Main

Global progress in reducing maternal mortality has stagnated since 2016 and it is unlikely that the Sustainable Development Goal target of 70 maternal deaths per 100,000 livebirths will be achieved by 20301. The highest maternal mortality ratio is in sub-Saharan Africa (SSA), with 545 deaths per 100,000 livebirths, accounting for 70% of all maternal deaths globally. The highest risk of maternal and perinatal death is at the time of childbirth from causes that include severe bleeding, hypertensive disorders, obstructed labor and sepsis2. Most of the nearly 300,000 maternal and 5 million perinatal (stillbirths and newborn) deaths can be prevented if women receive adequate care during pregnancy and give birth in facilities that are able to manage complications3.

Half of maternal deaths and three-quarters of intrapartum stillbirths are preventable with timely access to high-quality emergency obstetric care4. Availability of basic and comprehensive services within 2–3 hours is currently considered a reasonable standard5,6. Substantial efforts have been directed to characterize travel time and distance, locate hotspots and vulnerable areas for interventions7, and understand the role of physical access in maternal and perinatal health outcomes8,9.

Most research on geographic accessibility to health facilities has focused on rural areas9,10,11,12,13 due to the role of geographic distance, lack of roads and transport, and lower density of health facilities. However, urban settings face specific challenges in ensuring accessible, equitable and high-quality maternal and newborn care. Two-thirds of the world’s population is expected to live in urban areas by 2050. Nearly 90% of these additional 2.5 billion urban residents will concentrate in Africa and Asia14. Large-scale rural-to-urban migration leads to informal settlements, exacerbation of inequalities, and the spread of infectious and non-communicable diseases. Emerging evidence shows that the urban health advantage in survival is diminishing in SSA countries15,16.

There is a recognition that potentially long travel times are obfuscated by short geographic distances to health facilities in urban areas due to poor road infrastructure and traffic congestion8,10,11,12. Several studies have been conducted in urban areas of low- and middle-income countries, including 15 cities in Nigeria13,14,15, Nairobi in Kenya16 and Cali in Colombia17,18. Two studies focused on disaggregating accessibility metrics by the degree of urbanization19 and informal settlements20 across SSA. While useful for national and regional comparisons, both studies used a database of public health facilities curated between 2012 and 201821.

So far, there has been no comprehensive evaluation of geographic accessibility to facilities offering childbirth care nor linking of these metrics to socio-economic indicators in major urban areas of Guinea. This is despite Guinea having a very high maternal mortality ratio of 553 per 100,000 livebirths1, equivalent to more than 2,500 maternal deaths in 2020. A modeling study reported that about 40% of women were marginalized from hospital care, that is, living outside a 2-hour radius from the nearest facility. This was based on generic (non-specific) travel speeds that were similar in both rural and urban areas20.

In this study, we aimed to comprehensively assess geographic accessibility to childbirth care in the Grand Conakry metropolitan area in the Republic of Guinea (Figs. 1–3). Specifically, our objectives were to (1) estimate travel time to facilities offering childbirth care, disaggregated by facility level and sector, using a least-cost path algorithm at a 10-m spatial resolution, (2) compute the percentage of women of child bearing age (WoCBA) and pregnant women living within 15, 30 and 60 minutes of the nearest such facility, and (3) assess the intersection between geographic marginalization and poverty in the metropolitan area. For policy relevance, the estimates are also presented for subnational units (commune and district).

Fig. 1: Communes and population density in the Grand Conakry metropolitan area, Guinea.
figure 1

Map showing the population density at the 100 m grid level, ranging from low (1.8 people per 100 m grid) to high (455.3 people per 100 m grid). Commune boundaries are delineated in black. Inset: map highlighting the location of the Grand Conakry area within Guinea. The map was plotted by the authors in ArcGIS Pro (v3.3.1, Esri). The Grand Conakry boundaries were digitized based on secondary data, as described in the Data availability section. The base layer of the Conakry boundary was obtained from geoBoundaries under a Creative Commons license CC BY 4.0 (ref. 37); the Guinea boundary was from OCHA West and Central Africa (ROWCA), Humanitarian Data Exchange (HDX) under a Creative Commons license CC BY 3.0 (ref. 38).

Full size image
Fig. 2: Flowchart for modeling travel times to health facilities and assessing inequalities.
figure 2

The analytical flowchart used to estimate travel times to health facilities in Grand Conakry, Guinea. The flowchart outlines the sequence of steps for modeling travel times to health facilities and integrating these estimates with population distribution, relative wealth index and administrative boundaries.

