Conditional cash transfer (CCT) programmes seek to reduce poverty in the short term and to break intergenerational transmission of poverty( Reference Barham and Maluccio 1 – Reference Gaarder, Glassman and Todd 3 ) by requiring parents to meet certain conditions related to health and/or educational components( Reference Attanasio, Oppedisano and Vera-Hernández 4 , Reference Leroy, Ruel and Verhofstadt 5 ). CCT programmes often require that mothers fulfil a schedule of regular primary health-care visits for preschool children, such as adherence to vaccination, growth monitoring and attendance to informative sessions, and they require that children regularly attend school( Reference Handa and Davis 6 – Reference Minujin, Davidziuk and Delamonica 9 ).
Studies worldwide have demonstrated large impacts of CCT programmes on child education, health and nutritional outcomes directly associated with pre-specified conditions (further referred to as ‘conditionalities’)( Reference Castineira, Nunes and Rungo 2 , Reference Leroy, Ruel and Verhofstadt 5 , Reference Bourguignon, Ferreira and Leite 10 , Reference Gertler 11 ). For example, the Mexican Oportunidades programme increased the number of growth-monitoring visits by 60 % in rural areas( Reference Gertler 11 ) and by 52 % in urban areas( Reference Leroy, Ruel and Verhofstadt 5 ), and CCT programmes in Colombia( Reference Attanasio, Benavides and Borda 12 , Reference Attanasio, Gómez and Heredia 13 ) and Honduras( Reference Morris, Flores and Olinto 14 ) have been shown to increase adherence to child immunization schemes, all of which may translate into improvements in health( Reference Gertler and Boyce 15 ). Despite the wide range of evidence on child health, the impact of CCT programmes on well-known ‘determinants’ of child health has been less well explored or evidence has been mixed or inconsistent( Reference Bourguignon, Ferreira and Leite 10 , Reference Barber and Gertler 16 – Reference Skoufias and Di Maro 21 ). In particular, it is not well established whether CCT influence only the use of health and education services associated with programme conditionalities, or whether CCT programmes have wider impacts on behaviours, attitudes and social factors – determinants that may contribute to better child health( Reference Schubert and Slater 7 ). For example, most CCT programmes transfer money to the mother in the household and through this mechanism they may increase women’s decision-making power in relation to child health and well-being. CCT programmes may also influence mothers’ employment decisions, and improve knowledge and awareness of caregiving practices. So far evidence of effects of CCT on these behaviours is mixed and inconsistent( Reference Bourguignon, Ferreira and Leite 10 , Reference Barber and Gertler 16 – Reference Skoufias and Di Maro 21 ).
Familias en Accion (‘Families in Action’; FA) is the CCT programme in Colombia and includes both a health and an educational component. The health component provides cash to mothers of poor households on the condition that children younger than 7 years regularly attend growth and development check-ups as well as vaccination programmes, and mothers attend educational workshops on nutrition, hygiene and contraception. For the education component, children aged 7–17 years must also regularly attend school for at least 80 % of the school year. Families with children aged 7–17 years receive about 14 000 Colombian pesos (about $US 5·5) for every child attending primary school and 28 000 Colombian pesos (about $US 11) for every child attending secondary school( Reference Forde, Chandola and Garcia 22 ). All transfers are delivered to the mother in the household. Early evidence from FA suggested that the programme was associated with a 16·5 % increase in the purchase of protein-rich foods, as well as with increased height-for-age (Z-score increase of 0·16) among the youngest and poorest children. In addition, the programme reduced symptoms of diarrhoea and increased rates of DPT (diphtheria, pertussis and tetanus) vaccination( Reference Attanasio, Gómez and Heredia 13 , Reference Attanasio and Mesnard 23 – Reference Attanasio, Fernández and Fitzsimons 26 ). Yet, there is limited evidence on the impact of the programme on other determinants of child health not directly associated with the specific behavioural conditions, such as mother’s employment, empowerment and knowledge of caregiving practices.
