Cash transfer and microfinance interventions for tuberculosis control: review of the impact evidence and policy implications

Abstract

OBJECTIVE

To quantify the impact of cash transfer and microfinance interventions on a selected list of tuberculosis (TB) risk factors and assess their potential role in supporting TB control.

DATA SOURCE

Published and unpublished references identified from clinical and social electronic databases, grey literature and web sites.

METHODS

Eligible interventions had to be conducted in middle- or low-income countries and document an impact evaluation on any of the following outcomes: 1) TB or other respiratory infections; 2) household socio-economic position; and 3) factors mediating the association between low household socio-economic position and TB, including inadequate health-seeking behaviours, food insecurity and biological TB risk factors such as human immunodeficiency virus (HIV) and adult malnutrition. Interventions targeting special populations were excluded.

RESULTS

Fifteen cash transfer schemes (four unconditional and 11 conditional) and seven microfinance programmes met the eligibility criteria. No intervention addressed TB or any other respiratory infection. Of 11 cash transfer and four microfinance interventions, respectively seven and four reported a positive impact on indicators of economic well-being. A positive impact on household food security was documented in respectively eight of nine and three of five cash transfer and microfinance interventions. Improved health care access was documented respectively in 10 of 12 cash transfer and four of five microfinance interventions. The only intervention evaluating impact on HIV incidence was a microfinance project that found no effect. No cash transfer or microfinance interventions had an impact on adult malnutrition.

CONCLUSIONS

Cash transfer and microfinance interventions can positively impact TB risk factors. Evaluation studies are urgently needed to assess the impact of these social protection interventions on actual TB indicators.

THERE IS EVIDENCE that the reduction in tuberculosis (TB) mortality observed in Europe and North America before World War II resulted from the successful combination of economic growth and specific public health policies such as patient isolation and the elimination of bovine TB.^1,2 The importance of a combined approach has recently been reinforced by a number of studies suggesting that while DOTS programmes have significantly contributed to the reduction of TB prevalence and TB mortality, socio-economic development still remains the main determinant of TB incidence decline in many regions of the world.^3-5 Despite this, literature on interventions addressing structural and social determinants⁶ of TB is currently virtually non-existent. Through the improvement of material living conditions, psychosocial circumstances and health-seeking behaviours, these interventions have the double potential to improve access to quality TB care and also reduce people’s vulnerability to TB.

A broad range of poverty alleviation strategies may achieve the above objectives; among them, cash transfer and microfinance programmes have gathered considerable visibility in the last decade, due to both the large number of individuals enrolled and the increasing number of studies formally documenting their impact.

Cash transfers are innovative forms of social protection based on the provision of money to poor or vulnerable households and individuals (such as the elderly and children) with the aim of enabling them to move out of poverty by protecting and building their financial, physical and human capital assets.⁷ Cash transfers can be unconditional (given without obligation) or conditional on some behavioural requirements, such as school enrolment, health services utilisation and/or the attendance of health education workshops. Microfinance programmes are considered poverty alleviation measures aimed at directly averting deprivation.⁸ These programmes are based on the provision to the poor of small loans to enable them to engage in productive activities that can ultimately contribute to their long-term economic growth and productivity.⁸

Despite the impressive scale-up of these interventions and their acknowledged potential public health relevance,^9,10 their relevance for TB control largely depends on evidence of their effect on outcomes epidemiologically linked to TB and the critical assessment of the challenges limiting their implementation in the context of TB.

This review aimed to address these issues by: 1) systematically quantifying the impact of these interventions and 2) evaluating their main design and implementation features from the perspective of TB control.

METHODS

Conceptual framework

Hargreaves et al. discuss the mechanisms by which cash transfer and microfinance interventions might contribute to strengthening the international TB response, which outcome and risk factors may be potentially impacted by such interventions, and which effect should be expected depending on the stage of TB targeted (i.e., TB exposure, infection, disease and outcomes).¹¹ Building on this conceptual framework, we designed this review to formally assess the impact of cash transfer and microfinance interventions on the hypothesised TB outcomes and risk factors. In particular, we focused on household socio-economic position and those risk factors that mediate the association between household socio-economic position and TB disease, including inappropriate health-seeking behaviours, food insecurity¹² and biological risk factors impairing host immune system defences (such as human immunodeficiency virus [HIV] and adult malnutrition; Figure).

Figure.

Conceptual framework. In this review we focus on tuberculosis (TB) disease and interventions targeting socio-economic position at household level. It is hypothesised that cash transfer, either conditional or unconditional, and microfinance interventions can support TB control by improving a household’s socio-economic position and thereby reducing the exposure to biological risk factors (such as human immunodeficiency virus and malnutrition) and improving the household food security and health-seeking behaviour.

