Introduction
Pregnancy in type 1 diabetes is associated with increased complications, including pre-eclampsia, preterm birth, macrosomia, congenital malformations and stillbirths [
1‐
7]. However, few studies have explored the relationship between maternal age and pregnancy outcomes in type 1 diabetes.
In the general population, pregnancies at the extremes of reproductive age are associated with increased complications [
8‐
10]. Teenage pregnancy is associated with increased complications including low birthweight, premature birth and neonatal death [
11,
12]. Contributing factors include high levels of deprivation and unplanned pregnancies, inadequate engagement with antenatal care and high smoking rates [
11,
13]. Similar factors may adversely affect pregnancy in teenagers with type 1 diabetes. In addition, glycaemic control strongly influences pregnancy outcomes and is particularly poor in teenagers [
14,
15]. Unplanned pregnancy may reduce the opportunities to optimise glycaemic control before conception. Pregnancy outcomes may therefore be particularly poor in teenagers with diabetes. Understanding the consequences of teenage pregnancy in diabetes is important as even small numbers of adverse outcomes, such as congenital malformations, can have major consequences.
The published literature on pregnancy outcomes in teenagers with type 1 diabetes is limited. A recently published study of pregnancy outcomes in teenagers with pregestational diabetes reported that adverse outcomes are more common in teenagers with pregestational diabetes than in teenagers without diabetes, but did not compare outcomes between teenagers with diabetes and older women with diabetes [
16]. The largest published study to date comparing outcomes in teenagers and older women with type 1 diabetes included just 18 teenage pregnancies, and suggested that the children of teenagers with type 1 diabetes are at higher risk of congenital malformations [
17]. Well-designed community-based studies are required to validate this, and to provide a comprehensive description of teenage pregnancies in type 1 diabetes and their outcomes.
The Brecon Group has a near-complete (98%) register of children with type 1 diabetes diagnosed prior to age 15 years in Wales since 1995 (
n = 3289). This national community-based approach, along with minimal cross-border movement of participants, means it is representative of all individuals with childhood-onset type 1 diabetes in Wales. Linkage of the Brecon Register with other population-based datasets through the Secure Anonymised Information Linkage (SAIL) Databank has previously been used to demonstrate excess hospital admissions in children with type 1 diabetes [
18]. Linkage with national datasets within SAIL Databank facilitates a cohort study of all pregnancies in Wales within a defined period, providing a more complete record of outcomes and avoiding the potential sources of bias observed in previous studies. In particular, pregnancy rates and outcomes can be compared with those of the background maternity population. The aim of this study was to use this approach to describe the characteristics of pregnancy in teenagers with and without childhood-onset type 1 diabetes, and to compare pregnancy outcomes between teenagers and young adult women. We hypothesised that outcomes would be poorest among teenage mothers with type 1 diabetes, due to poor glycaemic control and other adverse factors.
Discussion
This study represents a large, community-based study comparing pregnancy characteristics and outcomes between teenage mothers with childhood-onset type 1 diabetes (14–19 years) and young adult women with childhood-onset type 1 diabetes (20–32 years). In our study, the proportion of teenage girls with diabetes with a pregnancy resulting in legal birth was half that of teenage girls without diabetes. Teenage pregnancy in girls and women with diabetes was not associated with social deprivation, in contrast to teenage pregnancy in those without diabetes. Obstetric outcomes were poor in teenagers with diabetes but not worse than in young adult mothers with diabetes, likely reflecting poor glycaemic control in both groups. The infants of teenagers with type 1 diabetes had an excess of hospital admissions during the first year of life.
Little is known about the frequency of teenage pregnancy in girls and women with type 1 diabetes. In the general population, although teenage pregnancy rates across Europe are falling, they remain high in many countries [
32,
33]. We report a mean annual rate of teenage pregnancies resulting in legal births of 18 per 1000 teenage girls for teenagers without diabetes during the period 1999–2013. This is comparable with data published regarding teenage pregnancy in Wales by the Office for National Statistics for England and Wales [
32]. Teenage pregnancy rates are lower in other European countries, (e.g. rates for teenage pregnancies resulting in birth in 2011 were 5 per 1000 girls and women aged 15–19 years in Denmark, 6 per 1000 in Sweden and 10 per 1000 in Spain) but higher in the USA (34 per 1000) and New Zealand (26 per 1000) [
33].
In our study, teenagers with type 1 diabetes had half the pregnancy rate resulting in legal birth seen in teenagers without diabetes. This finding has not been previously documented and the reasons behind it are unclear, but may represent the influence of parents/carers or healthcare professionals.
Teenage pregnancy is reportedly more common in deprived communities [
11,
34‐
36]. We replicated this observation in women without type 1 diabetes. However, in our study, teenage pregnancy in the context of diabetes was not associated with higher levels of deprivation than in young adults with diabetes. Since our study contained a relatively small number of teenagers with diabetes, larger studies are required to validate this finding.
In our cohort, glycaemic control around the point of conception was poor among teenagers, as reported elsewhere for non-pregnant teenagers [
14,
15]. Glycaemic control was equally poor among young adult women. As such, the mean HbA
1c for all women with type 1 diabetes in our cohort was relatively high, far from targets set by the National Institute for Health and Care Excellence [
37] and worse than reported in the UK National Pregnancy in Diabetes Audit [
38]. This may reflect the relatively young age of our cohort (mean age 22.8 years vs a median age of 30.0 years in the 2016 National Audit) [
38]. Data from the Type 1 Diabetes Exchange demonstrate that glycaemic control is typically poorest in teenagers, with significant improvements tending not to occur until a person’s late 20s [
14]. Data from the National Pregnancy in Diabetes Audit demonstrate that women who achieve HbA
1c targets during pregnancy are typically older and live in less deprived regions [
39]. Therefore, the relatively young age of our ‘young adult mothers’, as well as the fact that our cohort represents a moderately deprived community, likely account for our high mean HbA
1c levels.
