Person-Centred Maternity Care in a Poor-Resource Setting: Evidence from a Cross-Sectional Study in Enugu State, Nigeria

Background Women are not getting adequate person-centred maternity care (PCMC) in low-income and middle-income countries despite being important in reducing maternal morbidity and mortality. This study assessed perceptions and predictors of PCMC among childbearing women in Enugu State, Southeast Nigeria. Methods The study was conducted in two health districts in Enugu State using a descriptive, cross-sectional survey design. We conveniently selected 450 childbearing women, within 9 weeks post-partum, from 11 health facilities purposively selected based on high maternal and child health attendance. Data was collected from women using an interviewer administered 30-item PCMC scale (scores 0–90) consisting of: dignity and respect (6 items, scores 0–18), communication and autonomy (9 items, scores 0–27), and supportive care (15 items, scores 0–45). We categorized full PCMC and each sub-scale into “low, medium and high” using 25th and 75th percentile of the summative scores. The questionnaire also collected data on women characteristics, facility characteristics and service type. Data were analysed using descriptive statistics, t- tests, analysis of variance and Generalized Linear Models. Results Women had medium scores on full PCMC scale and sub-scales. The lowest score was in communication and autonomy. A quarter of women perceived PCMC as high. Marrying at age 20– 29 years (β = 5.83, ρ = 0.002), self-employed women (β = -10.23, ρ = 0.004), starting antenatal care in the third trimester (β = -7.93, ρ = 0.008), high participation in household decisions (β = -4.98, ρ = 0.011), domestic violence experience (β = 5.41, ρ = 0.002), delivery at health centre (β = 7.09, ρ = 0.000), delivery at private/mission hospital (β = 17.47, ρ = 0.000), delivery by non-skilled attendant (β = -9.61, ρ = 0.001); delivery by community health workers (β = -7.65, ρ = 0.001), and experience of pregnancy complication (β

Women who delivered in health centres reported higher PCMC than those who were delivered in public hospitals in Kenya and India, but no signi cant difference was observed in Ghana [8,11]. Whereas women who delivered in private health facilities reported higher PCMC than those who were delivered in public hospitals in Kenya [8,11], no signi cant difference was observed in Ghana [11]. Disrespect and abuse were more likely to be reported by women delivering at hospitals than health centres in Ethiopia [18]. In Pakistan, the risk of reporting disrespect and abuse was twice in public health facilities as compared to private [23]. Mothers who gave birth in primary level facilities have a tendency to be more satis ed than those who gave birth in hospitals [29]. In this study, we hypothesize that PCMC will be lower in public hospitals than in health centres and private hospitals.
Evidence that type of services such as non-emergency or emergency care and experience of pregnancy complications in uence PCMC are mixed. In Pakistan, the type of delivery service did not in uence women's experiences of disrespect and abuse [23]. In contrast, women who experienced pregnancy complications reported higher PCMC than those who did not experience complications in Ghana, but not in Kenya and India [11]. However, severe pregnancy complication signi cantly predicted higher PCMC [8].
Also, In Gambia, women who had a normal vaginal delivery were more likely to have a higher perception of autonomy and supportive care than those who had instrumental delivery [30]. Conversely, higher incidents of disrespectful and abusive care were reported from women who had and who had a delivery with complications, longer labour durations and Caesarean birth [19,22,31,32]. Our proposition in this study is that women with pregnancy complications will have signi cantly lower PCMC than those without complications.
In Nigeria, increase in skilled birth attendance has not been matched with enough quality of maternity care. A systematic review of studies on respectful maternity care in Nigeria indicate that mistreatment of women during childbirth are common and not only undermine utilization of health facilities for delivery but also create psychological distance between women and health providers [16]. Yet, disrespect and abuse of women at childbirth has not been comprehensively studied in the Nigerian health system [12,16]. Equally, the broad concept of PCMC is still nascent in Nigeria, and to our knowledge, there is no study that has investigated the perception and predictors of PCMC in Nigeria. There is a need for evidence that extends studies on respectful maternity care to include effective communication and autonomy, and supportive care during labour and childbirth using a validated PCMC scale [1,6,7]. This study, therefore, assessed perceptions and predictors of PCMC among childbearing women in Enugu State, South-east Nigeria. As patient experience is sensitive to differences in quality care across different providers, institutions and time (5), evidence from this study, on what works for PCMC and what does not, will be helpful to health decision-makers, providers and service users in identifying gaps, designing interventions to promote positive childbirth experiences and evaluating changes in quality of maternity care.

