A mixed methods assessment of barriers to maternal, newborn and child health in Gogrial West, South Sudan

Human subjects protection

Ethics review was obtained from the Ministry of Health (MOH) and conducted in accord with the Declaration of Helsinki [6]. Verbal consent to participate in the study was obtained from all participants (quantitative and qualitative respondents). Every effort was made to ensure protection and confidentiality and to reduce any potential adverse consequence to the participants. Participants did not receive any material compensation. Participants were informed that participation or lack thereof would not affect their access to or the quality of the care they receive, and were explicitly given the right to refuse participation.

Geographic scope

This multi-stage clustered and randomized household study was conducted to represent all of the population in 8 of 9 accessible payams in Gogrial West, Warrap State, South Sudan. Within the payams, qualitative interviews were conducted to add nuance to the issues found in the household survey.

Sample size

To determine an appropriate sample size for the quantitative study, we assumed 50% of women in Gogrial West had accessed MNCH services. Therefore, the sample size required to estimate access to MNCH services through a random sample with 95% confidence (margin of error of ±5%) was 770 households assuming a design effect of two.

Target population

To better define and understand existing barriers to MNCH, household women who were pregnant or who had children under the age of 5 years or a male household member who had a pregnant wife and or children under the age of 5 years were randomly selected to be interviewed for this study.

Sampling frame

Gogrial West (one of 10 counties in Warrap State) is divided into nine payams and 28 bomas. Population information for the eight of the nine accessible payams was 221,795 persons and 229 “main” villages excluding Akon North, which was inaccessible due to flooding and security [7]. Villages were randomly selected from the MOH approved list of “main villages.” Due to the rainy season, many of the villages in this list could not be accessed; therefore, only villages accessible could be randomized. A list of randomized alternate villages was generated for cases where the randomized village could not be accessed for any reason. For any villages not accessible, the nearest neighbor was used for substitutions. Substitutions for initial randomized villages occurred in 5 of 30 clusters (17%) due to inaccessible roads due to the rainy season.

To allow an approximate probability-proportional-to-size sampling by payam strategy, existing information on the community sizes, as provided by the Republic of South Sudan National Bureau of Statistics, was used to create relative weights to allocate population based proportional distribution of the clusters.

Quantitative participants were selected by via systematic random sampling which utilized a 30 × 30 cluster sampling methodology (30 randomly selected clusters of 30 households) to generalize to the largest population possible. To account for refusals, we targeted our sample size to be 900 households. Households within villages were sampled according to a modified World Health Organization (WHO) EPI Method [8]. Limited time in each cluster (4–6 h) due to security constraints and distance, and poor accessibility and flooding required using purposive and systematic transect selection aimed at covering as many zones as possible. Surveyor transects were selected either from the village center (or a village main accessible point), or from different points of main roads. In villages with large populations, to reach households that were further from village center, surveyors were instructed to skip every second household on their transect.

Quantitative instrument

The quantitative survey was developed for the Gogrial West context to elucidate the barriers to MNCH and was based on the continuum of care approach that serves as a core principle of programs for maternal, newborn, and child health, and as a means to reduce maternal, neonatal and child deaths [9]. In doing so, these questions would elucidate the barriers to MNCH and highlight where there were barriers throughout the reproductive lifecycle (adolescence, pregnancy, childbirth, the postnatal period, and childhood) and between places of caregiving, which can include households and communities, outpatient and outreach services, and clinical-care settings. The instrument contained 98 questions on respondent demographics, reproductive health, antenatal care service use, postnatal care, breastfeeding, child health, sexual gender based violence (SGBV), and opinions. Given the surveyor’s preference for English (and the inability of the surveyors to write in Dinka), only key words were translated into Dinka. The instrument was verbally translated into Dinka by the group and individuals were tested to ensure standardization. They were monitored by two Dinka speaking individuals. Observed role-play and field testing was designed to both ensure functional equivalence of the questions across the two languages as well as adjust for cultural issues.

