Intended for healthcare professionals

Letters

Outcome of pregnancy in diabetic women

BMJ 2001; 322 doi: https://doi.org/10.1136/bmj.322.7286.614 (Published 10 March 2001) Cite this as: BMJ 2001;322:614

Authors did not define criterion for case selection

  1. Frank Johnstone, senior lecturer (fdj{at}srv1.med.ed.ac.uk)
  1. Department of Obstetrics and Gynaecology, University of Edinburgh, Centre for Reproductive Biology, Edinburgh EH3 9ET
  2. Division of Child Health, University of Bristol, Bristol BS8 1TQ
  3. Northern Diabetic Pregnancy Survey, Regional Maternity Survey Office, Newcastle upon Tyne NE2 4AA
  4. Medical Birth Registry of Norway, Hauletend Hospital, N-5021 Bergen, Norway

    EDITOR—Hawthorne et al claim to show that women with diabetes have a much more unfavourable outcome of pregnancy in England than Norway.1 But what does their study really show?

    Their criterion for case selection (“diabetes”) was not defined. But the prevalence differs hugely between the countries: 1 in 335 pregnant women in northeast England and 1 in 90 in Norway were reported to have diabetes. This suggests that selection in the two countries was based on different clinical criteria. In northeast England most women included in the northern diabetic pregnancy survey were taking insulin before pregnancy, and all cases were confirmed by the clinicians and from the patient record.2 The data from Norway are from the centralised medical birth registry, and the possible pitfalls of this are illustrated by experience in Scotland.

    I checked national registry listings of diabetes as part of the protocol for the SIGN (Scottish Intercollegiate Guidelines Network) guideline on management of pregnancy.3 Some women recorded as having diabetes did indeed have this, but some had only impaired glucose tolerance; some had had a glucose tolerance test but the result was normal; some had only a relative with diabetes; and some did not have diabetes, had not had a glucose tolerance test, and did not have a relative with the condition. Even if the diagnosis in the Norwegian registry was always reliable and was recorded before pregnancy, as stated, here is one further specific source of possible systematic error.

    Women who did not have type 1 or type 2 diabetes but who had had an abnormal result of a glucose tolerance test during a previous pregnancy may have been included as cases. The higher the proportion of women without diabetes in the case group the better the outcome will be.

    The outcome of pregnancy may be better in Norway, but this comparison looks flawed. The authors need to define the condition to be studied to ensure that data assembly includes only those cases, and thus like is compared with like.

    References

    1. 1.
    2. 2.
    3. 3.

    More investigation is needed into whether control of diabetes is really poorer in England than Norway

    1. Jean Golding, professor of paediatrics and perinatal epidemiology
    1. Department of Obstetrics and Gynaecology, University of Edinburgh, Centre for Reproductive Biology, Edinburgh EH3 9ET
    2. Division of Child Health, University of Bristol, Bristol BS8 1TQ
    3. Northern Diabetic Pregnancy Survey, Regional Maternity Survey Office, Newcastle upon Tyne NE2 4AA
    4. Medical Birth Registry of Norway, Hauletend Hospital, N-5021 Bergen, Norway

      EDITOR—Hawthorne et al compared the outcome of pregnancy in women with pregestational diabetes between northeast England and Norway.1 They note a wide discrepancy in both perinatal mortality and congenital defects, with Norway having much lower rates of adverse outcome than northeast England. They state that the registration system in Norway results in identical data being collected in the two countries.

      Apparent from the table—but not commented on by the authors—is the striking contrast in prevalence of maternal diabetes between the two countries: 0.3% (304/101 516) in northeast England and 1.1% (2019/179 754) in Norway. This raises the question as to whether this difference is a true reflection of the prevalence in the two countries. There are at least four possible alternative explanations: the Norwegian data include data for cases of gestational diabetes and the English data are for only cases of more severe uncontrolled diabetes; the Norwegian data have a higher rate of keying or editing errors (which will serve to increase the apparent numbers with a rare disorder and dilute any effect); the Norwegians do have a higher rate of diabetes in young women; or the criteria for diagnosis differ between the two countries.

      We hope that these four possibilities will be investigated before any conclusion is reached on poorer control of diabetes causing an excess of problems in the United Kingdom. We have strong reasons to think that the English data on prevalence are accurate since data from the prospective Avon longitudinal study of parents and children show a similar rate (0.4%) of pregnancies in women with non-gestational diabetes to that in Hawthorne et al's study.2

      Footnotes

      • For the ALSPAC (Avon Longitudinal Study of Parents and Children) Study Team

      References

      1. 1.
      2. 2.

      Authors' reply

      1. Gillian Hawthorne, consultant paediatrician (gillian.hawthorne{at}nth.northy.nhs.uk),
      2. L M Irgens, professor,
      3. R T Lie, professor of medical statistics
      1. Department of Obstetrics and Gynaecology, University of Edinburgh, Centre for Reproductive Biology, Edinburgh EH3 9ET
      2. Division of Child Health, University of Bristol, Bristol BS8 1TQ
      3. Northern Diabetic Pregnancy Survey, Regional Maternity Survey Office, Newcastle upon Tyne NE2 4AA
      4. Medical Birth Registry of Norway, Hauletend Hospital, N-5021 Bergen, Norway

        EDITOR—Johnstone suggests that the case selection for diabetes was not defined and that this invalidates the finding that the outcome of pregnancy is better in Norway than northeast England. He acknowledges that diabetes was confirmed in most cases from northeast England before entry into the study, but he questions the data from Norway. Golding et al draw attention to the contrast in prevalence of maternal diabetes between the two countries and suggest possible explanations for this.

        We accept that there are pitfalls in using centralised data. However, data collection by the medical birth registry in Norway has documented a decline in perinatal mortality for diabetic pregnancy, from 155.1/1000 for 1967–72 to 18.1/1000 for 1986-92.1 During this time the number of births delivered in hospitals with more than 3000 births a year has increased from less than 10% to 34%.

        During the 1990s gestational glucose intolerance was a particular focus in clinical work as well as in the registration. This has ensured the option of removing cases of gestational glucose intolerance from the analysis. The fact that there has been no secular decrease in the occurrence of macrosomia in the infants adds to the validity of the diagnoses registered by the medical birth registry.2

        There are clear clinical guidelines recommending centralisation of clinical care of diabetic pregnancy. These guidelines distinguish the management of diabetes and glucosuria during pregnancy.3 The improved mortality figures are thought to relate to an updated intensified follow up module for pregnant diabetic patients, improved diabetes care in general, home glucose monitoring, and measurement of haemoglobin A1c concentrations.

        Norway has a documented high and increasing incidence of diabetes in children4 and young people.5 In 1991 Joner and Sovik showed that there was a twofold increase in the incidence of diabetes mellitus in the 15–29 year age group.5 The finding that there is an increased prevalence of diabetic pregnancies in Norway compared with northeast England is consistent with these data.

        References

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