Modeling the relationship between women’s perceptions and future intention to use institutional maternity care in the Western Highlands of Guatemala

Women in low- and middle-income countries are increasingly giving birth alongside skilled workers in health facilities. However, many women and their families avoid institutional childbirth care leaving them at risk when complications arise. A frequently cited reason for giving birth at home is negative opinions of institutional childbirth care. This study investigates barriers and facilitators to giving birth in a health facility through the use of a household survey conducted among women of childbearing age in a largely indigenous rural population in the Western Highlands of Guatemala. Women who experienced institutional childbirth either at home or in a facility participated. Those surveyed were asked about specific factors thought to be most related to where women intended to give birth including: access to institutional care, whether they experienced or perceived disrespect and abuse was associated with institutional care, satisfaction with last childbirth experience and whether women believe in a need for facility based care. As part of the survey, women were also asked to explain certain answers. The results indicate women’s perceived need for and satisfaction with their last birth experience most influenced intended future delivery location; and experiencing disrespectful care in a health facility was a barrier to future facility birth. Improved methods of capturing women’s experiences and perceptions of care can help us better understand their past choices and, going forward, inform how to increase demand for and improve institutional childbirth services and related programming.

In 2015, it was estimated that 303,000 women across the globe died due to complications associated with childbirth [1]. Despite average gains, regional estimates mask disparities in health outcomes within countries. Those at highest risk of not receiving adequate care are the geographically isolated, rural poor, residing in certain low- and middle- income countries [2]. Failure to address the needs of “left behind” populations will hinder the achievement of national and global maternal health targets and goals such as universal health coverage (UHC) and the Sustainable Development Goals (SDGs) [3]. Poor outcomes for mothers and babies can largely be prevented through access to emergency obstetric care (EmOC) provided by a skilled birth attendant operating in a sufficiently equipped health facility [4].

To increase use of potentially lifesaving obstetric care we must understand why uptake of services is low. The answer is driven in part by the sociocultural beliefs related to childbirth practices, whether women and families have a perceived need for facility-based care, and if women (and those in their spheres of influence) determine care provided in health facilities is of sufficient or acceptable quality [5]. Despite decades of work invested in measuring experiences and perceptions related to the quality of institutional healthcare, conceptual and methodological issues remain [6]. The challenge has been to construct valid and useful measures of perceptions related to healthcare services, including those specific to the perceived quality of labor and childbirth [7]. Historically, the most common measure of an individual’s healthcare experience is ‘satisfaction.’ Despite popular use, measures of satisfaction have been criticized for a lack of definition and common conceptualization [8]. While it is generally accepted that ‘satisfaction’ represents a collection of distinct interactions and perceptions (which may include a variety of negative and positive experiences), it is often measured as a single-item quantitative measure [7]. Disentangling the confluence of factors that drive satisfaction ratings is a formidable challenge [9].

There are few examples of how to quantitatively measure particular dimensions of poor quality of institutional childbirth care in low- and middle-income countries [10]. One promising approach is the use of the disrespectful and abusive care typology proposed by Bowser and Hill [11]. This typology categorizes specific elements of care (as opposed to the overall experience) by areas identified in the literature as problematic e.g. non-consented care, non-confidential care (including lack of privacy), abandonment/neglect, non-dignified care (e.g. verbal abuse and poor communication), physical or sexual abuse, detention in health facilities for failure to pay (this category has since been expanded to “unfair requests for payment” [12]) and discrimination [13]. In measuring specific dimensions of poor healthcare quality, such as disrespect and abuse, we can hold health systems accountable for what might be masked by (often inflated) measures of satisfaction.

Another (less common) global measure of perceived healthcare quality is ‘willingness to recommend a facility’ to others [14]. ‘Willingness to recommend’ measures have utility in contexts where health service information is passed primarily through social networks. Some argue this is a more accurate measure of satisfaction and indication of future behavioral intentions [15]. Satisfaction and willingness to recommend have been shown to have varying degrees of correlation [14, 15]. Level of satisfaction may not necessarily translate into an equivalent willingness to recommend, suggesting the factors that influence each may be different [14]. Satisfaction measures may tap into an affective evaluation of care [8], while willingness to recommend, could indicate an openness or intention to return to a provider in the future [15]. Therefore, there is a case to be made for evaluating satisfaction and willingness to recommend as separate constructs in their association with institutional childbirth care [14].

