Towards cash transfer interventions for tuberculosis prevention, care and control: key operational challenges and research priorities

Establishing that cash transfers can effectively impact TB control is only the first step in deciding whether to, and how to, implement a cash transfer programme. Consistently with what suggested for HIV/AIDS [13, 35], three implementation models can be conceptualised in the context of TB elimination: TB-specific, TB-sensitive and TB-inclusive. TB-specific interventions are interventions targeted at TB-affected households/individuals with the precise intent of improving a number of TB indicators. TB-sensitive interventions are interventions that can potentially affect TB epidemiology and control by targeting people at high risk of TB. Finally TB –inclusive are interventions in which having TB or being a member of an household affected by TB is one of the inclusion criteria for the programme, albeit not the only one (Table 1). Each of these models is characterised by a different scope, targeting strategy, extent of integration with the local social protection platforms, and advantages and disadvantages that are likely to be highly setting-specific (Table 1).

It is likely that - as for HIV/AIDS – the most cost-effective approach will be to adapt existing cash transfer schemes to make them more TB-sensitive, rather than creating or scaling up de novo programmes. However, it remains unclear as yet how existing national TB programmes will build effective and cost-effective partnership models with the institutions in charge of the design, implementation and evaluation of governmental cash transfer programmes.

Furthermore, the potential to implement TB-sensitive cash transfer schemes seems to follow an “inverse care law” pattern [36] by which most impoverished countries highly affected by TB are least able to afford cash transfer implementation or expansion to encompass TB objectives without jeopardizing existing TB health care provision coverage or quality. For example, pilot cash transfer schemes launched in Zambia between 2003 and 2004 [37‐39], were effective in helping destitute HIV-affected households withstand the economic impacts of AIDS. However, these schemes have not been absorbed into the country’s welfare system [38], may have limited capacity for scale-up or sustainability [37] and may lack the human and financial resources for monitoring and evaluation. This limits the potential of these schemes to become TB-sensitive or TB-inclusive by including TB care, prevention and/or support objectives.

Even in countries where comprehensive government-led social protection programmes are in place, there may be obstacles to make these programmes more inclusive for TB patients. For example, South Africa has extensive cash transfer schemes, including the Temporary Disability Grant in which TB patients are eligible to receive cash when unable to work. While the scheme was initially widely implemented, over the years the number of TB patients eligible for the scheme has been reduced to include only patients with multi-drug resistant TB (MDR-TB) and/or severe TB-related complications [40]. The reasons for this reduction included lack of evidence of the impact on TB care and prevention, insufficient allocated resources to cope with the growing number of TB-HIV co-infected patients applying for the grant, the increasing roll-out of anti-retroviral therapy, and the anecdotal reports of fraud by healthy individuals buying sputum samples from TB patients to falsely receive a diagnosis of TB and thus qualify to receive (or continue to receive) the benefits [22].

A different picture emerges from Brazil. Experts of the Bolsa Familia Programme, the Brazilian conditional cash transfer programme for impoverished households [41], have recently argued that, in Brazil, TB is serious enough but also infrequent enough to incorporate TB into the existing Bolsa Familia logistical and financial capacity, thereby becoming more TB-inclusive or TB-sensitive [Soares S. personal communication]. Furthermore, the preliminary evidence of the impact of Bolsa Familia on major public health issues, including TB, are encouraging [42]. Nonetheless the operational challenges of this synergy have not yet been fully explored.

Choosing a TB-specific format may appear to be simpler from an operational perspective but it may cause a number of implementation issues in terms of management and delivery of financial incentives from the TB care providers. Such issues emerged from the process evaluation of an intervention in South Africa [43]: the trial results suggested a small, but not statistically significant effect of the provision of vouchers on TB treatment success rates in the intervention communities [44]. However, fidelity to the intervention was low with approximately 36 % of eligible patients not receiving any voucher [44]. Such low fidelity was attributed mainly to two factors: first, nurses modified the delivery of the vouchers, mainly by arbitrarily deciding who was eligible to receive the incentive and also rejecting the trial eligibility criteria, which were perceived by the nursing staff to be inequitable. Secondly, from the nurses’ perspective, it was logistically easier to give all the vouchers in one batch at the end of the month. This practical ground-level implementation meant that some patients were obliged to come back to the clinic more than once to collect the voucher, a situation that was often unfavorable for these patients. Furthermore, the coordination of delivery of vouchers to clinics and collection of vouchers from local shops required considerable organization and a dedicated staff complement [43].

