Beyond emotional support: predictors of satisfaction and perceived care quality following the death of a baby during pregnancy

Introduction

Assessing quality from the perspective of patients is a crucial part of delivering effective care that complements measures of health related quality of life and biological outcomes [1, 2]. Measuring patient satisfaction and perceptions of care quality can be effective ways of assessing health subjectivities, particularly as part of mixed-method designs and when placed in social and cultural context [3], [4], [5], [6]. Although considerable work has been done on the measurement of care outcomes in perinatal bereavement, there has been relatively little quantitative research on the underlying components and predictors of satisfaction and what constitutes quality of care from the perspective of women, or their partners. Partly, this relates to the underuse of established techniques and, in parallel, the lack of area-specific instruments to do so. This is an important gap because healthcare following perinatal loss is understood to have an important impact on grief and health outcomes [7]. This article describes the development of a preliminary scale to measure women’s subjective evaluations of care following late pregnancy loss – second and third trimester stillbirth and termination of pregnancy for medical reasons – and the results from its application through an online survey.

To date, most work on care outcomes following stillbirth has focused on the implementation or effectiveness of care interventions and patients’ subjective experiences. Through surveys, patient-reported outcomes have been used to measure the implementation or offer of particular care interventions, which can be divided into two categories: a) clinical and structural aspects of care (e.g. autopsy, mode of birth, accommodation), and b) care related to bereavement and loss (e.g. contact with the baby, memory objects) [8, 9]. Other studies have measured the effectiveness of bereavement care interventions, such as ‘the mother saw/held the baby’, in terms of mental health outcomes or satisfaction [10], [11], [12], [13], [14], [15], [16], [17]. While the experiences of women and their partners in hospitals have been widely assessed through both qualitative and quantitative research [18], [19], [20], only a few studies are based on established constructs of care satisfaction and perceived quality.

The studies that have used such constructs are problematic for a number of reasons. Firstly, the use of single-item measures of satisfaction [13, 21, 22] is an issue because perceptions of care are made up of various sub-components, meaning that the specific aspects of care that patients are assessing remain unknown, and, furthermore, satisfaction tends not to capture the full range of subjective evaluations [23, 24]. This is highly limiting as a guide to practice. Even when multidimensional scales are used, their appropriateness is questionable. For example, the PSQ-18 [25] and MPSQ [15] have been employed to assess perinatal bereavement care, but are designed to measure continuous primary care from doctors. While some studies are based on area specific scales, they are mostly published with little information on their development and validation [10], [11], [12, 26]. Recent work by Wool et al. in the related area of perinatal palliative care provides an interesting contrast [27, 28]. This indicates a twofold problem: lack of knowledge of the underlying components of subjective care quality and an absence of area-specific scales to carry out such measurement.

Studies on the dimensionality of ‘satisfaction’ or ‘perceived quality’ show that patients are principally evaluating functional, technical and physical (comfort, privacy, etc.) components of their subjective experience [5, 24, 29], [30], [31], [32], [33], [34], [35], [36]. On one side, the functional dimension relates to empathy and emotional care/support, including perceptions of being treated respectfully and as individuals. This also includes decision-making and equity in care encounters, which encompasses information provision, feeling listening to and respect for patient preferences. In perinatal bereavement care, emotional support and decision-making are the cornerstones of good practice, but are specifically oriented toward affirmation of identities (mother/father/child) and decisions that parents will most likley never have considered (autopsy, post mortem contact, etc.) [18], [19], [20]. The technical dimension of care relates to the technical competence of health professionals, administrative and structural competence in relation to coordination of care, and perceptions of health outcomes. In perinatal bereavement care, aspects such as continuity and coordination of care have been highlighted as important [18], [19], [20]. However, there is a lack of consideration for possible health outcomes beyond mortality and evaluations of technical aspects of care (task-directed skill related to clinical and loss-focused care), and how these may fit into overall perceptions of care, as well as considering how doctors and midwives work together to provide care.

Developing more detailed knowledge of the meaning of patient satisfaction and subjective quality in perinatal bereavement care could have significant advantages for learning more about patient needs and the ongoing development of care practices. In addition, at a methodological level, it may help to ensure that these constructs are better understood and that measurement is more reliable. This exploratory study aims to investigate: (1) if satisfaction alone is a sufficient measure of care quality in the context of loss; (2) what objective and subjective outcomes predict satisfaction; (3) what components of care are most important to women. The study included the development of a detailed questionnaire, including a custom scale to measure subjective evaluations of various aspects of care, and its application to a large sample in Spain.

Materials and methods

The population was defined as women who had experienced a late miscarriage, stillbirth, intrapartum death, or termination of pregnancy for medical reasons from 16 weeks of gestation and within the Spanish healthcare system. A cross-sectional descriptive study was carried out using an anonymous online survey, the only practical means to reach the population at a national level. The study took place as part of a broader ethnographic and mixed-methods research project that included participant observation and qualitative analysis of narrative style interviews [37].