In the present study we estimated the impact of the FA CCT programme in Colombia on a range of child health determinants including: (i) use of preventive health services; (ii) food consumption and diet diversity; (iii) knowledge, attitudes and practices about child caring; (iv) maternal employment; and (v) women’s empowerment. We focused on families with children younger than 7 years, as this represents a critical period of development that may have implications for outcomes later in life( Reference Campbell, Conti and Heckman 27 ). Initial reports suggested that the FA programme had stronger effects in rural areas, where access to health facilities is generally lower than in urban settings( Reference Attanasio, Gómez and Heredia 13 , Reference Attanasio and Mesnard 23 ). Therefore, we also examined whether the CCT programme in Colombia had different effects across rural and urban areas.
Methods
The Familias en Accion programme
Eligibility for the CCT programme is determined based on a scoring system known as the System for Identifying and Selecting Beneficiaries (SISBEN), a survey of low-income households used to derive a poverty score that ranges from 0 to 100 and enables targeting social welfare programmes( Reference Forde, Chandola and Garcia 22 , Reference Paes-Sousa, Regalia and Stampini 28 ). Implementation of the programme includes operational units at the national, departmental and municipality levels. The programme is managed and implemented by a national coordinating unit, but regional coordinating units in each department manage the programme and liaise with the national and municipal government. Programme monitoring is a joint responsibility of departmental and municipality governments and includes a comprehensive monitoring system that follows families through the various stages of programme implementation, such as: (i) beneficiaries’ registration and status; (ii) compliance with programme conditionalities; (iii) payment of transfers; and (iv) complaints and case management( Reference Paes-Sousa, Regalia and Stampini 28 ). Municipal liaison offices verify that mothers meet the conditions. If a mother fails to meet the requirements three consecutive times, she could be dismissed from the programme.
For the present study, we used data from the evaluation of the FA programme( Reference Attanasio, Benavides and Borda 12 ). The evaluation of the programme was carried out by the Institute for Fiscal Studies, an independent research institute in London, UK, using a matched-control design( Reference Attanasio, Benavides and Borda 12 ). At the beginning, the programme was targeted geographically. Of the 1060 municipalities in Colombia, 622 qualified for the programme based on their fulfilment of several conditions, which required municipalities to have: (i) a population smaller than 100 000 inhabitants; (ii) the health and education infrastructure to guarantee programme implementation; (iii) a bank to enable cash transfers; and (iv) up-to-date census, welfare and service infrastructure data. The evaluation is based on a detailed survey carried out in 122 municipalities with a baseline assessment in 2002 and follow-up assessments in 2003 and 2005–06. For the survey, a stratified and probabilistic sample of fifty-seven treatment municipalities representative of the 622 eligible municipalities was selected. These municipalities were matched to sixty-five control municipalities, based on similarities to treatment municipalities in observed characteristics. In practice, except for the requirement to have a bank, control municipalities were comparable to treatment municipalities in all other assessed dimensions. Further details of the evaluation are available elsewhere( Reference Attanasio, Benavides and Borda 12 ).
Assessments were carried out through a household survey among participants in both control and treatment municipalities. In addition, data on municipal services supply were collected based on a survey among health centres and schools. In 2002, baseline assessments were scheduled to take place before the programme started, but due to political pressure the programme started before in twenty-six out of the fifty-seven treatment municipalities. Because no baseline data were available for them, we focused on the thirty-one treatment municipalities with assessments prior to programme implementation. Children under 7 years of age in treatment (n 2394) and control (n 3197) groups were randomly selected. A first follow-up assessment was carried out in 2003 and included 2010 treated children and 2606 control children who had previously been measured in 2002. A second follow-up assessment was carried out between 2005 and 2006, and included 1450 and 1851 children in both control and treatment areas who were evaluated in 2002 and 2003. They were considered as the final sample for the current analysis. Children lost to follow-up were slightly older (5·1 v. 4·4 years, P<0·0001) and their families reported lower use of health services and attendance at educational workshops. However, children and mothers lost to follow-up were similar to those who remained in the sample in terms of sex, maternal and municipality characteristics investigated (see online supplementary material, Table S1).
We examine the impact of the CCT programme on determinants of child health. Figure 1 summarizes the outcomes assessed and how we hypothesize they may relate to programme exposure and child health. We examined impacts not only on outcomes that were associated with programme conditionalities such as use of health-care services and workshop attendance, but also on outcomes such as women’s empowerment, which were not directly required in order to receive the cash transfers but may have changed as a result of the CCT programme.
