Institutional and home delivery practices and their predictors among women in Benadir region, Somalia: a community-based cross-sectional study

Study area and design

This community-based cross-sectional study was conducted from February to April 2025 among residents of four purposively selected peripheral districts in the northern part of Benadir Region, namely Heliwa, Karaan, Warta Nabadda and Yaqshid. These districts were selected based on their high population density, diverse sociodemographic profiles and the presence of relatively few functional maternal health facilities. While some of these districts are located in peripheral areas, others are more centrally situated, allowing for a balanced representation of urban communities. The selection aimed to capture a range of maternal health service experiences and contextual factors that may influence institutional delivery service utilisation, thereby enhancing the relevance and generalisability of the findings.

Study population

The study population consisted of all women who had given birth within the last 12 months and were permanent residents of the specified districts.

Inclusion and exclusion criteria

Inclusion criteria were women aged 18–49 years who had given birth within the last 12 months and were permanent residents of the selected districts in Benadir at the time of delivery. Women who were not permanent residents, unable to provide consent or had severe illness preventing participation were excluded.

Sample size determination

A sample size of 710 women was calculated using a single population proportion formula: n=Z²(α/₂)p(1−p)/w², where n represents the sample size, z is the standard normal deviation (set at 2.58 for a 2% CI), w is the desired degree of accuracy (taken as 0.05), and p is the estimated prevalence of institutional delivery (32%) as reported by SDHS.13 A design effect of 1.1 was applied to account for the increased variance introduced by the multistage sampling technique. This value was chosen based on similar community-based studies conducted in comparable settings, where minimal clustering was expected. Additionally, a 10% non-response rate was considered to ensure adequate sample representation.

Sampling technique

The study employed a multistage sampling approach. First, four districts were randomly selected from the 20 districts in the Benadir region. Second, within each selected district, a list of all households was compiled. Households were then selected using systematic random sampling, with the sampling interval determined by dividing the total number of eligible households by the required sample size for that district. Third, in each selected household, one eligible woman (who had given birth within the last 12 months) was identified and invited to participate. If a selected household did not have an eligible woman, the next nearest household was approached. Non-responders were replaced with eligible participants from the same sampling frame, and each household was revisited two times before replacement.

This structured, probability-based method ensured that the sample was representative of the study population and minimised selection bias, thereby enhancing the reliability and generalisability of the findings within the districts of the Benadir region and to similar settings.

Data collection procedure

Data collection was conducted through face-to-face interviews by trained female enumerators in the participants’ residences. To ensure consistency and high data quality, enumerators underwent a 2-day intensive training on study objectives, ethical considerations, the questionnaire content and interview techniques.

Prior to each interview, enumerators explained the purpose of the study, emphasised confidentiality and obtained informed verbal consent. Interviews were carried out in quiet and private areas within the participants’ homes, minimising distractions and safeguarding the confidentiality and comfort of respondents. All interviews were conducted in Somali, the participants’ primary language, and were approached with patience and respect to encourage open, accurate and detailed responses.

Study instrument

A structured and pretested questionnaire was utilised to collect data. The instrument was developed after a comprehensive review of relevant literature20 and was reviewed by maternal health experts to ensure cultural and contextual relevance. The questionnaire was pretested on a sample of 10 women in a non-selected area similar to the study sites. Based on feedback from the pretest, adjustments were made to improve the clarity, sequence and appropriateness of the questions. Pretesting and expert review were conducted to ensure clarity, relevance and appropriateness of the questionnaire, and feedback was used to refine the instrument.

The development of the questionnaire was guided by Andersen’s Behavioral Model of Health Services Use, which categorises factors associated with health service utilisation into three domains: predisposing (eg, sociodemographic characteristics), enabling (eg, economic status and access to services) and need-based (eg, obstetric history and perceived health needs).21 This framework informed the selection and organisation of variables to ensure a comprehensive assessment of institutional delivery utilisation.

The finalised tool included sections on sociodemographic characteristics (eg, age, marital status, education, employment, occupation, family size, income and perceived economic status), obstetric history (eg, age at first marriage, age at first pregnancy, parity) and healthcare access and service utilisation (eg, distance to nearest health facility, ANC attendance, number of ante natal care visits, knowledge of pregnancy danger signs (severe bleeding, convulsions, headache/blurred vision, fever, swelling, breathing difficulty, reduced fetal movement, prolonged labour and loss of consciousness), place of delivery and type of birth attendant). The questionnaire predominantly used closed-ended questions with some probes to ensure completeness of responses.

Study variables were measured to identify factors associated with institutional delivery utilisation among women in the Benadir region. The dependent variable, institutional delivery utilisation, was assessed by determining whether women had given birth within the last 12 months in a healthcare facility attended by a qualified birth attendant, as reported by the participants. The independent variables, including sociodemographic, obstetric and healthcare access-related factors, were measured through structured interviews and questionnaires to assess their impact on institutional delivery utilisation.

