The study was conducted in the Kassena-Nankana Districts (East and West) located in northern Ghana. For the purposes of this study, the two districts shall be referred to by their former name - the Kassena-Nankana District (KND). The KND has an area of about 1,675 square kilometres with a population of about 152,000 people 21 . Subsistence agriculture is the mainstay of the people. The district is characterized by a high poverty and mortality burden. The district is in one of the poorest regions in Ghana with poverty incidence of 88% 22 , 23 . Maternal mortality ratio for the period 1995–1996 was estimated at 637/100,000 live births but it declined to 373 maternal deaths per 100,000 live births based on an estimate for the period 2002–2004, representing a 40% reduction in the ratio 24 .

With regards to health care, the KND has a district hospital located in the capital town (Navrongo) that serves as a referral point for all health facilities in the district. The hospital is the only health facility equipped to offer comprehensive emergency obstetric care in the district 11 . There are six health centres, one private clinic and twenty seven Community-based Health Planning and Services (CHPS) compounds. The CHPS initiative started in 1999 by the government of Ghana with the aim of increasing access to primary health care in the entire country. In this initiative, midwives and community health nurses are trained and sent to rural communities to provide basic preventive and curative services as well as doorstep services. These include antenatal care, delivery and postnatal services 25 .

A cross sectional quantitative survey design was employed in data collection. Data was collected between February and April 2014 from the patient’s perspective. Two graduate research officers conducted all the interviews after two weeks of training on the study tools. Women with pregnancy-related complications were defined as women who were diagnosed by health staff during pregnancy or delivery to have a maternal complication and were referred from the health centre to the hospital for treatment. Data on all women who had pregnancy-related complications at the six main health centres in the district between April 2012 and March 2013 (12 month period) were obtained from the six health centres.

A total of 296 women with maternal complications were referred from the health centres to the hospital within the period. However, contact information for 145 cases were never recorded by the health centre and therefore could not be traced in the community. In addition, 91 women had migrated from the district when their homes were visited. Thus only 60 women who were met during our visits were interviewed. Information on socio-demographic characteristics, direct and indirect costs of treating the complications were obtained. The reasons for referral reported by the women were also obtained. Since the aim was to capture official and unofficial payments made by women, all expenditures incurred within the hospital and outside the hospital were included.

This study was part of a larger project (QUALMAT project) which aimed to improve quality of maternal and prenatal care in Ghana, Tanzania and Burkina Faso by testing two interventions: a computer-assisted clinical decision support system and performance-based incentives for improvement of the quality of maternal health services provided 26 , 27 .

Ethical approval was obtained from the ethics committee of the University of Heidelberg (S-173/2008) and the Institutional Review Board of the Navrongo Health Research Centre in Ghana (NHRCIRB 085) before the study was conducted. In addition, individual oral informed consent was obtained from respondents before being interviewed.

Data were entered into Epidata 3.1 and exported into STATA 11.0^© for analysis. Descriptive analysis on background characteristics of respondents was done. The direct out-of-pocket costs for each pregnancy-related complication were estimated. Direct and non-direct medical costs were estimated by summing the costs and means calculated. The direct medical costs covered out-of-pocket payments for drugs, laboratory tests and medical supplies. Direct non-medical costs included all expenditure made on food during the health seeking process and transportation to the hospital and back home. The transportation cost included both the woman and the person who accompanied her to the hospital. Indirect costs associated with productivity lost were estimated by multiplying the number of days spent at the hospital by the daily minimum wage for the year 2013 (GH¢5.24/US$2.8). This was calculated for both the patient and the caretaker. Pre-referral costs were not collected and for that matter are not part of the analysis. Given that health centres are generally closer to the people, indirect cost such as transportation will be negligible. Also health centres do not have in-patient services hence costs related to in-patient care will be marginal.

To determine the financial impact of maternal complications for households, actual cost incurred by the household was measured in relation to average annual household expenditure obtained from the Ghana Living Standards Survey Report of the fifth round (GLSS 5)_ Gh¢1,918 (US$1009) 28 . Catastrophic Health Expenditure (CHE) was also assessed. CHE is when a household’s out-of-pocket (OOP) payments are so high relative to its available resources which would require the household to forego the consumption of other essential goods and services 29 . Thus total OOP health care payments in excess of a certain threshold of household resources (household income, expenditure or consumption) are catastrophic. There is no consensus regarding the threshold for defining catastrophic health expenditures. Most authors have used threshold levels of 2.5%, 5%, 10% 15% and 20% of total household income.