Diagram to identify determinants that might affect child health and nutrition through the FA programme. This diagram shows the impact of the CCT programme on multiple outcomes ranging from the conditionalities required to receive the transfers (shaded in grey) to more distal determinants (continuous line). It also includes potential confounders at the individual, household and municipality level (dashed borderline). (FA, Familias en Accion; CCT, conditional cash transfer)
Outcomes
Use of health-care services
Use of health-care services was measured by asking mothers whether they had visited child health-care services in the last year (yes/no) for attending the growth and development check-ups for children younger than 10 years, which was one of the conditions to receive the cash transfers. The number of check-ups was also collected. This information was available in the baseline and the first follow-up household surveys.
Workshop attendance
Mothers were asked about their attendance (yes/no) at specific educational workshops on nutrition, diarrhoea control, prenatal care and management of acute respiratory infections in the six months prior to the survey. This information was collected for each workshop in each follow-up; information about the nutrition workshop was available only at baseline and the first follow-up.
Child dietary intake
Child dietary intake was collected by asking mothers if their children consumed certain foods (yes/no) and how often they did during the seven days preceding the survey( Reference Attanasio, Benavides and Borda 12 ). The food items were classified into eight food groups which included: (i) cereals, roots and tubers; (ii) fruits; (iii) vegetables; (iv) legumes and nuts; (v) meat and poultry; (vi) fish; (vii) dairy; and (viii) eggs. The evaluation of dietary adequacy was assessed by creating a simple dietary diversity score (DDS), which is widely recognized as a key indicator of diet quality( Reference Oldewage-Theron and Kruger 29 , Reference Steyn, Nel and Nantel 30 ). We added the total number of days each of the eight food groups was consumed in the previous week. The sum of the number of days could range from 0 to 56. Based on previous application( Reference Rah, Akhter and Semba 31 ), we created tertiles of DDS to classify children into low, average and high diversity, using the following cut-offs (low=0–15; average=16–21; high=22–56).
Women’s empowerment
One of the expectations of CCT programmes is that by putting resources in the hands of poor women, the programme will promote gender equality within the household, resulting in large development pay-offs. To assess this, mothers were specifically asked to report who decides when to take a child to the doctor if sick and how much is spent on food. For each decision there were four possible answers: (i) only the father decides; (ii) only the mother decides; (iii) both decide; and (iv) other members of the household decide. We collapsed this information into two categories: (i) mother or both decide; and (ii) father or others decide.
Maternal employment
Women were asked to report in which activity they spent most of their time during the previous week. We reclassified responses so as to identify two groups: (i) women in the labour market, which included women currently at work, women in the labour force but temporarily not working, and women in the labour force but currently unemployed and looking for work; and (ii) women who were out of the labour market, including women who were retired, women who were studying, homemakers and the disabled.
Women’s knowledge, attitudes and practices
Mothers were asked to provide information about their knowledge, attitudes and practices concerning (i) diarrhoea and fluids replacement and (ii) diarrhoea and food consumption at each follow-up. We dichotomized each answer into a value of 1 if the mother’s answer was correct and 0 otherwise.
Control variables
Covariates at the individual, household and municipality levels were used as control variables. Children’s individual characteristics included age, sex and whether the child was participating in Hogares Comunitarios, a home-based childcare programme for children from poor families. We controlled for maternal characteristics including mother’s educational attainment, marital status and age. Mother’s highest level of education completed was categorized into: (i) no education; (ii) incomplete primary; (iii) completed primary; (iv) incomplete secondary; (v) completed secondary; and (vi) higher education. Covariates also included household size and household income, measured by asking respondents their income from all sources in the past month, including wages, salaries, retirement benefits, help from relatives and rent from property. To account for differences in the number of household members, gross income was equivalized by dividing all household income by the square root of household size( 32 ). In regression models, household income was log-transformed to account for non-linearities. At the municipality level, models included number of inhabitants, level of urbanization (urban/rural), availability of health-care services and geographic location (Central, Caribbean, Pacific or Eastern region).
