Participants for this study were recruited from the 11 ^th of April 2024 to the 1 ^st of November 2024 as part of the NeoRisk Study. The NeoRisk study is a prospective observational study, aiming to analyse the sociodemographic impact on neonatal and maternal adverse outcomes in a cohort of mother-baby pairs receiving perinatal care at Kawempe National Referral Hospital in Kampala, Uganda. A total of 1,070 participants were recruited during this period from the antenatal clinic and labour ward. Data was collected during patient visits through structured interviews just before or after delivery, with informed consent obtained from all participants. The collected data was securely stored in the NeoRisk dataset on RedCap.

To be eligible for recruitment, participants had to have delivered a live baby at 34 weeks’ gestation or more at Kawempe National Referral Hospital and had to provide informed consent. The data collected included a comprehensive range of variables such as medical and obstetric history, labour and delivery outcomes, neonatal outcomes, and demographic and lifestyle factors.

Ethical approval was obtained from the Uganda Virus Research Institute Research Ethics Committee (Ref: GC/127/996), the Ugandan National Council for Science and Technology (Ref: HS3278ES) and City St George’s, University of London Research Ethics Committee (Ref: 2024.0090). All participants provided informed consent, either written or via witnessed thumbprint if unable to read and write.

Kawempe National Referral Hospital, located in Uganda’s capital city of Kampala, was established in 2014 and officially opened to the public in 2016. It is the largest national referral hospital for pregnant women in the country ( Kyohere et al., 2020), offering specialised services in obstetrics, paediatrics, adolescent health, and HIV/AIDS care. The hospital largely serves the urban poor within Kampala City, which has a population of approximately 1.8 million inhabitants ( Uganda Bureau of Statistics, 2024b).

The facility features a maternity ward staffed with approximately 47 midwives, 20 specialist obstetricians and gynaecologists, and a variable number of residents, depending on the enrollment of affiliated universities utilising the hospital for clinical placements ( Ainebyona et al., 2024). The hospital records an average of 80 births daily ( Jijuuko, 2022). Between July 2022 and June 2023, the hospital reported a total of 1103 stillbirths and 125 maternal deaths ( Ainebyona et al., 2024).

All statistical analysis was conducted using RStudio version 2024.12.0. The analysis focused on understanding the relationships between key sociodemographic groups such as parental education, maternal age, maternal occupation, and previous birth outcomes (including total number of live births and deceased children). The term deceased children can be defined as the total number of children the mothers had who died after birth. Additionally, socioeconomic factors such as food security and access to sanitation were investigated against ANC attendance. Food security was assessed based on the self-reported frequency of concerns about running out of food for both the mother and their family in the past month and the number of meals skipped. Access to sanitation was investigated by examining the availability and type of facilities a family has, such as the use of toilets or latrines.

Pearson Chi-Squared tests, Fisher’s Exact Test (in variables with ≤5 observations) and Poisson regression analysis were conducted to identify and quantify potential predictors of antenatal clinic attendance. Variables such as maternal age, number of live births and deceased children were each analysed using both categorical and continuous approaches. Initially, variables were categorised, and associations with ANC attendance were assessed using chi-squared tests. Poisson regression was then performed, treating these variables as continuous. Statistical significance was determined at the threshold p < 0.05.

Continuous variables such as antenatal visits, age and previous birth outcomes were categorised for this analysis. To analyse the relationship between sociodemographic characteristics and the frequency of ANC visits, we categorised ANC visits into three groups: ≤4, 5-7 and ≥8. This classification aligns with the updated WHO guidelines recommending the minimum number of antenatal visits from four to eight visits ( World Health Organisation, 2016). Maternal age was grouped: 15-24, 25-34 and 35-43 years old. The total number of live births was categorised as 1-2, 3-4 and 5+. Deceased children were categorised as none, 1 and 2+.

Socioeconomic status was assessed using two domains: food security and access to sanitation. Food security included two components: concern about food sufficiency, which was categorised as a binary yes/no variable, and the total number of meals skipped. This was grouped into four categories: 1-5, 6-15 and 16+. Finally, access to sanitation was grouped as follows: •

Toilet: flush toilets connected to a sewer or compost, not shared.

Occupations were categorised using the International Standard Classification of Occupations (ISCO-08), which groups jobs into hierarchical categories based on skill level and specialisation ( International Labour Organisation, 2023). For this analysis, occupations were grouped as follows: •

Skill level 1: Includes elementary occupations requiring basic tasks and limited education, typically corresponding to primary or lower secondary education.

