Effects of a Midwife-Coordinated Maternity Care Intervention (ChroPreg) vs. Standard Care in Pregnant Women with Chronic Medical Conditions: Results from a Randomized Controlled Trial

The proportion of childbearing women with pre-existing chronic medical conditions (CMC) is rising. In a randomized controlled trial, we aimed to evaluate the effects of a midwife-coordinated maternity care intervention (ChroPreg) in pregnant women with CMC. The intervention consisted of three main components: (1) Midwife-coordinated and individualized care, (2) Additional ante-and postpartum consultations, and (3) Specialized known midwives. The primary outcome was the total length of hospital stay (LOS). Secondary outcomes were patient-reported outcomes measuring psychological well-being and satisfaction with maternity care, health utilization, and maternal and infant outcomes. A total of 362 women were randomized to the ChroPreg intervention (n = 131) or Standard Care (n = 131). No differences in LOS were found between groups (median 3.0 days, ChroPreg group 0.1% lower LOS, 95% CI −7.8 to 7%, p = 0.97). Women in the ChroPreg group reported being more satisfied with maternity care measured by the Pregnancy and Childbirth Questionnaire (PCQ) compared with the Standard Care group (mean PCQ 104.5 vs. 98.2, mean difference 6.3, 95% CI 3.0–10.0, p < 0.0001). In conclusion, the ChroPreg intervention did not reduce LOS. However, women in the ChroPreg group were more satisfied with maternity care.


Introduction
Worldwide the population of pregnant women living with chronic medical conditions (CMC) is increasing [1][2][3], and in Denmark, the current prevalence is estimated to be 16-21% of all childbearing women [2,4]. The reason for this rise in CMCs among pregnant women is multifactorial [1,2]. Still, an enhanced registration of diagnoses and awareness of specific conditions, better treatment options (including medications), fertility treatment, Int. J. Environ. Res. Public Health 2021, 18, 7875 3 of 16 by phone after the 12th pregnancy week by a research midwife for in-depth information. If the women agreed to participate, a written consent form was signed, and a baseline questionnaire was filled in before randomization.

Randomization and Blinding
Randomization was performed in a 1:1 ratio between the ChroPreg intervention and Standard Care. The randomly generated allocation sequence was computer-generated with concealed varying block sizes. Allocation to study group was centrally administered by the online clinical trial management and randomization software EasyTrial (Easytrial.net, Aalborg, Denmark).

Standard Care
In Denmark, maternity care is tax-financed and free of charge, and most women give birth at public hospitals (97%) [29]. Antenatal care is delivered in a primary care-based collaboration between hospital-based midwives and the general practitioner. If pregnancyrelated complications arise, women are referred to hospital-based obstetricians for clinical evaluation. In Denmark, midwives are authorized to independently provide care for lowrisk pregnant and laboring women [30]. Routinely, women with CMC receive collaborative care provided by obstetricians, medical specialists, and midwives in a tertiary hospital [12].
Women in the Standard Care group were referred to an obstetrician. An individual care plan was designed dependent on the character and severity of the respective CMC and the pregnancy-associated risks. Participants were also referred to a midwife for standard care antenatal consultations scheduled around weeks 14-18, 28, 35, 38, and 40. Consultations include physical examination, discussion about lifestyle and physical and mental wellbeing, symptoms of normal and complicated pregnancies, and breastfeeding-related issues. Antenatal classes consisted of two group sessions in an auditorium (90 min) delivered by midwives. Intrapartum care was provided by the midwives on call on the labor ward. Postnatal care was delivered by midwives, nurses, and obstetricians on the maternity ward. According to individual needs, postpartum consultations could be scheduled after hospital discharge at the woman's request. This could be a debriefing session with a midwife, or, for women who had experienced severe or unexpected complications, a postnatal consultation with an obstetrician was offered.

