Demand- and supply-side factors affecting utilisation and provision of maternal health services under the free maternal health care policy in one municipality in Northern Ghana

A free maternal health care policy was introduced in Ghana in 2008 to cater for the health needs 4 of pregnant women for the reduction of maternal deaths. Experiences from other regions and 5 countries show that demand- and supply-side factors often affect the success of such policies 6 although this is unknown in this context. The study aimed to assess demand- and supply-side 7 factors affecting the utilisation and provision of services during pregnancy under the policy. A convergent parallel mixed methods study was undertaken, collecting quantitative and 10 qualitative data. The study was carried out in the Kassena-Nankana East Municipality in Ghana. 11 Questionnaire were used to collect data from women (n=406) who utilised services during 12 pregnancy. Focus group discussions with women (n=10) and in-depth interviews with 13 midwives and nurses (n=25) were also conducted. Quantitative data were analysed and 14 presented using descriptive statistics. Qualitative data were audio-recorded, transcribed and 15 coded using themes and sub-themes. Demand- and supply-side factors were reported to impede the use and provision of services. 1 Government and stakeholders should prioritise building as well as expanding the infrastructure 2 of CHPS compounds. Emergency transport for women should also be provided. These together 3 may contribute to improving the use and provision of services, leading to a reduction in 4 maternal deaths and achievement of universal health coverage. 5 11

the NHIS for vulnerable groups like pregnant women, which comes under the free maternal 8 health policy. Given that insured women were more likely to use antenatal, childbirth and 9 postnatal services (3-6), leading to reduced maternal deaths (7)(8)(9).   Under the policy, pregnant women are given free registration with the NHIS, with no payment 21 of premium or processing charges. Health services, including drugs are provided free of charge 22 during the periods of pregnancy, childbirth and postpartum (11). Following the implementation 23 of the policy, a cumulative total of over 2 million women had benefitted at the end of 2013, 24 which is considered a significant achievement for the NHIS and the country (12). A recent 25 study in the Northern and Central regions of Ghana revealed that, the policy had promoted the 1 use of health services by women (13).

3
The success of health policies in terms of implementation or coverage are often affected by 4 demand-and supply-side factors. Demand-side factors are the individual, household or 5 community level factors which either promote or impede the use of health services. These 6 generally include distance and time taken to reach health facilities, cost of transport to health 7 facilities, means of transport available, education, information on health services or health 8 providers, community and cultural preferences and household expectations (14,15). Supply- 9 side factors refer to the characteristics of the health system which are beyond the control of 10 consumers of health services, but have an impact on the provision of health services (14,16,11 17). These health system factors include opening and closing hours of health facilities, cost of 12 health services, availability of health providers, drugs and supplies, equipment and   Municipality in Ghana (15,21,22). Aside that, a recent study demonstrated that women in the 6 Kassena-Nankana area had a continuum of care completion rate of 14.0%; that is, women who 7 made 4 or more antenatal visits, used a health facility for childbirth and were available for 8 postnatal services within 48 hours, and at two weeks as well as six weeks (23). Given these, 9 there is the need to carry out investigations in that direction. Thus the aim of this paper was to 10 assess demand-and supply-side factors affecting the utilisation and provision of free maternal 11 health services during pregnancy in the Kassena-Nankana East municipality. The views and 12 perceptions of women who utilised health services during pregnancy as well as frontline health 13 providers (midwives and nurses) were studied. 14 15

17
A convergent parallel mixed methods study was adopted, utilising quantitative and qualitative 18 data collection techniques (24). A mixed methods approach was adopted to ensure 19 comprehensiveness and confidence in the study findings (25-27). The use of the convergent 20 parallel mixed methods design allowed for data collection and analysis for the quantitative and 21 qualitative studies to be carried out in a parallel form; while the findings are integrated to assure 22 a comprehensive analysis of the research question (24); that is, factors affecting the use and 23 provision of maternal health services under the free maternal health policy. The study was 24 cross-sectional and data collection was carried out from March-August, 2016. A structured 25 questionnaire was administered to women who had utilised maternal health services during 1 pregnancy. The qualitative aspect involved focus group discussions (FGDs) with similar 2 women and in-depth interviews (IDIs) with frontline health providers (midwives and nurses) 3 in NHIS accredited health facilities, using semi-structured questions. The study was set in the Kassena-Nankana East Municipality which is located in the Upper 7 East region of Northern Ghana. The Kassena-Nankana East Municipality was selected as the 8 study area due to the fact that it was well mapped out by the Navrongo Health Research 9 Centre for research and hence convenient for the study team. The study was time bound and 10 financially constrained to select other districts or regions for investigation. The population of 11 the municipality was estimated to be 108,000; with males and females constituting 48.8% and 12 51.2% respectively (28).  The formula for a proportion in a single cross-sectional survey was used for the estimation of 1 the sample size for the quantitative study (32). A total of 406 women were recruited 2 representing women who utilised health services during pregnancy.   Quantitative data collection 25 9 The quantitative data were collected electronically using SurveyCTO. For a health facility to 1 be included in the study, it must have at least a midwife to offer antenatal, childbirth and 2 postnatal services, since the study sought to capture data pertaining to all these aspects of 3 maternal health services. Thus, the study recruited women from the main hospital, two (2) 4 health centres, and eleven (11) CHPS compounds across the municipality. Two trained 5 research assistants were responsible for visiting the health facilities on a daily basis for the 6 identification of women who gave birth and were discharged to go home. After consent, the 7 women were surveyed. All women who were invited agreed to participate in the study. The 8 inclusion criteria were that women should have utilised maternal health services during 9 pregnancy; to be able to report their experiences. Women should have also given birth   The guide for the FGDs in particular, was translated by language experts into the two dialects 17 (Kasem and Nankam) spoken in the study area. Hence the discussions were done in the two 18 dialects. The guide for the IDIs was not translated because all the midwives and nurses spoke 19 and understood English. All the FGDs and IDIs were audio-recorded with the permission of 20 the women, midwives and nurses. Field notes were also taken alongside the recordings.