Table 1b summarises the associations between various sociodemographic characteristics and the number of antenatal care (ANC) visits. Significant associations were observed between ANC attendance and several variables, including parental education, maternal age, the number of deceased children, concern about food security (running out of food, and the number of meals skipped).

Mothers’ education level was significantly associated with ANC attendance (chi-squared p = 0.018; Fisher’s exact p = 0.013). Women with secondary education were more likely to attend five or more ANC visits compared to those with no or only primary education. Paternal education was also significantly associated with ANC attendance (p = 0.007 for both tests), with higher paternal education correlating with increased ANC utilisation.

A statistically significant association was found between maternal age group and ANC visits, when determined by Fisher’s exact test (p = 0.039). ANC attendance varied across age groups. Women aged 25-34 were more likely to attend more visits. This increased from 45% of them attending four or fewer visits to 53.5% attending more than 8 ANC appointments. The proportion of women aged 15-24 years attending their ANC appointments remained relatively stable across the visit categories. In contrast, women aged 35-43 years were less likely to attend ANC appointments. This indicates that mothers aged 15-34 years are more likely to attend the recommended ANC visits compared to those older than 35.

No significant associations were found between the mother’s occupational skill level and attendance at ANC.

No statistically significant associations were observed between the number of live births (chi-squared p = 0.105; Fisher’s exact p = 0.120) and ANC attendance when the number of live births was analysed as a categorical variable. Conversely, both the chi-squared (p = 0.005), Fisher’s exact (p = 0.005) analyses showed a significant association between the number of deceased children and ANC attendance.

In terms of socioeconomic indicators, no significant association was found between access to sanitation and ANC appointment attendance. However, food security-related factors were significantly associated with the use of ANC. Women who were not concerned about running out of food had higher ANC attendance (p < 0.01). Likewise, the number of meals skipped was significantly associated with ANC attendance (chi-squared p = 0.007; Fisher’s exact p = 0.004), with women who skipped meals more frequently attending fewer ANC visits.

Table 2a presents the results of a Poisson regression analysis examining the relationship between ANC attendance and continuous variables: maternal age, number of live births, and number of deceased children.

In the adjusted model, a higher number of live births was significantly associated with fewer ANC visits (IRR 0.956; p = 0.0181), suggesting that women with more children were less likely to attend ANC appointments. Conversely, the number of deceased children was positively associated with ANC attendance (IR 1.16; p = 0.00337), indicating that women who had experienced a child loss were more likely to seek ANC services.

Maternal age, however, when treated as a continuous variable, showed no significant association with ANC attendance in either the unadjusted or adjusted models.

The regression models are also illustrated in Figure 1 (scatter plot).

The association observed between maternal age and ANC attendance in this study adds to the mixed evidence on how age influences maternal health service utilisation. Our analysis found a statistically significant association between maternal age groups and ANC attendance, as determined by Fisher’s exact test. Women aged 25-34 consistently showed the highest attendance for eight or more ANC visits, followed by those aged 15-24, while none aged 35-43 attended more than eight ANC appointments. The relatively high attendance within the 25-34 age group may be explained by various factors. For instance, they may have better health literacy and more autonomy in decision-making. Nevertheless, these results reveal that women in their late twenties and early thirties are more likely to attend more ANC appointments compared to older mothers, highlighting a gap that may need targeted intervention.

These findings are consistent with regional literature, for instance, Boadi et al. (2023) analysed the Ghanaian Demographic and Health survey dataset on pregnant women’s ANC visits for 2003, 2008 and 2014, and found mothers aged 25-29 years had the greatest proportion of attending more than eight ANC appointments.

However, there are studies which contradict our findings, reporting that older women are more likely to attend more ANC appointments than younger women. For example, Atuoye et al. (2020) reported that younger women, particularly those pregnant for the first time, have a significantly lower probability of accessing maternal health services. They found that women aged 45 were two times more likely to access early ANC and about three times more likely to meet the WHO-recommended maternal health service in 2016 compared to those aged 15.

These contrasting findings may be due to the influence of social dynamics on maternal health-seeking behaviour. In Uganda, many first-time mothers are teenagers or unmarried, and unlike older women, they often face barriers such as fear of punishment for early sexual activity, community stigma, and limited autonomy in making health-related decisions ( Atuoye et al., 2020).

Moreover, maternal age and ANC attendance can also be shaped by previous pregnancy experience. Depending on the outcome of earlier pregnancies, women may be inclined to seek ANC early in subsequent pregnancies. Factors such as economic well-being and social support are limited for younger mothers; hence, they seek fewer ANC appointments. Manyeh et al. (2020) found that older women were more likely to initiate antenatal care visits within the first trimester of gestation compared to the younger women. These findings highlight the need for more targeted, age-sensitive strategies to promote ANC attendance, ensuring that both younger and older mothers receive the necessary support to overcome the distinct barriers they face.

