Women’s characteristics and care outcomes of caseload midwifery care in the Netherlands: a retrospective cohort study

The data used in our study came from the Netherlands Perinatal Registry (Perined). The registry routinely collects and combines data on antenatal, intrapartum and postnatal care from four separate national registries; one for primary midwife-led care (LVR1), one for maternity care by general practitioners (LVR1h), one for obstetrician-led care (LVR2), and one for neonatal care (LNR). The Netherlands Perinatal Registry contains data on approximately 98% of all births in the Netherlands [16]. We selected all women who gave birth in 2015 and received antenatal care in a primary midwifery practice in the Netherlands. Exclusion criteria were birth at a gestational age under 28 weeks, or an antenatal transfer to secondary or tertiary care for pregnancy complications before 28 weeks. In this way we ensured that women experiencing serious medical problems in early pregnancy are not included. The study population, being in primary midwife-led care at 28 weeks of pregnancy, is, by definition, considered to be low risk.

Caseload midwives throughout the Netherlands were identified with a snowball method, starting from the professional network of the first author (PO). All potential participants were contacted and were included if their practice description fit the caseload definition [3], based on a short questionnaire. Midwives working in a team or as a couple were included only if they offered 24/7 one-to-one continuity of care throughout pregnancy and birth. Midwives working according to an agreed schedule or duty roster and who shared their clients with their colleagues – a customary arrangement in regular midwifery practices - were excluded from the caseload group. Participating caseload midwives gave informed consent and voluntary disclosed their LVR1 practice registration number, to enable the researchers to anonymously identify caseload clients in the study population.

Based on personal communication with caseload midwives, we expected their clients to be more often multiparous, older, with a higher social-economic status (SES) and more often of a Dutch background, compared to women in regular care. To reduce confounding and enhance comparability of characteristics in the study population, we matched each woman in caseload care to three women in regular midwife-led care. We used parity (0, 1 or more), maternal age (six categories) and background (Dutch or non-Dutch) as matching variables. Since SES is measured on neighborhood level only in the Perined database, we decided not to use SES as a matching variable. Because serious regional variation in maternity care has been described in the Netherlands [27, 28] we also included the postal code (first two out of four digits) in the matching procedure to minimize confounding based on these regional differences. Exact matching was performed randomly using SPSS for Windows, version 22.

The following maternal characteristics were collected from the database: maternal age; parity; ethnic background (Dutch or non-Dutch); SES and level of urbanisation of the neighbourhood in which women were living were determined using the four digits of the postal code [29]. The characteristics of birth we collected were: planned and actual place of birth, gestational age at birth, birthweight, multiple pregnancy and foetal presentation at birth. To explore whether women in caseload care may have switched to a caseload midwife at a later stage after starting their care in another midwifery practice or a secondary care hospital team [25], we also collected the gestational age at intake in the practice that recorded the birth.

The primary outcomes of interest were antenatal or intrapartum referral to obstetrician-led care and mode of birth (spontaneous vaginal, instrumental vaginal or caesarean). Other outcomes of interest were inductions of labour, maternal morbidity (blood loss and perineal trauma) and perinatal outcomes (perinatal mortality and Apgar score). In a sub-analysis among low risk women who were in midwife-led care at the start of labour, we also observed interventions during childbirth (augmentation of labour, pharmaceutical pain relief and episiotomy).

All analyses of outcomes are presented separately for nulliparous and multiparous women. In the main analysis, we used inferential statistics (Chi square; two-sided t-test) to compare the outcomes for women in caseload care and regular care. To capture the differences between intrapartum care in caseload and regular practices, we performed an exploratory sub-analysis among low risk women who started labour in midwife-led care. For this sub-analysis we excluded all women who experienced an antenatal referral after 28 weeks and women with the following risk factors: gestational age < 37 or > 42 weeks, non-vertex presentation, and multiple pregnancy. Given these exclusions, we could no longer use our matching procedure; for this reason we used only descriptive statistics (percentages) to illustrate the differences between the two groups.

The national database contained the records of 127,818 women in midwife-led care at a gestational age of 28 weeks and beyond. Of these women, we included all 657 women that were registered by caseload midwives in our study. The matched cohort contained 1954 women in regular midwifery care. If the matching procedure had been fully successful, the matched cohort would have contained 1971 controls. This means we missed 17 controls (0,9%). All cases had at least one exact match.

Women in caseload care were slightly older (mean 31.6 vs 30.4 year) when compared to women in the total national cohort, they more often had a Dutch background (79.5% vs 75.6%), and were somewhat more often nulliparous (48.6% vs 45.8%). These and other characteristics are shown in Table 1.

