Article Text

Download PDFPDF

Maternity care models in a remote and rural network: assessing clinical appropriateness and outcome indicators
  1. J Tucker1,
  2. A McVicar2,
  3. E Pitchforth3,
  4. J Farmer4,
  5. H Bryers5
  1. 1Dugald Baird Centre, Department of Obstetrics and Gynaecology, Division of Applied Health Sciences, School of Medicine and Dentistry, University of Aberdeen, Aberdeen Maternity Hospital, Aberdeen, UK
  2. 2Aberdeen Maternity Hospital, NHS Grampian, Aberdeen, UK
  3. 3LSE Health, London School of Economics and Political Science, London, UK
  4. 4UHI Millennium Institute, Centre for Rural Health, Inverness, UK
  5. 5Raigmore Hospital, NHS Highland, Inverness, UK
  1. Correspondence to Dr J Tucker, Dugald Baird Centre for Research on Women's Health, Department of Obstetrics and Gynaecology, University of Aberdeen, Aberdeen Maternity Hospital, Cornhill Road, Aberdeen AB25 2ZL, UK; j.s.tucker{at}abdn.ac.uk

Abstract

Background Little is known about performance of small rural maternity units, including stand-alone midwife units.

Aim To describe the proportions of women delivering locally, clinical appropriateness of model of care at delivery and outcome indicators for three rural staffing models of care.

Design Case note review.

Setting Remote and rural maternity units in NHS North of Scotland Region.

Subjects and methods 1400 deliveries to women from the catchments of eight rural units (stratified by staffing model) included those in local rural units and in associated distant referral units. Descriptive analysis examined women's risk, clinical appropriateness of model of care at delivery and outcomes aggregated by local catchment unit type and delivery unit type.

Results Local deliveries by staffing model were 31% (214/697) in midwife stand-alone units, 70% (236/336) in midwife units alongside non-obstetric medical support and 86% (317/367) in small obstetric-led units. Model of care at delivery was generally appropriate according to risk. Judged inappropriate were 3% (22/696) of women with complications delivering in midwife stand-alone units; and of referral unit deliveries, 6% (37/632) with suspected complications unconfirmed, plus 5% (31/633) discharged undelivered by referral hospital at >36 weeks' gestation. Risk profiles of catchment samples were similar, but caesarean section rates appeared lower and neonatal unit admissions higher for women from stand-alone midwife units.

Conclusions Rural women were generally referred appropriately for specialist care. These stand-alone midwife units provided intrapartum care for approximately one-third of rural women who remained without complications. Further evidence is needed about outcomes by staffing models of care.

  • Quality of care
  • pregnancy
  • rural health services
  • care networks
  • outcomes
View Full Text

Statistics from Altmetric.com

Wildeman et al1 described substantial differences in the average size of maternity units and proportions of home births between 12 developed European countries. They noted that variation in unit size within countries indicate differing network structures of maternity services, with the most highly centralised structures in Denmark, Sweden and Scotland. The organisation of perinatal care in the UK aims to achieve appropriate and efficient levels of care according to clinical need, ensuring access to acute care as required.2 The hierarchical levels of care are characterised by different staffing models, with increasing specialist medical provision for acute maternity and neonatal care offered at district general and tertiary hospitals.3 Women's care pathways include referrals up the hierarchy from small rural units to acute specialist units in networks of affiliated perinatal units that serve a defined geographical area.

Over the last 5 years, UK policy has also highlighted that medical staffing in small peripheral hospitals is difficult to sustain,2 4–8 and increasing centralisation of obstetric and neonatal services may further limit geographical access to acute intrapartum care for remote and rural populations. The National Service Framework in England and Wales, the Framework for Scotland and the Expert Group on Acute Maternity Services in Scotland (EGAMS) recommended the further development of tiered networks and supported local access to midwife-led intrapartum care for women with no complications.2 9 10 However, to date, there is scant evidence about the performance of maternity networks and conflicting reports about process and outcome indicators by unit size or staffing models,11–13 particularly for small rural and remote stand-alone midwife units. Much of the UK evidence is for preselected samples of eligible women with no complications (sometimes called “low risk”) and arises from studies of midwife units alongside obstetric units in tertiary and urban settings. We noted that evidence is unlikely to be generalisable to stand-alone midwifery units in rural or remote rural settings and that further research was required about staffing and quality issues in rural maternity services.14

This prospective study aimed to describe the proportions of women delivering locally and report on the clinical appropriateness (in relation to standards and outcome indicators) by different staffing models of maternity care in one rural NHS Scotland Region.