Both maternal and paternal education were significantly associated with increased ANC attendance. Women with secondary or higher education were more likely to complete the recommended number of ANC visits compared to those with no or only primary education. These findings are consistent with previous literature. For example, Bhutada et al. (2024) found that in three out of eight studies reviewed, higher maternal education was significantly associated with greater ANC uptake. These findings suggest that improving access to education in particularly for women, can be a long-term strategy to enhance maternal healthcare utilisation.

Paternal education also emerged as a strong predictor of ANC attendance. In line with our findings, Apanga et al. (2022) reported that partners with at least secondary education were 26% more likely to support timely ANC attendance in Northern Ghana. This could be explained by several interrelated factors, including greater health literacy, more stable income, and greater involvement in family decision-making among better-educated men. Education and resulting higher income may also alleviate financial barriers such as transport costs, which have been shown to limit access to care. In the short term, ANC promotion efforts should prioritise parents with limited education to raise awareness of the benefits and timing of ANC services.

Our analysis showed divergent associations between different birth outcomes and ANC attendance. Women with more previous live births were less likely to attend ANC, possibly due to a perceived sense of experience or reduced perceived need for care. Mwenebanda et al. (2024) conducted a qualitative study on factors influencing ANC uptake in women in Malawi. They found that parity appeared to be an influence on ANC attendance, with many women actively seeking more ANC during their first pregnancy due to uncertainty ( Mwenebanda et al., 2024). However, with subsequent pregnancies, multiparous women were less likely to seek ANC, possibly due to the sense of familiarity and increased confidence. Pell et al. (2013) further highlighted the complexity of ANC utilisation among primigravidae women from Ghana, Kenya, and Malawi. They observed that although some primigravidae were more likely to initiate ANC earlier on, others, however, felt that due to the sense of unfamiliarity with pregnancy symptoms, women would unintentionally delay seeking ANC. Nevertheless, Pell et al. (2013) noted that these decisions were not made in isolation, but instead were often influenced by older relatives such as their mothers.

In contrast, our study also found that those who had experienced the loss of a child were more likely to attend ANC appointments. This pattern may reflect a heightened sense of risk of perceived vulnerability following a negative pregnancy outcome, motivating women to attend more ANC appointments. This finding aligns with a study conducted in Mali, where women with a history of stillbirths were more likely to attend more than four ANC appointments ( Bognini et al., 2025). Collectively, these findings underscore the importance of tailoring ANC messages not only to first-time mothers but also to multiparous women, recognising that both overconfidence and vulnerability can influence whether a woman will seek ANC throughout different stages of their pregnancy.

Food insecurity was another important determinant of ANC utilisation. Women who reported concerns about running out of food or had skipped meals in the previous month were significantly less likely to attend ANC visits. This suggests that financial hardship and basic survival needs may compete with healthcare access. Similar results were reported by Kotiso et al. (2020) in South Ethiopia, where food insecurity was linked to reduced ANC attendance due to fatigue, malnutrition, and competing priorities. These findings highlight the need to integrate social protection and nutritional support with maternal health programmes, particularly in urban poor settings where informal employment and fluctuating incomes are common.

While this study contributes valuable insights, several limitations must be acknowledged. First, the cross-sectional design limits our ability to infer causality. The relationships identified represent associations, and the underlying drivers may be complex and multifaceted. Future research should incorporate qualitative methods—such as interviews or focus groups—to explore in depth the reasons why women attend or miss ANC appointments. Understanding individual, social, and systemic barriers can help tailor interventions to the needs of specific subgroups.

Second, some methodological limitations affected the statistical analysis. Several categorical variables had low cell counts (fewer than five observations), reducing the validity of chi-squared tests. To address this, Fisher’s Exact Test was used for small sample sizes, as it provides more accurate p-values in such contexts ( Technology Networks, 2025). Although this approach strengthens the reliability of our findings, the limited statistical power in certain comparisons may affect generalisability.

Another limitation of this study lies in the reliance on self-reported measures for certain variables, such as food security. Self-reporting can introduce bias, potentially leading to an over- or underestimation of true effects. For instance, some women may feel embarrassed or ashamed to admit that they had skipped a meal and may therefore provide socially desirable responses rather than accurate ones. This can compromise the external validity of the data and obscure the true relationship between food security and ANC utilisation.