Qualitative instruments and sampling

Qualitative interviews added nuance to the issues identified in the quantitative instrument and were administered by researchers with the aid of a translator. The qualitative instrument contained 12 open-ended questions relating to barriers to care with regard to MNCH, causes of malnutrition in communities, roles within families, decision-making, cultural norms, traditional practices and community needs. Interviews were used to identify common themes regarding gender barriers and cultural norms affecting MNCH within the context of Gogrial West, Warrap State. Convenience sampling was used to allow flexibility in accessing individuals who might be important for informing the quantitative data. As these individuals were sampled and used to add nuance and context to the quantitative survey, there was no set number of individuals needed. A total of 72 key informants and 25 community members-10 males and 15 females, were interviewed. Interviews included women and men at the village level, key informants including community leaders/healers (religious/tribal/ethnic), beneficiaries, NGO personnel, Ministry of Health, United Nations, World Bank, healthcare providers, traditional birth attendants (TBAs), and community health workers (CHWs).

The information from these individuals was handwritten in the field and used to provide additional insights into the themes and findings that emerged from the quantitative survey. Handwritten notes were then transcribed into a PDF form on the computer that could easily summarize narrative testimonies into themes within all the interviews transcribed.

Training

Four women and six men to serve as surveyors were recruited from the Warrap State. Training included a 4-day training with a standardized curriculum including observed role-play and field testing [10]. Supervisors were present continuously throughout the survey period to ensure data quality and veracity.

Household sampling

Teams arrived early in the day and in some cases interviewed respondents while they worked in gardens. Upon arrival to a village, the boma chief was contacted to request permission for the survey. A small token of appreciation was extended to the chief and in return they asked residents to be available for the survey. Surveyors conducted a one-on-one, anonymous interview with a randomly selected male or female household member in a setting that afforded privacy and confidentiality. If no one met the criteria, the next household to the right was surveyed. If only one sex or adult was present at the time of the household visit, that person was interviewed regardless of sex. Surveyors were not assigned by sex. Interviews took approximately 30–40 min and were conducted in Dinka. Surveyors completed 6–7 surveys per day. All surveys were reviewed for completeness and correctness at the end of each cluster.

Statistical analysis

Quantitative household data were analyzed using XLSTAT. Analysis involved the estimation of weighted population means and percentages (descriptive statistics and cross-tabulations), using 95% confidence intervals and considering p-values as significant when less than 0.05, when comparisons by age, age of marriage, wife status and wealth were to be established. Weighting was necessary to keep the per payam proportion of male (of ages 20 years or above) and female (of ages 15–49 years) population in our collected sample equal to that of the actual population in Gogrial West (excluding Akon North). Wealth was measured in a comprehensive and context adapted manner, through the use of both a wealth indicator and wealth groups constructed through a Principal Components Analysis and a k-means classification respectively. The latter were calculated based on size of crops, livestock, mobility and communications assets and monthly income of households.

Qualitative narratives transcribed onto a standardized PDF form were compiled into a master Excel spreadsheet. Themes were hand coded by issues identified in the household quantitative study, analyzed and used to add nuance to the quantitative data.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Demographics

A total of 860 households consented to the study (860 women and 144 men) with a response rate of 99.4% among women and 99.3% among men (Fig. 1). The average age of female respondents was 28.5 years and 40 years among males (Table 1). Women had less than 1 year of formal education (0.3 years), whereas men had just over a year of formal education (1.1 years). Qualitative interviews revealed that once married, women and/or girls do not continue in school. Agreements are made in some circumstances to allow continued education after marriage but in practice most husbands do not pay school fees. In addition, although women and/or girls may be in school after marriage, as soon as she is pregnant, she can no longer attend school.

