The authors have declared that no competing interests exist.
Conceived and designed the experiments: KM SB. Performed the experiments: KM SB. Analyzed the data: KM SB SZ. Contributed reagents/materials/analysis tools: KM SB MR MA SZ AK. Wrote the paper: KM SB SP RH.
This study examines the effect of diabetes in pregnancy on offspring weight at birth and ages 1 and 5 years.
A population-based electronic cohort study using routinely collected linked healthcare data. Electronic medical records provided maternal diabetes status and offspring weight at birth and ages 1 and 5 years (n = 147,773 mother child pairs). Logistic regression models were used to obtain odds ratios to describe the association between maternal diabetes status and offspring size, adjusted for maternal pre-pregnancy weight, age and smoking status.
We identified 1,250 (0.9%) pregnancies with existing diabetes (27.8% with type 1 diabetes), 1,358 with gestational diabetes (0.9%) and 635 (0.4%) who developed diabetes post-pregnancy. Children whose mothers had existing diabetes were less likely to be large at 12 months (OR: 0.7 (95%CI: 0.6, 0.8)) than those without diabetes. Maternal diabetes was associated with high weight at age 5 years in children whose mothers had a high pre-pregnancy weight tertile (gestational diabetes, (OR:2.1 (95%CI:1.25–3.6)), existing diabetes (OR:1.3 (95%CI:1.0 to 1.6)).
The prevention of childhood obesity should focus on mothers with diabetes with a high maternal pre-pregnancy weight. We found little evidence that diabetes in pregnancy leads to long term obesity ‘programming’.
Diabetes in pregnancy is linked to complications in 2–5% of all pregnancies
Investigations into whether these adverse effects persist into later infancy and childhood are scarce and show inconsistent findings. Previous studies focusing on the Pima Indian population have produced strong findings of an independent effect of exposure to maternal diabetes on offspring obesity
We carried out a large-scale record linked database analysis of health data to investigate the association between existing diabetes, gestational diabetes and maternal diabetes developed post pregnancy (a marker for exposure to poor lifestyle factors rather than diabetes) and offspring being large (above the 90th centile) at birth, age 12 and age 60 months. A second aim was to examine whether any observed association was independent of maternal pre-pregnancy BMI, age and smoking status.
The study population consisted of all women with GP (General Practitioner) data (registered longer than 12 month period) and a corresponding pregnancy available in the National Community Child Health Database (NCCHD). Due to the nature of our study design ethical approval and participant consent was waived by the approving IRB (Institutional Review Board), the Information Governance Review Panel (IGRP). Mother-child pairs were identified using delivery records on the NCCHD, a resource encompassing child health data from Local Health boards throughout Wales. Mothers of twins were included in the study as two separate mother-child pairs (Mum –Twin 1 and Mum- Twin 2). Mothers were also duplicated if more than one child was born throughout the designated time frame. To ensure we had a complete record of the pregnancy and birth period, only records where the mother was registered with a GP for the 12 month period before the birth were selected. The use of electronic data to identify diabetes diagnosis has been examined by others
Offspring weight and maternal diabetes status were examined using linkages within the Secure Anonymised Information Linkage (SAIL) databank
All statistical analyses were performed using STATA version 12 (STATA, Texas, USA). Differences in distributions of baseline characteristics (maternal and offspring) in relation to maternal diabetes status were analysed using