The current study explores how women’s first-hand experiences of institutional childbirth care and their perceptions of institutional childbirth care (derived from hearsay, second/third hand accounts of others) are associated with intentions to give birth in a health facility. While intention is an imperfect predictor of behavior, evidence from Bangladesh and Ethiopia suggests intention to give birth in a facility is a significant predictor of whether a woman delivers in a health institution¹ [16, 17]. The present study draws on quantitative and qualitative community-based survey data from the Western Highlands of Guatemala. This paper is one of the first attempts to model women’s experiences and perceptions of institutional childbirth care on future care-seeking intentions. Assessing women’s opinions of care in the context of other key care-seeking factors, such as individual characteristics and access, can shed light on the relative importance of and associations between them. This analysis can inform how to prioritize interventions to increase uptake of facility-based delivery services in an area with relatively low utilization.

The authors developed a conceptual model (post data collection), which illustrates the underlying hypotheses to be tested. The conceptual model shown in Fig. 1 is based on the care-seeking literature and theories of health behavior and health service utilization. Andersen’s model of health service utilization (1960) [5] frames the types of factors hypothesized to facilitate use of health services. These include predisposing characteristics (demographic, social structure and health beliefs), enabling resources (personal/family, community) and perceived need [5]. Building on this frame, the literature identifies key factors associated with where women give birth. These include socio-cultural factors [10, 18, 19]; geographic and economic access [10, 18, 19]; perceived need for facility-based delivery care [10]; previous childbirth experience [18]; and perceived quality of care in health institutions [18, 19]. The conceptual model is concordant with the theory of reasoned action and theory of planned behavior [20], which posit intention to do something is a function of attitudes (behavioral beliefs, evaluation of behavioral outcomes), subjective norms (normative beliefs and motivation to comply) and control over decision-making (control beliefs and perceived power) [21].

Figure 1 illustrates the Conceptual Model of Future Intention to Deliver in a Facility, including the variables from the dataset that are used to operationalize it for the purposes of analysis. Starting on the left-hand side of the figure, intention to use facility-based care is formed by structural and cultural factors (ethnicity, municipality), and individual characteristics (education, age, parity, Cesarean section); these factors shape perceived need for institutional childbirth, perceptions of maternity care quality, and access to skilled care. The conceptual model was developed prior to data analysis and reflects the authors’ hypothesis that the combination of these factors contributes to future intention to deliver in a health facility.

Data for the present analysis are drawn from a household survey concerning maternal and child health among women ages 16 to 51 from 15 villages in Ixil. The data were collected in 2014 as part of the Translating Research into Action (TRAction) Project managed by University Research Co., LLC. Study villages were purposively selected to represent a diversity of residences at different proximities to the only public birthing facility in the municipality. Villages ranged from rural (just outside the municipal centers) to remote (roughly an hour plus car ride from municipal centers, including some villages with poor/no vehicle access). None of the study villages were located on the fringes of the municipal centers.³ The quantitative and qualitative survey data were collected orally by enumerators in Spanish or Ixil (the predominantly spoken indigenous language) depending on respondent preference. Data collectors were women from Ixil possessing similar cultural and linguistic characteristics as the study respondents, but were not from the study villages. The data collectors were hired by the local NGO named COTONEB (Cooperativa Todos Somos Nebajenses), which (at the time) managed health services in Ixil as part of the government’s Extension of Services Program. Data collectors were trained by survey experts contracted by University Research Co., LLC (URC) based in Guatemala. All study participants provided consent before engaging in data collection. Ethical and technical approval were given by URC and the University of San Carlos in Guatemala prior to data collection. URC granted permission to use the data for the present analysis.

For the quantitative analysis, two complementary set of models are estimated using data from a home birth group and a facility birth group. The two analyses allow for considering differences in intention among those with recent experience using childbirth services for their last birth (facility birth group) and those who gave birth to their last child at home (home birth group). The home birth group’s intentions to deliver in a health facility in the future may be influenced by experiences with prior births, or their perception of facilty-based care informed by second-hand experience or hearsay, rather than first-hand experience. Since the questions refer exclusively to women’s last births, we we cannot rule out the former. We believe however, based on the overall prevalence of facility birth in the population and insights from a separate qualitative analyses (not shown) that the likelihood of this is low.