ChroPreg Intervention in Addition to Standard Care
The design of the ChroPreg intervention was based on a literature review of maternity care interventions aiming to improve the quality of maternity care for women with high-risk pregnancies and aiming to fit the individual care needs of women with CMC [19,[21][22][23]31]. The three main components of the ChroPreg intervention were: (1) Midwife-coordinated and individualized care, (2) additional ante-and postpartum consultations, and (3) specialized known midwives (Figure 1). What differentiated the ChroPreg intervention from Standard Care was that the specialized midwives, who undertook all antenatal and postpartum midwife consultations, had the role of care coordinators between all health care providers involved in providing maternity care for the women. During each visit, the specialized midwife followed up on appointments with, e.g., obstetricians and other medical specialists to help the woman understand her care plan and assist her in integrating information given during all consultations to assure that any questions or uncertainties would be addressed. When necessary, the midwife would coordinate additional communication or consultations with involved care providers. In addition to the routine visits, two additional (1-h long) visits were scheduled antenatally. Postpartum follow-up in the weeks after birth and a face-to-face postpartum debriefing session to facilitate evaluation and processing of the childbearing and birth experience were also planned [32]. To ensure flexibility and individualization of care, unlimited access to e-mail consultations and weekly telephone hours were available (Supplementary Materials Table S1 and the study protocol [25]). weeks after birth and a face-to-face postpartum debriefing session to facilitate evaluatio and processing of the childbearing and birth experience were also planned [32]. To ensu flexibility and individualization of care, unlimited access to e-mail consultations an weekly telephone hours were available (Supplementary Materials Table S1 and the stud protocol [25]).

Primary Outcome
The primary outcome was LOS, defined as the total number of whole days of hosp talization from study inclusion until two weeks after birth. Data for this outcome we collected from electronic medical records. LOS is widely used as an objective health uti zation outcome in health care research [33,34]. In this study, a potential reduction in LO encompassed two critical issues: the patient perspective with a reduced need for hospita ization and the health economic aspect with a possible reduction of costs [34].

Secondary Outcomes
All patient-reported outcomes were collected through self-administered electron questionnaires during the third trimester (33-37 weeks) and two months after birth. Ps chological well-being was assessed by the World Health Organization Well-being Inde (WHO-5) [35], the Edinburgh Postnatal Depression Scale (EPDS) [36], the Cambridg Worry Scale (CWS) [37], and the Short Form-12 Health Survey (SF-12) [38]. Satisfactio with maternity care was measured by The Pregnancy and Childbirth Questionnaire (PCQ [39], consisting of a total scale score and two subscales measuring satisfaction with mate nity care during pregnancy and delivery, respectively. The Pregnancy sub-scale is furth divided into two domains; Personal treatment, and Information and education. The D livery sub-scale measures the Personal treatment domain during delivery.
Secondary healthcare utilization outcomes and maternal and infant outcomes we collected from electronic medical records and described in detail in the study protoc [25].

Statistical Analysis
Data are presented by median values with Inter Quartile Ranges (IQR) and coun with percentages as appropriate. The power calculation was described in the study pr tocol [25]. Expecting a drop-out rate of 5% and allocating 129 women to each group, th power was 80%, in order to detect a reduction in LOS of 25%. LOS in the two groups w  The primary outcome was LOS, defined as the total number of whole days of hospitalization from study inclusion until two weeks after birth. Data for this outcome were collected from electronic medical records. LOS is widely used as an objective health utilization outcome in health care research [33,34]. In this study, a potential reduction in LOS encompassed two critical issues: the patient perspective with a reduced need for hospitalization and the health economic aspect with a possible reduction of costs [34].

Secondary Outcomes
All patient-reported outcomes were collected through self-administered electronic questionnaires during the third trimester (33-37 weeks) and two months after birth. Psychological well-being was assessed by the World Health Organization Well-being Index (WHO-5) [35], the Edinburgh Postnatal Depression Scale (EPDS) [36], the Cambridge Worry Scale (CWS) [37], and the Short Form-12 Health Survey (SF-12) [38]. Satisfaction with maternity care was measured by The Pregnancy and Childbirth Questionnaire (PCQ) [39], consisting of a total scale score and two subscales measuring satisfaction with maternity care during pregnancy and delivery, respectively. The Pregnancy sub-scale is further divided into two domains; Personal treatment, and Information and education. The Delivery sub-scale measures the Personal treatment domain during delivery.
Secondary healthcare utilization outcomes and maternal and infant outcomes were collected from electronic medical records and described in detail in the study protocol [25].

