A systematic review of person-centered care interventions to improve quality of facility-based delivery

PCC frameworks are complex and feature multiple domains, a fact that may hinder intervention design and result in slow translation of PCC objectives into practice. PCC frameworks span anywhere from 7 to 9 domains of experience [15] and are made up of challenging concepts to operationalize, such as “humanization” and “dignity” [5]. PCC domains could either extensively overlap or be at odds with each other depending on the legal, clinical, or cultural contexts [5]. Finally, there is little to no guidance as to how a given PCC objective, or combination of objectives, might plausibly impact outcomes.

The current complexity of concepts, contexts, and impact when designing PCC interventions could be clarified by using a consistent logic, or theoretical rationale, to inform intervention design. Sales et al. (2006) argue that developing a rationale can “provide a foundation for designing and planning strategies for intervention and selecting tools with a better than random probability of success in implementation [16].” We conducted this systematic review (1) to map the PCC objectives of past interventions using a current definition of PCC delivery care and (2) to explore the impact of PCC objectives on PCC and clinical outcomes.

Refinement of instruments and quantification of the maternity care experience are critical steps prior to development of interventions with a better than random chance of impact. In one recent systematic review Nilver et al. found a wide-ranging set of 36 existing instruments that measure the maternity care experience [17]. Sando et al. found extensive heterogeneity in the sampling techniques, eligibility criteria, and operational definitions of mistreatment during childbirth and advised caution in interpreting prevalence measures of mistreatment [5]. Adding to the complexity, existing studies span countries and regions, and there may be cultural differences in how women define positive person-centered experiences, not to mention the diversity of both biomedical and midwifery models in health systems around the world [18]. Similar to these reviews, we expected to find an expansive literature on the subject of person-centered care in birth facilities. Our paper uses a review method that is particularly suited to synthesizing disparate literatures and is a timely contribution to those seeking to design impactful interventions to improve the quality of maternity care around the world.

Person-centered delivery care in developed settings emerged almost exclusively in response to the impersonal and excessive medicalization of childbirth [2]. Only recently has PCC become an area of inquiry as a potential deterrent to facility-based childbirth [3]. Modern objections to the over-medicalization of childbirth are rooted in events of the early twentieth century when white women in Europe and America began to pursue ‘twilight sleep’ [19]. The woman’s movement of the 1960s and 1970s redefined birth into an event of social and personal significance, ideally controlled by an awake and empowered mother [20‐22]. Around the same time midwifery care re-emerged and became solidified to different extents in the official health systems of Europe and North America [23].

By the early 1990s efforts to reform the experience of maternity care and to shore up the profession of midwifery in developed nations were fully underway. A direct line between current definitions of PCC delivery care can be drawn to the United Kingdom’s National Health Service’s 1993 Changing Childbirth expert report. The report included both a strategy for midwifery care and a “Patient’s Charter” that laid out the rights of maternity patients [24]. Meanwhile, in the international context the 1994 International Conference on Population and Development in Cairo became a turning point for a rights-based approach to sexual and reproductive health [25]. Twenty years later a synthesis of international reproductive rights declarations produced the first publications on “respectful maternity care” (RMC) [3, 26]. The RMC frameworks were soon followed by statements from all the major international health organizations denouncing the mistreatment of women in childbirth [4]. The WHO, leveraging its power as a norm-setting organization, then published a framework to establish the experience of care as a pillar of quality maternity care [27].

Now with close to 30 years of discussion around person-centered care, several overlapping strains of PCC exist, but with gaps between the different approaches. We described one strain above that aligns with the re-emergence of midwifery care and includes different approaches to the provision of care (continuity midwifery models, centering in pregnancy, doula-supported childbirth). More recently, the Lancet Midwifery series examined the contributions of midwives in the global context, including resource-poor settings which have not been exposed to excessive medicalization [28]. A second strain of PCC in the global literature has emerged relatively separate from midwifery, namely the framework around mistreatment of pregnant women, as advanced by Bowser and Hill [3]. Bohren et al. revised the Bowser and Hill typology to develop the most comprehensive set of PCC categories in the maternity context [15]. However, between the midwifery approach and the mistreatment typologies there existed a persistent gap as to how these two areas might be inter-related.

