ABSTRACT
Objective
Addressing inequalities in maternal and newborn health is a UK public health priority. Evidence on effective multi‐interventional strategies is urgently needed. This study evaluated the impact of community‐based midwife continuity of care (CBMCOC) models for women and babies in ethnically diverse and socially disadvantaged areas of South London.
Design
We conducted a prospective cohort study using the eLIXIR, Born in South London, maternity–child data linkage.
Setting
United Kingdom.
Population
Pregnant women exposed to CBMCOC and standard care between 2018 and 2020.
Methods
Propensity score matching (1:4) was used to account for differences between CBMCOC and standard care cohorts and control for confounding bias. Conditional logistic regression estimated risk ratios. Subgroup analysis included women of Black, Asian and other ethnic minority groups, and those living in highly deprived areas.
Outcomes
The primary outcome was preterm birth (< 37 weeks' gestation). Secondary outcomes included other relevant maternal, perinatal, process and clinical variables.
Results
Before matching, 12 386 women were exposed to standard care and 1338 to CBMCOC; after matching, 5352 and 1338 were included, respectively. The risk of preterm birth was lower among women exposed to CBMCOC (unmatched: 4.6% vs. 10.3%, RR = 0.50, 95% CI: 0.38–0.64; matched: 4.6% vs. 8.4%, RR = 0.54, 95% CI: 0.40–0.70). Subgroup analyses showed reduced preterm birth rates among ethnic minority women and those in deprived areas when exposed to CBMCOC.
Conclusions
In this diverse population with a range of risk factors, locality‐based interventions integrating community‐based care and midwife continuity may reduce maternal and newborn health inequalities. Further trials of such models should be conducted.
Short abstract
Linked article: This article is commented on by Bone, pp. 974–975 in this issue. To view this article visit https://doi.org/10.1111/1471‐0528.70141.
1. Introduction
Global reports have described inequalities in maternal and newborn health outcomes and experiences for decades. Each year millions of women die from preventable causes that are not only related to complications in pregnancy, childbirth and the postnatal period, but also to manifestations of persistent global socioeconomic and health inequities [1]. The United Kingdom (UK) is no exception. Enquiries into maternal deaths have consistently found that women from the most disadvantaged and poorest backgrounds, and those from Black, Asian and minority ethnic groups are at greatest risk of severe mortality and morbidity [2, 3, 4, 5]. Their babies are also more likely to be stillborn, preterm, smaller than expected or die within their first month [3, 5, 6, 7, 8]. In addition, there are many challenges faced by recent migrants, and those with language barriers, emphasising the ongoing need for culturally sensitive care and improved access to services [7]. Integrative, holistic innovative solutions beyond maternity care are also needed to address underlying societal structures that impact health before, during and after pregnancy—such as housing, education and access to healthy environments [9, 10].
Three recent systematic reviews of policy, health and social care interventions to mitigate inequalities in maternal and newborn health among disadvantaged groups found that multicomponent approaches and interventions combining midwife continuity of care models and community‐based services could improve adverse pregnancy outcomes and increase access to care, attendance and engagement [11, 12, 13]. These findings are consistent with existing Cochrane reviews of community‐based intervention packages [14] and midwife continuity of care models [15], which highlight the need to integrate both maternal and neonatal care in community settings. These observations also align with health policy and the focus on people‐centred and place‐based models of maternity care that integrate health and care by shifting the way services are funded, managed and delivered ‘from health systems designed around diseases and health institutions towards health systems designed for people and communities’ [16, 17].
In England maternal policy includes an ambition to halve maternal mortality, neonatal mortality and serious brain injury in newborns by 2025 [18] and the Core20PLUS5 strategy to reduce inequalities in outcomes for mothers and babies [19]. There is some evidence on the effectiveness of continuity of care models (including place‐based care) for ‘at risk’ population groups, and limited understanding of care pathways and contextual factors surrounding severe mortality and morbidity among women with physical, mental and social risk factors [20, 21, 22]. We aimed to investigate the impact of community‐based midwife continuity of care models for women living in areas of ethnic diversity and social disadvantage in South London. To achieve this, we utilised a population maternal‐child data linkage of electronic health records, employing the method of propensity score matching to control for confounding bias.
2. Methods
2.1. Setting and Study Population
This prospective cohort study used the eLIXIR Born in South London (BiSL) programme [23]. eLIXIR is a unique population‐based database using opt‐out consent to collect real‐time, pseudonymised and routine maternity, neonatal and mental health data at two hospitals and primary care data from the Lambeth DataNet platform, enabling life course studies of physical and mental health in a large, diverse and inner urban population of South London, UK [23]. We included all pregnancies recorded between October 1, 2018 and March 1, 2020 (from the first antenatal appointment to discharge from maternity services). From an initial extraction of 13 795 pregnancies, we applied the following exclusion criteria: (1) multiple pregnancies (they wouldn't be eligible for community‐based midwife continuity of care but specialist care) and (2) pregnancies without complete data from the first antenatal appointment were excluded.
2.2. Data Sources
The eLIXIR Partnership database was utilised. Maternity and neonatal data for two maternity services are recorded on the BadgerNet electronic patient record system (CleverMed), which captures data from clinical records including clinical data, socio‐demographics, physical and mental health history, model of care and assessments for obstetric risk (e.g., a previous low birthweight or preterm infant, previous placental abruption or preeclampsia, gestational diabetes) and social risk (e.g., domestic violence, homelessness, drug misuse) [23]. Mental health data are obtained from the South London and Maudsley NHS Foundation Trust (SLaM) Clinical Records Interactive Search (CRIS) system, which generates variables extracted from electronic mental health records in SLaM. Primary care data are recorded from the Lambeth DataNet platform which captures, that is, clinical and development data, consultations, multimorbidity, long‐term conditions. For this study, relevant variables (detailed below) from BadgerNet systems, linked at the individual level as part of the eLIXIR partnership database, were extracted for the selected period, which represented the entire timespan available before the COVID‐19 pandemic and associated maternity services reconfiguration.
2.3. Exposure to Community‐Based Midwife Continuity of Care (CBMCOC) Models
There were eight models of CBMCOC in which the same midwife or team of midwives provided care to a woman from early pregnancy to the postnatal period; antenatal and postnatal care were predominantly provided in the community, and labour and birth care were at home or in the hospital. All models provided community‐based care based on women's geographical location and/or social vulnerability, and although many models were located in areas of high social deprivation, not all women under those models were considered vulnerable or had social risk factors. Some models included a team approach (a small team of midwives who share the caseload) while others included a caseload approach (a named and partner midwife who provide all the care), and the composition and modus operandi varied between the different models in terms of caseload size, team organization and on‐calls for childbirth care. However, in all models, midwives planned, organised and delivered comprehensive midwifery care in the community and hospital, that is, assessed needs, planned care, referred to other professionals, and coordinated services; they worked in partnership with the woman and with a multidisciplinary network of support. Women who developed complications in pregnancy, birth and the postnatal period were referred or escalated for obstetric care using the same clinical guidelines as in standard care; however the midwifery care continued to be provided from the team. Thus, while the models varied in composition and modus operandi, they were grouped together for the analysis. This approach was taken because all eight models shared the core principles of delivering midwife continuity of care within a community‐based setting, allowing for a sufficiently powered evaluation of this overall care type against standard care.
2.4. Exposure to Other Models of Maternity Care
Other models of maternity care included standard care models where obstetricians were the lead professionals for antenatal care and rostered midwives provided in‐hospital labour, birth and postpartum care for women having obstetrician‐led care (not necessarily by the obstetrician providing or leading antenatal care); and where midwives, GPs and obstetricians shared the responsibility for the organisation and delivery of care throughout the initial booking to the postnatal period provided in both hospital and/or community settings. These models are similar in that they do not aim to provide midwife continuity of care.
2.5. Outcomes
The primary outcome was preterm birth, defined as any birth that occurs before 37 completed weeks of gestation. Secondary maternal outcomes included onset of labour (spontaneous onset, induction, caesarean before labour), intrapartum analgesia/anaesthesia (none, epidural, spinal), mode of birth (spontaneous cephalic, vaginal breech, caesarean), section (elective and emergency) and instrumental (forceps and ventouse) birth, place of birth (hospital, home, other), perineal status after birth (intact perineum, first‐ and second‐degree tear, episiotomy, third‐ and fourth‐degree tear), estimated blood loss more than 500 mL, and maternal admission longer than 7 days. Secondary perinatal outcomes included stillbirth (born with no signs of life at or after 24 weeks of pregnancy) or neonatal death (death during the first 28 days), Apgar score at 5 min less than or equal to 7, low birthweight (< 2500 g), small for gestational age, skin‐to‐skin contact, first feed method (breast, bottle, other), admission to the neonatal unit. Secondary process outcomes included late booking for antenatal care (after 20 weeks), missed appointments, antenatal admissions (other than birth), and referrals (smoking, mental health and child protection). See Table S1 for BiSL variable definitions.
