Health literacy status among pregnant women in a Brazilian conditional cash transfer programme : a cross-sectional investigation

Objective: To investigate the overall score and the variables associated with health literacy of pregnant women benefited from a conditional cash transfer programme. Methods: A crosssectional survey was conducted with pregnant women benefited from a Brazilian conditional Research, Society and Development, v. 9, n. 11, e4979119650, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.9650 2 cash transfer programme. We used a validated short health literacy survey developed for research on public health and health promotion. The level of health literacy (high or low) was considered a determining variable. Descriptive, bivariate and multivariate analysis was applied to the data collected. Results: Only the two following variables were associated with health literacy: participation in health education activities and the choice of mothers to seek advice regarding pregnancy. Functional health literacy was higher when pregnant women participated in health education activities. Critical health literacy was higher when pregnant women sought advice from their mothers, friends, and other family members. Conclusion: Health education and a social network of pregnant women should be part of prenatal care.

The conditional cash transfer programme of the Brazilian government, known as Bolsa Família (Family Allowance Programme), has been associated with the improvement of a wide spectrum of health conditions (Alves & Escorel, 2013), such as reduced number of deaths attributed to poverty-related causes and decrease in overall childhood mortality rates (Rasella, et al., 2013).
The HL status may be related to higher equity levels in public health (Sørensen, et al., 2012) and it may help health professionals to promote adequate health education activities (Abel, et al., 2015). Hence, this study aimed to investigate contextual variables associated with the overall score and structural factors of HL in pregnant women benefited from a conditional cash transfer programme.

Methodology
This study was a cross-sectional survey (Celentano & Szklo, 2018)  The study included pregnant women recipients of a cash transfer payment from the Family Allowance Programme, living in upstate Minas Gerais (Human Development Index = 0.789) in the southeast region of Brazil, and whose last known menstruation date was between July and November 2017. This time frame was selected for covering one of the two annual periods when the local programme supervisor should send the consolidated data of pregnant women included in the Family Allowance Programme to the national supervisor. In this period, 128 pregnant women had participated in the programme.
The data were collected between January and May 2018 at the addresses provided by the Family Allowance Programme. Self-applied questionnaires were completed in the presence of a research assistant who aided the pregnant women, from minimal supervision to reading all the questions, when assistance was required.
Health literacy was considered the dependent variable, including overall and structural Development, v. 9, n. 11, e4979119650, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.9650 5 factors scores. Independent variables included the characteristics of pregnant women (gestational and actual age), socioeconomic characteristics (years of education, marital status, and family income), participation in health education activities, search for advice regarding pregnancy, occupational status, and religion. Age, gestational age, and level of education were collected as numerical variables. Other characteristics were recorded as nominal variables.
Health literacy (HL) was assessed with the instrument developed and tested by Abel et al. (2015) to improve the understanding on the distribution of HL in the general population. This instrument measures HL by eight Likert-scale items that assess four structural factors: 1) Health information understanding (or Functional HL-1), 2) Health information search (or Functional HL-2), 3) Health interactivity, and 4) Health critical knowledge. The instrument was translated and validated to Brazilian Portuguese (pHLAT-8) by Quemelo, et al. (2017).
The structural factors of HL were obtained by adding the questions of the pHLAT-8.
Functional health literacy related to understanding information (FHL-1) was obtained by adding questions one and two (maximum score is 10). Functional health literacy related to searching information (FHL-2) was obtained by adding questions three and four (maximum score is 8). Interactive health literacy (IHL) was obtained by adding questions five and six (maximum score is 10). Critical health literacy (CHL) was obtained by adding questions seven and eight (maximum score is 9). Overall, the maximum pHLAT score is 37.
Additionally, another instrument was used to collect contextual characteristics of pregnant women, such as age, gestational age, level of education, income, number of children, marital status, main person advising pregnant women, participation in collective health education activities, people living in the same household, occupational status, and religion.
The statistical analysis was performed with the STATA software, version 12.0 (StataCorp., College Station, TX, USA). Descriptive analyses were conducted considering frequency and measures of central tendency. Numerical variables were dichotomised using the median as cut-off. Originally, the pHLAT-8 did not classify HL as low and high, but our study provided such classification. Hence, we selected the median as cut-off (50th percentile), which was presented in high and low scores.
Bivariate analyses were conducted to test associations between predictors and outcomes using the chi-square test (p < 0.05). Poisson regression with robust variance was used for the multivariate analysis, with the calculation of prevalence ratios (PR) and respective 95% confidence intervals (CI). Variables with p-value < 0.20 in the bivariate analysis were incorporated into the multivariate analysis and those with p-value < 0.05 in the final model were considered significantly associated with the outcome. Research, Society and Development, v. 9, n. 11, e4979119650, 2020 (CC BY 4.

Results
There were 97 pregnant women participating in this study, representing 24.22% of loss, which occurred because either the pregnant women denied participation in the study (2.34%) or their address and phone number from the registration data in the Family Allowance Programme was incorrect or outdated (21.88%). Additional attempts to find the missing pregnant women were performed through the contact with primary health care professionals and neighbours. Despite the efforts, these women were not found. However, the families enrolled in the Family Allowance Programme present very low income and low level of education. Thus, the pregnant women who could not be found should not be different from those participating in the study. Table 1 presents the other contextual characteristics examined in the pregnant women.  The highest mean values observed were for Health Information Understanding and Interacting Health Literacy. However, these structural factors also present greater amplitude between the maximum and minimum values. Table 3 shows the results of the bivariate analyses among all the independent variables analyzed and the structural factors of HL. Table 4 presents results of the multivariate analyses for overall HL, FHL-1, and FHL-2. Research, Society and Development, v. 9, n. 11, e4979119650, 2020 (CC BY 4.  to those who did it. Table 5 presents the results of multivariate analyses for IHL and CHL.  Research, Society and Development, v. 9, n. 11, e4979119650, 2020 (CC BY 4. Pregnant women who sought advice from friends and family members were twice more likely to present an IHL outcome (PR = 2.20; 95%CI:1.26-3.83). Considering a CHL outcome, the mothers were mainly responsible for advising pregnant women (PR = 2.29; 95%CI: 1.45-2.92).