After reporting on complete-case data we conducted sensitivity analysis by imputing missing data for mother weight and weight of the child at age 5 years through a process of multiple imputations. Mother-child pairs with missing data were more likely to be deprived whilst some missing at random depend on data captures of weight at school entry and school rather than any difference between child weights. In our analysis, the multiple sets of point estimates for previous weights and their standard errors were combined to obtain a single point estimate. The MICE (Multivariate imputation by chained equations
There were 408,461 individuals born in Wales between the years 2000 and 2012 of which 160,887 (39.2%) had mothers with GP records for longer than a 12 month period around the child birth date. Those with high glucose values suggestive of hyperglycaemia that did not develop diabetes post pregnancy were removed and the remaining 147,773 women were analysed in this prospective longitudinal study (see
The average age of mothers at time of delivery was 28.3 years (range 12.6 to 51 years, see
No diabetes | Existing diabetes | Gestational diabetes | Diabetes developed post pregnancy | ||
N | 147,773 | 144,530 | 1,250 | 1,358 | 635 |
Age at delivery (years) | 147,773 | 28.3±6.1 | 30.5±6.1 | 31.7±5.8 | 29.5±6.2 |
WIMD score* | 139,714 | ||||
1 (most affluent) | 18.4(25,070) | 16.0(190) | 19.5(251) | 13.1(80) | |
2 | 18.8(25,760) | 16.7(199) | 19.2(247) | 13.6(83) | |
3 | 19.5(26,622) | 20.9(247) | 20.3(261) | 20(122) | |
4 | 21.5(29,439) | 21.1(251) | 20.5(263) | 24.9(152) | |
5 (most deprived) | 21.8(29,741) | 25.3(300) | 20.5(263) | 28.4(173) | |
Mulitparity (≥2 partus) | 147,773 | 76.3(110,241) | 71.8(898) | 71.2(967) | 81.6(518) |
Smoking during pregnancy | 73,864 | 25.7(19,014) | 35.3(230) | 30.3(178) | 22.2(82) |
Pregestational weight (kg) | 91,101 | 68.2±16.2 | 79.8±21.8 | 82.2±22.5 | 86.6±22.1 |
Pre-pregnancy BMI>24.9 (kg/m2) | 85,056 | 41.6 (35,125) | 65.3 (589) | 71.9 (655) | 81.8 (240) |
Male sex | 147,773 | 52(64,463) | 51 (633) | 52(709) | 51(325) |
Birth weight (kg) | 147,773 | 3.4±0.6 | 3.4±0.8 | 3.5±0.6 | 3.5±0.7 |
Gestational age (wks) | 142,596 | 39.3±2.2 | 37.0±2.9 | 38.4±1.9 | 38.4±2.7 |
Preterm delivery (<37 weeks) | 142,596 | 7.4 (10,412) | 31.2(380) | 10.7(143) | 13.9(84) |
Results are expressed as mean±SD or %(number) *Welsh Index of Multiple Deprivation (overall deprivation rank in Wales).
Children were more likely to be born large if their mother had diabetes (mothers with existing diabetes (adjusted OR:2.96 (95%CI 32.7 to 38.3)), gestational diabetes (adjusted OR:1.82 (95%CI 25.1 to 30.0)) or if their mother developed diabetes post pregnancy (adjusted OR:1.42 (95%CI 23.6 to 30.1))) (See
No diabetes (95%CI) | Existing diabetes (95%CI) | Gestational diabetes (95%CI) | Diabetes developed post pregnancy (95%CI) | |
N | 144,530 | 1,250 | 1,358 | 635 |
Small for gestational age | 2.13% (3,004) | 1.88% (21) | 1.07% (14) | 2.18% (13) |
(2.1 to 2.2) | (1.2 to 2.8) | (0.6 to 1.8)* | (1.28 to 3.69) | |
Large for gestational age | 12.6% (17,718/140,184) | 35.5% (396/1117) | 27.4% (360/1312) | 27.0% (161/596) |
(12.5 to 12.8) | (32.7 to 38.3)* | (25.1 to 30.0)* | (23.6 to 30.1)* | |
N | 97,692 | 838 | 873 | 406 |
>90th centile by age 12 months | 20.6% (20,121) | 15.2% (127) | 20.6% (180) | 20.0% (81) |
(20.4 to 20.9) | (12.9 to 17.8)* | (18.1 to 23.4) | (16.4 to 24.1) | |
N | 54,269 | 465 | 321 | 306 |
>90th centile at age 60 months | 14.9% (8079) | 17.6% (82) | 24.6% (79) | 16.3% (50) |
(14.6 to 15.2) | (14.4 to 21.4) | (20.2 to 29.6)* | (12.6 to 20.9) | |
N | 81,302 | 753 | 561 | 482 |
>90th centile at age 60 months including imputed data | 17.2% (13,997) | 23.2% (175) | 25.5% (143) | 25.9% (125) |
(17.0 to 17.5) | (20.4 to 26.4)* | (22.1 to 29.3)* | (22.2 to 30.0)* |
Risk factor | Large for gestation | Large at age 12 months | Large at age 60 months | |||
Odd ratio | 95%CI | Odd ratio | 95%CI | Odd ratio | 95%CI | |