Ethical approval for the evaluation study was granted by a local institutional ethics committee. Adults provided signed informed consent to participate in the study. Data from the evaluation are made publically available by the Planning Department of the Colombian Government with no identifiable information on survey participants (https://www.dnp.gov.co).
Statistical methods
We used a difference-in-differences (DID) approach, an analytical method that compares changes between baseline and follow-up between treatment and control, rather than differences in post-treatment outcomes only. The DID estimate is based on the difference in outcome in the treatment group before and after treatment minus the difference in outcome in the control group over the same period. The assumption is that the change observed in the control group is a good counterfactual of the change we would have observed in the treatment group if they had not been exposed to the programme, because this removes biases in post-treatment comparisons between the treatment and control group that could result from permanent differences between those groups, as well as biases from comparisons over time in the treatment group that could be the result of trends( Reference Card and Krueger 33 ). This approach has been commonly applied in the evaluation of CCT programmes in different countries( Reference de Brauw and Hoddinott 34 , Reference Gitter and Barham 35 ).
To assess the quality of the DID approach, the ‘common trend assumption’ should be tested. The common trend assumption means that the outcome variable would have evolved in the same way between baseline and follow-up in both treatment and control municipalities had the FA programme not taken place. Although we could not test this assumption because we did not have data for the outcomes in our study prior to enrolment, an indirect test was performed to examine trends prior to programme implementation. We used data sources provided by the national statistics agency (DANE), which collects and harmonizes data on all mortality information (under-5 mortality rate) from all regions. We estimated trends in under-5 mortality rate and urbanization in control and treatment municipalities between 1997 and 2001, before the programme started. Trends in these indicators of health and living conditions were similar prior to the programme (P value >0·05); therefore this provides an indication that the common trend assumption holds (data not shown).
We used linear regression to model continuous outcomes and logistic regression to model dichotomous outcomes. For a continuous outcome variable Y and individual i, we estimated the following model:
$$Y_{{it}} {\equals}\beta _{0} {\plus}\beta _{1} treat_{i} {\plus}\beta _{2} wave_{{it}} {\plus}\beta _{3} treat_{i} {\times}wave_{{it}} {\plus}\beta _{4} X_{i} {\plus}ε_{{it}} ,$$
where wave (β 2)=0 if the period is the baseline assessment (2002) and wave=1 if it is a follow-up assessment (2003 or 2005–06); treat (β 1) is equal to 1 if the individual lives in a treatment municipality and 0 otherwise; and X (β 4) refers to a vector of individual, household and municipality baseline variables. β 3 is an interaction term between treatment and wave, which measures the difference in trends before and after the programme between treatment and control, the DID estimate of interest.
We carried out all analyses separately for urban and rural areas, given prior evidence that effects may differ by level of urbanization( Reference Attanasio, Benavides and Borda 12 , Reference Attanasio, Gómez and Heredia 13 , Reference Attanasio and Mesnard 23 ). All analyses were performed using the statistical software package SAS version 9·3. We incorporated appropriate sample weights to account for differential selection probabilities. We estimated robust standard errors clustered at the municipality level in an intention-to-treat analysis.
Results
At baseline, the children’s mean age was 4·4 years in treatment municipalities and 4·5 years in control municipalities. The use of health services was higher in the treatment than in the control group. In rural areas, mothers participated less in educational workshops in treatment than in control municipalities. In urban areas, children in treatment municipalities consumed fewer vegetables than in control municipalities. Mothers in treatment municipalities had also higher empowerment and knowledge regarding childcare decisions than mothers in control municipalities. In rural areas, mothers in treatment municipalities had lower rates of labour force participation, but when employed worked more hours per week than mothers in control municipalities (Table 1).
Baseline characteristics; Familias en Accion conditional cash transfer programme, Colombia, 2002
Significant P values are shown in bold font.
* ‘Did you use health-care services in the last year?’
† ’Is your child registered in the growth and development programme?’
‡ ’Did you attend a workshop regarding prenatal care, diarrhoea control, management of acute respiratory infections and nutrition in the six months prior to the survey?’
§ ’Did your child consume those food items in the prior week?’
