Menu
Menu
Menu
Menu
Introduction Access to emergency obstetric care can lead to a 45%-75% reduction in stillbirths. However, before a pregnant woman can access this care, she needs to travel to a health facility. Our objective in this study was to assess the influence of distance and travel time to the actual hospital of care on stillbirth. Methods We conducted a retrospective cross-sectional study of pregnant women who presented with obstetric emergencies over a year across all 24 public hospitals in Lagos, Nigeria. Reviewing clinical records, we extracted sociodemographic, travel and obstetric data. Extracted travel data were exported to Google Maps, where typical distance and travel time for period-of-day they travelled were extracted. Multivariable logistic regression was conducted to determine the relative influence of distance and travel time on stillbirth. Results Of 3278 births, there were 408 stillbirths (12.5%). Women with livebirths travelled a median distance of 7.3 km (IQR 3.3-18.0) and over a median time of 24 min (IQR 12-51). Those with stillbirths travelled a median distance of 8.5 km (IQR 4.4-19.7) and over a median time of 30 min (IQR 16-60). Following adjustments, though no significant association with distance was found, odds of stillbirth were significantly higher for travel of 10-29 min (OR 2.25, 95% CI 1.40 to 3.63), 30-59 min (OR 2.30, 95% CI 1.22 to 4.34) and 60-119 min (OR 2.35, 95% CI 1.05 to 5.25). The adjusted OR of stillbirth was significantly lower following booking (OR 0.37, 95% CI 0.28 to 0.49), obstetric complications with mother (obstructed labour (OR 0.11, 95% CI 0.07 to 0.17) and haemorrhage (OR 0.30, 95%CI 0.20 to 0.46)). Odds were significantly higher with multiple gestations (OR 2.40, 95% CI 1.57 to 3.69) and referral (OR 1.55, 95% CI 1.13 to 2.12). Conclusion Travel time to a hospital was strongly associated with stillbirth. In addition to birth preparedness, efforts to get quality care quicker to women or women quicker to quality care will be critical for efforts to reduce stillbirths in a principally urban low-income and middle-income setting.
Lagos State in the southwestern part of Nigeria has diverse geographical terrains (including land and water) and settlement types (including its central megacity, suburbs, slums and towns). While principally urban, Lagos state has some rural parts in its extreme east and west. The state has 20 local government areas and a population of approximately 26 million (estimated in 2019).15 The most common means of travel in Lagos is by road. However, in many parts of the state, the road infrastructure is poor, evidenced by presence of multiple potholes that sometimes make roads impassable for commuters. Severe traffic congestions are a common feature, with flooding during the rainy season making road conditions even worse. Road renovations are at best a stopgap and sometimes cause even more travel disruptions.16–18 Our study was a statewide multifacility retrospective cross‐sectional study that identified pregnant women who presented with obstetric emergencies at one of the 24 public hospitals with capacity for 24/7 all provision of EmOC services in the state. These 24 public hospitals include 20 general hospitals which are secondary health facilities with either a general obstetric unit or a dedicated Maternal Childcare Centres and four teaching hospitals which are tertiary health facilities (Details of the hospitals are in online supplemental table 1). According to the Health Facility Monitoring and Accreditation Agency, there are 1329 accredited private hospitals in Lagos. However, government health facilities manage 42% of deliveries in the state, while private health facilities take up about 28%.19 Two studies that estimated institutional SBR in Nigerian public hospitals reported 39.6 and 61.8 per 1000 births, respectively.20 21 bmjgh-2021-007052supp001.pdf We collected data from all 24 public hospitals over a 6-month period. The data collection team comprised consultant obstetricians, resident doctors and medical officers who had clinical experience working in the obstetric units of the hospitals and were familiar with the patient records system in Lagos public health facilities. The data collection team members were all trained on the standard operations protocol to guide data collection and ensure consistency across the different hospitals, use of the pretested online data collection tool in Google Forms and the ethical procedures guiding the research. From clinical records of all pregnant women with gestational age of 28 weeks or more who presented with an obstetric emergency between 1 November 2018 and 30 October 2019, we obtained data on sociodemographic characteristics, obstetric history, travel to reach the health facility (including day of travel and period-of-day when journey to the facility commenced, street name of women’s self-reported place of residence, referral facilities if any, the destination facility (one of the 24 public hospitals)), obstetric complication managed, mode of delivery and pregnancy outcomes. All pregnant women who presented at the obstetric emergency room and had a live or stillbirth at or after gestational age of 28 weeks were included. For the outcome, stillbirth, we aligned with WHO’s definition applicable in many LMIC settings defining a stillbirth as a baby born with no signs of life at 28 weeks of pregnancy or more.4 We excluded 51 cases with perinatal deaths that occurred after the baby was born alive (early neonatal deaths), because these