OUP user menu

Preconception care and the risk of congenital anomalies in the offspring of women with diabetes mellitus: a meta‐analysis

J.G. Ray, T.E. O'brien, W.S. Chan
DOI: http://dx.doi.org/10.1093/qjmed/94.8.435 435-444 First published online: 1 August 2001


Offspring of women with pregestational diabetes mellitus are at increased risk for congenital malformations, largely attributable to poor periconceptional glycaemic control. We assessed the effect of preconception care in reducing congenital malformations, in a meta‐analysis of published studies of preconception care in women with diabetes mellitus. Articles were retrieved from Medline (1970 to June 2000) and Embase (1980 to June 2000), and data abstracted by two independent reviewers. The rates and relative risks (RR) for major and minor congenital malformations were pooled from all eligible studies using a random effects model, as were early first‐trimester glycosylated haemoglobin values. In 14 cohort studies, major congenital malformations were assessed among 1192 offspring of mothers who had received preconception care, and 1459 offspring of women who had not. The pooled rate of major anomalies was lower among preconception care recipients (2.1%) than non‐recipients (6.5%) (RR 0.36, 95%CI 0.22–0.59). In nine studies, the risk for major and minor anomalies was also lower among women who received preconception care (RR 0.32, 95%CI 0.17–0.59), as were the early first‐trimester mean glycosylated haemoglobin values (pooled mean difference: 2.3%, 95%CI 2.1–2.4). Women who received preconception care were, on average, 1.8 years older than non‐recipients, and fewer smoked (19.6% vs. 30.2%). Only one study described the routine use of periconception folic acid. Out‐patient preconception care probably reduces the risk of major congenital anomalies among the offspring of women with pregestational diabetes mellitus. Because many women with diabetes neither plan their pregnancy nor achieve adequate glycaemic control before conception, strategies are needed to improve access to these programs, and to maximize those interventions associated with improved pregnancy outcome, such as smoking cessation and folic acid use.


The offspring of women with type 1 and type 2 diabetes mellitus (DM) are at increased for congenital anomalies.1–,3 Clinicians favour preconception care (PCC) and strict periconceptional glycaemic control to limit the number of congenital birth defects.4,,5 Because many women with DM fail to obtain appropriate counselling and care before pregnancy,6 it is argued that a more aggressive policy, with provision of easier access to PCC, is needed.6 In order to define the overall benefit of PCC programs, specifically in terms of a reduction in the risk for major congenital anomalies, we conducted a systematic review of all published PCC studies. We also defined the various types of PCC programs and their relative efficacy. Third, we searched for differences between women who received PCC and those who did not, to determine whether other factors might explain the relative difference in the rates of birth defects between these two groups.


Literature search

Two authors (TO and JR) independently searched Medline from 1970 to June 2000, and Embase from 1980 to June 2000, merging the following textword and MeSH headings: (‘diabetes’ or ‘diabetes mellitus’) and (‘anomalies’, ‘congenital anomalies’, ‘malformations’, ‘congenital malformations’ or ‘birth defects’) and (‘prevention’, ‘preconception’, ‘preconception care’ or ‘counseling’). Review articles and the references of all studies were examined for further potential citations. English language studies that met all of the following criteria were included: (i) participants had pregestational DM, at least 50% with type 1 DM; (ii) a PCC intervention was explicitly studied; (iii) a concurrent group of women with DM, but who did not receive PCC, was also studied; and (iv) the rate of major congenital anomalies during the index pregnancy was reported for both groups.

Data abstraction

Data were abstracted by two authors (TO and JR), and are listed in Tables 1 to 5. We collected information on the frequency of major and minor congenital anomalies, as well as major congenital anomalies alone, and their methods of detection. A major congenital anomaly was typically defined in each study as one which causes death or a serious handicap necessitating surgical correction or medical therapy.7 We also abstracted data on maternal smoking8–,10 and maternal age,11 two other known risk factors for congenital malformations, in order to explore baseline differences between women who received PCC and those who did not. Finally, we collected data on the earliest available first‐trimester glycosylated haemoglobin values in the PCC and non‐PCC groups.