Full size image
Fig. 3: Spatial distribution of health facilities in Grand Conakry included in the analysis.
figure 3

Map showing the health facilities that offered childbirth care by level and sector in 2023. The map was plotted by the authors in ArcGIS Pro (v3.3.1, Esri). The Grand Conakry boundaries were digitized based on secondary data, as described in the Data availability section. The base layer of the Conakry boundary was obtained from geoBoundaries under a Creative Commons license CC BY 4.0 (ref. 37); the Guinea boundary was from OCHA West and Central Africa (ROWCA), Humanitarian Data Exchange (HDX) under a Creative Commons license CC BY 3.0 (ref. 38).

Full size image

Results

The primary data used to compute travel speeds were based on 102 trajectories. Over 1,255 km were covered in total, with a median of 11.9 km per trajectory (lower quartile = 5.5 km, upper quartile = 18.5 km). Estimated travel speeds, by road type, are shown in Table 1. Overall, travel speeds in Grand Conakry were low, regardless of road type, all under 60 km h−1. Average speeds ranged from 14 to 28 km h−1, depending on road type. There was large variability in speeds, spanning from just above 2 km h−1 (highest traffic levels, worst scenario) to 60 km h−1 (lowest traffic levels, best scenario).

Table 1 Travel speeds based on travel trajectories in Grand Conakry across different road type and land cover type
Full size table

We derived travel times to the nearest facility for 20 scenarios, summarized by the average speed, the minimum and maximum speeds (extreme margins), and the lower and upper quartiles (moderate margins) for the Grand Conakry metropolitan area, and by district and commune. We report the results of two scenarios (the best and worst scenarios) in Table 2 and Fig. 4. The data for the remaining 18 scenarios are presented in Supplementary Table 1.

Fig. 4: Travel time to the nearest healthcare facilities in Grand Conakry.
figure 4

The upper panels show the travel time to the nearest public and private health facilities, while the lower panels focus on public hospitals only. The maps are based on travel speeds calculated for three scenarios: average, maximum and minimum travel speeds. Travel time is categorized into six intervals: ≤5, >5–15, >15–30, >30–60, >60–120 and >120 minutes. Dark red represents longer travel times, indicating regions with limited healthcare accessibility. The map was plotted by the authors in ArcGIS Pro (v3.3.1, Esri). The Grand Conakry boundaries were digitized based on secondary data, as described in the Data availability section. The base layer of the Conakry boundary was obtained from geoBoundaries under a Creative Commons license CC BY 4.0 (ref. 37); the Guinea boundary was from OCHA West and Central Africa (ROWCA), Humanitarian Data Exchange (HDX) under a Creative Commons license CC BY 3.0 (ref. 38).

Full size image
Table 2 Travel times to the nearest health facilities for different speed scenarios in Grand Conakry
Full size table

At the pixel level, on average, the travel time to the nearest (public or private) health facility offering childbirth care ranged from 0 to 35 minutes. The longest estimated travel time was 126 minutes for the lowest speed. When these estimates were aggregated at the conurbation level, the average travel time was 8 minutes. This increased fourfold to 30 minutes for the minimum speed. Across the 14 communes, the average travel time varied widely from 3 minutes (8 minutes at the lowest speed) in the Conakry communes of Kaloum, Matam and Dixinn to 15 minutes (19 minutes at the lowest speed) in Manéah in the Coyah district (Table 2).

In the most restricted scenario estimating travel to the nearest public hospital, the average travel time ranged from 0 to 68 minutes at the pixel level and increased to 222 minutes at the lowest speed. The aggregated average travel time at the conurbation level was 22 minutes, while the longest estimated travel time was 80 minutes at the minimum speed. Across the communes, the average ranged from 5 minutes (inner communes of the Conakry district) to 33 minutes in some communes in the Coyah and Dubréka districts. At the minimum speed, the travel time to the nearest public hospital exceeded 60 minutes in 7 of the 14 communes of Grand Conakry (Table 2).

Table 3 shows the geographic coverage of WoCBA within specified travel time thresholds (15, 30 and 60 minutes) of the nearest health facility and public hospital by district and commune. Further results, including the geographic coverage of pregnant women, are presented in Supplementary Tables 2 and 3. Approximately 94% of WoCBA living in Grand Conakry were within 15 minutes of the nearest health facility based on the average speed scenario, and all WoCBA were within the 30 minutes threshold. This coverage reduced to 32% (15 minutes) and 70% (30 minutes) for the minimum travel speed scenario (Table 3).