Characteristic	N/Total Respondents, Female	Females Weighted % (95% CI)	N/Total Respondents, Male	Males Weighted % (95% CI)
Agea, mean, y	852/860	26.0 (16.0,47.0)	142/144	40.0 (18.0,75.0)
Education
None	833/860	93.0 (91.2,94.7)	137/144	77.7 (70.8,84.7)
Years (formal)	833/860	0.3 (0.2,0.4)	137/144	1.1 (0.7,1.6)
Unable to read	803/860	95.4 (94.0,96.9)	136/144	83.3 (77.0, 89.6)
Marital status	851/860		144/144
Never				0.8 (0.0,2.2)
Married		95.5 (94.1,96.9)		97.6 (95.0,100.0)
Widowed		3.8 (2.5,5.1)		0.8 (0.0,2.2)
Otherb		0.7 (0.1,1.2)		0.9 (0.0,2.4)
Wife number	850/860
First		62.4 (59.2,65.7)
Second		21.1 (18.4,23.9)
Third or greater		16.5 (14.0,19.0)
Number of wives			137/144
Mean				1.8 (1.7,2.0)
None				0.8 (0.0,2.3)
One				43.4 (35.1,51.7)
Two				35.1 (27.1,43.0)
Three or more				20.7 (14.1,27.7)
Bride Pricec	819/860	25.9 (25.0,26.9)
Age of marriage, y	819/860	17.7 (17.6,17.9)
Occupation	843/860		143/144
Farmer		75.5 (72.6,78.4)		63.2 (55.3,71.1)
Homemaker		15.7 (13.2,18.2)		4.6 (1.2,8.1)
Small business		5.9 (4.3,7.5)		13.4 (7.8,19.0)
Government		0.2 (0.0,0.5)		10.9 (5.8,16.1)
Otherd		2.7 (1.6,3.8)
Monthly income, mean USD	763/860	11 (10,12)
Water
Enough availablee	840/860	68.4 (65.2,71.5)
Distance to source (mins)	787/860	32.7 (30.0,35.3)
Time fetching/d (hrs)	765/860	2.6 (2.5,2.8)
Fetched water while pregnant	844/860	97.4 (96.3,98.4)
Water source while pregnantf	857/860
Protected source		83.9 (81.5,86.4)
Unprotected source		16.1 (13.6, 18.5)

^aQualitative interviews revealed that female respondents did not know their age. Respondents stated “unless a girl or woman is in school, she may not know her age” or “you will need to ask my mother;” therefore, women’s age is likely an estimate

^bOther: living with partner, spouse missing/killed in conflict/divorced/separated

c This estimate assumes one cow has a value of 250 USD, and was equivalent to 5 goats or 12 sacks of durah which was counted as half a cow. This price represent the price paid in cows to the wife’s family for marriage

^dOther: day labor, student

^eEnough, yes or no for drinking, washing hands, bathing for everyone in the household

^fProtected water source: borehole, protected well/spring; Unprotected water source: Unprotected well, Unprotected spring, River/stream, Lake/pond, Rain catchment

Contraception

More than 90% of women and men (96.6%) did not want contraception. Therefore, of the remaining 10%, only 1.2% of women aged 15–49, who wanted family planning, were able to meet their need for family planning (Table 2). The most common barriers included, lack of desire for contraception, contraception is “not traditional”, spouse’s refusal to allow use, and the lack of availability. Women married before 18 years, were less interested, or desired using contraceptives compared with women who married after 18 years [6.4%; 95% CI (3.9,8.9%) vs 9.3%; 95% CI (6.6,12.0%), p < 0.05]. A 21-year-old woman stated: “…men value themselves by the number of cows and wives they have; women by the number of children.” For men, the reasons were not about worth but for fame “…Our name will be known and we [our name] will not disappear…if we marry a girl who is young, we are assured to have many children and we will be famous.”

Care during pregnancy

On average, pregnant women presented for antenatal care (ANC) 2.3 times and rarely by skilled providers (Table 2). Only 31% of women had verifiable (card present) two doses of tetanus toxoid immunization (TT2+). Forty-seven percent households had a mosquito net and even fewer (21.5%) had an insecticide treated nets (ITN). Less than 10% of pregnant women slept under ITN mosquito nets. For household members, only 39.4% slept under a mosquito net if it was present in a household. Pregnancy danger sign recognition (≥3 signs) was 53.4% among men and 49.6% among women. Qualitative interviews revealed that ANC were primarily sick visits. Women stated costs of services ranged from 2SSP to 150 SSP (20 cents to 15 USD).

