ImPROving TB outcomes by modifying LIFE-style behaviours through a brief motivational intervention followed by short text messages (ProLife): study protocol for a randomised controlled trial

The primary objective is to assess the effectiveness and cost-effectiveness of the ProLife programme compared to usual care in improving pulmonary TB (PTB) treatment outcomes. We also aim to assess the effectiveness and cost-effectiveness of the programme in achieving abstinence from smoking, reducing harmful and hazardous drinking, and improving TB and ART medication adherence.

This is a pragmatic, prospective, multicentre, two-arm, parallel, individual RCT taking place in 27 purposively selected primary care clinics with the highest TB case-load in three districts in South Africa: Welkom in the Free State; Bojanala in the North West province; and Sedibeng in Gauteng province. To be eligible for inclusion in the trial, TB clinics had to be under the control of the provincial or local government (i.e. not a mining TB clinic). The intervention will be delivered by LHWs and three district coordinators who will each cover 1–2 clinics.

Participants will be randomly allocated to one of two arms:

The ProLife programme is a complex behavioural intervention aimed at improving treatment outcomes in TB patients who smoke tobacco, drink alcohol at harmful or hazardous levels, or do both. The ProLife programme has been developed in line with the Medical Research Council (MRC) guidance on developing and evaluating complex interventions [35] and has undergone a period of development and feasibility testing in South Africa. The paper reporting the results of this evaluation has not yet been published. The conceptual framework used to develop the ProLife programme assumes that smoking cessation, reducing harmful alcohol use and improved adherence to TB and HIV treatment will result in improved TB treatment outcomes.

Participants randomised to the intervention arm will receive three brief MI counselling sessions, lasting 15–20 min, each one month apart, from a trained LHW at their TB clinic. In the initial MI session, occurring at the start of TB treatment, the counsellor will establish the participant’s tobacco smoking habits and problem drinking; potential obstacles and facilitators for medication adherence or treatment initiation (both TB treatment and ART) will be determined. The participant will determine which factor should be prioritised (i.e. agenda setting), which could be a plan either to quit tobacco smoking, reduce or quit drinking or deal with barriers relating to ART or TB medication adherence. For participants who are HIV-infected and not yet on ART, beliefs and attitudes regarding HIV-testing or ART will be explored to facilitate ART initiation and adherence. The second session will build on the previous one and will focus on the previous agenda before moving on to the next behavioural problem (tobacco, alcohol or medication adherence) where applicable. The third session will deal with the last identified problem.

The individual counselling sessions will be re-enforced with SMS text messages regarding information, motivation and behaviours (IMB) supporting tobacco cessation, alcohol use and medication adherence. Text messages will be delivered twice a week over 12 weeks. All participants will first receive 10 TB-related messages. These messages will be followed by seven alcohol or smoking-related messages depending on whether the participant smokes or drinks. Co-joint users will receive all sets of messages (i.e. 24 in total).

Participants in the second group are the controls and will receive usual TB treatment and support offered to TB patients. Control participants are seen by a TB nurse and receive the same biochemical investigations and medical treatments as the intervention arm. However, they will not receive the MI and SMS package of care as described above. Usual care also includes HIV testing with pre- and post-test HIV test counselling by a lay counsellor or a nurse (this varies by district).

In addition, health education is given on:

The inclusion criteria for participants are:

Exclusion criteria:

Participants will be recruited consecutively from the starting date of the trial until the required sample size for each clinic has been achieved. The sample size per clinic will vary depending on the workload for each clinic. TB patients who initiate treatment (or who have been on treatment for < 1 month) will be asked if they would like to participate. If they agree, they will then be screened for eligibility by trained fieldworkers immediately after the TB nurse at the clinic has initiated TB treatment and started the routinely used TB treatment record. Eligibility screening will involve being assessed in line with the inclusion criteria listed above and being asked about their:

If eligible, patients will be invited to join the trial. For those who wish to be enrolled into the trial, written informed consent will be obtained (Additional file 1).

Patients will be randomised using a randomised sequence generator performed by the trial statistician (MK) who will remain blind to the arm allocation. We will use block randomisation with varying block sizes stratified by the clinic to achieve equal numbers in intervention and control groups within each clinic (see the ‘Sample size’ section). Allocation concealment will be done with consecutively numbered, sealed, opaque envelopes.

Potential participants will be approached and given an introduction to the study and basic eligibility criteria required for participation. Participants who meet the eligibility criteria for language, age, PTB, treatment duration and mobile phone access will be given a consent form (Additional file 1) to sign before screening for alcohol and tobacco use eligibility as the alcohol-related questions are sensitive and the fieldworkers must access the patient treatment record. If recruited to the trial, participants will be ask to consent to enter the trial, including consent for audio recording of the MI counselling sessions (Additional file 1). A witness signature will be required where the participant is unable to read or write.

