Outcomes of pregnancies in women with type 1 diabetes in Scotland: a national population-based study

Authors


* Dr G. Penney, Scottish Programme for Clinical Effectiveness in Reproductive Health, Aberdeen Maternity Hospital, Office 66, Cornhill Road, Aberdeen AB25 2ZD, UK.

Abstract

Objective To determine the outcomes of pregnancies in women with pre-existing, type 1 diabetes.

Design Prospective, population-based cohort.

Setting Scotland.

Population All 273 women with type 1 diabetes with a pregnancy ending (in miscarriage, abortion or delivery) during the 12 months (from April 1, 1998 to March 31, 1999).

Methods Pregnancies identified prospectively by clinicians in each hospital; outcome data collected from case records and from Scottish national data sets.

Main outcome measures Perinatal and infant mortality, congenital anomaly and birthweight.

Results Of the 273 pregnancies, 40 (14.7%) ended in miscarriage, 20 (7.3%) in abortion and 213 (78%) in delivery. Three deliveries were twin births, thus 216 babies were born. Stillbirth rate (4/216): 18.5 (95% CI 5.1–46.8) per 1000 total births; perinatal mortality rate (6/216): 27.8 (95% CI 10.2–59.4) per 1000 births. There were 13 verified congenital anomalies (in six abortions and seven live births), anomaly rate: 60 (95% CI 32–101) per 1000 total births. Among 208 singleton, live born infants, the mean birthweight was 3427 g. Standardised birthweight scores, relative to a reference population, showed a unimodal distribution, shifted to the right (mean, 1.57 SD).

Conclusions In an unselected population, adverse outcomes remain more common among the infants of mothers with type 1 diabetes than in the general population. The targets of the St Vincent Declaration of 1989 have not been met. Improvements may be gained by increases in provision of prepregnancy care and in the proportion of pregnancies that are planned. However, further research is needed to clarify the root causes of adverse outcomes in the pregnancies of women with diabetes.

Introduction

In 1989, the St Vincent Declaration included, as a five-year goal ‘the outcome of diabetic pregnancy should approximate that of the non-diabetic pregnancy’1. This statement stimulated the Scottish Intercollegiate Guidelines Network (SIGN) to develop its ninth evidence-based clinical guideline, published in 1996, on the topic, Management of Diabetes in Pregnancy2. During 1998–1999, the Scottish Programme for Clinical Effectiveness in Reproductive Health (SPCERH) undertook a national audit to assess the extent to which the recommendations and targets in the ninth SIGN guideline were being met in Scottish maternity services. Audit findings relating to organisation of services and to the quality of care provided have been reported previously3. These findings indicated that antenatal and delivery care for women with diabetes were organised in line with guideline recommendations—services in each unit were provided by a multidisciplinary team led by a designated consultant obstetrician and physician. In addition, antenatal surveillance and intrapartum care practices matched up well against the SIGN recommendations. However, arrangements for formal prepregnancy care were less well organised and the majority of pregnancies appeared to be unplanned.

In this article, we present data on the outcomes of the pregnancies in women with diabetes that were identified and studied during the national audit.

Methods

For a one-year period (1.4.98 to 31.3.99), volunteer diabetes nurse specialists or midwives working in each of Scotland's 22 consultant-led maternity units prospectively identified all clinically recognised pregnancies (whether progressing to delivery, abortion or miscarriage) among women with type 1, insulin-dependent diabetes that pre-existed prior to pregnancy. At the end of each pregnancy, a comprehensive data form was completed by the nurse/midwife by extraction of relevant information from the hospital case records. Data items included pregnancy outcome (miscarriage, abortion or delivery) and, for deliveries, live birth/stillbirth status, birthweight, details of any apparent congenital anomalies present at birth and infant deaths occurring prior to discharge from hospital. Further details of data extraction and processing are provided in our previous article3.

Apparent congenital anomalies recorded by the volunteer nurses/midwives were verified by consultant medical staff by reference to the infants' hospital records approximately six months after birth. Stillbirths and infant deaths were verified by reference to the Scottish Stillbirth and Infant Death data routinely collected by the Information and Statistics Division (ISD) of the Scottish NHS. This data source also permitted identification of infant deaths occurring after discharge from hospital. Information on congenital anomalies not apparent at birth, but becoming apparent later during the first year of life, was sought using ISD's Scottish Morbidity Record (SMR01) system. Admissions to hospital during the first year of life for babies in the audit data set were identified from SMR01. The ICD-10-coded reasons for these admissions were scrutinised to identify anything that might relate to previously unrecognised anomalies. Any apparent anomalies were verified by reference to the hospital record by consultant medical staff, as before.