Search strategy and selection criteria

A wide range of electronic databases and websites were searched (Table 1). Eligible interventions had to: 1) address the following outcomes of interest: TB (or other respiratory infections), household socio-economic position or any of the postulated mediating risk factors (Figure); 2) explicitly document an impact evaluation (i.e., the assessment of any change in a given outcome among the intervention beneficiaries compared to the non-beneficiaries); and 3) take place in low- and middle-income countries.

Table 1.

Search strategy. As the terminology in socio-economic interventions is often imprecise, technical (non-medical) and constantly evolving, we first conducted a scoping search with a provisional set of text word terms.¹³ We then refined the search strategy by adding indexed terms and text words as suggested by the studies identified in the preliminary search.

		Queries launched through electronic databases
Electronic  databases	Biomedical  EMBASE  MEDLINE  Global Health  HMIC  EPPI  DoPHER  TRoPHI Social sciences  Social Policy & Practice	Socio-economic  interventions Social protection/  livelihood  strengthening	(Socioeconomic OR Social OR Economic OR Complex OR Structural) AND (Programme OR Program OR Project OR Intervention OR Prevention) AND (Evaluation OR Impact) (Microcredit OR Microfi nance OR Cash OR economic empowerment  OR Income) AND (Poverty OR food security OR Access OR Health-seeking behaviours  OR TB OR Respiratory OR HIV OR Nutrition) AND (Reduction OR Progr* OR Project OR Prevention OR Integration) AND (Evaluation OR Impact)
Websites	http://scholar.google.com; http://www.omni.ac.uk; http://www.sosig.ac.uk; http://www.eelv.ac.uk; http://www.undp.org;  http://www.worldbank.org; http://www.eldis.org; http://www.usaid.gov; http://www.dfid.gov.uk; http://www.ifpri.org
Grey literature and expert consultation

Experimental and observational studies with a controlled or uncontrolled design providing a retrospective or a prospective evaluation of the intervention of interest were included. No time or language restriction was applied. Studies conducted in specific special settings (e.g., hospitals, prisons, workplaces, schools) or in specific population groups (e.g., homeless, professional categories, students) were excluded. Given the heterogeneity of the study design and study results, the evidence collected was summarised through a ’narrative synthesis’ approach.^14,15

RESULTS

Approximately 150 documents of potential interest were appraised. Through them, we identified 15 cash transfer and 7 microfinance intervention evaluations meeting the inclusion criteria, from respectively 13 and 5 different countries (Table 2). Evidence on microfinance was integrated with data from a recent review on the impact of microfinance on poverty and household socio-economic position.¹⁶

Table 2.

