An assessment of geographical access and factors influencing travel time to emergency obstetric care in the urban state of Lagos, Nigeria

Previous efforts to estimate the travel time to comprehensive emergency obstetric care (CEmOC) in low- and middle-income countries (LMICs) have either been based on spatial models or self-reported travel time, both with known inaccuracies. The study objectives were to estimate more realistic travel times for pregnant women in emergency situations using Google Maps, determine system-level factors that influence travel time and use these estimates to assess CEmOC geographical accessibility and coverage in Lagos state, Nigeria. Data on demographics, obstetric history and travel to CEmOC facilities of pregnant women with an obstetric emergency, who presented between 1st November 2018 and 31st December 2019 at a public CEmOC facility were collected from hospital records. Estimated travel times were individually extracted from Google Maps for the period of the day of travel. Bivariate and multivariate analyses were used to test associations between travel and health system-related factors with reaching the facility >60 minutes. Mean travel times were compared and geographical coverage mapped to identify ‘hotspots’ of predominantly >60 minutes travel to facilities. For the 4005 pregnant women with traceable journeys, travel time ranges were 2-240 minutes (without referral) and 7-320 minutes (with referral). Total travel time was within the 60 and 120 minute benchmark for 80 and 96% of women, respectively. The period of the day of travel and having been referred were significantly associated with travelling >60 minutes. Many pregnant women living in the central cities and remote towns typically travelled to CEmOC facilities around them. We identified four hotspots from which pregnant women travelled >60 minutes to facilities. Mean travel time and distance to reach tertiary referral hospitals were significantly higher than the secondary facilities. Our findings suggest that actions taken to address gaps need to be contextualized. Our approach provides a useful guide for stakeholders seeking to comprehensively explore geographical inequities in CEmOC access within urban/peri-urban LMIC settings. The study was conducted in Lagos State, southwest Nigeria. The Lagos State Bureau of Statistics estimated that 25.6 million people resided in Lagos State in 2019: a density of 6871 residents per square kilometre (km) (LASG, 2019). The state is further divided into 20 local government areas (LGA). Lagos state is highly urbanized and has a mix of different geographical terrains, including city and suburb, metropolis and slums, as well as land and riverine areas. The central areas form the Lagos metropolis, which is surrounded by several suburbs. In contrast, the extreme western and eastern parts of the state are made up of less built-up towns (Figure 1). Map of Lagos state showing variation in geographical settlements and location of the public CEmOC facilities. Compared to the national maternal mortality ratio (MMR) of 512 maternal deaths per 100 000 live births (year—2017) (National Population Commission, ICF International, 2019), MMR in Lagos State has been estimated as 450 (95% CI 360–530) per 100 000 live births (Oye-Adeniran et al., 2011). When disaggregated by LGAs, MMR ranges from 356 per 100 000 live births in Ikeja LGA to 826 per 100 000 live births in Alimosho LGA. Similar to global patterns, hypertension, spontaneous abortions and ectopic pregnancies were the most commonly reported causes of death during pregnancy, while haemorrhage and prolonged or obstructed labour were more commonly reported as causes of death during childbirth in Lagos (Odeyemi et al., 2014; Okonofua et al., 2017). In terms of available CEmOC facilities, there are 24 public CEmOC facilities in Lagos State, including 20 secondary health care facilities (general hospitals/maternal childcare centres) and four tertiary health care facilities (teaching hospitals/apex referral centres), all of which are expected to provide CEmOC services 24 hours a day (Figure 1 and Supplementary Data). There is also a complement of three military hospitals and about 35 private hospitals that can be classified as CEmOC facilities with specialists who can provide CEmOC services 24 hours a day, as per the database of the State Ministry of Health. However, for this study, we focus only on public sector hospitals providing CEmOC, as they form the bedrock of universal health coverage in LMICs (Sachs, 2012). In any case, as per 2018 Nigeria Demographic Health Survey (NDHS), excluding home delivery (59.0%), twice the number of women in Nigeria deliver in public hospitals (26.4%) compared to private hospitals (13.0%) (National Population Commission, ICF International, 2019). Data for this study were collected over 6 months and based on a review of patient records of all pregnant women who presented in the obstetric emergency rooms of all 24 public CEmOC facilities in Lagos state with any major pregnancy and childbirth complication between 1st November 2018 and 30th October 2019. However, some facilities were being built or renovated during this one-year study period. First, the Institute of Maternal and Child Health (IMCH, commonly referred to as Àyìnkę House) was only re-opened for service after a 9-year closure for renovation on 24th April 2019 (Okoghenun, 2016; Ugvodaga, 2019). As such, we could only extract 3 months of data from the facility (1st July 2019 to 30th September 2019). Second, Eti-Osa Maternal and Child Care Centre (MCC) was newly built and commissioned (Bassey, 2019). Data collected from this facility were for the period 1st September 2019 to 31st December 2019. The data were mined from the records by members of the research team supported by trained research assistants who were qualified medical doctors conversant with the patient records system in Lagos public health facilities. Using a pre-tested data extraction tool, we collected data on demographic characteristics, obstetric history, travel to reach the health facility (month of the year, day of the week—weekday or weekend and period of day when the journey to the facility commenced—morning, afternoon, evening or night), street name of women’s self-reported start location (place of residence, unless otherwise stated), other facilities visited en route (referral points) if any and the destination facility (one of the 24) from clerking notes recorded in the patient folders. In cases where clarification was needed, we solicited the support of the medical doctors in charge of the obstetric emergency room. We geo-coded the origin, any facilities visited along the way (referral points) and destination locations for each woman. The points of origin were based on the stated street name of the women’s self-reported start location, most commonly her home. We used Google Maps to find the exact location and selected