Intended for healthcare professionals

Editorials

At last—maternity statistics for England

BMJ 1998; 316 doi: https://doi.org/10.1136/bmj.316.7131.566 (Published 14 February 1998) Cite this as: BMJ 1998;316:566

Some trends are apparent but the data still have too many gaps

  1. Alison Macfarlane, Reader in perinatal and public health statistics
  1. National Perinatal Epidemiology Unit, Radcliffe Infirmary, Oxford OX2 6HE

    After a long gestation the Department of Health has finally delivered a bulletin on maternity statistics.1 Few routine data on maternity care in England have been published since 1985, the last year of the old Hospital In-patient Enquiry. This lack has been equally frustrating to people who want to compare local performance with national statistics and those who want to monitor national trends. 2 3 Not surprisingly, the new statistics show many changes in maternity care since 1985, while revealing inadequacies in the data.

    The Maternity Hospital Episode Statistics system,4 through which the data are collected, started in April 1989, and the new publication contains data for 1994-5 and trends since 1989-90. It is restricted to data about care during delivery, but a further bulletin with data for 1995-6 and 1996-97, including antenatal and postnatal episodes and data about newborn babies, is promised for the autumn. Subsequent bulletins will then appear annually.

    The most dramatic changes are in caesarean section rates and the methods of operative vaginal deliveries (see figure). The caesarean section rate, which levelled off at just above 10% of deliveries in the early 1980s, rose from 11.3% in 1989-90 to 15.5% in 1994-5, mainly due to an increase in the emergency caesarean section rate from 5.5% to 9.0%. The overall level of operative vaginal deliveries changed much less, accounting for 9.7% of births in 1985 and 10.8% in 1994-5. A ventouse was used for only 0.7% of deliveries in 1985 compared with 4.8% in 1994-5, while forceps use declined from 9.1% to 5.8%. Induction rates, which had fallen since the 1970s, fluctuated between 17% and 20% of all deliveries in the early 1990s, while elective caesarean section rates rose from 4.9% in 1985 to 6.8% in 1994-5.

    The percentage of deliveries coded as “normal deliveries with no antenatal or postnatal complications” fell from 30.8% in 1989-90 to 18.5% in 1994-5. This could reflect changes in coding practice, but it is ironic that it occurred just when the Changing Childbirth initiative was emphasising that childbirth is essentially a normal process.5 Probably because of the rise in caesarean section rates, the percentage of women delivered by midwives fell from 75.6% in 1989-90 to 72.3% in 1994-5.

    Despite this increase in obstetric intervention, the length of stay has fallen. Three quarters of women delivering in 1994-5 spent fewer than four days in hospital afterwards, compared with just under half in 1985 and a third in 1975. In 1994-5, 10% left hospital on the day they gave birth and 27% the next day.

    The more detailed analyses for 1994-5 include some by gestational age and clinical parity, items that are absent from the Office for National Statistics' birth registration data. An estimated 6.2% of singleton deliveries, 46% of twin deliveries, and 79% of triplet and higher order deliveries took place before 37 weeks' gestation. Tabulations by region reveal considerable variations in obstetric practice, but the widest differences occur in the tables that give modes of onset of labour and of delivery for individual trusts.

    The most striking feature of these trust tables, however, is their many empty spaces. Records with usable data were available for only 67% of deliveries in England in 1994-5. This ranged from 22-29% of deliveries in the former Yorkshire, North East Thames, and South East Thames regions to over 90% in the former North West Thames, South West Thames, Wessex, West Midlands, and North Western regions. More worryingly, national coverage, which rose from 57% in 1989-90 to 78% in 1992-3, has actually decreased since then. The abolition of regional health authorities, which used to coordinate data collection, may have contributed to this decline.

    Figure1

    Operative delivery rates 1955 to 1994-5. Data from Maternity Hospital In-patient Enquiry and Hospital Episode Statistics

    Several other factors contribute to the gaps. In some units maternity data are collected in stand alone systems that are not linked to the hospital systems from which data are extracted for Hospital Episode Statistics. In others data collection is not computerised. A survey in December 1995 found that only 55% of maternity units in England had a computer system and that only half of the rest were considering buying one.6

    Given the poor quality of the data, the department's reluctance to publish them earlier is unsurprising. The bulletin's authors, led by Lesz Lancucki, have done much to check not only for non-response bias but also for major errors. Initially they found that 2% of mothers were coded as male, for example, and some caesareans were coded as taking place at home or without any anaesthesia.

    Maternity hospital episode statistics have other deficiencies which arise from their episode based nature. The first is the lack of linkage nationally between data on delivery and data about antenatal and postnatal care. Linkage is now becoming possible, as the new NHS number offers the potential for building up a pregnancy based record. Even then, valid comparisons cannot be made between hospital based intervention rates without data about referral patterns and the social composition of the catchment population. These data could be derived, as in Scotland, through linkage with birth and death registration data collected by the Office for National Statistics.

    None of this can happen while maternity data remain so incomplete. The bulletin takes a step in the right direction by using the available data to show important trends, while exposing the gaps to public view. These gaps must be filled before this new series of publications can fulfil its promise as a useful resource for monitoring maternity care.

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    View Abstract