Influence of conditional cash transfer program on prenatal care and nutrition during pregnancy: NISAMI cohort study

ABSTRACT BACKGROUND: There are few studies on the influence of a cash transfer program on nutritional outcomes from pregnancy. OBJECTIVES: To analyze how a Brazilian conditional cash transfer program (Bolsa Familia Program, BFP) was associated with changes in body mass index (BMI) and food consumption among pregnant women. DESIGN AND SETTING: Cohort study on 250 pregnant women (≥ 18 years of age) in Brazilian prenatal services. METHODS: A food frequency questionnaire was used to evaluate dietary intake. Weight was measured in each gestational trimester. Generalized estimation equations and structural equation modeling were used for statistical analyses. Correlations were analyzed using standardized coefficients (SCs). RESULTS: Women benefitting from the BFP were of greater age and had lower education. The BFP exerted a direct negative effect on the pregnant women’s consumption choices regarding refined grains, regional foods, vegetable oil, sausages, salted meats and snacks (SC = -0.10) and on maternal BMI (SC = -0.12). Among the intermediate variables, we observed that the time elapsed since pregnancy and the month of prenatal onset had direct negative effects; and that the number of visits to doctors, family income and number of years of education had direct positive effects. CONCLUSIONS: Beneficiaries were less likely to increase their BMI outside of the recommended standards and had a greater tendency to receive prenatal care. Participation in the BFP had a direct negative effect on adherence to unhealthy diets.


INTRODUCTION
Social protection programs for health have grown in popularity around the world, especially for pregnant women within a context of vulnerability to hunger and food insecurity. [1][2][3][4][5] These programs have important effects with regard to improving maternal and child health and nutrition and diminishing child and maternal mortality rates. 6 However, there is an important gap in knowledge regarding interventions through such programs on the diets and gestational weight gains of women living in low and middle-income countries. Thus, epidemiological studies on the repercussions of social protection policies on the health and nutrition of pregnant women are important for evaluating these intervention programs.
There is evidence to suggest that social protection policies such as the Brazilian conditional cash transfer program (Bolsa Familia Program, BFP) attenuate the effects of poverty and ensure the human right to adequate food, thereby promoting food and nutritional security (FNS). 7,8 In order to remain in the BFP, families must comply with requirements within the areas of education and health. With regard to health, these requirements include prenatal monitoring for pregnant beneficiaries, attendance at scheduled visits to doctors and educational activities relating to breastfeeding and adequate and healthy feeding during pregnancy and early childhood. 9 The results from some studies have shown that pregnant women in the BFP begin prenatal care earlier and have a greater number of visits to doctors 6,10 than those who are not program beneficiaries. The results indicate that the BFP constitutes a protective factor for the health of the mother-child binomial. This is due to early attention to improvement of health, thus resulting in adequate and healthy nutrition; and to adoption of preventive measures with regard to risk factors that compromise proper development of gestation, especially in the initial cycles of fetal formation. In addition, beneficiaries of this program have made progress in seeking out nutritional care. In 2012, approximately 165,000 Brazilian pregnant women were monitored by healthcare teams. Among these women, 99% were up-to-date with their prenatal care and 80% had undergone evaluation of their nutritional status. 11 Regarding the destination of the funds provided by the BFP, the results show that most families used the benefit to purchase foodstuffs. This proportion was higher in the northeastern region and among families in situations of greater food and nutritional insecurity (FNI). 12,13 An epidemiological survey conducted in Brazil revealed changes to the population's diet after its integration into the BFP. Information on the influence of the program on pregnant women's food consumption and anthropometric patterns, however, is only just beginning to emerge.
Given the impact of the BFP on pregnant women's health, our key hypothesis was that women who were beneficiaries of the BFP would have greater number of consultations in prenatal services, which would provide these women with a greater amount of nutritional counseling and monitoring from the healthcare teams. Therefore, women who were beneficiaries would achieve a positive impact on their nutrition and the beneficiaries would attain better control of weight gain during pregnancy. In addition, we postulated that the cash transfers would increase pregnant women's food and nutritional security, through their adoption of healthier eating habits during pregnancy.
Thus, the BFP can be considered to be a social protection policy that contributes positively towards ensuring FNS. In addition, it is probable that the BFP can influence these relationships directly or indirectly, or in an intermediary fashion.

OBJECTIVE
The aim of this study was to examine whether the Brazilian conditional cash transfer program was associated with changes to body mass index, food consumption and prenatal care, among pregnant women.

