Table 2 Available evidence and CRESIPT project operational decisions relating to cash transfers

Cash transfer schemes

• Cash transfer schemes were implemented in Latin America in the 1990s to tackle the socioeconomic consequences of financial crises [21]. Schemes include: Bolsa Familia (Brasil, 1995); Oportunidades (Mexico, 1997); Red de Protección Social (Nicaragua, 2000–2005); Bono de Desarrollo Humano (BDH Ecuador, 2004); Red Solidaria (El Salvador, 2005); and JUNTOS (Peru, 2005)

• We investigated the use of food or other vouchers/cards but found very few existing systems in the study site. Those that were in place could only be redeemed in supermarkets (felt in FGDs to be inappropriate for the study population due to infrequent use, limited access and higher costs of goods)

• Our systematic review revealed only one controlled trial of TB-specific cash transfers from South Africa that showed no significant increase in successful TB treatment outcome [41]. During treatment, vouchers (15 US Dollars) that could be exchanged for foodstuffs were given to patients by local health post nursing staff. The authors opted for vouchers over cash due to: posing a lower security risk; not being able to be spent on unhealthy items such as alcohol or cigarettes; being easy to monitor; and “public health sector clinics would not have bank accounts, making electronic transfers difficult” [41]

• Based on our experiences and the limited published evidence, we opted for a bank cash transfer scheme. Bank transfers reduce the likelihood of fraud, robbery or security risk (a concern in impoverished shantytowns in Lima, Peru) [37, 41] and are a reliable way to maintain accurate transfer records for cost-effectiveness analysis. We also felt opening a bank account and having freedom of choice to decide on how transfers were spent was empowering [44]

• We decided not to impose conditions on how the cash transfers were spent. Successfully funded social protection interventions related to TB (especially MDR TB) have mainly focused on mitigating non-medical direct costs associated with having TB such as food or travel [17, 44]. There is some evidence that even when money rather than food vouchers is given as a form of social protection, it is commonly spent on food and travel costs anyway [45]

Conditionality of cash transfers

• Cash transfers can be unconditional, conditional (requiring specific behavioural, education or health actions) or combined [46, 47]

• Perú has an exemplary, well-established and organised National TB program. Learning from ongoing collaboration with regional heads of the TB program, we decided that our cash transfers conditions would relate to National TB Program treatment and prevention goals and selected project activities

• Unconditional cash transfer schemes include: Ecuador’s BDH targeting those below poverty threshold or by location; [48, 49] and a village bank loan scheme for TB-affected households in Cambodia [50]

• We chose to use conditional cash transfers that mixed both hard and soft conditions to be more inclusive: “hard” in that if participants met the condition with “perfect behaviour” then a double cash transfer was provided and “soft” in that if participants met the condition with adequate behaviour, then a simple cash transfer incentive was provided (Fig. 4a and b)

• Conditions can be “hard” (if the condition is not met, the transfer is not made) or “soft” (less stringent conditions where transfers may be made even when a condition is unmet). Soft conditionality may be preferable in settings with poor healthcare infrastructure [21, 46, 51]

How much cash to give

• Minimal evidence exists on the size of cash transfers. In Latin America, total amounts have varied widely in previous projects: 6-10% of annual income in Ecuador; [49] 21.8% in Mexico; and 29.3% in Nicaragua [47]

• We aimed to establish an amount for the cash transfers that was too small to act as a perverse incentive [34, 35], but large enough so that poverty-related TB risk factors in TB-affected households were reduced and the households were both incentivized and enabled to achieve National TB program and project goals

When to give cash

• Most initiatives deal more with poverty than a finite illness such as TB, so evidence of duration and frequency of TB-specific cash transfers is scarce. Longer duration and more frequent cash transfers may have greater impact in TB-affected households [31]

• We decided to provide the majority of the cash transfer incentives in the intensive treatment phase (the first 2 months of treatment) and to continue monthly cash transfers specific for treatment adherence throughout treatment. This meant the intervention was designed to increase equity for people with TB-HIV co-infection and MDR TB whose treatment lasted longer than 6 months

• Our previous work in the study setting showed that hidden TB costs were mainly incurred pre-diagnosis or early in treatment [7]