Figure 2 summarizes trends in health-care services use, growth and development check-ups and educational workshops. Between 2002 and 2006, health-care services use increased in the treatment group, while it remained constant in the control group. Enrolment and attendance at the growth and development check-ups increased substantially in the treatment group in the period between 2002 and 2003, while it did not change for children in the control group. Mothers’ attendance at educational workshops generally increased for the treatment group, while it declined in the control group between 2002 and 2006.
Trends in the conditionalities for control and treatment municipalities, Familias en Accion, Colombia, 2002–2006. Trends in (a) health-care services use, (b) frequency of growth and development check-ups, (c) registration in growth and development programme, (d) attendance at diarrhoea workshop, (e) attendance at nutrition workshop, (f) attendance at acute respiratory infection workshop and (g) attendance at prenatal care workshop. 
, 2002 (baseline); 
, 2003 (first follow-up); 
, 2005–06 (second follow-up). Values are percentages with their 95 % confidence interval represented by vertical bars, or means with their standard deviation represented by vertical bars (b)
Figure 3 shows trends in food intake between 2002 and 2006. In the treatment group there was an increase in consumption of almost all food items. By contrast, there were no changes or smaller increases in the consumption of dairy products, meat, vegetables, eggs and legumes for the control group. Fruit consumption declined in both treatment and control groups. These results were confirmed with the use of the dietary diversity score. At baseline, there were no differences in control and treatment groups. However, in 2003 and 2006 there was an increase in dietary diversity in the treatment group, resulting in almost half of the children in the treatment group from rural areas having a high dietary score. In contrast, more than 40 % of children in the control group in rural areas remained at a low diversity score (Fig. 4). Figure 5 shows that women’s empowerment increased in both groups, with a slightly larger increase in the control group. There were no clear changes in labour force participation and working hours in either group.
Trends in child food consumption, Familias en Accion, Colombia, 2002–2006. Trends in (a) dairy products, (b) meat, (c) fish, (d) eggs, (e) legumes, (f) vegetables and (g) fruits. , 2002 (baseline); , 2003 (first follow-up); , 2005–06 (second follow-up). Values are percentages with their 95 % confidence interval represented by vertical bars
Trends in child dietary diversity, Familias en Accion, Colombia, 2002–2006. Trends in dietary diversity score (range 0–56) categorized in tertiles as (a) low (score=0–15), (b) average (score=16–21) and (c) high (score=22–56). , 2002 (baseline); , 2003 (first follow-up); , 2005–06 (second follow-up). Values are percentages with their 95 % confidence interval represented by vertical bars
Trends in women’s empowerment, maternal employment, knowledge, attitudes and practices, Familias en Accion, Colombia, 2002–2006. Trends in (a) mother’s empowerment on food, (b) mother’s empowerment on childcare, (c) maternal employment, (d) mother’s working hours, (e) mother’s knowledge, attitudes and practices on diarrhoea and fluids replacement and (f) mother’s knowledge, attitudes and practices on diarrhoea and food intake. , 2002 (baseline); , 2003 (first follow-up); , 2005–06 (second follow-up). Values are percentages with their 95 % confidence interval represented by vertical bars, or means with their standard deviation represented by vertical bars (d)
Table 2 shows the estimates of DID analysis on the conditionalities. In rural areas the FA programme was associated with an increase in use of health-care preventive services (OR=2·63, 95 % CI 1·31, 5·27). The FA programme increased attendance at the growth and development check-ups (OR=5·09, 95 % CI 2·88, 8·99) as well as the frequency of these check-ups (β=1·36, 95 % CI 0·76, 1·95) in both rural and urban areas. Likewise, the programme was associated with a higher mother’s attendance at educational workshops on diarrhoea (OR=2·36, 95 % CI 1·42, 3·92), prenatal care (OR=2·92, 95 % CI 1·74, 4·89) and acute respiratory infections (OR=2·57, 95 % CI 1·44, 4·60) in rural and urban areas. There was no effect on attendance at nutrition workshops.
Difference-in-differences (DID) estimates on the conditionalities for control and treatment municipalities; Familias en Accion (FA) conditional cash transfer programme, Colombia, 2002–2006
Variables included in the model: age, child’s sex, participation in Hogares Comunitarios (home-based health care), mother’s marital status, mother’s age, mother’s education, household income, level of urbanization, availability of health services, number of inhabitants and region.
