Do incentives improve tuberculosis treatment outcomes in the Republic of Moldova?

Abstract

Setting: Tuberculosis (TB) health facilities in the Republic of Moldova, where various incentives were provided to TB patients to improve treatment outcomes.

Objective: To compare treatment outcomes among new drug-susceptible TB patients registered for treatment before (2008) and after (2011) introduction of incentives.

Design: Retrospective cohort study using data from the national electronic patient database and incentive registers.

Results: Of 2378 patients registered in 2011, 1895 (80%) received incentives (cash, food vouchers, travel reimbursement). Compared to 2008 (no incentives, n = 2492), the patients registered with incentives in 2011 had higher treatment success (88% vs. 79%, P < 0.001) and lower proportions of unsuccessful outcomes: loss to follow-up (5% vs. 10%, P < 0.001), death (5% vs. 6%, P = 0.03) and failure (2% vs. 5%, P < 0.001). In multivariate analysis (log-binomial regression) using the intention-to-treat approach, provision of incentives was independently associated with an overall reduction in unsuccessful outcomes of 50% (RR 0.5, 95%CI 0.45–0.62, P < 0.001), after adjusting for other confounders such as sex, age, education, occupation, residence, homelessness, type of TB and human immunodeficiency virus status.

Conclusion: Provision of incentives to TB patients significantly improved treatment success rates and needs to continue. Treatment retention increased, thus potentially preventing drug resistance, a serious problem in the Republic of Moldova.

Tuberculosis (TB) re-emerged as an important public health problem in the Republic of Moldova after the breakup of the Soviet Union. Between 1990 and 2005, the number of notified TB cases (new and relapses) doubled. While there has been a decrease in notified cases since 2005, the rate of decline has been modest and slow.¹ TB incidence was estimated at 160 (range 132–190) cases per 100 000 population in 2012.² In addition, the problem of drug-resistant TB has assumed alarming proportions, and has been increasing. The results from the 2012 Drug Resistance Surveillance showed that 24% of newly diagnosed and 64% of previously treated patients had multidrug-resistant TB (MDR-TB),¹ higher than in 2006 (19% and 51%, respectively).³ During 2007–2011, the loss to follow-up rates among patients on first-line anti-tuberculosis drug regimens were high, at 10–12%, and even higher, at around 25%, among MDR-TB patients.¹

To improve treatment adherence, a system of incentives for patients, all provided after hospital care, was introduced in 2009 and had reached countrywide coverage (with the exclusion of the region of Transnistria) only by the end of 2010. Incentives varied depending on actual availability and source of funding (the Global Fund to Fight AIDS, Tuberculosis and Malaria [Global Fund], National Health Insurance Company [NHIC], local public administrations) and type of patient (drug susceptibility, urban/rural residence, treatment phase). Concern has been raised in the international arena about the impact of these incentives on anti-tuberculosis treatment outcomes. Previous studies on the use of incentives to improve treatment outcomes have shown conflicting results, with some showing benefit and others not.^4–6 A systematic review of randomised clinical trials published in 2012 concluded that there was limited evidence to support the use of material incentives in improving adherence to anti-tuberculosis treatment.⁷ One of the main recommendations of the external review of the National Tuberculosis Programme (NTP) conducted in 2013 was to improve consistency of the incentives used in the country and measure their effectiveness on improving treatment outcomes.⁸

The aim of the present study was to assess the system of incentives in the Republic of Moldova and guide the Ministry of Health to make an informed decision about continuing the system while improving effectiveness and sustainability. The specific objectives of the study were:

1. To describe the system of incentives given to TB patients in 2011.

2. Among drug-susceptible new TB patients registered in 2011:

   • a) to assess the number (proportion) of patients who received incentives during treatment
   • b) to describe the type of incentives among those who received them
   • c) to describe the treatment outcomes stratified by the type of incentives received

3. To compare treatment outcomes among new drug-susceptible TB patients registered for treatment before (2008) and after (2011) the introduction and countrywide coverage of incentives in the Republic of Moldova.

METHODS

Study design

This was a retrospective cohort study.

Setting

The Republic of Moldova, situated in south-eastern Europe between Romania and Ukraine, has a population of approximately 4 million. After independence from the former Soviet Union, the country went into economic decline, with increasing poverty and ill health in the population, including TB. However, during the last decade, the gross domestic product per capita has increased progressively, and with it, a stabilisation of poverty levels.⁹

The Ministry of Health has the primary responsibility for TB control in the country. It acts through the central unit of the NTP, located at the ‘Chiril Draganiuc’ Institute of Phthisiopneumology, which is responsible for NTP management, including collaboration with other national and international partners. TB services are provided via a network of specialised facilities and primary health care facilities. The NTP follows the World Health Organization guidelines for diagnosis and treatment of TB.^10–12

Study population and study period

The study population was composed of all new adult (aged ⩾18 years), drug-susceptible TB patients in the country who were registered for treatment in 2008 (before the introduction of incentives) and in 2011 (after the introduction of incentives). Incentives were introduced in 2009 and gradually scaled up in a phased manner to achieve nationwide coverage by the end of 2010. Given the variable implementation of incentives in 2009 and 2010, we chose 2008 as the control cohort, to reflect TB patients who had definitively not received any incentives. Patients from the region of Transnistria and incarcerated patients were excluded, as there was no system of incentives for these populations during the study period. Those patients not confirmed to have TB after registration or diagnosed later with MDR-TB were excluded. The study was conducted from June 2013 to March 2014.

