Identifying Distal and Ultimate-Level Correlates of Gestational Diabetes Among Latinas in California

Introduction: Gestational diabetes mellitus (GDM) is a pregnancy-induced hyperglycemia that occurs in as many as 9.2% of all U.S. pregnancies, and poses significant health threats to mother and baby (DeSisto, 2014). GDM disproportionately affects Latinas, yet research addressing disparities has primarily focused on intrapersonal correlates (e.g., genetic and behavioral risk factors; Chasen-Taber, 2010). As such, there is a need to examine broader influences of GDM among Latinas.  

Methods: This study was a cross-sectional, secondary analysis of data from the 2011-2012 California Health Interview Survey [CHIS]. The analytic sample contained 5,531 Latinas in California. A nested logistic regression model was estimated whereby predictors of GDM were grouped into blocks based on the community (e.g., produce accessibility, produce affordability and neighborhood safety), interpersonal (e.g., physician listening and difficulty understanding physician) and intrapersonal (e.g., diet, physical activity and education) levels of the Social Ecological Model. Supplemental analyses were conducted on only Latinas born in Mexico (N = 2,620). Model fit, contribution of each block, and significance of predictors were explored.

Results: In the community level-only block of the analysis, neighborhood safety was significantly associated with history of GDM. In the second block containing community and interpersonal variables, neighborhood safety remained statistically significant.  Additionally, difficulty understanding the physician was significantly associated with GDM. Within the final block including all levels, difficulty understanding the physician (AOR = 2.04 (1.18, 3.53), p = .011), overweight status, increased produce consumption (an unexpected result) and being foreign-born were associated with GDM; each block contributed significantly to the model.  In the supplemental analysis, none of the intrapersonal correlates were significant in the final model, while neighborhood safety, and difficulty understanding the doctor were significant.

Conclusion: Results of these analyses reveal the complexities of GDM etiology among Latinas. The effect of neighborhood safety was attenuated in the final model, suggesting this association may be mediated by intrapersonal variables. Difficulty understanding one’s physician remained significant across all models, suggesting a direct effect on GDM. While the results of the analyses including intrapersonal variables paralleled previous GDM research, novel correlates of GDM were identified by drawing from a broader model. The supplemental analysis exposed the possible moderating role being Mexican-born has on gestational diabetes, and should be investigated in future research.