Methods

Setting and sample

The diverse geography of the North of Scotland is characterised by mainly small conurbations, separated by great tracts of uninhabited and often mountainous regions or sea, with relatively poor infrastructure; it covers 50% of Scotland's landmass and approximately 23% of the population. It presents particular challenges to the planning and provision of healthcare.

Rural maternity units are small, widely dispersed and cater for sparse populations, so rural women are most often served by only one type of facility in their locality.

All 12 rural maternity care units in NHS North of Scotland region were identified and characterised using routine data from our previous national scoping exercise.14 They represented geographic coverage of the region and are set in very remote rural, remote rural and accessible rural settings according to the Scottish Executive eight-fold urban/rural classification.15 No rural unit had a specialist neonatal service. Of the 12 units, stratified by EGAMS levels of care staffing models,2 10 a sample of 8 was selected: 2/2 small, remote obstetrician-led district hospitals with 200–300 deliveries per annum (non-air travel time to referral unit >2.5 h (EGAMS level 2a)), 2/3 midwife-led alongside non-obstetric community hospitals on very remote island settings with 90–150 deliveries per annum (non-air travel time to referral unit >12 h (EGAMS level 1c)) and 4/7 stand-alone midwife-only community unit with a range of 20–110 deliveries per annum (accessible rural settings, non-air travel time to referral unit 40–60 min (EGAMS level 1b)). All home births (EGAMS level 1a) associated with any of the study units were recorded, although Scottish rates of planned home births were low (<1%). Of the four excluded units, two stand-alone midwifery units were closing for deliveries, judged unsustainable because of low throughput. The remaining stand-alone midwifery unit also had fewer than 20 births per annum and the level 1c hospital fewer than 40 births per annum.

We undertook a case note review of a prospective series of women living in the catchment areas of the study remote and rural units for 10 months up to February 2005. Women's postcode was used to attribute cases to local catchment hospitals. We included all consecutive births in the eight rural study units. We also included consecutive births to rural women referred from the study units who delivered in any of the three large associated referral specialist units in the region (two district general and one tertiary teaching hospital in urban settings and outside the rural study unit catchments).

Data collection

Data abstraction forms were developed using national and regional guidelines that identified problems arising during pregnancy that would require exit from community-based low risk care and referral to acute obstetric units (table 1).10 A pilot study resulted in few changes—for example, in item wording to avoid ambiguity, providing duplicate fields for multiple births and adding one open-format question for comments. Fourteen volunteer link midwives from all 11 units (8 rural and 3 referral) attended training, and initial checks of intracoder and intercoder reliability were completed using anonymised dummy case notes. Over a total of 31 items pretest at training, intercoder reliability per item ranged between 69% and 100% (item total correct mean, 88%); by post-test on training day, intercoder reliability ranged between 68% and 100% (item total correct mean, 90% correct). Intracoder reliability was 89% on training day (pretest and post-test). Feedback was given on items with uncertainty. Retest 2 months later showed that reliability in item total mean of 94% correct. Checks about the completeness of the identified series were undertaken in each unit at site visits using units' birth registers.

Table 1

Complications that indicate requirement for consultation or change to consultant-led acute care for delivery

Analysis

Data were entered and analysed using SPSS V.12. Descriptive and comparative analyses were undertaken. The proportions of mothers by catchment hospital type who delivered in their local rural unit are presented. The overall case mix of women, by both catchment hospital type and by hospital of delivery type, are described as those women with no complications (“low risk”) and those with complications (“high-risk” women who require referral for specialist consultation or care). “Risk status” is presented at three stages in the pregnancy episode: complications at the outset of pregnancy, those arising throughout pregnancy and those arising around onset or during labour.

We tested if the process of risk assessment and staffing model of care at delivery was clinically appropriate in relation to guideline-defined conditions. One measure of quality of care in the network is the proportions of cases that could be judged as clinically inappropriate16: first, women with complications who nevertheless delivered in the rural midwife-led stand-alone units (false negatives); and second, women referred and delivered at specialist hospitals either with no confirmation of condition at arrival or who were discharged undelivered at >37 weeks' gestation and subsequently readmitted there for delivery (false positives).16

Finally, outcome indicators are described by catchment hospital type (describing outcomes for local population samples served by different types of local units) and by hospital of delivery type (describing outcomes for samples after risk assessment selection within the network). Outcome indicators described include onset of labour, mode of delivery, birth outcome and neonate admitted to neonatal unit (NNU) >48 h.

