Abstract
Aims/hypothesis:
To determine whether a history of gestational diabetes mellitus (GDM) is associated with cognitive function in midlife.
Methods:
We conducted a secondary data analysis of the prospective Nurses’ Health Study II. From 1989–2001, and then in 2009, participants reported their history of GDM. A subset participated in a cognition sub-study in 2014–2019 (wave 1) or 2018–2022 (wave 2). We included 15,906 parous participants (≥1 birth at ≥18 years) who completed a cognitive assessment and were free of cardiovascular disease, cancer, and diabetes before their first birth. The primary exposure was a history of GDM. Additionally, we studied exposure to GDM and subsequent type 2 diabetes mellitus (T2DM) (neither GDM nor T2DM, GDM only, T2DM only, and GDM followed by T2DM) and conducted mediation analysis by T2DM. The outcomes were composite z-scores measuring psychomotor speed/attention, learning/working memory, and global cognition obtained with the Cogstate brief battery. Mean differences (β and 95% confidence intervals) in cognitive function by GDM were estimated using linear regression.
Results:
The 15,906 participants were a mean of 62.0 years (SD 4.9) at cognitive assessment, and 4.7% (n=749) had a history of GDM. In models adjusted for age at cognitive assessment, race and ethnicity, education, wave of enrollment in the cognition sub-study, socioeconomic status, and pre-pregnancy characteristics, women with a history of GDM had lower performance in psychomotor speed/attention (β, −0.08; 95% confidence interval, −0.14 to −0.01) and global cognition (β, −0.06; −0.11 to −0.01) than those without a history of GDM. The lower cognitive performance in women with GDM was only partially explained by the development of T2DM.
Conclusions/interpretation:
Women with a history of GDM had poorer cognition than those without GDM. If replicated, our findings support future research on early risk modification strategies for women with a history of GDM as a potential avenue to decrease their risk of cognitive impairment.
Keywords: Gestational diabetes mellitus, cognitive function, women’s health, type 2 diabetes, dementia
Introduction
Dementia has a greater impact on women globally, with 1.7 women affected for every man.[1] This trend is likely to continue as the number of people with this condition is expected to rise from 57 million in 2019 to 153 million by 2050.[1] To address this disparity, it is crucial to identify and address the factors that contribute to the higher burden of dementia in women. Adverse pregnancy outcomes, such as gestational diabetes mellitus (GDM), are among these factors. However, the evidence to support this claim is still limited.[2, 3]
GDM is an increasingly common condition in the United States (US), affecting 2 to 10% of pregnancies yearly.[4] This condition can lead to adverse maternal and infant outcomes such as preterm delivery, cesarean section, and macrosomia.[5] Studies also suggest that GDM may affect cognitive performance during pregnancy,[6, 7] and could lead to long-term complications that can impact cognitive function. For instance, women with GDM are at higher risk of developing type 2 diabetes mellitus (T2DM),[8] cardiovascular and cerebrovascular diseases,[9] which are established risk factors for cognitive impairment and dementia.[10, 11]. The development of these conditions, particularly T2DM, which affects cognition through various pathways such as hyperglycemia-related neurotoxicity, oxidative stress, and inflammation,[12] provides a potential link between GDM and adverse cognitive outcomes. Despite the increasing prevalence of GDM and its potential association with cognitive function, only one study has been conducted on this topic. The study involved 730 US women, and the authors did not find differences in cognitive function by history of GDM.[3] However, this study was limited by a small sample of women with GDM (n=67) and an assessment of cognitive function at a relatively young age (47.7 years). To address this knowledge gap, we studied a large cohort of parous participants from the Nurses’ Health Study II (NHS II) to determine if a lifetime history of GDM was associated with cognitive function in midlife. We hypothesized that women with a history of GDM would have lower cognitive function than those without a history of this condition and that T2DM would largely explain the associations.
Methods
Participants
We studied participants from the NHS II, an ongoing cohort of 116,429 US female nurses aged 25 to 42 years at enrollment in 1989. Participants complete mailed or online questionnaires biennially.[13] From 1989–2001, and then in 2009, participants reported their history of GDM. A subset of the full cohort who had previously completed questionnaires on trauma and posttraumatic stress disorder were invited to complete an online cognitive assessment during 2014–2019 (wave 1) or 2018–2022 (wave 2).[14, 15] Details on the eligibility criteria for completing the trauma and posttraumatic stress disorder questionnaires are provided in the Appendix (electronic supplementary material [ESM] Appendix). A total of 20,282 participants had cognitive data available. We restricted the analysis to parous women (≥1 birth at ≥18 years of age) who were free of cardiovascular disease (CVD), including coronary heart disease and stroke, cancer, and T2DM before their first birth. After exclusions, we included 15,906 participants in the analysis (ESM Figure 1). Participants in our study differed from those excluded (n=100,523) in several characteristics, including a lower history of GDM (4.7% vs. 5.6%) compared to women not included in the analysis (ESM Table 1).
The study protocol was approved by the Institutional Review Boards of the Brigham and Women’s Hospital and the Harvard T.H. Chan School of Public Health. Completion of questionnaires implied consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
Measures
Gestational diabetes
History of GDM was captured in 1989 and biennially through 2001 by self-report of a physician’s diagnosis. These reports were validated against medical records in a subgroup of NHS II participants.[16] In the 2009 assessment, participants reported their lifetime reproductive history, including the diagnosis of GDM and year of pregnancy. We derived our primary exposure, history of GDM, by combining reports from 1989–2001 and 2009. For secondary analyses that required a date of GDM occurrence, we only considered reports from the 2009 assessment.
Cognitive function
Our study focused on one evaluation of cognitive function obtained with the Cogstate brief battery, an online cognitive assessment with proven clinical utility in identifying cognitive impairment and dementia.[17–19] The battery has shown good construct and criterion validity against standardized neuropsychological tests.[18, 20] The battery comprises the tasks of detection (measuring psychomotor function and information processing speed), identification (measuring visual attention and vigilance), one card learning (measuring visual learning and short-term memory), and one back (measuring attention and working memory).[20, 21] To improve normality, we transformed the task scores and internally standardized them to have a mean of 0 and SD 1. We then created three composite scores by averaging the standardized z-scores: 1) psychomotor speed/attention (mean of detection and identification z-scores); 2) learning/working memory (mean of one card learning and one back z-scores); 3) global cognition (mean of all four z-scores).[14, 15, 21] Higher z-scores reflect better cognitive performance.