All costs were collected in Ghana Cedis (GH¢) and results presented in US$. The US$ conversion was based on the average exchange rate for 2013 (1US$ = 1.9GH¢). Given that the numbers interviewed were small, results on expenditures are presented in median.

Of the 60 women, majority were married 88% (53). The mean age was 26 years (median = 25; range = 16–39 years) with standard deviation of 6. Most of the women 34% (20) were between 21–25 years, 23% (14) were above 30 years and 23% (14) were less than 20 years old. Majority 57% (34) of the respondents were farmers/traders, 25% (15) were unemployed, 10% (6) were employed in the formal sector, 5% (3) were artisans and 3% (2) were students. In terms of education, 56% (34) of the respondents had basic level education (primary and junior secondary level), 22% (14) had post-basic education (secondary education or higher), and 22% (13) had no formal education. Majority of the women who had complications were primiparous 47% (28) (Table 1).

As shown in Table 2, more than half (60%) of the pregnancy-related complications reported by respondents resulted from prolonged labour (labour lasting more than 12 hours). About 17% (10) was due to severe abdominal pain, 10% (6) due to anaemia/malaria, and about 7% (4) due to pre-eclempsia. About 80% (49) of respondents sought care at a hospital and the remainder at a regional hospital.

Table 3 presents the main cost components for each of the five maternal complications. Of the 60 women who had complications, only 30% (18) incurred direct medical costs and 70% (42) reported zero direct medical cost. Out of the 18 women, the total direct medical cost emanating from drugs, laboratory services and scanning was US$543.68 and the median cost was US$8.68(IQR = 43.68) per treatment. Thus, 21% of the total cost of treatment was spent on direct medical costs (Figure 1). Most of these direct medical costs were incurred outside the hospital because of shortage/non-availability of prescribed drugs or non-availability of equipment. For instance, 2 years preceding the survey, the Navrongo hospital operated without an ultrasound scan. Accordingly, women had to obtain their scans from private sources which made them incur additional costs.

The median transportation cost was US$13.48 (IQR = 16.05) per person (patient and person accompanying the patient) representing 32% of the total cost (Figure 1). Majority of the respondents were transported to the hospital with the official vehicle of the health centre (37%). Respondents reported paying between US$11 and US$13 to fuel the vehicle to the referral point. The median expenditure made on food for both patient and caretaker was estimated at US$9.47(IQR = 14.21) per person. In addition, median indirect cost attributed to productivity losses was estimated at US$5.2 (IQR = 8.27) per person.

Overall, the median expenditure made by households on both direct and indirect costs per complication was US$ 32.03(IQR = 42.55) per person. Indirect costs accounted for the largest proportion of total cost (79%). Disaggregating costs by type of complication, women who had pre-eclampsia spent more (median = US$58.33; IQR = US$64.65) than those who had other types of complications. Women with haemorrhage incurred the least cost (median = US$6.84; IQR = 6.32).

As shown in Table 4, the median number of days spent in the hospital was two days (mean = 3 days, IQR = 3). The median days varied from zero for haemorrhage to six days for pre-eclampsia.

Table 5 presents costs of maternal complication as a proportion of annual household expenditure for 2008 (US$1009). Payments for maternal complications amounted to about 3% of annual household expenditure. Pre-eclampsia and anemia/malaria accounted for the greatest burden on households (6% of annual household expenditure). Haemorrhage accounted for the least burden on households (1% of annual household expenditure). About 33% (6) of households spent more than 5% of annual household expenditure and therefore faced catastrophic payments (threshold at 5%).

One limitation was the possibility of recall bias. Some expenditure could either have been overestimated or underestimated. We however recognize that health expenditure and hospitalization are critical events for low income families and therefore can easily be recollected. Field staff training also put emphasis on probing strategies to help mitigate the effect of recall bias.

Also, the study could not contact about 49% (145) of all cases due to the absence of contact details. It is worth noting that, as in many developing countries, home addresses in Ghana is a major challenge. However, the health centres are located within a demographic surveillance system which has developed a unique home address system with identity cards for all individuals in the district. This should have facilitated the recording of contact information which is part of the standard of care within the health care system. The nurses at the health centres however complained that most patients do not come with their identity cards and recalls often led to wrong addresses.

We did not include intangible costs in our analysis which therefore underestimated our cost results. Intangible costs relates to the reduced quality of life due to illness. It includes pain, psychological pressure, reduced joy of life and social prestige due to the illness 30 . However, this cost component, though important, is difficult to measure and is therefore not usually included in cost studies.