Pre-pregnancy weight (per kg increase in weight) | 1.02 | 1.0–1.0 | 1.01 | 1.0–1.0 | 1.02 | 1.0–1.0 |
Pre-existing diabetes | 2.96 | 0.62 | 1.03 | 0.7–1.5 | ||
Gestational diabetes | 1.82 | 0.93 | 0.7–1.2 | 1.10 | 0.6–1.9 | |
Diabetes after pregnancy | 1.42 | 0.49 | 0.57 | 0.3–1.1 | ||
Mothers age at delivery | 1.02 | 1.0–1.0 | 0.99 | 1.0–1.0 | 1.00 | 1.0–1.0 |
Smoking in pregnancy | 0.79 | 0.94 | 0.95 |
Prepregnancy weight | Large child % (n) | Non large child % (n) | OR (95%CI) | |
Lower tertile | Gestational | 15.3 (9) | 84.7 (50) | |
No diabetes | 8.7 (831) | 91.3 (8,712) | 1.89 (0.9–3.8) | |
Middle tertile | Gestational | 27.1 (16) | 72.9 (43) | |
No diabetes | 14.8 (1,457) | 85.2 (8,378) | 2.14 (1.2–3.8)* | |
Higher tertile | Gestational | 37.3 (22) | 62.7 (37) | |
No diabetes | 21.8 (2,022) | 78.2 (7,235) | 2.1 (1.25–3.6)* |
Imputation of missing values of weight at age 5 years based on previous child weight measures and maternal characteristics. Imputing missing weight values at age 5 years provided an extra 49% (n = 27,385) of mother-child pairs for analyses. Binary logistic regression revealed that children born to mothers with any form of diabetes across the life span were at a significantly increased risk of being large at age 5 years. However, analysis adjusting for confounding factors (maternal age, smoking and pre-pregnancy weight) showed no effect of diabetes on infant weight at age 5 years (
Risk factor | Large at age 60 months | |
Odd ratio | 95%CI | |
Pre-pregnancy weight (per kg increase in weight) | 1.03 | 1.0–1.0 |
Pre-existing diabetes | 1.16 | 0.9–1.5 |
Gestational diabetes | 0.87 | 0.6–1.3 |
Diabetes after pregnancy | 0.86 | 0.6– 1.3 |
Mothers age at delivery | 0.99 | 1.0–1.0 |
Smoking in pregnancy | 0.94 |
Adjusted odds ratio for large size at age 60 months using imputed values for age 60 months.
Prepregnancy weight | Large child | Non large child | OR (95%CI) | |
% (n) | % (n) | |||
Lower tertile | No diabetes | 9.2 (1,472) | 90.8 7(14,447) | 1.00 (Ref) |
Existing | 8.4 (7) | 91.6 (76) | 0.9 (0.4 to 2.0) | |
Gestational | 8.3 (5) | 91.7 (55) | 0.9 (0.4 to 2.2) | |
Post pregnancy | 12.5 (3) | 87.5 (21) | 1.4 (0.4 to 4.7) | |
Middle tertile | No diabetes | 19.1 (2,619) | 80.9 (13,682) | 1.00 (Ref) |
Existing | 16.9 (27) | 83.1 (133) | 1.1 (0.7 to 1.6) | |
Gestational | 14.5 (11) | 85.5 (65) | 0.9 (0.5 to 1.7) | |
Post pregnancy | 20.5 (8) | 79.5 (31) | 1.3 (0.6 to 2.9) | |
Higher tertile | No diabetes | 25.6 (4,093) | 74.4 (11,893) | 1.00 (Ref) |
Existing | 30.5 (94) | 69.5 (214) | 1.3 (1.0 to 1.6)* | |
Gestational | 35.2 (82) | 64.8 (151) | 1.6 (1.2 to 2.1)* | |
Post pregnancy | 28.8 (53) | 71.2 (131) | 1.2 (0.9 to 1.6) |
No diabetes (95%CI) | Existing diabetes (95%CI) | Gestational diabetes (95%CI) | Diabetes developed post Pregnancy (95%CI) | |
N | 308,070 | 1,771 | 4,442 | 2,694 |
>90th centile at age 60 months | 19.7 (60,683) | 24.8 (439) | 24.2 (1,073) | 23.3 (627) |
(19.6 to 19.8) | (22.8 to 26.9)* | (22.9 to 25.4)* | (21.7 to 24.9)* |
Imputation of weight at age 5 years and inclusion of women who were originally excluded as they did not have a full 12 month GP record (in order to accurately classify maternal diabetes category). Combining imputed values with those mother-child pairs originally excluded from our study provided a total of 316,977 mother-child pairs. There was no effect of diabetes on infant weight at age 5 years when adjusting for confounding factors (maternal age, smoking and pre-pregnancy weight) (
Risk factor | Large at age 60 months | ||
Odd ratio | 95%CI | ||
Pre-pregnancy weight (per kg increase in weight) | 1.0 | 1.0–1.0 | |
Pre-existing diabetes | 1.2 | 0.8–1.4 | |
Gestational diabetes | 0.9 | 0.7–1.2 | |
Diabetes after pregnancy | 0.8 | 0.6– 1.2 | |
Mothers age at delivery | 1.0 | 1.0–1.0 | |
Smoking in pregnancy | 0.9 | 0.9–1.0* |
Adjusted odds ratio for large size at age 60 months (sensitivity analysis).