Covariates
Demographic factors included age at cognitive assessment and self-reported race and ethnicity. We recognize that race and ethnicity are social and not biological factors but include them because GDM and dementia prevalence varies across people from different backgrounds. We considered participants’ socioeconomic status (SES) during childhood and adulthood, which included parental education and occupation reported in 2005, marital status, and neighborhood SES at enrollment.[22] We used the participants’ highest educational level reported in 2018 to measure cognitive reserve. Information on parity (pregnancies ≥6 months) and age at first birth was obtained at enrollment and updated biennially. We used height, self-reported in 1989, and weight, queried in 1989 and biennially, to calculate body mass index (BMI) at each assessment cycle (weight in kilograms divided by height in meters squared).[23] Cigarette smoking was assessed in 1989 and biennially. At enrollment, participants reported their regular alcohol consumption and strenuous physical activity at various age intervals. To determine pre-pregnancy characteristics (i.e., before the first birth), including BMI, smoking, alcohol consumption, and physical activity, we used the information from the assessment cycle or age interval closest to but preceding the participants’ first birth. We recorded self-reported physician-diagnosed health conditions in 1989 and biennially. We identified cases of hypertension before the first birth and T2DM diagnosed after the first birth but before the cognitive assessment. T2DM cases were confirmed using a supplemental questionnaire that included questions about diabetes symptoms, diagnostic tests, and hypoglycemic therapy. The protocol for confirming T2DM cases has been previously reported and validated.[24] We assessed depressive symptoms in 2013 with the Center for Epidemiologic Studies Depression (CESD) Scale.[25] We imputed missing covariate values using GDM group-specific means or modes for continuous and categorical covariates, respectively.
Statistical analysis
We estimated mean differences (β, 95% confidence intervals [CIs]) in cognitive composite z-scores by history of GDM (yes vs. no [reference]) using multivariable linear regression. In model 1, we adjusted for age at cognitive assessment, non-Hispanic White vs. non-White race and ethnicity (due to sparsity in the other categories), participants’ education (associate’s degree, bachelor’s, master’s, or doctorate), and wave of enrollment in the cognition sub-study. In model 2, in addition to model 1 covariates, we accounted for parental education (high school or less, some college, four years of college or more), parental occupation (unskilled labor/service, farmer, mechanic/machine/military/sales, or professionals), and participants’ neighborhood SES (quartiles). In model 3, we built on model 2 and further adjusted for pre-pregnancy characteristics (i.e., before the first birth), including BMI (kg/m2), smoking status (never, past or current smoker), alcohol intake (no. drinks: none, ≤1/wk, 2–6/wk, ≥1/day), strenuous physical activity (never, 1–3, 4–6, 7–9, 10–12 mo/yr), and hypertension (yes vs. no). In a supplementary model, we accounted for depressive symptoms, which are independent predictors of cognitive function but may lie in the pathway between GDM and cognitive health outcomes.[26] Age is an important determinant of cognitive function. However, we did not find evidence of an interaction between GDM and age at cognitive assessment for any composite score (P for interaction >0.05). Consequently, we did not include the interaction term in the final models. To contextualize our results, we used linear regression to estimate mean differences in cognitive function associated with a one-year increase in age while adjusting for race and ethnicity.
We conducted two secondary analyses to understand the role of developing T2DM after GDM on cognitive function. First, we cross-categorized exposure to GDM and subsequent T2DM as neither GDM nor T2DM [reference], GDM only, T2DM only, and GDM followed by T2DM. Second, we evaluated the proportion of the GDM-cognitive function association explained by T2DM by conducting a formal mediation analysis using the Causalmed procedure implemented in SAS.[27] This procedure fits generalized linear models for cognitive function (continuous) and T2DM (binary) based on the counterfactual framework for causal mediation analysis. Provided that the models are correctly specified and there is no unmeasured exposure-outcome confounding, mediator-outcome confounding, exposure-mediator confounding, and mediator-outcome confounding affected by the exposure, the procedure allows the decomposition of total effects into direct and indirect effects. We tested for interactions between GDM and T2DM. Since the interaction terms were not significant for any of the cognitive composite scores (P for interaction >0.05), we did not include these terms in the models for mediation analysis. We conducted both secondary analyses in the subset of participants with 2009 reproductive information (n=15,303) as only this subset provided an exact date of GDM occurrence to establish temporality with T2DM diagnosis.
We conducted several sensitivity analyses. First, we used inverse probability weighting (IPW) to control for selection bias in our main analysis, given that the prevalence of a history of GDM differed by participation in the cognition sub-study. Participation in this sub-study was the most important factor limiting inclusion in our study and, consequently, in the overall analytical sample. We followed the recommended two-step approach.[28, 29] We first used a logistic regression model to predict the probability of completion of the cognitive assessment among those invited to participate in the sub-study. This model included significant predictors of participation such as history of GDM, race and ethnicity, parental education and occupation, participants’ education, marital status, neighborhood SES, age at first birth, parity, and pre-pregnancy physical activity. Next, we calculated the weight as the inverse of the probability of selection and included it in GDM-cognitive function linear regression models.
To understand the influence of BMI, an important risk factor for GDM associated with cognitive impairment,[30, 31] we studied the joint associations of GDM and pre-pregnancy BMI (i.e., before the first birth). We then studied the number of pregnancies complicated by GDM as the exposure (0 [reference], 1, ≥2) in the subset of participants with 2009 reproductive information and tested for linear trend by treating the categories as an ordinal variable.
In the last set of sensitivity analyses, we considered different covariates to rule out potential residual confounding. Instead of adjusting for depressive symptoms in a supplementary model, we accounted for a self-reported history of physician-diagnosed depression prior to the cognitive assessment (i.e., by 2013 for wave 1 and 2017 for wave 2). History of depression was reported every two years beginning in 2003. Next, in addition to the main modeĺs covariates, we adjusted for BMI (kg/m2) and lifestyle factors assessed before cognitive testing. These included diet quality measured via the Alternate Healthy Eating Index score, alcohol intake (g), smoking status (never, past, current), and physical activity (<3, 3–8.9,9–17.9,18–26.9, ≥27 MET-h/wk). Briefly, BMI and smoking were assessed biennially, while physical activity and diet were assessed every four years. We selected the assessments closest but preceding the cognitive assessment. For example, we used BMI, smoking, and physical activity data from 2013 and 2017 for participants in wave 1 and 2, respectively. For diet and alcohol, we used data from 2011 for wave 1 and 2015 for wave 2 participants. In the last sensitivity analysis, we additionally adjusted the main model for lifetime parity.
We conducted data analysis in SAS statistical software version 9.4 (SAS Institute).