Table 3 Geographic coverage of WoCBA within 15, 30 and 60 minutes of any health facility (best scenario) and any public health hospital (most restricted scenario) by commune within Grand Conakry
Full size table

At the commune level, at the average speed, the geographic coverage at the 15-minute threshold was 100% in 9 of the 14 communes, all within the Conakry district. In the minimum speed scenario, this coverage declined to <30% in five communes with a further three communes having <50% coverage. At the minimum speed, the percentage of WoCBA within 30 minutes of the nearest health facility providing childbirth care varied from 31% in Manéah (Coyah District) and 40% in Kagbélen (Dubréka District) to 100% in several communes in the Conakry district.

Considering the geographic coverage of public hospitals, 44%, 82% and 100% of WoCBA were within 15, 30 and 60 minutes, respectively, of the nearest public hospital at average speed. These percentages dropped to 2%, 7% and 30% for the minimum speed scenario. Substantial variations across the communes were noted across the three (15, 30 and 60 minutes) thresholds. For example, 0% WoCBA in Sanoyah (Coyah) and Kagbélen (Dubréka) were within 15 minutes of a public hospital, increasing to 67% (Gbéssia) and 100% in Kaloum, Matam, Dixinn and Ratoma (districts of Conakry). In addition, all WoCBA in Kaloum, Matam and Dixinn were within 60 minutes of the nearest public hospital, regardless of the speed. However, <2% of WoCBA living in the communes of Sanoyah (Coyah) and Kagbélen (Dubréka) were within 60 minutes of the nearest public hospital in the minimum speed scenario.

Figure 5 shows the distribution of travel time by wealth quintiles for three travel speed scenarios (minimum, average and maximum), by the four types of health facilities and in the three districts. The results for the lower and upper quintile speeds are presented in Supplementary Fig. 2. The data show pro-rich inequalities that vary in magnitude by facility type and travel speed. In Grand Conakry, the inequality was largest for the minimum speed, followed by the average and maximum speeds. Inequalities in travel time to hospitals were larger than to any facility offering childbirth care. For example, travel time to the nearest public facility was 22 minutes for the fifth quintile (Q5, richest) and 52 minutes for the first quintile (Q1, poorest), equating to a difference of 30 minutes, while travel time to the nearest public hospital was 54 minutes for Q5 and 105 minutes for Q1 for the minimum speed, a difference of 51 minutes.

Fig. 5: Variation in travel time with the relative wealth index in Grand Conakry.
figure 5

a–c, Equiplots of travel time (geographic accessibility) by relative wealth index for different travel speeds, facility type and subnational areas, highlighting socio-spatial disparities. Travel times at minimum speeds (a), average speeds (b) and maximum speeds (c) are displayed for public and private facilities, public facilities only, public and private hospitals, and public hospitals only across four areas: Grand Conakry, Conakry, Coyah and Dubréka. Each data point represents the median travel time stratified by wealth quintiles (Q1 (poorest), Q2, Q3, Q4 and Q5 (richest)). The colors of the data points correspond to wealth quintiles.

Full size image

When disaggregated by district using equiplots, the level of pro-rich inequality varied, with Conakry having the smallest differences between the richest and poorest quintiles by travel speed scenario, sector (private and public) and facility level (hospital and lower; Fig. 5). The least inequalities were observed when the estimated travel time to the nearest public/private facility was based on maximum speed, with all five wealth quintiles having estimates of <10 minutes in the three districts. The largest inequality in travel time to the nearest public/private health facility with minimum speed was in Coyah, with 21 minutes of travel for Q5 (richest) increasing to 53 minutes for Q1 (poorest).

Restricting the analysis to the nearest private/public hospital increased inequalities and reversed the direction of inequalities in some districts. For example, the shortest travel time to the nearest private/public hospital was consistently observed for Q4 in Dubréka, while the longest was for Q1 (75 minutes versus 143 minutes for the minimum speed). Inequalities generally increased when limiting the destination to public hospitals in the three districts.

Discussion

In this analysis, we assessed the spatial accessibility and geographic coverage of health facilities providing childbirth care based on geotraced travel speeds in the Grand Conakry metropolitan area in the Republic of Guinea. We identified spatial heterogeneities in access by sector and facility level and linked these to the relative wealth index. Most WoCBA needed on average 15 minutes to reach the nearest facility offering childbirth care in any of the communes. However, under heavy traffic conditions, this increased to about 50 minutes. Under these conditions, accessibility declined rapidly when WoCBA were in need of affordable comprehensive emergency obstetric care, mostly available in public hospitals, with average travel times exceeding 2 hours in communes of the Dubréka district.