In contrast to HIV/AIDS policy [35, 45], there is insufficient evidence to advocate for the adoption of a TB-sensitive approach over a TB-inclusive or TB-specific approach. However, the choice between TB-sensitive, TB-inclusive or TB-specific interventions may be guided by a preliminary programmatic assessment of both the TB epidemic profile and social protection environment based on a preliminary list of questions suggested in Table 2. This programmatic assessment should help countries to evaluate whether the TB control objectives can be realistically incorporated into existing schemes and how this can be best achieved. Ideally, National TB control programmes can have a key role in this initial assessment because they are well placed to understand how cash transfer interventions can be best tailored to TB patients and/or people at risk for TB. Ministries of Finance and Development may then advise on the foreseen issues such as the operational, financial and ethical barriers associated with the implementation of either a TB-sensitive or a TB-specific intervention.

Given the limited operational evidence on the feasibility of TB-sensitive cash transfer interventions it may be more appropriate to establish initially TB-specific cash transfer interventions run by national TB control programs and co-implemented with a governmental or non-governmental organization, with the intent of addressing certain TB-specific issues such as poor case-finding, low cure rates or TB clustering in high-risk groups (e.g. homeless, drug abusers, indigenous populations). While the long-term objective will be most likely to move towards a TB-sensitive approach, TB-specific initiatives can still be used as a valid starting point to gain insight and knowledge on the expected TB impact of these interventions as well as the costs and other important aspects to be considered when designing and implementing cash transfer interventions for TB control purposes (whether conditional or unconditional). An example of TB-specific intervention is the recent evaluated trial that in South Africa providing vouchers to TB patients [43]. Another example, the CRESIPT study in Peru, is currently under impact evaluation and summarised in Table 3 [46].

The effectiveness and cost-effectiveness of conditional versus unconditional cash transfer schemes is under debate. For example, conditionality does not seem to be essential to meet HIV-related objectives [26, 47]. Furthermore, the implementation of conditionality to achieve public health objectives, including those related to TB programmes, may be hampered by existing administrative and health system weaknesses. Therefore, cash transfer programme interventions may not have the financial or administrative capacities to monitor the health-seeking behaviour conditionality or offer adequate health systems infrastructures to make these conditions feasible or ethical [6, 48‐50]. For example, in settings with limited TB care services, implementing conditional cash transfer schemes to improve treatment TB adherence may have limited impact if treatment or access to treatment is not consistently available.

Conditionality may also raise important questions concerning choice, autonomy and decisional capacity of the poor without due flexibility determined by cultural and socioeconomic factors [48]. Conditionality may stimulate poor people to seek TB care while neglecting other important health priorities in the household such as maternal/child health or to act against their common belief or will [51, 52]. Evidence concerning this is inconsistent: the low uptake of the Janani Suraksha Yojani programme in India among the poorest and least educated women demonstrates that conditioning the cash transfers do not necessarily overcome cultural or geographical barriers that hamper people’s capacity to meet behavioural conditions [14]. Conversely, in Malawi, monetary incentives of less than a tenth of a day’s wage dramatically changed people’s health seeking behaviours by compensating the psychological and economic costs of HIV testing and reduced the gender gap in accessing HIV services [53]. Thus, cash incentives or enablers for seeking TB care may compensate for the direct material and the indirect psychological costs of TB diagnosis and ultimately increase gender equity in access to TB diagnosis [53]. In conclusion: a) There is currently no direct, rigorous evidence to guide to guide the circumstances under which TB-related cash transfer interventions should be conditional and if so, on what TB-related behaviours conditionalities should be applied and how stringently. b) Even evidence from the HIV/AIDS experience is not conclusive and the issue has not yet been rigorously explored [13]. Finally, c) the effectiveness of utilizing conditionality in cash transfers depends on the outcome under evaluation, the feasibility of the condition, the program context, and the target population [34, 54, 55].