Thus, a broader approach was necessary if the model of care and mistreatment categories were going to be useful in a range of resource settings and in health systems with different proportions of technological obstetric care and primary midwifery care. Given this tension between local contexts and universal frameworks for PCC and for models of care, Sudhinaraset et al. [29] conducted a trans-disciplinary review to create the most comprehensive and adaptable PCC framework to date, which they call The Person-Centered Framework for Reproductive Health Equity. In this framework Sudhinaraset et al. link the provision of care to the experience of person-centered care, using a similar typology to Bohren et al. However, they go beyond the typology approach and link the provision of care to PCC. The provision of care encompasses evidence-based care, both the over and underuse of technology, information and referral systems, infrastructure, human resources, and the medical supply chain. Sudhinaraset and colleagues also situate PCC within the context of a community’s experiences with care, as a community’s specific history with discrimination can determine care-seeking behaviors. Ongoing experiences with mistreatment in the facility can in turn influence a community’s care-seeking behaviors. Finally, the authors link the facility and care- seeking behaviors to societal and community determinants of health equity, including gender and violence norms [30, 31]. In this review, we chose to use Sudhinaraset and colleagues’ framework as we see theirs as the most comprehensive and flexible PCC framework. Importantly, for a global review of PCC delivery interventions, the Sudhinaraset et al. framework points to the ways in which PCC components need to be contextualized within specific health systems, gender and violence norms, and community behaviors [29]. Finally, this systematic review protocol and the PCC Framework for Reproductive Health Equity were developed contemporaneously within a cooperating research group.

In order to be included, an article had to: (1) contain original data (quantitative or qualitative), (2) consist of an evaluation, (3) have at least one PCC objective designed into the intervention and (4) be facility-based. We defined quantitative data as using inferential statistics and qualitative data as primary narratives from participants. We defined an “evaluation” as any quantitative study that utilized a control group (experimental, quasi-experimental). Quasi-experimental quantitative studies needed to collect longitudinal and/or cross-sectional data from treatment and comparison groups. A qualitative evaluation had to be associated with a new person-centered delivery intervention, but we did not require a control group. We defined an objective as the primary goal that the intervention sought to impact. We defined “person-centered” objectives using a current definition grounded in the literature, encompassing: dignity, autonomy, privacy/confidentiality, communication, social support, supportive care, trust, and the health facility environment (See Additional file 1: Table S1 for definitions) [29]. We defined “facility-based” as having some linkage to a hospital or birth center and, required that outcomes be measured at the level of the facility. We defined person-centered outcomes according to the same criteria as the objectives. We defined clinical outcomes to include: labor and delivery, perinatal, and maternal mental health. If a PCC intervention was conducted in the prenatal setting, the measured outcomes had to cross over into the delivery setting.

We excluded quantitative studies that lacked a valid control or comparison group, exclusively examined prenatal outcomes (e.g., ambulatory prenatal diabetes care) or postpartum outcomes (e.g., breastfeeding). We recorded the number of excluded studies and the reason for exclusion at each stage.

We designed a search strategy to maximize the number of primary sources. We searched the English language literature from 1990 until 2016. We conducted a first search in October 2015 and a second search in April 2016 in order to identify any new publications. We systematically searched peer-reviewed literature in PubMed, CINAHL, EconLit, and EMBASE using controlled search terms and free-text terms combining three main components: (a) pregnancy and delivery care (b) person-centered care and (c) interventions (See Additional file 2: Search Strategy). The final keyword chain from April 2016 differed slightly in that we added terms for group prenatal care and birth plans. Otherwise, the 2016 search was only adjusted for the date of publication. The same keywords were used in CINAHL, EconLit, and EMBASE according to their respective search engine requirements.

We hand searched the Cochrane database for all studies related to maternity care. We extensively searched the gray literature, including reports from relevant governmental and non-governmental organizations’ websites by using Google Scholar keyword searches (See Additional file 2: Search Strategy). We searched dissertations and theses in the ProQuest database. Finally, we used bibliographic back referencing to identify additional studies not captured by any of the above searches. We maintained a search diary describing the search methods, keywords used, and search results.

