Travel time to maternity care and its effect on utilization in rural Ghana: A multilevel analysis
Introduction
Ghana faces high burdens from maternal and child mortality. In 2008, its maternal mortality rate was 409 deaths per 100,000 women. This rate was below the West African regional average of 629 deaths per 100,000 live births in 2008, but exceeded the global average of 251 (Rajaratnam et al., 2010). Its neonatal mortality rate was estimated at 28.1 per 1000 live births, accounting for over a third of all child deaths in the country (Hogan et al., 2010).
Ghana has made significant progress in recent years. From 1990 to 2008, child mortality decreased by 2% per year and maternal mortality decreased by 1.4% per year. Despite this progress, the burden of childbirth complications could be further reduced by increasing rates of in-facility delivery (IFD) (Campbell et al., 2006, Lawn et al., 2005). In rural Ghana, approximately 60% of births occur in the home, away from facilities equipped to deal with complications (Ghana Statistical Service, Ghana Health Service, & ICF Macro, 2009). Given the established benefits of IFD, understanding the determinants of its use (or lack thereof) is a highly relevant research question that could lead to improvements in maternal and neonatal health.
Distance has been recognized as a determinant of utilization, especially in rural areas (Gabrysch and Campbell, 2009, Gage, 2007, Hodgkin, 1996, The White Ribbon Alliance for Safe Motherhood, 2011). Descriptive statistics from the 2008 Ghana Demographic and Health Survey (GDHS) support this conclusion: 33% of rural mothers cite distance as the factor for not seeking birth services, more than any other reason (Ghana Statistical Service et al., 2009). Previous studies have analyzed the effect of distance, but few have quantified the effect of travel time on utilization. Travel time encompasses not only distance, but also the mode and difficulty of travel. It is a methodological advancement over matching facilities and their surrounding population using straight-line distance (Alegana et al., 2012, Astell-Burt et al., 2011, Lovett et al., 2002, Tanser et al., 2006). Relative to distance alone, travel time better reflects the decision-making process to utilize IFD.
Our analysis contributes to a small but growing body of work that estimates the effect of travel time on service utilization (Alegana et al., 2012, Astell-Burt et al., 2011, Lovett et al., 2002, Tanser et al., 2006). Lovett et al. (2002) calculate travel times to health facilities in East Anglia, United Kingdom, but their study is limited to a descriptive analysis. Tanser et al. (2006) calculate travel times to health facilities and estimate their effect on utilization in KwaZulu-Natal, South Africa. They find higher travel times have a significantly negative effect on utilization. Astell-Burt et al. (2011) explore whether higher travel times to primary health care centers are associated with lower rates of hepatitis C detection in France. They also find a significantly negative effect. Finally, Alegana et al. (2012) model the effect of travel time on seeking treatment for fever among children under the age of 5 years in Namibia. They find the probability of facility attendance remains relatively high for up to three hours of travel time, but decreases steadily thereafter.
Given our covariates and methodological strategy, our analysis resembles Gabrysch, Cousens, Cox, and Campbell (2011), who estimate the influence of distance and level of care on IFD in Zambia. In our analysis, we introduce two consequential improvements over their strategy. First, we employ travel time as an independent variable, rather than straight-line distance. Second, uncertainty exists in the precise location of GDHS households because the geographic coordinates are scrambled (MEASURE DHS, 2011). We address this uncertainty by treating the location selection as an imputation problem. To our knowledge, this study is the first analysis to seriously account for that uncertainty rather than ignore it.
Section snippets
Theoretical motivation
Throughout sub-Saharan Africa, pregnant women perceive that benefits exist from utilizing IFD (Kruk and Prescott, 2012, Magadi et al., 2007). This value is derived from many factors, one of which is the capacity of the facility to provide multiple types of services (Parkhurst et al., 2005). However, utilizing IFD is costly. The precise costs vary across settings, but they may include user fees and the time required to travel to the facility, among other factors (Nyonator & Kutzin, 1999).