After matching for parity, age and background, women in caseload care showed some differences in other demographic characteristics with the women in regular care. They more often lived in a neighbourhood with a higher SES (24.8% vs 19.6%), more often in an urban neighbourhood (29.5% vs 18.4%) and less often in a rural neighbourhood (14.2% vs 23.9%). There were no significant differences in mean birthweight, birthweight categories, small for gestational age (<percentile 5) or large for gestational age (>percentile 95).

Other characteristics seem to confirm that women in caseload care are a distinct group. They more often opted for a home birth (44.5%% vs 20.3%), and more often had a late intake in the practice (11.4% vs 2.1% at ≥28 weeks), likely an indication of a switch in care provider during antenatal care. Gestational age at birth was slightly higher (39.5 vs 39.3 weeks; p-value 0.004), and more women in caseload care gave birth at or after 42 + 0 weeks (5.2% vs 1.6%). This suggests that they more often opted for expectant management in case of a prolonged pregnancy, despite existing recommendations in the Netherlands [29].

Outcomes of women in caseload care were compared to the matched cohort (Table 2). We found that women in caseload care were less often referred to obstetrician-led care (p-value < 0.001). A small majority (53.1%) stayed in primary midwife-led care without a referral to obstetrician-led care (nulliparous women: 40.3%; multiparous women: 65.2%), compared to 34.3% in the matched cohort (nulliparous women 23.8%; multiparous women: 44.2%). The overall referral rate was 46.9% in the caseload group and 65.7% in the matched cohort. A lower referral rate was observed both in the antenatal period and in the intrapartum period, for nulliparous and for multiparous women. In the antenatal period, 24.2% of women in caseload care experienced a referral to obstetrician-led care and 22.8% were referred intrapartum, versus 37.4 and 28.3% in regular care.

Interventions at the start of labour differed between the two groups (p-value < 0.001). Women in caseload care were more likely to experience a spontaneous start of labour (83.7%) compared to women in regular care (73.2%). They were less often induced (13.2% vs 21.0%) and less often had an elective caesarean section (3.0% vs 5.8%). A similar pattern existed for nulliparous and multiparous women. The place of birth was also different (p-value 0.001). A larger proportion of women in caseload care had an out-of-hospital birth (39.4% at home and 2.1% in a birth centre) compared to women in regular care (16.1% at home and 1.6% in a birth centre). More multiparous women than nulliparous women had an out-of-hospital birth, both in caseload practices and in regular care. The larger proportion of hospital births in obstetrician-led care in regular care (66.1% vs 47.5%) reflects the higher referral rate discussed earlier. A postpartum haemorrhage (> 1000 ml) occurred in 5.0% (nulliparous women 6.9%; multiparous women 3.3%) versus 6.9% in the regular care group (nulliparous women 8.3%; multiparous women 5.6%). This difference was not statistically significant (p-value 0.088). Fewer women in caseload care experienced an episiotomy (16.0% vs 26.4%, p-value < 0.001) and more had an intact perineum (41.3% vs 28.2%; p-value < 0.001). There was no significant difference in 3rd or 4th degree perineal ruptures (2.4% vs 2.8%; p-value 0.687). Nulliparous women were more likely to experience perineal damage compared to multiparous women, both in the caseload group and in the matched cohort.

Unfavourable perinatal outcomes were rare in both groups. Perinatal mortality occurred in one case (0.2%) in the caseload group and in seven cases (0.4%) in the matched cohort. A low Apgar score (< 7 at 5 min) was registered in 8 cases (1.2%) in the caseload group and in 35 cases (1.8%) in the matched cohort. These differences were not statistically significant.

In a sub-analysis we analysed births of women who had no antenatal referral to obstetrician-led care, describing births attended by a primary care midwife in caseload care (n = 496) and births from the regular care group (n = 1214). We excluded 39 (7.9%) births with a known risk factor (23 pregnancies ≥42 weeks) from the caseload group and 73 (6.0%; 7 pregnancies ≥42 weeks) from the regular care group (see Fig. 1 for a flowchart). As a result of this selection of women without known risk factors at the start of term labour, the groups are no longer matched and the results should be interpreted with caution.

