Methods
Study design, study population
This was a multisite cohort study in Western Sydney with enrolled women from four participating hospitals.
Pregnant women aged 18 years and over who had GDM in their first pregnancy and had an expected date of delivery from 1st March 2017 to 30th April 2019 were recruited. Recruitment occurred from 20 weeks’ gestation. Exclusion criteria included pre-existing diabetes, pre-term first pregnancy (<37 weeks), multiple birth first pregnancy, a history of more than three miscarriages, fetal anomaly or stillbirth. Women who had care in their first pregnancy in a country other than Australia were also excluded from the study.
All potentially eligible women were identified through the electronic maternity database from information provided at their booking visit. Eligible women were approached in the antenatal clinic or in hospital during the first two days postpartum at the four participating hospitals.
At participating hospitals, GDM was diagnosed on the basis of the 1998 Australasian Diabetes in Pregnancy Society criteria.19This involves a 75g oral glucose tolerance test (OGTT) with a fasting blood glucose level (BGL) of ≥5.5 mmol/L and/or a 2 h BGL ≥8.0 mmol/L. The OGTT at 1 h was also collected and results made available to clinicians, however, it is not routinely used for diagnostic purposes. All pregnant women were tested for GDM at 24–28 weeks’ gestation, and earlier if clinically indicated. Management of GDM with insulin or metformin was defined as medical treatment inclusive of dietary modifications. Standard treatment for GDM includes blood glucose monitoring, dietician review, insulin treatment if glucose targets are not met. All women are advised to maintain a similar healthy diet after the birth of their baby and have follow-up with their family doctor.
Questionnaire
An expert panel, including lactation consultants (International Board Certified Lactation Consultants, IBCLC), endocrinologists and a maternal fetal medicine specialist, designed a task-specific, four-page health and lactation questionnaire with breastfeeding intensity assessment (Supplemental S1). Questions primarily related to the participants’ previous pregnancy and were based on maternal recall. Co-design input was through a consumer trial of questionnaires, with revision from feedback, to a final questionnaire design. After revision, 92% (11/12) of consumers agreed or strongly agreed that the questions were easy to understand and ‘accurately records how I fed my baby’.
Length of breastfeeding was assessed by a question on baby’s age when all breastfeeding stopped. The Breastfeeding Length Intensity Scoring System (BLISS) was used to assess breastfeeding intensity in the first three months postpartum. The level of intensity is a determination of the amount of maternal breastfeeding inclusive of expressing, in relation to formula feeding. The highest intensity is mostly or exclusively breastfeeding, and the lowest intensity is mostly formula feeding. The scoring of intensity and determination of four levels of intensity was validated by two consultants (IBCLC) who assessed typical optimal feeding patterns and correlated the BLISS score to length of breastfeeding with a maximum score of 25. Assessment of infant feeding by BLISS occurred via maternal recall for the time periods as follows; the first week post-partum (score/6), during the first month (score/7), one to three months (score/6) and at three months (score/6). Breastfeeding only at any time point was given full marks, formula only feeding scored nil, and scores were incrementally reduced for 1–4 formula feeds a week, 1–3 per day, with four or more formula feeds per day scoring similar or equal to full formula feeding (Supplemental S1). High intensity feeding was classified as a score between 19–25, mild-moderate intensity 7–18, and a low-intensity score was defined as between 0–6.
Data Collection
Participating women completed the self-administered questionnaire. Women who utilised an interpreter for their obstetric appointment were verbally given the questionnaire with the assistance of the interpreter. Demographic information collected included self-identification of ethnicity, country of birth and length of time living in Australia if they were a migrant. Second pregnancy details were collected from the participating hospitals’ electronic records and included booking BMI, diagnosis of GDM and pregnancy OGTT results.
Statistical Analysis
Demographic and clinical variables were summarised between outcome groups, with percentages for categorical variables and median and interquartile range (IQR) for continuous variables, as all those included were not normally distributed. The relationships between length of breastfeeding and the BLISS score with the OGTT results from the subsequent pregnancy were analysed at each OGTT measurement time separately using linear regression. This was done after the relationship between both length of breastfeeding and BLISS score and the OGTT measurement was found to be different for each of the 3 OGTT measurements (interaction) were found to be different using a general linear model to account for the correlation within a woman. All relationships with the diagnosis of GDM were assessed with a general linear model with a log link and the results summarised as relative risks. Both length of breastfeeding and BLISS score were dichotomised for these models due to the relationships using the continuous variable not being linear. Both the models for OGTT and for diagnosis of GDM were fitted unadjusted and then adjusted for medical treatment of GDM in the first pregnancy, BMI as ≥30 kg/m2 or <30 kg/m2, age at current pregnancy ≥30 or <30 years old, and ethnicity as self-identified Caucasian, South Asian or other. Family history of diabetes was assessed but did not contribute to any model so was excluded. There was no imputation of missing data and no adjustment for multiple comparisons. All analyses were done with SAS 9.4 SAS Institute Inc., Cary, NC, USA.