This study was conducted at Kiswa HCIII & China-Uganda Friendship Hospital-Naguru located within Nakawa division, Kampala and are the biggest public health facilities offering maternity services for adolescents including those residing within this division. These facilities attend to more than 100 adolescent mothers on a monthly basis. These were purposively selected due to their strategic location as catchment areas for adolescent mothers within this area to receive maternal health services. This was a facility-based descriptive cross-sectional study. We included all adolescent mothers aged 10 - 19 years attending the antenatal clinics of the selected health facilities and were residents of Nakawa division.

Adolescent mothers who were sick and unable to give information at the time of the study were excluded.

We used systematic random sampling to enroll eligible adolescent mothers in the two facilities. Probability proportionate to size sampling was used to determine the required respondents for each facility. The sampling interval was calculated based on the facility’s recent quarterly report of pregnant adolescent attendees. In both facilities, the 2^nd mother of the day was interviewed only after determining the first by simple random sampling. A facility-based sampling frame was chosen to efficiently capture pregnant adolescents already engaged in ANC services at these high-volume public facilities serving urban slums, enabling direct assessment of care timing and related factors among service users. However, this approach could limit generalizability of these findings to all slum-dwelling adolescents in Kampala, as it excludes those not attending these facilities (e.g., due to barriers like stigma or distance) or using private/non-selected providers.

Sample size estimation was done using the Kish Leslie formula basing on a prior study by Atusiimire et al. to assess determinants of facility-based deliveries within urban slums in Kampala district [6] . Using the Kish-Leslie formula (n = Z² *p *(1 − p)/d²) parameters were based on a prior study, as mentioned above, estimating a 39% prevalence (p = 0.39) of early ANC initiation in Kampala slums, with Z = 1.96 (95% confidence level) and d = 0.05 (5% margin of error). This resulted in an initial sample size of 366, which was adjusted downward for a finite population of approximately 365 adolescent attendees (from Jan 2022 to April 2022) across the two facilities (using finite population correction) to 183. It was then further increased by 10% to account for non-response, resulting in the final sample size 202.

The main outcome variable was timing of ANC initiation which was categorised as early (<12 weeks of gestation) or late (≥12 weeks of gestation). This was based on the adolescent mother’s responses to the question—At what gestational age did you first visit the ANC clinic for antenatal care?

The study analysed independent variables including socio-demographic variables like mother’s age, education, parity, religion, wealth index, employment status, husband/partner’s education, exposure to mass media, and individual perceptions of danger signs during pregnancy. It also assessed obstetric factors like gestation age, parity, expected delivery date, number of children, age at first childbirth, and complications in the last pregnancy.

Health system factors assessed included distance to health facility, waiting time and quality of care.

Data analyses were done at three levels: univariate, bivariate and multivariate levels.

Univariate analyses generated frequency counts and percentage distributions.

Bivariate analyses tested the association between the variables using the Pearson chi-square and Fisher’s exact tests, while the t-test compared age between groups. P-value < 0.05 was considered statistically significant. The modified Poisson model was used to identify significant variables with a p-value of 0.2 or less for the multivariate model. The significant variables included religion, education level of the respondent and partner, partner employment status, marital status, living with the partner, gestation age, and waiting time at the ANC clinic. In addition, parity, quality of care, age, and persons that advised the adolescent mothers to start ANC.

We used a modified Poisson regression model to identify which variables predict the likelihood of an adolescent mother making the first antenatal care (ANC) visit in the first trimester. The model had robust standard errors and was fitted at the multivariate level. The modified Poisson regression with robust standard errors was preferred over binary logistic regression because the outcome (early ANC initiation) was not rare (>20% prevalence), making odds ratios from logistic models prone to overestimating prevalence ratios for common binary outcomes. For bivariate and multivariate analyses, the level of statistical significance was set at p ≤ 0.2 and p < 0.05 respectively.

The models were tested for goodness of fit using the link test and the Hosmer-Lemeshow test. Multicollinearity was checked using the Variance Inflation Factors (VIF), and any predictor variables with a VIF of 10 or greater were excluded from the models.