2.6. Co‐Variables
The following variables were adjusted for in all models: participants' socio‐demographic characteristics included maternal age, self‐reported ethnicity (grouped into: White, Black, Asian, Mixed and Other), Index of Multiple Deprivation quintile (IMD, a method used to measure social and economic deprivation in small areas of England and Wales; a score of 1 indicates the most deprived and a score of 5 the least deprived) [24]. Clinical and other characteristics at the first antenatal visit included: parity, previous preterm birth, existing physical or mental health conditions (including Whooley questions for identification of possible depression [25]).
2.7. Ethics Approvals
Ethical approval eLIXIR is granted by the Oxford Central Research Ethics Committee (23/SC/0116). The Health Research Authority Confidentiality Advisory Group (HRA CAG Ref: 18/CAG/0040) provided approval under Section 251 (s251) of the NHS Act (2006). This study was also approved by the eLIXIR Oversight Committee (RAF: DL021R).
2.8. Statistical Analysis
The sample size required was calculated based on the preterm birth rate with an expected 2% reduction aligned with the national maternity policy safety ambition policy [26]. We anticipated 240 women per year from each community‐based midwife continuity of care model and 2 years of data collection, meaning 1920 women exposed to community‐based models, and (with 4:1 matching) 7680 women exposed to standard care models, to provide 86.4% power to detect a reduction in preterm birth from 8% to 6%. Propensity scores were calculated for each woman using multivariable logistic regressions based upon the covariates: age, ethnicity, indices of deprivation (IMD), parity, social risk factors, Whooley positive, pre‐existing medical and mental conditions, along with an interaction term between Whooley positive and pre‐existing mental health condition. To ensure that all covariates were measured, a total of 4122 cases in which one covariate was unmeasured were removed from the sample, Data were deemed to be missing completely at random and therefore complete case analysis was adopted over multiple imputation [27]. Matching was performed in R using the packages, MatchIt [28], Optmatch [29] and MatchThem [30]. Nearest neighbour matching was used at a ratio of 1:4 and caliper of 0.2; given the much larger untreated group this ratio was chosen to reduce selection bias [31, 32]. Analysis included all women based on their assigned care model at the beginning of pregnancy, regardless of whether they fully adhered to that assigned model.
Demographic variables were compared using standardised mean differences (SMD) to assess differences in characteristics before and after the matching process (a SMD greater than 0.1, or 10% would indicate an imbalance). We fitted logistic regression models to compare key maternal outcomes (e.g., onset of labour, mode of birth) and perinatal outcomes (e.g., preterm birth, stillbirth, birthweight, Apgar score) between the two care groups, when the outcome was binary. Multinomial logistic regressions were fitted to investigate the impact of the intervention on non‐binary (multiple level) outcomes (e.g., perineal trauma, type of delivery, onset of birth), producing bootstrapped 95% confidence intervals. A sub‐group analysis by social deprivation and ethnicity was conducted, to assess the effect of the care model on preterm birth. Women from Black, Asian, mixed or other ethnic backgrounds were compared to women of White ethnicity. Women living in IMD quintile 1 or 2 were compared to those living in IMD quintiles 3, 4 and 5. Before and after propensity scoring models were also adjusted for age, ethnicity, deprivation, parity, prior preterm birth, social risk factors, Whooley positive result, pre‐existing medical and mental health conditions. To account for unmeasured confounding E values were calculated for all models. All models contained a random intercept of women's ID to account for shared variance across pregnancies. As recommended for sensitivity analysis for observational research [33, 34], we also calculated E‐values, which represent the strength of association that an unmeasured confounder would require with the exposure and outcome to attenuate main associations to non‐significance. R was used for data manipulation and analysis using the following packages, lme4 [35], tidyverse [36], naniar [37], jtools [38] and afex [39].
3. Results
17 917 pregnancies were initially extracted with birth outcomes; 4122 were excluded due to multiple pregnancies or missing data for the covariates of ethnicity and IMD from antenatal booking. Data relating to 13 795 pregnancies from 1 October 2018 to 1 March 2020 were finally extracted. 251 completed duplicates were removed from the dataset. The first pregnancy recorded within the eLIXIR database was considered the index pregnancy. Overall, 13 609 pregnancies with completed data from their first antenatal appointment were included in the final dataset and analysis.
3.1. Baseline Characteristics
The baseline characteristics of women exposed to standard care and women exposed to CBMCOC before propensity score matching (12 386 and 1338, respectively) and after matching (5352 and 1338, respectively) are presented in Table 1. Before matching, there were differences between women in CBMCOC and standard care in some sociodemographic and clinical characteristics. Women in CBMCOC were more likely to be primiparous, White and less likely to be Asian, more likely to be born in the UK and have English as their primary language compared to women in standard care. They were also more likely to have a Whooley positive and prior mental health condition, showing moderate imbalance in comparison to standard care. Following matching score adjustment, there were negligible imbalances between the groups.
TABLE 1.
Maternal baseline characteristics.
| Socio‐demographics | ||||||
|---|---|---|---|---|---|---|
| Unadjusted | Propensity‐adjusted | |||||
| Standard care n = 12386 | CBMCOC n = 1338 | SMD | Standard care n = 5352 | CBMCOC n = 1338 | SMD | |
| Age at booking (years) | 32.71 (5.19) | 32.73 (5.36) | 0.00 | 32.81 (5.42) | 32.73 (5.36) | 0.01 |
| Ethnicity a | ||||||
| White | 6352 (51.3) | 778 (58.1) | 0.14 | 3091 (57.7) | 778 (58.1) | 0.00 |
| Black | 2662 (21.5) | 265 (19.8) | 0.03 | 1116 (20.8) | 265 (19.8) | 0.03 |
| Asian | 1127 (9.1) | 65 (4.9) | 0.18 | 300 (5.6) | 65 (4.9) | 0.08 |
| Mixed | 561 (4.5) | 78 (5.8) | 0.09 | 280 (5.2) | 78 (5.8) | 0.06 |
| Other | 867 (7.0) | 73 (5.4) | 0.08 | 269 (5.0) | 73 (5.5) | 0.05 |
| Not stated | 817 (6.6) | 79 (5.9) | 0.03 | 296 (5.5) | 79 (5.9) | 0.02 |
| IMD quintile | ||||||
| 1 (most deprived) | 2424 (19.6) | 263 (19.7) | 0.01 | 1045 (19.5) | 263 (19.7) | 0.00 |
| 2 | 5123 (41.4) | 567 (42.4) | 0.01 | 2223 (41.5) | 567 (42.4) | 0.01 |
| 3 | 3156 (25.5) | 369 (27.6) | 0.03 | 1491 (27.8) | 369 (27.6) | 0.01 |
| 4 | 1225 (9.9) | 99 (7.4) | 0.09 | 414 (7.7) | 99 (7.40) | 0.02 |
| 5 (least deprived) | 458 (3.6) | 40 (3.0) | 0.07 | 179 (3.3) | 40 (3.0) | 0.06 |
| Born in the UK (yes) | 5154 (41.6) | 740 (55.3) | 0.28 | 2883 (53.9) | 740 (55.3) | 0.02 |
| Primary language English | 8545 (69.0) | 999 (74.7) | 0.12 | 4064 (75.9) | 999 (74.7) | 0.01 |
| Support status at booking b | ||||||
| Supported | 11114 (89.0) | 1181 (88.3) | 0.01 | 4848 (90.6) | 1181 (88.3) | 0.08 |
| Unsupported | 169 (1.4) | 19 (1.4) | 0.01 | 90 (1.7) | 19 (1.4) | 0.02 |
| Sheltered accommodation | 5 (0.0) | 5 (0.4) | 0.12 | 3 (0.1) | 5 (0.4) | 0.07 |
| Other | 144 (1.2) | 23 (1.7) | 0.15 | 85 (1.6) | 23 (1.7) | 0.05 |
| Clinical and social characteristics | ||||||
| Gestational age at booking (weeks) | 12.08 (7.01) | 11.00 (5.41) | 0.17 | 11.06 (5.72) | 11.00 (5.41) | 0.01 |
| BMI at booking | 24.20 (6.37) | 24.03 (6.15) | 0.03 | 24.04 (6.16) | 24.03 (6.15) | 0.00 |
| Primiparous | 5732 (43.6) | 793 (59.3) | 0.26 | 3111 (58.13) | 793 (59.3) | 0.07 |
| Smoking at booking | 462 (3.7) | 67 (5.0) | 0.11 | 266 (5.0) | 67 (5.0) | 0.00 |
| Pre‐existing physical conditions c | ||||||
| Hypertension | 193 (1.6) | 7 (0.5) | 0.08 | 80 (1.5) | 7 (0.5) | 0.08 |
| Asthma | 32 (0.3) | 5 (0.4) | 0.02 | 15 (0.3) | 5 (0.4) | 0.02 |
| Autoimmune disease | 151 (1.2) | 10 (0.7) | 0.04 | 55 (1.0) | 10 (0.7) | 0.03 |
| Diabetes | 273 (2.2) | 10 (0.7) | 0.12 | 101 (1.9) | 10 (0.7) | 0.09 |
| Chronic renal disease | 95 (0.8) | 4 (0.3) | 0.05 | 38 (0.7) | 4 (0.3) | 0.05 |
| Haematology disorders | 604 (4.88) | 40 (2.9) | 0.10 | 191 (3.6) | 40 (2.9) | 0.03 |
| Cardiac disorders | 52 (0.42) | 6 (0.45) | 0.00 | 16 (0.30) | 6 (0.4) | 0.00 |
| Pre‐existing mental health conditions | 2170 (17.5) | 401 (30.0) | 0.32 | 1676 (31.3) | 401 (30.0) | 0.02 |
| Whooley Positive | 977 (7.9) | 243 (18.2) | 0.37 | 934 (17.5) | 243 (18.2) | 0.02 |
| Obstetric risk | 1785 (14.4) | 181 (13.5) | 0.03 | 743 (13.9) | 181 (13.5) | 0.01 |
| Previous preterm birth | 346 (2.8) | 49 (3.66) | 0.05 | 192 (3.6) | 49 (3.66) | 0.03 |
| Social risk | 1424 (11.5) | 172 (12.7) | 0.04 | 3131 (58.5) | 799 (59.7) | 0.02 |
Note: Data are n (%), Age at booking and gestational age at booking: mean + standard deviation, BMI at booking: median (inter‐quartile range).