Given the numerous knowledge gaps, the strict application of conditionality to increase TB case detection and treatment success rate may be unjustified in countries with limited administrative and health care capacities. In these settings, an option may be a “soft” form of conditionality in which conditions are simple or made less stringent (e.g. penalties for non-adherence do not imply the exclusion of participants from the programme) [48, 55]. Another possibility may be to indirectly encourage the attendance of TB care services by making the cash transfer conditional upon the attendance of workshops and training sessions concerning TB that may help TB-affected families to overcome the fear and the stigma of approaching the TB services.

Should conditionality be deemed appropriate, it may be useful to assess the features of the local TB epidemic and available care. This may be accomplished with qualitative research with TB-affected families and main TB stake-holders to test the feasibility and acceptability of conditionalities from the target population and to design them so to meet better the specific challenges experienced by TB-affected families.

Cash transfer programs may be made TB-sensitive by targeting people at high risk of TB, or may be transformed to TB-inclusive programs by specifically including TB patients among their target population. These programs may broadly target TB-patients and/or their households; restrict their target to TB-patients subgroups (e.g. drug-resistant TB, TB-HIV co-infected or TB patients affected by other co-morbidities such as mental health issues or diabetes); or restrict to TB-patients during their highest risk time period (e.g. the time period when treatment loss to follow up or onward TB transmission is most likely).

Depending on the targeting strategy, issues concerning accuracy, costs, cost-effectiveness, equity, fairness, sustainability, unintended consequences and stigma may arise: for instance, using TB as a targeting criterion may be more efficient, but could be stigmatising. Furthermore, it may foster perverse incentives in which patients attempt to remain sick assuming that this may entitle them to receive the cash benefits. This concern was specifically raised by TB care providers in the intervention from Lutge et al. [43], although this was not observed and conversely patients receiving vouchers more often tended to be more likely to achieve treatment success [44].

Conversely, “passive” targeting, based only on poverty criteria, may be inefficient or impractical. For example, a 2010 analysis in Brazil cross-linking the national TB registry with the Single Registry for Social Programmes (a census database to identify people enrolled in social protection programmes in Brazil) showed that only one fourth of TB patients who live below the poverty line are already enrolled in government’s social protection programmes, with only 14 % of them being amongst the beneficiaries of Bolsa Familia. This unexpectedly low overlap between the two data sources suggest that in Brazil targeting based only on socioeconomic criteria may be insufficient to capture most TB patients in needs of social protection for a number of reasons discussed by Torrens and colleagues [29]. Moreover, even if such poor families are enrolled in Bolsa Familia, it is often the case that they do not have the necessary requirements to actually receive the transfers (e.g. a billing address, identification documents, or proof of income documents) [29, 56]. In contrast, in Zambia targeting for vulnerability to HIV based on poverty levels may be considered a viable option to also capture a substantial proportion of households affected by TB as well as HIV that are in need of financial support that avoids stigmatisation [38].

There is very limited evidence to inform the best targeting strategy for cash transfer interventions aiming at enhancing TB elimination; however the most suitable approach seems to be highly setting-specific and pretty much dependant on the TB issues to be addressed. For example, in high-TB burden countries preventing TB transmission may require targeting people at high-risk for TB; on the other hand, in settings where TB priorities are more circumscribed (i.e. addressing TB-related catastrophic costs, enhancing TB treatment adherence), and/or the epidemic is restricted to selected groups of the population (i.e. drug users, homeless, prisoners) then it may be more beneficial to target directly TB-affected households or TB patients.

As a general rule, cash transfer interventions, and more broadly social protection initiatives, should be delivered in way to be non-stigmatising, non-discriminatory and aim to optimise inclusiveness and equity [35]. In order to achieve this ultimate goal, counties are often encouraged to use multiple criteria [48] such as multidimensional targeting that identifies beneficiaries based on the objectives of the intervention and various forms of deprivation affecting the target population [57]. In the case of TB-sensitive or TB-inclusive cash transfer intervention, a multiple targeting strategy could include a combination of the following criteria: A. living in extreme poverty as assessed by econometric or asset-based indices; B. not having household members able to work; C. not having other forms of social assistance; and D. the household TB history or a high household risk score based on TB risk factors (such as household prevalence of HIV, alcohol abuse, smoking, diabetes) [48].