Informed consent was obtained from the study participants before enrolment into the study. Although some of the study participants were below 18 years of age, they were considered emancipated minors and as such consent was obtained in line with UNCST guidelines [12] . Approval to conduct the study was obtained from Makerere University School of Public Health IRB protocol number 117.

The study found that most adolescent mothers delay the initiation of their first antenatal care (ANC) visit, suggesting the need to sensitize pregnant mothers to start antenatal care immediately.

Whereas WHO in 2016 endorsed guidelines that recommends all pregnant women to initiate ANC in the first trimester [13] , our findings revealed that most of the adolescent mothers delayed initiating the first ANC visit. The study found that only 35% of the pregnant adolescents started ANC in the first trimester which is lower than other similar studies in this population. Although most studies had shown that ANC initiation amongst adolescents is lower amongst rural dwellers than their urban counterparts; this proportion is lower than the findings from a number of studies conducted amongst adolescents in rural settings. A case in point is a study done in Luuka district by Kayemba et al. (2023) that showed that only 47.2% of adolescents attended ANC in the first trimester [14] while a study done among adolescents in South Africa showed early initiation of ANC among rural adolescents at 46% [15] . Another study in India showed that 63% of pregnant adolescents aged 15 - 19 years registered their pregnancy in the first trimester [16] . The study results are also consistent with another study done in Ethiopia that showed that residents within urban slums have inadequate antenatal care services with most having have late initiation of antenatal care [11] . This may be explained by the loss of the urban advantage whereby health services which were presumed to be better off in the urban compared to rural areas are not exactly that [10] .

Delayed attendance of ANC exposes adolescent mothers to an increased risk of perinatal mortality, particularly, stillbirth since the potential risks cannot be identified and managed early by skilled health personnel [17] . Although according to the 2022 Uganda Demographic Health Survey, the maternal mortality ratio reduced from 336 to 189 per 100,000 live births, this is still a slow pace to realisation of the sustainable development goal (SDG) target of reducing the ratio to 70 per 100,000 live births by 2030.

This study reveals a significant relationship between the person who advised ANC initiation and timing of initiation highlighting the importance of external support like family members [18] . Similar findings were found in a study done in Bangladesh regarding family influences on health and nutrition practices among pregnant adolescents [19] . This is also consistent with existing literature where family members like grandmothers have a central influence on pregnant women [20] . A recent scoping review conducted in Rwanda also showed that those who lack family support are 70% more likely to delay antenatal care [21] .

The study also found that factors such as gestation age, staying with husband/partner, education level of husband/partner and waiting time at the health facility influence early ANC initiation.

Adolescent mothers’ gestational age significantly impacts their likelihood of initiating ANC in the first trimester with delayed revealing of unintended pregnancies impacting the utilization of ANC services [22] .

Living with a partner boosts early antenatal care (ANC) attendance, while not living with a partner reduces ANC initiation by 57% in the first trimester [23] .

Optimal waiting time at health facilities increases the likelihood of initiating antenatal care (ANC) in the first trimester, as it encourages better utilization of ANC. Short waiting time at the antenatal care clinics has been identified as an incentive for better utilisation of ANC [24] .

Education of husband/partner significantly impacts adolescent mothers’ ANC timing, with 65% less likely to start in the first trimester. Higher education levels lead to better utilization of ANC services, as shown in Ghana [25] . A recent systematic review indicated that several studies have identified low educational attainment of the partner or husband as a barrier to adolescents fully utilizing antenatal care [26] .

The study has limitations, including potential social acceptability bias and recall bias. To reduce these, research assistants from other departments were used and information was verified using antenatal appointment cards.

This cross-sectional study, limited in its cause-and-effect determination, provided some evidence about antenatal care utilization in urban settings, despite limitations such as selecting only public health facilities, despite being predominantly urban. The study was also subject to potential selection bias, as it excluded adolescents who were too ill to attend and those accessing care outside the selected public facilities. These groups may differ systematically from the study population in terms of health status, health-seeking behavior, and access to care, which could affect the representativeness and generalizability of the findings.

The study showed low timely initiation of antenatal care (ANC) among pregnant adolescents, indicating the need for increased community awareness and communication structures. Family support and positive health worker attitudes were key facilitators.