The first step was to estimate descriptive statistics (means and proportions) for all variables used in the analysis, followed by an inspection of the bivariate associations between the independent and dependent variables for each of the two analyses. Inferential statistics are presented to illustrate differences when comparing population characteristics and outcomes between the home and the facility group. Then, we estimate multivariate models to detect independent and potentially mediating effects of perceived need, perceived quality, and access on future intention to deliver in a health facility. The multivariate analyses are informed by the conceptual framework and tested in the form of a series of nested logistic regression models. Marginal and discrete change in probabilities are reported.⁴

While inferential statistics are used, these data concern the entire population of women with a child born in the past five years located in the 15 study villages. As such, the estimates presented are population parameters. While inferential statistics are presented in the analyses that follow, these may be best interpreted as approximate measures of the variability around the parameters they represent; from a counterfactual perspective, this population may be thought of as representative of similar (mostly indigenous) populations in the country.

Across both the home and facility birth groups, the results suggest the care seeking factors of interest—access, perceived need and perceived quality—demonstrate generally expected associations with future intention to deliver in a health facility, with a few exceptions. In terms of access, living in relatively closer proximity to the health facility increases the probability of intending to give birth among the home birth group, but the association was less compelling for the facility birth group. Perceived need, represented by willingness to recommend, was associated with a higher probability of intending to have a future facility-based birth among both groups. For women’s perceptions of care quality, satisfaction with last childbirth experience reinforced intention to return to the same location; and as expected, women who reported experiences of disrespect and abuse during their last facility based delivery had a lower probability of intending to return to a facility in the future. The measures of perceived disrespect and abuse (abandonment and unfair requests for payment) among the home birth group did not demonstrate conclusive associations in this context.

The results indicate the original conceptual model could be modified for each group to better capture the estimated relationships among the variables in this context. For example, the analysis presented here suggests that, in this particular population, the association between cultural, structural, and individual characteristics and future intentions to use facility based maternity care are partially mediated by perceived need for institutional care (in both groups) and satisfaction with prior childbirth experience in the home group. For the facility group, physical access to care and perceived care quality (disrespect and abuse and satisfaction) additionally had independent effects on future intentions to use facility-based maternity care.

The two types of perceived quality of care measures—disrespect and abuse and satisfaction—generated mixed results. For the home group, satisfaction with home birth is the main driver of intentions to give birth at home again. The two measures of perceptions of disrespect and abuse did little to help explain the home group’s future intentions. As originally hypothesized, the facility group analysis indicates poor satisfaction ratings and reports of disrespect and abuse are a deterrent to seeking facility-based care in the future. Further, these results suggest satisfaction and the disrespect and abuse measures capture different aspects of perceived quality of care. This is evidenced by the fact that when both measures are estimated simultaneously, satisfaction is adjusted downward in the presence of disrespect and abuse.

The findings from this analysis may be generalizable to other rural indigenous populations in Guatemala, but several limitations must be kept in mind. While a limited number of variables were included, (due to the relatively small sample size,) key factors identified in the literature as being associated with uptake of facility-based childbirth care informed the conceptual model and are included in the models. Other important care-seeking factors may have been omitted from the current analyses, and some measures (structural and social factors and individual characteristics) may be imperfect proxies of these constructs. Going forward it would be useful to include additional information about the facility group care experiences such as duration of the stay in the health facility, cost of care, degree of obstetric complication and health outcomes of mother and baby. These are factors that may negatively influence future intention to seek facility-based care in the future [33]. Additionally, controlling for whether women’s preferences were accommodated (i.e. presence or absence of respectful practices), such as being allowed to give birth in one’s preferred position and observe cultural practices as outlined in the Healthy Motherhood Law [25], may increase future intention to return to a health facility to give birth. For the home birth group, it could be useful to control for religion or spiritual beliefs/values [36] and measures of empowerment or autonomy (given the low status of women in Guatemalan society) [33] as those factors may influence a woman’s ability or intention to use institutional care.

Another limitation may stem from recall bias when questioning women about experiences that may have occurred within a five-year time-frame. However, such biases may be minimal. A study from Ghana suggests that women’s recollections of obstetric events were found to be accurate, even more than five years later, though this may pertain to more tangible events rather than subjective feelings [37]. There is potential uncontrolled for heterogeneity within the home birth group given the possibility that some women may have had a facility birth experience prior to her last home birth, though the assumption is that this was rare and the subsequent effect on the results unsubstantial. The normalization of disrespect and abuse combined with low expectations and relatively less experience with institutional childbirth care may have contributed to “under-reporting” of disrespect and abuse. This should be partially adjusted by the inclusion of education/literacy and wealth; additionally, we are concerned with subjective reports of women’s experiences as they are what drive care-seeking (as opposed to “objective” measures of disrespect and abuse or satisfaction). Courtesy bias may have affected responses to the questions about willingness to recommend or intention to use maternity care in the future. However, the variation and distribution of responses increase confidence in the findings. For example, only 25% would recommend facility birth and intend to deliver in a facility in the future, and 37% of the total study population would neither recommend nor intend the former. While there is potential for reverse causality of the association between willingness to recommend and future intention to deliver in a health facility, the analysis of the accompanying qualitative data reveals willingness to recommend is a proxy for perceived need for institutional care. A woman is not likely to report an intention to do something for which she does not perceive a need.⁸