Abbreviations: BMI, body mass index; CBMCOC, community‐based midwife continuity of care; CI, confident interval; IMD, index of multiple deprivation; RR, risk ratio; SMD, standardised mean differences.
817 in standard care models and 79 in community‐based models were missing, not recorded or not stated (propensity‐adjusted: 296 and 79 respectively).
1123 in standard care models and 110 in community‐based models were missing (propensity‐adjusted: (propensity‐adjusted: 326 and 110 respectively)).
Poorly recorded and reported the most common pre‐existing conditions.
3.2. Primary Outcome
Women in CBMCOC experienced a significantly reduced risk of preterm birth compared to those in standard care. Before adjustment, preterm birth occurred in 4.6% of the CBMCOC group and 10.3% of the standard care group, corresponding to a risk ratio (RR) of 0.50 (95% CI: 0.38–0.64) and an absolute risk difference (RD) of −5.8 percentage points. After propensity score adjustment, the risk of preterm birth in CBMCOC remained significantly lower at 4.6% versus 8.4% in standard care (RR 0.54, 95% CI: 0.40–0.70; RD –3.8 pp). These results indicate that CBMCOC was associated with an absolute reduction of 38–58 fewer preterm births per 1000 women compared with standard care (Table 2).
TABLE 2.
Primary outcome.
| Unadjusted | Propensity‐adjusted a | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Standard care n = 12386 | CBMCOC n = 1338 | RR (95% CI) | RD | p | E value | Standard care n = 5352 | CBMCOC n = 1338 | RR (95% CI) | RD | p | E value | |
| Preterm birth (< 37 weeks) | 1277 (10.3) | 61 (4.6) | 0.50 (0.38, 0.64) | −5.7 | < 0.001 | 3.41 | 450 (8.4) | 61 (4.6) | 0.54 (0.40, 0.70) | −3.8 | < 0.001 | 3.11 |
Note: Data are n (%).
Abbreviations: CBMCOC, community‐based midwife continuity of care; CI, confident interval; RD, risk difference; RR, risk ratio.
Adjusted for age, ethnicity, deprivation Index quintile, parity, previous preterm birth, social risk factors, Whooley positive, pre‐existing medical and mental conditions.
3.3. Secondary Outcomes
Secondary maternal outcomes are presented in Table 3. After adjustments and compared to women in standard care, women in CBMCOC were more likely to experience spontaneous onset of labour (59% vs. 47% RD +12 pp), spontaneous vaginal birth (63% vs. 49%, RD +13 pp) and skin‐to‐skin contact after birth (85% vs. 79%, RR 1.09, 95% CI: 1.03–1.16; RD –6.0 pp); and they were less likely to use intrapartum analgesia/anaesthesia (14% vs. 9%; RR 1.68, 95% CI: 1.42–1.99; RD +5.6 pp) or experience induction of labour (24% vs. 27%; RR 0.71, 95% CI: 0.61–0.82, RD –2.6 pp), caesarean birth (25% vs. 36%; RR 0.53 95% CI: 0.46–0.61; RD 13.6 pp) and instrumental birth (11% vs. 14%; RR 0.65 95% CI: 0.53–0.79, RD –2.9 pp). In terms of perineal outcomes, CBMCOC women were less likely to have an episiotomy (12% vs. 16%, RR 0.71, 95% CI: 0.59, 0.85; RD –4.2 pp) and more likely to have first‐ or second‐degree tears (41% vs. 32%, RR 1.25, 95% CI: 1.08–1.44; RD +9.0 pp), with no differences in intact perineum and third‐ or fourth‐degree tears. The risk of blood loss over 500 mL was also reduced in CBMCOC (38% vs. 45%, RR 0.85, 95% CI: 0.77–0.93; RD –7.0 pp). Women exposed to CBMCOC were also more likely to give birth at home (15.7% vs. 0.9%; RR 17.41, 95% CI: 12.61–24.03; RD +14.8 pp), with corresponding reductions in hospital birth and prolonged postnatal stay in comparison to those in standard care (9% vs. 11%, RR 0.71, 95% CI: 0.57–0.87).
TABLE 3.
Secondary maternal outcomes.
| Unadjusted | Propensity‐adjusted a | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Standard care n = 12386 | CBMCOC n = 1338 | RR (95% CI) | p | RD | E value | Standard care n = 5352 | CBMCOC n = 1338 | RR (95% CI) | p | RD | E value | |
| Onset of labour | ||||||||||||
| Spontaneous b | 5669 (45.7) | 787 (58.8) | 13.1 | 2492 (46.6) | 787 (58.8) | 12.2 | ||||||
| Induction | 3197 (25.8) | 324 (24.2) | 0.72 (0.63, 0.83) | < 0.001 | −1.60 | 2.12 | 1432 (26.8) | 324 (24.2) | 0.71 (0.61, 0.82) | < 0.001 | −2.6 | 2.17 |
| Planned caesarean birth before onset of labour | 3404 (27.5) | 227 (17.0) | 0.47 (0.40, 0.56) | < 0.001 | −10.5 | 3.68 | 1428 (16.7) | 227 (17.0) | 0.49 (0.42, 0.58) | < 0.001 | 0.3 | 3.50 |
| Analgesia/anaesthesia | ||||||||||||
| None b | 1023 (8.3) | 192 (14.4) | 1.68 (1.43, 1.96) | < 0.001 | 6.1 | 2.75 | 466 (8.7) | 192 (14.4) | 1.68 (1.42, 1.99) | < 0.001 | 5.7 | 2.75 |
| Epidural | 3649 (29.5) | 273 (20.4) | −9.1 | 1751 (32.7) | 273 (20.40) | −12.31 | ||||||
| Spinal | 2942 (23.7) | 220 (16.4) | −7.3 | 1438 (26.9) | 220 (16.4) | −10.5 | ||||||
| Type of birth | ||||||||||||
| Spontaneous Cephalic vaginal birth b | 5705 (46.1) | 844 (63.1) | 17.0 | 2651 (49.5) | 844 (63.1) | 13.6 | ||||||
| Breech birth | 65 (0.5) | 7 (0.5) | 0.82 (0.37, 1.80) | 0.612 | 0.0 | 1.74 | 23 (0.4) | 7 (0.5) | 0.97 (0.41, 2.27) | 0.937 | 0.1 | 1.21 |
| Caesarean birth | 4565 (36.8) | 334 (25.0) | 0.50 (0.44, 0.58) | < 0.001 | 17.2 | 3.41 | 1909 (35.7) | 334 (25.0) | 0.53 (0.46, 0.61) | < 0.001 | 13.6 | 3.18 |
| Planned | 2032 (16.4) | 135 (10.1) | −6.3 | 883 (16.5) | 135 (10.1) | −6.4 | ||||||
| Unplanned | 2533 (20.4) | 199 (14.9) | −5.5 | 1026 (19.2) | 199 (14.9) | −4.3 | ||||||