Ethical issues

This study was part of a wider evaluation of maternity services in the North of Scotland that also explored women's preferences and user costs.17 18 Submission MREC/04/017 was judged to be comparative audit/evaluation and therefore not requiring full ethical approval. Information about the on-going study was displayed in all participating units and labour wards.

Results

Response rates and completeness

All eight rural study units and the three associated referral units in NHS North of Scotland hierarchical Maternity Network agreed to participate. Of 1423 confirmed consecutive deliveries for women living in the catchments of the study units, 23 cases were excluded. Nine women had incomplete information as they were referred to hospitals outside the North of Scotland network, and a further 14 were forced referrals during a temporary closure of one rural study unit for 4 weeks. Data are available for 1400 women (98%).

Number (%) of the service areas' population delivering locally

Table 2 shows the total number of births to women from the geographical catchments served by the eight remote and rural units and the number and percentage of those who gave birth locally in those units. The percentage of women delivering locally is lower in locations with no specialist obstetricians or medical/surgical staff to deliver acute perinatal care. Whereas midwifery units alongside non-obstetric medical services (level 1c) and small obstetric-led units (level 2) provided services at intrapartum for 70% and 86%, respectively, of the women living in their rural catchments, only approximately one-third of women living in the catchments of midwife-led stand-alone units (level 1b) delivered in those units. The remaining two-thirds were referred or transferred for delivery to tertiary referral units.

Table 2

Number of cases in catchment and number (%) delivered locally, by catchment unit type

The risk profile of women by local catchment unit type

The number (%) of women with and without complications at the three stages in the pregnancy episode is shown by local catchment hospital type (figure 1). The risk profiles by catchment samples are similar by the end of the antenatal period; between 42% and 45% of women remain with no complications. Similarly, approximately one-third of women (31–36%) were identified with no complications (ie, “low risk”) at the onset and then through labour by all types of catchment unit. Notably, of 696 women with complete data from the catchments of midwife stand-alone units, 67% had no complication at outset of their pregnancy, and 44% and 36% continued with no complications throughout the pregnancy and labour, respectively. These proportions are similar to the “low-risk” profile of the total sample of 65% at outset, 44% throughout pregnancy and 34% through labour.

Figure 1

Risk profile by local catchment unit type. (Number (%) women/without complications arising.)

Risk and assessment of clinical appropriateness of model of care at birth

Analysis by model of care at delivery shows the proportion of rural women who delivered in specialist referral units for acute care and had confirmed complications was high (89%) and that most of the women who delivered in midwife-led stand-alone units (95%) appropriately had no complications (figure 2).

Figure 2

% (n) Women with/without complications by delivery unit type.

Those cases that might be judged clinically inappropriate; first, 22 of 696 (3% of catchment sample) of women had complications who nevertheless had intrapartum care at the midwife stand-alone units. Nine of those 22 were clearly noted as due to maternal preference. Thirteen arrived at the local midwife units in well-established labour, of whom 11 delivered before they could be transferred (11/203 (5%) of deliveries at midwife stand-alone units (figure 2)).

Second, by hospital of delivery and for those transferred to the referral obstetric and neonatal specialist units, 6% (37/633) of women had no confirmation of the suspected condition at arrival but nevertheless delivered at referral hospitals. A further 5% (31/633) were transferred but initially discharged undelivered at >36 weeks' gestation (a total 11% (68/633) of “false positives” (figure 2)). Overall, in this study, home deliveries were rare (13, <1%). There were three “born before arrivals”, two during attempted ambulance transfer and one in a car before reaching any unit. A third of the home deliveries and BBAs had noted complications by onset or during labour.

Midwife-led units alongside non-obstetric hospitals but with anaesthetic and surgical/general practice cover (level 1c) are of note. They cared for low-risk women and higher percentages of women with at least one noted complication. Both these units were in remote and island settings, and the complications often appeared to be intermediate—for example, post-dates resolved by membrane sweeps and spontaneous onset, or elective caesarean section on the island and endorsed by the obstetric consultant in the tertiary referral unit.

Process and outcome indicators

Table 3 shows process and outcome indicators for rural women analysed by local catchment unit type and hospital of delivery type.