Study setting
The study took place in two districts of Enugu State, South-east Nigeria. We purposively chose two health districts: Enugu metropolis, the capital city, and Isi-Uzo district following our previous maternal and child healthcare study [33], and to compare urban with rural settings. The two districts have, each a general hospital and a network of cottage hospitals and primary health facilities. In 2019, the estimated population of Enugu State was about 4.8 million people. Enugu Metropolis and Isi-Uzo had 1,061,256 and 217,952 populations respectively, out of which women of childbearing age constitute 47.2% and 43.1% respectively [34]. About 94.7% of women in Enugu received ante natal care from a skilled provider, while skill birth attendance is about 93% [34]. However, the maternal mortality rate in Enugu is 645/100, 000 live births [35], higher than the national rate of 512/100, 000 live births [34].

Research design
The study adopted a facility-based cross-sectional survey design using an interviewer administered questionnaire.

Study population and sampling strategy
Women of childbearing age (WCBA), aged 15-49 years, who delivered in 9 weeks preceding the study constituted the study population. Using 54.5% prevalence for non-consented care among women during childbirth from a previous study in Enugu [12], 95% con dence limit, allowable error of 0.05 and 10% nonresponse rate, the minimum sample size was calculated to be 419. We, however, sampled 450 eligible WCBA.
In each district, we purposively selected a general hospital and four primary health centres with the highest maternal and child healthcare attendance based on routine health management information system. Additionally, the sample in Enugu metropolis purposively included the state teaching hospital because of its central location which made it very accessible. The sample was equally allocated to the two districts. Eligible WCBA were recruited by convenience at the health facilities as they leave immunisation clinics using healthcare providers as gatekeepers.

Data collection.
Data was collected from eligible WCBA from January to March 2019 using an interviewer administered PCMC scale made up of 30 items measuring three domains of PCMC: dignity and respect (6 items), communication and autonomy (9 items), and supportive care (15 items). The PCMC scale has been validated in similar low-resource context (1,6). In our sample, Kaiser-Meyer-Olkin measure of sampling adequacy was 0.963 (X2 = 16083.11, ρ = 0.000). On exploratory factor analysis with principal component analysis, all 30 items yielded communalities ≥ 0.4, which was deemed adequate [36]. The cumulative variance was 73.3%. Oblimin rotation with Kaiser Normalization showed all 30 items loaded ≥ 0.4 and were retained since a rotated factor loading of 0.32 is considered signi cant [36]. The reliability coe cient of the PCMC scale was 0.964. Each item is on a 4-point response scale-0: "no, never," 1: "yes, a few times," 2: "yes, most of the time," and 3: "yes, all the time." For each respondent, responses from the PCMC scale were summed up into one composite PCMC score with interval-like properties [37]. The possible score on the PCMC scale range from 0 to 90, with a lower score implying poorer PCMC. The range of possible scores on the sub-PCMC scales are: 0-18, 0-27 and 0-45 for respect and dignity, communication and autonomy, and supportive care correspondingly. We categorized full PCMC and each sub-scale into "low, medium and high". Low was de ned as scores in the approximate lower 25th percentile and scores in the top 75th percentile de ned as high [38].
The questionnaire also included information on socio-demographic characteristics such as age, marital status, residence, religion, age at marriage, education, literacy, occupation, partner's education, partner's occupation, and maternal health care-seeking behaviour. Other information collected include facility characteristics (facility type and provider type), service types, household wealth index, women's participation in household decisions, domestic violence tolerance and experience.
Household wealth index was measured using 11 questions on Nigeria equity tool and its accompanying syntax used to create wealth quintiles [39]. Participation in household decision-making was assessed using questions on different household decisions such as respondent's healthcare, visit to family and relatives, large household purchases, how respondent's earning will be used, and how husband/partner's earning will be used [40]. Each question was assigned the following scores: 0 -if the decision was made by husband/partner alone, someone else or other; 1 -if the decision was jointly made by respondent and husband/partner; and 2 -if the respondent alone made the decision. Participation score ranged from 0-10. Also, attitudes towards domestic violence were measured using ve variables describing whether beating was justi ed if the wife: goes out without telling her husband; neglects the children; argues with her husband; refuses sex with her husband; and burns food [40]. Women who answered 'Yes' and 'Don't know' were scored 0 while women who responded 'No' were scored 1. Domestic violence tolerance score ranged from 0-5. The value of either the participation score or domestic violence tolerance was transformed into 0-1 interval [40]. The median values were used to dichotomise the scores into low and high participation as well as domestic violence tolerant and intolerant categories. The questionnaire was pre-tested on women with recent deliveries in a different district to check for ease of understanding. Five trained research assistants administered the questionnaires, while the authors checked questionnaires for completeness daily.

