Gestational Diabetes Mellitus (GDM) affects around 5% of women in Europe and is defined as glucose intolerance that is first diagnosed in pregnancy [1]. GDM increases the risk of complications for the mother and child, including a seven-fold increased risk of type 2 diabetes (T2D) among women who have had GDM compared to those without [2]. Rates of GDM have increased over recent years with one Canadian study reporting that prevalence almost doubled in a decade from 4.6% in 2006 to 8.2% in 2016 [3] and the international diabetes federation reporting that around 21 million live births were complicated by GDM in 2021 [4]. The benefits of breastfeeding for both women and their babies are well documented, and it has been suggested that there are specific benefits for women with pregnancies complicated by GDM including a reduction in the incidence of T2D [5,6,7]. There is a growing body of research to suggest that women with GDM are less likely to initiate and continue breastfeeding than those who have not had GDM [6], however findings are mixed.
Only one study has been identified in the UK that assesses frequency of breastfeeding in a small group of women with GDM and a comparison group [8], and none report the association of breastfeeding with incidence of T2D. The study by Logan et al. [8] with 86 infants did not find any significant difference in exclusive or predominant breastfeeding at 8 to 12 weeks between women with GDM and those without. This study and other non-UK studies have not adequately adjusted for potential confounders such as obesity and insulin use in pregnancy when assessing the relationship between breastfeeding and T2D in women with GDM [9]. It is important to understand these issues in the UK context, where the prevalence of breastfeeding is relatively low for developed countries [10]. In the most recent Scottish infant feeding survey for 2021/2022 66% of babies were breastfed at birth, 37% were being exclusively breastfed at 2 weeks and 32% at 8 weeks in 2021 [11]. An understanding of breastfeeding frequency among women with GDM could inform intervention development and evaluation for these women. This study aims to use routinely collected UK health care data to investigate a historical cohort to assess the frequency of breastfeeding in women with GDM and to assess how exclusive breastfeeding influences the risk of T2D among women with GDM with adjustment for potential confounders.
Methods
Design
Historical cohort study using routinely collected, anonymised health care data for the population of pregnant women in the Fife and Tayside Health Boards in Scotland, United Kingdom (UK). Data were provided by the Health Informatics Centre (HIC) of the University of Dundee who have developed the record linkage of routinely collected health care datasets. Data held by HIC are anonymised to ensure confidentiality and meet data protection legislation.
Population
The study population was all women with a diagnosis of GDM in Tayside and Fife Health Boards, Scotland, between September 1993 to May 2015 (n = 2499) and a matched comparator cohort of women giving birth during the same time period who did not have GDM (n = 2499). NHS Tayside health board has a current population of approximately 416,000 and the neighbouring NHS Fife health board has a population of approximately 374,000 [12]. A validated population-based diabetes clinical information system, SCI-DC, was used to identify women diagnosed with GDM during the study period. The original SCI-DC database for Tayside had 95% sensitivity at identifying people with diabetes but only a small subset of women with GDM were included in the validation study [13].
Data from SCI-DC were then linked to SMR02, which is the maternity inpatient and day case dataset in Scotland, to provide demographic and clinical information for the study population such as mother’s age, deprivation category, Body Mass Index (BMI) at first antenatal appointment, and insulin use. Women with serious maternal health problems (heart disease, alcohol dependence, syndrome/alcoholism, substance abuse, HIV and hepatitis B; n = 12) or neonatal death/stillbirth complications (n = 3) were excluded from the study leaving a final population of 2484 women with GDM. Women without GDM were selected from the SMR02 dataset and matched to women with GDM based on their Scottish Index of Multiple Deprivation (SIMD) quintile. SIMD is an area-based measure of relative deprivation calculated using 30 indicators across seven domains: income, employment, health, access to services, crime and housing [14]. Where possible women with GDM were also matched to those without GDM according to parity. Exact matches on parity were made for 63% of pairs of women (n = 1564) and 19% of pairs (n = 477) were matched more generally on nulliparity and parity. It was not possible to make a match on parity for 16% of pairs (n = 397) or for 2% of pairs (n = 46) who had missing parity data. The Child Health dataset provided data on breastfeeding status (exclusive breastfeeding, mixed, bottle, or other) for the first feed after birth, upon discharge following birth, at visit from the midwife (between discharge and 10 days postpartum), at the first health visitor visit (around 2 weeks postpartum) and at the 6 to 8 week visit from the health visitor. Bottle feeding in the present study refers to formula fed via bottle. Missing data on breastfeeding status in the Child Health dataset were supplemented with data from SMR02 where it was available for the first feed after birth and feeding status at discharge. Women with GDM and those without were followed up for a diagnosis of T2D using SCI-DC until a date of diagnosis of T2D was made, or until the end of the study or date of death. A diagnosis of T2D in SCI-DC was made using the World Health Organisation (WHO) criteria, but the precise glucose levels used depended upon the criteria in use at the time of diagnosis. Similarly, diagnoses of GDM were made based on clinical guidance in use at the time of the study.
Analysis
Differences in the demographic characteristics and feeding status (exclusive breastfeeding, mixed feeding, bottle, or other) of women with GDM and those without were explored using chi-square tests of independence. Multinomial logistic regression was used to estimate crude and adjusted odds ratios for breastfeeding for the whole sample with exclusive breastfeeding duration as the dependent variable and diagnosis of GDM, maternal age, BMI, parity, deprivation category and baby birthweight as predictor variables. The association between breastfeeding and development of T2D in women with GDM was assessed by univariate and multivariate Cox regression from which hazard ratios (HRs) and 95% CIs were calculated. Breastfeeding duration, maternal age, parity, deprivation category, BMI, family history of T2D were entered as independent variables, with diagnosis of T2D as the dependent variable. Statistical analyses were carried out using SPSS for Windows version 25.
Add Comment