For each live born baby, a standardised birthweight score was calculated. Scores were based on published birthweight standards for a Scottish (Aberdeen) population4 and served to correct for gestational age, sex of baby and parity of the mother. Standardised birthweight scores could not be calculated for five babies born at less than 32 weeks of gestation, due to the limits of the published data tables.

Results

A total of 273 pregnancies in women with pre-existing, type 1, insulin-dependent diabetes that ended during the 12-month audit period were identified. (This total represents an update on the 268 pregnancies described in our previous report3, following receipt of five delayed data collection forms.) Of these pregnancies, 213 (78%) progressed to delivery of a live or stillborn infant while the remainder ended in miscarriage (40, 14.7%) or induced abortion (20, 7.3%). In 1999, there were 56,551 total births (live and stillbirth) in Scotland. Using this denominator, the rate of ‘diabetic pregnancy’ (213/56,551) was 3.8 per 1000 or 1 in every 265 deliveries.

The 213 deliveries comprised 210 singleton births and three twin births, thus 216 babies were included in the study. Of these babies, four were still stillborn, two died in the early neonatal period (first week of life) and there was one postneonatal infant death (between four weeks and one year of age). Mortality rates for babies in the audit series, for the overall Scottish population (1999)5 and for other published UK series of diabetic pregnancies are summarised in Table 1.

Table 1.  Mortality rates for Scottish babies born of mothers with diabetes (n= 216), all Scottish babies (1999) and comparable published series. Values given are mortality rates (95% CI).
SeriesStillbirth ratePerinatal mortality rate (per 1000 total births)Infant mortality rate (per 1000 live births)
Scottish national audit (1998–1999) (n= 216)18.5 (95% CI 5.1–46.8)27.8 (95% CI 10.2–59.4)14.2 (95% CI 2.9–40.8)
All Scottish births (1999) (n= 55,433)5.27.65.0
Northwest England (1990–1994) (n= 360)625.036.119.9
North England (1994) (n= 104)719.248.058.8

Six pregnancies were terminated because of antenatally detected fetal anomalies (three central nervous system, two chromosomal and one cardiac). The remaining 14 terminations were undertaken on socio-economic grounds. Thirteen apparent congenital anomalies were noted at birth. Of these, seven were verified by review of case records. The remaining six ‘anomalies’ comprised ‘clicking hips’ or functional cardiac murmurs, which were found, at follow up, to be of no significance. Review of the national SMR01 data set identified a total of 123 admissions in the first year of life among 49 of the babies in the audit series. Most admissions related to respiratory tract infections or minor accidents. Two babies with congenital anomalies apparent at birth each had multiple admissions. One, with aortic stenosis, had eight admissions and the other, with Hirschsprung's disease, had 16 admissions—all before the age of one year. However, the review of SMR01 admissions revealed no additional congenital anomalies that were not already noted at birth. The types of the seven confirmed anomalies in live born babies are summarised in Table 2. Thus, there was a total of 13 confirmed anomalies among fetuses and infants in the audit data set comprising six abortions for known anomalies and seven verified anomalies present at birth. The anomaly rate, using this numerator, was 13/216 total births (60 per 1000, 95% CI 32–101).

Table 2.  Verified congenital anomalies identified in infants of mothers with diabetes.
Sacral agenesis, neurogenic bladder and bilateral talipes
Ambiguous genitalia and partial androgen insensitivity
Hirschsprung's disease
Postaxial polydactyly
Intramuscular ventricular septal defect
Ventricular septal defect
Aortic stenosis and mitral regurgitation

Birthweight was documented for all 208 singleton, live born infants in the audit series. The mean birthweight was 3427 g and the median was 3440 g (range 570–5190 g). Eighty six percent weighed above the 50th centile and 55% weighed above the 90th centile for babies in a Scottish population (Aberdeen). The mean standardised birthweight score (z score) was 1.57 and the median was 1.36 (range −2.62 to 8.18). Figure 1 shows the distribution of standardised birthweight scores for the 203 babies for which scores could be calculated. The distribution is unimodal, approximately normal and markedly shifted to the right compared with the reference population. Rather than being distributed about a mean of zero (the predicted mean from the reference population data), the z scores are distributed about a mean of 1.57 standard deviations above the reference population mean.

