Factors associated with intended use of a maternity waiting home in Southern Ethiopia: A community-based cross-sectional study

Background: Although Ethiopia is scaling up Maternity Waiting Homes (MWHs) to reduce maternal and perinatal mortality, women’s use of MWHs varies markedly between facilities. To maximize MWH utilization, it is essential that policymakers are aware of supportive and inhibitory factors. This study had the objective to describe factors and perceived barriers associated with potential utilization of an MWH among recently delivered and pregnant women in Southern Ethiopia. Methods: A community-based cross-sectional study was conducted between March and November 2014 among 428 recently delivered and pregnant women in the Eastern Gurage Zone, Southern Ethiopia, where an MWH was established for high-risk pregnant women to await onset of labour. The structured questionnaire contained questions regarding possible determinants and barriers. Logistic regression with 95% Confidence Intervals (CI) was used to examine association of selected variables with potential MWH use. Results: While only thirty women (7.0%) had heard of MWHs prior to the study, 236 (55.1%), after being explained the concept, indicated that they intended to stay at such a structure in the future. The most important factors associated with intended MWH use in the bivariate analysis were a woman’s education (secondary school or higher vs. no schooling: odds ratio [OR] 6.3 [95% CI 3.46 to 11.37]), her husband’s education (secondary school or higher vs. no schooling: OR 5.4 [95% CI 3.21 to 9.06]) and envisioning relatively few barriers to MWH use (OR 0.32 [95% CI 0.25 to 0.39]). After adjusting for possible confounders, potential users had more frequently suffered complications in previous childbirths (adjusted odds ratio [aOR] 4.0 [95% CI 1.13 to 13.99]) and envisioned fewer barriers to MWH use (aOR 0.3 [95% CI 0.23 to 0.38]). Barriers to utilization included being away from the household (aOR 18.1 [95% CI 5.62 to 58.46]) and having children in the household cared for by the community during a woman’s absence (aOR 9.3 [95% CI 2.67 to 32.65]). Conclusions: Most respondents had no knowledge about MWHs. Having had complications during past births and envisioning few barriers were factors found to be positively associated with intended MWH use. Unless community awareness of preventive maternity care increases and barriers for women to stay at MWHs are overcome, these facilities will continue to be underutilized, especially among marginalized women. A community-based cross-sectional study was undertaken between March and November 2014 among women who had given birth in the three years prior to the study or who were pregnant at the time of the study. Participants had to be able to communicate in the national language Amharic. Ethical approval for the study was obtained from the Southern Nations Nationalities and People Regional State Health Bureau in Hawassa, Ethiopia. The study took place in the Eastern Gurage Zone, a predominantly rural area in the Southern Nations Nationalities and People Regional State of Ethiopia, with an estimated population of over 500,000. Administratively, this zone is divided into four districts: Meskan, Mareko, Soddo, and Butajira town (the largest urbanized area), with approximately 46,000, 19,000, 42,000 and 11,000 women in the reproductive age group, respectively (personal communication from Zonal Health Bureau Welkite, 2 May 2015). While no regional data are available to calculate a Crude Birth Rate (CBR) for the area, the national CBR was 32 births per 1000 people in 2016 [5]. A total of 119 health posts, 20 health centres and two general hospitals (one governmental and one faith-based) served the population. The closest tertiary referral hospital was in the capital, Addis Ababa, at approximately two hours’ distance by ambulance. At the time of the study, ambulances were available at hospital and district level, but these had difficulties accessing remote areas, particularly in the rainy season. Project Mercy Hospital, located in a small village near Butajira town, opened an MWH in 2012. This study was conducted for Butajira General Hospital, located in Butajira town, which established an MWH on its grounds in 2015. At the time of the study, the hospital had a catchment area population of around one million, serving people from the Eastern Gurage Zone and neighbouring zones. Butajira Hospital provides 24-h comprehensive emergency obstetric and new-born care and the number of births was approximately 3000 in 2014. Delivery services became free-of-charge in the first quarter of 2014. Access to the MWH is provided at no cost. A sample size of 383 was calculated using Epi Info StatCalc, with a 5% margin of error and a 95% confidence interval (CI), by using the estimated number of women in the reproductive age group in the Eastern Gurage Zone (118,000). Since the true rate of expected MWH-use was unknown, the expected frequency was set at 50%, which gives the largest possible sample size. A design effect of 1.0 was used. In total, 428 respondents were conveniently sampled from each of the four districts in the Eastern Gurage Zone: 120 women from Butajira town, 108 from Meskan district, 100 from Soddo district, and 100 from Mareko district. In Butajira, participants were selected from each of the five ‘Kebeles’ or neighbourhoods. In the other three districts, data collection took place near health centres. Five health centres were randomly selected from Meskan and Soddo and four from the smallest district Mareko. In Soddo, one of the randomly selected health centres was not accessible by public transport at that time, which led us to purposely choosing another centre at a similar distance. Participants were selected by visiting every third household. If a woman in that household had given birth in the three years prior to or was pregnant at the time of the survey, and could communicate in the national language Amharic, she was asked to participate. Whenever more than one eligible woman was found in the same household, one was randomly selected and included in the study. If no one in that specific household fulfilled the inclusion criteria, the neighbouring house was visited. One woman declined participation, stating that she needed permission from her husband, who was not available. Informed written consent was obtained from all participants, through their signature or fingerprint. Variables were formulated using the Adapted Three Delay Model, which describes possible delays in (1) deciding to seek birth care, (2) trying to identify and reach a health facility, and (3) receiving adequate and appropriate treatment. This Adapted Three Delay Model was formulated by Gabrysch et al., who expanded the original model arguing that the latter implicitly considers home births with complications and that (possibly reduced) delays of what the authors call “preventive facility births” should be more explicitly included in the model. These factors are grouped into four themes: sociocultural factors, perceived need/benefit, economic and physical accessibility [12, 13]. A structured questionnaire was developed in English, translated into the