All WoCBA in Grand Conakry appear to reside within a travel time of 30 minutes to the nearest facility. However, when the traffic is heavy, this may decline to less than 60% in 2 of the 14 communes. Similarly, despite all women being within an hour of a public hospital, in 6 communes, only 3 in 10 women were within the same threshold under the lowest travel speeds. Private health facilities contributed considerably to increasing geographic coverage, especially in the peri-urban Coyah district. Finally, while we have shown that women from wealthier areas were, on average, closer to any health facility, paradoxically, poorer women in Coyah were closer to public hospitals.

Compared with travel time in other cities in SSA, such as in Nigeria13,22, the estimates from Grand Conakry are not substantially different from average travel speeds. Spatial access to public comprehensive emergency obstetric care varied between 10 and 41 minutes across 15 Nigerian cities, while in Grand Conakry, travel time to public hospitals varied between 5 and 33 minutes across the 14 communes13,22. Similarly, under high traffic conditions, 71% of WoCBA were within 1 hour of a childbirth care facility in Grand Conakry, while this number ranged from 83% to 100% across the 15 most populated cities in Nigeria13. Our study has several applications and implications for research and implementation to improve access to facility-based childbirth care.

Irrespective of the approach used to estimate travel time in SSA23, a frequent issue is the lack of observed data that characterize journeys to facilities offering childbirth care to parametrize travel time models24. Among the observed data is travel speed, arguably the most important factor contributing to the magnitude of the estimated travel time. Due to a lack of observed travel speeds, most studies rely on generic speeds7,14,25,26. Overall, the travel speeds that we measured in the Grand Conakry metropolitan area were lower than those previously described in the literature7,26. Travel speeds within Grand Conakry did not vary substantially between main road types such as trunk and primary roads, with an average speed of less than 28 km h−1 and a maximum speed of 60 km h−1. Yet, average speeds of 100 km h−1 have been used previously for major roads7,26. Therefore, it is likely that previous estimates of geographic accessibility to healthcare that included Grand Conakry underestimated travel time.

The low speeds observed in Grand Conakry could be explained by its geographical configuration. Grand Conakry is among the African metropolises facing the most complex challenges regarding land use and transport, primarily due to its geographical location and linear shape27. It is a peninsula, extending 40 km in length and barely 5 km in width, with spatial growth constrained by Mount Kakoulima to the east and mangroves to the west, north and south. Furthermore, administrative functions are predominantly centralized within the commune of Kaloum, the narrowest part of the city, with a few decentralized offices in Dixinn and Matam. Consequently, during rush hours, a substantial portion of the population either travels to or departs from Kaloum, which is the most constrained area of the capital in terms of road infrastructure. This results in traffic congestion that can last for hours during peak times and beyond.

Under heavy traffic conditions, travel times to the nearest facility and nearest public hospital offering childbirth care were approximately 30 and 80 minutes, respectively. There was high heterogeneity within the metropolitan area, with travel times ranging from 12 to 48 minutes for all facilities and from 21 to 123 minutes for public hospitals by commune. Women residing in these communes are within the recommended 2-hour radius, as proposed by strategies for ending preventable maternal mortality5,28. However, at high resolution, some pixels fell outside the 2-hour threshold, highlighting the need to prioritize these hotspots. The communes in the Dubréka and Coyah districts had the longest travel times and least coverage of WoCBA within either 30 or 60 minutes of the nearest facility, especially under heavy traffic conditions. There could be a number of reasons for this.

First, the geographic distribution of health facilities is skewed, with these two districts having fewer facilities and a substantial number of WoCBA. For instance, Kaloum (Conakry district) represents 0.15% of the population of Grand Conakry but hosts 6.9% of all facilities (18% of all public hospitals). In contrast, Kagbélen (Dubréka district), with over 17% of the population, has only 7% of the facilities (no public hospital), and all primary level facilities are concentrated in the eastern part of the commune. Even more striking are the Sanoyah and Manéah communes (Coyah district), each representing about 10% of the population and each hosting only one public health facility (1.2%) and no private facilities. These disparities are even wider when focusing on public hospitals, which provide lifesaving obstetric care functions such as caesarean section and blood transfusion free of charge. In fact, over half of the nine referral hospitals (55.6%) are in the three smallest communes of Grand Conakry where only 1.3% of WoCBA reside.

Second, the districts with poor geographic access metrics are in the peri-urban or relatively newly urbanized areas. Kaloum is the oldest urbanized commune, while Kagbélen, Sanoyah and Manéah are newer and represent a key urbanization frontier where growth happens. These peri-urban areas have a poor road network, making it harder to reach the health service providers where roads are not well developed.