Results
The 15,906 participants had a mean (SD) age of 62.0 (4.9) years at cognitive assessment, they were predominantly non-Hispanic White (96.2%), and 4.7% had a lifetime history of GDM. Women with GDM were slightly older at their first birth (28.0 vs. 26.7 years) and had a higher pre-pregnancy BMI (23.2 vs. 22.0 kg/m2) than those without GDM (Table 1).
Table 1.
Participants’ Characteristics by Lifetime History of Gestational Diabetes Mellitus Among 15,906 Parous Women in the Nurses’ Health Study II
| History of Gestational Diabetes Mellitus | |||
|---|---|---|---|
| Total (n=15,906) | Yes (n=749, 4.7%) | No (n=15,157, 95.3%) | |
| Characteristic | |||
| Age at cognitive assessment, mean (SD), years | 62.0 (4.9) | 62.2 (4.8) | 62.0 (4.9) |
| Age at first birth, mean (SD), years | 26.8 (4.7) | 28.0 (4.8) | 26.7 (4.7) |
| Parity, mean (SD), births | 2.3 (0.9) | 2.4 (1.0) | 2.3 (0.9) |
| Pre-pregnancy BMI, mean (SD), kg/m2 a | 22.0 (2.9) | 23.2 (4.2) | 22.0 (2.8) |
| Depressive symptoms score, mean (SD) b | 5.5 (4.4) | 5.9 (4.7) | 5.5 (4.4) |
| Race and ethnicity, n (%) | |||
| Non-Hispanic White | 15,305 (96.2) | 712 (95.1) | 14,593 (96.3) |
| Black | 83 (0.5) | 7 (0.9) | 76 (0.5) |
| Hispanic | 152 (1.0) | 9 (1.2) | 143 (0.9) |
| Asian | 168 (1.0) | 10 (1.3) | 158 (1.0) |
| Other | 198 (1.2) | 11 (15) | 187 (1.2) |
| Parental education, n (%) | |||
| High school or less | 8,158 (51.3) | 390 (52.1) | 7,768 (51.3) |
| Some college | 3,782 (23.8) | 157 (21.0) | 3,625 (23.9) |
| Four years college or more | 3,966 (24.9) | 202 (27.0) | 3,764 (24.8) |
| Parental occupation, n (%) | |||
| Unskilled labor, service | 1,710 (10.8) | 90 (12.0) | 1,620 (10.7) |
| Farmer | 984 (6.2) | 34 (4.5) | 950 (6.3) |
| Mechanic, machine, military, sales | 7,550 (47.5) | 357 (47.7) | 7,193 (47.5) |
| Professionals | 5,662 (35.6) | 268 (35.8) | 5,394 (35.6) |
| Married, n (%) | 13,696 (86.1) | 643 (85.8) | 13,053 (86.1) |
| Participant’s education, n (%) | |||
| Associate’s degree | 3,570 (22.4) | 163 (21.8) | 3,407 (22.5) |
| Bachelor’s degree | 7,255 (45.6) | 356 (47.5) | 6,899 (45.5) |
| Master’s degree | 4,459 (28.0) | 196 (26.2) | 4,263 (28.1) |
| Doctorate degree | 622 (3.9) | 34 (4.5) | 588 (3.9) |
| Neighborhood socioeconomic status, n (%) | |||
| Quartile 1 | 3,689 (23.2) | 159 (21.2) | 3,530 (23.3) |
| Quartile 2 | 3,605 (22.7) | 280 (37.4) | 3,325 (21.9) |
| Quartile 3 | 4,365 (27.4) | 136 (18.2) | 4,229 (27.9) |
| Quartile 4 | 4,247 (26.7) | 174 (23.2) | 4,073 (26.9) |
| Pre-pregnancy smoking, n (%) a | |||
| Never smoker | 11,005 (69.2) | 526 (70.2) | 10,479 (69.1) |
| Past smoker | 1,475 (9.3) | 80 (10.7) | 1,395 (9.2) |
| Current smoker | 3,426 (21.5) | 143 (19.1) | 3,283 (21.7) |
| Pre-pregnancy alcohol (# of drinks), n (%) a | |||
| None | 4,192 (26.4) | 210 (28.0) | 3,982 (26.3) |
| ≤1/wk | 6,224 (39.1) | 283 (37.8) | 5,941 (39.2) |
| 2–6/wk | 4,459 (28.0) | 207 (27.6) | 4,252 (28.1) |
| ≥1/day | 1,031 (6.5) | 49 (6.5) | 982 (6.5) |
| Pre-pregnancy strenuous physical activity, n (%) a | |||
| Never | 4,721 (29.7) | 204 (27.2) | 4,517 (29.8) |
| 1–3 mo/yr | 4,886 (30.7) | 243 (32.4) | 4,643 (30.6) |
| 4–6 mo/yr | 2,680 (16.8) | 135 (18.0) | 2,545 (16.8) |
| 7–9 mo/yr | 1,845 (11.6) | 83 (11.1) | 1,762 (11.6) |
| 10–12 mo/yr | 1,774 (11.2) | 84 (11.2) | 1,690 (11.1) |
| Pre-pregnancy chronic hypertension, n (%) a | 119 (0.7) | 9 (1.2) | 110 (0.7) |
| Cogstate composite z-scores c | |||
| Psychomotor speed/attention, mean (SD) | −0.04 (0.9) | −0.12 (0.9) | −0.03 (0.9) |
| Learning/working memory, mean (SD) | −0.03 (0.7) | −0.10 (0.8) | −0.03 (0.7) |
| Global cognition, mean (SD) | −0.03 (0.7) | −0.11 (0.7) | −0.03 (0.7) |
BMI, body mass index.
These characteristics reflect status before the first birth.
Depressive symptoms by the 10-item Center for Epidemiologic Studies Depression scale applied in 2013.
Standardized z-scores; higher z-scores indicate better cognitive performance.
Compared to women without a history of GDM, those with a history of this condition scored lower in both psychomotor speed/attention (β, −0.08; 95% CI, −0.14 to −0.01) and global cognition (β, −0.06; 95% CI, −0.11 to −0.01) with consistent results across models (Table 2). The learning/working memory composite estimates were in the same direction but slightly weaker (β, −0.05; 95% CI, −0.10 to 0.01). For reference, a one-year increase in age has a negative impact of −0.04 z-scores on psychomotor speed/attention and global cognition, and −0.03 z-scores on learning/working memory. Therefore, compared to women without a history of GDM, the associations of GDM with psychomotor speed/attention and global cognition are equivalent to those of being 2 and 1.5 years older, respectively. The results were consistent after accounting for depressive symptoms (ESM Table 2) and when we used IPW to account for selection bias (ESM Table 3).
Table 2.
