Companionship during facility-based childbirth: Results from a mixed-methods study with recently delivered women and providers in Kenya

Background: Research suggests that birth companionship, and in particular, continuous support during labor and delivery, can improve women’s childbirth experience and birth outcomes. Yet, little is known about the extent to which birth companionship is practiced, as well as women and providers’ perceptions of it in low-resource settings. This study aimed to assess the prevalence and determinants of birth companionship, and women and providers’ perceptions of it in health facilities in a rural County in Western Kenya. Methods: We used quantitative and qualitative data from 3 sources: surveys with 877 women, 8 focus group discussions with 58 women, and in-depth interviews with 49 maternity providers in the County. Eligible women were 15 to 49 years old and delivered in the 9 weeks preceding the study. Results: About 88% of women were accompanied by someone from their social network to the health facility during their childbirth, with 29% accompanied by a male partner. Sixty-seven percent were allowed continuous support during labor, but only 29% were allowed continuous support during delivery. Eighteen percent did not desire companionship during labor and 63% did not desire it during delivery. Literate, wealthy, and employed women, as well as women who delivered in health centers and did not experience birth complications, were more likely to be allowed continuous support during labor. Most women desired a companion during labor to attend to their needs. Reasons for not desiring companions included embarrassment and fear of gossip and abuse. Most providers recommended birth companionship, but stated that it is often not possible due to privacy concerns and other reasons mainly related to distrust of companions. Providers perceive companions’ roles more in terms of assisting them with non-clinical tasks than providing emotional support to women. Conclusion: Although many women desire birth companionship, their desires differ across the labor and delivery continuum, with most desiring companionship during labor but not at the time of delivery. Most, however, don’t get continuous support during labor and delivery. Interventions with women, companions, and providers, as well as structural and health system interventions, are needed to promote continuous support during labor and delivery.

