Inequities and Disparities: An Investigation of Antenatal Visits in Mozambique

2.1. Source of Data and Sampling Design

This cross-sectional analytical study used secondary data from the Mozambique National Demography and Health Survey (2022-23). The survey was conducted as part of the eighth round of demographic and health statistics (DHS). For the survey, 6168 women were interviewed about their socio-demographic background, marriage and sexual life, family planning, and so on. The DHS employs two-stage cluster sampling: the first stage at the enumeration areas and the second stage at the households within the selected enumeration areas. The samples were representative of national, regional, and resident levels [14].

2.2. Variables Selected for the Model

The dependent variable was the number of antenatal visits by women. The independent variables were selected using the individual-interpersonal-community framework of the socio-ecological model. Sun et al. and Jalu et al. suggested that age, education, income, ethnicity, and women’s ability to make decisions are individual factors. They recommended social support, husband’s education, marital status, and religion as variables at the interpersonal level, whereas the community level comprised access to information, location of the household, and perceived barriers to care [15, 16].

Based on the adapted theoretical framework, the variable inclusion criteria were as follows:

2.3. Tools used for the Study

RStudio (Version: 2024.04.01) and Microsoft Excel (2019) were used for the analysis. The zero-inflated Poisson regression (ZIP) was used to examine the factors associated with the number of antenatal visits in Mozambique. Zero-inflated and negative binomial models were used when there was overdispersion (mean<variance) in the dependent variable caused by excess zeroes and non-zeroes [17]. Table 1 provides the descriptive statistics of the variable ‘number of visits to antenatal care,’ where the overdispersion ratio (variance/mean) is greater than 1 (1.05), indicating overdispersion in the presence of zeroes, and it resolves when zeroes are removed (overdispersion ratio = 0.67 <). Thus, zero-inflated or negative binomial models provide a better fit than regular Poisson regressions [18, 19].

The Zero-inflated Poisson Regression (ZIP) was selected after considering Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC), and theoretical considerations. Four-count regression models were considered in the study, and hurdle models were not used for the analysis. The zero-inflated model accommodates and distinguishes both sampling and structural zeroes, as some women would not visit antenatal care in any circumstance (structural zeroes) and some women did not visit antenatal care by chance during the study period, even though they might visit them in the future (sampling zeroes). The Hurdle model considers zeroes to be sampling zeroes, which may not fit the theoretical framework [17].

Table 2 shows the model fit of various count regression model statistics. Poisson and negative binomial regressions have significantly higher AIC and BIC values, along with a lower log likelihood with fewer degrees of freedom, indicating a poorer fit. ZIP provided a better fit to the model, with the lowest AIC and BIC values and a relatively higher log likelihood.

3.1. Changes in Antenatal Visits Over the Years

The multiple bar diagram in Fig. (1) presents the relative change in the number of antenatal visits over time. The number of zero visits increased from 7.23% in 2015 to 13.75%; however, the percentage of 1-3 visits fell by 10% at the same time, and the percentage of four or more visits increased by 1.6%. The number of zero visits decreased to 8.14% in 2018 and increased marginally to 9.17% in 2022, whereas the percentage of four or more visits increased to 59.5%, indicating effective health management.

3.2. The Respondent Profile

The respondent profile is shown in Table 3, and 4040 samples were selected for the study after eliminating missing values from the datasheet. In the age category, most respondents belonged to the 20-24 (29.8%) and 25-29 (22.37%) categories, with more than 50% of the respondents. Approximately 76% of the respondents had education at the primary level (47.6%), or no education at all (28%), and most of them lived in rural areas (70%). The wealth quintile was almost equally distributed, except for the richest category (16%), with approximately 20% of the respondents. Religion-wise distributions were evangelical (28.36%), Catholic (23.54%), and Islamic (22%). The respondents mostly had partners with an education of either primary (38%) or secondary (28%) level, and 20% had partners with no education. Most respondents lived with their partners (63%) instead of a formal marriage (37%). When it comes to reading newspapers, most of the respondents revealed (92%) that they had never read it; similarly, the majority did not listen to the radio (64%), watch TV (69%), or use the Internet (83%). However, when it came to using them regularly, TV had the highest number of users (23%). Most of the respondents had no problem getting permission (94%), nor did they need a company to visit a hospital (90%). Distance to the facility was a problem for 45% of respondents.