Associations Between Lifetime History of Gestational Diabetes Mellitus and Cogstate Composite z-Scores, Among Parous Women in the Nurses’ Health Study II a
| β (95% CI) b | |
|---|---|
| Psychomotor speed/attention (GDM, yes [n=748] vs. no [n=15,129]) | |
| Model 1 c | −0.08 (−0.15 to −0.01) |
| Model 2 d | −0.08 (−0.15 to −0.02) |
| Model 3 e | −0.08 (−0.14 to −0.01) |
| Learning/working memory (GDM, yes [n=748] vs. no [n=15,151]) | |
| Model 1 c | −0.05 (−0.11 to 0.00) |
| Model 2 d | −0.05 (−0.10 to 0.00) |
| Model 3 e | −0.05 (−0.10 to 0.01) |
| Global cognition (GDM, yes [n=747] vs. no [n=15,123]) | |
| Model 1 c | −0.07 (−0.12 to −0.02) |
| Model 2 d | −0.07 (−0.12 to −0.02) |
| Model 3 e | −0.06 (−0.11 to −0.01) |
GDM, gestational diabetes mellitus.
Analysis conducted in the total analytical sample (n=15,906). Samples for each composite z-score may not add up to 15,906 due to missing outcome data.
Higher z-scores reflect better cognitive performance.
Model 1: age at cognitive assessment, non-Hispanic White race and ethnicity, participants’ education and wave of enrollment in the cognition sub-study.
Model 2: model 1 + childhood socioeconomic status (parental education, parental occupation) and neighborhood socioeconomic status.
Model 3: model 2 + pre-pregnancy characteristics (body mass index, smoking status, alcohol intake, strenuous physical activity, and chronic hypertension).
Women who experienced GDM with subsequent T2DM scored lower in all the composites compared to those with neither GDM nor T2DM; in fact, these women had the lowest cognitive performance among the groups studied with mean differences of −0.13 (95% CI, −0.28 to 0.02), −0.11 (95% CI, −0.23 to 0.01) and −0.12 (95% CI, −0.24 to −0.01) for the psychomotor speed/attention, learning/working memory and global cognition composites, respectively (Table 3). Compared to women with neither GDM nor T2DM, those who experienced GDM without subsequent T2DM also scored lower in psychomotor speed/attention (β, −0.08; 95% CI, −0.16 to 0.00), global cognition (β, −0.06; 95% CI, −0.12 to 0.00) and learning/working memory (β, −0.04; 95% CI, −0.11 to 0.03), although the estimates for the latter composite were weaker in magnitude. The mediation analysis showed that T2DM explained an important proportion of the GDM-cognitive function association, 12.66% (95% CI, −1.79 to 27.11%) for psychomotor speed/attention, 19.61% (95% CI, −8.82 to 48.05%) for learning/working memory and 14.98% (95% CI, −0.60 to 30.57%) for global cognition. However, it should be noted that T2DM did not account for the majority of the relationship between GDM and cognitive function (Table 4).
Table 3.
Associations Between Exposure to Gestational Diabetes Mellitus and Subsequent Type 2 Diabetes Mellitus and Cogstate Composite z-Scores a
| β (95% CI) b | ||||
|---|---|---|---|---|
| Model 1 c | Model 2 d | Model 3 e | ||
| No. | ||||
| Psychomotor speed/attention | ||||
| Neither GDM nor T2DM | 13,779 | [reference] | [reference] | [reference] |
| GDM only | 465 | −0.08 (−0.16 to 0.00) | −0.08 (−0.16 to 0.00) | −0.08 (−0.16 to 0.00) |
| T2DM only | 897 | −0.10 (−0.16 to −0.04) | −0.10 (−0.16 to −0.04) | −0.09 (−0.16 to −0.03) |
| GDM + T2DM | 134 | −0.15 (−0.30 to 0.00) | −0.14 (−0.30 to 0.01) | −0.13 (−0.28 to 0.02) |
| Learning/working memory | ||||
| Neither GDM nor T2DM | 13,798 | [reference] | [reference] | [reference] |
| GDM only | 466 | −0.04 (−0.11 to 0.03) | −0.04 (−0.11 to 0.03) | −0.04 (−0.11 to 0.03) |
| T2DM only | 899 | −0.09 (−0.14 to −0.04) | −0.09 (−0.14 to −0.04) | −0.08 (−0.13 to −0.03) |
| GDM + T2DM | 133 | −0.12 (−0.25 to 0.00) | −0.12 (−0.24 to 0.00) | −0.11 (−0.23 to 0.01) |
| Global cognition | ||||
| Neither GDM nor T2DM | 13,773 | [reference] | [reference] | [reference] |
| GDM only | 465 | −0.06 (−0.12 to 0.00) | −0.06 (−0.12 to 0.00) | −0.06 (−0.12 to 0.00) |
| T2DM only | 897 | −0.09 (−0.14 to −0.05) | −0.09 (−0.14 to −0.05) | −0.09 (−0.13 to −0.04) |
| GDM + T2DM | 133 | −0.13 (−0.25 to −0.02) | −0.13 (−0.25 to −0.02) | −0.12 (−0.24 to −0.01) |
GDM, gestational diabetes mellitus; T2DM, type 2 diabetes mellitus.
Analysis conducted in the subset of participants with reproductive information reported in 2009 (n=15,303). Samples for each composite z-score may not add up to 15,303 due to missing outcome data.
Higher z-scores reflect better cognitive performance.
Model 1: age at cognitive assessment, non-Hispanic White race and ethnicity, participants’ education and wave of enrollment in the cognition sub-study.
Model 2: model 1 + childhood socioeconomic status (parental education, parental occupation) and neighborhood socioeconomic status.
Model 3: model 2 + pre-pregnancy characteristics (body mass index, smoking status, alcohol intake, strenuous physical activity, and chronic hypertension).
Table 4.
Estimated Proportion of the Associations Between Lifetime History of Gestational Diabetes Mellitus and Cogstate Composite z-Scores Explained by Type 2 Diabetes Mellitus a, b
| Estimate (95% CI) c’ d | |
|---|---|
| Psychomotor speed/attention (n=15,275) | |
| Total effect | −0.08 (−0.16 to −0.01) |
| Direct effect | −0.07 (−0.15 to 0.00) |
| Indirect effect | −0.01 (−0.02 to 0.00) |
| Proportion mediated | 12.66 (−1.79, 27.11) |
| Learning/working memory (n=15,296) | |
| Total effect | −0.05 (−0.11 to 0.01) |
| Direct effect | −0.04 (−0.10 to 0.02) |
| Indirect effect | −0.01 (−0.01 to 0.00) |
| Proportion mediated | 19.61 (−8.82, 48.05) |
| Global cognition (n=15,268) | |
| Total effect | −0.06 (−0.12 to −0.01) |
| Direct effect | −0.05 (−0.11 to 0.00) |
| Indirect effect | −0.01 (−0.02 to 0.00) |
| Proportion mediated | 14.98 (−0.60 to 30.57) |
Type 2 diabetes mellitus diagnosed during follow-up.
