The MMSH is a tertiary hospital located in Kano, northern Nigeria, and the population this hospital serves is ~11 million. Within the MMSH, there are 1,000–1,200 general hospital beds, 17 neonatal intensive care unit (NICU) beds, 133 maternity beds, and 22 delivery cubicles. Each month, there are around 550 deliveries and there are four midwives on shift at any one time, two for complicated deliveries and two for uncomplicated deliveries.

Data were collected through two exploratory studies and ethical approval for both studies was obtained locally. Study 1 (S1), a retrospective case-control study, was conducted without funding; to investigate associations and to understand the complexities surrounding data collection. Data and experiences from S1 informed the design of Study 2 (S2); a prospective cohort study, whereby incidence could be determined.

The S1 (case-control) stillbirth data were collected between 13/09/2017 and 31/12/2017. Trained research assistants were present at all deliveries and if the neonate was confirmed stillborn by a medical professional, the mother was approached to participate in this study, and a pre-tested paper-based questionnaire was completed after obtaining informed consent. Stillbirths were retrospectively matched to a livebirth from BARNARDS by time (target of 24 h time variation) to establish a case-control study with a 1:1 ratio. The maximum time between stillbirth and matched livebirth was 29 days, due to BARNARDS ceasing enrolment in early December 2017. BARNARDS procedures¹ were followed to enrol controls. Sufficient relevant data were collected within BARNARDS to ensure that matched controls need not repeat the questionnaires.

The S2 (prospective cohort study) data were collected between 06/03/2018 and 22/04/2018 from all mothers presenting to MMSH while in labour, with the outcome being live or stillbirth. The mothers were enrolled during labour after informed consent was obtained, an initial pre-tested paper-based questionnaire was completed during labour, and another after delivery, dependent on delivery outcome.

The questionnaires captured the participant's information in three domains: maternal, household, and neonatal/birth. Clinical notes were not exclusively used for data collection; however, they were used for some data points (e.g., gestational status). The domains were selected to provide a holistic approach to the mother's characteristics, demographics, health background, and delivery, and they are linked to unaddressed issues of three delays in the Kano state: (i) recognition of danger signs and associated decisions to seek appropriate healthcare and knowledge of how to seek healthcare; (ii) access to healthcare (distance, cost, security [concern after usual hours], and transport cost; (iii) lack of adequate resources and equipment and supplies.

The household income was defined as extremely low, very low, low, and average, based on local expertise and knowledge. The average monthly income for Kano is US$250.

Sample size calculations were not performed for either study, due to non-existent benchmark data. Both studies were based on recruitment duration rather than sample sizes. The intention of this study was to provide preliminary data to inform future work.

Descriptive data are presented both overall and by outcome (livebirth/stillbirth) as frequencies and percentages. For each variable of interest, a univariable logistic regression was used with stillbirth as the outcome. Conditional logistic regression was used for the case-control study, to account for matching. Results are presented as odds ratios (OR), 95% CI, and p-values and are described within their corresponding domain. Additionally, relationships between area type (urban, semi-rural, and rural) and the other variables were investigated through multivariable logistic regressions in two ways. First, the area type and each of the other variables, in turn, were included as main effects to identify the impact of adjusting for the other. Second, only interaction terms were included to identify differential effects. We examined differences in model coefficients without and after accounting for multiple births within the mothers and based on negligible differences chose to not account for these. Statistical analyses were conducted using Statistical Package for the Social Sciences (SPSS) Version 25 and R Version 3.5.1.

The funder had no role in study design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; nor in the decision to submit the paper for publication. The corresponding author confirms that she had full access to all the data in the study and had final responsibility for the decision to submit for publication.

A total of 548 mothers; 274 stillbirths and 274 livebirths were included, as shown in Figure 1. Cases are defined as stillbirths and controls as matched livebirths. Data collection for controls ceased in December 2017, so cases born in this period were matched with controls born in late November 2017.

Of the mothers, 12% (n = 67) were <20 years old and 5% (n = 29) were ≥40 years, 20% (n = 108) were primigravida. The mothers who reported “any disease” accounted for 60% (n = 329), and 40% (n = 219) reported malaria. Over half (53% n = 291) of the mothers reported a minimum education level as secondary school/university, 26% (n = 142) reported limited schooling, and 21% (n = 115) reported no education. The mothers living in extremely low-income households represented 47% (n = 259), 46·4% (n = 254) reported living in low/very low, with only 6% (n = 35) living at average or above household income. Those without access water (defined as anything other than municipal network) represented 69% (n = 380), and 26% (n = 144) reported to live in shacks. Most were from urban (53%, n = 288) or semi-rural settings (37%, n = 203) with 10% (n = 57) residing in a rural setting. Observing deliveries, 77% (n = 424) were full-term (37–42 weeks), 13% (n = 69) were pre-term (<37 weeks) (17), and 10% (n = 55) were post-term >42 weeks) (18). There was no variation in gender distribution; 55.8% (n = 153) of stillbirths were male and 56.6% (n = 155) of livebirths were male (Table 1).