Data
To generate estimates of travel times to facilities within Ghana, we divided the country's territory into gridded cells of 1 square kilometer. We utilized two data sources to calculate the travel times from each cell to each facility: the first dataset reports facility locations, and the second dataset reports geographic information about each cell, which we used to reflect the difficulty of traveling between cells. To generate regional-level descriptive statistics, we used disaggregated
Methods
To test our theoretical implication, we employed a multilevel random-intercept logistic regression model. This model explicitly accounted for the hierarchical and nested structure of the data (Rabe-Hesketh & Skrondal, 2012). In our data, women were nested within households, and households were nested within communities. This clustering suggested that women in the same households are very likely to have similar socioeconomic and community-level characteristics (including travel time). Therefore,
Results
Results from our population-level analysis (which is independent of the GDHS data) are reported in Fig. 1 and Table 2. Fig. 1 illustrates district-level travel times to the nearest facility. Table 2 reports regional-level population estimates and travel times to the nearest facility. Reported values are the population-weighted average of cells in that administrative unit, assuming vehicular travel on roads. In metropolitan areas such as Accra and Kumasi, average travel times were relatively low
Conclusion
Geographical features, the architecture of the road network, and road quality can substantially modify the ease of travel in many locations. Consequently, the travel times in this analysis represented a more realistic measure of health-facility accessibility than Euclidean distance. The factors influencing a woman's utilization of IFD services are complex, but our results suggest travel times affect behavior. Due to the stated effect of travel time on behavior, policy evaluations should account
Acknowledgments
We thank Andy Tatem for his help and guidance during the creation of the friction surface and cost path analysis. We thank the Ghana Health Service and UNICEF Ghana for providing the EMOC needs assessment dataset. We thank Kelsey Moore and other members of the IHME team for their insight and contributions. This research was supported by the Institute for Health Metrics and Evaluation's core funding from the Bill & Melinda Gates Foundation.
References (35)
- et al.
Does geographic access to primary healthcare influence the detection of hepatitis C?
Social Science & Medicine
(2011) - et al.
Strategies for reducing maternal mortality: getting on with what works
The Lancet
(2006) Barriers to the utilization of maternal health care in rural Mali
Social Science & Medicine
(2007)- et al.
Maternal mortality for 181 countries, 1980–2008: a systematic analysis of progress towards millennium development goal 5
The Lancet
(2010) - et al.
4 Million neonatal deaths: when? where? why?
The Lancet
(2005) - et al.
Car travel time and accessibility by bus to general practitioner services: a study using patient registers and GIS
Social Science & Medicine
(2002) - et al.
A comparative analysis of the use of maternal health services between teenagers and older mothers in sub-Saharan Africa: evidence from demographic and health surveys (DHS)
Social Science & Medicine
(2007) - et al.
Health systems factors influencing maternal health services: a four-country comparison
Health Policy (Amsterdam, Netherlands)
(2005) - et al.
Neonatal, postneonatal, childhood, and under-5 mortality for 187 countries, 1970–2010: a systematic analysis of progress towards millennium development goal 4
The Lancet
(2010) - et al.
Modelling and understanding primary health care accessibility and utilization in rural South Africa: an exploration using a geographical information system
Social Science & Medicine
(2006)
Spatial modelling of healthcare utilisation for treatment of fever in Namibia
International Journal of Health Geographics
A mixed logit model of health care provider choice: analysis of NSS data for rural India
Health Economics
A new framework for managing and analyzing multiply imputed data in Stata
The Stata Journal
A simple and flexible GIS tool for deriving accessibility models
Still too far to walk: literature review of the determinants of delivery service use
BMC Pregnancy and Childbirth
The influence of distance and level of care on delivery place in rural Zambia: a study of linked national data in a geographic information system
PLoS Medicine
Geographical access to care at birth in Ghana: A barrier to safe motherhood CPC working paper 21
Cited by (63)
The projected cost-effectiveness and budget impact of HPV vaccine introduction in Ghana
2022, VaccineCitation Excerpt :Societal-level treatment costs included direct medical costs, plus opportunity costs of women’s time for procedures (including convalescence if they had hospital stays or recovery time), time waiting for test results (if provided on the same day as procedures), and time costs associated with travel to/from facility visits. The amount of time women spent traveling to/from facilities for cervical cancer treatment was estimated based on a 2013 study presenting average travel times to/from facilities in Ghana for maternity care [30]. For opportunity costs of women’s time, we applied the Ghana 2018 minimum wage, reported as 10.65 Ghanaian cedis per day (USD 2018 $2.32) [31].
The long road to health: Healthcare utilization impacts of a road pavement policy in rural India
2021, Journal of Development EconomicsEducational expansion and health disparities in Ethiopia, 2005–2016
2019, Social Science and MedicineCitation Excerpt :These cultural differences manifest in large and small ways, ranging from an individual woman's ability to travel alone to visit a healthcare facility to the redefinition of traditional gender roles and the social relations that define who can access healthcare services (Woldemicael and Tenkorang, 2010). Finally, the distance women must travel to receive treatment is another important determinant of healthcare utilization, generating stark disparities in access rates between rural areas and urban centers (Bobo et al., 2017; Masters et al., 2013; Okwaraji et al., 2015; Stock, 1983). Previous studies do not elaborate on how educational expansion might shift rural health disadvantages, but educational expansion in rural areas may improve physical accessibility to healthcare services, as educational expansion often accompanies general improvements in infrastructure (Barber, 2004).
Distance to healthcare facilities, healthcare use and health outcomes: The case of Papua New Guinea
2023, Australian Economic PapersPredictors for achieving adequate antenatal care visits during pregnancy: a cross-sectional study in rural Northwest Rwanda
2023, BMC Pregnancy and Childbirth