View this table:
Table 1

Design and maternal characteristics from 16 studies of women with diabetes mellitus (DM) who did and did not receive preconception care (PCC)

Statistical analysis

For dichotomous data, such as the rate of major congenital anomalies, the unadjusted relative risk (RR) and its 95%CI were calculated between PCC and non‐PCC study groups, and then pooled (RRp) using a random‐effects model by DerSimonian and Laird.12 We conducted three sensitivity analyses. The first analysis excluded all retrospective studies, and the second excluded study programs that had an in‐patient phase. The third analysis only considered studies where study investigators were blinded to maternal PCC status during the assessment of congenital malformations. Because the absolute rate of malformations was low, the RRp and its 95%CI were used.13 In a post hoc analysis, the odds ratios were found to be closely approximated by the RRp (data available upon request). The crude event rates in the PCC and non‐PCC groups were also pooled using the random effects model. Heterogeneity across studies was assessed using the Breslow and Day test,14 with statistical significance set at p<0.20. Continuous data, such as mean differences in first trimester glycosylated haemoglobin values between PCC and non‐PCC groups, were pooled using an inverse variance‐weighted method.15 In those situations where the standard deviation (SD) of the mean was not provided, the SD was approximated from the average SD of all other studies, and then substituted therein. The applied statistical software programs were Meta‐Analyst 0.98816 and Metanal.17


Study and participant characteristics

A total of 154 citations were retrieved, of which eight retrospective and eight prospective cohort studies were finally included (Table 1). Their countries of origin were Europe (five studies), the UK (three studies), the US (7 studies), and Israel (one study). All but five studies were conducted at a single centre. Most study participants had type 1 DM, but three studies also included women with type 2 DM.26,29,,31 The mean age of the women who received PCC was approximately 27 years, an average of 1.8 years older (95%CI 1.3–2.2) than those who did not receive PCC (Table 1). Four studies reported an overall lower rate of current smoking among PCC recipients (pooled rate 19.6%, 95%CI 14.6–25.9) than non‐recipients (30.2%, 95%CI 20.1–42.5).25,29,30,,32

Table 2 details the various kinds of interventions in the PCC and non‐PCC groups. Three studies included an early in‐patient phase for PCC followed by out‐patient care,18,19,,22 while the remaining 13 studies focused on out‐patient care alone. There were several different approaches to the type and duration of PCC, as well as the methods used to assess the adequacy of glycaemic control. For example, although most studies provided some maternal education about the pregnancy risks associated with poor glycaemic control, several failed to define the dietary7,18,22–24,32–,34 or insulin7,23–26,28,29,,32 regimens for their participants. Only one study mentioned the use of periconception folic acid or multivitamins as a part of their program.27 For women who did not receive PCC, eight studies explicitly defined the type of care that they received after conception.7,18,19,26–28,30,33,,34 Similarly, only five studies described the gestational period at which non‐recipients were eligible for post‐conception care, which varied from a minimum of 37 to 8 weeks,18,,19 to a maximum of 928 to 14 weeks.22

The methods used to assess congenital anomalies were explicitly outlined in eight studies, and were mostly limited to a physical examination of all stillborn and liveborn infants7,20–22,24–,28 (Table 3). In the remaining eight studies, it was likely that a physical examination was performed on all newborns. Only one study described the use of ultrasonography for the antenatal detection of fetal anomalies.26 Three studies used masking between the type of PCC care that a woman received and the subsequent assessment of her offspring for a congenital anomaly7,27,,28 (Table 3).

View this table:
Table 2

Interventions from 16 studies of women with diabetes mellitus who did not receive preconception care (PCC)

View this table:
Table 3

Methods used to detect for congenital anomalies in 16 studies of women with diabetes mellitus did and did not receive preconception care (PCC)

Glycosylated haemoglobin values

Seven studies reported on the early first‐trimester HbA1 or HbA1C values in the PCC and non‐PCC groups (Table 4). In each study, the mean glycosylated haemoglobin level was lower in the PCC group, as was the pooled absolute mean difference (2.3%, 95%CI 2.1–2.4). Heterogeneity was present for this pooled estimate, however (p<0.20).

View this table:
Table 4

Early first trimester glycosylated haemoglobin (GHb) values from eight studies of women with diabetes mellitus who did and did not receive preconception care (PCC)

Congenital anomalies

Among 2104 offspring, the pooled rate for major and minor anomalies was 2.4% (95%CI 1.2–4.6) in the PCC group, compared to 7.7% (95%CI 6.3–9.4) among non‐PCC recipients (RRp 0.32, 95%CI 0.17–0.59; heterogeneity absent) (Table 5). Among 2651 offspring, the pooled rate for major malformations was lower in the PCC group (2.1%, 95%CI 1.4–3.2) than the non‐PCC group (6.5%, 95%CI 4.5–9.2) (RRp 0.36, 95%CI 0.22–0.59; heterogeneity absent) (Table 5, Figure 1).