This research contributes to the literature by testing two types of self-reported quality of care measures: a ‘global’ measure of satisfaction and specific measures related to poor perceptions of quality that capture disrespectful and abusive service provision. While there is conceptual overlap between satisfaction with facility care and reporting specific instances of disrespect and abuse, the facility group analysis suggests the measures largely capture different aspects of the relationship between perceptions of quality and future delivery location preference. Global satisfaction ratings often lack variation and are skewed to the positive end of the spectrum [38]. This appears to be true in this case as more than 90% of women reported high to very high satisfaction with their last birth experience at home or in a health facility.

Reported high satisfaction with facility care is positively associated with intent to give birth in a health facility again in the future. However, the facility group’s high satisfaction ratings may be less of an indication of stellar service provision and more a function of low expectations or other confounders [39]. The qualitative data suggest the expectation in the case of a facility-delivery is tied to survival, instead of the traditional expectations of childbirth care in the home. Reports of high satisfaction among the facility group may be a better measure of ‘acceptable’ care instead of ‘highly satisfactory’ care [8]. High satisfaction among the home group was the most substantively and statistically significant factor associated with future birth location. Satisfaction with home care may represent the overall preference for home birth, which is why it is a major deterrent to future intent to seek institutional childbirth care. An analysis of the qualitative open-ended responses reinforces why women prefer to give birth at home.⁹ As hypothesized, satisfaction with (or preference for) home birth was found to mediate perceived need for institutional care, distance and municipality. Future research could further examine the relationship among specific instances of disrespect and abuse experienced during home birth (which was likely partially captured by home satisfaction ratings) and intention to seek facility birth care.

The nested model results from the facility group indicate the satisfaction estimate is adjusted by the inclusion of disrespect and abuse, meaning disrespect and abuse explains variation apart from that explained by satisfaction. Further, only 9 % of the facility group reported average to very poor satisfaction with their care compared to the 18% who reported disrespect and abuse. This supports the hypothesis that satisfaction may represent an affective evaluation of quality that could discount or fail to capture specific (negative) experiences [40]. The results indicate the measures of perceived quality (satisfaction versus experiences of disrespect and abuse), when operationalized in this context, are interpreted as different constructs likely limiting issues of potential endogeneity due to a correlation between the perceived quality measures and the outcome.

The perceptions of disrespect and abuse incorporated in the home group analyses were not significant or substantively associated with future intentions to use facility-based childbirth care. Perceived abandonment in health facilities had the anticipated negative association, but a virutally null estimate. The isolating effects of abandonment were potentially captured by the language variable and municipality of Nebaj where poor care is most commonly perceived. Perceptions that one must pay or give something for better care (unfair requests for payment) is associated with higher likelihood of facility-based delivery intentions. This could be due to the general expectation of costs related to facility-based care. “Unfair requests” may not be viewed as distinct from other payments. Future research should explore other potentially important perceptions of quality that may explain contributors to low intended use of facility-based childbirth care.

The decision about where to give birth is a complex social process that can lead to delays in seeking emergency obstetric care. The qualitative data suggest respondents must balance sub-optimal quality of services with believing health facilities are safer, which is concurrent with other research indicating acceptability of facility birth in the event of an obstetric complication [19]. The qualitative and quantitative responses suggest perceived need is an important mediating factor for both groups’ future care-seeking. Willingness to recommend or perceived need for institutional care demonstrates an awareness of the susceptibility to and severity of obstetric complications, making them more likely to utilize facility-based care [17]. Sixty one percent of the respondents, from both groups, are willing to recommend a facility, suggesting an acceptance of and perceived need for institutional care. However, less than a third of women intend to have a facility-based delivery. This figure is not much higher than the proportion of women from the Department of Quiché who reported a facility birth during the last maternal and child health survey [27].