Study design and main study findings: an overview

					Impact findings*
Intervention	Country†	Type of intervention	Evaluation design	Follow-up period	Significant positive impact	Uncertain positive impact‡	No effect	Opposite impact to that expected
Poverty/household socio-economic position§
RMP	Bangladesh	CCT	H-QEXP	25 months¶	• •	—	—	—
BF	Brazil	CCT	H-QEXP	2004–2005	—	—	•	—
PFA	Colombia	CCT	C-QEXP	2002–2003	• •	—	—	—
BDH	Ecuador	CCT	CRT	2003–2005	—	—	•	—
PRAF	Honduras	CCT	CRT	2000–2002	—	—	•	—
PROGRESA	Mexico	CCT	CRT	1998–2000	—	•	—	—
RPS	Nicaragua	CCT	CRT	2000–2004	• • •	—	—	—
PSNP#	Ethiopia	SCT	H-QEXP	01/2005–06/2006	—	—	•	•
MCHINJI	Malawi	SCT	H-QEXP	2007–2008	• • •	—	—	—
DECT	Malawi	SCT	CSS	12/2006–04/2007	—	•	—	—
SCT	Zambia	SCT	H-QEXP	2004–2007	—	• •	—	—
CFRP/TUP	Bangladesh	Microfinance	C-QEXP	2002–2004	•	—	—	—
RDP	Bangladesh	Microfinance	H-QEXP	1993 and 1996	• •	—	—	—
IGVGD	Bangladesh	Microfinance	CSS	1994, 1996, and 1999	—	• • •	—	—
IMAGE	South Africa	Microfinance	CRT	2001–2005	—	• •	—	—
Food security/household food consumption patterns**
RMP	Bangladesh	CCT	H-QEXP	25 months¶	• •	—	•	—
BF	Brazil	CCT	H-QEXP	2004–2005	—	• • •	—	—
PFA	Colombia	CCT	C-QEXP	2002–2003	• •	—	—	—
PROGRESA	Mexico	CCT	CRT	1998–2000	•	• • •	—	—
RPS	Nicaragua	CCT	CRT	2000–2004	• • •	—	—	—
PSNP#	Ethiopia	SCT	H-QEXP	01/2005–06/2006	—	—	• • •	—
MCHINJI	Malawi	SCT	H-QEXP	2007–2008	• • •	—	—	—
DECT	Malawi	SCT	CSS	12/2006–04/2007	—	• • •	—	—
SCT	Zambia	SCT	H-QEXP	2004–2007	—	• • •	—	—
CFRP/TUP	Bangladesh	Microfinance	H-QEXP	2002–2004	• • •	• •	—	—
Grameen Bank††	Bangladesh	Microfinance	CSS	Not reported	•	—	—	—
WISDOM	Ethiopia	Microfinance	CSS	10 and 20 months¶	—	—	• • •	—
IMAGE	South Africa	Microfinance	CRT	2001–2005	—	—	•	—
Kafo Jigenew††	Mali	Microfinance	CSS	12 and 24 months¶	• •	—	—	—
Health-seeking behaviours/health care access††
BF	Brazil	CCT	C-QEXP	2004–2005	—	—	• •	—
PFA	Colombia	CCT	C-QEXP	2002–2003	•	—	—	—
PRAF	Honduras	CCT	CRT	2000–2002	• •	—	•	—
JSY	India	CCT	H-QEXP	2002/4–2007/9	• • •	—	—	—
PATH	Jamaica	CCT	H-QEXP	2004–2005	•	—	•	—
—	Malawi	CCT	RT	Not applicable§§	• • •	—	—	—
PROGRESA	Mexico	CCT	CRT	1998–2000	• •	• • •	—	—
SDIP	Nepal	CCT	CSS	2005–2008	• • •	—	—	—
RPS	Nicaragua	CCT	CRT	2000–2004	•	•	—	—
MCHINJI	Malawi	SCT	H-QEXP	2007–2008	• •	—	—	—
DECT	Malawi	SCT	CSS	12/2006–04/2007	—	•	—	—
SCT	Zambia	SCT	H-QEXP	2004–2007	—	—	•	—
RDP/ICCDR,P	Bangladesh	Microfinance	H-QEXP	1992,1996, 2001	—	—	—	•
CFRP/TUP	Bangladesh	Microfinance	H-QEXP	2002–2004	• • •	• •	•	—
CHC	Cambodia	Microfinance	CSS	1999–2001	• •	• •	—	—
SHG	India	Microfinance	CSS	≤48 and >48 months¶	•	—	•	—
IMAGE	South Africa	Microfinance	CRT	2001–2005	•	—	—	—
Factors impairing host defences: adult malnutrition¶¶ and HIV
RMP	Bangladesh	CCT	H-QEXP	25 months¶	—	—	•	—
PROGRESA	Mexico	CCT	CRT	1998–2000	—	—	—	•
MCHINJI	Malawi	SCT	H-QEXP	2007–2008	—	—	•	—
CFRP/TUP	Bangladesh	Microfinance	H-QEXP	2002–2004	—	—	•	—
WISDOM	Ethiopia	Microfinance	CSS	Not applicable§§	—	—	•	—
IMAGE	South Africa	Microfinance	CRT	2001–2005	—	—	•	—

^*• = evidence from one indicator; •• = evidence from 2 indicators; ••• = evidence from 3 or more indicators.

^†Countries listed in alphabetical order and by type of intervention.

^‡Measures of effect did not reach significance level or it was not reported in the reference.

^§Indicators include household consumption expenditure (monthly or annual, per capita or total), household income (monthly or annual, per capita or total), proportion of households below the poverty line or the extreme poverty line, number of assets owned and proportion of households in debt.

^¶Length of time of beneficiaries’ participation at the time of the survey.

^#The impact evaluation of the PSNP was conducted using three different definitions of ‘treatment’ households. In this article we reviewed the impact data following from the first definition (i.e., households that received any cash for public works in the period from June 2005 to May 2006). This was due to the second definition adopted providing data only from a sub-region of Ethiopia, whereas the third did include households receiving PSNP completed by other food security interventions, making it difficult to understand the net impact of PSNP.

^**Indicators include food quantity (i.e., number of meals per day, grams of food per day) and quality (food diversity, consumption of meals containing proteins), food availability, food shortage, mean Kcalories intake per capita, food share, monthly expenditure on food.

^††Results from the Goldberg review¹⁶ are synthesised in the text.

^‡‡Indicators include mean total number of consultations per month (overall and by specific health care provider), health care expenditure, antenatal care, infacility births, skilled birth attendance, post partum check-ups, preventive health care visits, child growth monitoring visits, HIV testing and results collection, exclusion from health care, mean case notification rates, proportion of extra-pulmonary cases detected, diagnostic delay.

^§§In this intervention, respondents in rural Malawi were offered free door-to-door HIV testing and were randomly given US$0–US$3 redeemable upon obtaining HIV testing results after 2–4 months from sample collection.

^¶¶As TB is an adult disease, we focused on adult malnutrition. Indicators include prevalence of acute malnutrition measured by body mass index (BMI) and mid-upper arm circumference (MUAC).