When we considered the eight prospective studies alone (Table 5), the risk for major malformations remained significantly lower among PCC recipients than non‐recipients (RRp 0.42, 95%CI 0.24–0.74). Upon the exclusion of three in‐patient studies,18,19,,22 there was a significantly lower risk for both major (RRp 0.36, 95%CI 0.22–0.60) as well as major and minor (RRp 0.29, 95%CI 0.12–0.72) congenital anomalies among PCC recipients. Finally, when we analysed those studies in which infant examiners were unaware of the mothers’ PCC status,7,27,,28 there was a significantly lower risk for major malformations in the PCC group (RRp 0.38, 95%CI 0.17–0.87). In a post hoc analysis, the lowest risk for major anomalies was observed in the one study that described the use of periconception folic acid among its PCC recipients (RR 0.11, 95%CI 0.01–0.82).27 In that study, some PCC recipients began to receive folic acid supplements as late as 8 weeks gestation.

Figure 1.

Relative risk (RR) for major congenital abnormalities from 14 studies of women with diabetes mellitus who did or did not receive preconception care (PCC).

View this table:
Table 5

Rates and relative risk (RR) for congenital anomalies from 16 studies of women with diabetes mellitus who did and did not receive preconception care (PCC)


Principal findings

In an analysis of 16 published studies, PCC was associated with a significantly lower risk of major and minor congenital anomalies among the offspring of women with established diabetes. This was mirrored by significantly lower first‐trimester glycosylated haemoglobin values among the PCC recipients. Compared to women who received PCC, non‐recipients were younger and more likely to be cigarette smokers.

Strengths and weaknesses

Our meta‐analysis was prone to several possible sources of bias, including publication bias. Furthermore, some congenital anomalies thought to be related to DM may have been attributable to other factors, such as smoking8–,10 and heritable genetic malformations,11 neither of which were assessed nor adjusted for in our analysis. However, our pooled estimate of 6.5% for major anomalies in the non‐PCC group is probably valid, since it closely approximates those of other large epidemiological studies of DM in pregnancy.1,2,,35 Also, since women who received PCC were, on average, nearly two years older than those who did not, one might have expected a slightly higher rate of malformations in these older women,11 which was not the case. Nonetheless, the observed demographic differences between groups suggested that women who access PCC likely possess different attributes from those who do not, and that other unmeasured differences might have explained the apparent benefits of PCC.6

Another source of bias in the studies included herein may have been the systematic exclusion of certain PCC recipients at greatest risk for fetal malformations, leading to an overestimation of the apparent benefits of tight preconception glycaemic control. For example, if a PCC participant was found to have diabetic vasculopathy, then she may have postponed or avoided conception altogether,2 thereby excluding herself from further study. At the same time, a woman who did not receive PCC, and whose pregnancy was unplanned, may have unknowingly exposed her fetus to a teratogenic substance, such as ethanol. Few study investigators could account for such confounders in their analysis. However, the relatively lower first‐trimester glycosylated haemoglobin values seen in the PCC groups suggests that early glycaemic control probably had some role in the prevention of fetal malformations. Further support comes from published data that have shown a rising risk for congenital anomalies in parallel to higher periconception glycosylated haemoglobin values.1,,2 Reducing these and other sources of bias by means of a randomized clinical trial is unlikely, however, since the ethics of asking some women to forgo the potential benefits of PCC would be challenged by most at this time.

Sources of heterogeneity

We detected significant statistical heterogeneity in the pooled glycosylated haemoglobin values obtained during the first trimester of pregnancy. This was probably attributable to differences in the type of care provided between studies, as well as the use of varying assays to measure glycosylated haemoglobin at different gestational periods. The presence of heterogeneity in the pooled rate of major congenital anomalies in the non‐PCC group may be explained by differences in both the timing and degree of post‐conception care, as well as the methods used to detect for fetal and newborn malformations. Furthermore, unlike those who received preconception counselling,7,18,20,,21 non‐recipients typically were not instructed to monitor their basal body temperature, nor to obtain an early pregnancy test after a missed menstrual cycle. Thus, a greater number of early spontaneous abortions, as a consequence of more lethal congenital malformations,3 could have been missed among some of the non‐PCC recipients.