Within the maternal domain, associations with higher odds of stillbirth were identified among younger mothers (<20 years old) (OR: 2.39 95% CI: 1.32–4.29 p = 0.004) compared with those aged 20–24 years old, no formal education, when compared with secondary education (OR: 3.92 95% CI: 2.44–6.29 p < 0.001), and among the mothers who had experienced the previous stillbirth(s) (OR: 2.90 95% CI: 1.90–4.43 p < 0.001) compared with no previous stillbirth. The mothers who presented with “any disease” were three times more likely to deliver a stillbirth (OR: 3.04 95% CI: 2.13–4.34 p < 0.001) compared with those with no reported disease. On exploring specific maternal health conditions, hypertension (inclusive of pre-eclampsia) was the only condition that presented higher odds of stillbirth when compared with those without hypertension (OR: 3.84 95% CI: 1.97–7.49 p < 0.001) (Table 1).

The findings within the household domain associated with higher odds of stillbirth included a household income of average or above and a low household income both compared with an extremely low household income [(average or above: OR: 8.67 95% CI: 3.27–23.01 p < 0.001), (low income: OR: 2.11 95% CI: 1.86–3.21 p < 0.001)]. Those residing in semi-rural or rural locations had higher odds of stillbirth compared with an urban area [(semi-rural: OR: 2.21 95% CI: 1.54–3.19 p < 0.001) (rural: OR: 5.98 95% CI: 3.03–11.80 p < 0.001)] (Table 1).

Pre-term delivery compared with term delivery was associated with higher odds of stillbirth (OR: 11.9 95% CI: 5.31–26.59 p < 0.001) and when comparing membranes rupturing ≤1 h before delivery anything over 4 h prior to delivery was associated with a higher odds of stillbirth [(>4 to ≤12 h OR: 6.02 95% CI: 2.71–13.37 p < 0.001) (>12 to ≤24 h OR: 2.15 95% CI: 1.11–4.16 p = 0.023) (>24 h OR: 6.39 95% CI: 3.17–12.87 p < 0.001)] (Table 1).

The area type was associated with stillbirth independent of the following confounding variables: access to clean water, maternal education level, house type, maternal disease, previous stillbirth, and household income (Supplementary Tables 1–6).

The odds of stillbirth for the mothers living in a semi-rural area with access to clean water were seven times higher than the mothers in an urban area with access to clean water (OR: 7.78 95% CI: 1.73–34.91 p = 0.007) as shown in Figure 2. The interaction between education level and area type indicated that the mothers with no formal education had the highest odds of a stillbirth and this association was amplified for those from rural settings compared with those in an urban environment with secondary school level education (OR: 10.20 95% CI: 3.65–28.55 p < 0.001) (Figure 2). The odds of stillbirth were higher for rural mothers living in a shack compared with the mothers living in an urban area in a house (OR: 8.84 9% CI: 3.14–24.87 p < 0.001) (Figure 2). Living in a rural area with disease(s) was associated with higher odds of stillbirth when compared with those living in an urban area without a disease (Figure 2) (OR: 24.76 95% CI: 6.54–93.73 p < 0.001). In semi-rural and urban areas, previous stillbirth(s) was the largest predictor of subsequent stillbirth. Yet, in rural areas, mothers who had previously experienced a stillbirth(s) and lived in a rural area were over five times more likely to deliver a stillborn baby compared with mothers living in an urban setting and not having had a previous stillbirth (OR: 5.39 95% CI: 1.73–16.88 p = 0.004). Mothers who had not previously had a stillbirth but reported living rurally had higher odds again compared to mothers living in an urban setting and having not had a previous stillbirth (OR: 12.99 95% CI: 4.59–36.71 p < 0.001) (Figure 2). The interaction among area type, household income, and stillbirth found the mothers from rural areas with very low household income were 40 times more likely to deliver a stillborn baby than the mothers living in an urban setting with extremely low income (OR: 40.44 95% CI: 3.61–453.67 p = 0.003) (Supplementary Table 6).

Within S2, 1,468 births were enrolled, comprising of 265 stillbirths and 1,203 livebirths, as shown in Figure 3. The stillbirth incidence was 180/1,000 births. The mothers aged <20 years represented 23% (n = 334) of the cohort, 6% ≥40 years (n = 85), 21% (n = 306) of the mothers were primigravida. The majority reported having “any disease” 62% (n = 916), with 20% (n = 294) reported having malaria. Those with no formal education made up 20% (n = 296), 21% (n = 304) had limited schooling, and 59% (n = 868) were educated to secondary school/university level. With 10% (n = 141) living in a household with average or above income, 45% (n = 663) did not have access to clean water, and 26% (n = 376) live in shacks. The majority lived in urban 53% (n = 781) and semi-rural settings; 38%, (n = 559) with only 9% (n = 128) in rural settings. This study included 788 male neonates; 139 stillbirths, and 678 female neonates; 124 stillbirths, the sex of two stillbirths were undetermined. The majority, 77% (n = 1,137), were delivered at term, 6% (n = 93) pre-term, and 13% (n = 238) post-term (Table 2).