RMP = Rural Maintenance Programme; CCT = conditional cash transfer; H-QEXP = Household–Quasi-experimental design; BF = Bolsa Familia; PFA = Programa Familias en Acción; C-QEXP = Community-Quasi-experimental design; BDH= Bono de Desarollo Humano; CRT = Community Randomised Trial; PRAF = Programa de Asignación Familiar; PROGRESA = Programa de Educación, Salud y Alimentación; RPS = Red de Proteción Social; PSNP = Productive Safety Net Program; SCT = Social Cash Transfer; DECT = Dowa Emergency Cash Transfer Project; CSS = Cross Sectional Survey; CFRP/TUP = Challenge the Frontiers of Poverty Reduction/Targeting the Ultra Poor; RDP = Rural Development Programme; IGVGD = Income Generation for Vulnerable Group Development; IMAGE = Intervention with Microcredit for AIDS and Gender Equity; JSY = Janani Suraksha Yojana; PATH = Programme for Advancement Through Health and Education; RT = Randomised Trial; SDIP = Safe Delivery Incentive Programme; ICCDR,B = International Centre for Diarrhoeal Research, Bangladesh; CHC= Cambodia Health Committee; SHG = Self-Help Group.

With the exception of one microfinance intervention in Cambodia,¹⁷ none of the interventions we reviewed targeted individuals with any kind of health conditions. For both microfinance and cash transfer interventions, the selection of the target population was based mainly on socio-economic criteria, the specificity of which varied depending on the intervention objectives and resources available.

Four of the 15 cash transfer programmes were unconditional, and these were all reported from African countries, including Ethiopia,^18,19 Zambia²⁰ and Malawi^21,22 (Table 2). The only African study of a conditional programme was from Malawi, in which conditionality was based on HIV testing and the collection of HIV results.²³

Virtually all conditional cash programmes from Latin America required child school enrolment, the attendance of nutritional education sessions and health care uptake for children (<5 years of age) and pregnant/lactating women; adult health conditionalities were less common, and included generic health care,^24,25 in-facility medically attended delivery,^26,27 HIV testing and result recollection.²⁸

Apart from the Dowa Emergency Cash Transfers (DECT) intervention in Malawi,²¹ cash transfer schemes all employed complex impact evaluation designs, including community randomised trials and quasi-experimental studies (Table 2). Cross-sectional surveys were more commonly employed in microfinance interventions. The impact data refer to a follow up period of between 1 and 3 years of observation in most interventions (Table 2). Only in the microfinance intervention in Cambodia was the impact of the intervention measured against TB indicators.¹⁷

Impact on poverty and household socio-economic position

Evidence was gathered from 11 cash transfer and four microfinance interventions (Table 2).

In Nicaragua, Red de Protección Social beneficiary households found a significant net increase in both the total and per capita annual consumption (Table 3).²⁹ A net increase in the overall monthly household expenditure was observed among the beneficiary households of PROGRESA in Mexico²⁴ and the Programa Familias en Acción in Colombia,³² whereas Rural Maintenance Programme (RMP) beneficiaries in Bangladesh experienced a 31% increase in their monthly per capita expenditure (Table 3).³⁴ In Malawi, Mchinji households showed a significantly higher annual expenditure, a higher income from all sources,²² as well as a higher number of assets owned compared to the control households.²² These results were also observed in Zambia (Table 3).^20,37 The proportion of households living below the poverty line significantly decreased in Nicaragua,²⁹ Colombia³² and Bangladesh³⁴ (Table 3). No effect was detected among cash transfer participants in Honduras,³⁸ Ecuador,³⁸ Brazil³³ or Ethiopia¹⁸ (Table 2).

Table 3.

Positive impact findings of conditional and social cash transfer: a quantitative summary