| Instrumental birth | 1930 (16.5) | 153 (11.4) | 0.61 (0.51, 0.74) | < 0.001 | −5.1 | 2.66 | 766 (14.3) | 153 (11.40) | 0.65 (0.53, 0.79) | < 0.001 | −2.9 | 2.45 |
| Perineal status c | ||||||||||||
| Intact perineum b | 3813 (30.8) | 455 (34.0) | 3.2 | 1766 (33.0) | 455 (34.0) | 1.0 | ||||||
| 1st/2nd degree tear | 3943 (31.8) | 548 (41.0) | 1.21 (1.06, 1.38) | < 0.01 | 9.2 | 1.71 | 1711 (32.0) | 548 (41.0) | 1.25 (1.08, 1.44) | < 0.01 | 9.0 | 1.81 |
| 3rd/4th degree tear | 237 (1.9) | 27 (2.0) | 1.21 (0.80, 1.84) | 0.364 | 0.1 | 1.71 | 83 (1.6) | 27 (2.0) | 1.34 (0.85, 2.10) | 0.205 | 0.4 | 2.01 |
| Not known | 32 (0.3) | 2 (0.1) | 0 (0.0) | 2 (0.1) | ||||||||
| Episiotomy | 2265 (18.3) | 163 (12.2) | 0.62 (0.52, 0.73) | < 0.001 | −6.1 | 2.61 | 877 (16.4) | 163 (12.2) | 0.71 (0.59, 0.85) | < 0.001 | −4.2 | 2.17 |
| Blood loss > 500 mls | 5825 (47.0) | 504 (37.7) | 0.83 (0.75, 0.91) | < 0.001 | −9.3 | 1.70 | 2391 (44.7) | 504 (37.7) | 0.85 (0.77, 0.93) | < 0.001 | −7.0 | 1.63 |
| Place of birth (where known) | 34.31 | |||||||||||
| Home | 78 (0.6) | 211 (15.7) | 21.79 (16.80, 28.55) | < 0.001 | 15.1 | 43.07 | 46 (0.9) | 211 (15.7) | 17.41 (12.61, 24.03) | < 0.001 | 14.8 | |
| Hospital | 9162 (74.0) | 853 (63.8) | −10.2 | 3987 (74.5) | 853 (63.8) | −10.7 | ||||||
| Missing | 3146 (25.4) | 274 (20.5) | 1327 (24.8) | 274 (20.6) | ||||||||
| Skin to skin established after birth | 10375 (83.8) | 1251(79.3) | 1.09 (1.03, 1.16) | 0.003 | −4.5 | 1.40 | 4566 (85.3) | 1251(79.3) | 1.09 (1.03, 1.16) | 0.005 | −6.0 | 1.40 |
| Postnatal admission longer than 7 days | 1409 (11.4) | 118 (8.8) | 0.73 (0.59, 0.88) | < 0.01 | −2.6 | 2.17 | 601 (11.2) | 118 (8.8) | 0.71(0.57, 0.87) | < 0.001 | −2.3 | |
Note: Data are n (%).
Abbreviations: CBMCOC, community‐based midwife continuity of care; CI, confident interval; RD, risk difference; RR, risk ratio.
Adjusted for age, ethnicity, deprivation Index quintile, parity, previous preterm birth, social risk factors, Whooley positive, pre‐existing medical and mental conditions.
Variable selected for conditional logistic regression.
4361 in standard care models and 306 in community‐based models were missing or not recorded (propensity‐adjusted: 1845 and 306 respectively).
Perinatal outcomes are shown in Table 4. CBMCOC and standard care groups did not differ for the risk of stillbirth or neonatal death. The proportion of small for gestational age babies and low birthweight babies was significantly lower in CBMCOC compared to standard care (5.7% vs. 7.6%, RR 0.75, 95% CI: 0.59–0.95; RR −1.9 pp and 5.7% vs. 9.4%; RR 0.63, 95% CI: 0.50–0.39; RD –3.7 pp, respectively). There were neither differences in first feed method nor in admission to the neonatal unit, or five‐minute Apgar score.
TABLE 4.
Secondary perinatal outcomes.
| Unadjusted | Propensity‐adjusted a | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Standard care n = 12386 | CBMCOC n = 1338 | RR (95% CI) | p | RD | E value | Standard care n = 5352 | CBMCOC n = 1338 | RR (95% CI) | p | RD | E value | |
| Birth outcome | ||||||||||||
| Livebirth b | 12142 (98.0) | 1336 (99.8) | 5.82 (1.87, 30.76) | 0.802 | 1.8 | 11.11 | 5313 (99.3) | 1336 (99.8) | 5.87 (1.85, 31.69) | 0.852 | 0.5 | 3.11 |
| Stillbirth or neonatal death | 128 (1.0) | 2 (0.1) | −0.4 | 39 (0.7) | 2 (0.1) | −0.2 | ||||||
| Five minutes Apgar score > 7 | 11519 (93.0) | 1279 (97.8) | 1.27 (0.89, 1.87) | 0.833 | 4.8 | 1.85 | 5040 (94.2) | 1279 (97.8) | 1.31 (0.90, 1.93) | 0.831 | 3.6 | 1.45 |
| Low birthweight (< 2.500g kg) | 1240 (10.0) | 76 (5.7) | 0.57 (0.44, 0.72) | < 0.001 | −4.3 | 2.90 | 501 (9.4) | 76 (5.7) | 0.60 (0.46, 0.75) | < 0.001 | −3.7 | 2.72 |
| Small for gestational age | 1117 (9.0) | 76 (5.7 ) | 0.63 (0.50, 0.79) | < 0.001 | −3.3 | 2.78 | 406 (7.6) | 76 (5.7) | 0.75 (0.59, 0.95) | < 0.05 | −1.9 | 2.50 |
| First feed method | 1.02 (0.97, 1.09) | 0.362 | 1.16 | 1.03 (0.96, 1.09) | 0.415 | 1.21 | ||||||
| Breast b | 10606 (85.6) | 1222 (91.3) | 5.7 | 4642 (86.7) | 1222 (91.3) | 4.6 | ||||||
| Bottle | 986 (7.9) | 83 (6.2) | −1.7 | 450 (8.4) | 83 (6.2) | −2.2 | ||||||
| Other: NGT, etc | 794 (6.4%) | 33 (2.5) | −3.9 | 8 (0.1) | 33 (2.5) | 2.4 | ||||||
| Admission to the neonatal unit | 964 (7.7) | 73 (5.5) | 1.40 (0.53, 3.72) | 0.647 | −2.2 | 6.87 | 427 (8.0) | 73 (5.5) | 1.45 (0.53, 3.97) | 0.633 | −2.5 | 5.33 |
Note: Data are n (%).
Abbreviations: CBMCOC, community‐based midwife continuity of care; CI, confident interval; NGT, nasogastric tube; RD, risk difference; RR, risk ratio.
Adjusted for age, ethnicity, deprivation Index quintile, parity, previous preterm birth, social risk factors, Whooley positive, pre‐existing medical and mental conditions.
Variable selected for conditional logistic regression.
Process outcomes (Table 5) showed no significant differences after adjustments in late booking for antenatal care (16% vs. 18%; RD –2.0 pp). Missed appointments were significantly lower in CBMCOC (5.4% vs. 15.7%; RR 0.66, 95% CI: 0.55–0.78; RD –10.3 pp) and antenatal admissions (excluding birth) were rare (< 1%) in both groups. Referral patterns differed; women in CBMCOC were more likely to receive referrals for mental health services (5% vs. 3%; RR 1.80, 95% CI: 1.26–2.37) with no differences in referrals for smoking or child protection.
TABLE 5.
