Effect of a phone reminder system on patient-centered tuberculosis treatment adherence among adults in Northwest Ethiopia: a randomised controlled trial

Trial design

We implemented a two-arm randomised controlled trial (RCT) to evaluate the effect of weekly pill refilling and daily medication reminder systems on patient’s adherence to patient-centred medication, provider–patient relationships and treatment outcomes.

Participants and setting

The study population includes all adult, drug-sensitive and new patients with TB on the continuation phase at health centres and hospitals in Central Gondar Zone and Gondar town administration. These included eight districts from Central Gondar Zone and Gondar town administration. The districts include Gondar Zuria, Tach Armachio, Wegera, East Dembia, West Belesa, Takusa, East Belesa, Alefa and Gondar town administration. The study area consists of 74 functional public health facilities (health centres and hospitals) serving more than 2.9 million people overall in the area. Eligibility criteria include all new drug-sensitive patients enrolled in continuation phase, aged 18 years and above, who own mobile phone, willing and able to provide informed consent. The study excluded patients with no/limited mobile networks at their home, enrolled in multidrug-resistant TB, and already enrolled in another study that could affect the study’s outcomes. To reduce attrition bias, we excluded patients who planned and decided to transfer to another health facility.

Interventions

We developed a web-based reminder system tailoring the local context, piloted the prototype and published it before implementation.33 The basic features of the system include patient registration, scheduling reminders, visualisation and generating reports. This trial was implemented using the web-based reminder system. Development of the system followed a waterfall technique34 with stakeholders’ involvement, including TB focal care providers, information technology and health informatics experts.

Participants in the intervention arm, in addition to routine care, received weekly pill refilling and daily medication reminders using graphics-based and text messages in the local language (Amharic). The graphic reminder was intended to engage illiterate patients who cannot read text messages. In figure 1, ‘A’ represented pills for daily medication followed by Amharic text to mean ‘Please take your pills’; while ‘B’ represented a strip of pills to refill followed by Amharic text to mean ‘your clinic appointment is tomorrow, please come on time’.

Request permission

Before initiating the intervention, each participant on the intervention arm received a face-to-face orientation on the use of graphic and text messages for 5 min during enrolment. Immediately after enrolment, a welcoming message was sent to the patients in the intervention group, which aimed to improve their engagement in the treatment through motivation. Medication reminders were sent every day from 07:30 to 08:00 to remind the conventional time (08:30) for taking daily TB pills; whereas pill-refilling reminder was sent a day before the due date (appointment date) from 18:30 to 19:00. The intervention was provided for 4 months of the continuation phase.

Routine care

The control arm received only routine care. Routine care in continuation phase TB treatment means patients take their daily medication at home with the help of TB treatment supporters (TTS). TTS is either a health extension worker (community health worker), family member, neighbour, workmate or community figure trained to observe, educate and remind patients for the optimal TB treatment administration at home.35

Outcomes

The primary outcome of the study was adherence to patient-centred TB medication during the continuation phase. During the continuation phase, we measured adherence to TB treatment using the shorter form, 11-item Adherence to Refill and Medication Scale (ARMS).36 37 The items include the following:

1. How often do you forget to take your TB medicine?

2. How often do you decide not to take your TB medicine?

3. How often do you forget to get prescriptions filled?

4. How often do you run out of TB medicine?

5. How often do you skip a dose of your TB medicine before you go to the doctor?

6. How often do you miss taking your TB medicine when you feel better?

7. How often do you miss taking your TB medicine when you feel sick?

8. How often do you miss taking your TB medicine when you are careless?

9. How often do you change your TB medicines’ dose to suit your needs (like when you take more or fewer pills than you’re supposed to)?

10. How often do you put off refilling your TB medicines because they cost too much money?

11. How often do you plan ahead and refill your medicines before they run out?

The original ARMS tool consists of 12 items, including two subscales, an 8-item medication-taking subscale and a 4-item refill subscale.36 One of the items, ‘How often do you forget to take your medicine when you are supposed to take it more than once a day?’ was not relevant to TB medication since TB pills often are taken once a day. Each of the items was structured as a Likert scale with responses of ‘none’, ‘some’, ‘most’, or ‘all’ of the time, which were given values from 1 to 4.