Clinical and financial considerations

Because a woman may continue to receive PCC for months or years, the financial cost of these programs has been questioned.34,,36 Upon comparing combined PCC and prenatal care vs. prenatal care alone, Elixhauser et al. found a net cost savings of $1720 (1989 US$) per hypothetical enrollee, with a desirable benefit‐cost ratio of 1.86.36 More recent estimates by Herman and colleagues, using real‐time, direct cost measures, suggested a savings of approximately $34 000 (1992 US$) per PCC recipient.34 A debate then emerges as to what constitutes the ‘best’ PCC, not only in terms of benefits to the fetus, but also related to the use of health‐care resources and maternal compliance.

Two PCC study programs conferred the lowest relative risk for major congenital anomalies.25,,27 In both, pregnancy planning and contraceptive use were encouraged, proper diet was emphasized, and insulin doses were adjusted to achieve pre‐meal glucose levels <6.0 mmol/l. Recent data confirm the benefits of insulin injections four times daily,37 the positive effect of preconception and first‐trimester glycaemic control on neonatal birth weight,38 and the safety of out‐patient DM care throughout pregnancy.39 Because such programs emphasize rigorous glycaemic control, concomitant education to improve hypoglycaemia awareness and treatment is of utmost importance. Although intermittent maternal hypoglycaemia does not appear to adversely effect the fetus,37,40,,41 neuroglycopenia and hypoglycaemia unawareness may increase the maternal risk for automobile accidents or hypoglycaemic coma during sleep.42 Practical ways to avoid such undesirable side‐effects include small insulin dose adjustments (e.g. 2 unit increases per dosing interval), ensuring that a non‐perishable glucose source is on hand, and educating the patient, her family and work colleagues about the use of a glucagon injection device.42 Furthermore, women who appear prone to hypoglycaemia should be encouraged to routinely check their blood glucose levels before driving.

This review and other studies6 have demonstrated that a substantial percentage of women with DM do not access PCC programs. For example, in the statewide California Diabetes and Pregnancy Programme, only 20% of women registered before conception.26 Women who do not receive PCC tend to have higher rates of unplanned pregnancies, are poorer and less formally educated, and often lack support from a stable partner.6,25,,33 Thus, exploring new ways of increasing PCC awareness among these women is crucial.43 This might include an annual mailed reminder to all adult females with DM aged <45 years, providing them with written materials in appropriate languages, and printing messages directly on the packages containing insulin or oral hypoglycaemic agents. Future studies might evaluate the benefits of free or low‐cost insulin, insulin injection devices and glucometer strips for women with financial problems.

Beyond differences in certain sociodemographic features, it appeared that a substantial proportion of women with DM failed to consume prenatal folic acid or multivitamins, just as in the general population.44,,45 Interestingly, we observed in a post hoc analysis that the risk for major congenital malformations was lowest in the one study that described the use of periconceptional folic acid among its PCC recipients (RR 0.11).27 Randomized clinical trials have established the role of prenatal folic acid for the prevention of neural tube defects,46 and perhaps major renal and cardiac defects,47 each of which are more common in the offspring of women with DM.1–,3 Accordingly, greater efforts are needed to facilitate preconception folic acid use in this population of women. Because the offspring of women with DM are at high risk for congenital anomalies, and since many pregnancies are unplanned, we wonder whether a universal policy of supra‐dietary folic acid supplementation48 should be encouraged for all women with DM who are of reproductive age. The higher number of smokers within the non‐PCC groups also suggests that greater focus on pre‐pregnancy smoking cessation is needed.8,,9

Future research

We believe that a large multicentre cohort study can optimally address some issues raised in this review. Specifically, the inclusion of a large number of women with pregestational DM in a prospective fashion will also enable the evaluation of other important maternal (e.g. spontaneous abortion, Caesarean delivery, toxaemia) and perinatal (e.g. preterm birth, gestationally adjusted birthweight, diseases of prematurity) outcomes.49 By adjusting for specific confounders and covariates, such as DM type and duration, maternal education, income, smoking status and folic acid use, important information about the influence of PCC on adverse pregnancy outcomes may be collected. Separately, effort should be made to study why some women with pregestational DM do not access PCC, and how we might best reduce their risk for fetal malformations once conception has occurred.


  • Address correspondence to Dr J.G. Ray. e‐mail: rayjgmcmaster.ca


View Abstract