Study design and sample
This was a prospective cohort study using a dynamic popula-

Exclusion and inclusion criteria
The study included women who lived in the municipality's urban area, were aged 18 years or over, had gestational ages of up to 34 weeks at the time of enrollment and were receiving prenatal care through the public healthcare system.   excluding pregnant women from the study. Gestational age was recorded from the first ultrasound, which was performed by the end of the first trimester and documented in the prenatal services.
Thus, 379 pregnant women met these criteria and were invited to participate in the study. They were invited to answer a closed-end questionnaire and to send blood samples to a clinical laboratory in the city after overnight fasting. Among these women, 107 refused to have blood collected and another 13 women refused to answer the food frequency questionnaire (FFQ). Consequently, a total of 259 pregnant women were eligible to participate. There were nine losses during the follow-up.
Among these, six pregnant women did not fully complete the questionnaires and three had miscarriages. After these exclusions, 250 pregnant women were effectively included in the study and monitored for nine months (Figure 1).

Bolsa Familia Program assessment
The Bolsa Familia Program (BFP) is a conditional cash transfer program that was created in October 2003 through Provisional Measure No. 132. It was aimed at poor and extremely poor families and used per capita family income as the inclusion criterion. 16 The exposure variable consisted of receipt of BFP benefits.

Outcome assessments
We adopted two continuous response variables: gestational body mass index (BMI) variation during gestation (first, second and third gestational trimesters); and dietary intake pattern relating to fatty acid, at the time of enrollment.

BMI variation during pregnancy
For maternal weight measurements in the first, second and third gestational trimesters, we used a scale with a capacity of 150 kg and sensitivity of 100 g (Filizola, model 31 mechanical, Brazil). To measure height, a stadiometer was used, with a capacity of 2000 cm and sensitivity of 0.1 cm (Sanny, Brazil).
Anthropometric measurements were made in duplicate. A maximum variation of 0.5 cm was accepted for length measurement, and a maximum variation of 100 g for weight. 17

Dietary intake assessments
To evaluate fatty acid consumption, the semi-quantitative FFQ was used. 18 This instrument assesses 89 dietary sources of these lipids, with 13 possible responses for consumption, ranging from rarely/never to ≥ 3 times per day.
A photograph album of food portions and kitchen utensils was used to assist in making estimates of the portion sizes consumed, from the interviewees' memory. Data on frequencies of consumption of foods and the portions consumed were inserted into a spreadsheet and calculated, using an adaptation from Santana et al. 19 To analyze consumption, we used daily food consumption.
Thus, all time intervals relating to polyunsaturated fatty acid consumption were converted into the daily frequency of polyunsaturated fatty acid consumption. 18 In this, daily food intake frequency was assigned a value of one. For the weekly and monthly time intervals, the mean of the interval was divided by the period of fatty acid consumption: when weekly, it was assigned the value of 7, and when monthly, the value was 30.
The daily frequency of consumption of each food was used to form food groups that had the same nutritional characteristics, namely: milk and dairy products, fish, fruits and vegetables, olive oil, oilseeds and whole grains, refined grains, foodstuffs that were part of regional dishes, vegetable oil, sausages, salted meats and foods belonging to the snack and processed foods group.

Statistical analyses
The power of this sample to detect an association between the in order to evaluate the influence of the BFP on BMI variation during gestation. This strategy was adopted because the model in the SEM did not fit when the variable "gestational weight gain variation over time" was included in the confirmatory equation of the factorial analysis.
Dropout analyses were performed to investigate the presence of selection bias, through comparing the mothers who completed the study with those who were lost or excluded during the monitoring.
The following variables were considered: age and consumption of mono and polyunsaturated fatty acids.
The baseline consumption of polyunsaturated fatty acids gave rise to two latent variables. These two variables are here denominated "Pattern 1" (milk and dairy products, fish, fruits and vegetables, olive oil, legumes, oilseeds and tubers and roots) and "Pattern 2" (refined grains, regional foods, vegetable oil, sausages, salted meats and snacks). These patterns were established internally in SEM by using confirmatory factorial analysis and were included in analyses in the continuous form. The composition of the fatty acid consumption pattern and its respective standardized coefficients are presented in Table 1.
The direct, indirect and total effects of the relationships stud- Informed consent was obtained for experimentation with human subjects and the privacy rights of human subjects were observed.

Description of participants
Out of the 259 pregnant women considered for the study, 250 women were included (Figure 1), and these participants con- Information on the daily frequencies of consumption of food groups that were present in the diet of the pregnant beneficiaries and non-beneficiaries of the BFP is set out in Table 2.
The pregnant beneficiaries of the BFP had basic food groups such as milk and dairy products, grains, legumes and oilseeds  (i.e. components of the healthier pattern 1) in their daily consumption more frequently.