As a secondary outcome, treatment success rate (TSR) was assessed. According to the WHO TB treatment guideline, treatment outcomes are measured as TSR, which is defined as the sum of patients with TB cured and/or completed their treatment. Cured is a patient with pulmonary TB with bacteriologically confirmed TB at the beginning of treatment which was smear or culture negative in the last month of treatment and on at least one previous occasion. Treatment completed is a patient with TB who completed treatment without evidence of failure but with no record to show that sputum smear or culture results in the last month of treatment and on at least one previous occasion were negative, either because tests were not done or because results are unavailable. Defaulted is a patient whose treatment was interrupted for 2 consecutive months or more. Treatment failure is a patient with TB whose sputum smear or culture is positive at month 5 or later during treatment. While death is a patient with TB who dies for any reason before starting or during the course of treatment.38

It was also planned to measure patient attendance to TB clinics39 ; however, it was not analysed because the weekly appointment scheme was not consistently practised by TB clinics. From our observation, some TB clinics appoint patients every 2 weeks, some every month, and some others allow family/treatment supporters to refill on the patient’s behalf. One of the reasons for the inconsistency was during the rainy season, patients were not willing to come every week. Patients used to delegate their treatment supporter during social events (like funerals) and whenever feeling sick.

As an additional analysis, we assessed the provider–patient relationship using a 7-item provider–patient relationship questionnaire adapted from previous studies.11 40 The items include patients’ satisfaction with the care provided, patient’s trust in care provider, frequency of phone call to the patient, the convenience of appointments, frequency of calls by the patient to care provider, patient feeling shame to question, and patients’ perceived compassionate and caring provider. It aimed to evaluate the effect of the intervention on smoothing the relationship between patients and care providers, as the provider–patient relationship was a factor affecting adherence to TB treatment.10 11 40

Sample size and sampling procedure

We calculated the sample size using STATA V.14 considering the assumption of superiority design. The sample size in the published protocol was 262.32 It was calculated using the proportion of adherence (p) from other similar studies and applying power (β=80%).9 The sample size was recomputed using our baseline proportion of patient adherence to TB treatment during the continuation phase (p1=64.2%) and p2=79.2% considering 15% minimum detectable effect size. The total sample size was 324 (162 in each study arm), considering α=5%, power (β=90%) and 10% attrition rate. The study tried to involve urban and rural settings to ensure representativeness in both hard-to-reach and urban settings. Eligible patients were recruited from 22 health facilities (15 health centres and 7 hospitals) using an open cohort from 15 July to 8 November 2019.

Randomisation, allocation and blinding

The participants’ eligibility was assessed and they were recruited using enrolment guides during the intensive phase aimed to initiate the intervention immediately at the start of the continuation phase. The study used a covariate adaptive randomisation (minimisation technique) to control and balance covariates’ influence on the study outcome.41 Simple randomisation (lottery method) was applied at the beginning for allocating the first subjects at each arm and in the middle when the subsequent patient’s characteristics appear not important to minimise imbalances across the covariate groups. Otherwise, every subsequent patient was assigned to the groups with an imbalance across the covariate groups using a minimisation algorithm. The covariates, including participants’ educational status (able and unable to read and write), sex (male, female), residence (urban, rural) and type of facility (health centre and hospital), were considered during randomisation. The research assistant provided an identification number to identify participants and carried out the randomisation process using a covariate algorithm. To reduce information contamination, we planned to include one, if more than one patient attending TB treatment from a household; however, the situation did not happen. The outcome assessors (care providers in TB clinic and endline data collectors) did not participate in patient enrolment, and endline data were collected using participant codes to mask each participant’s randomisation status purposely to reduce detection bias. The study participants have not masked due to the intervention’s nature that a study subject requires overt participation.

Data collection

Baseline data were captured using a participant enrolment guide from eligible patients during recruitment. Endline data were collected using phone-call interviews and patient registers in TB clinics at the end of 4-month follow-up (after completing the treatment course). In the protocol, we planned to conduct a face-to-face interview; however, data were collected using a phone-call technique due to the COVID-19-related movement restrictions. Training was provided for data collectors to reduce measurement bias related to the phone-call interview.

TB treatment outcomes were collected using checklists from registers in TB clinics. Endline data were collected from 20 January to 24 April 2020.

