. The research assistants will have ≥5 years of qualitative research experience and hold a minimum diploma in health sciences. KIIs will be conducted in person at the workplace in English.

Sample size will depend on the saturation principle, a point at which no new information emerges even when additional data are collected.20 However, the a priori sample size estimate to reach the saturation point is presented in table 4. We will perform verbatim transcription of qualitative data within 24–48 hours, verify through re-reading and reading, and rectify any discrepancies. The data will be organised and analysed using the CFIR domains through content analysis by two analysts (JI and NW) to prevent subjective biases. JI is a public health specialist with nearly 10 years of experience in mixed-methods research. NW is a sociobehavioural scientist with 12 years of experience in qualitative research. In the analysis, approximately 5–10 transcripts will be first read by the analysts several times to achieve familiarity, enabling them to develop the initial codes that will form the primary codebook attained through discussions and consensus. The analysts will read the remaining transcripts several times and perform both inductive and deductive coding using the existing primary codebook. Newer codes will be allowed to emerge and the analysis might take different and unpredictable directions. Once coding is completed, the codes will be combined to form categories and then basic, organising, central and global themes, all aligned with the CFIR domains and presented with texts. We will adhere to the guidelines set by the Consolidated Criteria for Reporting of Qualitative Research.21

Quality control measures and research transparency and reproducibility

We will use pretested and validated tools for data collection. The research assistants and TB focal persons will be trained in responsible research conduct, study protocol, recruitment and data collection procedures. We will replay all audio recordings, summarise the main findings within the first 48 hours and transcribe within 5 days. A team leader will be hired to oversee the research assistants and review completed data collection tools to ensure completeness during data collection. We will input the quantitative data into Epi-Data, along with quality control measures such as skips, alerts, range and legal values. We will create a preanalysis plan and prespecify the regression model to prevent p-hacking, deposit the dataset and statistical analysis codes in Dryad and publish the dataset as Data-in-Brief.