Statistical Analysis
Data are presented by median values with Inter Quartile Ranges (IQR) and counts with percentages as appropriate. The power calculation was described in the study protocol [25]. Expecting a drop-out rate of 5% and allocating 129 women to each group, the power was 80%, in order to detect a reduction in LOS of 25%. LOS in the two groups was compared using a t-test on the log scale to allow for a right-skewed distribution. As prespecified, we also performed the comparison adjusted for age and parity. Further, a chi-square test was used to compare the proportion of women with LOS ≤ 3 days in each group.
The secondary patient-reported outcomes (WHO-5, EPDS, CWS, SF-12) were compared at 33-37 weeks of gestation and two months postpartum. The comparisons were performed using a covariance pattern model, including the baseline measurement as an outcome. The models included the interaction between allocation group and time (baseline/33-37 weeks/two months postpartum) with the constraint that the means at baseline are equal due to randomization [40]. An unstructured covariance pattern was used to model the correlation between the three measurements for each woman. The mean PCQ and CWS scores were compared between the allocation groups using the Welch t-test. Maternal and infant outcomes and secondary health utilization outcomes were compared using the Chi-square, Fisher, or Wilcoxon test as appropriate. The analysis of the mode of delivery was supplemented by a logistic regression analysis adjusting for parity. Calculations were performed using R version 4.0.2.

Changes Due to COVID-19
In response to the global Covid-19 Pandemic [41], from 13 March 2019-8 June 2020, a national lock-down was issued in Denmark [42], affecting certain areas of maternity care, including the ChroPreg intervention. All ChroPreg consultations were retained and converted from face-to-face to telephone consultations. At routine antenatal visits, women were not allowed to bring their partners. During birth and postpartum, women could bring only one person for support. These changes affected both groups equally.

Study Population
From 12 October 2018-22 January 2020, all women referred for antenatal care at Rigshospitalet were screened for eligibility (n = 6608). Of these, 357 women were invited to participate. After verbal and written information, 63 women declined to participate. Between invitation and randomization, 32 women were excluded. In total, 262 women were randomized to either the ChroPreg intervention (n = 131) or Standard Care (n = 131). In each group, one woman withdrew consent after randomization. Moreover, the screening process had failed to identify that two women had pre-existing type 2 diabetes and they were therefore excluded. Thus, 258 women were included in the Intention-to-treat (ITT) analysis ( Figure 2).
The study population was primarily well-educated, employed, non-smoking, cohabiting, with a normal BMI and a planned pregnancy. In the ChroPreg group, 50.8% were nulliparous compared with 59.4% in the control group. A total of 18.5% of all had conceived by assisted reproductive technologies (ART). Most women had one CMC, and the most prevalent types were endocrinological conditions, neurological conditions, and rheumatological conditions (Table 1). Women in the ChroPreg intervention had a median of six antenatal consultations with their midwife, 90% participated in the postpartum telephone follow-up, and 85% in the debriefing session. The optional intervention components of weekly telephone hours and e-mail consultation were utilized by 27% and 33% of the ChroPreg participants, respectively ( Table 2). Table 2. Adherence to the ChroPreg intervention (n = 130).

Primary Outcome
We found no differences in LOS between the two allocation groups. The median LOS during the study period was three days in both groups (Table 3). Women in the ChroPreg group had 0.1% lower LOS than women in the Standard Care group (95% CI −7.8 to 7%, p = 0.97). When adjusting for age and parity, women in the ChroPreg group had 1.8% higher LOS (95%CI −5.0-9.1%) than the Standard Care group. In the ChroPreg group, 84/130 (65%) women had LOS ≤ 3 days compared with 75/128 (59%) in the Standard Care group (p = 0.32).  The study population was primarily well-educated, employed, non-smoking, cohabiting, with a normal BMI and a planned pregnancy. In the ChroPreg group, 50.8% were nulliparous compared with 59.4% in the control group. A total of 18.5% of all had conceived by assisted reproductive technologies (ART). Most women had one CMC, and the most prevalent types were endocrinological conditions, neurological conditions, and rheumatological conditions (Table 1).

Secondary Outcomes
The overall response rate to the online questionnaires was 96.5% in the 3rd trimester and 94.6% two months postpartum. We found no differences in any patient-reported outcomes measuring psychological well-being or health-related quality of life (Table 4). An increased level of satisfaction with maternity care measured by the PCQ was found in the ChroPreg group compared with the standard group; mean total PCQ score 104.5 vs. 98.2 (MD difference 6.3 points 95% CI 3.0-10.0, p < 0.0001). A higher level of satisfaction was also found concerning the two domains of the pregnancy subscale; women in the ChroPreg group scored 3.3 points higher in Personal treatment (48.8 points vs. 45.5, mean diff. 3.3, 95% CI 2.0-5.0, p < 0.0001) and 2.0 points higher in Information/education (25.8 points vs. 23.8 points, CI 1.0-3.0, p = 0.003). We found no difference in the delivery subscale. Table 4. Patient-reported outcomes for the intention-to-treat population according to group allocation. All analyses are based on the constrained covariance pattern model.   A non-significant difference was found in the mode of birth, where 98 women in the ChroPreg group had a vaginal birth compared with 84 women in the Standard Care group (75% vs. 66%, p = 0.10). Due to a higher proportion of multipara in the ChroPreg group, an adjusted analysis was performed, and parity could not explain this tendency (p = 0.11).

