Initiation, scale-up and outcomes of the Cambodian National MDR-TB programme 2006–2016: hospital and community-based treatment through an NGO–NTP partnership

Setting

Programme origins

Prior to 2006, there was no mechanism for public sector MDR-TB testing or treatment in Cambodia (figure 1). The CHC applied to the WHO Green Light Committee (GLC) in 2006 on behalf of patients enrolled in the CAMbodian Early vs. Late Initiation of Antiretrovirals (CAMELIA) trial11 testing optimal timing of HIV treatment in immunosuppressed patients with TB, including 13 found to have MDR-TB through TB culture and drug resistance testing performed per trial protocol. CHC received GLC approval in 2007 for a second application with the National Tuberculosis Program (NTP) to extend MDR treatment to new patients outside the CAMELIA study.12 CHC then established MDR treatment sites throughout the country in partnership with the NTP, initially establishing three hospitals as treatment centres and constructing and renovating 29 isolation rooms at these sites. As of 2017 there were 65 isolation rooms in 11 referral hospitals throughout Cambodia. A third successful CHC application to the GLC in 2010 further expanded and began national scale-up of DR-TB treatment. In 2011, the NTP assumed operational leadership of the MDR-TB programme to create the national MDR-TB treatment programme in partnership with CHC, with the support of the WHO. The NTP is now responsible for overall programmatic management, including screening suspected MDR cases for drug resistance and baseline contact screening. CHC continues to provide essential programme services including patient follow-up, treatment support and education, clinical follow-up of patients, side effect management, home infection control and reporting, as well as technical assistance to the NTP.

Request permission

Figure 1

Timeline of development of key elements of the Cambodian MDR-TB treatment programme. *GLC1 application for patients enrolled in the CAMbodian Early vs. Late Introduction of Antiretrovirals (CAMELIA) clinical trial only. CHC, Cambodian Health Committee; cx, culture; GLC, Green Light Committee of the WHO for MDR-TB Treatment; GXP, GeneXpert MTB/RIF; INH, isoniazid; IPC, Institut Pasteur du Cambodge; LJ, Lowenstein-Jensen solid culture method; MDR-TB, multidrug-resistant tuberculosis; MSF, Medecins Sans Frontieres Belgium and France programmes; NTP, National Tuberculosis Program; RIF, rifampicin; sm+, sputum smear positive.

Patient screening and identification

Prior to 2011, MDR-TB cases were detected through drug-susceptibility testing (DST) of smear-positive retreatment TB cases. Patients were notified to the CHC through a combination of CHC staff examining provincial clinic TB registers as well as NTP staff contacting CHC when such patients were identified. Screening criteria were expanded to include: (1) all pulmonary retreatment cases; (2) patients with TB who are persistently smear positive after months 2–3 of TB treatment; (3) close contacts of MDR-TB cases with cough >2 weeks; and (4) HIV-infected smear-positive patients with TB.13 HIV counselling and testing was performed at the time of TB diagnosis. In 2012, the screening algorithm was revised to recommend screening any HIV-infected patient with signs/symptoms of TB (cough, weight loss, fever, night sweats) using GeneXpert MTB/RIF (GXP).

Medecins Sans Frontieres (MSF) Belgium and MSF France programmes provided MDR-TB treatment in two provinces between 2007 and 2009. Patients who were still on treatment (n=28) when MSF withdrew from this sector were transferred to the CHC programme and are included in this analysis.

Laboratory diagnosis and follow-up

Initially, in 2006, Lowenstein-Jensen culture and DST by the proportional method were performed at the Institut Pasteur du Cambodge (IPC), in a laboratory externally validated by the College of American Pathologists and the Japanese Supranational Reference Laboratory. In 2007, solid TB culture became available at the NTP laboratory, and in 2008 liquid culture was introduced. DST was thereafter mainly conducted in the national laboratory beginning in 2011. Second-line drug susceptibility testing has been performed at IPC using GenoType MTBDRsl since 2012. The CHC obtained the first GXP platform in Asia in 2010 for detection of Mycobacterium tuberculosis and rifampin (RIF) resistance through a research protocol.14 This GXP platform was then transferred to the NTP for use for public sector MDR testing. The NTP subsequently acquired additional GXP platforms, and in 2011, an MDR screening algorithm was adopted to use GXP as the initial screening test for suspected MDR.15

Diagnostic criteria

MDR was defined as infection with M. tuberculosis resistant to isoniazid (INH) and RIF as detected by DST. Patients were empirically started on MDR treatment on the basis of strong clinical suspicion or, starting in 2011, GXP evidence of RIF resistance. If DST detected first-line drug resistance not meeting criteria for MDR, the treatment regimen and duration was modified accordingly. TB with GXP-detected RIF resistance for which DST or genetic resistance testing was not available was treated presumptively as MDR.