national language Amharic and then translated back into English to check for consistency [see Additional files 1 and 2 for the English and Amharic versions of the questionnaire]. All questions except those concerning relative household wealth, likelihood of staying at an MWH and envisioned barriers to MWH use were taken from the Ethiopian Demographic Health Survey [14]. Questions regarding the use of an MWH were formulated based on a previous Ethiopian study [9]. The questionnaire was pre-tested twice among pregnant and recently delivered women in Butajira Hospital, first by a medical doctor and thereafter by a data collection team. They read questionnaires out loud and completed these in the presence of an observer, which led to improvements in the questionnaire’s layout and explanatory texts. Data collection with regard to socio-cultural factors comprised of the respondent’s estimated age category, marital status, and her and her husband’s educational level. Decision-making power was determined by a combined score of answers to two questions regarding who is involved in deciding on family earnings and in matters of maternal and child health. Women who generally made decisions independently or jointly with their husbands were considered to have decision-making power. Health education is one of the included factors relating to perceived need/benefit: women were asked if they had received information about signs of pregnancy complications during antenatal care visits and if they could name any (eight options were provided for the data collector: vaginal bleeding, vaginal flush of fluid, severe headache, blurred vision, fever, abdominal pain/preterm contractions, decreased foetal movement, oedema/body swelling, plus the option ‘other, specify…’). Parity was defined as the number of times a woman had given birth, including intrauterine deaths and stillbirths. History of facility delivery was recorded as ‘birthing location’. Primigravida were recorded as not having a history of home or facility birth. If a respondent’s births all took place at the same location, the last birth was explored in terms of the reason(s) why she delivered at home or at a facility and, if applicable, which complications she suffered. If there had been a change in birthing location, we prompted for reasons why she had previously given birth both at home and at the facility, and, if applicable, which complications she had suffered. Answers were recorded using a multiple response set: haemorrhage, prolonged labour, obstructed labour, hypertensive disorder, puerperal infection, foetal distress, intrauterine foetal death, and ‘other, namely’. For the analyses, complications were clustered into a yes-no score. Economic accessibility was assessed by asking respondents to compare the wealth of their household with those around them on a four-point scale (very wealthy, wealthy, poor, very poor). In the analyses, a combined score was used. Physical accessibility was defined by a respondent’s travel time from her household to the nearest hospital. Urban/rural residency was based on the 2007 Population Census [15]. The questionnaire contained specific questions regarding the likelihood of staying at an MWH and perceived social and economic barriers to using an MWH. First, respondents were asked if they had ever heard of an MWH. Regardless of their answers, they were then explained the concept of an MWH: “A Maternity Waiting Home is a place for high-risk pregnant women to await birth in their last weeks of pregnancy, close to 24/7 emergency obstetric care. Possible reasons to stay are for example a previous caesarean section or haemorrhage, previous stillbirth or neonatal death, breech presentation, twin pregnancy, or living far from a hospital.” Respondents were then asked if they knew an MWH in the region and if they believed there were advantages to staying at an MWH and if so, what these advantages would be. Subsequently they were asked how likely it would be for them to stay at an MWH during the last two to four weeks of their current or next pregnancy using a four-point scale (very likely, likely, unlikely, very unlikely). In the analyses, a combined score was used. Finally, respondents were asked to imagine staying at an MWH for two to four weeks and how they might arrange transport and food, bring their own cooking utensils, stay for that length of time, bring an attendant to accompany them, and arrange for others at home to take care of their children and household chores. Envisioned barriers were measured with a dichotomous scale (possible/affordable, not possible/not affordable). The data collection team comprised of one female supervisor and five female data collectors from Butajira town who completed at least ten years of education. Data were entered by two staff members. All field research staff enrolled in a two-day training that included study objectives, topics related to maternal health, interviewing skills, role-play, and test questionnaires. For the data-entry staff, specific training was given on SPSS. Completed questionnaires were checked for completeness in the field and households were revisited to complement incomplete data. Quantitative data were then computerized using SPSS 22. Subsequently, all data were double-checked variable by variable and cross-checked between variables by the primary investigator (TV). To investigate which factors were associated with the intention to use the MWH intervention, the sample was divided into two nominal categories: (1) women who indicated they were unlikely to use an MWH (“Potential Non-Users”) and (2) women who stated they were likely to use one (“Potential Users”). In our analyses, these categories are the outcome of interest. Variables were selected based on a literature review, considering their importance in the Ethiopian setting (e.g. previous facility delivery, previous complications), previously found associations in various directions (e.g. decision-making power), and/or to be able to adjust for potential confounders (e.g. maternal age, wealth). Bivariate and multivariable logistic regression analyses were performed using all selected variables from the Adapted Three Delay Model, in order to investigate which of these (women’s decision-making power, previous place of delivery, etc.) influence the outcome of interest. Women with missing responses in any of the selected variables were excluded from multivariable regression. The proportions were calculated using the total number of respondents. Due to some missing responses, percentages will not always add up to 100.0%. Envisioned social and economic barriers of using an MWH (transport to and from an MWH, arranging your own food at an MWH, having to bring your own cooking utensils, etc.) were included separately in a model, to show which of these possible barriers have the greatest influence on the likelihood of utilizing an MWH during the current or next pregnancy. Proportions were calculated using the total number of Potential Users and Potential Non-Users. Using logistic regression, crude Odds Ratio (OR) and adjusted Odds Ratio (aOR) with 95% CI were calculated to measure the effect of each independent variable on the target outcome variable.