Childbirth and postnatal care

Just over 75% of women had a birth attendant, however only 4.6% had a skilled attendant at delivery. Most women (51.6%) used a TBA/village midwife (Table 2). Wealth had no significant effect on access or use of MNCH services. Interviews revealed clinics were not well-equipped for births, as they did not have skilled attendants for delivery. In addition, interviews indicated cultural preferences for home deliveries. TBAs were both trained and untrained; many used unsterile techniques. Village surgeons (Arets) were consulted for obstructed births. Arets used a scalp hook to dislodge live infants, an unsterilized spear for dismemberment and removal of stillbirths. Spiritual and traditional leaders treated a variety of maternal health issues around pregnancy and childbirth, most often through casting spells and medicinal roots. Immediate breastfeeding rates were high (98.2%) and exclusive rates low (5.1%). Interviews revealed that only women who were ill or had “no milk” did not breastfeed. A variety of supplements were used for infants less than 6 months; water being the most common. Commonly held myths regarding nutrition prohibit breastfeeding women from consuming groundnuts. Children are not fed eggs or chicken as the belief is that children will not talk or will have elevated blood pressure if chicken is consumed. In general, men determine breastfeeding length. Women believe that malnutrition results from engaging in sex while breastfeeding, therefore, many try to abstain from sex while breastfeeding but struggle as decision-making power lies with the male partners. Only 7.9% of women accessed postnatal care, which was largely done by unskilled providers. The danger signs of childhood illnesses and essential newborn care were largely unknown among male and female respondents.

Child health

Violence

Brief summary of findings

Barriers to MNCH were more than lack of access to focused ANC. Women in Gogrial West had minimal education and little desire for contraception which impact maternal health and mortality [11, 12]. ANC was largely non-existent and consisted of sick visits instead of focused ANC and clinical facilities with a dearth of appropriate skilled healthcare workers to administer ANC. Mosquito net use, although high for the area, still represented only a small proportion of households putting women and children at risk, and especially pregnant women at risk for malaria. Immunization status of both women and children was low. Exclusive breastfeeding rates of 5% put a significant number of newborns and infants at risk for diarrhea and death [13, 14]. The overall lack of knowledge of danger signs, low use of skilled attendance at birth and the preferred use of largely untrained TBAs put women and newborns at significant risk in Gogrial West and represented barriers to MNCH that given the current situation will be difficult to change [13, 14]. In addition, a dearth of facilities that meet basic or emergency obstetric (EmOC) care criteria, lack of transportation and access, especially during rainy season, for obstetric emergencies leave women at the hands of untrained birth attendants and delivery at home. The time and distance needed to fetch clean water, especially while pregnant and the overall lack of clean water will continue to hamper MNCH. Finally, the presence of systematic and accepted violence in the community, largely addressed to women and children due to negative social norms will continue to add to barriers to health for women, newborns and children.

Context of findings

Programmatic implications

Limitations

Flooding during the rainy season limited access to areas randomly chosen for surveying. Akon North was not represented in this study as access required movement into volatile areas and the time to reach Akon North did not support ICRC guidelines for security. Although data collectors were careful to explain that there will be no material or other gain by participation in the assessment, respondents might have exaggerated or underestimated responses if they believed it would be in their interest to do so. In some instances, (e.g., when interviewing intended beneficiaries), responses might have been constrained due to fear of reporting or stigma such as with questions around SGBV, however, based on our qualitative study, it was noted that SGBV was normalized and responses were less likely to be constrained. Although it is possible that differences within the interviewer due to ethnicity, sex, or overall comfort level during the interview could bias the results, this was mitigated largely through the use of local data collectors, properly trained in interviewing techniques who did not interview in areas they are familiar with. According to most versions of customary law, death does not terminate a marriage. A widow is therefore still considered married to her deceased husband. This made getting true numbers of widow headed households difficult as the data collectors marked them as “married.” Data cannot be extrapolated to represent other counties or the entire State. However, the population of Gogrial West is homogeneous and therefore it is likely the study has practical application to Dinka populations beyond the sampling frame.