Programme, country, reference	Household socio-economic position		Food security/household food consumption patterns
	Income/expenditure	Poverty level	Food security	Food expenditure	Food quantity	Food quality	Health care seeking
PROGRESA,  Mexico24,30	+ 13% median monthly  total expenditure†			+ 11% monthly median  increase/person‡	+ 7.8% mean Kcal/day/  person*†	+30% monthly median  value of animal  products consumed/  person‡ +16.3% monthly median  value of vegetables &  fruit consumed/person‡	+2 visits/day/public clinic on  average*† 78 quality score of health  procedures received vs. 72  among non-beneficiaries* 8% increase in prenatal visits*†
PRAF,  Honduras31							+ 18.7% antenatal care visits*† + 20.2% child care visits*†
RPS, Nicaragua29	+US$322 annual total  household expenditure*† + US$77 annual total per  capita expenditure*†	22% points  reduction* †		+ US$68 annual per  capita* † +4.7% food share*†		+2.2% food share on  meat*† +2.4% on fats*†  (compared to baseline)	+16% child care visits in 2001  vs. 2000* +8.4% child care visits in 2002  vs. 2000
PFA, Colombia32	+19.5% and +9.3%  average monthly total  expenditure in rural and  urban beneficiaries*†	6% points reduction  in both the rural  and urban areas		+US$27 and +US$15  monthly expenditure  per rural and urban  household*†			+ 23% and +33% children  (respectively 24 and 24–  48 months of age) with an  up-to-date schedule of  preventive health care visits*
B F, Brazil33				+23.2 Brazilian Real  among intervention  households
PATH, Jamaica25							38% increase in health care visits  for children (0-6 years)
RMP,  Bangladesh34	+ 31.4% average monthly  per capita expenditure*†	16% reduction*†		+113 Taka/person/  month† among  benefi ciary households  compared to the  control group	+ 271 Kcal/day/person*†
JSY, India§27							+11% net increase in the  frequency of women reporting  three antenatal care visits*† +43.5% net increase in births in  health facilities*† +36.2% net increase in skilled  birth attendance*†
SDIP, Nepal35							+13% increase in skilled birth  attendance −5% decrease in delivery at  home +24% increase in delivery in  governmental institutions¶
SCT, Zambia  SCT‡20,36				+64% increase among  beneficiary households  compared to the  control group	13% households with  1 meal/day vs.  19% at baseline 35% households still hungry  after meal vs. 56% at  baseline	Average weekly fat  consumption rose from  0.7 days/week to  almost 2 days/week Percentage of households  consuming proteins  7 days/week increased  from 23% to 35%
MCHINGJI,  Malawi#22	MK15 103 net annual  non-food expenditure  increase among  intervention households  compared to controls Intervention households  reporting higher total  annual income  compared to control  households (MK27 079  vs. MK3528)*		88% of the intervention  households had food  store vs. 57% of  control households*	MK3125 net increase in  monthly food  expenditure among  intervention households  compared to controls*	93% of intervention house-  holds reported food  consumption improved  compared to 11% of  control households* 37% of control households  reporting hunger after meal  compared to 7.5% of  intervention households* Mean number of days without  enough to eat in the past  month: 1.2 (intervention  households) vs. 5.2  (comparison households)* 44% of intervention  households had three  meals/day compared to  8% of control households*	8.1 food groups on  average consumed by  intervention households  vs. 4.9 in the control  households* Intervention households  reported to have  meat, chicken and fish  2.1 days/week vs.  0.3/week in the control  households*	Beneficiary households more  likely to get health care when  members were sick compared  to control households (84%  vs. 10%)* 63% of control households  reporting spending nothing on  health care for adult illness  compared to 25% of  intervention households*
DECT, Malawi21				64% of cash transfer #x2003;spent on food	Average number of meals rose #x2003;from 1.5 (2006) to 2.4 #x2003;(2007) and respectively #x2003;from 1.4 to 2.3 in male- #x2003;headed households and #x2003;from 1.2 to 2.5 in female- #x2003;headed households	Number of different food #x2003;groups from 2.5 (2006) #x2003;to 4.0 and 3.6, #x2003;respectively, in male- #x2003;headed households #x2003;and female-headed #x2003;households (2007)
Malawi23							27% increase in HIV testing #x2003;43% increase in VCT attendance #x2003;9.1% increase in VCT attendance #x2003;for every US$ of incentive

^*Statistically significant result.

^†Result adjusted for confounding factors.

^‡In Zambia, it was also documented that the average debt of beneficiary households dropped from approximately 13 000 Zambian Kwacha (USD 2.6) to 8000 (USD 1.6) and that the average number of assets per households rose from four to five.²⁰

^§In the JSY programme in India the net impact has been estimated through three different methods. In the case of antenatal care visits, the net impact was equal to approximately 11% for all the three methods; in the case of in-facility birth the net impact was equal to 43.5%, 43.9% and 49.2%, depending on the method used; in the case of skilled birth attendance the net impact was equal to 36.6%, 36.2% and 39.3%, depending on the method used. For each indicator we have reported the lower estimate.²⁷

^¶The impact evaluation in Nepal was conducted through propensity score matching and time series analysis. Results in the table refer to the propensity score matching analysis. The time series analysis showed that the implementation of SDIP in Nepal had an immediate positive impact on the utilisation of institutional deliveries in governmental health facilities; however, the long-term effect is negative, with the SDIP effect disappearing in just 3 years from implementation. This has been attributed to the fact that in the hill district (accounting for most of the national deliveries) the implementation of SDIP was followed by a dramatic growth in the utilisation of home delivery care. Two possible explanations include the inefficient implementation in the programme in the first two years and the payment delays, which may have reduced programme’s credibility among women.³⁵

^#In the Mchiji programme (Malawi) it was also documented that after one year of intervention more intervention households owned every type of asset and some households owned multiple assets per category.²²

PRAF = Programa de Asignación Familiar; RPS = Red de Proteción Social; PFA= Programa Familias en Acción; BF = Bolsa Familia; PATH = Programme for Advancement Through Health and Education; RMP = Rural Maintenance Programme; JSY = Janani Suraksha Yojana; SDIP = Safe Delivery Incentive Programme; SCT= Social Cash Transfer; MK= Malawi Kwacha; DECT = Dowa Emergency Cash Transfer Project; HIV = human immunodeficiency virus; VCT= voluntary counselling and testing.