Table 3

Number (%) of rural women with process and outcome indicators by (a) local catchment unit type and (b) delivery unit type

By hospital of delivery

Predictably, there are significant differences in both onset of labour and mode of delivery by hospital of delivery type. With higher levels of acute service, there are increasing rates of induction of labour, instrumental and operative deliveries, and admissions of neonates to neonatal care units for >48 h (table 3, data set in boldface), reflecting case-mix differences and acute management of increasing proportions of women with complications (figure 2).

By local catchment type

Of note are any differences in process and outcome indicators when analysed by women's local catchment type provision. Despite overall similar risk profiles of the rural catchment samples of women (figure 1), there appear to be lower percentages of caesarean sections and higher rates of neonatal admissions of >48 h from the catchment sample of midwife stand-alone units (table 3, shaded). However, considering the neonatal admission data by delivery unit type, few of those neonates were born in remote and rural catchment hospital births (∼3%) and then transferred. This indicates most care conformed to best practice with antepartum transfer and delivery in referral hospitals, and of those, 92 (15%) of neonates were admitted to NNU >48 h.

Discussion

This population-based analysis by both local catchment unit type and by delivery unit type allows insights into how the hierarchical maternity care network system is working in this rural setting. In relation to established guidelines, 34% of women appeared without complications throughout the whole of the pregnancy episode, and approximately one-third of women living in the catchment of rural stand-alone midwife-led units delivered in the local unit. The proportions of the rural catchment population cared for locally at intrapartum increased markedly with models that provided some level of acute medical services for births. The proportion delivering in these rural stand-alone midwife units was small and, linked to very low throughput, clearly has implications for sustainability. Policy alternatives may include closure, substituting stand-alone midwifery units with midwife-managed home births or merging neighbouring stand-alone midwife-only units (within travel time thresholds) to increase throughput.

Women were generally referred in a clinically appropriate way in relation to guideline recommendations. Few (3%) of women from the catchment of midwife stand-alone units had complications and delivered locally, and details show this was mainly attributable to maternal preference against advice or precipitous deliveries. Our qualitative and quantitative study findings on women's preferences in this study are reported elsewhere. They suggested that above all else, women valued safety and were willing to travel (within limits) to obtain specialist consultant-led care.18 Of the 11% “false-positive” referrals for acute care, 5% of women did not have the indication confirmed at referral and 6% experienced discharge from the referral hospital undelivered at >36 weeks' gestation. In terms of outcome indicators, analysis by hospital of delivery indicates the increase of obstetric interventions and neonatal admissions at hospitals with increasing levels of specialist service at delivery are as expected, and related to high-risk selection and resultant case mix in specialist referral centres. However, by hospital of catchment (and given the similarities in risk profile of local population catchment samples by hospital types), the suggestions from these data that there may be lower rates of elective caesarean section and higher rates of neonatal admission to NNU >48 h for women from the catchments of midwife stand-alone units are of note and require further study. It might be that midwifery support for normality in child birth may influence women's preferences or access in relation to elective caesarean sections, but the reasons for any higher rate of neonatal admissions by rural midwifery care catchment groups remain unclear. These findings require further investigation and confirmation in larger studies to see if there are reliable and significant differences in these outcomes associated with different models of care at intrapartum, such as the on-going National Birthplace Studies Programme in England.19

One recent report indicates one-third of the population of Scotland lives in rural areas.20 One strength of this study is that it explores stratified maternity units in NHS North of Scotland region network that well represents the challenges of providing acute maternity services in rural and remote rural settings, and data completeness suggests minimal selection bias because of loss to follow-up. The study region represents nearly 50% of Scotland's landmass and provides services for nearly one-fourth of Scotland's population.

Because of the small numbers and observational design of this study, we cannot exclude the possibility of random effects or confounding that might explain any apparent differences between levels of care in the outcome indicators described. Stillbirths and neonatal deaths are reported, but these are rare and not sensitive indicators in this study. The problem of very low throughput and rare events pose major methodological problems for research including smaller remote and rural hospitals. Thus, although our overall risk profiles appear similar between catchment populations, we cannot exclude the possibility that further undetected variations in case mix, variation in professional risk assessment and behaviour or increasing distance from acute maternity service might further explain any apparent differences. Although we undertook checks on reliability and completeness of the series, the quality of our data relies on the completeness and accuracy of clinical information in the records. The link audit midwives were staff members of their units and, although we cannot exclude the possibility of observer bias, they did show high levels of reliability in coding.