We excluded duplicate references. Next, we independently reviewed titles, abstracts, and executive summaries; we excluded all references that were clearly not relevant. Following this, two team members independently applied the pre-specified inclusion/exclusion criteria to the remaining abstracts. Disagreements regarding the inclusion status of any article were presented to a third team member for a final decision. When the abstract did not contain sufficient information for inclusion, the full text was retrieved.

We first categorized interventions according to their PCC objectives. All interventions were assigned a primary PCC objective, such as “autonomy” or “supportive care,” through a consensus process. An intervention frequently was assigned multiple secondary PCC objectives. Then, undergoing a data reduction process, interventions that shared a primary objective were grouped together and then sub-categorized into conceptual groupings. Following Sudhinaraset et al.’s framework, we formed the conceptual groupings into established models of care.

We categorized outcomes using the same categories as the PCC objectives. We summarized overarching themes and directions of PCC and clinical outcomes. We conducted a thematic analysis of qualitative data [35], a process that included reading repeatedly to extract concepts, categories, and metaphors used to describe or interpret PCC as experienced by the women interviewed. We integrated the quantitative and qualitative data into a final summary. We restricted our search terms by intervention type, rather than outcomes and thus included studies that measured a range from outcomes, from clinical maternal and perinatal outcomes to satisfaction and mental health. Given the diversity of quantitative outcomes, we are not able to combine results to make statements about effect sizes.

Two researchers independently assessed the risk of selection, confounding, performance, and reporting bias; we coded the studies as low, medium, or high risk for each of the four types of bias. Qualitative data were independently appraised by two researchers using the 9-item Critical Appraisal Skills Programme Qualitative Research Checklist [36]. Risk of bias and quality assessment summaries can be found in the Additional file 3: Figure S1 and Table S2.

The initial database searches yielded 11,409 articles with N = 9378 after duplicates were removed. After title-abstract screening was performed, N = 947 remained, and after full text reviews N = 100 potentially eligible studies were found. Of the final included studies (N = 47), 34 resulted from database searches and were supplemented by hand-searches (N = 7), bibliographic back referencing (n = 5), and theses and dissertations (n = 1). The review process and descriptive characteristics of the studies are reviewed in Fig. 1 and Table 1.

We categorized interventions (both quantitative and qualitative findings) into five primary PCC objectives: autonomy (N = 20), supportive care (N = 17), social support (N = 11), the health facility environment (N = 2), and dignity (N = 1). We did not find any intervention designed with these primary objectives: trust, privacy/confidentiality, or communication. Nonetheless, these three domains were addressed as secondary objectives. Below brief summaries of the results are organized from most common primary PCC objective to least common (for definitions of PCC objectives/outcomes and more detailed descriptions of the interventions, see Additional file 1: Table S1). In Table 2 we present detailed information about each quantitative study, PCC domain, intervention type and description, and the effect size of the outcomes; Table 3 presents the comparable information for the qualitative studies. In Tables 5, 6, 7 and 8 we briefly summarize the quantitative interventions and in Table 9 the qualitative interventions. Each table (Tables 5, 6, 7, 8 and 9) represents a single primary PCC objective; most tables contain a sub-grouping of interventions according to a model of care (for instance, “continuity midwifery” under the PCC objective of “autonomy”). Each intervention has two rows: the first row contains the PCC objectives (marked with an “X”); the second row contains PCC and clinical outcomes. The key below explains how to interpret the direction of the outcomes.

There are several limitations to this systematic review. First, the nature of our search terms and exclusion strategy lead to many papers from low-resource settings being excluded and thus the majority of papers are from high-resource settings. While there is a longer history of interventions on person-centered care from the developed world and subsequently more literature, it is unclear if these findings are as relevant to a developing world setting. As mentioned above, women in different settings may desire different aspects of person-centeredness, so interventions in one setting may not be relevant in another. Relatedly, while our definition of PCC was based on findings from around the globe, our domains might not appropriately represent all women’s expectations. Furthermore, we limited our search to interventions in facilities and to outcomes measured post-delivery. There are many exciting person-centered interventions that were exclusively based in the community, prenatal or post-natal settings. These diverse PCC interventions deserve attention and critical review. Finally, we found that the qualitative literature was very narrow based on our search terms; we recommend that a future mixed-methods systematic review use broader search terms, especially to identify relevant studies that were not directly attached to an intervention.