Higher odds of stillbirth within the maternal domain were found among the mothers aged ≥35 years, compared with the mothers aged between 20 and 24 years; (35–39 OR: 1.83 95% CI: 1.11–3.00 p = 0.018) (≥40 OR: 1.84 95% CI: 1.07–3.18 p = 0.027), and the mothers with no formal education compared with those with secondary school/university level (OR: 2.65 95% CI:1.94–3.62 p < 0.001). Mothers who have previously been pregnant, yet pregnancy did not result in a livebirth, had higher odds of stillbirth compared with mothers who had had a livebirth (OR: 2.99 95% CI: 1.21–7.42 p = 0.018). Previous stillbirth(s) was associated with higher odds of subsequent stillbirth (OR: 2.84 95% CI: 2.08–3.88 p < 0.001) when compared with no previous stillbirth. The mothers who reported having “any disease” had higher odds of stillbirth (OR: 1.69 95% CI: 1.23–2.20 p < 0.001) when compared with those with no reported disease. Hypertension was the only medical condition that presented with higher odds of stillbirth when compared with those without hypertension (OR: 1.72 95% CI: 1.17–2.51 p = 0.005) (Table 2).

Within the household domain, findings associated with a higher odds of stillbirth included living in either semi-rural or rural settings compared to living in an urban setting (semi-rural OR: 1.69 95% CI: 1.26–2.25 p < 0.001) (rural OR: 3.11 95% CI: 2.04–4.74 p < 0.001), living in a shack compared to a house (OR:1.95 95% CI:1.40–2.73 p < 0.001). Having no access to clean water, compared to having access to clean water had higher odds of stillbirth (OR: 1.35 95% CI: 1.04–1.77 p = 0.026). Having an extended travel time to the hospital was associated with higher odds of stillbirth; compared with <1 h journey time. Any journey over 1 h had higher odds (1–2 h OR: 1.59 95% CI: 1.13–2.24 p = 0.007) (>2 h OR: 1.91 95% CI: 1.09–3.36 p = 0.024) (Table 2).

Highers odds of stillbirth within the neonatal/birth domain included pre-term delivery (OR: 7.38 95% CI: 4.45–11.48 p < 0.001) compared with term delivery, membranes rupturing between >4 and ≤12 h before delivery (OR: 2.60 95% CI: 1.62–4.18 p < 0.001) compared with <1 h before delivery. Any duration of >12 h before delivery compared with <1 h highlighted a positive trend, yet no statistically significant association (Table 2).

The area type was associated with stillbirth independent of the following confounding variables: access to clean water, maternal education level, type of house, time traveled to the hospital, maternal disease, previous stillbirth, and household income (Supplementary Tables 7–13).

Mothers living in a rural area with no access to clean water were over three times more likely to experience stillbirth than the mothers living in an urban area with access to clean water (OR: 3.3 95% CI: 1.87–5.82 p < 0.001), as shown in Figure 4. The interaction between education level and area type indicated that the mothers with no formal education had the highest odds of stillbirth, and this association was magnified for those in rural settings compared to mothers residing in an urban environment with secondary school education (OR: 5.86 95% CI: 3.33–10.24 p < 0.001) (Figure 4, Supplementary Table 8). The mothers living in a rural area living in a shack were almost four times more likely to have a stillbirth than the mothers from urban areas living in a house (OR: 3.78 9% CI: 2.15–6.64 p < 0.001) (Figure 4). The interaction analysis found that the mothers from rural settings with >2 h travel had the highest odds of stillbirth compared with the mothers living in an urban area with <1 h to travel (OR: 4.58 95% CI: 2.32–9.04 p < 0.001) (Figure 4). The mothers in a rural area with or without disease compared with the mothers who lived in an urban area without disease had higher odds of stillbirth (Figure 4), and those with a disease in a rural area had the highest odds of stillbirth (OR: 4.04 95% CI: 2.22–7.34 p < 0.001). The mothers with previous stillbirth(s) and living in a rural area were 15 times more likely to have a stillbirth compared with the mothers in an urban setting with no previous stillbirth (OR: 15.00% CI: 6.15–36.60 p < 0.001) (Figure 4). The mothers from an extremely low income household and living in a rural environment compared with those from an urban environment with extremely low income had the highest odds of stillbirth (OR: 4.75 95% CI: 2.13–10.92 p < 0.001) (Figure 4).