Analysis conducted in the subset of participants with reproductive information reported in 2009 (n=15,303). Samples for each composite z-score may not add up to 15,303 due to missing outcome data.
Estimates for total, direct and indirect effects represent β coefficients. Higher z-scores reflect better cognitive performance.
Adjusted for age at cognitive assessment, non-Hispanic White race and ethnicity, participants’ education, wave of enrollment in the cognition sub-study, childhood socioeconomic status (parental education, parental occupation), neighborhood socioeconomic status, and pre-pregnancy characteristics (body mass index, smoking status, alcohol intake, strenuous physical activity, and chronic hypertension).
In a sensitivity analysis, we studied the joint associations of a history of GDM and pre-pregnancy BMI. Compared to women without a history of GDM and with normal pre-pregnancy BMI, women without GDM and with overweight before pregnancy scored lower in learning/working memory and global cognition. Women without GDM and with pre-pregnancy obesity and those with a history of GDM who had normal weight before pregnancy had lower performance in psychomotor speed/attention and global cognition. Counter to our expectations, we did not observe consistent associations with cognitive function for women with a history of both GDM and pre-pregnancy overweight or obesity (ESM Table 4). When the number of pregnancies complicated by GDM was the exposure, we observed associations consistent with the main analysis among women with one complicated pregnancy only (ESM Table 5). Additionally, when we accounted for a self-reported history of physician-diagnosed depression, the results were consistent with the main analysis (ESM Table 6). The main results were also unchanged after adjusting for BMI and lifestyle factors before the cognitive assessment (ESM Table 7) and parity (ESM Table 8).
Discussion
In this study of 15,906 women, we evaluated the associations between a lifetime history of GDM and cognitive function assessed at a mean age of 62.0 years. We found that women with a history of GDM had lower psychomotor speed/attention and global cognition performance than their counterparts without a history of GDM. The observed associations were equivalent to an additional 1.5 to 2 years of aging and align with those of known factors affecting cognitive function. For instance, hypertension has been reported to negatively impact psychomotor speed/attention with a difference of −0.09 z-scores compared to people without hypertension,[32] similar to the association for GDM in our study (β, −0.08). Furthermore, although T2DM explained an important proportion of the associations between GDM and cognitive function, they were not entirely dependent upon the development of this condition.
To our knowledge, only one study has been conducted on the association between a history of GDM and cognitive function.[3] In the study of 730 US women from the Bogalusa Heart Study, women with a history of GDM had lower performance in multiple measures of cognitive function assessed at a mean age of 47.7 years, including working memory and processing speed (β, −0.18; standard error [SE] 0.11), attention (β, −0.03; SE 0.10) and executive function (β, −0.08; SE 0.11) compared to women without a history of GDM, although the results were not statistically significantly.[3] In our study, women with a history of GDM scored lower in the psychomotor speed/attention and global cognition composites. Our findings were comparable to those from the Bogalusa Heart Study; however, their lack of significant results may be due to limited statistical power as they only had a sample of 67 women with GDM compared to our study, which had 749 participants with this condition.
Women with GDM have a 10-fold times higher risk of subsequent T2DM,[8] and this condition is a strong predictor of cognitive impairment and dementia.[11, 33] Therefore, we hypothesized that the development of T2DM would be a mechanism linking GDM and cognitive function. Our findings of lower cognitive performance, particularly in global cognition, in women with a history of GDM subsequently diagnosed with T2DM support our hypothesis and align with prior research showing that T2DM affects memory, attention, and processing speed.[34, 35] A recent study by van Gennip et al. found that processing speed was lower in 10,663 individuals with T2DM compared to 77,193 control subjects within the UK Biobank study (β, −0.11; 95% CI, −0.13 to −0.19).[36] In a replication analysis within the same study, they compared 1,327 individuals with T2DM to 3,732 controls from the Maastricht Study and found that those with T2DM also had lower processing speed, executive function, and memory.[36] Macrovascular disease, hyperglycemia-related neurotoxicity, oxidative stress, endothelial dysfunction, and inflammation are among the factors contributing to lower cognitive function in people with T2DM.[10, 12]
Interestingly, women with GDM who did not develop T2DM also had lower cognitive performance, particularly in the psychomotor speed/attention and global cognition composites, than those who never experienced GDM or T2DM. The mediation analysis confirmed that the development of T2DM did not explain a substantial proportion of the GDM-cognitive function associations and suggested a strong and direct path between GDM and cognitive function. There are some potential explanations for these findings. After pregnancy, women with a history of GDM have higher levels of cardiometabolic risk factors, including blood pressure, glucose, dyslipidemia, insulin resistance, and C-reactive protein.[37–41] All these have been linked with lower cognitive function and a higher risk of cognitive impairment and dementia.[10, 12, 42–45] For example, insulin resistance could potentially cause an insulin-resistant brain by altering insulin transport and receptor activity.[46] This may cause the accumulation of amyloid β protein, which can contribute to cognitive decline.[46] There is evidence that some cardiometabolic risk factors may already be present in women with GDM within the first year after delivery [41] and even before gestation.[47, 48] These data suggest that women with GDM may experience chronic exposure to an adverse risk factor profile associated with poor cognitive outcomes.
Another possible explanation is that GDM and T2DM represent phenotypes with a common pathophysiology that differs only in the degree of severity. This idea is reinforced by the observation that cognitive function seemed to gradually decrease in women with GDM alone, to T2DM alone, and both conditions combined in the joint GDM-T2DM analysis.
Our research contributes to the growing body of evidence that shows adverse long-term health consequences associated with GDM in women. While further studies are needed to confirm our results, our findings emphasize the importance of preventing GDM in the first place as a potential avenue for reducing the prevalence of cognitive impairment in women.[1] Additionally, our findings highlight the importance of an ongoing postpartum follow-up after GDM to timely identify and intervene in women at higher risk of developing T2DM and cardiometabolic risk factors for cognitive impairment. Targeting these risk factors earlier in life, for example, right after the affected pregnancy and throughout the reproductive years, may have implications for cognitive outcomes. Research shows that exposure to cardiometabolic risk factors earlier in life, such as between ages 44 −54, which coincides with the end of the reproductive lifespan, is more strongly associated with dementia risk than similar factors in late life.[49]
In the NHS II, repeated cognitive tests are available for a subset of participants as part of the cognition sub-study. While our analysis focused on cognitive reserve, specifically cognitive health at baseline, in subsequent studies we will examine changes in cognitive function and cognitive decline in relation to GDM. Future research should be conducted to understand the mechanisms underlying the associations between GDM and cognitive health later in life, and how the development of T2DM contributes to these associations. Furthermore, the role of lifestyle should be explored as a potential avenue for preventing adverse cognitive health outcomes after GDM.