The anonymous online questionnaire sought to collect five main types of data: (1) socio-demographic; (2) reproductive history and details of the loss/death; (3) objective measures of care focused on care interventions related to: the diagnosis, accommodation, birth and mode of delivery, administration of sedatives, investigation of the cause of death, disposition of the body, contact with the baby, memory objects; (4) intensity of contact with support charities; and (5) subjective evaluations of care through the development of a custom battery called the Pregnancy Loss Care Quality Scale (PLCQ scale), which uses a 5-point Likert agreement scale. As well as the PLCQ, a separate five-item scale was developed to measure specific types of information. Finally, respondents were also asked if they believed there had been medical negligence in their case (response options: No; Yes, a claim was made; Yes, no claim was made).

The questionnaire was developed through a review of the literature, exploratory qualitative interviews with four healthcare professionals and six parents, and content and face validation [32, 38, 39]. For the PLCQ scale development, a panel of 15 expert health professionals were presented with 57 items, broadly related to the care dimensions outlined in the introduction, as well as items related to diagnosis and decision-making on post mortem contact. The panel rated and qualitatively evaluated the items. Following a second round, 29 items were retained in the final questionnaire.

In addition, six global measures or dependent variables (DVs) were included. Firstly, both ‘satisfaction’ and ‘willingness to recommend’ the hospital to other couples were included as standard, yet differentiated, measures of care performance [40]. In the analysis, these two items were first compared and subsequently collapsed into one overall global measure of care (GLB1). As the study was also concerned with patients’ perception of professional competence, two items were added that sought to evaluate doctors and nurses/midwives individually: ‘the doctors were competent in their work’ (GLB2); ‘the nurses/midwives were competent in their work’ (GLB3). At a hypothetical level, it was thought that both ‘competence’ and ‘satisfaction/willingness to recommend’ might not capture evaluations of loss-focused care. This conjecture was based on exploratory interviews and theory on satisfaction measurement which suggests that patients rate the difference between their expectations and performance [29, 41]. As most women, and their partners, are not prepared for this experience or have knowledge of appropriate care, it was posited that they may differentiate between standard expected care and loss-specific aspects of care. Therefore, two further items were introduced: ‘the doctors knew how to deal with cases of loss’ (GLB4); ‘the nurses/midwives knew how to deal with cases of loss’ (GLB5). A mixed panel of 11 health professionals and eight bereaved women validated the draft questionnaire, which was subsequently piloted by 18 women. A full list of the socio-demographic data, reproductive/loss variables, care interventions and the full questionnaire has been published elsewhere [37].

Fieldwork took place between June 2013 and June 2016. Respondents were recruited using convenience sampling (snowballing), including advertisements on support associations’ websites, social media and through direct contact with parents and health professionals. In total, 796 women, whose baby had died within the previous 5 years, completed the survey. However, to give greater stability to the regression analysis, only the 610 cases occurring within 24 months previous to participation, and which had no more than one missing data point, were analysed. Further detailed information on sampling procedures, fieldwork and data purification are available elsewhere [37].

As the analysis was concerned with care practices, variables related to socio-demographics, pregnancy history, details of the loss, and intensity of contact with support associations were not included. It should be noted, however, that exploratory analysis indicated that, of these, only second trimester loss had strong predictive value over satisfaction. For the purpose of the regression analysis, objective variables were converted to dummies (0,1). Prior to performing the regression, Mann-Whitney 2-tailed tests of independence (p<0.05) and Cohen’s effect size [42] were used to test the 51 objective independent variables (IVs) against the DVs and to screen for inclusion in the initial objective regression model. Multiple sequential (forward method) regression was used to assess the ability of objective and subjective IVs to predict the five global measures of care (GLB1, GLB2, GLB3, GLB4, GLB5). The sample size was sufficiently large to meet a recommended IV to DV ratio of 18:1 [43]. In order to reduce contributions from weak or unstable variables, the alpha criteria for inclusion in the regression model was set at p<0.1. Missing data was substituted by mean scores. Analysis of tolerance (IVs<0.1) and inflation (VIF > 10) found that multicolinearity was not present in any of the models and Durbin–Watson scores were all acceptable (≥1.5 and ≤ 2.5). Residual scores were within the acceptable range (3.3–3.3) and plot analysis and Mahalanobois and Cook distance tests indicated no significant issues [43, 44].

Results

Discussion

The models produced by this exploratory study are robust and provide important insights to hospital care, one of the main social spaces for defining the meaning of pregnancy loss and for providing appropriate resources for grieving. The study has implications for both the provision of care and quality assessment. Before discussing the results it should be taken into account that the non-probabilistic online sampling procedure means that the study has coverage limitations that reduce its generalizability. However, some confidence can be gained from the knowledge that the sample is a very close match to national data in terms of gestational age, autopsy rate and caesarean section rate [37, 45], and that exploratory analysis of socio-demographics found only very small predictive values. The results should also be interpreted as situated within a particular set of cultural values and expectations, which may be similar to or differ from other countries. Finally, the limitations of surveys themselves should be considered as they tend to reduce complex patient experiences to simplistic and impersonal dimensions [3, 46]. In this project, the results of the survey are triangulated with qualitative methods, showing good convergence and some divergence [47, 48].