Treatment protocol and programme components

The MDR-TB treatment regimen in Cambodia consisted of a standardised regimen (box 1). If a patient had culture-positive MDR-TB or no clinical improvement at 4 months, second-line DST was performed and the regimen was modified to include only susceptible drugs. In 2012, the injection phase of treatment was extended to 8 months for all patients based on WHO recommendations.16

As part of the initiation of the countrywide MDR programme, MDR Management Committees (MMC) were established at each provincial hospital designated as an MDR referral centre in order to safely expand treatment initiation capacity. MMCs included a government physician, a CHC physician, a nurse with expertise in MDR and/or a radiologist at the site. The MMC reviewed laboratory, clinical and imaging data for each patient prior to initiating treatment and agreed on the appropriate treatment profile and dosing for the patient.

Community versus hospital-based care

Patients without hospital-level clinical requirements for managing comorbidities, available home-based support and preference for outpatient care could elect to start treatment as outpatients without hospital admission. The community-based treatment group was defined as patients who initiated treatment in the community or who were discharged from the hospital less than 30 days after treatment initiation (community arm). Patients who were hospitalised for 30 days or more, or who died before discharge were considered to be in the hospital-based treatment group (hospitalised arm). The decision to discharge a patient home was made in consultation with the patient, the inpatient clinicians and the community care team.

At the time of initiation of outpatient treatment, a treatment contract was signed by the patient, a treatment supporter (usually a nurse) from the health centre closest to the patient, a second treatment supporter designated by the patient and CHC staff. The treatment supporters supervised daily medication administration, provided motivation, maintained awareness of side effects and contacted the health centre or CHC staff in the event of clinical deterioration. Local health centre staff administered daily injectable medications in the intensive phase of MDR-TB therapy, monitored adherence during this phase and reported any clinical problems to the CHC outpatient monitoring team by phone.

For patients returning home prior to sputum culture conversion, education and supplies were provided for home-based infection control, including masks for patients and family members, recommendations for maintaining adequate ventilation and isolating small children from the patient. Assistance with creating an adjoining shelter for the patient was also provided if there were inadequate existing ventilated space for the patient. Symptom screening of household contacts was conducted at monthly visits. Contacts suspected to have TB had sputum collected by the visiting staff on the spot or were directed to the local health centre for screening. Direct and financial support was provided for nutrition (including baby formula for patients who could not breast feed due to treatment toxicity) and transportation for laboratory tests and hospital visits.

Patients with MDR-TB with HIV infection were referred to local centres providing antiretroviral therapy (ART) for clinical follow-up. MDR-TB clinicians worked closely with ART providers to monitor combined treatments, side effects and dose adjustments.

Ongoing education and training in MDR diagnosis, treatment and clinical follow-up including side effect management was provided by CHC to clinicians at the regional treatment sites and to local health staff. CHC designed and implemented nationwide training for the NTP in 2006 on MDR-TB diagnosis and treatment and continues to provide ongoing clinical mentoring and care of patients countrywide.

Patient follow-up and monitoring

The CHC mobile outpatient monitoring team of doctors and/or nurses and local government health staff made visits to patient homes monthly to resupply medication, screen for side effects, monitor clinical progress and collect sputum for treatment monitoring. Visits were more frequent if clinically indicated for adverse events or worsening of clinical symptoms. A CHC physician made periodic home visits to patients as necessary. CHC field nurses set up community care by linking the patient to the local health centre and village health support group in their village and trained the treatment supporter. Any clinical or adherence issues were reported to the CHC team for problem solving. Patients visited the health centre quarterly for physical examination. Any treatment or adherence issues that could not be readily solved were reported to the CHC monitoring team by local government health centre staff.

Treatment outcomes were defined and reported according to WHO definitions for MDR-TB.17 Outcomes for the cohort are reported for patients initiating therapy by June 2015 who therefore had the opportunity to complete 24 months of follow-up by the time of data censoring in June 2017. For analyses of associations with treatment outcome, we defined a combined outcome of ‘treatment success’, defined as either ‘cure’ or ‘completion’. Outcomes of ‘died’, ‘lost to follow-up’ or ‘failed’ were defined as ‘no treatment success’. Patients who transferred out were excluded from this categorisation scheme as their outcomes were unknown.

Statistical analysis

Demographic, clinical, laboratory and microbiological data were routinely abstracted from patient records into a clinical database maintained by the CHC. An anonymised subset of this database was entered into a Microsoft Excel spreadsheet with range and error-checking formulas. Only the first episode of treatment for MDR-TB was included in this analysis for each person. Statistical analyses were conducted using Stata (College Station, TX). Descriptive frequencies were calculated and compared using Χ² tests. Medians of continuous variables were compared using the Mann-Whitney test. Predictors of treatment success were assessed using bivariate and multivariate logistic regression. Factors of a priori interest were included in the multivariable model.