The identity of participants will be protected by allocating each participant a unique trial number, which will be used on all research documents and will ensure anonymity for the data analysis. Participation will be voluntary; participants will be informed of their right to withdraw from the study at any time without giving a reason for their withdrawal. The project is unlikely to be directly harmful to TB patients with the exception of inconvenience in terms of time spent on the counselling sessions. Participants will receive ZAR 60 (US$4.19) travel and other expenses reimbursement for each MI counselling session and follow-up visit related to the study.

Figure 1 Additional file 4 outlines the overall participant flow through the trial. Eligible and consenting patients will be enrolled in the trial and the baseline interviewer-administered questionnaire and record review completed. Participants will receive a baseline interview on the same day as the recruitment or on the nearest available date suitable to both the participant and the LHW. The initial part of the questionnaire covers socioeconomic and demographic factors. We will also assess for depression using the CES-D10 [38], a validated tool, as depression may be a potentially moderating or mediating factor affecting treatment outcome. See Additional file 2 for the full case report form (CRF).

The next part will cover medical history, particularly TB and HIV history and current medications. Participants identified as current smokers at eligibility screening will be asked more in-depth about their smoking habits and quit history. All participants will be asked about their use of smokeless tobacco and exposure to second-hand smoke. Participants will also be asked about time and money spent on TB-related healthcare visits in the past three months and health-related quality of life (see the ‘Economic evaluation’ section). After administering the baseline interviews, fieldworkers will draw the allocation envelopes and organise the follow-up sessions for the participants. Each of the interviews on health status, tobacco and alcohol use at baseline, three months and six months is not expected to last > 25 min.

The primary outcome is TB treatment success at 6–9 months of follow-up, measured using the routinely collected programmatic TB treatment outcomes as defined by the World Health Organization (WHO) and adopted in South Africa. This is a binary variable defined as either successful treatment (cured or treatment completed) or failed treatment, death, acquired drug resistance, loss to follow-up or ‘default’, or not outcome evaluated. Table 1 defines the different treatment outcomes according to the South African Department of Health National Tuberculosis Management Guidelines [2].

Individual TB treatment records will be used as the primary source of information. The TB treatment record is the routinely used clinical document initiated by the TB nurse when the patient first presents at the TB clinic and includes demographic information as well as a comprehensive overview of the patient’s treatment over time. The TB treatment outcomes will be obtained 6–9 months after the TB treatment start dates. This is to allow for short periods of treatment interruption and time needed to confirm the TB treatment outcomes status. For example, the TB nurse may have to wait for end of treatment TB sputum or culture results or time may be required to determine whether a participant died. In some cases, TB patients may also undergo a longer TB treatment regimen.

Routinely reported outcomes will be verified for correctness by checking the actual individual diagnosis and treatment records. For example: if the outcome is classified as cured, then the criteria for diagnosis and outcome definitions should have been adhered to. Attempts will be made to verify the information of patients classified as ‘defaulted’ or who are lost to follow-up by contacting them telephonically or sending them a short text message. Sputum cultures and smears will be performed at baseline, two months, three months and six months per routine care.

The following outcome measures will also be recorded at the six-month follow-up:

Adherence will be measured using an adherence index calculated by the formula (using the four-day recall table):

ART patients with at least 95% of adherence will be considered as having optimal adherence and those with < 95% will be considered as having low (or suboptimal) adherence. For the TB medication regimen, patients with at least 95% of adherence will be considered as having optimal adherence and those with < 95% will be considered as having low (or suboptimal) adherence:

The following outcomes will be measured at three months:

The schedule for enrolment, intervention and assessments is presented in Fig. 2. Table 2 summarises the methods of data collection and analysis for each of the primary and secondary outcomes.

Data will be collected and recorded by field workers equipped with Android phones with a mobile data collection application installed. MI counselling and data collection will take place in a well-ventilated private area inside or outside the clinic (but within the grounds of the health facility); MI sessions will be audio-recorded where consent has been obtained. Fieldworkers and LHWs will be provided with high particulate respirator masks to minimise the risk of infection.

Fieldworkers will follow-up all participants in both arms at three and six months within a window period of two weeks before and two weeks after the ideal three-month and six-month visit. Participants will receive SMS reminders three days before each planned visit. Participants will also be in a position to send ‘please call me’ messages to the district coordinators, who will then call the participant to solve problems that may have arisen with the appointment.

The electronic data captured will be stored on secured and password-protected storage servers and mobile phones which ensure data privacy through only allowing authorised research staff access to the data. The electronic data collection system used for the study requires an SMS gateway to send and receive messages to the research participants. Consenting participants’ phone numbers, participant IDs and associated SMS messages will be stored on the SMS gateway’s secured and password-protected server.

