The influence of distance and quality on utilisation of birthing services at health facilities in Eastern Region, Ghana

Objectives Skilled birth attendance is the single most important intervention to reduce maternal mortality. However, studies have not used routinely collected health service birth data at named health facilities to understand the influence of distance and quality of care on childbirth service utilisation. Thus, this paper aims to quantify the influence of distance and quality of healthcare on utilisation of birthing services using routine health data in Eastern Region, Ghana. Methods We used a spatial interaction model (a model that predicts movement from one place to another) drawing on routine birth data, emergency obstetric care surveys, gridded estimates of number of pregnancies and health facility location. We compared travel distances by sociodemographic characteristics and mapped movement patterns. Results A kilometre increase in distance significantly reduced the prevalence rate of the number of women giving birth in health facilities by 6.7%. Although quality care increased the number of women giving birth in health facilities, its association was insignificant. Women travelled further than expected to give birth at facilities, on average journeying 4.7 km beyond the nearest facility with a recorded birth. Women in rural areas travelled 4 km more than urban women to reach a hospital. We also observed that 56% of women bypassed the nearest hospital to their community. Conclusion This analysis provides substantial opportunities for health planners and managers to understand further patterns of skilled birth service utilisation, and demonstrates the value of routine health data. Also, it provides evidence-based information for improving maternal health service provision by targeting specific communities and health facilities.

The tables 1a and b below compares the characteristics of women that were successfully geocoded with women who were not. The results show similar proportions across the groups in each variable. This suggests that the women who were geocoded or not are similar. Therefore, the analysis might not be affected by systematic bias.  Measuring quality of maternal health delivery can be based on two themes that are provision and experience of care 2 . The provision of care is usually measured with standards and protocols outlined for service delivery whereas the experience of care focuses on aspects of service delivery that could influence the perceived quality of clients. This type of framework has been implemented to measure the quality of delivery services and report basic proportions in India 3 , Nigeria 4 and Ghana 5 . Also, other measures of maternal health created summary indices from a group of variables to assess inequalities [6][7][8] . The bed complement of hospitals has also been included in gravity models as a proxy for the size or capacity of hospitals 9 10 .
Since the objective of creating the quality index was to plug it into a gravity model, a summary measure was deemed more appropriate. Thus a composite index was created with the provision of care, the experience of care and physical size variables.
Staff strength and bed complement were selected as measures of the physical capacity of the hospitals 11 . The core staff involved in maternal health care and skilled delivery was included in the index. These group of staff are obstetrician doctors, general practitioner, general surgeon, midwives, community health nurses, medical assistant, and anaesthesiologist. The scores were normalised with range standardisation to rescale the scores between 0 and 1. Similarly, the total bed complement, maternity beds, and delivery beds were standardised as shown in Table 2.
Essential drugs used in performing specific EmONC indicators were also assessed. Oxytocin was chosen as the major uterotonic, Diazepam for anticonvulsants and antibiotics. The drugs availability questions were Yes/No types and coded as 1 for yes and 0 for no. Hence, the three variables were added up to have a total maximum score of 3.
Furthermore, medical supplies that aid in the performance of maternal health services and promotes the safety of the child were also added to the index. Ambubags were included in the index because they were essential for newborn resuscitation. Other variables in this group were partograph for monitoring of births, weighing scale to weight newborns, stethoscope to check pulse, and a cup for expressing breastmilk. These five variables were added to create one group for medical supplies.
Hospital infrastructure is an important indicator of perceived quality of care 12 . Availability of electricity, water, running water, partitioning for privacy, functioning patient latrine, and waiting area for family and friends were added to create a group for hospital infrastructure.
Infection prevention items for healthcare workers were included. Variables chosen were the availability of antiseptic, soap for handwashing, bleach for disinfection, single-use towels, Veronica bucket and a covered contaminated waste bin with a pedal.
The final group of indicators added was the EmONC indicator that included the nine signal functions. Since the objective was to create an index and not classify the health facilities by their EmONC status, the nine EmONC signal functions were summed into one variable. The signal functions were the administration of parenteral antibiotics, administration of uterotonics, administration of parenteral anticonvulsants, manual removal of placenta, removal of retained products, assisted vaginal delivery, newborn resuscitation, blood transfusion, and performance of caesarean section. Table 2 shows the variables, score and weights used to calculate the quality of care index. To prevent the weights giving any group in the final index an unfair advantage, a range standardisation was applied to all the categories rescaling them from zero to one before multiplying by their weights. The normalised values were calculated as shown in Equation Where: Z= standardised score, X = score, min(x) = minimum score for the variable, and max(x) = maximum score for the variable.
The final quality of care index was calculated using the normalised scores and weight assigned to the variables. The scores were weighted for each group of indicators, divided by the highest possible score (12) and multiplied by 100 to arrive at a percent score for each health facility. The formula used for calculating the index is shown in Equation 3.6 below: 3 + 3B + 3C + 3D + 2E + F + 2G + 3H + I = × 100 12  framework. Variables that could easily influence the perceived quality of care such as functioning patient toilet, waiting area for family and friends, privacy, electricity and water and physical size of the hospital were given a higher weight 13 . The other variables that measure the quality of care delivery with standards and protocols were weighted less since they might not be ranked higher by clients and their family.

Appendix C:
Flow diagram showing excluded data