The Reproductive Risk Factors for Incontinence Study at Kaiser (RRISK) enrolled 2270 women, aged 40–78 years, between 2003 and 2008. Women were recruited from within the Kaiser Permanente Medical Care Program (KPMC) of Northern California, an integrated health maintenance organization with over 3 million members that serves approximately 25–30% of the population in the geographic area served. The design and sampling methods have been previously described.12
Briefly, random samples within age and race strata were drawn such that a roughly equal number of women would be in each 5-year age group, and the race distribution would be approximately 40% white, 20% African American, 20% Hispanic, and 20% Asian. In addition, women with diabetes were oversampled within each age and race sampling strata to reach the goal of at least 400 diabetic women.
Data for the study were collected by self-reported questionnaires and in-person interviews. Physical measurements of height and weight, which were used to calculate body mass index (BMI), were obtained by a trained research assistant. Most interviews took place either in the home of the participant (57%) or at a Kaiser Permanente Medical Care Program clinic (28%). Informed consent was obtained by phone and in writing at the time of the interview. The Institutional Review Boards at University of California, San Francisco, Kaiser Permanente, and the University of Pittsburgh approved this study.
Detailed information on lactation was collected by asking women who reported one or more live births, “Did you breastfeed any of your children for more than a month?” Women who answered, “yes,” were then asked to specify for each infant “the number that best describes your baby's age in months at the time when you stopped nursing, breastfeeding, expressing or pumping breast milk for your baby at least once a day” (duration of lactation) and “the number that best describes your baby's age in months at the time when he/she first received a bottle of formula, cow's milk, water, juice, tea, or cereal at least once a day” (duration of exclusive lactation). This information was used to calculate the average number of months a woman lactated following the birth of each of her children. In addition, we examined consistency of lactation by considering 3 groups of women: those who consistently breastfed all of their children for >1 month, those who breastfed some (but not all) of their children for >1month, and those who had never lactated.
Women were identified as having diabetes if they self-reported a history of diabetes and were currently taking insulin or oral medications at the time of the interview, had fasting glucose and/or HbA1C laboratory values taken within one year of the interview date suggestive of diabetes, or had previously been identified for inclusion in the Kaiser Permanente Diabetes Registry, which uses an algorithm based on age at diagnosis, length of time between initial diagnosis and start of insulin treatment, intervals of 3 months or longer off insulin after initiation, and obesity at diagnosis. To avoid the chance of undiagnosed type-2 diabetes, all RRISK participants who reported no diagnosis of diabetes had a fasting glucose measured, and the population of non-diabetic RRISK participants was limited to those who self-reported no diabetes history, were not taking a diabetes medication, and had a fasting glucose < 100 mg/dl.
Variables initially considered as covariates included age, parity, race, education, income, physical activity, tobacco and alcohol use, history of hysterectomy, family history of diabetes, and body mass index. However, due to the large amount of missing income data, this variable was dropped from the final multi-variable models.
Descriptive statistics were used to characterize the study population. We excluded 37 subjects who were missing data on parity and on lactation from further analyses. The remaining 2233 women, who reported one or more live births and had complete lactation data (n=1828) or had never been pregnant (n=405), were included in the following analyses. We examined the significance of differences in demographic variables by lactation history using Chi-square tests for categorical variables, Student t-test for continuous variables, and the Wilcoxon test when appropriate. We also used descriptive statistics to assess whether duration or exclusivity of lactation varied by parity.
We used multivariable models to analyze the effect on risk of developing diabetes of average duration per child of total lactation and exclusive lactation among parous women. As there was not a linear relationship between risk of diabetes and duration of either exclusive or total lactation, we categorized exclusive lactation as ≥3 months, 1 to <3 months, or never, and total duration of lactation as ≥6 months, 1 to <6 months, or never. We then examined risk of type-2 diabetes among nulliparous women and parous women who had or had not consistently breastfed their children using bivariate analyses. We used separate multivariable logistic regression models to examine these relationships while controlling for age, parity, race, education, physical activity, tobacco and alcohol use, history of hysterectomy, family history of diabetes, and body mass index. In addition, we examined whether the relationship between lactation and diabetes differed by race, or there was a differential effect by race, by entering an interaction term in all models. P ≤0.05 was considered statistically significant in all models. All analyses were performed using SAS 9.2 (SAS Institute, Cary, NC).