Statistical analysis

We used EpiData V.3.1 software for data entry and exported it to STATA V.14 for analysis. The family wealth index was calculated using principal component analysis technique.42 The study applied per-protocol analysis to evaluate patients’ adherence to patient-centred TB treatment using ARMS with 11 items dichotomised as 11 or >11. The provider–patient relationship was ascertained and dichotomised using a median score. Intention to treat was applied to evaluate treatment success. Absolute difference (AD) was computed to evaluate changes of outcome variables among intervention and control groups. Log-binomial regression was used to estimate the adjusted effect size controlling other variables. As a superiority design trial, we computed a one-sided hypothesis testing. Superiority tests were often computed as one sided in which the alternative and null hypotheses are reported as one of the study arms is better by more than a stated margin. Hence, we reported the lower bound of 95% CI with a one-sided p value.43

Recruitment and patient flow

Three hundred forty-six patients were assessed for their eligibility to be included in the study from 23 July to 8 November 2019. Of them, we communicated with 318 potentially eligible patients, and 12 declined to participate in the study. Hence, we randomly allocated 306 eligible patients: 152 in intervention group and 154 in control group. Adherence to TB medication and provider–patient relationships were assessed using 282 participants (139 intervention group and 143 control group). The rest 3 (1 in intervention group and 2 in control group) died, 4 patients from the intervention group did not receive messages at all, 17 (8 in intervention group and 9 in control group) did not respond to the phone call-based endline assessment. We captured TB treatment outcomes for all 306 participants from facility records (figure 2).

Request permission

Baseline data

The study included 306 (152 intervention and 154 control) participants for analysis. The majority of participants, 83 (54.6%) in the intervention group and 86 (55.8%) in the control group, were aged between 18 and 29 years. Of the total participants, 84 (55.3%) in the intervention group and 75 (48.7%) in the control group were men. According to the type of facility, 116 (76.3%) in the intervention group and 105 (68.2%) in the control group were enrolled in hospitals. Participants’ residence indicated that 115 (75.7%) in the intervention group and 117 (76%) in the control group were urban dwellers (table 1).

Estimated effect size

Multivariable analysis

We run the adjusted log-binomial regression analysis to calculate the adjusted relative risk (RR) holding other covariates constant. We analysed the data using both two-tailed and one-tailed tests. A two-tailed test was conducted for baseline characteristics. A one-sided test was analysed for the intervention and control groups, where a superiority hypothesis was formulated.

The finding has shown that the intervention had a significant association with patient adherence to patient-centred TB treatment with one-tailed RR (95% upper CI) (RR=1.632 (2.294); p 0.018). All other covariates have no significant association with adherence to patient-centred TB treatment (table 3).

Implication to research and practice

This study was the first of its kind in Ethiopia that could initiate similar studies to improve TB care and treatment. More RCTs are needed considering other messaging schemes like voice and video messages and a combination of different types. Studies are also needed to address the different and complex contexts in Ethiopia. It also required exploring acceptability, feasibility and implementation challenges of the system for TB treatment support. Our study tried to address the mHealth system from the patient side for adherence support. It also needed to evaluate the mHealth system’s effect from the care provider’s perspective whether it improves healthcare delivery.

A cost-effectiveness analysis is required for the intervention considering the complex contexts and the need to understand the underlying barriers of implementation before scaling up to routine practice. Despite this, mHealth technologies were generally suggested as cost-effective solutions to resolve communication barriers in the developing world.59 The previous study suggested that a graphics-based medication reminder system was more beneficial, cost-effective and accepted for use by most patients, including those with limited education.46

Limitations

This study applied a patient self-reported adherence to TB treatment that could underestimate the problem due to recall and social desirability biases.60 We tried to validate the instrument prior to actual data collection to overcome the social desirability biases. The composite items of the tool could trigger memory and responses that could reduce both recall and social desirability biases. This study included only patients who own a mobile phone that could affect the study findings’ generalisability to the study area’s general population. Studies indicated that phone sharing was a common practice in the study area.61 This study, however, did not apply this approach as phone sharing could not allow absolute access and use of the phone by the patients. The small sample size and considerably higher lost to follow-up of study participants could reduce the power of the test. This study did not evaluate the effect of the intervention on the sputum conversion rate on the 5th month, perceiving that needs the reminder system to involve care providers than patients.