Limitations and Strengths
Our study has limitations. First, history of GDM was self-reported and could be prone to recall bias, leading to misclassification. However, physician-diagnosed GDM has been validated in the NHS II, and recalled reproductive history has been shown to be reliable in other studies.[16, 50] Second, depressive symptoms were assessed in 2013 and may not be representative of symptoms at the time of cognitive testing, particularly for participants in wave 2; however, adjustment for history of physician-diagnosed depression yielded identical results as those of the main analysis. Third, selection bias is a concern due to differences in the history of GDM based on participation in the cognition sub-study and inclusion in the analytical sample. However, using IPW to account for potential selection bias led to results identical to the main analysis. Fourth, the results of the mediation analysis are valid provided the models are correctly specified and the assumptions of no unmeasured confounding are met. In our analysis, T2DM was specified as a binary mediator, and we assumed a linear association with cognitive function. This model specification may oversimplify the relationship between T2DM and cognitive function, as there could be threshold, or nonlinear effects influenced by factors such as T2DM duration and glycemic control that are not captured. Additionally, as with most observational studies, it is not possible to completely rule out residual confounding due to covariates not included in the models. Lastly, since our study population included predominantly non-Hispanic White and highly educated women, our findings might be generalizable only to women of a similar background. Studies in more diverse racial and ethnic samples would be informative, as rates of both GDM and dementia vary substantially across different races and ethnicities. Additionally, our study population consisted of nurses whose risk profiles for GDM and cognitive health may differ from those of the general US female population, which further limits the generalizability of our findings.
Our study has several strengths. This study is among the first examinations of the association between a history of GDM and cognitive function. The large sample of women and availability of several important confounders, potential mediators, and predictors of cognitive function are key strengths. Having information on the diagnosis of T2DM and its temporality with GDM enabled us to examine a potential mechanism through which GDM might affect cognitive function.
Conclusions
History of GDM was associated with lower performance in psychomotor speed/attention and global cognition in participants from the NHS II. The observed associations were only partially explained by subsequent T2DM. If our findings are confirmed, they support further research on early risk modification strategies for women with a history of GDM as a potential avenue to decrease their risk of cognitive impairment.
Supplementary Material
Research in context.
What is already known about this subject?
Gestational diabetes mellitus (GDM) affects cognitive performance during pregnancy, and it may also have long-term implications in cognitive function, for example, through the development of type 2 diabetes mellitus (T2DM). Evidence of the association between a history GDM and cognitive function in midlife is scarce.
What is the key question?
Is there an association between a history of GDM and cognitive function in midlife?
What are the new findings?
A history of GDM is associated with lower performance in psychomotor speed/attention and global cognition among parous women in midlife. The associations between GDM and cognitive function were only partially explained by the development of T2DM.
How might this impact on clinical practice in the foreseeable future?
The results suggest that women who experience GDM during their reproductive lifetime may be at higher risk of poor cognition in midlife and may benefit from early risk modification strategies to prevent cognitive impairment.
Acknowledgements:
Data and/or research tools used in the preparation of this manuscript were obtained from the National Institute of Mental Health (NIMH) Data Archive (NDA). NDA is a collaborative informatics system created by the National Institutes of Health to provide a national resource to support and accelerate research in mental health. Dataset identifier(s): 10.15154/mmdn-dh30. This manuscript reflects the views of the authors and may not reflect the opinions or views of the NIH or of the Submitters submitting original data to NDA.
We acknowledge the Channing Division of Network Medicine of the Department of Medicine, Brigham and Women’s Hospital, and Harvard Medical School for managing the Nurses’ Health Study II.
Funding:
This study was supported by grants R01MH101269, R01MH078928 and 1R21MH102570 from the National Institute of Mental Health, U01CA176726 and U01 HL145386 from the National Institute of Health, and R01ES017017 and R01ES028033 from the National Institute of Environmental Health Sciences. During this work, Dr. Soria-Contreras was supported by the National Research Service Award T32 HD 104612. She is currently supported by grant U54 AG062322, funded by the National Institute on Aging and the Office of Research on Women’s Health. Dr. Siwen Wang is supported by the Irene M. & Fredrick J. Stare Nutrition Education Fund Doctoral Scholarship and Mayer Fund Doctoral Scholarship. The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Abbreviations
- BMI
Body Mass Index
- CESD
Center for Epidemiologic Studies Depression
- CVD
Cardiovascular Disease
- GDM
Gestational Diabetes Mellitus
- IPW
Inverse Probability Weighting
- NHS II
Nurses’ Health Study II
- SES
Socioeconomic Status
- T2DM
Type 2 Diabetes Mellitus
- US
United States
Footnotes
Competing interests declaration: Since the time that the work for this manuscript was completed, Dr. Purdue-Smithe became a full-time employee of Merck & Co.
Data Sharing Statement:
Because of participant confidentiality and privacy concerns, data cannot be shared publicly and requests to access Nurses’ Health Study II data must be submitted in writing. Further information including the procedures to obtain and access the data is described at https://www.nurseshealthstudy.org/researchers (contact email: nhsaccess@channing.harvard.edu).