ChroPreg
Women in the ChroPreg group had a total median of 11 (IQR 9-13) planned consultations with midwives and obstetric doctors vs. 9 (IQR 8-12) in the Standard Care group (Table 6). Women in the ChroPreg group had a median of 4 (IQR 3-5) planned telephone consultations vs. 2 (IQR 1-4) in the Standard Care group. No differences between the groups were detected regarding unscheduled consultations. During the study period, two women had a late termination of pregnancy (week 22-24) on fetal indication, and one woman experienced a stillbirth (week 30). These cases were found to be unrelated to the intervention.

Main Results
In this randomized controlled trial, we found no evidence to support that a midwiferycoordinated maternity care intervention could reduce the total LOS for childbearing women with CMC. Women in the ChroPreg group were, however, more satisfied with the maternity care they had received. Women in the ChroPreg group were especially satisfied with the care received during pregnancy in relation to the two domains, Personal treatment and Information, and education. The intervention was found to be safe, and adherence was high.

Interpretation of Results
Worldwide, the average LOS during pregnancy and childbirth has gradually declined [43]. In 2001 the average LOS for all Danish pregnant women was 4.0 days, and in 2018 it had decreased to 2.7 days [44]. Women with CMC are, however, expected to have longer LOS [45,46]. We based our sample size calculation on local data with an average LOS of 3.9 days in the target population. To show a significant reduction in LOS, we considered one day of reduction in LOS (25%) to be clinically significant [25]. During the study period, women in both groups reduced the median LOS by 0.9 days compared to the data initially used to calculate the sample size. Therefore, it may not have been possible or even advisable to reduce LOS further, as a further reduction might potentially negatively impact neonatal readmissions [33], breastfeeding rates, and patient experience [47]. A substantial part of the explanation for the overall reduction of LOS may be ascribed to recent local organizational changes at the study site (December 2019), where all women with uncomplicated deliveries were routinely discharged shortly after birth for outpatient postpartum care [48]. This included 29% of the women in this trial (data not shown) who previously would have been offered a routine postpartum stay.
In line with other studies on midwife-led maternity care interventions [18,19,49,50], we found an increased level of satisfaction with maternity care among women who received the ChroPreg intervention. Maternal satisfaction is an important outcome when measuring patient experience [51,52]. It is widely recognized as one of three essential pillars in the quality of healthcare and clinical effectiveness, and patient safety [53]. A review of studies evaluating maternal satisfaction with maternity care found that the primary component of satisfaction was a positive relational experience with the primary healthcare provider [54], allowing for an intimate and trusting relationship in which women felt involved in decisionmaking in antenatal care. This is in line with our findings from the pregnancy domain of the PCQ Personal treatment, which contains items related to the relationship with the midwife and perceived involvement in the decision-making process, rated higher in the intervention group. In the previous studies [18,19,49,50], satisfaction was measured in midwifery continuity of care models, where known midwives provide continuity of care throughout the childbearing experience from antenatal intrapartum and postpartum care [18,19,55] for women at both high and low risk. The ChroPreg intervention included only continuity of care provided by a known specialized midwife throughout all antenatal and postnatal consultations but not intrapartum, which may explain why we found no difference in satisfaction with care during birth. However, in line with our findings, a recent qualitative study from the UK evaluated the implementation of midwifery continuity of care for women of all risks throughout pregnancy and postpartum that did not include intrapartum care and found that women were highly satisfied with the relational continuity, the extra time and flexibility in consultation compared with women receiving Standard Care [56].