The results make it clear that when women rated ‘satisfaction’ or ‘willingness to recommend’ the hospital they were not evaluating the totality of loss-focused care. This fits with the disconfirmation thesis [23, 41] and suggests a ‘halo’ effect [49] whereby patients assess aspects of care that fall within their frame of reference or expectations. So, while useful, the results suggest that satisfaction tends to provide overly positive evaluations and should be balanced by a measure such as ‘willingness to recommend’ and, much more importantly, a specific item to measure loss-focused care. In addition, single-item global evaluations had limited analytic capacity, making a strong argument for the need to use constructs that capture the broad range of underlying components/dimensions of care following pregnancy loss. The PLCQ scale developed in this study, for example, provided a much better measure of the value of specific care interventions to women.

It is also apparent that when the participants were rating care they were largely, though not exclusively, evaluating the process of care rather than the outcomes themselves. This reinforces the idea that the human aspects of interactions with health professionals are crucial to care and experiences of grief/loss. However, in this study, these interactions were much broader in scope than the expression of empathy and emotional support, encompassing the organisation of care, teamwork, technical competence (including perceptions of fault) and bereavement related interactions. The rating of ‘emotional expression’ over ‘emotional support’, ‘being well informed’ over ‘information’, and the novel inclusion of ‘teamwork between doctors and nurses/midwives’ point to a number of important findings for care practice.

At a general level, the results suggest a transversal dimension related to the ‘atmosphere of the care setting’ and its significance in the unique situation of death/loss. As women commonly experience great sadness, trauma, fear and anxiety in the period after diagnosis [50], [51], [52], it should be considered that the primary goal of many women may be to just get through a threatening situation that seems to exceed coping capacity [53]. On this basis, being ‘well informed’ of the process of care before it happens may help to reduce the anxiety and fear that women often feel. Similarly, the participants seemed to readily detect harmony or tension between health professionals, which appeared to reinforce or undermine confidence in the quality of care. Lack of teamwork may relate to conflict between obstetricians and midwives on what constitutes appropriate care [54, 55]. This finding highlights the extreme importance of collaborative teamwork and a unified approach during encounters with patients. It is also noteworthy that the importance of teamwork was not as evident in the qualitative research [37], showing the benefits of mixed-method approaches. Finally, as the highest ranked predictor, feeling free to ‘express emotions’ in front of health professionals is differentiated from ‘receiving’ emotional support and suggests an atmosphere of care that creates a space for loss and reduces feelings of shame and stigma. Collectively, these three predictors seem to define a dimension of care that traverses functional and technical components and that creates security and confidence in carers. It might be said to represent the basis or groundwork of good care. Each of these predictors also gives a sense of women as astute observers of healthcare provision rather than just receivers of care. This questions the directionality of traditional thinking on care provision, which tends to focus on direct provider actions, for example, when examining trauma in childbirth [56].

The results also demonstrate considerable variations in expectations of the roles of nurses/midwives compared doctors, whose actions had a greater influence over ratings. In this sample, doctors not only held greater clinical authority – e.g. in expectations of the provision of clinical information – but they also seemed to have greater power to define social identity [6]. This was seen in the way that insensitivity with language was more influential in doctors’ models and highlights the challenge to doctors to be empathetic and supportive of the emotional experience of loss, as observed elsewhere [57]. Nurses/midwives, on the other hand, were assigned to subordinate positions associated with traditional caring and support roles. From the perspective of implementing ‘midwife-led’ bereavement care in Spain this is problematic. It indicates a requirement for stronger teamwork between professionals and that doctors validate or reinforce the role of midwives during initial interactions with women and their families.

Finally, the study helps our understanding of how women view and assess technical-professional competence, in particular evaluations of how care is managed and organised and whether fault for the death is attributed to health professionals. This contradicts any notion that patients are incapable of making technical assessments [58, 59]. Given the impact of ‘perceived negligence’, future studies may consider including it as one way of measuring health outcome. On the other hand, the relatively low ranking of ‘information’ shows that other aspects of care are prioritised in the peritraumatic period. It seems relevant here that parallel qualitative research found that low cultural expectations of equity in decision-making and naturalised paternalism are common in obstetric care following pregnancy loss in Spanish hospitals [37].

To conclude, the results indicate that a transversal dimension ‘atmosphere of care’ is the single greatest influence on satisfaction/quality, which probably relates to the unique context of loss, grief and trauma. It is necessary to use specific loss-focused global variables to adequately assess bereavement care and a custom scale to measure the influence of care interventions. Future work will focus on further development and testing of the PLCQ scale.