References
- 1.GBD 2019 Dementia Forecasting Collaborators (2022) Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health 7(2):e105–e125. 10.1016/S2468-2667(21)00249-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Peterson A, Tom SE (2021) A Lifecourse Perspective on Female Sex-Specific Risk Factors for Later Life Cognition. Curr Neurol Neurosci Rep 21(9):46. 10.1007/s11910-021-01133-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Harville EW, Guralnik J, Romero M, Bazzano LA (2020) Reproductive History and Cognitive Aging: The Bogalusa Heart Study. Am J Geriatr Psychiatry 28(2):217–225. 10.1016/j.jagp.2019.07.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Centers for Disease Control and Prevention (2021) Gestational diabetes. Available from https://www.cdc.gov/diabetes/basics/gestational.html. Accessed 15 Mar 2022
- 5.Ye W, Luo C, Huang J, Li C, Liu Z, Liu F (2022) Gestational diabetes mellitus and adverse pregnancy outcomes: systematic review and meta-analysis. BMJ 377:e067946. 10.1136/bmj-2021-067946 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Keskin FE, Ozyazar M, Pala AS, et al. (2015) Evaluation of Cognitive Functions in Gestational Diabetes Mellitus. Exp Clin Endocrinol Diabetes 123(4):246–251. 10.1055/s-0034-1395634 [DOI] [PubMed] [Google Scholar]
- 7.Sana S, Deng X, Guo L, Wang X, Li E (2021) Cognitive Dysfunction of Pregnant Women with Gestational Diabetes Mellitus in Perinatal Period. J Healthc Eng 2021:2302379. 10.1155/2021/2302379 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Vounzoulaki E, Khunti K, Abner SC, Tan BK, Davies MJ, Gillies CL (2020) Progression to type 2 diabetes in women with a known history of gestational diabetes: systematic review and meta-analysis. BMJ 369:m1361. 10.1136/bmj.m1361 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Xie W, Wang Y, Xiao S, Qiu L, Yu Y, Zhang Z (2022) Association of gestational diabetes mellitus with overall and type specific cardiovascular and cerebrovascular diseases: systematic review and meta-analysis. BMJ 378:e070244. 10.1136/bmj-2022-070244 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zlokovic BV, Gottesman RF, Bernstein KE, et al. (2020) Vascular contributions to cognitive impairment and dementia (VCID): A report from the 2018 National Heart, Lung, and Blood Institute and National Institute of Neurological Disorders and Stroke Workshop. Alzheimers Dement 16(12):1714–1733. 10.1002/alz.12157 [DOI] [PubMed] [Google Scholar]
- 11.Xue M, Xu W, Ou Y-N, et al. (2019) Diabetes mellitus and risks of cognitive impairment and dementia: A systematic review and meta-analysis of 144 prospective studies. Ageing Res Rev 55:100944. 10.1016/j.arr.2019.100944 [DOI] [PubMed] [Google Scholar]
- 12.McCrimmon RJ, Ryan CM, Frier BM (2012) Diabetes and cognitive dysfunction. Lancet 379(9833):2291–2299. 10.1016/S0140-6736(12)60360-2 [DOI] [PubMed] [Google Scholar]
- 13.Bao Y, Bertoia ML, Lenart EB, et al. (2016) Origin, Methods, and Evolution of the Three Nurses’ Health Studies. Am J Public Health 106(9):1573–1581. 10.2105/AJPH.2016.303338 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Sumner JA, Hagan K, Grodstein F, Roberts AL, Harel B, Koenen KC (2017) Posttraumatic stress disorder symptoms and cognitive function in a large cohort of middle-aged women. Depress Anxiety 34(4):356–366. 10.1002/da.22600 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lawn RB, Jha SC, Liu J, et al. (2022) The association of posttraumatic stress disorder, depression, and head injury with mid-life cognitive function in civilian women. Depress Anxiety 39(3):220–232. 10.1002/da.23233 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Solomon CG, Willett WC, Rich-Edwards J, et al. (1996) Variability in diagnostic evaluation and criteria for gestational diabetes. Diabetes Care 19(1):12–16. 10.2337/diacare.19.1.12 [DOI] [PubMed] [Google Scholar]
- 17.Hammers D, Spurgeon E, Ryan K, et al. (2012) Validity of a brief computerized cognitive screening test in dementia. J Geriatr Psychiatry Neurol 25(2):89–99. 10.1177/0891988712447894 [DOI] [PubMed] [Google Scholar]
- 18.Maruff P, Thomas E, Cysique L, et al. (2009) Validity of the CogState brief battery: relationship to standardized tests and sensitivity to cognitive impairment in mild traumatic brain injury, schizophrenia, and AIDS dementia complex. Arch Clin Neuropsychol 24(2):165–178. 10.1093/arclin/acp010 [DOI] [PubMed] [Google Scholar]
- 19.Maruff P, Lim YY, Darby D, et al. (2013) Clinical utility of the cogstate brief battery in identifying cognitive impairment in mild cognitive impairment and Alzheimer’s disease. BMC Psychol 1(1):30. 10.1186/2050-7283-1-30 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Fredrickson J, Maruff P, Woodward M, et al. (2010) Evaluation of the usability of a brief computerized cognitive screening test in older people for epidemiological studies. Neuroepidemiology 34(2):65–75. 10.1159/000264823 [DOI] [PubMed] [Google Scholar]
- 21.Koyama AK, Hagan KA, Okereke OI, Weisskopf MG, Rosner B, Grodstein F (2015) Evaluation of a Self-Administered Computerized Cognitive Battery in an Older Population. Neuroepidemiology 45(4):264–272. 10.1159/000439592 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.DeVille NV, Iyer HS, Holland I, et al. (2023) Neighborhood socioeconomic status and mortality in the nurses’ health study (NHS) and the nurses’ health study II (NHSII). Environ Epidemiol 7(1):e235. 10.1097/EE9.0000000000000235 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Rimm EB, Stampfer MJ, Colditz GA, Chute CG, Litin LB, Willett WC (1990) Validity of self-reported waist and hip circumferences in men and women. Epidemiology 1(6):466–473. 10.1097/00001648-199011000-00009 [DOI] [PubMed] [Google Scholar]
- 24.Manson JE, Rimm EB, Stampfer MJ, et al. (1991) Physical activity and incidence of non-insulin-dependent diabetes mellitus in women. Lancet 338(8770):774–778. 10.1016/0140-6736(91)90664-b [DOI] [PubMed] [Google Scholar]
- 25.Andresen EM, Malmgren JA, Carter WB, Patrick DL (1994) Screening for depression in well older adults: evaluation of a short form of the CES-D (Center for Epidemiologic Studies Depression Scale). Am J Prev Med 10(2):77–84 [PubMed] [Google Scholar]
- 26.Shimada H, Park H, Makizako H, Doi T, Lee S, Suzuki T (2014) Depressive symptoms and cognitive performance in older adults. J Psychiatr Res 57:149–156. 10.1016/j.jpsychires.2014.06.004 [DOI] [PubMed] [Google Scholar]