the relevant coordinates if the street was discoverable on the platform. For streets that were difficult to find, we used local persons who were familiar with the various communities to check for any spelling errors and re-attempted to locate the street. If, despite our best efforts, we could not locate the street, the record of the woman, along with those that did not have an address marked in their patient files, was labelled untraceable (4% of the sample). We identified the exact entry point of the obstetric emergency ward for each destination CEmOC facility by visiting the obstetric emergency ward of each facility and geocoding its location. Geographical coordinates of the CEmOC facilities were collected using a free mobile application, ‘Easy GPS’ (TopoGrafix, Stow, Massachusetts, USA), which automatically logged longitude and latitude values of the CEmOC facilities (see Supplementary Data). In cases in which pregnant women went to a referral point on her path to a CEmOC facility, we used the same approach as was used for geocoding the points of origin to geocode such referral points. However, for those who had multiple referral points, we only traced their journeys from their places of residence to the facility from where they were referred to the final destination. Stopovers made to informal settings (e.g. church or mosque) were not geo-coded. For pregnant women whose journeys could be traced, estimated travel time between the origin and destination (including referral points) were extracted from Google Maps using the ‘typical time of travel’ tool for the time and day that the woman commenced her journey to the CEmOC facility. Motorized vehicle was used as means of transport in Google Maps, as private cars (25%) followed by taxis (21%) are the most popular means of transportation to health facilities in Lagos, as per the 2018 NDHS (National Population Commission, ICF International, 2019). While journeys that required travel by boat were identified (0.14% of the sample), these could not be traced on Google Maps. To collect travel time estimates for the period of the day when journey to the facility commenced, we used 9.00 a.m., 3.00 p.m., 6.00 p.m. and 9.00 p.m. time slots for morning, afternoon, evening or night journeys, respectively. For journeys in which we could not tell the time of the day that women commenced their journeys to the facility (33% of the sample), travel time was extracted for the afternoon (3.00 p.m.), as it offered a middle-ground estimate in between the two known peak periods for travel in Lagos (6.30 a.m. and 11.30 am—morning peak period and 3.00 p.m. and 7.30 pm—evening peak period) (Asiyanbola et al., 2012). Most up to date (as of 2017) shapefiles capturing administrative boundaries, population, road networks, and water bodies within Lagos were retrieved from the State’s Ministry of Urban and Regional Planning. These files formed the platform for which the geographical analysis was conducted. Categorical variables, which included demographic data, obstetric and travel history of the included women, were summarized using frequencies and proportions and presented in summary tables. Continuous variables were summarized using means and medians with their interquartile ranges (IQR). Individual-level, pregnancy-related and health systems-related factors as groups of independent factors that can be associated with travel time (Sacks et al., 2016; Geleto et al., 2018; Ahmed et al., 2019; Banke-Thomas et al., 2019). Individual-level and pregnancy-related factors relate to socio-demographic and obstetric history, respectively, while health system factors comprise referral, skilled health personnel and type of facilities providing care. In addition, the season, day, period of the day when the journey took places and road conditions may also impact women’s total travel time to reach care. As our study objective was focused on system-level factors, we did not report individual-level and pregnancy-related factors as part of our analysis. We used Chi-square test (bivariate analysis) to test the null hypothesis that there is no association between day, period of day or health systems factors with reaching (or not reaching) the destination CEmOC facility within the 60 minutes. The choice of 60 minutes as benchmark, as opposed to 120 minutes, was based on the established evidence that pregnant women with obstetric emergencies can escalate in less than 2 hours (Khan and El-Rafaey, 2006; UNFPA, 2012) and that further delays could occur upon reaching the CEmOC facilities (Thaddeus and Maine, 1994; Gabrysch and Campbell, 2009). As such, a narrower window of travel will be helpful for effective service planning and policymaking decisions. In any case, other authors have used the 60 minute travel time benchmark for analysis (Chowdhury et al., 2017; Niyitegeka et al., 2017; Ouma et al., 2018). For our analysis, associations between the independent and dependent variables were tested at a 95% confidence interval (CI), with a P-value of significance set at ≤0.05. Multivariate analysis was conducted to identify the factors associated with the travel time category to reach health facilities. Using the actual travel time and the distance to reach each facility, a linear regression model was conducted to show any statistically significant differences in mean travel time and distance to CEmOC facilities. Linear regression was also conducted to compare mean travel time and distance to CEmOC facilities for women living in the area surrounding the newly established facility before and after its commissioning. In cases where specific data were not retrieved from the patient records, such missing data were excluded from the analysis. All statistical analyses were done using STATA SE 15.0® (StataCorp, College Station, Texas, USA). For visualization, we visually identified locations of high concentration of long travel time of >60 or >120 minutes to the destination facility. Data points were disaggregated by day of the week, period of the day, and referral. All maps were drawn with the ‘gg’ package, including the tile server for Stamen Maps, in R version 4.0.2 (R Development Core Team, Auckland, New Zealand). Data layers were projected into the spatial reference frame, WGS84/ UTM Zone 35S. Ethical approval for this study was obtained from the Research and Ethics Committees of the Lagos State University Teaching Hospital (LASUTH) (LREC/06/10/1226) and Lagos University Teaching Hospital (LUTH) (ADM/DCST/HREC/APP/2880). Social approval for the study was received from the Lagos State Government (LSHSC/2222/VOLII/107). As this study was based on patient records, we minimized the risk of patient identification by not collecting data on patient names and specific street numbers. In mapping, we selected the mid-point of streets of origin to ensure anonymity. Random displacements of the sort are typically used in similar large surveys (Burgert et al., 2013).