Data quality will be ensured by providing fieldworkers with standard operating procedures (SOPs), training and ongoing support on the importance of data quality, data collection and data collection problem solving. Training will consist of a four-day session before the commencement of the pilot and a one-day training session focusing on problems identified during the pilot preceded the trial. The data manager will continuously monitor the captured data for missing variables and inconsistencies in order to resolve any data problems.

The data manager will export the data from the secured server, conceal the participants study arm allocation and de-identify the data before sharing the data in STATA and R compatible formats. The exported de-identified data will be stored in Dropbox, a secure cloud storage platform, for sharing with the lead trial statistician at the University of York for analysis.

All research data and documents referring to the ProLife trial will be stored and maintained in a secured storage space at SMU for a minimum of 15 years from the end of the ProLife trial. Study materials will be destroyed 15 years after the study.

This is a complex behavioural intervention and the team dynamics mean that all team members work very closely with one another. As such, LHWs and participants cannot be blinded to the intervention. The determination of the primary outcome will be done by the TB nurses who are blinded to the intervention status of the participants based on routinely collected data. Blinding of the field researchers collecting other questionnaire data (Additional file 2) will not be possible. The statistician will be blinded to the intervention or control arm allocation of participants during the analysis stage.

There is a potential of biased outcome reporting on self-reported tobacco smoking or alcohol consumption favouring the intervention arm. This will be minimised by training the fieldworkers in open communication and standardised data collection. Tobacco cessation will also be verified by exhaled CO monitoring. Given the use of the TB treatment record as a primary source of information pertaining to TB outcomes, there is a risk of non-differential misclassification. However, this risk is equal in both the treatment and control arms and therefore the decision has been taken that the TB treatment record can be used as a primary data source (i.e. bias would be towards the null).

The sample size has been set at 696 participants (348 participants per arm). This sample size is sufficient to detect a 10% difference in TB treatment success rates (0.86 vs 0.76) in the ProLife group versus the control group with 80% power, a significance level of 0.05 and 25% attrition. The sample size per clinic was in the range of 14–74 participants per clinic with a median of 24. The assumed success rates in the control group are based on actual success rates in TB patients in the studied provinces.

We will summarise baseline data descriptively by trial arm; however, we will not undertake statistical comparisons. For continuous measures, we will report means and standard deviations (SD); for skewed data, we will also provide medians and interquartile ranges. For categorical data, we will report counts and percentages. For the primary outcome, we will conduct analysis on an intention-to-treat basis. We will use binary logistic regression to compare the main outcome between the intervention and the usual care arm.

We will carry out similar analyses for the secondary outcomes with appropriate regression techniques: logistic regression for categorical outcomes and linear regression for continuous outcomes. We will also adjust for baseline characteristics and other covariates (HIV status, sex, alcohol versus tobacco versus both, district) if these differ between trial arms at baseline.

In case of missing data, we will employ a number of methods including multiple imputations to assess the sensitivity of the results. We will conduct subgroup analyses to determine whether TB treatment outcomes differ between subgroups, as follows: HIV-positive versus HIV-negative participants; participants with an alcohol problem only versus smokers only versus participants who are conjoint smokers and drinkers; and participants who were GeneXpert positive versus participants who were GeneXpert negative at baseline.

Each centre will be responsible for its own data entry and local trial management. Monitoring and site training will be carried out at each site within specified intervals. The project manager will visit each district every third month with the counselling supervisor. The district coordinators will visit each clinic bi-weekly; the site leads and data manager will visit the sites when required.

The monitoring will adhere to the principles of Good Clinical Practice and will follow an agreed monitoring plan. During their bi-weekly visits, the district coordinators will be guided by a checklist to confirm adherence to protocol, review eligibility verification and consent procedures, and provide additional training as needed. Any adverse events will be formally recorded and reported where appropriate.

The mobile data collection tool was improved based on findings from the pilot phase and feedback from the field staff capturing the data. Validation was added to the mobile questionnaires to ensure that all questions should be answered, participants only answer question relevant to them and that no unusual values (range checks) are entered. Coordinators can use the mobile data collection tool on their phones to monitor incomplete data activities, outstanding and upcoming appointments with participants and basic project performance indicators per clinic.

The data captured during the interviews with the participants are only captured on the electronic mobile data collection tool and can therefore not be verified by the site coordinators. However, the district coordinators will require access to all (enrolled TB) patient medical records including, but not limited to, laboratory test results and prescriptions. They will do spot checks by comparing the captured data with the medical records and look for missed outcomes reporting: verify completeness; consistency; and accuracy of data being entered on CRFs. The site leads (or delegated personnel) should work with the district coordinators to ensure that any problems detected are resolved.

The data will also be checked by the data manager for missing or unusual values (range checks) and checked for consistency within participants over time. If any such problems are identified, the individual centres will be contacted and asked to verify or correct the entry.