Microfinance appeared to significantly reduce extreme poverty, as demonstrated by the Challenge the Frontiers of Poverty Reduction/Targeting the Ultra Poor (CFRP/TUP) and the Rural Development Programme (RDP) interventions in Bangladesh^39,40 (Tables 3 and 4). The percentage of Income Generation for Vulnerable Group Development (IGVGD) households earning more than 300 Taka rose from 7% (1994) to 64% (1996); however, it dropped to 31% 3 years after the intervention.⁴⁷ Additional evidence from Goldberg’s review showed the consistently positive effect of microcredit on household poverty and socio-economic position.¹⁶ The Intervention with Microfinance for AIDS (acquired immunedeficiency syndrome) and Gender Equity (IMAGE) study in South Africa appeared to somewhat improve the economic well-being of beneficiary households (Table 4).⁴²

Table 4.

Positive impact findings of microfinance interventions: a quantitative summary

Programme, country, reference	Household socio-economic position		Food security/household food consumption patterns
	Income/expenditure	Poverty level	Food security	Food expenditure	Food quantity	Food quality	Health care seeking
IMAGE, South Africa  41,42	65% of IMAGE households able to  spend >200 South African Rand  on food and clothing vs. 54% of  the control group†						IMAGE clients: OR = 1.6,  95%CI 1.1–2.6 for HIV testing*†
CFRP/TUP,  Bangladesh43-45		30% reduction*	20% of households living in  chronic food deficiency  compared to 60% at baseline 14.9% of households couldn’t eat whole day vs. 62.1% at baseline	+13.5 Taka/person/day vs. 8.7 at baseline*	85 g/person/day  animal food vs.  22 at baseline* 1019 g/person/day vs.  706 at baseline* 2138 Kcal/person/day  vs. 1750* at baseline 3.2% of energy from  animal source vs.  1.3% at baseline*	5.6 different food  items/day vs. 3.7*	−7.2% self-care visits*† +9.2% formal care visits†
RDP, Bangladesh46		25% of RDP house-  holds moved out of  extreme poverty* 8% of RDP households  moved into extreme  poverty compared  to 30% of the  control group*
IGVGD,  Bangladesh47	% of households earning >300 Taka  rose from 7 (1994) to 64 (1996) Household monthly income rose from  75 Taka (1994) to 717 Taka (1996)
GRAMEEN BANK,  Bangladesh16				+8% compared to  non-clients in  Grameen Villages +35% compared to  non-clients in non  Grameen Village
SHG, India48							Exclusion from health  care among SHG clients  OR = 0.6 (0.4–0.9)*†
Cambodia Health  Committee17							13% of extra-pulmonary  cases detected through  home-DOTS vs. 4%  through hospital/health  care package* 6 months diagnostic delay  with home-DOTS vs.  30 months HHC*
KAFO JIGINEW,  Mali16			12% of clients in food insecurity vs.  30% of incoming clients* 0.25 months in food insecurity  (clients) vs. 1.2 (incoming clients)

^*Statistically significant result.

^†Result adjusted for confounding factors.

IMAGE = Intervention with Microfinance for AIDS and Gender Equity; OR = odds ratio; CI = confidence interval; HIV = human immunodeficiency virus; CFRP/TUP = Challenge the Frontiers of Poverty Reduction/Targeting the Ultra Poor; RDP = Rural Development Programme; IGVGD= Income Generation for Vulnerable Group Development; SHG = Group.

Impact on food security and household food consumption patterns

Evidence was gathered from nine cash transfer and five microfinance interventions (Table 1), of which respectively eight and three reported a positive impact. An increase in food expenditure was documented in the conditional transfer programmes of Mexico,³⁰ Nicaragua,⁴⁹ Colombia,³² Brazil³³ and Bangladesh³⁴ and both the social cash transfer schemes in Zambia^36,37 and the Mchinji in Malawi²² (Table 3). Beneficiaries of PROGRESA in Mexico and RMP in Bangladesh experienced a significant increase in the average daily energy intake.^30,34 Except for Ethiopia,¹⁸ all the cash transfer schemes in Africa documented a significant increase in the daily number of meals consumed and a reduction in the number of households still hungry after a meal.^20-22 Higher food quality and diversity are also an indicator of food security,⁵⁰ and were both found to be improved in the cash transfer programmes of Mexico,³⁰ Nicaragua,⁴⁹ Zambia,³⁶ and both the DECT and Mchinji cash transfer schemes in Malawi (Table 3).^21,22