Program influence on fatty acid dietary intake patterns and anthropometric outcomes
The composition of the fatty acid consumption pattern and its respective standardized coefficients are shown in Table 3.
Pattern 1 consisted of the following food groups: milk and dairy products, fish, fruits and vegetables, olive oil, legumes, oilseeds and whole grains. The root and tuber food groups had the greatest contribution towards formation of this construct and the olive oil and fish groups contributed to a lesser extent. Pattern 2 consisted of refined grains, foodstuffs that were part of regional dishes, vegetable oil, sausages, salted meats and foods belonging to the snack and processed foods group. The food groups that contributed the most notably to this pattern were salted meats, processed foods and those that make up snacks.
The Bolsa Familia Program had a direct negative effect on consumption pattern 2 (SC = -0.10; P = 0.034) and on BMI during pregnancy (SC = -0.12; P = 0.001). This indicated that pregnant women who were beneficiaries of the program had lower adherence to pattern 2 fatty acid dietary intake (refined grains, caruru, vatapá, vegetable oil, sausages, salted meats and snacks) and presented lower BMI during the gestational cycle ( Table 3).
Among the other variables evaluated, there were direct negative effects from age (SC = -0. 16 Table 3).

DISCUSSION
This study was one of the first in Brazil to investigate the influ-

ence of pregnant women's participation in the Bolsa Familia
Program on BMI during pregnancy. The results indicate that pregnant beneficiaries of the BFP had lower adherence to food In this context, the results suggest that, as a social policy, the Bolsa Familia Program exerts a protective effect on maternal nutritional health, through increasing access to and consumption of the traditional basic foodstuffs within a healthy diet among Brazilian families, consisting of milk and dairy products, beans, meat, eggs and grains. These food groups have an outstanding physiological function within the development of adequate gestation and maintenance of women's weight throughout pregnancy. It is also worth noting that the results from this study can be interpreted in the light of proposals for social protection programs (PTCRs).
These programs focus not only on direct cash transfers to families in order to alleviate poverty over the short term, but also on requirements that encourage beneficiaries to access healthcare and educational services. 6,16 Population-based studies have shown that PTCRs improve the economic conditions of poor households, which in turn promotes greater access to food and contributes towards ensuring these households' FNS. 15 Accordingly, the results from this study substantiate the direct relationship between the protective aspect of cash transfer programs and the underprivileged population's health and nutrition.
Among the direct effects of these programs are those relating to the gestational cycle, which ensure that pregnant women have access both to food and to prenatal and postpartum consultations, in addition to ensuring their participation in educational interventions relating to nutrition and health within the public healthcare network. 7,8,16 In a qualitative and quantitative study carried out in municipalities of the state of Bahia, it was found that the Bolsa Familia Program was one of the programs within the National Food and Nutrition Policy that had the greatest coverage in the municipalities evaluated. 22 There was greater monitoring of its requirements, specifically in the area of healthcare, in municipalities where the Family Health Strategy covered more than 70% of the population.
This showed that dialogue between these two public health programs favored interaction between positive health and nutritional actions in the population. 22 Thus, adherence to the population's traditional diet and reduc-  Another limitation related to non-inclusion of rural pregnant women in the study sample. Inclusion of rural pregnant women in the sample should be considered in the future, since groups may have different access to healthcare services according to where they live. This can also occur with regard to food consumption. In studies in other countries, with different epidemiological contexts, it has been reported that pregnant women have differing health and nutrition vulnerabilities according to their area of residence. [33][34][35] In our study, however, these differences may not have been significant, given that Family Health Strategy units that serve the population through health monitoring programs, including prenatal care, also exist in rural areas of the municipality. Regarding the losses during the follow-up, we can attest that these were small and were unlikely to have introduced any selection bias in the study.
The robust statistical analysis in our study was a strength.

It can thus be concluded that studies that address the Bolsa Familia
Program's impact on pregnant women's diet and nutrition are at an early stage. We also noted that there was a lack of study designs with robust methodology for dealing with primary data. From this perspective, the present investigation has contributed to filling the gap, through a longitudinal study on the influence of the Bolsa Familia Program on pregnant women's diets and on maternal weight during pregnancy.

CONCLUSION
This study showed the important role of health conditionalities in expanding access to prenatal services and nutritional guidance during pregnancy, as well as for monitoring and controlling weight gain during pregnancy. Accordingly, the improvement of pregnant women's parameters may have resulted from the Bolsa Familia Program's impact on families' economic conditions, through increasing the population's access to food staples and providing orientation sessions for improving the beneficiary population's health and nutrition.
This study advances the understanding of the positive influence of a social protection policy on the food and nutrition of specific populations, such as the group of pregnant women in urban areas.
However, further studies are needed, especially among pregnant women in rural areas, since these women may have different access to healthcare services.