We employed mixed methods for this study. Mixed methods combine elements of qualitative and quantitative approaches to provide breadth and depth in our understanding of phenomenon [22]. Quantitative methods allow objective measurement of a particular reality, with a large enough sample to increase generalizability [23]. Quantitative methods are however limited in explaining the “why” of phenomena and how personal viewpoints, context and meaning produce multiple realities [23]. Qualitative methods help to address this weakness, but have limited generalizability because of practical limitations of sample size. Mixed methods were therefore appropriate to harness the strengths and counterbalance the weaknesses of quantitative and qualitative methods to provide a holistic understanding of birth companionship [23]. The quantitative methods provide data for measuring prevalence and examining statistical associations and the qualitative data help to explain the quantitative findings and provide rich descriptions of views and beliefs about birth companionship. In addition, we collected data from recently delivered women and maternity providers to obtain the complementary perspectives of both the recipients and providers of care. Data for this study were part of a larger study examining community perceptions of quality of maternity care in a rural county in western Kenya. The County has a population of about one million and an estimated 40,000 annual births [24]. Approximately 43% of the population live below the poverty line and only about 3% of women of reproductive age have more than a secondary education. About 24% of women of childbearing age are 15–19 years [25]. The total fertility rate of the county is 5.3 compared to the national average of 3.9. The County is one of 15 counties in Kenya that account for over 60% of maternal deaths in the country—with an estimated maternal mortality ratio of 673 deaths per 100,000 live births compared to the national average of 495 [26]. The number of nurses, clinical officers, and doctors, per 100,000 people in the county is 32, 19, and 4 respectively [27]. About 53% of births in the County occur in health facilities, compared to the national average of 61% [25]. Several factors account for low use of facility deliveries, including low perceived need of facility deliveries, poor physical and financial access to health services, and low perceived quality of care [28–30]. Additionally, sociocultural factors such as norms and values related to childbirth (e.g. beliefs about the causes of pregnancy complications) and women’s autonomy (e.g. women not being final decision makers on place of birth) interact with socioeconomic factors such as wealth and education to determine whether or not women seek health care during pregnancy or delivery [5, 31–33]. Thus, the low status of most childbearing women in the County (young, uneducated, poor, and unemployed) is likely a key determinant of the low use of services in the County. These factors also affect how women are treated in the health facility, which affects future health seeking behavior. Given evidence that birth companionship improves women’s birth experiences [5, 9, 17], understanding and respecting women’s desires and preferences for birth companions could positively impact health seeking behavior during childbirth in the County. We used 3 data sources: surveys and focus group discussions with recently delivered women and in-depth interviews with providers. Eligible women were 15–49 years and delivered in the 9 weeks preceding the study. Eligible providers worked in a maternity unit in the county. Ethical approval for the study was provided by the authors’ institutions. All participants provided written informed consent after receiving information about the study and received 200 KES (~$2) for their participation. Survey data were collected between August and September 2016 with eligible women. A multistage sampling approach was used to select women. First, the County was divided into eight strata based on its 8 sub-counties. Next, ten health units were randomly selected from each strata. Eligible women were then identified and recruited from the selected health units. The sampling procedures are documented elsewhere [34]. Twelve data collectors were trained to conduct the interviews—in English, Swahili, and Luo in private spaces in health facilities or in respondents’ homes. Data were collected using the REDCap electronic data capture tools hosted at the University of California-San Francisco [35]. Approximately one thousand women were interviewed, with a response rate above 98%. We use data from women who delivered in a health facility (894) and had no missing data on relevant variables for this analysis (N = 857–877 for different outcomes). Between October and November 2016, we conducted eight FGDs—one in each sub-county—with 58 eligible mothers. We chose a focus group design to obtain qualitative data from women because our field experience in the region suggested women were more willing to discuss their experiences in groups with their peers than with an individual interviewer not familiar to them. The FGD procedures are documented elsewhere [36]. We recruited respondents from one randomly selected health unit in each sub-county that was not included in the survey. Women who met the eligibility criteria were purposively selected from the selected health unit. Each focus group consisted of six to ten women and lasted about 90 min. Two female research assistants with college degrees were trained by the first author to moderate each group discussion; one led the discussion using a discussion guide, and the other took notes and managed audio recording. Women were asked, among other things, if they were allowed to have a companion present during labor and delivery and how they felt when they were allowed or disallowed a companion. They were also asked if they desired birth companionship and why they desired or did not desire it. The discussions were conducted in Swahili or Luo, in private spaces in the community or health facility. Discussions were audio recorded, and simultaneously translated and transcribed. We conducted 49 interviews with maternity providers from 18 facilities across all the 8 sub-counties, which were selected for an intrapartum quality improvement project based on their relatively higher volume of deliveries. With permission from the county and facility heads to conduct the study, the research assistants approached maternity providers in selected facilities, briefed them about the study, and invited them to participate in the interviews. The interviewers used an interview guide with both closed and open-ended questions. Providers were asked, among other things, if women were allowed CSLD, their perceptions of CSLD, and barriers to providing it in their facilities. Interviews were conducted in English, Swahili or Luo—in private spaces in each health facility—and lasted about an hour. The structured responses were directly entered into the REDCap application during the interview. Interviews were also audio-recorded and transcribed (with simultaneous translation where necessary). Our dependent variables are from several survey questions on birth companionship during: (1) labor, (2) delivery, and (3) post-delivery. These include questions on whether they were accompanied by a companion to the facility, their relationship to the companion, whether they were allowed continuous support during labor and during delivery, whether they desired companionship during labor, delivery, and after delivery, and their preferred type of companion. The variables are described in Table 1 with exact wording of questions in Additional file 1. Description of Birth Companionship variables These include various socio-demographic, facility, and provider related factors that might be associated with the dependent variables. These were selected based on existing literature and theoretical reasoning. They include age, marital status, parity, education, literacy, employment, household wealth, tribe, religion, prior facility delivery, antenatal attendance, pregnancy complications, labor and delivery complications, type of delivery facility, type and sex of delivery providers, and women’s perceptions of crowding in the facility. In multivariate analysis, we controlled for potential reporting bias by place of interview and post-partum length. First, we conducted descriptive analysis to examine the characteristics of the sample and to assess the prevalence of birth companionship and CSLD, as well the proportion of women who desired companionship and their preferred companions. Next, we used cross tabulations and bivariate logistic regressions to examine the bivariate associations between the dependent and independent variables. We also examined whether the type of companion was associated with being allowed CSLD and women’s desire for companionship. Finally, we used multivariate logistic regression to examine the factors associated with being allowed CSLD, net of other factors. Using the Braun & Clarke approach (2006), we analyzed the qualitative data thematically to identify patterns within data [37]. We coded data inductively, considering both the semantic (surface) and latent (underlying) meaning of the text, focusing on salience rather than frequency. We iteratively read and re-read the transcripts and coded line-by-line across the entire data set. Two coders (the first and second authors) double coded half of the transcripts and compared codes to check consistency. We then analyzed initial codes to generate categories and identify themes [37]. We again reviewed transcripts until no new themes emerged. See Additional file 2 for how open codes were used to generate categories and themes. This was however an iterative process with constant reviews and revisions. Throughout the process, we wrote analytic and reflexive memos to capture emerging ideas and examine our assumptions, preconceptions, and reactions to the data. For example, we started the study with the preconception that all women will be unhappy about being denied companions. We were thus surprised by some women’s reactions to not being allowed companions and the many reasons they gave for not desiring companions. We therefore wrote a memo on this early on to ensure we adequately captured women’s perceptions. We managed data using Atlas.ti. (COREQ checklist in Additional file 3).

• afulani, patience a.
• kusi, caroline
• kirumbi, leah
• walker, dilys