Among microfinance interventions, food expenditure increased significantly among CFRP-TUP⁴⁴ and Grameen Bank clients (Tables 2-4),¹⁶ whereas no effect was observed among the IMAGE⁴² and WISDOM beneficiaries (Table 4).⁵¹ CFRP-TUP in Bangladesh³⁹ and the Kafo Jijinew in Mali¹⁶ resulted in a significant reduction of the food insecurity experienced by clients in terms of acute food shortage,^16,39 chronic food deficit³⁹ and duration of food insecurity (Table 4).¹⁶ No effect on food security was observed among the IMAGE⁴² and WISDOM clients⁵¹ (Table 3). CFRP-TUP beneficiaries consumed significantly higher food quantity and quality intake compared to the control group (Table 4).⁴⁴

Impact on health-seeking behaviours and health care access

Evidence was derived from 12 cash transfer and six microfinance interventions (Table 1). In Mexico, PROGRESA resulted in an 18% increase in child and adult health care visits in the intervention areas compared to control areas.²⁴ A significant increase in the number of child-care visits was documented from Honduras,³¹ Nicaragua,²⁹ Colombia⁵² and Jamaica²⁵ (Table 3). Maternal health care also appeared to be improved, in terms of the frequency of antenatal visits,^24,27,31 in-facility delivery,^26,27 skilled birth attendance^26,27 and the quality of services requested and received by women (Table 2).⁵³ Mchinji beneficiaries in Malawi were significantly more likely to access heath care when sick compared to control households (Table 3).²² Unlike Zambia, where no effect was detected,³⁶ Mchinji beneficiaries showed a significant increase in the household expenditure on each adult illness (Table 2).²² In the other Malawi study, monetary incentive resulted in a significant 27% increase in HIV testing and a 43% increase in voluntary counselling and testing attendance (Table 3).

CFRP-TUP microfinance clients experienced a significant increase in the use of formal medical care and health expenditure per ill person (Table 4),⁵⁴ whereas, unexpectedly, the RDP/International Centre for Diarrhoeal Research Bangladesh intervention resulted in a significant increase in the proportion of households using self-care as an elective health care strategy (Table 2).⁵⁵ Self-help group clients in India were significantly less likely than the comparison group to have experienced health exclusion.⁴⁸ The Cambodia Health Committee intervention was associated with significant increases in both extra-pulmonary TB detection and earlier diagnosis compared to its hospital-based counterpart (Table 4).¹⁷ Finally, at follow-up, the IMAGE study showed a significant increase in the voluntary counselling and testing uptake in the intervention group compared to controls (Table 4).⁴¹

Impact on adult malnutrition and HIV

Impact on adult nutritional status, assessed as body mass index (BMI), was documented in three cash transfer schemes in Mexico,⁵⁶ Bangladesh,³⁴ Malawi,²² and two microfinance interventions in Ethiopia⁵¹ and Bangladesh.⁵⁷ However, none of them appeared to have a significant positive impact on this indicator (Table 2). In Mexico, doubling cash transfer was associated with higher BMI among both male and female PROGRESA participants, but it was also associated with a higher prevalence of excess weight and obesity (Table 2).⁵⁶ No significant reduction in HIV incidence was detected in the villages covered by the IMAGE intervention (Table 2).⁴²

DISCUSSION

This review shows a lack of studies on microfinance and cash transfer interventions that specifically address TB or other respiratory infections. While the documented effect on risk factors epidemiologically relevant for TB may be an informative way of inferring the potential impact on TB of such interventions,⁵⁸ this evidence cannot replace actual studies specifically addressing TB. The evidence gathered comes from a relatively small, heterogeneous pool of studies, making the data synthesis methodologically challenging and difficult to interpret. Also, given the relatively short follow-up period in many of the studies, little can be concluded concerning the sustainability of these impact findings. As a result, conclusions on the two main questions posed by this review should be drawn with caution.

Do cash transfer and microfinance interventions have a quantifiable impact on outcomes epidemiologically linked to TB?

Both conditional and unconditional cash transfer programmes have reported influences on risk factors for TB in countries with a major TB burden. In principle, if rolled out in communities with a high TB burden, these interventions have the potential to improve people’s access to TB service access and reduce people’s vulnerability to TB, by improving households’ socio-economic position and food security (in terms of both food availability and food diversity). The importance of the latter effect is supported by at least three historical studies that, despite their ecological design, present fairly convincing evidence that a low-protein diet has a direct effect on TB incidence.⁵⁹ Although less consistently, this review also suggests that microfinance could benefit TB-affected communities by turning the loans received into income-generating activities. This may result in the improvement of household material well-being and also reduce household vulnerability.⁶⁰ As discussed below, there may be significant barriers to successfully designing microfinance programmes for TB-affected families.