Indicators related to enrolment, premature withdrawal, motivational interviewing completion, CRF completion and SMS delivery will be analysed and included in monthly reports.

As the ProLife programme is a low-risk intervention, the decision was taken not to appoint an external data monitoring committee.

Eighteen LHWs, three district coordinators and one person who will be focusing on counselling supervision underwent MI training over five days, which covered: basic knowledge of TB; treatment adherence; tobacco and alcohol use; overview of the overall spirit of MI and communication style; core interviewing skills; evoking change talk, hope and confidence; developing a change plan; and strengthening commitment. The LHWs’ knowledge in terms of TB, alcohol use, tobacco smoking, TB and ART treatment adherence, and motivational interviewing was assessed before and after the training

To ensure that all the essential steps, techniques and skills of an MI session are covered during each session, the LHWs will be provided with a checklist to use during each session (Additional file 3). At the end of each session, they will also complete a post-session semi-structured form onto which they will indicate the extent to which they implemented each element of MI, as well as their general qualitative impressions of that particular session.

During the trial, the LHWs will receive regular supervisory support from the district and national study coordinators. In addition, an appointed supervisor with proficiency in MI will be providing monthly counselling supervision to the LHWs in each site, by travelling to each of the three sites and then providing support to the LHWs in a group, followed by individualised support to each LHW separately. All lay counsellors completed grade 12 and have at least one year of prior counselling experience in TB and/or HIV.

In order to assess the fidelity of the counsellors’ delivery of motivational interviewing during the trial, the validated Motivational Interviewing Treatment Integrity Coding Manual 4.2.1 (MITI) tool will be used [41]. We will tape record each LHW’s counselling sessions and then randomly select 5% of each counsellor’s patients and then use the recordings of all three sessions with those patients for the MITI task. Those selected recordings will be transcribed and translated into English. Two independent raters who are Setswana/Sesotho speakers will listen to the recordings and code a randomly selected 20-min portion of the written transcript. In the case of shorter counselling sessions, the entire recording will be assessed. The coding will entail making ‘Global Ratings’ (on four dimensions: Cultivating change talk; Softening sustain talk; Partnership; and Empathy) and ‘Behaviour Counts’ (with respect to the items: giving information; persuade; persuade with permission; question; simple reflection; complex reflection; affirm; seeking collaboration; emphasising autonomy; and confront). A score will be assigned to each of these items and the scores compared against the competency and proficiency thresholds that are specified in the MITI manual.

The economic evaluation will be a within-trial incremental cost-effectiveness analysis of the ProLife package over and above usual care conducted alongside the RCT. In each arm of the trial, we will estimate the costs of delivering the intervention to participants. Costs consist of the delivery of ProLife intervention (staff costs, overheads and consumables) in the intervention arm and the costs of behavioural support and support (staff time, overheads and consumables) in the control arm. Costs will be calculated based on the length of time of contacts and the unit cost of the healthcare worker delivering the intervention. These costs should demonstrate some within patient variability as cost will be partly dependent upon the length of the recorded appointments.

We will also record contacts with health services for TB by patients in each trial arm, using the questionnaires adapted from those used in a previous trial of TB [42]. Applying unit costs of healthcare to quantities recorded in the service use questionnaire will produce a healthcare cost profile for each patient at baseline, three-month and six-month follow-ups. Healthcare cost profiles based on TB contacts can then be used to investigate differences in total costs between the treatment and control groups. The base case analysis will use healthcare costs as the study perspective. This is justified by presenting the cost implications of delivering the ProLife intervention from a healthcare decision maker’s perspective. However, we will also record participants’ travel costs associated with making the journey to the intervention site. This will estimate the wider societal cost to patients associated with receiving treatment.

We will estimate the cost per additional successful TB outcome by computing an incremental cost-effectiveness ratio (ICER) combining treatment and wider healthcare costs with outcome data from the trial. The analysis will also use measures of health utility. We have registered on the EuroQol website to use the country specific version of EQ-5D-3 L for South Africa to measure health-related quality of life at baseline, three-month and six-month follow-ups. We will use EQ-5D-3 L to estimate changes in quality-adjusted life years (QALY) using the area under the curve by using the utility score at each time point to create a profile [41]. A cost utility analysis is then performed by combining incremental health outcomes using QALYs with the incremental cost. The cost utility analysis will estimate the cost per QALY that estimates the value for money afforded by administering the ProLife intervention over and above usual care. We will plot incremental cost against incremental outcome using a cost-effectiveness plane. We will also conduct sensitivity analyses to assess the robustness of the ICER and use bootstrapping techniques to calculate cost-effectiveness acceptability curves. Supporting analysis will also present the costs to patients incurred by making visits to both trial arms to present the wider cost burden. However, these costs will not be included in the base case scenario as the objective is to present the value for money from the perspective of the purchaser or commissioner.