Data analysis.
Data were analyzed using SPSS (version 20, IBM, New York, USA). Characteristics of respondents were presented using frequencies and percentages. Mean PCMC scores and standard deviation were calculated and compared across various socio-demographic characteristics of respondents, facility characteristics and service type using t-tests and analysis of variance (ANOVA). Parametric tests were deemed appropriate since the single composite PCMC scores have interval-like properties. Generalized Linear Models was used to the test relationship between PCMC and the parameters that were signi cant on bivariate analysis. Statistical signi cance was set at alpha 0.05 level.

Ethical consideration
The study was approved by the Health Research Ethics Committee of University of Nigeria Teaching Hospital, Enugu, Nigeria. Written, informed consent was obtained from all respondents.

Characteristics of respondent
The response rate was 100%. Table 1 shows the characteristics of respondents. About 58% of women were in the 20-29 years age group and educated to at least secondary or vocational school. Most women were married, Christians, Igbo, and married in their 20 s. Nearly half of women could read and write very well. About a fth of women were unemployed. Over 50% of women had low participation in household decisions. Whereas 90% of women were intolerant to domestic violence, about 30% had experienced domestic violence. Over 50% of women started antenatal care during or after the second trimester. Few women (10.2%) were delivered by non-skilled provider, while about 8% reported pregnancy complications. Distribution of individual PCMC items among women (N = 450) As shown in Table 2, most women were treated with respect, but 20% were verbally abused, and seven percent were physically abused. Most women reported good visual privacy and record con dentiality. Most women reported that providers introduced themselves, called women by their names, involved women in care decisions, sought consent to procedures, talked to women in language that women understood, explained examinations and medicines (Table 2). However, whereas 79% were able to ask questions, only 13% of women were able to choose their delivery position.
Most women indicated that providers promptly initiated care, paid attention when needed, talked to women about their feeling, took the best care of them and could be trusted (Table 2). Most women also reported that health facilities were safe, clean, had electricity and clean water. Although most women reported that health facilities have few staff, health facilities did not allow labour and childbirth support ( Table 2).
Distribution of full PCMC scale and sub-scales Table 3 shows that about a quarter of women perceived PCMC as high. Nearly two-third of women perceived respect and dignity as high. While less than a third of women perceived communication and autonomy as high, almost a third perceived supportive care as high. As shown in Table 4, women had medium scores on full PCMC scale and sub-scales.  Bivariate analysis Table 5 shows mean score differences in PCMC disaggregated by predictors. Women who married at age ≥ 30 years had signi cantly lower PCMC score than those who married before 30 years (ρ = 0.000). Selfemployed women (ρ = 0.000) and women married to unemployed partners (ρ = 0.027) had signi cantly the least PCMC scores among occupational categories. Women with high participation in household decision-making had signi cantly lower PCMC score than those with low participation (ρ = 0.000).
Women who had experienced domestic violence had signi cantly lower PCMC score than those with no experience (ρ = 0.001). Women who were delivered in health centres had higher PCMC scores than those delivered in public and private hospitals (ρ = 0.000). Women who were delivered by doctors had higher PCMC scores than those delivered by other providers (ρ = 0.039). Women who had pregnancy complications had higher PCMC scores than those without pregnancy complications (ρ = 0.045). Women who started antenatal care during the third trimester had signi cantly lower PCMC score than other women (ρ = 0.000). Predictors of person-centred maternity care Table 6 shows the parameters that predicted perceived PCMC among women in this study.