In addition to the quantitative data, the study also qualitatively explored women’s reasons for choosing home delivery. Structured interviews included open-ended probes that captured personal, cultural and logistical factors associated with delivery decisions. These insights were analysed to complement the quantitative findings and provide a deeper understanding of barriers to institutional delivery.

Data management and analysis

All collected data underwent rigorous checks for accuracy, completeness and consistency. Missing or erroneous entries were addressed through systematic data cleaning procedures. After preparing the dataset, statistical analyses were performed using SPSS V.25.

Descriptive statistics—frequencies and percentages—summarised participants’ demographic, obstetric and healthcare characteristics as well as institutional delivery utilisation. To identify factors associated with institutional delivery, Poisson regression with robust variance was applied to estimate adjusted prevalence ratios (aPRs). Confounders included in the multivariable model were selected based on literature review, clinical relevance and statistical significance (p<0.20) in bivariate analysis. These included maternal age, education level, parity, ANC attendance, distance to health facility and employment status.

The aPRs, derived from the regression model, quantified the strength of associations while accounting for potential confounders. 95% CIs were computed to assess estimate precision, and statistical significance was set at p<0.05. The unit of analysis was individual women, and no clustering adjustments were required as sampling was independent. Missing data were minimal (<5%) and handled through complete-case analysis.

Institutional delivery

The findings from the present study, conducted on mothers who gave birth in the last 12 months, revealed that merely 292/710 (41%) of mothers delivered their most recent child at healthcare facilities (figure 1).

Factors associated with institutional delivery

Women’s likelihood of institutional delivery was strongly influenced by ANC attendance, knowledge of obstetric danger signs and proximity to health facilities. Those who attended ANC were more than three times as likely to deliver in a health facility (aPR=3.20; 95% CI 2.30 to 4.40), and women aware of danger signs had two times the likelihood (aPR=2.00; 95% CI 1.50 to 2.70). Living within walking distance significantly increased facility delivery (aPR=3.01; 95% CI 2.10 to 5.02), highlighting geographic access as a key determinant. Birth type also mattered—single births were less likely to occur in a facility compared with multiple births (aPR=0.38; 95% CI 0.22 to 0.64). Maternal education showed a strong inverse association, with women lacking formal education far less likely to deliver in a facility (aPR=0.19; 95% CI 0.09 to 0.58).

Other factors, including marital status (aPR=1.97; 95% CI 0.91 to 3.40), household size (aPR=0.54; 95% CI 0.42 to 1.47), maternal working status (aPR=1.16; 95% CI 0.77 to 1.74), husband’s employment (aPR=1.91; 95% CI 0.94 to 3.90), monthly income (aPR=0.56; 95% CI 0.36 to 1.33), age at first pregnancy (aPR=1.30; 95% CI 0.81 to 2.54) and parity (aPR=1.32; 95% CI 0.75 to 2.92) were not statistically significant after adjustment (table 2).

Reasons behind the decision for home delivery

The findings reveal that the most common reasons for home delivery among women in the Benadir region are primarily linked to financial constraints, accessibility issues and personal comfort. A significant number of respondents cited previous safe home deliveries, the high cost of private hospitals and long distances to health facilities as key factors affecting their choice. Sudden onset of labour and lack of transport also played major roles, indicating logistical barriers to timely facility access. Cultural beliefs, trust in TBAs and feeling more supported at home further reinforced home delivery preferences. In contrast, fewer women reported reasons related to gender dynamics, such as fear of male providers or lack of spousal permission, and concerns about medical procedures like caesarean sections (figure 2).

Strength and limitations

This study has several strengths that enhance the reliability and relevance of its findings. The use of a community-based cross-sectional design allowed for the inclusion of a diverse sample of women from various settings within the Benadir region, making the findings reflective of real-world conditions. By targeting women who had given birth in the past 12 months, the study minimized recall bias and ensured the data were recent and contextually meaningful. Importantly, the analysis applied both bivariate and multivariable Poisson regression with robust variance, allowing for the identification of independent factors associated with institutional delivery while adjusting for potential confounders. This strengthens the analytical rigour and the interpretability of the results. Additionally, conducting research in a post-conflict setting like Somalia adds value by capturing maternal health realities shaped by disrupted systems and limited service access.

Importantly, the study provides context-specific evidence on institutional delivery in Benadir that can inform local maternal health programming and policy. By identifying modifiable factors associated with facility use, it offers actionable insights to strengthen ANC uptake, community awareness, and access to skilled birth services, while also contributing evidence from a fragile, post-conflict setting where data remain limited.

Nevertheless, certain limitations should be acknowledged. The cross-sectional nature of the study limits the ability to draw causal inferences from the associations observed. It also restricts the ability to assess temporal trends or determine the directionality of relationships between variables. The study was also geographically limited to the Benadir region and may not reflect the experiences of women in rural areas or displacement settings, where access and cultural norms may differ. Furthermore, unmeasured confounding variables may have influenced the observed associations, and the reliance on self-reported data introduces the possibility of recall and social desirability bias.