Data variables and data sources

The term ‘incentive’ in this paper indicates the rewards and reimbursements to cover indirect costs of care provided to patients to encourage them to adhere to treatment.⁷ Data on incentives to patients were collected from the registers or other relevant records kept at the local level by non-governmental organisations implementing the Global Fund TB grant and by TB facilities implementing NHIC funds. Data on socio-demographic and clinical characteristics of TB patients were extracted from the System for Monitoring and Evaluation of TB (SYME-TB), the national database used by the NTP to store patient information.

Data analysis

Data were extracted into and analysed using Statistical Package for the Social Sciences, Version 20.0 (IBM Corp, Armonk, NY, USA). Anti-tuberculosis treatment success rates were compared between patients registered in 2008 (when no incentives were provided) and those in 2011 (when incentives were provided). Univariate analysis was conducted to describe the socio-demographic and clinical characteristics of patients. Continuous variables were summarised as mean (standard deviation) or median (interquartile range); categorical variables were indicated as proportions. To examine whether incentives were independently associated with improvement in treatment success after adjusting for other possible known confounders, we performed a multivariate analysis (log-binomial regression) using Stata v12.1 (Stata Corp, College Station, TX, USA) per both intention-to-treat and per-protocol approaches. Relative risks (RRs) with 95% confidence intervals (CIs) were calculated as measures of the association; P < 0.05 was considered statistically significant.

Ethics approval

Ethics approval was obtained from the Ethics Advisory Group of the International Union Against Tuberculosis and Lung Disease, Paris, France.

RESULTS

Table 1 shows five different types of incentives (‘small’ cash, ‘bigger’ cash, vouchers for food/hygiene products, reimbursement for transport costs, other support) provided in four different combinations and time periods by the Global Fund and in six combinations by the NHIC, in addition to other support provided by local administrations. The various combinations of incentives could be grouped into four main categories: 1) small cash only; 2) small and bigger cash; 3) non-cash incentives (reimbursement for transport costs and/or voucher for food and hygiene products); and 4) a combination of cash and non-cash incentives (including other support from local administrations). The patients in the ‘small’ cash group received a monthly sum of 315 Moldovan lei (MDL, USD22.49) during their out-patient treatment only; this incentive was stopped if they interrupted treatment for ⩾5 days. The patients in the ‘small’ and ‘bigger’ cash group received 1450 MDL (USD102.94, equivalent to the average monthly salary of a nurse) in addition to the above at treatment completion. The patients in the ‘non-cash incentive’ group were reimbursed for transport costs and/or vouchers for food/hygiene products, depending on their area of residence, i.e., daily if living in an urban area, or monthly. Some patients received a mix of all of the incentives mentioned above. Twelve patients also received ‘other support’, such as clothes, wood for cooking, etc., as decided by the local administration.

TABLE 1.

Incentives given to tuberculosis patients registered for treatment, Republic of Moldova, 2011

Of the 2378 new drug-susceptible TB patients registered for treatment in 2011, 1895 (80%) received different types of incentives. Among these, 1291 (68%) were male; the mean age was 41 years. Among those who received incentives, nearly 95% received cash with or without non-cash incentives. A total of 483 patients received no incentives (Table 2).

TABLE 2.

Type of incentives provided to tuberculosis patients registered for treatment, Republic of Moldova, 2011

Table 3 shows the anti-tuberculosis treatment outcomes stratified by the combination of incentives received. It should be noted that treatment success rates were similar (in the range of 92–97%), irrespective of the incentives received. Among the 483 patients who did not receive any incentive in 2011, 280 (58%) were successfully treated, 11 (2%) were treatment failures, 87 (18%) were lost to follow-up and 105 (22%) died.

TABLE 3.

Treatment outcomes among new drug-susceptible tuberculosis patients, stratified by incentives received, Republic of Moldova, 2011

The baseline socio-demographic and clinical characteristics of the 2008 (no incentives) and 2011 (with incentives) cohorts of TB patients are compared in Table 4. There were some differences in baseline characteristics between the two cohorts with respect to residence, age group, homelessness, human immunodeficiency virus (HIV) status and sputum smear results. Compared to 2008 (no incentives), the patients registered in 2011 (with incentives) had a higher treatment success rate (88% vs. 79%, P < 0.001) and lower proportions of unsuccessful outcomes: loss to follow-up (5% vs. 10%, P < 0.001), death (5% vs. 6%, P = 0.03) and failure (2% vs. 5%, P < 0.001) (Table 5).

