People–centred care versus clinic–based DOT for continuation phase TB treatment in Armenia: a cluster randomized trial

The Innovative Approach to Tuberculosis Care in Armenia (IATCA) was an open–label stratified cluster randomized controlled non–inferiority trial with two parallel equal arms (intervention and control). It targeted the Continuation Phase of TB Treatment (CPT) and was implemented through Armenia’s TB outpatient centres (TBOC). We conducted the trial in accordance with good clinical practice guidelines and the Declaration of Helsinki. The American University of Armenia Institutional Review Board (IRB) reviewed and approved the study protocol. This study is registered with ClinicalTrials.​gov (NCT: 02082340) and its methods are described in greater detail elsewhere [21].

The study included all TBOCs in Armenia that had more than five drug susceptible TB patients in 2012 (52 TBOCs out of the 60 total TBOCs). We stratified the centres by patient load (low [up to 10 patient per year], average [11 to 25 patients per year], high [more than 25 patients per year]) and treatment success [22] (up to 64%; from 65 to74%; above 74%) yielding nine distinctive categories. Computer assisted block randomization resulted in 26 centres in each arm with intracluster correlation equal to 0.04 and an estimated design effect of 1.6 [21]. Stratifying TBOCs by patient load also improved the balance between the number of physicians and nurses in the clinics across the study arms, which varied between 2 health care workers in TBOCs with a small patient load to 7 health care workers in TBOCs with a high patient load. Interviewers assessing non–clinical outcomes were not blinded at baseline but were at follow-up. Independent of the research team, each TBOC physician assessed and reported treatment outcome to the National Tuberculosis Control Centre (NTCC).

Study participants were drug susceptible pulmonary TB patients starting their CPT TB treatment in a selected TBOC between March and December 2014, at least 18 years old at enrolment, and able to communicate in Armenian. A patient’s residence determined assignment to a specific TBOC. The corresponding TBOC physician and NTCC psychologist verified the eligibility of each patient.

We conservatively calculated the sample size needed to conduct a superiority trial. We assumed a 10% difference in the primary outcome, [23] an alpha of 0.05 and power of 80%. Considering the potential for losses to follow-up, we used a design effect of two (rather than 1.6); thus, requiring 190 for each arm.

Where available, intervention arm patients identified family supporters who participated in a counselling session and supported them during their treatment period. Trained interventionists (a psychologist and a nurse) conducted a single educational and psychological counselling and empowerment session for the TB patient and, as available, his/her supportive family member, and other interested family members. Session topics included: TB symptoms, route of transmission, treatment strategies, the importance of treatment adherence, prevention and infection control, addressing TB–related stigma and common myths, handling side effects, and a detailed explanation of the treatment protocol. The session lasted 120 min on average, exceeding the initially estimated duration by 30 min. The longer duration of the session was due to participants’ significant interest in the topics covered as well as an extended question and answer period at the end of the session. After the session, patients began self–administered drug–intake supervised by a trained family member. Once per week, patients visited the TBOC healthcare provider to receive the next week’s medication in a pill–minder box. The research team sent daily morning SMS messages to intervention patients. The team placed daily phone calls to supporting family members to inquire about patients’ adherence to the prescribed treatment as well as any potential side effects. Patients lacking a supporting family member were responsible for all activities defined for the supportive family member.

Control arm patients received DOT during their continuation phase treatment, in accord with Armenia’s national strategy (comparable to implementations of WHO TB treatment guidelines worldwide).

The primary outcome was treatment success at the end of the continuation phase. Following WHO conventions, we considered “cured” or “treatment completed” a “treatment success.” [22] WHO Definitions and reporting framework for tuberculosis [22] are presented in the supplemental material. In both intervention and control arms, the TBOC treating physician clinically assessed and reported the outcome independent of the research team.

We utilized face–to–face surveys administered by trained research staff to collect non–clinical (secondary) outcome data: self–reported (patient–reported) adherence to TB treatment, knowledge, depressive symptoms, stigma, quality of life, and social support [24‐27]. We measured all non–clinical outcomes (except self–reported adherence) at baseline and follow–up. We measured treatment adherence by self–reported (patient–reported) and family member–reported frequency of receiving and taking TB medications. We adapted the knowledge variable from previous work [24, 25], with a score range of 0 to 31 (see the Additional file 1). We assessed depression symptoms using a validated modified Armenian version of the Centre for Epidemiological Studies Depression (CES–D) scale which had a score range of 0 to 60, with higher scores indicating higher levels of depression symptoms [26]. We modified the Van Rie stigma scale to measure stigma [28]. The original scale measures stigma of tuberculosis from the community perspectives. We modified the questions to measure the behaviour of family members toward the tuberculosis patient from patients’ perspective. Its score ranged from 0 to 27 (see the Additional file 1). We measured quality of life using the EQ-5D [29]. The responses to the quality of life scales yield an index utility score anchored at 0 for death and 1 for perfect health. The scores were multiplied by 100 to increase precision of the reported estimates. We used the Berlin social support scale to measure social support of family members toward TB patients. Scores ranged from 0 to 45 [30].

We used SAS 9.4 statistical package for our analyses. We included only complete cases (those without a missing value for the outcome under analysis). We assessed the distribution of socio–demographic characteristics and other potential confounding variables across the two arms using frequency and proportion for categorical variables, and mean and standard deviation for continuous variables. We analysed the clinical outcome using a Generalized Estimating Equations (GEE) binomial regression with identity link, considering TB clinics as the clusters following the intention to treat and per protocol principles [31]. We used an exchangeable correlation structure to account for the correlation between treatment outcomes within the clusters [32].

Following per–protocol analysis principle, linear mixed effect models including random effect for TB clinics and subjects (random intercepts) tested the non–clinical outcomes. For each non–clinical outcome, we adjusted for baseline values and the change over time. The final results present the change in the outcome over the study period (from baseline to follow-up) in each arm, adjusted for the baseline measures [33].

We determined absolute differences for outcomes based on efficacy established in other reported TB non–inferiority or equivalence trials [34] and our previous research experience [24, 26, 27]. We obtained the non–inferiority margin of − 3% (i.e. − 0.03) on the absolute difference scale for the clinical outcome (treatment success). We declared non–inferiority if the lower limit of the 95% CI was greater than the corresponding non–inferiority margin for the positive outcomes (i.e., treatment success, knowledge score, social support, and quality of life scores). For the negative outcomes (i.e., stigma score, depression score), we declared non–inferiority if the upper limit of the 95% CI was smaller than the corresponding non–inferiority margin. Similarly, we defined non–inferiority margins for knowledge and social support to be 1.5, and for stigma to be − 1.5. We applied non–inferiority margins of 2.0 for depression scores, and 15 for quality of life scores.

We conducted sensitivity analyses using multiple imputation and last observation carried forward to assess the potential impact of missing data in the non-clinical outcomes (see Additional file 1 for details). We reported our study using the Consolidated Standards of Reporting Trials (CONSORT) guidelines.