Figure 1.

Distribution of standardised birthweight scores (n= 203).

Discussion

In this one-year audit series, pregnancies in women with diabetes were identified prospectively by members of the clinical teams caring for them. To our knowledge, it represents a complete series from the total Scottish population. We are unaware of any ascertainment biases. The women included represent a relatively homogeneous group—they all had pre-existing, insulin-dependent, type 1 diabetes.

Mortality rates were higher than those for the background population, showing a similar pattern to rates reported in other UK population series6,7.

The anomaly rate of 60 per 1000 births described in this article was calculated using a numerator based on live births, stillbirths and abortions and a denominator of total births. The Greater Glasgow Health Board's congenital anomalies register routinely calculates anomaly rates for its population area on the same basis and contributes data to the European Registration of Congenital Anomalies (EUROCAT)8. The published Glasgow rate for 1990–1996 was 28 per 1000 births. Casson et al.6 published an anomaly rate, calculated on a similar basis, of 97 per 1000, in their northwest England study. The equivalent anomaly rate calculated from the data of Hawthorne et al.7 in their northern England study was 86 per 1000 (9/104). Our own findings mirror those of the other UK studies in demonstrating an anomaly rate in excess of that for the cited comparison group.

Similar to our own findings, Hawthorne et al.7 found that a disproportionate number of babies weighed above the population 95th centile. The finding of Casson et al.6 that only 13.7% of their study sample weighed less than the median birthweight for gestational age of the reference population is very similar to our own figure of 14%. These authors expressed birthweights in terms of z scores in a similar manner to us. Our mean score of 1.57 is broadly similar to their value of 1.36 for infants without congenital malformations. Johnstone et al.9 have published birthweight data for an Edinburgh cohort of 302 infants born to mothers with type 1 diabetes (1975–1992). They found that only 17% weighed <50th centile with a mean z score of 1.26. Again, these findings resemble our own.

Our findings mirror those of other UK series studied over the past decade and show that, among unselected populations of pregnant women with type 1 diabetes, adverse pregnancy outcomes remain higher than in the general population. Evidently, the St Vincent Declaration goal for pregnancy1 was not being met in Scotland at the end of the 20th century.

The outcome data presented here were collected alongside clinical audit data that served to monitor the clinical management of these women in relation to the recommendations in a national evidence-based guideline2. The poor pregnancy outcomes and high rate of macrosomia demonstrated in this study are particularly disappointing as the audit indicated that organisation of care and the monitoring and control of diabetes during pregnancy matched well with national recommendations3. However, the audit also demonstrated that arrangements for formal prepregnancy care are generally poor and that less than half of the pregnancies were reported as being ‘planned’3. Our findings suggest that exemplary care and diabetic control during pregnancy cannot compensate for adverse glycaemic status around the time of conception. Alternatively, perhaps poor pregnancy outcomes are related to the diabetic condition in some fundamental way that cannot be influenced by clinical care.

SIGN has recently published updated recommendations for the care of pregnant women with diabetes within its 55th guideline10 and we plan to re-audit this topic nationally during 2002–2003. We await with interest the 21st century information on the provision of prepregnancy care, attainment of planned pregnancies and any associated changes in pregnancy outcomes.

Conclusion

In an unselected population, adverse outcomes remain more common among the infants of mothers with type 1 diabetes than in the general population. The targets of the St Vincent Declaration of 1989 have not been met. Improvements may be gained by increases in provision of prepregnancy care and in the proportion of pregnancies that are planned. However, further research is needed to clarify the root causes of adverse outcomes in the pregnancies of women with diabetes.

Contribution

Gillian Penney and Donald Pearson were responsible for the conception and design of the study. Gillian Penney drafted the article. Grant Mair undertook data analysis. All three authors contributed to the interpretation of data, revision of the article and approval of the final version.

Acknowledgements

The Scottish Programme for Clinical Effectiveness in Reproductive Health (SPCERH) and the Scottish Diabetes in Pregnancy Audit were funded by the Clinical Resource and Audit Group of the Scottish Executive Health Department.

The authors thank Gordon Lang for participation in the design and conduct of the audit, Jane Carmichael for data processing and Jim Chalmers and colleagues at ISD for access to routinely collected data. The following members of the Scottish Diabetes in Pregnancy Group co-ordinated the audit and collected data in individual maternity units:

Ancillary