TABLE 4.

Comparison of socio-demographic and clinical characteristics between new tuberculosis patients registered for treatment in 2008 and 2011, Republic of Moldova

TABLE 5.

Treatment outcomes among new drug-susceptible tuberculosis patients registered for treatment in 2008 and 2011, Republic of Moldova

In the multivariate analysis (log-binomial regression), using the intention-to-treat approach, we found that incentives were independently associated with treatment outcomes and reduced the proportion of unsuccessful outcomes by 50% (RR 0.5, 95%CI 0.45–0.62, P < 0.001), after adjusting for other confounders such as place of residence, sex, age, occupation, homelessness, HIV status and type of TB. In a stratified analysis, we found that the protective effect of incentives was seen among both employed (RR 0.53, 95%CI 0.52–0.67) and unemployed TB patients (RR 0.59, 95%CI 0.51–0.68). We performed a secondary analysis using a per-protocol approach and found that the effect of incentives was even more pronounced in reducing unsuccessful outcomes (RR 0.20, 95%CI 0.16–0.25, P < 0.001).

DISCUSSION

This is the first study from the Republic of Moldova to provide clear, conclusive evidence that patient incentives were independently associated with improved anti-tuberculosis treatment outcomes. Among those who actually received incentives, treatment success rates were around 95%. This is likely to have prevented the development of drug resistance, a serious problem in the Republic of Moldova. This addresses the concerns of national policy makers about the effectiveness of such treatment support measures, and provides strong justification for the continuation of the incentive system in the country. It also provides the necessary evidence for international donors, such as the Global Fund, to continue and even scale up similar support in other countries.

Several issues merit further discussion. First, as expected, the main effect of patient incentives was observed in the reduction in loss to follow-up among TB patients. In addition, incentives reduced treatment failure rates moderately and death rates minimally, and this may also have been related to improved treatment adherence among patients. There was a large difference in unfavourable outcomes between patients who received incentives and those who did not in 2011. This may be due to the fact that many patients might have died or were lost to follow-up early in the course of treatment and hence may not have been available to receive incentives. We did not have well-documented information on when incentives were provided or the exact time of treatment outcome; we cannot therefore comment on whether this difference in outcomes was related to when the incentives were provided. This is a limitation of the study. Second, while there was clear evidence indicating that incentives worked, treatment success rates between the types of incentives provided were not statistically significantly different. Overall, cash incentives appeared to work marginally better than non-cash incentives; this difference was not statistically significant, due to the small number of patients who received non-cash incentives. A previous systematic review on this issue has shown that cash incentives work better than non-cash incentives.¹³ Third, our study provided an overview of the system of incentives provided and the large variation in the types provided owing to the different stakeholders involved. While this could be viewed as a positive sign of flexibility on the part of the NTP in adapting to different patient needs, an over-diverse system of incentives can pose logistical challenges in planning, organisation and distribution of incentives and predicting the financial resources required. Furthermore, the future sustainability of the system of incentives in the Republic of Moldova will depend not only on availability of domestic funds but also on a more rational plan in using them.

There were several strengths to our study. First, this is the first report from the Republic of Moldova on the impact of incentives on anti-tuberculosis treatment outcomes. Second, as we had a countrywide sample covering all registered TB patients, our findings were nationally representative and generalisable to settings with demographic and socio-economic characteristics similar to those of the whole of the Republic of Moldova. There could be concerns about the generalisability of the findings beyond Moldova, given the relatively high unemployment rate (~60%) among TB patients. However, stratified analysis indicates that the protective effect of incentives was seen among both employed and unemployed TB patients. Third, we followed STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for reporting the study, including ethics guidelines.¹⁴

There were several limitations related to the retrospective, operational nature of the study and the necessary reliance on existing reporting forms/registers as data source. First, we did not have information on the period during which incentives were given to patients. Second, we had to rely on a historical control for comparing outcomes. It is well known that many other changes could have occurred between 2008 and 2011 that could also have contributed to the improvements in treatment success. While we took adequate care to measure patient-related characteristics and adjust for any potential confounders and mitigate the impact of this issue, this limitation remains. Third, it was a challenge linking data between the TB records (electronic) and the ‘social support’ records (paper-based) to assess the impact of incentives on TB treatment outcomes. The existence of a good national TB database, the special efforts made to retrieve paper records and the support of the Structured Operational Research and Training Initiative (SORT IT) project made this possible. In the future, efforts should be made to design documentation systems that are electronic and enable real-time, routine monitoring of the effectiveness of such interventions.

The provision of incentives to TB patients significantly improved treatment success rates and needs to be sustained. This has the potential to improve patient adherence and treatment retention, thus preventing drug resistance, a serious problem in the Republic of Moldova.