Third, the assessment of geographic access to healthcare in peri-urban areas is further compounded by the fact that open data on road networks are likely to be incomplete due to less volunteered geographic information in the suburbs14 compared with the developed core urban area of Conakry city. The disadvantage recorded in the peripheral areas compared with in the inner city has also been observed in cities across Nigeria13 and Cali, Colombia18.

Fourth, one of the communes in the peri-urban areas (Kagbélen) is an industrial zone with numerous cement factories and a poor road network predominantly consisting of tertiary and residential roads frequented by many trucks. Long travel times result in delays in accessing life-saving care for both women and their babies in an appropriate environment. Finally, in these peri-urban areas with poor road connectivity, some sections of the journey that entail walking are at a high elevation based on the topography, thus increasing the difficulty of walking, which could be particularly challenging for pregnant women before and/or during labor or for those who are referred.

Private facilities improve geographic access to care in some of the communes, although they do not provide free care. For example, in the Kagbélen commune in the Dubréka district, about 45% of WoCBA were within 30 minutes of public hospitals; however, when both public and private hospitals were considered, the coverage improved to 68%. In 2022, the private sector accounted for 2% of the overall health infrastructure in Guinea. However, the distribution of health infrastructure is different within urban areas. In Grand Conakry, 28% of the health facilities providing childbirth care are private and more than half (54%) of the hospitals are private. According to the Ministry of Health, these private facilities provide about 30–40% of maternal, neonatal and infant care29.

The high number of private health facilities in Grand Conakry could provide alternatives and reduce the travel time needed to access childbirth care. However, this has implications for the geographic and financial accessibility of such care. The issue of quality of care, including the application of national guidelines and reporting into the routine data system by the private facilities, becomes even more critical given their importance.

The travel time to any health facility was shortest for the wealthiest 20% of women, irrespective of the speed scenario, indicating pro-rich inequalities. These inequalities were smaller in Conakry, while WoCBA in the peri-urban communes faced further challenges and were particularly more vulnerable, facing widespread inequalities. Similar findings have been observed in Nigerian cities15 and in Cali, Colombia16. While the same pattern was observed for public hospitals, there were a few exceptions where paradoxically the poorer were closer to public hospitals in Grand Conakry. This can be explained by the fact that public hospitals are often located in historical urban centers where the poorest residents live. In contrast, those in the top three wealth quintiles tend to move to newly urbanized areas with modern buildings, less crowding, reduced noise, better sanitation systems and schools. In Conakry, this pattern is evident in communes such as Ratoma, Lambagny and Sonfonia, where there is only one public hospital providing comprehensive emergency obstetric and newborn care. The building of new public hospitals has not kept up with the population expansion on the periphery of the city. However, private-for-profit health providers see opportunities in these areas, mostly because the people who live in these newer neighborhoods might be able to afford to pay for care. This in turn increases the overall number of health facilities, creating a wealth-related inequality in access to healthcare, but with unclear consequences for care quality.

Strengths and limitations

Our analysis has several strengths. First, we used a validated and updated list of health facilities from the Ministry of Health and Public Hygiene, ensuring accuracy regarding whether facilities provide childbirth care and their geographic locations. Furthermore, we incorporated public and private facilities, whereas the majority of previous studies relied on public health facilities because of the ease of mapping and validating these public facilities. Second, we defined the Grand Conakry metropolitan area (conurbation) as the urban core (Conakry) and the surrounding suburbs30 (the urban areas of Dubréka and Coyah), reflecting the lived experiences of WoCBA when seeking childbirth care.

Third, we captured and applied realistic travel speeds that reflect the local context. We provided measures of uncertainty around the average estimates using two extremes (minimum and maximum speeds) and two conservative estimates (25th and 75th percentiles). This approach to data collection is cost-effective, especially when embedded with other routine tasks and projects, and has the potential to contribute to more realistic estimates, advancing the frontier of spatial accessibility31. The approach complements and advances other approaches where speeds are elicited from local experts to provide an informed guess of speeds across different roads14,32 or the use of application programming interfaces to indirectly account for realistic travel speeds13,14,15,16,17,18,22.

However, our findings have limitations. First, we acknowledge that we did not integrate elements of intrafacility referral, that is, that some women giving birth will first seek care at a lower-level health facility and will then be referred to a hospital. This means that we might have underestimated the time to reach appropriate care. Second, we could not account for ‘waiting time’ in terms of waiting for transport/money, so travel times are again likely underestimated. Third, we computed the travel time to the nearest health facility, although some women may bypass the nearest facility. We collected data during the dry season only. Therefore, the real travel times could be longer than we estimated. Fourth, due to incomplete data on the road network in the peri-urban districts14 compared with in the urban core, travel times might be overestimated. Finally, in the equity analysis, we assumed the same travel speeds for everyone, but this is probably not the case, that is, richer women are more likely to travel by private car and poorer women by shared transport (such as taxi or motorcycle).