3.3. The Zero-inflated Poisson Regression Model

Table 4 presents the results of the zero-inflated Poisson regression model. The column count model shows the incidence rate ratio (IRR) of the frequency of visits, and the zero-inflated column shows the odds of zero visits at antenatal care. Among the individual factors, wealth, education, and occupation were significant factors in both the incidence of visits and the odds of zero visits. The poorer (IRR: 1.061 and OR: 0.6937), middle (IRR: 1.143 and OR: 0.6063), and richer (IRR: 1.239 and OR: 0.1314) classes had a higher incidence of visits and lower odds of non-visits than the poorest. The richest group had a significantly higher incidence of visits (IRR: 1.279), but despite having lower odds of non-visits (OR: 0.59), it was not statistically significant, possibly due to the small sample size. Education is another variable that influences both the incidence of visits and the odds of zero visits; primary education (IRR: 1.042 and OR: 0.5623) and secondary education (IRR: 1.057 and OR: 0.3436) had both lower odds of zero visits and marginal yet significantly higher incidence of continued visits (significant at 10%), compared to women with no education. Women who had jobs either in agriculture (OR: 0.341) or other sectors (OR: 0.3436) had lower odds of zero visits than women with no job, and the latter also had a higher incidence of antenatal visits (IRR: 1.077). However, those who work in agriculture would have the same incidence of visiting antenatal care as women with no job. Age was an insignificant parameter in both models.

Among interpersonal factors, religion and the education of partners were significant predictors of both the incidence and frequency of visits. The variables ‘getting permission’ and ‘not wanting to go alone’ were significant predictors of non-visits among Mozambican women. Catholic women had a similar incidence of visiting antenatal care as Muslim women, but women who believed in African Zionism (IRR: 1.092), Pentecost (IRR: 1.112), and other religions (IRR: 1.095) had a higher incidence of antenatal visits than Catholic women. However, apart from Islam (OR: 0.2362), all other religions had insignificantly lower odds of visiting antenatal care compared to Catholic women. Women with partners with higher than secondary levels of education had a higher incidence of visits (IRR: 1.052), and those with primary (OR: 0.546) and secondary (OR: 0.264) levels of education had significantly lower odds of not visiting antenatal care at least once. Women who faced significant problems in getting permission (OR: 2.358) and not wanting to go alone (OR: 1.977) had significantly higher odds of non-visits than those with no problems. Marital Status was statistically insignificant.

Community factors, such as Internet use and distance to health facilities, were marginally significant parameters in determining the incidence of visits (significant at 10%). The location, frequency of listening to the radio, and watching TV were significant predictors in determining whether they visited antenatal care at least once. People who used the Internet within the last 12 months of the survey had a higher incidence of antenatal care visits (IRR: 1.046) than those who never used it. People who faced difficulties in visiting antenatal care due to their distance had a lower incidence of visiting them (IRR: 0.966) than those who did not face any problems. Women living in rural areas had higher odds of not visiting antenatal care at least once than urban women (OR: 2.38), and women who listened to radio (OR: 0.5) and those who watched TV (OR: 0.234) had lower odds of non-visits than women who never used it. Regular users of either gadget had lower odds, but they were insignificant, likely because of the low sample distribution. The frequency of reading newspapers was neither a significant predictor of incidence nor visits to zero.