22
The women were invited to participate in the FGDs with the assistance of midwives and 23 nurses. The visit of the women to participate in the FGDs were planned to coincide with their 24 routine health facility visits. The FGDs were conducted in private rooms in the health 25 facilities, without the presence of health providers. This allowed participants to freely express 1 themselves on the issues discussed. Approximately 5-12 members formed a group for each 2 of the FGDs. All group members were encouraged by the investigator to partake in the 3 discussions.  After the FGDs and IDIs, a summary of the issues that emerged from the discussions and 12 interviews were made and presented back to the participants for their validation. New issues 13 emerging from the discussions and interviews which were considered to be important were 14 added to the questions on the interview guides for the next round of discussions and interviews.

15
Data saturation was reached in the FGDs and IDIs when discussants and interviewees had 16 nothing to say or talk about upon prompts and probes from the main investigator (37).

18
Data analysis and management 19 The data were cleaned and analysed using STATA 14. Descriptive statistics were used to 20 present the findings. The qualitative data were manually analysed. The audio data were 21 transcribed verbatim into English. The transcriptions and field notes were examined to 22 identify patterns in the data. To assure validity and accuracy of the data, a random sample of 23 5% of the recordings were selected, listened to and compared with the transcriptions, of 24 which differences were corrected before coding. However, a negligible difference was 25 observed between the recordings and the transcripts. A follow up reading of the transcripts 1 and interview notes were carried out, noting important issues brought out by the participants. 2 A coding system was then adopted using themes and sub-themes (where necessary) and this 3 was presented in text tables. The coding system based on the themes and sub-themes formed 4 the basis for the presentation of the findings, including essential quotes emanating from the 5 participants.  Time taken to reach nearest health facility irrespective of mode of transport 18 Most of the women perceived time taken to reach nearest health facility as a challenge. 19 Altogether, 58% (236/406) of the women reported that they had travelled between 30 minutes 20 and one hour or one to two hours to reach the nearest health facility, without regard to mode of 21 transport (Figure 1). Around 5% (20/406) of the women perceived that they had travelled for 22 more than two hours. Over two-thirds of the women (69.2%; 281/406) reported walking to nearest health facility for 6 maternal health services during pregnancy, with variation by perceived distance (Table 2). 7 More than a quarter of the women (26.1% (106/406) perceived distance to be either far or very 8 far, using various means of transport to get to nearest health facility. Thus, distance to health 9 facilities was also a barrier to the use of health services in the study area.   'But if its distance, I don't think now distance is much a problem -we're living closer 13 to them. And they're familiar with usthey know us; when they meet us anywhere, 14 they're free to talk to us, and we can also talk to them' (IDI, midwife).

16
Notwithstanding, health providers pointed that it was a challenge when there was the need to 17 refer women from lower to higher level health facilities (that is, from CHPS compounds/health 18 centres to the district or regional hospital Waiting times at nearest health facility during pregnancy 2 Over half of the participants perceived waiting times to be either short or normal (Figure 2).

3
Approximately 6% (24/406) of the women perceived that their waiting times were 'very long'. In the IDIs, it was reported that the main (and only) hospital in the municipality had in place a It was reported in the FGDs that women who visited private laboratories for tests were required 10 to pay for them, which was a challenge. Some of the women were unable to pay for these tests.

11
A woman said: The women confirmed the unavailability of drugs and were often given prescription forms by 1 health providers to buy drugs in pharmacies and chemical shops. A woman indicated: Some women reported that they were unable to buy the needed drugs immediately and had to 8 wait until they had money for the purchase. Some of the women revealed they never bought 9 the prescribed drugs due to their unaffordability. Availability of transport for emergency cases 23 The lack of transport for the use and provision of maternal health services was another barrier.

24
A suitable means of transport like ambulances, were not available at the community level to 25 21 convey women with emergency conditions to higher level health facilities. If even transport 1 was available, women were not always able to afford payment for the service. A nurse said: Appropriate transport is not available to carry pregnant women from their communities 5 to this place' (IDI, nurse). It was also difficult for lower level health facilities to refer women to higher level health 8 facilities due to unavailability of transport. Most of the lower level health facilities did not have 9 their own means of transport for emergency referrals. A midwife said:  supplies, and other basic essential items including reagents. There is the need for the NHIS to 3 timely pay health facilities to allow them to purchase these basic essential inputs for service 4 provision.

6
In particular, CHPS compounds had infrastructural challenges relating to the availability of Our study demonstrated the unavailability of appropriate transport for emergency referrals 21 between communities and health facilities as well as between lower and higher-level health 22 facilities. Other studies have found the same. For instance, transport for emergency cases was