Discussion
This research was designed to investigate the contextual variables associated with the health literacy (HL) of low-income pregnant women recipients of the conditional cash transfer programme of the Brazilian government. Only two variables were associated with HL: participation in health education activities and the choice of mothers to seek advice regarding pregnancy from family and friends. Health education was responsible to improve the understanding and search for information, while advice from mothers, friends, and other family members showed an improvement of interactive and critical HL.
Health education during prenatal care has been reported by several authors as the way to improve maternal, obstetric, and children outcomes (Quemelo, et al., 2017;McFadden, et al., 2017;Silva, Lima & Osório, 2016;Lumbiganon, et al., 2016;Bahri, et al., 2015;Brasington, et al., 2016;Al-Ateeq & Al-Rusaiess, 2015). Pregnant women who received health education during prenatal care presented reduced complications during pregnancy  (Brasington et al., 2016;Al-Ateeq & Al-Rusaiess, 2015), increased duration and exclusivity of breastfeeding, babies with higher birth weight (Silva, Lima & Osório, 2016), and improvement in oral health beliefs and behaviours (Bahri, et al., 2015). The results of the present study help understanding the mechanism that connect health education to improved outcomes in maternal and child health.
Social networks (mothers, friends, and other family members) were associated with improved HL results, which is favourable because mothers prefer to receive information from their peers than from health professionals (Duncanson, Burrows & Collins, 2014). Social networks act as mediators of HL for developing the autonomy of patients, communication with health professionals, and acceptance of potential diseases. This mediation happens because people rely on the health literacy skills of other people to search for, understand, and use health information (Edwards, et al., 2012). Moreover, peer support is sufficient to increase the knowledge of pregnant women and promote healthy breastfeeding practices (Kempenaar and Darwent, 2013).
Traditionally, there is still no cut-off point to determine whether the values obtained for the structural factors of HL are good or not when using the pHLAT-8. However, our results may be compared with validation studies of HLAT-8. Our study population presented lower results in all structural factors than the studies by Abel, et al. (2015), Guo, et al. (2018) and Quemelo, et al., (2017). This result may be explained by the lower level of education of pregnant women investigated, considering there is a direct relationship between HL and levels of education (Duggan, et al., 2014;Castro-Sánchez, et al., 2018;Akbarinejad, Soleymani & Shahrzadi, 2017). It is worth noting that, despite the impact of level of education on HL, improving this parameter alone is not enough to improve HL levels (Lupattelli, et al., 2014).
This lower HL status observed in pregnant women recipients of a conditional cash transfer programme is concerning and consistent with other studies (Lori, et al., 2015;Akbarinejad, Soleymani & Shahrzadi, 2017) observed that more than half of pregnant women analysed presented inadequate HL. Lori, et al. (2015) indicated that pregnant women with low HL interpret the information received incorrectly, execute health skills incorrectly during prenatal care, and do not understand that health services may help them and their families.
Thus, strategies should be directed to pregnant women to improve HL, and our study indicates two ways: health education and social networks.
Although other variables did not present statistically significant values, contextual factors must be also considered for understanding HL. Guo, et al., (2018) observed that intact families were associated with a higher HL status. Additionally, self-efficacy and school environment were important in predicting functional HL, whereas social support was an important predictor of the structural factors of interactive and critical HL. Akbarinejad, Soleymani & Shahrzadi, 2017) observed an indirect relationship between HL and age, which may be linked to the reduction in cognitive performance and sensory abilities. However, Duggan, et al. (2014) indicated that HL improved as age increased. Thus, age seems to be a contradictory predictor of HL in pregnant women, considering the present study used a similar age range and did not observe a relationship between age and HL.
This study presents limitations. The first one is the associated with use of a selfreported survey, which implies the possibility of participants to over-or undervalue their responses. However, this is a common problem for observational studies using self-reported instruments. The second limitation was the failure to locate some pregnant women (n=31); however, their sociodemographic characteristics are very similar to those participating in the study. Finally, the cross-sectional design does not allow establishing cause-effect (reverse causality) (Celentano & Szklo, 2018). Thus, other study designs are suggested to investigate the association between HL and contextual characteristics of pregnant women.
Despite the limitations, it is important to highlight that this study used a validated instrument and it was the first to use the pHLAT-8 (or versions of the original instrument in any language) compared to other validation studies; also, multivariate analyses were performed. In addition, the population of the present study was affected strongly by social inequities, configuring a priority population in need of the creation and implementation of health promotion strategies.

Conclusions
Our study on HL among low-income pregnant women benefited from a cash transfer programme of the Brazilian government shows lower HL than other studies using the same instrument. Participation in health education activities and the involvement in social networks for advice regarding pregnancy are two predictors of HL in this population.
As practice implications, the results reinforce the importance of health teams to develop health education activities among pregnant women, especially low-income mothers.
This initiative may improve the understanding and search for information. Such health education programme may benefit from the incorporation of social networks of pregnant women, who may serve as models for peer education. Development, v. 9, n. 11, e4979119650, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.9650