Setting

Khayelitsha is a peri-urban township outside of Cape Town, South Africa with a population of approximately 450,000 people, most of whom reside in informal settlements[28]. Approximately 200 patients are diagnosed with RR-TB annually with a case notification rate of 55/100,000[21]; HIV co-infection rates are 70%[29]. The standard RR-TB treatment regimen provided to patients contained all or most of the following drugs: kanamycin, moxifloxacin, pyrazinamide, ethambutol, terizidone, ethionamide and high dose isoniazid. This regimen did not differ for patients in the SAT versus SOC cohorts.

Standard of care: Directly observed therapy

All patients treated for RR-TB in Khayelitsha, South Africa are to receive DOT as per the national treatment guidelines [27]. This treatment model requires patients to attend the clinic five days per week, where treatment is administered with the support of clinical medical staff. DOT is to be provided for the entire 24-months of treatment. All patients receiving DOT receive four standardized counseling sessions provided by trained RR-TB counselors upon diagnosis, treatment initiation, during the intensive phase of treatment and upon the completion of the intensive phase of treatment. Once monthly these patients were assessed clinically by a medical officer and laboratory parameters were tested in order to monitor treatment response. Attempts to trace patients LTFU are made telephonically or via home visits by counselors or local community care workers (CCWs) upon treatment interruption; patients not linked back to care are assigned an outcome of LTFU after two months of consecutive treatment interruption and unsuccessful tracing efforts.

For the sake of this analysis patients who were treated in clinics where DOT was the recommended model of care, were considered to be enrolled in the SOC-cohort.

Self-administered treatment

Following completion of the intensive phase of treatment, a RR-TB counselor conducted a tailored counseling session with the patient to congratulate and encourage the patient to continue treatment and discuss the option of receiving SAT. Patients were assessed for SAT based on: treatment adherence record (for RR-TB and concomitant diseases), clinical status and adverse events requiring ongoing monitoring. Local CCWs were assigned to potential SAT patients; prior to SAT enrollment the CCW conducted a home visit to assess the social situation, identify a treatment supporter, and determine adherence barriers. Weekly meetings in each clinic were attended by CCWs, doctors, RR-TB professional nurses and MSF counselors to discuss patients for SAT and those approved were enrolled in SAT for the remainder of treatment if they provided verbal consent. Patients were enrolled if they were no longer receiving an injectable agent, gave verbal consent and met the above stated assessment criteria. At initial implementation in each clinic SAT was offered immediately to eligible patients already in the continuation phase, regardless of injectable completion date.

Once patients were enrolled in the SAT pilot programme they received an adherence counseling session by a dedicated MSF counselor, where medications were reviewed, a pillbox was issued and adherence barriers were addressed. Enrolled patients received a weekly or monthly supply of RR-TB medications, depending on clinic and patient preference. CCWs visited SAT patients weekly initially, and monthly once patients were deemed to be doing well in the programme, for the duration of treatment to offer support and identify adherence challenges. These processes were unique to the patients enrolled in the SAT pilot programme; all other patients received DOT and were required to go to the clinic daily to collect and take treatment.

Patients found to be interrupting treatment (>2 weeks) were referred for review by a doctor and enhanced adherence counseling by the RR-TB counselor, and if necessary, were returned to clinic DOT. If LTFU, these patients are traced according to the SOC, thus there were no differences in the ascertainment of LTFU based on the RR-TB treatment model (SOC versus SAT). All SAT patients attended their clinic monthly for clinical assessment and routine monitoring tests which were also conducted monthly for those receiving DOT (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Outline of activities included in the SAT pilot programme.

https://doi.org/10.1371/journal.pone.0178054.g001

SAT implementation was progressive across 5/10 primary care clinics in Khayelitsha from 2012 to 2015; initial pilot clinics were chosen based on available resources, functionality, and willingness of staff to participate.

Cohort selection

Patients were considered for the SOC-cohort if they initiated treatment at least 6-months prior to SAT implementation in their specific clinic, in order to limit the number of patients in the SOC-cohort who received SAT. Treatment initiation times for inclusion in the SOC-cohort ranged from January 2010-July 2013. To avoid bias, patients were considered for the SAT-cohort if they started treatment in their clinic at least 6-months after SAT was implemented so that patients who were enrolled in SAT near the end of their treatment course were not included in the cohort. Treatment initiation times for inclusion in the SAT-cohort ranged from January 2012-December 2014.Patients were excluded from the SOC and SAT-cohorts if they had an outcome within 6-months of RR-TB treatment initiation (had an outcome before the end of the intensive phase of treatment). Patients in the SOC and SAT-cohorts were included in the analysis of final outcomes if they had a final treatment outcome before January 1, 2017.