Among the structural, cultural and individual characteristics, only literacy/some education was seen to be associated with a higher likelihood of intending to deliver in a health institution (for both groups). It is not surprising that the estimated probability of intending to give birth in a health facility for those who are literate/have some education is twice as large for the facility group compared to the home group. Among the facility group, indigenous language was a significant deterrent to future intentions to repeat a facility delivery, indicating indigenous populations may not feel comfortable with the institutional childbirth processes and interactions—potentially a result of their inability to communicate [28] or because of differential treatment of indigenous populations [30]. Women with high parity from both groups have a relatively weak lower discrete probability of intending to deliver in a health facility in the future compared to those with fewer children. This could be a function of having sufficient success at home that one does not perceive a need for facility-based care or lack of intention to have more children. Living in the municipality of Nebaj is also strikingly associated with decreased intentions (overall) to deliver in a health facility, especially among the facility group. This could be because Nebaj’s referral hospital is associated with poorer treatment and/or infrastructure (suggested in a separate analysis of qualitative data, not shown). Alternatively, this health facility offers more surgical and other medical interventions that may be feared by respondents. The potential stigma surrounding the Hospital of Nebaj is unfortunate as this is the only public health facility capable of managing a complicated delivery in Ixil. Age was not significantly associated with future childbirth intentions for either group.

For hard-to-reach communities, access played less of a role in determining future intention than anticipated. Living in closer proximity to health facilities was associated with higher discrete probability of intending to have a facility birth for both groups. Proximity to the health facility was a stronger and more significant predictor of childbirth intentions for the home group compared to the health facility group. Closer access may be a more significant pull factor for the home group because better physical access means less of a perceived investment in time and money to reach care, making the thought of facility-based care utilization more likely. While closer physical access to health facilities does increase likelihood of use, the facility group already may have navigated physical and geographic access barriers to utilize facility care in the past, potentially minimizing the importance of proximity to care when considering future intentions. Notable is the relatively small magnitude of association between wealth and future delivery intentions for both groups. This could be explained by the fact that facility delivery is not a preference, but driven by obstetric emergencies, in which case families find a way to invest in facility-based delivery. This interpretation is reinforced by the strength of the magnitude and direction of ‘willingness to recommend,’ which again, points to the perception of need driving intention to use facility-based care. The findings suggest the access variables may be better categorized as controls since wealth and distance do not appear to play a mediating role as originally hypothesized.

Women’s perceptions of care quality influence future care-seeking for childbirth services. Facility group women had a lower probability of returning to a health facility if they were dissatisfied with their last care experience or reported being disrespected. Further, the home birth group’s satisfaction with home childbirth care is the driving force for maintaining a preference for giving birth at home. Interestingly, the qualitative data revealed the majority of dissatisfied women who gave birth at home indicated they were mistreated. Whether mistreatment at home could be a driver of facility delivery is an area for future investigation.

This study illustrates that measurement of a complex phenomenon like perceived quality (e.g. examples of disrespect and abuse) can be successfully conducted in geographically hard-to-reach places and executed in multiple languages (e.g. Spanish, Ixil) at a relatively low cost and in a short time period. The willingness to recommend a facility birth question could serve as an indicator of increasing demand in the event of an obstetric emergency and may foreshadow a shift in preferences. This is an example of another indicator that could be further tested more widely in this and other contexts. The use of qualitative data validated the understanding of the quantitative willingness to recommend survey question along with the overarching disrespect and abuse question. There is potential to adapt or build upon the willingness to recommend and disrespect and abuse indicators for future population-based or more routine data collection efforts. These measures could provide a source of data to monitor the implementation of laws, policies and programs; for example, the operationalization of elements related to mistreatment or respectful care programming and practice outlined by the Healthy Motherhood Law and other existing intercultural care guidelines in Guatemala. At the health facility-level, screening for past dissatisfaction and negative experiences as part of the antenatal history to address demand issues could be incorporated as a strategy for increasing facility care. Feedback from clients and continued measurement of specific negative experiences (instead of just satisfaction) can inform efforts required to increase uptake and quality of care.

It is unlikely that the prevailing preference in historically marginalized rural communities will rapidly shift from home birth to facility birth. While utilization is still relatively low, it is promising to see over half of the women from the 15 rural communities signal the acceptance of facility care in the case of an obstetric emergency. The key implication of this work is that a failure to address perceptions of quality and how women are treated in health facilities can ultimately result in a delay or avoidance of seeking institutional childbirth care the future—thus rendering women and newborns more vulnerable to mortality and morbidity related to childbirth. Now is the time to enhance efforts to measure women’s experiences with health facilities and to focus on accountable implementation of policies and guidelines related to advancing respectful care.