Data from the National Audit demonstrate that glycaemic control typically improves as pregnancy progresses [
38]. Similarly, in our cohort, glycaemic control improved with each trimester of pregnancy. The improvement in glycaemic control was more significant in young adult mothers, and by the third trimester young adult mothers had significantly better glycaemic control than teenage mothers. While this might suggest that interventions to improve glycaemic control during pregnancy were less successful in teenagers in our cohort, the data should be interpreted cautiously due to a high proportion of missing data. Missing data likely reflects a combination of not all general practices in Wales being registered with SAIL and poor recording of HbA
1c during pregnancy in primary care in Wales. Further studies are required to provide a more comprehensive description of the relationship between maternal age, changes in HbA
1c during pregnancy and pregnancy outcomes.
Duration of diabetes is likely to confound the relationship between maternal age and pregnancy outcomes. Older women may have a longer duration of diabetes and more microvascular complications, such as nephropathy, which is particularly associated with poor pregnancy outcomes [
40,
41]. However, in our cohort, exploratory analyses revealed no significant difference in glycaemic control or pregnancy outcomes between women with a duration of diabetes of less than 10 years and those with a longer duration (ESM Table
5). Data regarding signs of nephropathy were limited, with only 59 (17.9%) mothers with diabetes having a urine albumin/creatinine ratio available for the year prior to conception; of these, 14 had a result >3 mg/mmol.
A recent report from an insurance-based cohort of 119 million people in the USA described pregnancy outcomes in 639 teenagers with pregestational diabetes, and confirmed increased pregnancy complication rates compared with teenagers without diabetes [
16]. Consistent with our findings, the researchers reported an increased risk of pre-eclampsia, preterm delivery and high birthweight [
16]. However, they did not differentiate between types of pregestational diabetes and no socioeconomic data were included. Furthermore, this study did not compare pregnancy outcomes between teenagers and older women with diabetes.
Our cohort includes three times more teenage mothers with type 1 diabetes than the largest previously published study comparing pregnancy outcomes between teenagers and older women with type 1 diabetes [
17]. Our findings differ from those of Carmody et al, who observed worse glycaemic control during pregnancy in teenagers than older mothers and an increased rate of congenital malformations [
17]. However, their study included fewer teenagers (
n = 18) and compared them with all mothers with type 1 diabetes, whereas our cohort was censored at the age of 35 years, resulting in a lower mean age for ‘older mothers’ (24.0 vs 31.0 years). It is possible that the relatively good outcomes for older mothers in Carmody et al’s cohort was influenced by their better glycaemic control [
17]. The researchers did not report socioeconomic data or rates of teenage pregnancy in type 1 diabetes.
We report an excess of hospital admissions during the first year of life among babies of teenage mothers with type 1 diabetes. It is recognised that, in the general population, babies born to teenage mothers have more hospital admissions, partly because of socioeconomic factors [
42]. However, our data show that maternal type 1 diabetes has an additive effect, conferring an increased risk of admission during the first year of life beyond that observed among the offspring of teenage mothers in the background population. Since this remained significant even after discounting admissions in the first 28 days of life, the excess of admissions cannot be entirely explained by perinatal complications. Increased morbidity after the perinatal period or differences in healthcare-seeking behaviour may be contributing factors.
It is clear that pregnancy and early infant outcomes remain poor among teenage and young adult women with type 1 diabetes in Wales, and poor glycaemic control is likely the main reason for this. To improve outcomes, a multifaceted approach combining education with the latest advances in pharmacology and technology is required. Ensuring that young women with diabetes have access to effective contraception is essential. In addition, it is important that young women can access advice regarding sexual health and pregnancy, and that early referral to specialist services occurs as soon as pregnancy is reported. Newer technologies, such as continuous glucose monitoring, may help to address uncertainty about adjusting insulin dosage in the face of changing insulin requirements during pregnancy, and translate to better obstetric outcomes [
43]. Finally, in order to reduce the excess of admissions seen among the infants of teenage mothers with diabetes, support for young mothers with diabetes should continue beyond delivery.
The main limitation of this study is the relatively small number of teenage pregnancies in girls and women with type 1 diabetes. In addition, since the Brecon cohort includes only individuals diagnosed with type 1 diabetes prior to age 15 years, the background population includes individuals with type 1 diabetes diagnosed after age 15 years and with type 2 diabetes. However, these individuals are unlikely to account for a large number of women in the background population, particularly given the age distribution of our cohort. Since our cohort contains only pregnant women up to the age of 35 years, and contains more women at the younger age of the spectrum, we have not described pregnancy outcomes in women at the older end of the reproductive-age spectrum, in whom glycaemic control and therefore pregnancy outcomes may be better. Further studies are required, containing larger numbers of teenage and older mothers, to advance our understanding of the relationship between maternal age and pregnancy outcomes in type 1 diabetes. Our cohort included only pregnancies resulting in legal births, as pregnancies resulting in miscarriage or termination are not recorded within the SAIL Databank. Therefore, while our data provide an accurate representation of pregnancies resulting in legal births, further work is required to explore whether the proportion of pregnancies conceived is similar between teenagers with and without type 1 diabetes. Finally, a high proportion of missing data for HbA1c and urine albumin/creatinine ratio results within the primary care dataset limited our ability to incorporate these into our regression model.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.