deaths may have more to do with quality of care, as opposed to travel to the health facility. We excluded a further 22 cases which had missing data regarding the outcome of the pregnancy or gestational age could not be established. All recorded data captured in Google Forms was subsequently exported as a Microsoft Excel file. Additional data gathering on distance and travel time were required to fully characterise the travel to reach the hospital. Studies that estimated distance and travel time of pregnant women to reach EmOC facilities in LMIC settings have mostly been based on women’s self-reports or spatial models,10 with the accuracy of both approaches questioned by several authors.22–25 Compared with spatial model estimates, distance and travel time estimates using global positioning satellite navigation software like Google Maps (Alphabet, Mountain View, California, USA) have been shown to be closer to reality in an LMIC urban setting.26 Building on this evidence, we georeferenced the place of residence, referral points and destination facility for each woman in Google Maps, based on the data extracted from their clinical records. For addresses that were not discoverable on Google Maps, we contacted local persons who were well acquainted with the neighbourhoods to check for any spelling errors and reattempted to locate the street. For pregnant women with traceable journeys (meaning location of the home address and all referral points were known), we extracted distance (in kilometres (km) and travel time (in minute (min) for their journeys from Google Maps using the ‘typical time of travel’ tool for the time and day that the woman commenced her journey to the hospital, as per data extracted from the clinical records. To collect travel time estimates for the period-of-day of travel, we used specific time slots (9:00, 15:00, 18:00 and 21:00 hours for morning, afternoon, evening or night journeys, respectively). In cases in which this data was not available (27% of the sample), we assumed the woman travelled in the afternoon (15:00 hours), as it was a midpoint estimate between the two known travel peak periods in Lagos (6:30 and 11:30 hours (morning peak period) and 15:00 and 19.30 hours (evening peak period)).27 For means of transport, we assumed that all pregnant women used motor vehicle, since private cars (25%) and taxis (21%) are the most popular means of transport to health facilities, emergency or otherwise, and is almost always the transport means used by pregnant women in emergency situations in Lagos, especially as motorcycles have been banned.9 28 In cases in which it was not possible to find specific points of travel of the women, we labelled the case as untraceable (4% of the sample). Following data cleaning in Microsoft Excel (Microsoft Corporation, Redmond, USA), we used the extracted geocoordinates to map and visually inspect places of origin of all women with stillbirths relative to the location of public hospitals and produced maps in ArcGIS 10.6 (Esri, Redlands, California, USA). We then conducted descriptive analysis for all theoretically relevant sociodemographic and obstetric characteristics, travel path to facility and mode of delivery, indicating frequencies and percentages for categorical variables. The mean and SD as well as median and IQR for distance and travel time were computed. For interpretation, priority was given to the median values as these are known to be robust to the outliers.29 All continuous variables were subsequently converted into categorical variables. We conducted bivariate logistic regression to test the null hypothesis that there is no statistically significant association between each of the independent variables and stillbirth. The stillbirths were subsequently categorised into fresh and macerated stillbirths as extracted from patient records, as we theorised that the fresh type was more likely related to travel, since these occurred after the onset of labour but before birth, when the woman would have been en route to a health facility.4 Where there were discrepancies in stillbirth classifications for multiple gestations (eg, one fresh and one macerated), the stillbirth status of the first baby was used in the classification. Finally, we conducted multivariate logistic regression to determine the relative influence of the independent variable categories on stillbirths while controlling for other variables. The logistic regression models were built stepwise incorporating variables that showed a statistically significant association with stillbirths as an outcome in the bivariate analysis. Four models were fitted. Model I incorporated only significant sociodemographic and obstetric variables, model II added travel distance to significant sociodemographic and obstetric variables, model III added travel time to significant socio-demographic and obstetric variables and model IV included both travel time and distance to significant sociodemographic and obstetric variables. We reported both p values and 95% CIs of ORs derived from regression coefficients to show strength of evidence and considered differences between observations as statistically significant when the p value was <0.05. We also conducted a subgroup analysis by referral status and by stillbirth category (fresh and macerated). Missing data were excluded from the analysis. We conducted all statistical analysis in STATA SE V.15.0 (StataCorp). Patients and/or the public were not involved in the design, conduct, reporting or dissemination of this research.
N/A