Secondary outcomes: processes.
| Unadjusted | Propensity‐adjusted a | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Standard care n = 12386 | CBMCOC n = 1338 | RR 95% CI | p | RD | E value | Standard care n = 5352 | CBMCOC n = 1338 | RR 95% CI | p | RD | E value | |
| Late booking for antenatal care | 1870 (15.1) | 211 (15.8) | 0.99 (0.93, 1.05) | 0.720 | −0.7 | 1.11 | 954 (17.8) | 211 (15.8) | 0.99 (0.93, 1.06) | 0.806 | −2.0 | 1.11 |
| Reason: (descriptive) | ||||||||||||
| Not letter/reminder | 26 (1.4) | 3 (1.4) | 9 (0.9) | 3 (1.4) | ||||||||
| Late referral | 360 (19.3) | 46 (21.8) | 133 (13.9) | 46 (21.8) | ||||||||
| Moved within the UK | 162 (8.7) | 22 (10.4) | 48 (5.0) | 22 (10.4) | ||||||||
| Appointment re‐scheduled | 72 (3.6) | 9 (4.3) | 25 (2.6) | 9 (4.3) | ||||||||
| Unaware/unsure of pregnancy/PV bleed | 217 (11.6) | 15 (7.1) | 88 (9.2) | 15 (7.1) | ||||||||
| Unwell/too sick to attend | 5 (0.3) | 0 (0.0) | 2 (0.2) | 0 (0.0) | ||||||||
| Other or not stated | 1028 (55.1) | 116 (55) | 649 (68.0) | 116 (55.0) | ||||||||
| One or more missing appointments | 1085 (8.8) | 71 (5.4) | 0.66 (0.56, 0.79) | < 0.0001 | −3.4 | 2.40 | 841 (15.7) | 71 (5.4) | 0.66 (0.55, 0.78) | < 0.001 | −10.3 | 2.40 |
| One or more antenatal admissions (other than birth) | 63 (0.5) | 2 (0.1) | 0.27 (0.04, 0.88) | 0.072 | −0.4 | 6.87 | 23 (0.43) | 2 (0.1) | 0.34 (0.05, 1.15) | 0.142 | −0.3 | 5.33 |
| Referrals | ||||||||||||
| Smoking | 373 (3.0) | 63 (4.7) | 1.38 (1.04, 1.79) | < 0.05 | 1.7 | 2.10 | 210 (3.9) | 63 (4.7) | 1.14 (0.86, 1.51) | 0.352 | 0.8 | 1.54 |
| Mental health | 234 (1.9) | 70 (5.2) | 2.44 (1.85, 3.19) | < 0.001 | 3.3 | 4.31 | 185 (3.5) | 70 (5.2) | 1.80 (1.26, 2.37) | < 0.001 | 1.7 | 3.00 |
| Child protection | 15 (0.1) | 4 (0.3) | 3.24 (0.89, 9.54) | < 0.05 | 0.2 | 5.93 | 11 (0.21) | 4 (0.3) | 2.51 (0.79, 7.97) | 0.118 | 0.0 | 4.46 |
Note: Data are n (%). n/N (%) indicates that the denominator only includes participants with a relevant measurement for that variable.
Abbreviations: CBMCOC, community‐based midwife continuity of care; CI, confident interval; RD, risk difference; RR: risk ratio.
Adjusted for age, ethnicity, deprivation Index quintile, parity, previous preterm birth, social risk factors, Whooley positive, pre‐existing medical and mental conditions.
3.4. Subgroup Analysis
Subgroup analyses (Table 6) suggested that reductions in preterm birth in women in CBMCOC were evident across disadvantaged populations. Among women of Black, Asian, mixed, or other ethnicities, preterm birth was significantly reduced from 9.5% in standard care to 6.4% in CBMCOC (RR 0.66, 95% CI: 0.44–0.95; RD –3.1 pp) compared to women of Black, Asian, mixed, or other ethnicities exposed to standard care. Similarly, among women in the most deprived quintiles, preterm birth fell from 8.2% to 5.1% (RR 0.60, 95% CI: 0.43–0.82; RD –3.1 pp) compared to those in standard care.
TABLE 6.
(A) Subgroup 1: Effect among Black, Asian, mixed and other ethnic groups of women on preterm birth and outcome. (B) Subgroup 2: Effect among women living in high deprivation (IMD 1 and 2) on preterm birth outcome.
| (A) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Unadjusted | Propensity‐adjusted a | |||||||||||
| Standard care n = 5217 | CBMCOC n = 481 | RR 95% CI | p | RD | E value | Standard care n = 1965 | CBMCOC n = 481 | RR 95% CI | p | RD | E value | |
| Preterm birth (< 37 weeks) | 550 (10.5) | 31 (6.4) | 0.60 (0.41, 0.85) | < 0.01 | −4.1 | 2.72 | 186 (9.5) | 31 (6.4) | 0.66 (0.44, 0.95) | < 0.05 | −3.1 | 2.40 |
| (B) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Unadjusted | Propensity‐adjusted a | |||||||||||
| Standard care n = 7447 | CBMCOC n = 830 | RR 95% CI | p | RD | E value | Standard care n = 3268 | CBMCOC n = 830 | RR 95% CI | p | RD | E value | |
| Preterm birth (< 37 weeks) | 672 (9.0) | 42 (5.0) | 057 (0.41, 0.77) | < 0.001 | −4.0 | 2.90 | 268 (8.2) | 42 (5.1) | 0.60 (0.43, 0.82) | < 0.01 | −3.1 | 2.72 |
Note: Data are n (%).
Abbreviations: CBMCOC, community‐based midwife continuity of care; CI, confident interval; RD, risk difference; RRs, risk ratio.
Adjusted for age, ethnicity, deprivation Index quintile, parity, previous preterm birth, social risk factors, Whooley positive, pre‐existing medical and mental conditions.
4. Discussion
4.1. Main Findings
This study found that women living in South London who were exposed to models of community‐based midwife continuity of care (CBMCOC), after adjustment were significantly less likely to have a preterm birth compared to women exposed to models of standard maternity care. They were also significantly more likely to have a spontaneous onset of labour, a spontaneous cephalic birth, 1st and 2nd degree tears, skin‐to‐skin contact established and home birth; and significantly less likely to miss appointments, use intrapartum analgesia/anaesthesia, or experience induction of labour, caesarean section, instrumental birth, episiotomy, blood loss of more than 500 mL, and postnatal admission longer than 7 days. It is probable that the reduction of blood loss over 500 mL reflects the lower caesarean birth rate in CBMCOC. There were no differences in stillbirth or neonatal death but infants of women in CBMCOC were significantly less likely to be low birthweight or small for gestational age. In terms of process outcomes, women in CBMCOC had significantly more referrals to mental health services, and fewer missed appointments compared to those in standard care. The likelihood of having a preterm birth was also significantly lower among Black, Asian and other ethnic minority groups, and women living in the most deprived areas exposed to CBMCOC, compared to similar groups of women exposed to standard care.
4.2. Interpretation
Many of our findings align with a previous metanalysis and individual studies. The recent 2024 update of a Cochrane review of randomised controlled trials (RCTs) of continuity of care models in participants from heterogeneous populations found women receiving these models were less likely to experience a caesarean birth, instrumental birth and episiotomy, and more likely to have a spontaneous vaginal birth and report a positive experience but no differences in preterm birth [15]. The review found no RCT of continuity of care targeting women from specific ethnic groups or disadvantaged backgrounds; however there is promising evidence from observational and synthesis data. Rayment‐Jones et al. [40], for example, found women with complex social factors who received community‐based caseload continuity of midwife care experienced more spontaneous births, water use for pain relief, skin‐to‐skin contact, early access to care and referral to support services. Follow‐up research highlighted the protective nature of community‐based midwife continuity of care in reducing preterm birth, low birthweight, and induction of labour, particularly for women with the highest level of social risk [22]. Various observational studies, including two smaller UK studies of caseload midwifery in inner‐city deprived and diverse communities have also shown decreases in preterm birth and caesarean birth [41, 42], and increases in spontaneous births, home births and Apgar scores of > 8 at 5 min [41]. Research with Australian Indigenous women showed that collaborative models of midwife continuity of care integrated with Indigenous governance, family services, and community‐based hubs improved antenatal attendance, reduced preterm birth and increased breastfeeding on discharge [21, 43]. An analysis of the Maternity Services Dataset covering 922 149 women in England found that midwife continuity of care increases the uptake of the first breast milk feed and may reduce stillbirth rates for Black women, providing key insights for future healthcare policy [44]. Finally, a recent systematic review of targeted health and social care interventions for women and infants impacted by health inequalities in high‐income countries found multi‐interventional approaches could enhance a targeted approach for at‐risk populations, in particular combining midwifery models of care with community‐centred approaches, to enhance accessibility, earlier engagement, increased attendance and improved outcomes [11]. Deprivation and ethnicity remain key drivers of inequalities in maternal health, and prevention strategies need to address social and structural determinants in areas of high deprivation and minority ethnicity groups [45].
Recent WHO, ICM and FIGO good practice recommendations highlighted that midwife continuity of care within existing, context‐appropriate care models, in primary as well as secondary care, is pivotal to delivering high‐quality care across the pregnancy continuum, prior to conception, through pregnancy and birth and beyond [46, 47]. Models or packages of care, such as CBMCOC, are complex healthcare interventions and understanding the mechanisms by which they influence outcomes, particularly among women from diverse and disadvantaged groups, is crucial. Synthesising findings from several realist reviews and studies of midwife continuity of care can offer a structured framework for this, and mechanisms can be grouped into three core themes: (1) the woman‐midwife partnership; (2) improved maternity pathways and processes and (3) enabling system resources. At the heart of the model is the woman‐midwife partnership, where relational continuity is key; this ongoing relationship engenders mutual trust and confidence between women and midwives. The response to this trust is that women feel safer, less anxious, more respected, and empowered, and this trusting environment facilitates the disclosure of sensitive social risk factors and eases women's perceptions of stigma or surveillance, particularly for those with social care involvement—this allows for the provision of practical and emotional support that enables women to become active participants in their care [48, 49, 50, 51]. Second, these models improve maternity pathways and processes. The known midwife acts as an effective care coordinator and advocate, helping women navigate what is often a fragmented and unfamiliar system, and this mechanism appears to be enhanced when care is based in the community. Midwives in community settings report better integration with local services, which in turn helps women build a wider support network [48, 50, 52]. Finally, the success of these models depends on enabling system resources. This includes not just organisational infrastructure and partnerships but also addressing the professional and systemic challenges that arise from changing models of care. Implementing continuity models disrupts established professional roles and power structures, which can create role ambiguity and conflict [49], thus overcoming these barriers requires strong leadership, a shared philosophy among providers, and clear policies to support midwives and ensure the model is sustainable [50, 51, 52].