Data concerning existing cash transfers costs are limited, not always comprehensive, and lacking for low compared to middle income countries. Where available, evidence suggests an average cost between 12.5 US$ and 26.5 US$ per month per beneficiary (with 4 to 20 % of the monthly household consumption expenditure of the beneficiaries covered), which represents from 0.36 to 1.54 % of GDP for any given implementing country [24, 58]. Analyses of costs of social protection packages generally conclude that social protection is affordable in most countries, including deprived settings such as Sub Saharan low income countries, as long as costs are less than 1 % of countries’ GDP share [58, 59]. In Zambia, for example, the annual costs of the three pilot studies for the national unconditional cash transfer were US$40-$82 million [38]. A 2009 analysis showed that even if a pilot in the Katete district was expanded throughout the country to target 450,000 beneficiaries with transfers of an average of US $15/month, this would represent only 2.4 % of the Zambian National budget or 0.7 % of the GDP in 2007 [38]. These data suggest that implementing the Social Cash Transfer scheme on a national scale in Zambia is likely to be affordable. However public spending in Zambia on social assistance represents <0.1 % of GDP [38] suggesting that scaling up the Social Transfer Scheme across the country requires increased financial investment, but also political will [37].

It remains also unclear whether economic assessments should distinguish between different types of costs (e.g. administrative versus transfer costs) as well as distinguish between programs supported by aid provided by international donors versus domestic funding. Generally, government-led schemes are considered to have greater long-term sustainability [37, 38]. However, it is not yet clear how many government-led cash transfer programs may have the financial and logistic resources to sustainably encompass specific public health objectives, such as TB [60]. The incorporation of TB beneficiaries in existing cash transfer initiatives or linking them with TB activities is likely in fact to require additional costs for setting up new operating procedures, staff training and the creation of new collaborative forms of social and medical management. Countries may not be able to afford these costs without mobilizing extra-funds. In these circumstances, funding sources may be hesitant to sponsor collaborative projects if these payments may interfere with the performance and objectives of the original programme [60].

Cost concerns are legitimate; Referring to TB specifically, globally there are about 9 million new cases of TB each year. Hypothetically, even if $100 worth of social protection were to be additionally provided to every patient globally then the resultant expense of would be approximately $900 million. This is not a modest expense in the current global TB investments. However, costs should be evaluated through a proper extended cost-effectiveness analysis (ECEA), which accounts not just for the health gain against the cost of an interventions, but also the financial protection achieved as a result of that intervention (both in terms of poverty aversion and equity improvement) [61, 62]. In the case of TB, the potential financial benefit arising from proper control measures seems considerable: ecological studies have reported an effect of 0.2–0.4 % decreased economic growth for every 10 % higher TB incidence [63]. This corresponds to an annual loss of US$1.4-2.8 billion in economic growth worldwide. In addition, the World Bank estimates that loss of productivity attributable to TB is 4-7 % of some countries' GDP [61]. In a more recent study, Verguet and colleagues have positioned TB treatment among the best interventions at preventing medical impoverishment (i.e. 96 cases of poverty averted per every US$100,000 spent) [62].

At the micro-level the impact of cash transfers is expected to be even more significant: in most settings the average costs encountered by TB sufferers exceed 10 % annual household expenditures [64]. In worst cases this burden can exceed 20 %, a threshold that has been defined as “catastrophic” not only because associated with a further impoverishment of TB-affected households, but also associated with an increased risk of TB mortality, treatment loss to follow up, or TB recurrence [65]. Cash transfers can reduce financial vulnerability and potentially improve TB treatment adherence and outcome [65]. Furthermore, it should be considered that even TB-specific cash transfer interventions are likely to achieve multiple health and development impacts, not just TB [12, 65]. The capacity to produce multiple benefits is the strongest argument to propose co-financing mechanisms through which various sectors pool funds or engage in joint budgeting to fund interventions with multi-sectoral benefits [12, 66]. Under this approach several models have been proposed: one may be to consider the TB-related costs separately from the core services of the original intervention. International and/or domestic donors may be willing to specifically cover these marginal TB-related costs [18, 66, 67]. Alternatively it may be worth exploring the creation of a partnership among donors as theorised in the UNAIDS Investment framework to realise health and development funding synergies [68].