Results are displayed in Table 3. In this sub-analysis the lower intrapartum referral rate in caseload care among nulliparous women (42.6% vs 63.0%) and multiparous women (13.3% vs 25.7%) is largely explained by a lower referral rate in the first stage of labour: 24.1% vs 44.4% for nulliparous women and 11.2% vs 19.8% for multiparous women. Both nulliparous and multiparous women in caseload care experienced fewer interventions during labour, and more often had a spontaneous vaginal birth. As in the main analysis, maternal morbidity was lower in the caseload group.

The main challenge in this comparison of results from caseload care and regular midwife-led care is the comparability of women in both groups. Women were included when in primary care at 28 weeks gestation and are therefore likely to have a relatively low risk profile. Several factors may still cause differences in risk profiles, although our matching procedure successfully minimized confounding by parity, age, or background. To control for regional variation in interventions we added the first two digits of the postal code in the matching procedure. Because our measure for SES and urbanization both are based on the postal code as well, we chose not to match on these variables. Based on the somewhat higher SES in the caseload group compared to the matched controls in regular care, a slightly more favorable risk profile cannot be ruled out. We could also not describe, nor control for, risk factors such as obesity, smoking or other life style factors since these are not reliably registered in the Perined database. On the other hand, we observed no important differences in preterm births, mean birthweight, and SGA (<p5) between the national cohort, the caseload group and the matched cohort. Therefore, we assume that the risk profiles of the caseload group and the matched cohort are not very different and do not lead to important bias.

Nevertheless, women in caseload care may be different from women in regular care in other important ways that cannot easily be assessed. The observed characteristics confirm that women in caseload care are a distinct group, with a higher than average motivation for a physiological birth, including homebirth and expectant management in prolonged pregnancy beyond 42 gestational weeks. Seeking the care of a caseload midwife may be a part of this inclination [25]. It is not possible to assess to what extent these preferences resulted in continuation of primary care or opting for homebirths in situations that usually lead to referral to secondary care in the hospital. We did find a slightly higher percentage of women with known risk factors- mainly gestational age ≥ 42 weeks - in the caseload group. Altogether, the lower referral and intervention rate we found will at least partly be a result of specific preferences among this group of women.

The study is too small for a reliable comparison in perinatal mortality or serious perinatal morbidity, since these outcomes are rare. At the same time, our results do not suggest that perinatal safety is compromised in caseload care: percentages of both perinatal mortality and low Apgar scores in the caseload and control groups in the caseload and control groups were similar. In a larger, preferably prospective, study it would be interesting to analyze perinatal results with the possibility to control for several risk factors including maternal lifestyle and to audit cases with severe perinatal morbidity or mortality.

While our study has certain methodological limitations – e.g. self-selection – our results are supported by high quality studies of caseload midwifery care. For instance, randomized trials [1, 2, 30, 31] showed no differences in neonatal outcomes such as low Apgar scores or admission to neonatal intensive care. A meta-analysis showed a higher spontaneous vaginal birth rate in caseload midwifery care [3]. Observational studies of caseload midwifery in various settings consistently show fewer interventions during childbirth, without compromising perinatal safety [32‐35].

We were not able to assess the level of one-to-one continuity of care in the caseload group because this information is not available in the Perined database. However, we are confident that the majority of the caseload clients received a high level of continuity of care throughout pregnancy and childbirth. In the inclusion procedure, participating midwives confirmed that they remained involved in the care when antenatal obstetric consultation or referral was needed, and all offered continuity of care during labour, even when there was an intrapartum referral. Most of them also inform women of this service on their practice website. Fontein [26] observed a similar pattern in practices with one or two midwives, where continuity of care was higher, and the midwife-woman relationship was experienced more positively when compared to group practices.

A recent meta-synthesis of 13 qualitative articles examining women’s perspectives on continuity of care models confirms the importance of such relationships for women, concluding that ‘the midwife–woman relationship is the vehicle through which trust is built, personalised care is provided, and the woman feels empowered’ [36]. Fontein et al. describe this type of relationship as a key feature of woman-centered care [37]. Such relationships are more likely to evolve in continuity models than in fragmented care offered in busy institutions or by large midwifery teams [38]. Indeed, caseload midwives in the Netherlands report that they offer this kind of woman-centered care [20‐23] in order to strengthen women and give them more control in the care process [39]. Our data suggest that a strong woman-centered model of care – with the midwife in a supporting and facilitating role – results in more control by women themselves and a more optimal, less medicalized pregnancy and birth process, with no compromise in perinatal safety. Improving woman-centeredness is an important point in the quality improvement agenda for maternity care in the Netherlands [40]. Based on our results, finding ways to implement caseload midwifery on a larger scale should be considered.