Significant results are shown in bold font.
* Values are odds ratios and 95 % confidence intervals.
† Values are regression coefficients (β) and 95 % confidence intervals.
‡ Effect of FA programme in the first follow-up only.
In rural areas, children in treatment municipalities had larger increases in the consumption of meat, eggs and dairy products than children in control municipalities. The FA programme was associated with an increased dietary diversity among children in rural areas (OR=2·13, 95 % CI 1·25, 3·65). In urban areas, children in treatment municipalities experienced larger increases in the consumption of fish, eggs and vegetables than children in control municipalities. Nevertheless, we did not find differences in the dietary diversity between control and treatment groups (OR=1·42, 95 % CI 0·85, 2·37). There was no evidence that the programme had any positive impacts on women’s perceived decision-making power within the household on issues related to child nutrition and care. If anything, in rural areas the programme was associated with a decline in mother’s involvement in decisions regarding childcare (OR=0·90, 95 % CI 0·83, 0·98). There was no evidence of an impact of the programme on labour force participation or working hours (Table 3).
Difference-in-differences (DID) estimates on other determinants for control and treatment municipalities; Familias en Accion (FA) conditional cash transfer programme, Colombia, 2002–2006
Variables included in the model: age, child’s sex, participation in Hogares Comunitarios (home-based health care), mother’s marital status, mother’s age, mother’s education, household income, level of urbanization, availability of health services, number of inhabitants and region.
Significant results are shown in bold font.
* Values are odds ratios and 95 % confidence intervals.
† Values are regression coefficients (β) and confidence intervals.
Discussion
The present study suggests that the FA programme increased the use of preventive health-care services, growth and development check-ups, and mothers’ participation in educational workshops, particularly in rural areas. The programme increased children’s food consumption in both rural and urban areas, as well as dietary diversity among children in rural areas. This is consistent with previous studies in other countries such as Mexico, Nicaragua and Brazil, which have also found stronger effects in rural areas( Reference Lagarde, Haines and Palmer 8 , Reference Barber and Gertler 16 , Reference Gitter and Barham 35 – Reference Adato, Roopnaraine and Becker 38 ). On the other hand, we found no evidence that the programme had significant effects on maternal employment, women’s empowerment, and women’s knowledge, attitudes and practices about caregiving practices.
The FA programme significantly increased the use of preventive health services. A possible explanation is that in the absence of a financial incentive, families prefer non-conventional, alternative medicines. Traditionally, 40 % of Colombians use non-conventional or alternative medicines( Reference Payyappallimana 39 ), while the use of health services is more related with the perception of a serious illness or injury( Reference Garcia-Subirats, Vargas and Mogollón-Pérez 40 ). Similar results have been found in other countries. For example, in El Salvador, CCT beneficiaries reported that they treated their acute episodes of illness at home in the first instance and attended health-care services only if complications arose( Reference Adato, Roopnaraine and Becker 38 ). Although families in urban areas may have easier access to health centres and services than families in rural areas, the programme may not be sufficient to additionally incentivize these families to attend health services.
We found that the programme increased dietary diversity among children in rural areas. To our knowledge, only one study has assessed the effect of CCT programmes on household dietary diversity. That study showed that beneficiary households of cash transfers in Ethiopia had better household dietary diversity scores. However, the latter study was cross-sectional and did not evaluate changes over time( Reference Baye, Retta and Abuye 41 ). Consistent with previous studies( Reference Adato and Bassett 42 , Reference Leroy 43 ), we found that the programme increased consumption of some healthy foods that may be translated into an improvement of diet quality. However, we found a reduced intake of fruits, suggesting that food consumption is largely driven by a direct income effect rather than by improved nutritional knowledge and attitudes.
Our results suggest that the FA programme did not have effects on maternal employment. This is consistent with findings from Mexico’s Oportunidades programme, which did not increase mother’s employment rates but was also not a disincentive to work( Reference Skoufias and Di Maro 21 ). These findings suggest that other structural or cultural factors that are not amenable to intervention through cash transfers may be more important determinants of maternal employment. For example, Colombian women’s labour participation has been associated with lack of access to childcare, local crime rates, lack of public transport services and low economic activity in region of residence( Reference Morales and Cardona-Sosa 44 ).