Recent policy in England has focused on improving access to midwife continuity of care for women from ethnic minority groups and those living in deprived areas [16, 18]. Although current evidence shows that ethnicity is associated with socioeconomic deprivation, Black and minority ethnic women who are not socially deprived still experience worse outcomes than their white counterparts [53]. This could be due to area levels of deprivation being used rather than individual indicators, that overlook determinants of health such as wealth, social status, isolation and social capital [48]. Understanding the impact that these measures have on birth outcomes for ethnic minority groups will enable maternity providers to target women who are most at risk whilst avoiding stereotypical assumptions and racial profiling. It is important to recognise and build on the strengths of the most disadvantaged or marginalised in society. They are often the main target population for interventions, and find it hardest to access and engage with services [53], thus early involvement with representatives of ‘under‐served’ groups, intermediaries and advocates is crucial to ensure acceptability.
4.3. Strengths and Limitations
There is a paucity of research investigating the effect of community‐based midwife continuity of care models on preterm birth and other outcomes among women living in areas of social disadvantage and ethnic diversity in the UK. Our study used all eligible records from the eLIXIR maternity‐child data linkage and, uniquely, controlled for potential confounders through propensity score matching. This approach has more power than conventional regression modelling when, as in this study, the number of events is low and there are seven or fewer events per confounder, as this produces less biased and more precise estimates [54]. Using linked NHS records reduced selection bias, but data quality depended on clinicians' reporting, and limitations included potential misclassification of ethnicity, underreporting of sensitive issues (e.g., mental health), and missing contextual psychosocial information, which may have led to incomplete identification of risk factors [55]. All women in CBMCOC were successfully matched in a 1:4 ratio to women in standard care (much larger cohort). We grouped more than 20 ethnic groups into four main categories to make the analysis feasible and meaningful for our research purpose. While randomised trials can perfectly balance intervention and control groups on both measured and unmeasured unknown variables, propensity score matching can only account for the measured variables that were included in the analysis; this means there is a higher chance of unknown sources of bias remaining in the analysis. Following matching score adjustment, we found no baseline differences. Propensity score matching does have limitations given how it relies on the completeness and quality of the data available (e.g., accuracy), meaning unadjusted confounders may exist due to unmeasured factors influencing maternity model allocation [56]. E‐values for key associations exceeded the strength of known confounders in our models, which ranged in adjusted RRs from 0.00 to 2.72. For example, E‐values were 3.68 for planned caesarean birth, 2.90 for low birthweight and 4.31 for mental health referrals, suggesting that unmeasured confounding would need to be stronger than any measured covariate to nullify these findings [35]. Analysing women by their assigned care model enhances real‐world applicability and reduces bias, though it cannot account for variations in intervention fidelity; this is a recognised limitation when pragmatically evaluating complex healthcare interventions. We were unable to differentiate between spontaneous and medically indicated preterm births; and this is an important distinction, as the underlying causal pathways and the potential impact of the care model may differ for each; future research should aim to capture this level of detail to better elucidate the potential mechanisms of action.
5. Conclusion
Our findings support the current policy drive to increase continuity of midwife care, and that adding community‐based care may further improve outcomes for women at increased risk of health inequalities. However, future trials should evaluate the effectiveness, implementation, scale‐up, and cost of these models to understand their real‐world impact, and how they may benefit ‘at risk’ women and babies throughout their life course by improving short‐and long‐term health outcomes and social determinants, and contributing to mitigate inequalities in maternal and newborn health in the UK.
Author Contributions
C.F.T. conceptualization, data curation, investigation, methodology, project administration, visualisation, writing (original draft), writing (review and editing). S.B. conceptualization, data curation, formal analysis, methodology, software, visualisation, writing (original draft), writing (review and editing). Z.K. conceptualization, methodology, writing (review and editing). Z.V. conceptualization, methodology, writing (review and editing). M.N. conceptualization, writing (review and editing). P.T.S. conceptualization, methodology, supervision, writing (review and editing). H.R.‐J. conceptualization, methodology, visualisation, writing (review and editing). J.S. funding acquisition, conceptualization, methodology, supervision, validation, visualisation, writing (review and editing). A.E. funding acquisition, conceptualization, methodology, supervision, visualisation, validation, writing (review and editing).
Funding
J.S., C.F.T., A.E., S.B., Z.V., Z.K. and this research was funded by the National Institute for Health Research (NIHR) Applied Research Collaboration South London (NIHR ARC South London) at King's College Hospital NHS Foundation Trust. The views expressed are those of the author[s] and not necessarily those of the NIHR or the Department of Health and Social Care. The work was supported by the Early Life Cross Linkage in Research, Born in South London (eLIXIR‐BiSL) Partnership developed by an MRC Partnership Grant (MR/P003060/1) and the MRC Longitudinal Population Study Grant (MR/X009742/1). The eLIXIR platform is also part‐supported by the National Institute for Health Research (NIHR) Biomedical Research Centre at the South London and Maudsley NHS Foundation Trust and King's College London. HRJ is funded by a NIHR Advanced Fellowship (NIHR 303183).
Ethics Statement
Ethical approval eLIXIR is granted by the Oxford Central Research Ethics Committee (23/SC/0116). The Health Research Authority Confidentiality Advisory Group (HRA CAG Ref: 18/CAG/0040) provided approval under Section 251 (s251) of the NHS Act (2006). This study was also approved by the eLIXIR Oversight Committee (RAF: DL021R).
Consent
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: bjo70101‐sup‐0001‐TableS1.docx.
Acknowledgements
We would like to thank all members of the eLIXIR‐BiSL Partnership (see below) for the constant support over the years; clinical teams at King's College Hospital NHS Foundation Trust (Rachel Gallimore, Octavia Wiseman, Michelle Harrison, Miriam McCarthy) and Guy's and St Thomas' NHS Foundation Trust (Gina Brockwell, Nina Khazaezadeh, Lia Brigante, Rachel Smith) for their advice and guidelines in the characteristics of the different continuity models including LEAP (Carla Stanke). We thank women and families from all participating sites, for sharing their data and supporting this programme, and our patient and public involvement and engagement group for their feedback and advice (Tania Sutedja, Vanathy Nathan, Tannice Hemming, Amy Dignam).
Members of the eLIXIR‐BiSL Partnership: Professor Lucilla Poston, Professor of Maternal and Fetal Health, Department of Women and Children's Health, School of Life Course and Population Sciences, King's College London; Professor Laura A. Magee, Professor of Women's Health, Department of Women and Children's Health, School of Life Course and Population Sciences, King's College London; Professor Robert Stewart, Professor of Psychiatric Epidemiology and Clinical Informatics, Department of Psychological Medicine, Institute of Psychiatry, Psychology and Neuroscience, King's College London and NIHR Maudsley Biomedical Research Centre, South London and Maudsley NHS Foundation Trust, London. Consultant Psychiatrist at South London and Maudsley NHS Foundation Trust, London; Professor David Edwards, Chair in Paediatrics and Neonatal Medicine, Department of Perinatal Imaging and Health, King's College London. Neonatal Consultant at Guy's and St. Thomas' NHS Foundation Trust; Professor Mark Ashworth, Professor of Primary Care, Department of Population Health Sciences, School of Life Course and Population Sciences, King's College London; Professor Jane Sandall, Professor of Social Science and Women's Health, Department of Women and Children's Health, School of Life Course and Population Sciences, King's College London; Dr. Ingrid Wolfe, Clinical Senior Lecturer, Department of Women and Children's Health, School of Life Course and Population Sciences, King's College London and Consultant in Children's Public Health Medicine and Director of the Evelina London Children's Healthcare; Dr. Cheryl Gillett, Head of Tissue Banking, Department of Comprehensive Cancer Centre, School of Cancer and Pharmaceutical Sciences, King's College London; Dr. Michael Absoud, Paediatric Consultant at Evelina London Children's Healthcare; Dr. Lucy Pickard, Consultant Paediatrician, King's College Hospital NHS Foundation Trust; Ms. Amanda Grey, Lay member of the eLIXIR Oversight Committee; Ms. Sarah Spring, Lay member of the eLIXIR Oversight Committee; Ms. Toyin Kazeem, Information Governance Operations Lead, South London and Maudsley NHS Foundation Trust, London; Ms. Amelia Jewell, Clinical Data Linkage Service Lead, NIHR Maudsley Biomedical Research Centre, South London and Maudsley NHS Foundation Trust; Mr. Matthew Broadbent, CRIS Clinical Informatics Lead, NIHR Maudsley Biomedical Research Centre, South London and Maudsley NHS Foundation Trust, London; Ms. Finola Higgins Research Informatics Programme Manager, Guy's and St Thomas' NHS Foundation Trust; Mr. Leonardo de Jongh, Data Warehouse Manager, Guy's and St. Thomas's Hospital NHS Foundation Trust; Ms. Tisha Dasgupta, Research Associate and eLIXIR Coordinator, Department of Women and Children's Health, School of Life Course and Population Sciences, King's College London; Dr. Carolyn Gill, School Bioresource Manager, School of Life Course and Population Sciences, King's College London.