Discussion
The study found that women had medium scores on full PCMC scale and sub-scales. The least score was communication and autonomy sub-scale, while highest score was respect and dignity dimension.
Our ndings are comparable to evidence from previous studies [1,6,10,11]. Although, most women in this study were treated with respect, as much as 20% were verbally abused, which is slightly higher than ndings of 11-18% in prior studies [6,10,11], but consistent with high incidence of disrespect and abuse in other studies [12][13][14][15][16][17][18][19]. Similarly, our nding that 7% of women were physically abused was higher than results elsewhere [6,10,11,15], but lower than ndings reported in previous studies [12,[16][17][18]. The sharp decrease in physical abuse might re ect underreporting as this behaviour may be accepted as normal and not considered as abuse or disrespect by some women in Nigeria [16], or due to a rising awareness of litigation among care providers [15]. Lower score on communication and autonomy is accounted for mainly by a quarter of women who received limited consented care, explanation of procedure and medicines; and another third who were least involved in their care, not called by name and received inappropriate introduction from providers. Conversely, previous studies found higher prevalence of unconsented care, limited explanation of medicine and procedures, not calling of women by name and inappropriate introduction by providers [6,10,11]. Consistent with other studies [6,10,11], women were greatly concerned about lack of labour support, restrictive childbirth companionship and inadequate sta ng. While many areas of strength exist across the three domains of PCMC, meaningful changes to improve PCMC would involve reducing verbal and physical abuse, improving communication with women, and addressing gaps in facility-level drivers of PCMC.
This study revealed that marriage at 20-29 years had a signi cant positive relationship with women's perception of person-centred maternity care, which is consistent with a higher incident of disrespect and abuse among women aged 20-34 years [22]. The observed relation between age at marriage and PCMC might not simply re ect age, but also economic and educational empowerment given that nearly 60% of women in our sample belong to rich quintiles and majority of women have a minimum of secondary education. In Nigeria, women with no education marry six years earlier than women with secondary education, whereas women in the lowest wealth quintile marry more than eight years earlier than women in the highest quintile [34]. We argue that women in this study, who marry at age 20-29 years, are better empowered, more likely to live in areas with better quality of care, have higher expectation of care and can recognise low-quality care and advocate for improved care [21].
We found that self-employment had signi cant, but an inverse relationship with women's perception of person-centred maternity care. An increase in self-employment would result in decrease in PCMC among self-employed women. This nding is comparable with evidence in Kenya which found that employment status predicted women's perception of PCMC [8]. However, while the study in Kenya dichotomized occupation into unemployed and employed, our study used ve occupational categories. Two factors could explain our ndings. First self-employment could enhance women's participation in household decision-making for their own healthcare [41]. Secondly, self-employment increases women's economic empowerment, which means that women can effectively demand better maternity care [21,41]. As perception of PCMC varies with socio-economic status [8,11], an increase in women's labour participation that promotes self-employment is needed to improve person-centred maternity care.
High participation in household decision-making was found to have an inverse relationship with women's perception of PCMC in this study. Women with high participation have signi cantly lower PCMC score than those with low participation in household decisions. Our results contrast ndings of a prior study in Kenya which found that PCMC was not signi cantly related to participation in household decisionmaking [8]. In Nigeria, healthcare decisions for women are mostly made by their husbands/partners without women's involvement [42]. It might be that low women's decision-making autonomy limits women's expectation of quality of maternity care, social power between women and providers, and women's capacity to demand better care in Nigeria. Conversely, women who participate highly in household decisions are better aware of their rights to person-centred care and tend to have increased self-con dence thereby reducing power differential between health providers and women [16].
This study further revealed that the experience of domestic violence was inversely related to women's perception of person-centred maternity care. Women who had no domestic violence experience had signi cantly higher PCMC score than those who experienced domestic violence. Our ndings compare to a similar association of domestic violence experience and PCMC in Kenya [8]. These ndings are expected because women who experience gender-based violence are disempowered and more vulnerable to dominance by providers who use coercive strategies to control women or punish women for perceived disobedience during childbirth [43]. Women who experience domestic violence are emotionally challenged. Women even when receiving technically sound care but lacking in emotional support perceive it as low-quality care [23,44]. In Nigeria, childbearing women patronise providers with sympathy and compassion regardless of their level of competence [16]. Also, domestic violence limits women's decisionmaking power regarding their reproductive health and have been associated with poor maternal health outcomes [45]. For instance, Sipsma et al found that women's experience of physical abuse is associated with inadequate use of antenatal care [46].