Strengths and Limitations
Recruitment of participants for the current trial progressed according to plan, and eligible women with CMC were interested in participating (81%). Adherence to all in-tervention elements was overall high, indicating that the women found the intervention meaningful and relevant.
We had access to complete data on the primary outcome and all other outcomes collected from electronic medical records, ensuring a high internal validity. High response rates were also obtained for self-reported data collected through electronic questionnaires.
We used validated psychometric tools to measure the patient-reported outcomes [57]. The PCQ has solid psychometric properties [39]. As well as measuring overall satisfaction with maternity care, it distinguishes between antenatal and delivery care [58], allowing for in-depth exploration of elements of maternal satisfaction, which has previously been requested in a Cochrane review evaluating the effect of midwife-led models of care [17]. However, a limitation of using the PCQ is that it does not assess women's satisfaction postpartum, which was also an intervention component. Further, the PCQ does not provide cut-off values of satisfaction (high or low levels) but measures satisfaction as a qualitative continuum [39]. Therefore, the results do not suggest that women in the Standard Care group were unsatisfied-the results merely imply that women in the ChroPreg group were more satisfied with the care they received.
Another strength of the intervention is the potential reduction in the need to transfer pregnant women to other care teams. The experience from this trial showed that only one woman in the ChroPreg group (vs. 15 in the Standard Care group) needed a referral to other specialist midwives. This is in line with findings from a previous review of interventions of integrated maternity care [31].
However, some limitations of the study design and methods should be discussed. First, we chose LOS as our primary outcome for this trial. Based on studies in mixed populations and populations with women of high risk that found a reduction of LOS in midwife-led interventions, we wished to assess if an intervention given to a population of women with CMC would reduce overall LOS. We hypothesized that, through improved self-reported psychological well-being and health-related QoL, women would feel empowered and more ready to cope with everyday life in pregnancy and the immediate postpartum period [25]. However, we found no difference in LOS or self-reported psychological well-being or health-related QoL, and our hypothesis was rejected. We may have overlooked the clear association between medical and pregnancy-related problems as the primary causes of hospitalization in the study period, and we do not have any reason to believe that the intervention could change the pathophysiology of any of the CMCs or their interaction with pregnancy.
Secondly, we excluded women with cardiac conditions, diabetes mellitus, substance abuse disorders, and psychiatric conditions as only CMC, because these women had already received maternity care which included some of the elements included in the ChroPreg intervention. Therefore, standard care would differ between these women and women with other CMCs. These conditions are, however, associated with risk of severe adverse maternal and neonatal outcomes [13,59], and therefore the inclusion of these women would have been desirable when evaluating the effect of the intervention on LOS as well as secondary maternal and neonatal outcomes.
We included a wide variety of CMCs. According to national recommendations, to be eligible for the trial the pregnant woman should have at least one CMC that needed tertiary obstetrician-led care [12]. By including CMCs such as asthma and thyroid disease, which are not as strongly associated with adverse maternal and neonatal outcomes as, e.g., hypertensive disorders and epilepsy [7,13], there is a risk that this may have affected the results in the direction of no differences between groups in any secondary maternal and infant outcomes. Third, we were not able to test a full scale midwife continuity of care intervention that also included intrapartum care. At the recruiting hospital it was not an organizational option of care at the time of the trial, but future studies should assess the effect of a midwife continuity of care model for women with CMC.
Generalizing the findings from this trial to other populations of childbearing women with CMC must be done cautiously. It is necessary to consider that the study population was well-educated, averagely weighted, and non-smoking. Therefore, the results may not be generalizable to pregnant women with less advantageous characteristics. Blinding of participants and specialized midwives was not possible and introduced a risk of bias. We can, therefore, not preclude a degree of researcher or social desirability bias. We aimed to reduce this risk of bias by measuring outcomes with validated instruments and introducing an objectively measurable primary endpoint.
For this trial, it was not possible to conduct a full-scale economic evaluation of the cost-effectiveness of the intervention [60,61], as we did not have access to data on total health costs. This analysis might have added further to the complete picture of the effects of the intervention and should be included in future studies. The detected difference of a median of two additional visits with midwives and obstetricians during the study period corresponds to the two additional planned antenatal visits, which were an essential component of the intervention, and therefore expected. However, adding two additional visits to the maternity care model for women with CMC is not without financial costs and may hinder the more widespread implementation of the intervention elements in clinical settings. However, interventions to improve the patient experience of maternity care towards a positive experience of the childbearing process are highlighted by the WHO as in important aim for maternity care and should be considered when building future models of maternity care [51].

Conclusions
Childbearing women with CMC are considered to have high-risk pregnancies, and maternity care for this population requires a high level of flexibility and specialization to ensure optimal outcomes. To our knowledge, this RCT was the first to compare a midwifecoordinated maternity care intervention with Standard Care for women with CMC. We found no difference in the LOS between the groups. We did, however, find that women in the ChroPreg group were overall more satisfied with maternity care. The intervention was feasible, safe, and easy to implement in a clinical setting. Further studies among women with CMC should consider measures of satisfaction with postpartum care and explore the effects of continuity of midwife-led care throughout the childbearing experience, including intrapartum care.