4.1. Individual Factors

The wealth quintiles were significant in both whether they visited the antenatal care unit and the incidence of visits, as women belonging to higher wealth quintiles had a higher chance of visiting them frequently than the poorest of the population. Over 70 percent of the population in Mozambique is currently living in poverty, and with the evidence from this study, it can be inferred that antenatal visits are more of a luxury than a necessity for an average Mozambican woman [11]. Education is another significant predictor of antenatal visits; women with higher levels of education have lower odds of non-visits, and a higher incidence of antenatal visits is observed slightly more in women with secondary and higher levels of education. Similar findings were reported by Biza et al. in their qualitative study on antenatal visits in southern Mozambique, as the less educated are less aware of the importance of antenatal care and may have cultural beliefs that interfere with their judgement [10]. The results of this study align with those of Tesfay et al.’s study in Ethiopia; however, the nature of this relationship differs. They found that primary education had a higher incidence of antenatal visits, but the odds of visiting antenatal care were similar to those of women with no education. However, this study found that the odds and incidence rates in women with primary education were higher than in women without education [21]. Sun et al., in their review of the application of the socio-ecological model to pregnant women’s health, discussed low wealth and education as major barriers to physical activity due to its influence on accessibility and awareness, and similar studies conducted on antenatal visits in sub-Saharan Africa validated these findings [12, 15, 16].

Women working in sectors other than agriculture had a higher incidence of visits and lower odds of non-visits compared to women with no job, and women working in the agriculture sector had lower odds of non-visits but a similar incidence of higher visits. In their review of antenatal visits in sub-Saharan Africa, Okedo-Alex et al. noted that the low incidence of visits by women working in agriculture may be due to their lack of or delayed initiation [22]. The reason for delayed initiation could be wealth constraints and a lack of accessibility to healthcare, as agriculture is predominantly a rural activity [22, 23]. Age was an insignificant parameter in both models, indicating that women of all ages had a similar incidence of use and odds of non-use, which was also observed in Guinea and Mali [12]. This finding contradicts that of similar studies conducted in this region [16, 21-23].

4.2. Interpersonal Factors

Catholics in the country have a significantly lower incidence of visits to religions other than Muslim women; however, the latter is more likely to visit antenatal care once than the former, indicating that Catholic women are least likely to visit the facilities at least once or use them frequently. This is likely due to the orthodox nature of the Catholic women, which would prevent them from using modern reproductive services, these findings are corroborated by studies conducted in Malawi, Zambia, Mali, and Ethiopia [12, 24, 25]. However, this finding is contradictory in other sub-Saharan African countries, such as Nigeria and Zimbabwe, where Christian women are more likely to use them over Muslims [12, 22, 24]. Women with an educated partner are more likely to have a higher incidence of visits and lower odds of non-visits, indicating that women with highly educated partners will have a better support system at home than women with partners with no education [12, 15, 23]. Similarly, women who face restrictions from their partners/parents in obtaining permission to visit a hospital or have problems going there alone have higher odds of non-visits, indicating a lack of support systems [12, 15].

4.3. Community Factors

Evidence suggests that women living in rural areas are twice as likely as their urban counterparts to not visit antenatal care. The study also found that women who perceived distance to the health facility as a big problem had marginally higher odds of non-visits. This clearly indicates a lack of geographic accessibility for health services in the rural areas of Mozambique owing to a lack of public infrastructure, natural disasters, and healthcare [10, 11]. Geographic inaccessibility is a common theme in sub-Saharan Africa, and studies conducted in Nigeria, Ethiopia, Zambia, Mali, Malawi, and Guinea have reported similar findings [12, 22-24]. Women who used the Internet, radio, or television were more likely to visit antenatal care facilities at least once than women who never used them. Sun et al. pointed out that access to electronic media can positively influence behaviors deemed healthy for pregnant women [15], which is supported by several studies conducted in the sub-Saharan region [12, 25]. However, Belay et al., in their study in Ethiopia, found that women are more prone to the recommendations and advice of their partners than mass media or health professionals (Supplementary Material) [23].