Prepregnancy weight | Large child | Non large child | OR (95%CI) | |
% (n) | % (n) | |||
Lower tertile | No diabetes | 10.1 (2,257) | 89.9 (20,197) | 1.00 (Ref) |
Existing | 7.8 (8) | 92.2 (95) | 0.8 (0.4 to 1.6) | |
Gestational | 9.3 (13) | 90.7 (127) | 0.9 (0.5 to 1.6) | |
Post pregnancy | 8 (5) | 92 (58) | 0.8 (0.3 to 1.90 | |
Middle tertile | No diabetes | 16.7 (3,843) | 83.3 (19,137) | 1.00 (Ref) |
Existing | 17.3 (32) | 82.7 (153) | 1.0 (0.7 to 1.5) | |
Gestational | 19.6 (37) | 80.4 (152) | 1.2 (0.8 to 1.7) | |
Post pregnancy | 19.4 (13) | 80.6 (54) | 1.2 (0.7 to 2.2) | |
Higher tertile | No diabetes | 26.5 (5,828) | 73.5 (16,124) | 1.00 (Ref) |
Existing | 31 (116) | 69 (258) | 1.2 (1.0 to 1.6)* | |
Gestational | 33.2 (151) | 66.8 (304) | 1.4 (1.1 to 1.7)* | |
Post pregnancy | 26.6 (62) | 73.4 (171) | 1.0 (0.7 to 1.3) |
We found evidence that diabetes in pregnancy is associated with larger birth weight. However, mothers with existing diabetes were less likely to have a large infant at 12 months. There is limited evidence that diabetes in pregnancy leads to a large weight of the child at age 5 years as only in women in the highest pre-pregnancy weight tertile was there an association between maternal diabetes and a large child. Therefore the influence seen at birth does not last into childhood. The finding that there is an increased risk of macrosomia as a result of foetal exposure to maternal diabetes, is already supported by an extensive field of research establishing maternal glucose as a key determinant of growth and adiposity in the offspring at birth
Type 2- and gestational diabetes have been shown to arise from the same risk factors, concur the same genetic susceptibility and are thus deemed aetiologically indistinct
We found at age 1 year offspring born to mothers with existing diabetes were less likely to be large in comparison with those born to mothers without diabetes, whilst no association was evident at age 5 years using complete data. This supports Manderson and colleagues
We did observe an association between gestational diabetes and an increased risk of being large at age 5 years. Our sensitivity analyses using imputed weight also showed an association between existing diabetes and being large at age 5 years. In agreement, findings in multi-ethnic
Examining the effect of maternal pre-pregnancy weight in our study revealed those children born to mothers with gestational diabetes in the middle or highest weight tertile were significantly at risk of being large, a finding in agreement with Boerschmann et al
Our findings suggest that only when maternal diabetes is accompanied by a high pre-pregnancy weight, rather than glucose levels throughout pregnancy alone, the likelihood of a large child is affected. Consequently our study provides limited support for the Pedersen hypothesis
A major strength of the study is the large scale population of which we were able to track both retrospective and prospective health records regarding both mother and child
Secondly to increase the sensitivity and capture reliable diagnoses, a large proportion of women who had not been registered with a GP for longer than one year prior to the pregnancy were originally excluded, significantly impacting upon our sample size. Pregnancies excluded due to lack of complete GP records (GP record for less than 12 months before the birth of child) were comparable to those included in the analysis in terms of age (28.5 years vs. 28.3), deprivation (17% vs. 21% in the most deprived ranking) and birth weight (3.34 kg (SD 6.1) vs. 3.4 kg)). However, we did include these women later in the sensitivity analysis.
Thirdly, a large number of children did not have electronic records of weight at age 5. These children were shown to be comparable to those with recorded weights in terms of gender (49% vs. 48.2% female), birth weight (3.37 compared to 3.36 kg) and gestational age (39.30 vs. 39.24). However, those with missing data were more likely to be deprived (p = 0.001) and data is likely to be missing at random depending on data captures of weight at school entry/school rather than any difference in child weight. However, we did use imputed estimates of child weight.
Lastly poor documentation of fields within the electronic records prevented us from assessing the effects of potential confounders e.g. ethnicity, gestational weight gain and infant feeding regimes within our findings. We are therefore unable to rule out the effects of unmeasured confounding. Maternal BMI was only available for half of our study population and infant height measurements are not available meaning we could not look at BMI of child. Therefore our findings should be interpreted with caution.
Our findings provide insight into the complex relationship between maternal diabetes and the later risk of offspring being large. Indicating an association between maternal diabetes, pre-pregnancy weight and childhood weight, our study demonstrates the need for future research on the multi-factorial parameters surrounding maternal diabetes and offspring size.
(DOCX)
This study makes use of anonymised data held in the Secure Anonymised Information Linkage (SAIL) system, which is part of the national e-health records research infrastructure for Wales. We would like to acknowledge all the data providers who make anonymised data available for research.