Conclusions

In Grand Conakry, a city with 2.6 million inhabitants, major disparities exist in geographic access to health facilities providing childbirth care, driven by the skewed spatial distribution of health facilities, heavy traffic and socio-economic disadvantage. The average travel time to facilities is generally within the internationally defined thresholds under normal traffic conditions. However, during heavy traffic, travel times exceed 2 hours in some areas, with likely negative consequences for maternal and perinatal survival and well-being. The peri-urban communes (Dubréka and Coyah districts) are near medical deserts due to the low number of facilities, particularly public hospitals providing comprehensive emergency obstetric and newborn care. Furthermore, wealthier populations live closer to facilities providing childbirth care, while, surprisingly, poorer populations living in older, more centrally located urban areas benefit from greater proximity to public hospitals.

Our findings are useful for healthcare planning within Grand Conakry in terms of the geographic distribution of health facilities offering childbirth care vis-a-vis WoCBA and their socio-economic situation. For example, private health facilities improved geographic access in underserved areas. However, the effectiveness of this integration requires the alignment of private health facilities with public health standards for maternal health services and the financial accessibility of these services to the broader population. This study underscores the need for targeted interventions to address these accessibility gaps, particularly in the peri-urban districts of Coyah and Dubréka where infrastructure development lags behind. For example, increasing the number of public health facilities, strengthening existing private facilities, upgrading existing roads and enhancing public transportation could mitigate some of these challenges, ensuring that WoCBA have timely access to childbirth care, regardless of their socio-economic status or the time of day.

Methods

Study context

The Republic of Guinea is divided into eight administrative regions, including the special region of Conakry. These 8 regions are further subdivided into 33 prefectures in the countryside (rural) and 10 communes in Conakry (the main urban area comprising Kaloum, Dixinn, Matam, Gbéssia, Matoto, Tombolia, Ratoma, Lambagny, Sonfonia and Kassa Island). The setting of this study was Grand Conakry, a metropolitan area encompassing the urban core of Conakry city (excluding Kassa island) and the peri-urban area around the city (Fig. 1). Grand Conakry had an estimated population of over 2.6 million (49.5% of whom were women) in 2024. This represents about a fifth of the population of Guinea33. In Guinea, the proportion of the population living in urban areas has steadily increased from 10% in 1960 to 37% in 2022. More than half of this urban population (54%) is based in Conakry city33.

In 2024, Grand Conakry consisted of 14 urban communes: 9 in Conakry city and 5 in the urban zones of the Dubréka (Kagbélen and Dubréka) and Coyah prefectures (Coyah, Sanoyah and Manéah). Grand Conakry is bounded to the west by the Atlantic Ocean, to the south by the islands of Kaback, Kakossa and Matakang, to the north by the rural zone of the Dubréka prefecture and to the east by the rural part of the Coyah prefecture. The landscape of the area is mainly characterized by hills and coastal plains. Roads are the primary means of transportation used in Grand Conakry. However, many roads are in poor condition, making travel difficult, particularly during periods of heavy rain. Traffic jams are common in Grand Conakry and have a substantial impact on socio-economic activities. For example, in 2017, traffic jams were estimated to cost 5% of Guinea’s gross domestic product34.

Health care in Grand Conakry is provided by both the public and private sectors. According to the Ministry of Health and Public Hygiene, Grand Conakry has 155 health facilities: 58 public, 84 private and 13 private not for profit. Maternal care is mainly provided by the public sector. Antenatal care visits and basic emergency obstetric care are offered at health posts, health centers and private not-for-profit centers. Comprehensive emergency obstetric care is available mainly in public hospitals (district, regional and national) and in a few private hospitals. Healthcare utilization for maternal health services is high: according to the 2018 Demographic and Health Survey (DHS)35, 90% of births in Conakry took place in health facilities (25% in public hospitals). The total fertility rate was 3.8 in all urban areas of Guinea and 3.2 in Conakry35.

Overall methodological approach

We undertook four steps to model travel time, extract geographic coverage metrics and link the modeled travel time to the relative wealth index (Fig. 2). First, we defined the boundaries of Grand Conakry. Second, we assembled health facilities that offer childbirth care and factors that affect travel time, including road network, elevation, land cover and travel speeds. Third, we used a geospatial framework to compute travel time to the nearest heath facility disaggregated by level and sector for five travel-speed scenarios. In the last step, we linked the travel time to the population distribution (WoCBA and pregnant women) and relative wealth index.