Ethics

As this study was based on existing programme records collected as part of routine clinical care, and no patient identifying information was collected, ethical clearance was not required according to local regulations.

Patient demographics

Baseline characteristics were similar between the community and hospitalised treatment groups (table 1). The median duration of hospitalisation for community patients was 6 days (IQR 0–12 days) and for hospitalised patients was 84 days (IQR 45–180 days). There were more than twice as many men as women (63% male vs 37% female) in the cohort overall, with women making up a greater proportion of the community patients (41% vs 33%, p=0.046). The age distribution was similar between community and hospitalised groups; the median age at the time of starting MDR treatment was 44 (IQR 33–55), including three children younger than 15 years and 48 adults older than 65 years. HIV infection was similar in the two groups and present in 117 (20%) of patients initiating MDR treatment, with 110 (94%) taking ART at or within 4 weeks of MDR treatment initiation. Patients were severely underweight with a median body mass index (BMI) of 17.2 (IQR 15.2–19). BMI was lower in the hospitalised group (16.9 vs 17.4, p=0.04). The hospitalised group had greater overall drug resistance compared with the community group (57% with additional resistance beyond INH and RIF vs 30%, p<0.001). There were four cases of extensively drug-resistant (XDR) TB identified and confirmed by second-line drug testing and one case of pre-XDR-TB. A history of prior TB treatment was present in 306 (53%) of patients initiated on MDR-TB treatment; 276 (47%) were being treated for their first episode of TB, and this did not differ between community and hospitalised groups.

Treatment outcomes

Of 486 patients initiated on MDR-TB treatment, 374 (77%) had treatment success, with 28 (6%) lost to follow-up, 3 (0.6%) failed and 77 (16%) died (table 2). The proportion of each outcome did not differ significantly between patients in the community and hospitalised groups (p=0.46). One patient was lost to follow-up, was subsequently re-enrolled after 2 years and treatment was ongoing at the time of data censoring. Among the 117 HIV-coinfected patients, 65 (65%) had treatment success, 32 (27%) died, 1 (1%) failed and 8 (7%) lost to follow-up, and 11 (9%) continued on treatment at the time of data censoring.

Non-MDR outcomes

Patients with mono or poly-DR-TB not meeting criteria for MDR were treated in the DR-TB programme using second-line drugs and were also eligible to participate in community or hospitalised care. Treatment regimens and durations varied according to resistance pattern and clinical status, with median treatment duration of 13 months (IQR 9–19 months). Among 148 patients with DR-TB who were initiated on treatment for DR-TB, 117 (77%) were cured, 7 (5%) completed, 5 (3%) were lost to follow-up, 14 (9%) died, 4 (3%) failed, 3 (2%) transferred out and 3 (2%) remained on treatment at the time of data censoring. There was no significant difference in the proportion of each outcome according to community (n=79) or hospitalised status (n=74). One patient with treatment failure had initiated treatment for DR-TB but was found to have MDR after repeat DST; the patient was transitioned to MDR treatment and the outcome of that treatment episode was included in the cohort of MDR patients.

Predictors of treatment success

Predictors of treatment success were evaluated using logistic regression (table 3). The crude odds for treatment success were similar between the hospitalised and community groups, and this similarity persisted after adjusting for sex, age, HIV status, resistance pattern and BMI (adjusted OR (aOR) 1.0 community vs hospitalised, 95% CI 0.63 to 1.60, p=0.99).

Factors strongly associated with decreased odds of treatment success included: age ≥44, HIV infection and baseline BMI <16. While not statistically significant, the relative odds of treatment success for patients with XDR-TB were 31% compared with patients with INH and RIF resistance alone or RIF resistance diagnosed by GXP (table 3). The relative odds of treatment success were only 33% in HIV-infected patients compared with HIV-uninfected patients, a relationship that persisted after adjustment for potential confounders (p<0.001) (table 3). Patients with BMI less than 16 (severely underweight) at the time of treatment initiation were only 40% as likely to have treatment success compared with patients with a baseline BMI ≥16, after adjusting for confounders (p<0.001). Older age was also significantly associated with lower odds of treatment success (aOR 0.40, p<0.001).

To reduce bias due to early in-hospital deaths in the comparison between community and hospitalised groups, the treatment cohort was restricted to patients who survived for at least 1 month after starting treatment in a sensitivity analysis. Twenty patients, including 11 HIV-infected patients (nine of whom were on ART), died in hospital less than 1 month after starting treatment, and were thus excluded from this sensitivity analysis. The aOR for treatment success in the community versus hospitalised groups was not different from the estimate from the entire cohort (aOR 0.86, 95% CI 0.53 to 1.40), and this was still not statistically significant (p=0.55). ORs of all other predictors were similar.