Definitions

Pulmonary tuberculosis (PTB) and extra pulmonary tuberculosis (EPTB) were defined as Mycobacterium tuberculosis (M. tuberculosis) which affects the lungs and other sites, respectively; patients with both PTB and EPTB were defined as M. tuberculosis which affects both the lungs and other sites[30]. RR-TB treatment outcomes including treatment success (treatment cure or completion), LTFU and death were defined in line with WHO recommended definitions; LTFU was defined as a patient whose treatment was interrupted for two consecutive months or more[4],[30]. Treatment failure was defined according to South African National Drug Resistant TB guidelines as failure to culture convert after 6–8 months of RR-TB treatment or 6–8 months of consecutive positive cultures[27]. The intensive phase was defined as the first 6-months of RR-TB treatment. Standard and modified regimens were provided as per National Department of Health SOC for RR-TB treatment[27]. Baseline CD4 count was defined as the CD4 count taken within 2 months of RR-TB diagnosis.

Outcome of interest

The outcome of interest in this study was a final RR-TB treatment outcome of LTFU; we aimed to determine if there were differences in LTFU among those treated in the SOC and SAT-cohorts.

Data collection & analysis

Routine programmatic RR-TB data collected for the decentralised RR-TB programme (retrospective) were linked with data collected on paper registers for this pilot (prospective). Data collected for the decentralised programme included date of RR-TB diagnosis, date of RR-TB treatment initiation, sex, age at RR-TB diagnosis, TB treatment history, RR-TB disease classification, RR-TB regimen, RR-TB disease site, HIV status, CD4 count at RR-TB diagnosis, anti-retroviral (ART) initiation date, final RR-TB outcome, and date of RR-TB treatment outcome. These data were collected from the national electronic RR-TB register, paper registers, and clinical files. Data collected specifically for the SAT programme included date of MDT presentation, outcome of MDT presentation (placed out without a CCW, placed out with a CCW, not placed out), and reversion to clinic DOT. These data were collected by MSF counselors involved in the identification of patients for presentation to the MDT committee for SAT enrollment.

STATA/IC version 14.1 was used for this analysis; differences in clinical and demographic characteristics and final treatment outcomes assigned before January 1, 2017, stratified by patients initiating RR-TB treatment in the SOC and SAT- cohorts were described. A sub-analysis investigated final outcomes among patients who received SAT only. Chi-squared and Fisher’s exact tests were used for statistical comparison of categorical variables, while Wilcoxon rank-sum tests were used for continuous variables; P-values <0.05 were considered statistically significant. Kaplan Meier (KM) Curves were used to investigate differences in time to 24-month LTFU (censor: death, treatment failure, transfer out or treatment completion) among patients who completed at least 6-months of treatment in the SOC and SAT-cohorts and among those HIV-infected and uninfected.

Ethics

The analysis of this programme was covered by pre-existing ethical approval for the ‘Evaluation of a decentralised programme for rifampicin-resistant tuberculosis care and treatment in Khayelitsha’ (HREC 540–2010) granted by the University of Cape Town Human Research Ethics Committee. Informed consent was not required by the ethics committee for this study as it involved a retrospective analysis of routinely collected, programmatic data. This study was conducted in line with the STROBE checklist for reporting on observational cohorts (http://www.strobe-statement.org/).

Limitations

This study had several limitations, however, the study might have been too small to see an effect of these limitations. As this was a non-randomized comparison, there are threats to the validity of the results presented. As the criteria to receive SAT were subjective and patients who were doing well on or adherent to RR-TB treatment were selected for the intervention, there might have been an overestimation of the effect of SAT. This was potentially minimized however, by conducting an intent-to-treat analysis. Due to the slow and phased implementation and initial reluctance by care providers to endorse the pilot, some eligible patients in pilot clinics were never offered SAT; limited resources to prepare and support large numbers of patients initiating SAT at once made this impossible. The eligibility criteria for enrollment in SAT were very subjective, based on adherence and clinical response to treatment, resulting in selection bias for enrollment into SAT. Time on SAT was a potential confounder in this analysis as some patients only enrolled onto SAT much later in treatment. However, median time to SAT enrollment was 7.8 months, indicating that patients completing the intensive phase were prioritized over those nearing continuation phase treatment completion. Thus the small number of patients who received SAT later in treatment likely had little effect on the results. Additionally, patients in the SOC-cohort might have received an informal version of SAT as facilities occasionally provided a supply of medications for self-administration to relieve pressure on the clinic, despite clinic DOT being the SOC. This contamination may have resulted in an underestimation of the true effect of SAT. These patients however, did not receive the specialized counseling and ongoing community support integral to the SAT pilot programme. Additionally, due to phased implementation of the pilot some patients in the SOC-cohort actually were offered SAT through the programme. This study was based on operational data and has limited statistical power but provides sufficient evidence to encourage future research of this intervention in larger studies. Despite these limitations, this was a study of routine, programmatic data which provided an overview of final RR-TB treatment outcomes among patients who received SAT in pilot clinics in a programmatic setting. Inclusion of all patients still on treatment at the end of the intensive phase in both cohorts limited any overestimation of the impact of SAT. Further research, including qualitative studies, is planned to determine patients’ and care providers’ perspectives regarding SAT, the impact of SAT implementation on healthcare providers and facilities and the utility of SAT among less adherent patients.