Data

Grand Conakry boundaries

We considered the boundaries of the main city (Conakry) and the adjoining suburbs that form Grand Conakry. However, there were no open access vector files of these boundaries. Therefore, we defined and digitized the external boundaries of Grand Conakry and its subdivisions (communes) based on secondary data sources36,37,38 (Fig. 1). We loaded a hard-copy map showing Grand Conakry based on the Urban Sector Review into ArcGIS Pro (version 3.3.1, Esri). The hard-copy map was overlaid on OpenStreetMap (OSM) together with shapefiles from geoBoundaries37 and digitized. We used information from a presidential decree creating new communes in Grand Conakry to digitize the boundaries of the communes39,40.

Health facilities providing childbirth service

We aimed to define a geocoded list of health facilities that provide childbirth care within the defined boundaries of Grand Conakry. To achieve this, we obtained two lists of health facilities and annual births per facility in Grand Conakry from the Ministry of Health and Public Hygiene through the Strategic and Development Office. The two lists were harmonized to create a master list showing only the facilities that offered childbirth care in 2023 and their attributes (district, facility name, level and ownership). Public health facilities included health posts, health centers and hospitals (district, regional and national), while the private sector included clinics and faith-based facilities (managed by religious groups).

We geocoded the list using a variety of approaches. We extracted coordinates from the open SSA database21, validated them while at the same time geocoding other facilities not in the SSA lists based on online gazetteers and basemaps (Google Maps, GeoNames, OSM, Bing Maps and HERE Map). For the remaining facilities that could not be geocoded through the SSA list or online sources, we collected their global positioning system (GPS) coordinates at the entrance of the facility. The final list contained 86 facilities. The number, level and sector of health facilities in Grand Conakry providing childbirth care are shown in Supplementary Table 4 and their locations are shown in Fig. 3.

Travel times and inequalities were assessed for the four categories of health facilities shown in Supplementary Table 4. These included (1) public and private facilities providing childbirth care (the most non-restricted scenario), (2) public facilities providing childbirth care, (3) public and private hospitals providing childbirth care and (4) public hospitals (the most restricted scenario), which provide care for complications free of charge. This allowed analysis of both access and equity to primary and hospital care within the metropolitan area.

Factors that affect travel time

We used publicly available geospatial data of factors that could affect travel time to facilities offering childbirth care23,26,41,42. These factors included road networks from OSM from 202343, Sentinel-2 land cover at 10-m spatial resolution from 202344, Shuttle Radar Topography Mission digital elevation model (DEM) at 30-m spatial resolution45 and travel barriers (water bodies and flooded vegetation)44. The roads were reclassified into trunk, trunk link, primary, secondary, tertiary, residential, others (minor roads) and non-motorized based on road attribute data from OSM, local context information and travel-speed data43. Sentinel land use/land cover was used to represent areas where no roads existed, while the DEM was used to adjust walking speeds uphill and downhill46. The water bodies and flooded vegetation derived from the land use/land cover map44 were treated as barriers, except in the presence of a bridge. Maps of these factors are shown in Supplementary Fig. 1.

Travel speed

To derive realistic speeds across different types of road, 18 researchers from the African Center of Excellence for the Prevention and Control of Communicable Diseases (CEA-PCMT), Gamal Abdel Nasser University, Conakry geotraced trajectories within Grand Conakry between 13 May and 3 June 2024 (dry season) at different times of day. Specifically, we used tablets and smartphones with the KoboCollect App (v2024.2.4)47, geotrace option, which marked GPS points along trajectories at specific time intervals and precisions as the researchers traveled throughout Grand Conakry in their routine research activities. Time intervals were set at 1 minute and the precision at 20 m (that is, a GPS point would be recorded every 1 minute if the precision was less than or equal to 20 m).

Each submission (form) was designed to represent one trajectory; a total of 176 forms were completed and submitted. Researchers were prompted to enter on the form the exact time of the start and end of the trajectory. Additional details of the mode of transportation, road quality, travel speed and other characteristics of the trajectory were entered through either open or closed-ended questions. Researchers were encouraged to take a variety of road types and modes of transportation to ensure that a variety of scenarios were covered. To address the issue of time entries occurring after the end of the trajectory, time information from the data collector was compared with the number of geopoints for each trajectory, as the number of recorded geopoints was expected to be the same as the time recorded by the data collector. Submissions with a time difference of more than 20 minutes (72 submissions) were excluded.