We found no evidence that the FA programme improved women’s perceptions of their decision-making power related to child health and well-being, as suggested elsewhere( Reference Barber and Gertler 16 ). A possible explanation may be that transfers to mothers increased fear for potential conflict and domestic violence, and male partners may use this mechanism to preserve control over transferred money( Reference Camacho and Rodriguez 18 ). In Zimbabwe, for example, women who participated in a CCT programme reported that they did not feel free to make decisions by themselves due to fears of ‘family disintegration’, and they continued with their traditional gender roles as child caregivers while men maintained control over household decisions and budget( Reference Brady 45 ).
Women living in treatment municipalities were more likely to attend educational workshops, but the latter were not effective in bringing about changes in knowledge and practices regarding child health and nutrition. Evidence from Mexico’s Oportunidades programme has also reported some problems with the educational sessions; for example, doctors informed that although mothers attended the workshops, many of them did not pay attention during the sessions. Likewise, cultural barriers also prevented doctors from discussing sensitive topics such as family planning and pap smear tests( Reference Adato, Coady and Ruel 46 ). This raises questions on the effectiveness of workshops as a way to improve mothers’ knowledge of caring practices.
Limitations
Despite several strengths, some limitations should be considered in our study. First, there was a 40 % loss to follow-up due to the high mobility of participants. Difficulties in re-contacting participants partly reflect the unstable living conditions of a migrating workforce. Nevertheless, we found that children contacted at follow-up did not differ from those lost to follow-up with respect to several key baseline characteristics. Some of the outcomes measures may also have been influenced by biases (e.g. reporting, memory, perception).
A crucial assumption of the DID approach is that a similar trend between treatment and control should be observed if the FA programme had not taken place. Although the common trend assumption could not be tested directly, we found no significant differences in trends for under-5 mortality rate and urbanization rate between control and treatment municipalities before the programme started. While not conclusive, it is reassuring that treatment and control municipalities did not differ prior to treatment in these key outcomes.
Our study was based on a strong research design that aimed to ensure internal validity. However, whether results are externally valid and generalizable to other countries is difficult to assess. We expect our findings to be of relevance to other Latin American countries running similar programmes such as Ecuador and Brazil, but our findings may not be generalizable to countries in Asia where the political and socio-economic context may be substantially different.
Conclusion
The FA programme increased the use of preventive health-care services and attendance at child growth and development check-ups. This suggests that CCT programmes may be efficient in improving child nutrition through growth and development check-ups as well as increasing access to preventive health services. Yet, the programme had no effects on other important determinants of child health such as women’s empowerment, knowledge, attitudes and practices, and women’s employment rates. These findings cast some doubt on the notion that CCT programmes have ‘spillover’ effects in broader determinants of child health not directly associated with programme conditionalities. Our findings highlight the need to develop CCT or other programmes that not only influence behaviours directly associated with pre-specified conditions, but also motivate households to further invest in other important determinants of child health and well-being.
Acknowledgements
Financial support: S.L.-A. was supported by the European Union, Erasmus Mundus Partnerships Programme Erasmus–Columbus (ERACOL) and Fundación para el Futuro de Colombia (COLFUTURO) at Erasmus MC in the Netherlands. M.A. was supported by the European Research Council (ERC) (grant number 263684), the National Institute on Ageing (award numbers R01AG040248 and R01AG037398), and the LIFEPATH project funded by the European Union’s Horizon2020 research and innovation programme under grant agreement 633666. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Conflict of interest: The authors declare that they have no conflicts of interest. Authorship: S.L.-A. and .M.A. were responsible for the study conception, design, analysis and interpretation of the data, as well as the drafting of the article. A.B. and F.J.v.L. intensively revised the manuscript. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects/patients were approved by a local institutional ethics committee. Adults provided signed informed consent to participate in the study. Data from the evaluation are made publically available by the Planning Department of the Colombian Government with no identifiable information on survey participants (https://www.dnp.gov.co).
Supplementary Material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S1368980016000240