Both cash transfer and microfinance interventions had a modest positive impact on adult malnutrition, probably because nutritional status is a function not only of food availability, but also of health status and health care services utilisation.⁶¹ It is also known that lending programmes are unlikely to significantly improve nutritional status unless they are linked to education to promote nutritionally important behavioural changes.⁶¹ This last notion makes the adverse effect observed in the PROGRESA intervention even more surprising, given that participants did receive workshops on nutrition.^56,62

The only study attempting to affect HIV incidence did not show the desired impact. Such results have been attributed to the fact that the microfinance and the gender and HIV training components of the I MAGE study did not directly aim to reach young people,⁶³ but only indirectly, through changes in the household economy and information sharing between IMAGE clients and young people.⁶³ Although both components were important elements of the interventions, such indirect targeting, together with the short follow-up period, may have been insufficient to instigate sexual behavioural changes, and therefore HIV decline, among the highest risk group.⁶³

What are the key design and implementation challenges to consider?

Targeting and conditionality are critical aspects in the design and implementation of cash transfer schemes and microfinance interventions. In the light of TB control they could be read as follows.

Targeting

All the interventions appraised applied some population targeting criteria. Targeting is a complex and expensive procedure, often leading to accuracy, equity, sustainability and even stigma problems; all these factors are likely to become even more relevant when adopting either microfinance or cash transfer for TB control. It is unclear, for example, whether these interventions are best targeted at TB-affected families or very poor households. While this choice largely depends on the objective we want to achieve (e.g., TB prevention among the very poor or TB care support in TB-affected families), both options can be challenging.

In the case of microfinance, some experiences with people affected by HIV/AIDS suggest that targeting vulnerable populations (such as TB-affected families) with microfinance may result in poor loan management, inadequate use of the loan for the intended purposes, and weak repayment systems.^64,65 This view is somewhat contradicted by the community-based programme conducted in Cambodia, where loan repayment rates approached 100% among TB-affected families benefiting from participation in village banks.¹⁷ Targeting TB-affected families with cash transfers may be similarly challenging from an ethical perspective, as unaffected families may be equally in need of assistance.⁶⁶

Targeting very poor households at high risk of TB is also controversial. Microfinance institutions are generally reluctant to enrol very poor households due to their uncertain repayment capacity.⁶⁷ It should be noted that there is a strong norm within the non-governmental microfinance sector to work towards a model of financial sustainability. High levels of loan repayment are essential, and this is of course quite different from cash transfer programmes. The IGVGD and CFRP-TUP microfinance interventions demonstrated that it is possible to bring even the most destitute households to a position where they can successfully access microfinance services.^40,47 Nonetheless, when programmes were appraised in more detail, the reason for programme success did not appear to be principally due to microfinance. Rather, the main incentive to participate in IGVGD appeared to be receipt of supplementary food, whereas microfinance was probably the least successful component of the intervention, with an 8% repayment default rate among the clients.⁴⁶ It is also noteworthy that the most successful microfinance programmes included in this review involved the creative and flexible collaboration between microfinance and safety net programmes (e.g., all the BRAC programmes)⁴⁷ or training/education components (e.g., the IMAGE intervention).⁴² In conclusion, the adoption of microfinance for TB control purposes is possible; however, to be effective, additional health, nutritional and educational support may be necessary.

Conditionality

The experiences from Honduras, Ecuador and Columbia demonstrated that successful conditional cash transfer programmes can be implemented in very poor settings; however, the scaling up of conditional cash transfer programmes in high TB burden countries, often characterised by weak health systems and administrative capacities, may be challenging.^68,69 Conditioning the cash transfer, for example, on TB care access and treatment adherence of TB-affected families may not only be impractical, but also unethical if TB services and management capacities are inadequate. In these contexts, while waiting for health systems strengthening, it may be more appropriate to test conditionality on a small scale and only under very specific TB control-related behaviour. Alternatively, conditionality could be completely removed and cash transfer used to reduce TB vulnerability by addressing household food insecurity and material living conditions.

Whereas the evidence provided from the social cash transfer schemes implemented in Zambia and Malawi^20-22 seems to demonstrate that conditionality may not be essential to achieve an impact on risk factors increasing TB vulnerability, it is unclear whether changes in TB care-seeking behaviours can be achieved without conditionality, and under which conditions conditionality should be applied to maximise impact.

CONCLUSIONS

The global TB control community is in need of innovative strategies to move towards TB elimination. The experiences reviewed here support the hypothesis that synergies between social protection interventions and TB control programmes could be effective. Nonetheless, several questions remain unanswered, most importantly: how can these partnerships be most cost-effectively implemented, and what would be their actual impact on TB outcomes? While funding to address these important research questions materialises, a preliminary and relatively inexpensive way to move forward would be to undertake proper studies specifically exploring what obstacles limit their integration with TB control programmes and, through the retrospective analysis of existing data, assess any unintentional impact these interventions may already have had on local TB trends.

Although policy recommendations are not possible at this stage, we hope that the evidence and critical considerations here reported will encourage a new generation of socio-economic interventions addressing key TB risk factors and supporting the existing biomedical strategy. Such studies can significantly advance our understanding of the social determinants of TB and inform broader and more effective TB control policies for the future.