Trimester of commencing antenatal care predicted women's perception of PCMC in this study. Women who commenced antenatal care during the third trimester were more likely to have a lower perception of person-centred maternity care than women who started antenatal care in their rst trimester. Our ndings are inconsistent with a previous Kenyan study showing that the onset of antenatal care is not signi cantly related to the perception of PCMC [8]. Every pregnant woman in developing countries should seek antenatal care during the rst trimester of pregnancy because failure to initiate antenatal care early is a potential risk for complications during pregnancy and childbirth [47]. Tailored group educational activities and peer support motivates behaviour change among pregnant women and increases women's satisfaction with maternity care [47]. In this study, late initiation of antenatal care meant that women are not familiar with the health system and might not have the bene t of psychological support and sharing of experiences which help women feel more empowered as decision makers during childbirth [48].
Moreover, women who were delivered in health centres and private/mission hospitals had higher PCMC scores than those delivered in public hospitals. Similar ndings of higher PCMC were also found in health centres and private hospitals in Kenya [8]. Also, mothers who gave birth in primary level facilities also tended to be more satis ed than those who gave birth in hospitals [29]. In Pakistan, the risk of reporting disrespect and abuse was also higher in public than private hospitals [23]. Our nding also compares to other studies that nd higher interpersonal quality of maternal healthcare in private than public hospitals [25,28]. Conversely, indices of clinical quality of maternal health were higher in public hospitals than private hospitals and health centres [25][26][27]. In this study, higher PCMC scores in health centres and private hospitals may be due to low provider-patient ratio which reduces the strain on provider-patient interaction [8]. Equally, higher PCMC scores in health centres might re ect closer ties between providers and women in closely knitted communities that health centres serve [8] and effect of citizen participation in governance of health centres [49]. In Nigeria, users have better perception of health workers in private facilities because private facilities greatly emphasize interpersonal quality [50].
Type of birth attendant was also found to predict women's perception of PCMC in this study. PCMC was inversely and signi cantly related to delivery by community health workers and non-skilled attendants, although we expected a direct relationship given that negative attitudes and behaviours are commonly ascribed to trained professionals especially doctors and nurses [51]. Although women who were delivered by doctors received higher PCMC than those delivered by nurses, delivery by doctors was not signi cantly predictive. By contrast, PCMC was directly and signi cantly related to delivery by doctors in Kenya [8].
Higher perception of PCMC among women delivered by doctors than nurses is consistent with a Nigerian study showing that healthcare users have a better perception of doctors than nurses [50]. It could be that negative attitudes and behaviours are more common among nurses than doctors as hostile and impersonal behaviour from nurses and midwives are common reasons for dissatisfaction with quality of maternal health services in South-east Nigeria [52].
Furthermore, our study revealed that women who had pregnancy complications had higher PCMC scores than those without pregnancy complications; and experience of pregnancy complication signi cantly predicted perception of PCMC. Comparable results were found in Kenya, where women with severe pregnancy complication reported higher PCMC than other women [8]. By contrast, we expected that women with pregnancy complications will have signi cantly lower PCMC than those without complications. Our expectation is consistent with ndings in previous studies showing that there were higher incidents of disrespect and abuse among women who experience pregnancy complications and longer labour durations requiring instrumental delivery and caesarean birth [19,22,[30][31][32]. It could be that survivors of pregnancy complication are more satis ed with their positive pregnancy outcomes and tend to report exaggerated positive patient experiences.
There is little evidence on the determinants of person-centred maternity care using a validated tool [8].
Hence, this study adds to the nascent scholarship on factors in uencing disparities in patients' experiences of maternity care in low-and middle-income countries. This notwithstanding, this study has some potential limitations. First, the study was limited to two health districts in one Nigerian state. The health facilities from which women were sampled are not representative of the entire state as nonprobability sampling approaches were used. Thus, respondents cannot be generalisable to all women.
Second, recall bias is common in retrospective data collection, though respondents in this study seemed to recall experiences of their pregnancy and childbirth vividly. Notwithstanding that women could recall childbirth experiences accurately within twenty years [53], we adopted 9 weeks post-partum used in a previous study [1]. Third, there may also have been sampling bias as only women who gave birth to live babies and attended immunization clinics were included. The study, therefore, potentially excluded women with stillbirths and neonatal deaths who may have had more negative childbirth experiences.

Conclusion
The purpose of this study was to assess predictors of PCMC in Enugu State, South-east Nigeria. Evidence from this study highlight the role ve patient characteristics (age at marriage, self-employed women, high participation in household decisions, domestic violence experience and initiation of antenatal care in the third trimester); two facility characteristics (facility type and provider type); and service type (pregnancy complication) as predictors of person-centred maternity care. This information should inform the design of interventions to promote positive childbirth experiences and can be used to evaluate changes in the quality of maternity care.