Fernandez Turienzo C., Burton S., Khan Z., et al., “The Impact of Community‐Based Midwife Continuity of Care Models for Women Living in Areas of Social Disadvantage and Ethnic Diversity in the United Kingdom: A Prospective Cohort Study,” BJOG: An International Journal of Obstetrics & Gynaecology 133, no. 5 (2026): 958–973, 10.1111/1471-0528.70101.
Linked article: This article is commented on by Bone, pp. 974–975 in this issue. To view this article visit https://doi.org/10.1111/1471‐0528.70141.
The eLIXIR‐BiSL Partnership is listed in Acknowledgement section.
Contributor Information
Cristina Fernandez Turienzo, Email: cristina.fernandez_turienzo@kcl.ac.uk.
the eLIXIR‐BiSL Partnership:
Lucilla Poston, Laura A. Magee, Robert Stewart, David Edwards, Mark Ashworth, Jane Sandall, Ingrid Wolfe, Cheryl Gillett, Michael Absoud, Lucy Pickard, Amanda Grey, Sarah Spring, Toyin Kazeem, Amelia Jewell, Matthew Broadbent, Finola Higgins, Leonardo de Jongh, Tisha Dasgupta, and Carolyn Gill
Data Availability Statement
The data accessed by eLIXIR remains within an NHS firewall and governance is provided by the eLIXIR Oversight Committee reporting to relevant information governance clinical leads. Subject to these conditions, data access is encouraged, and those interested should contact the eLIXIR Chief Investigator (Professor Lucilla Poston; lucilla.poston@kcl.ac.uk).
References
- 1. Souza J. P., Day L. T., Rezende‐Gomes A. C., et al., “A Global Analysis of the Determinants of Maternal Health and Transitions in Maternal Mortality,” Lancet Global Health 12, no. 2 (2024): e306–e316. [DOI] [PubMed] [Google Scholar]
- 2. MBRRACE‐UK , “Maternal Mortality 2020–2022,” Data Brief, updated January 11, 2024, https://www.npeu.ox.ac.uk/mbrrace‐uk/data‐brief/maternal‐mortality‐2020‐2022.
- 3. Jardine J., Walker K., Gurol‐Urganci I., et al., “Adverse Pregnancy Outcomes Attributable to Socioeconomic and Ethnic Inequalities in England: A National Cohort Study,” Lancet 398, no. 10314 (2021): 1905–1912. [DOI] [PubMed] [Google Scholar]
- 4. Knight M., Bunch K., Kenyon S., Tuffnell D., and Kurinczuk J. J., “A National Population‐Based Cohort Study to Investigate Inequalities in Maternal Mortality in the United Kingdom, 2009–17,” Paediatric and Perinatal Epidemiology 34, no. 4 (2020): 392–398. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Knight M., Kurinczuk J. J., Spark P., and Brocklehurst P., “Inequalities in Maternal Health: National Cohort Study of Ethnic Variation in Severe Maternal Morbidities,” BMJ (Clinical Research Ed.) 338 (2009): 392–398. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. MBRRACE‐UK , “Perinatal Mortality Surveillance UK Perinatal Deaths for Births From 1 January 2021 to 31 December 2021: State of the Nation Report,” https://timms.le.ac.uk/mbrrace‐uk‐perinatal‐mortality/surveillance/.
- 7. Felker A., Patel R., Kotnis R., Kenyon S., Knight M., and on behalf of MBRRACE‐UK , eds., Saving Lives, Improving Mothers' Care Compiled Report—Lessons Learned to Inform Maternity Care From the UK and Ireland Confidential Enquiries Into Maternal Deaths and Morbidity 2020–22 (National Perinatal Epidemiology Unit, University of Oxford, 2024). [Google Scholar]
- 8. Weightman A. L., Morgan H. E., Shepherd M. A., Kitcher H., Roberts C., and Dunstan F. D., “Social Inequality and Infant Health in the UK: Systematic Review and Meta‐Analyses,” BMJ Open 2, no. 3 (2012): e000964. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Fernandez Turienzo C., Newburn M., Agyepong A., et al., “Addressing Inequities in Maternal Health Among Women Living in Communities of Social Disadvantage and Ethnic Diversity,” BMC Public Health 21 (2021): 1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Rimmer A., “Maternal Death Rate in UK Rises to Highest Level in 20 Years,” BMJ 384 (2024): q62. [DOI] [PubMed] [Google Scholar]
- 11. Khan Z., Vowles Z., Fernandez Turienzo C., et al., “Targeted Health and Social Care Interventions for Women and Infants Who Are Disproportionately Impacted by Health Inequalities in High‐Income Countries: A Systematic Review,” International Journal for Equity in Health 22, no. 1 (2023): 131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Stevenson K., Edwards S., Ogunlana K., et al., “Public Health, Policy, and Clinical Interventions to Improve Perinatal Care for Migrant Women and Infants in High‐Income Countries: A Systematic Review,” EClinicalMedicine 78 (2024): 102938. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Sorrenti S., Prasad S., Elbarbary N., et al., “Interventions to Address Disparities in Perinatal Outcomes by Ethnicity: A Systematic Review,” BJOG: An International Journal of Obstetrics and Gynaecology 132 (2025): 2009–2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Lassi Z. S. and Bhutta Z. A., “Community‐Based Intervention Packages for Reducing Maternal and Neonatal Morbidity and Mortality and Improving Neonatal Outcomes,” Cochrane Database of Systematic Reviews 2015, no. 3 (2015): CD007754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Sandall J., Fernandez Turienzo C., Devane D., et al., “Midwife‐Led Continuity Models Versus Other Models of Care for Childbearing Women,” Cochrane Database of Systematic Reviews 2024, no. 4 (2024): CD004667. [DOI] [PubMed] [Google Scholar]
- 16. NHS England , “Placed Based Models of Care to Reduce Health Inequalities” (NHS England, 2019).
- 17. World Health Organisation , “Integrated People‐Centred Care,” accessed January 9, 2025, https://www.who.int/health‐topics/integrated‐people‐centered‐care#tab=tab_1.
- 18. NHS England , “The NHS Long Term Plan: Maternity and Neonatal Services,” (2019), accessed February 23, 2024, https://www.longtermplan.nhs.uk/online‐version/chapter‐3‐further‐progress‐on‐care‐quality‐and‐outcomes/a‐strong‐start‐in‐life‐for‐children‐and‐young‐people/maternity‐and‐neonatal‐services/.
- 19. NHS England , “Core20PLUS5 (Adults)—An Approach to Reducing Healthcare Inequalities,” accessed January 9, 2025, https://www.england.nhs.uk/about/equality/equality‐hub/national‐healthcare‐inequalities‐improvement‐programme/core20plus5/.
- 20. Gillibrand S., Dumville J., Wilson P., Kletter M., and Harkness E., “Evaluation of Enhanced Midwifery Continuity of Carer,” REVAL Insights Report (2024). [DOI] [PMC free article] [PubMed]
- 21. Roebuck C., Sandall J., West R., Atherden C., Parkyn K., and Johnson O., “Impact of Midwife Continuity of Carer on Stillbirth Rate and First Feed in England,” Communications Medicine 5, no. 1 (2025): 339. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Rayment‐Jones H., Dalrymple K., Harris J., Harden A., Parslow E., and Georgi T., “Project20: Does Continuity of Care and Community‐Based Antenatal Care Improve Maternal and Neonatal Birth Outcomes for Women With Social Risk Factors? A Prospective, Observational Study,” PLoS One 16, no. 5 (2021): e0250947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Carson L. E., Azmi B., Jewell A., et al., “Cohort Profile: The eLIXIR Partnership—A Maternity–Child Data Linkage for Life Course Research in South London, UK,” BMJ Open 10 (2020): e039583. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Consumer Data Research Centre , “Index of Multiple Deprivation 2024,” accessed November 18, 2024, https://data.cdrc.ac.uk/dataset/index‐multiple‐deprivation‐imd.