Our methods for judging care inappropriate are similar to those of the European Association of Perinatal Medicine.16 Our findings may be interpreted as evidence of high-quality functioning of a hierarchical network of associated units and demonstration of adherence to evidence-based and locally endorsed clinical guidelines. They compare favourably with the results of a previous study of appropriateness and costs of antepartum transfers in a managed clinical network in the USA of 15% of transfers with no maternal/fetal indication for transfer at arrival at tertiary centre.21 Beherenz et al21 also suggest that it is unrealistic to expect that un-indicated referrals and transfers will be reduced to zero, given the goal to avoid risking delivery at a site that is unable to provide appropriate acute obstetric or neonatal care.

However, it is important to highlight that “risk” is dynamic in pregnancy and lack of complications throughout is only applied retrospectively. Perception of risk may vary, but agreed identification of groups of women for whom particular models of care may be clinically appropriate is key to allow the development of an efficient and local service, especially for women with no problems. Of the women from the catchments of midwife stand-alone units, 67% had no complication at outset, and 44% and 36% continued with no complications throughout the pregnancy and labour, respectively. In the literature there are a number of reports of observed attrition of samples of women throughout a pregnancy episode due to complications arising from the outset of pregnancy, during pregnancy and at onset/during labour.

Thus, 33% of women in the catchment of midwife stand-alone units had complications from the outset of pregnancy compared with previous reports of 20–30% from other Scottish multicentre studies.22 23 Previous Dutch registration data analysis, however, estimated this outset high-risk proportion at 15%, with 85% of women first registered in the Dutch system designated “LVR1” level of care delivered by both independent midwives and general practitioners.24 Similarly, a further 21% of women in the catchment of midwife stand-alone units had complications arising during pregnancy, whereas previous reports from early studies of community-based maternity care ranged widely from 12% to 31%.25 Dutch registration data analysis noted that the proportion of women with problems arising during pregnancy to be 28%.24 Finally, a further 9% of women from the catchments of midwife stand-alone units had complications at onset/during labour—a lower rate than the 17% reported in the Dutch analysis24 or 16% transfer to consultant care in an alongside midwife-managed trial.26 Thus, the estimated proportions of women with complications at different stages of pregnancy in this study are broadly in line with previous reports except that complications arising in labour appear lower.

Clearly, the proportions will vary according to the criteria used, case mix and population studied (notably including parity), and require clear definition of the staffing and skill mix, model of care and context under consideration. It is notable that estimated fractions demonstrating “normality” and eligibility for community-based and midwife-led care will appear inflated in reports that only include the selected eligible patient group as “outset low-risk women”. The effect of excluding the initial proportion of non-eligible outset high-risk women may be to overestimate the proportions of local populations served by and judged appropriate for midwife-only care at intrapartum.

Conclusions

Comparing process and outcomes by local catchment hospital and delivery hospital types allowed exploration of quality of care and outcome indicators for geographical populations of women served by different care models. Women were mostly referred in a clinically appropriate way for acute care within this rural maternity network. Stand-alone midwife units in this study provided local intrapartum care appropriately for approximately one-third of rural women and those who remained at low risk. This proportion is small and, linked to low rural throughput, has implications for sustainability.

Despite similar overall risk profiles of catchment samples, our findings suggested lower rates of caesarean section and higher rates of neonatal admissions >48 h for those women from the catchment of the rural midwife stand-alone units. This is of concern and future research should include risk-adjusted comparisons of process and outcome indicators for models of care by staffing models and travel time to acute care. Further evidence is required on transfer safety, impact on women and risk-adjusted outcomes by different models of care.

Acknowledgments

We thank the staff and community members who gave their time to provide information for this study. We are grateful to link midwives for leading the review in their units and the midwives who abstracted case note data. The views expressed in this report are those of the research project team. This paper reports part of a project funded by the RARARI Project Board for Maternity Services, NHS Scotland: Brenda Thorpe, Tina Lavendar, Margaret McGuire, Fiona Dagge-Bell, Monica Thompson, Ian Bashford, Malcolm Wright and Malcolm Alexander. We thank the Board for their advice and support throughout the study. Ethical application MREC/04/0/017.

References

View Abstract

Footnotes

  • Funding RARARI Project Board for Maternity Services, NHS Scotland.

  • Competing interests None.

  • Ethics approval The North of SCotland MREC (NoS REC) reviewed the application but judged the anonymised case note review study did not require ethical committee opinion.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.