Based on this information, the average speed for each road segment was computed by dividing the total distance traveled by the time difference between the start and end of the journey. These speeds were linked to the OSM road network. The speeds by road type were aggregated to obtain the average speed, the maximum and minimum speeds and the 25th and 75th percentiles. These five scenarios represent a continuum from the lowest (due to severe traffic jams) to the highest (substantially reduced traffic jams or weekend travel) travel speeds within Grand Conakry. Furthermore, we extracted walking speeds for different land cover type from previously published studies14,25,41.

Relative wealth index

We used Meta’s Relative Wealth Index (RWI) in this analysis48. The RWI estimates the relative wealth of the people living in each micro-region of 2.4 km × 2.4 km relative to others in the same country. The estimates were constructed through machine learning based on data from satellite imagery, mobile phone networks, topography, and aggregated and deidentified connectivity data from Meta. The models were trained on data from the DHS Program49. The data are available at the HDX portal50.

Population distribution of WoCBA and pregnancies

We obtained the constrained version of the total number of females per 100 m grid broken down by 5-year age groupings in 2020 for Guinea from the WorldPop portal51,52. We summed the 5-year age groups that constitute WoCBA (15–49 years) and clipped to the extents of Grand Conakry. To obtain estimates for 2024, we projected the estimates of WoCBA for 2020 based on the growth rates for 2019–2020 that had been derived using similar age-disaggregated data.

To obtain an equivalent surface showing the distribution of pregnancies, we converted the 2024 estimates of the WoCBA population to a surface of pregnancies following the methodology described by James et al. in 201853. We did this by multiplying each 5-year population group by its corresponding age-specific fertility rate for women in urban areas35. The spatially distributed number of births was summed and multiplied by a constant (1.0382)54,55,56,57,58 to account for pregnancy loss and multiple pregnancies in the calculation of the spatially distributed number of pregnancies.

Cost distance algorithm modeling

We used a least-cost path algorithm to model a hybrid traveling scenario that included walking and motorized transport to the nearest facility. In the hybrid travel scenario, from a residence, a person will first walk across areas where there is no road network (for example, on a footpath) to the nearest road (for example, a bus stop) where the person will take motorized transport. For residential areas adjacent to a road that connects to a health facility, motorized speeds were applied. Likewise, walking speeds were applied in areas with no road network between a residential area and a health facility.

We generated 20 sets of models from the travel-time estimates using 4 sets of destinations for health seeking: all facilities providing childbirth care, public facilities providing childbirth care, public and private hospitals providing childbirth care, and public hospitals providing childbirth care. For each set of destinations, we applied five travel times: average, minimum, maximum, and 25th (lower quartile) and 75th percentiles (upper quartile). In this report, we have presented the average travel-time scenario data as the main results and refer to the ranges provided by the interquartile ranges and minimum/maximum speeds as the best/worst case scenarios.

Specifically, we implemented the least-cost path algorithm in AccessMod (version 5.8.0)59,60. AccessMod is a World Health Organization (WHO) tool used to model how geographically accessible the existing health services are to the target population, including other functionalities59. First, we merged the road network and land cover, including water bodies and flooded vegetation, using the ‘Merge land cover’ option in AccessMod, resulting in a merged gridded surface. The merged gridded surface, travel speeds (per scenario) and location of facilities (by level and sector) were used to compute cumulative travel time from each raster cell (10 m × 10 m) to the nearest heath facility considering the least cost (the cost being measured in terms of time).

The slope derived from the DEM was used to adjust walking speeds based on Tobler’s formulation46. The formulation decreases the upslope walking speed as the slope increases, while slightly increasing the speed for a slightly negative slope when walking downslope. The anisotropic travel times to the health facility were computed. ‘Knight’s move’ was specified, which allowed consideration of 16 neighbor cells for higher accuracy. Finally, we extracted the average travel times for Grand Conakry and the communes for all 20 scenarios.

Travel time and inequalities

Based on the assembled population distribution maps and modeled travel times, we extracted the percentage of WoCBA and pregnant women who were within 15, 30 and 60 minutes of the nearest facility for all of the 20 scenarios. The extraction was conducted for the entire Grand Conakry and for each of its 14 communes. Finally, we linked the travel time for each scenario with the RWI. Before linking, we resampled the travel-time raster to 240 m given the course resolution of the RWI. The resultant 4,464 raster cells of travel time were linked to the RWI value that was spatially closest to it. Based on the RWI, we created five wealth quintiles and generated equiplots to summarize the travel time by wealth quintile for Grand Conakry and the three districts (Conakry, Coyah and Dubréka) in R (version 4.4.1) using the ggplot2 package (version 3.5.1)61,62,63.

Ethics

This study used secondary publicly available data and primary data (travel speeds and location of health facilities), which did not require ethical approval.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.