- 25. Howard L. M., Ryan E. G., Trevillion K., et al., “Accuracy of the Whooley Questions and the Edinburgh Postnatal Depression Scale in Identifying Depression and Other Mental Disorders in Early Pregnancy,” British Journal of Psychiatry 212, no. 1 (2018): 50–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Department of Health , Safer Maternity Care. The National Maternity Safety Strategy—Progress and Next Steps (Department of Health, 2019). [Google Scholar]
- 27. Heymans M. W. and Twisk J. W., “Handling missing data in clinical research,” Journal of clinical epidemiology 151 (2022): 185–188. [DOI] [PubMed] [Google Scholar]
- 28. Ho D., Imai K., King G., and Stuart E., “MatchIt: Nonparametric Preprocessing for Parametric Causal Inference,” Journal of Statistical Software 42, no. 8 (2011): 1–28. [Google Scholar]
- 29. Hansen B. B. and Klopfer S. O., “Optimal Full Matching and Related Designs via Network Flows,” Journal of Computational and Graphical Statistics 15, no. 3 (2006): 609–627. [Google Scholar]
- 30. Pishgar F., Greifer N., Leyrat C., and Stuart E., “MatchThem: Matching and Weighting After Multiple Imputation,” R Journal 13 (2021): 292–305, 10.32614/RJ-2021-073. [DOI] [Google Scholar]
- 31. Ming K. and Rosenbaum P. R., “Substantial Gains in Bias Reduction From Matching With a Variable Number of Controls,” Biometrics 56, no. 1 (2000): 118–124. [DOI] [PubMed] [Google Scholar]
- 32. Chen W. C., Lu N., Wang C., et al., “Propensity Score‐Integrated Approach to Survival Analysis: Leveraging External Evidence in Single‐Arm Studies,” Journal of Biopharmaceutical Statistics 32, no. 3 (2022): 400–413. [DOI] [PubMed] [Google Scholar]
- 33. Bates D., Mächler M., Bolker B., and Walker S., “Fitting Linear Mixed‐Effects Models Using lme4,” Journal of Statistical Software 67, no. 1 (2015): 1–48. [Google Scholar]
- 34. VanderWeele T. J. and Ding P., “Sensitivity Analysis in Observational Research: Introducing the E‐Value,” Annals of Internal Medicine 167 (2017): 268–274. [DOI] [PubMed] [Google Scholar]
- 35. Chung W. T. and Chung K. C., “The Use of the E‐Value for Sensitivity Analysis,” Journal of Clinical Epidemiology 163 (2023): 92–94. [DOI] [PubMed] [Google Scholar]
- 36. Wickham H., Averick M., Bryan J., et al., “Welcome to the Tidyverse,” Journal of Open Source Software 4, no. 43 (2019): 1686. [Google Scholar]
- 37. Tierney N. and Cook D., “Expanding Tidy Data Principles to Facilitate Missing Data Exploration, Visualization and Assessment of Imputations,” Journal of Statistical Software 105, no. 7 (2023): 1–31.36798141 [Google Scholar]
- 38. Long J. A., “jtools: Analysis and Presentation of Social Scientific Data,” R Package Version 2.2.0 (2022).
- 39. Singmann H., Bolker B., Westfall J., and Aust F., “afex: Analysis of Factorial Experiments,” R Package Version 0.16‐1 (2016).
- 40. Rayment‐Jones H., Murrells T., and Sandall J., “An Investigation of the Relationship Between the Caseload Model of Midwifery for Socially Disadvantaged Women and Childbirth Outcomes Using Routine Data—A Retrospective, Observational Study,” Midwifery 31, no. 4 (2015): 409–417. [DOI] [PubMed] [Google Scholar]
- 41. Homer C. S., Leap N., Edwards N., and Sandall J., “Midwifery Continuity of Carer in an Area of High Socio‐Economic Disadvantage in London: A Retrospective Analysis of Albany Midwifery Practice Outcomes Using Routine Data (1997–2009),” Midwifery 48 (2017): 1–10. [DOI] [PubMed] [Google Scholar]
- 42. Hadebe R., Seed P. T., Essien D., et al., “Can Birth Outcome Inequality Be Reduced Using Targeted Caseload Midwifery in a Deprived Diverse Inner City Population? A Retrospective Cohort Study, London, UK,” BMJ Open 11, no. 11 (2021): e049991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Kildea S., Gao Y., Hickey S., et al., “Reducing Preterm Birth Amongst Aboriginal and Torres Strait Islander Babies: A Prospective Cohort Study, Brisbane, Australia,” EClinicalMedicine 12, no. 1 (2019): 43–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Kildea S., Gao Y., Hickey S., et al., “Effect of a Birthing on Country Service Redesign on Maternal and Neonatal Health Outcomes for First Nations Australians: A Prospective, Non‐Randomised, Interventional Trial,” Lancet Global Health 9, no. 5 (2021): e651–e659. [DOI] [PubMed] [Google Scholar]
- 45. Kayode G., Thilaganathan B., Burden C., et al., “Disparities in Stillbirths in England: Analysis of a Population‐Based Study of 1.3 Million Births,” BJOG: An International Journal of Obstetrics & Gynaecology 132, no. 8 (2025): 1130–1138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Anumba D., Soma‐Pillay P., Bianchi A., Valencia González C. M., and Jacobbson B., “FIGO Good Practice Recommendations on Optimizing Models of Care for the Prevention and Mitigation of Preterm Birth,” International Journal of Gynecology & Obstetrics 166, no. 3 (2024): 1006–1013. [DOI] [PubMed] [Google Scholar]
- 47. World Health Organization , Transitioning the Midwifery Models of Care: Global Position Paper (World Health Organization, 2024). [Google Scholar]
- 48. Rayment‐Jones H., Harris J., Harden A., Turienzo C. F., and Sandall J., “Project20: Maternity Care Mechanisms That Improve (or Exacerbate) Health Inequalities. A Realist Evaluation,” Women and Birth 36, no. 3 (2023): e314–e327. [DOI] [PubMed] [Google Scholar]
- 49. Rayment‐Jones H., Silverio S. A., Harris J., Harden A., and Sandall J., “Project 20: Midwives' Insight Into Continuity of Care Models for Women With Social Risk Factors: What Works, for Whom, in What Circumstances, and How,” Midwifery 84 (2020): 102654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Fernandez Turienzo C., Rayment‐Jones H., Roe Y., et al., “A Realist Review to Explore How Midwifery Continuity of Care May Influence Preterm Birth in Pregnant Women,” Birth 48, no. 3 (2021): 375–388. [DOI] [PubMed] [Google Scholar]
- 51. Bradford B. F., Wilson A. N., Portela A., McConville F., Fernandez Turienzo C., and Homer C. S., “Midwifery Continuity of Care: A Scoping Review of Where, How, by Whom and for Whom?,” PLOS Global Public Health 2, no. 10 (2022): e0000935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Simmelink R., Neppelenbroek E., Pouwels A., et al., “Understanding How Midwife‐Led Continuity of Care Can Be Implemented and Under What Circumstances: A Realist Review,” BMJ Open 15, no. 2 (2025): e091968. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Bonevski B., Randell M., Paul C., et al., “Reaching the Hard‐To‐Reach: A Systematic Review of Strategies for Improving Health and Medical Research With Socially Disadvantaged Groups,” BMC Medical Research Methodology 14 (2014): 1–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54. Cepeda M. S., Boston R., Farrar J. T., and Strom B. L., “Comparison of Logistic Regression Versus Propensity Score When the Number of Events Is Low and There Are Multiple Confounders,” American Journal of Epidemiology 158, no. 3 (2003): 280–287. [DOI] [PubMed] [Google Scholar]
- 55. Verma R., Au F., Green L., et al., “Understanding Challenges of Using Routinely Collected Health Data to Address Clinical Care Gaps: A Case Study in Alberta, Canada,” BMJ Open Quality 10, no. 1 (2021): e001150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. Haukoos J. S. and Lewis R. J., “The Propensity Score,” JAMA 314, no. 15 (2015): 1637–1638. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: bjo70101‐sup‐0001‐TableS1.docx.
Data Availability Statement
The data accessed by eLIXIR remains within an NHS firewall and governance is provided by the eLIXIR Oversight Committee reporting to relevant information governance clinical leads. Subject to these conditions, data access is encouraged, and those interested should contact the eLIXIR Chief Investigator (Professor Lucilla Poston; lucilla.poston@kcl.ac.uk).