- 27.Valeri L, Vanderweele TJ (2013) Mediation analysis allowing for exposure-mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods 18(2):137–150. 10.1037/a0031034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Nohr EA, Liew Z (2018) How to investigate and adjust for selection bias in cohort studies. Acta Obstet Gynecol Scand 97(4):407–416. 10.1111/aogs.13319 [DOI] [PubMed] [Google Scholar]
- 29.Bliddal M, Liew Z, Pottegård A, Kirkegaard H, Olsen J, Nohr EA (2018) Examining Nonparticipation in the Maternal Follow-up Within the Danish National Birth Cohort. Am J Epidemiol 187(7):1511–1519. 10.1093/aje/kwy002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Vats H, Saxena R, Sachdeva MP, Walia GK, Gupta V (2021) Impact of maternal pre-pregnancy body mass index on maternal, fetal and neonatal adverse outcomes in the worldwide populations: A systematic review and meta-analysis. Obes Res Clin Pract 15(6):536–545. 10.1016/j.orcp.2021.10.005 [DOI] [PubMed] [Google Scholar]
- 31.Pedditzi E, Peters R, Beckett N (2016) The risk of overweight/obesity in mid-life and late life for the development of dementia: a systematic review and meta-analysis of longitudinal studies. Age Ageing 45(1):14–21. 10.1093/ageing/afv151 [DOI] [PubMed] [Google Scholar]
- 32.Yassi N, Pase MP, Buckley RF, et al. (2022) Cardiovascular Risk Associated with Poorer Memory in Middle-Aged Adults from the Healthy Brain Project. J Alzheimers Dis 86(3):1081–1091. 10.3233/JAD-215375 [DOI] [PubMed] [Google Scholar]
- 33.Zhang J, Chen C, Hua S, et al. (2017) An updated meta-analysis of cohort studies: Diabetes and risk of Alzheimer’s disease. Diabetes Res Clin Pract 124:41–47. 10.1016/j.diabres.2016.10.024 [DOI] [PubMed] [Google Scholar]
- 34.Sadanand S, Balachandar R, Bharath S (2016) Memory and executive functions in persons with type 2 diabetes: a meta-analysis. Diabetes Metab Res Rev 32(2):132–142. 10.1002/dmrr.2664 [DOI] [PubMed] [Google Scholar]
- 35.Palta P, Schneider ALC, Biessels GJ, Touradji P, Hill-Briggs F (2014) Magnitude of cognitive dysfunction in adults with type 2 diabetes: a meta-analysis of six cognitive domains and the most frequently reported neuropsychological tests within domains. J Int Neuropsychol Soc 20(3):278–291. 10.1017/S1355617713001483 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.van Gennip ACE, Stehouwer CDA, van Boxtel MPJ, et al. (2021) Association of Type 2 Diabetes, According to the Number of Risk Factors Within Target Range, With Structural Brain Abnormalities, Cognitive Performance, and Risk of Dementia. Diabetes Care 44(11):2493–2502. 10.2337/dc21-0149 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Verma A, Boney CM, Tucker R, Vohr BR (2002) Insulin resistance syndrome in women with prior history of gestational diabetes mellitus. J Clin Endocrinol Metab 87(7):3227–3235. 10.1210/jcem.87.7.8684 [DOI] [PubMed] [Google Scholar]
- 38.Fraser A, Nelson SM, Macdonald-Wallis C, et al. (2012) Associations of pregnancy complications with calculated cardiovascular disease risk and cardiovascular risk factors in middle age: the Avon Longitudinal Study of Parents and Children. Circulation 125(11):1367–1380. 10.1161/CIRCULATIONAHA.111.044784 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Winzer C, Wagner O, Festa A, et al. (2004) Plasma adiponectin, insulin sensitivity, and subclinical inflammation in women with prior gestational diabetes mellitus. Diabetes Care 27(7):1721–1727. 10.2337/diacare.27.7.1721 [DOI] [PubMed] [Google Scholar]
- 40.Vilmi-Kerälä T, Palomäki O, Kankkunen P, Juurinen L, Uotila J, Palomäki A (2016) Oxidized LDL, insulin resistance and central blood pressure after gestational diabetes mellitus. Acta Obstet Gynecol Scand 95(12):1425–1432. 10.1111/aogs.13029 [DOI] [PubMed] [Google Scholar]
- 41.Pathirana MM, Lassi Z, Ali A, Arstall M, Roberts CT, Andraweera PH (2021) Cardiovascular risk factors in women with previous gestational diabetes mellitus: A systematic review and meta-analysis. Rev Endocr Metab Disord 22(4):729–761. 10.1007/s11154-020-09587-0 [DOI] [PubMed] [Google Scholar]
- 42.John CM, Yusof NISM, Aziz SHA, Fauzi FM (2018) Maternal cognitive impairment associated with gestational diabetes mellitus—A review of potential contributing mechanisms. Int J Mol Sci 19(12):1–16. 10.3390/ijms19123894 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Rasgon NL, Kenna HA, Wroolie TE, et al. (2011) Insulin resistance and hippocampal volume in women at risk for Alzheimer’s disease. Neurobiol Aging 32(11):1942–1948. 10.1016/j.neurobiolaging.2009.12.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Derby CA, Hutchins F, Greendale GA, et al. (2021) Cardiovascular risk and midlife cognitive decline in the Study of Women’s Health Across the Nation. Alzheimers Dement 17(8):1342–1352. 10.1002/alz.12300 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Crane PK, Walker R, Hubbard RA, et al. (2013) Glucose levels and risk of dementia. N Engl J Med 369(6):540–548. 10.1056/NEJMoa1215740 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Correia SC, Santos RX, Carvalho C, et al. (2012) Insulin signaling, glucose metabolism and mitochondria: major players in Alzheimer’s disease and diabetes interrelation. Brain Res 1441:64–78. 10.1016/j.brainres.2011.12.063 [DOI] [PubMed] [Google Scholar]
- 47.Harville EW, Viikari JSA, Raitakari OT (2011) Preconception cardiovascular risk factors and pregnancy outcome. Epidemiology 22(5):724–730. 10.1097/EDE.0b013e318225c960 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Retnakaran R, Shah BR (2021) Impact of pregnancy on the trajectories of cardiovascular risk factors in women with and without gestational diabetes. Diabetes Obes Metab 23(10):2364–2373. 10.1111/dom.14479 [DOI] [PubMed] [Google Scholar]
- 49.Gottesman RF, Albert MS, Alonso A, et al. (2017) Associations Between Midlife Vascular Risk Factors and 25-Year Incident Dementia in the Atherosclerosis Risk in Communities (ARIC) Cohort. JAMA Neurol 74(10):1246–1254. 10.1001/jamaneurol.2017.1658 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Chin HB, Baird DD, McConnaughey DR, Weinberg CR, Wilcox AJ, Jukic AM (2017) Long-term Recall of Pregnancy-related Events. Epidemiology 28(4):575–579. 10.1097/EDE.0000000000000660 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Because of participant confidentiality and privacy concerns, data cannot be shared publicly and requests to access Nurses’ Health Study II data must be submitted in writing. Further information including the procedures to obtain and access the data is described at https://www.nurseshealthstudy.org/researchers (contact email: nhsaccess@channing.harvard.edu).