Maternal Intake of Fried Foods and Risk of Gestational Diabetes Mellitus

Abstract

Purpose

We examined the relationship of maternal periconceptional (i.e., before conception and early pregnancy) intake of fried foods with GDM risk.

Methods

In a prospective birth cohort in Seattle and Tacoma, Washington State, USA, we assessed maternal periconceptional fried food intake using a food frequency questionnaire among 3,414 participants. We used multivariable generalized linear regression models to derive estimates of relative risks (RRs) (and 95% confidence intervals, 95% CI) of GDM in relation to the intake of different types of fried foods (i.e., fried fish, fried chicken, fried potatoes, chips and doughnuts).

Results

A total of 169 GDM incident cases were identified in this cohort (4.96%). Compared with no fried fish intake, fried fish intake >1 servings/month was associated with 68% higher GDM risk [adjusted RR and 95% CI; 1.68 (1.16, 2.45); P_trend=0.019]. After adjusting for confounders, the RRs (95% CI) of GDM relative to fried chicken intake were 1.0, 1.44 (0.98, 2.09) and 1.81 (1.22, 2.70) for none, ≤1 and >1 servings/month intake of fried chicken, respectively (P_trend=0.002). Dietary intake of fried potatoes, snack chips or doughnuts was not significantly associated with higher GDM risk. Limitations of our study include the lack of information about frying methods and the intake of fried foods at home and away from home.

Conclusions

Regular intake of fried fish and fried chicken are associated with elevated GDM risk.

Introduction

Fast food is a major part of the US diet [1,2]. Approximately, one in four Americans visit a fast-food restaurant daily [1,3]. Fast food restaurants commonly include fried foods in their menus such as french fries, fried chicken, fried fish, snack chips, and doughnuts, among others [3,4]. Frying is a complex cooking process that modifies the fatty acid composition of foods; and the oils used for frying increase the energy density and reduce foods’ water content [5]. Frying may also improve food palatability by making the food crunchy and aromatic, which may in turn lead to excess consumption [6]. Apart from the high-energy content, frequent consumption of fried food is considered an indicator of unhealthy lifestyle and dietary habit [7,8]. Fried food consumption has been shown to be associated with increased risks of type 2 diabetes [9], high blood pressure [10], and metabolic syndrome [11], as well as increased general and central obesity [6,10].

Gestational diabetes mellitus (GDM) is a common complication of pregnancy characterized by glucose intolerance with onset or first recognition during pregnancy [12]. Women with a history of GDM have more than a 7-fold increased risk of developing type 2 diabetes within 5 to 10 years after delivery [13]. Additionally, children born to mothers with GDM are more likely to develop type 2 diabetes later in life [14]. Risk factors for GDM include higher parity, advanced maternal age, family history of diabetes mellitus, nonwhite race/ethnicity, sleep disordered breathing, as well as pre-pregnancy overweight and obesity [15,16,17,18]. Modifiable risk factors such as diet before or during early pregnancy may serve as a target for preventing GDM. Specifically, a recent study reported that frequent fried food consumption before pregnancy was significantly associated with a greater risk of incident GDM [7]. However, few epidemiological studies have been focused on maternal outcomes associated with fried food consumption during pregnancy. Importantly, available literature, though sparse, do suggest associations of maternal habitual consumption of fried foods with adverse perinatal outcomes including infant low birth weight and preterm birth [19,20]. For example, in the European Prospective Mother-Child Study (NewGeneris), Pedersen and colleagues noted that maternal consumption of fried foods, such as fried potatoes, which are rich in acrylamide, was associated with cord blood acrylamide levels and reduced infant birth weight. Notably, a 1-unit increase in the acrylamide food score was associated with a 16 g decrease in birth weight (95% CI: −33, 1; p=0.066) [19]. Additionally, Martin and colleagues recently reported that maternal early pregnancy dietary pattern that includes fried fish and fried chicken consumption was associated with higher risk of preterm birth (OR=1.53; 95% CI: 1.02, 2.30) [20]. Given mounting epidemiologic evidence from studies of men and non-pregnant women supporting associations of fried food intake with higher risk of type 2 diabetes [8,9], and given these two recent studies conducted among pregnant women, we sought to study the extent to which, if at all, higher maternal fried food intake in the periconceptional period is associated with increased GDM risk later in the pregnancy. We hypothesized that periconceptional fried food intake is associated with an increased risk of developing GDM during the pregnancy.

RESEARCH DESIGN AND METHODS

The Omega study is a prospective pregnancy cohort designed to examine dietary risk factors of adverse pregnancy outcomes. Participants were women receiving prenatal care at clinics affiliated with the Swedish Medical Center and Tacoma General Hospital in Seattle and Tacoma, WA [21,22]. Eligible women were those who began prenatal care before 20 weeks of gestation, were able to speak and read English, were 18 years or older, and planned on delivering at either of the two hospitals. During early pregnancy (mean gestational age of 15 weeks), participants were asked to complete an interviewer-administered questionnaire. Participants also completed a 121-item previously validated semi-quantitative Food Frequency Questionnaire (FFQ) [23] to assess dietary habits in the past three months. Information on pregnancy outcomes was abstracted from medical records. Participants provided written informed consent and all procedures and study protocols were approved by the institutional review boards of the study hospitals.

Analytical Population

The analytical study population was derived from participants enrolled in the Omega Study between 1996 and 2008. During this period, 5,825 eligible women were approached and 4,602 (79%) agreed to participate. Women found to have physician-diagnosed pre-gestational diabetes (i.e., type 1, type 2 diabetes) and previous history of GDM (n=48), multi-fetal pregnancies (n=136), pregnancies lasting <20 weeks of gestation (n=45), and those with iron deficiency anemia (n=156) were excluded. We also excluded women who did not complete the FFQ (n=566), those who reported extreme levels of daily total energy intake [24] (<500 calories/day [n=27] or >3,500 calories/day [n=52]) and 158 women who moved out of the study area. A cohort of 3,414 women remained for analysis. The demographic and lifestyle characteristics of those who were included and excluded were similar (data not shown).

Data Collection

From structured questionnaire and medical records, we obtained covariate information including maternal age, educational attainment, height, pre-pregnancy weight, reproductive and medical histories, and medical histories of first-degree family members. We also collected information on maternal smoking during pregnancy. Self-reported pre-pregnancy weight and height were used to calculate pre-pregnancy body mass index (BMI: weight in kilograms divided by height in meters squared). Participants completed a self-administered, validated, and semi-quantitative food frequency questionnaire (FFQ) [23] at a mean gestational age of 15 weeks to assess periconceptional (i.e, three months before conception and up to three months post conception) diet. The WHI FFQ allows for assessment of intake, portion size, and food additives. Participants were provided clear instructions including photos of portion sizes for some of the items. The WHI FFQ has documented reliability of accurately recording intake over an extended period of observation [23]. Food composition values were obtained from the University of Minnesota Nutrition Coding Center nutrient database (Nutrition Coordinating Center, Minneapolis, MN).

We selected five fried food items measured in our study: fried potatoes, fried chicken, fried fish, doughnuts and snack chips. Fried fish was described as “fried fish, fish sandwich and fried shellfish (shrimp and oysters).” Fried potatoes included “french fries and fried potatoes.” Fried chicken was listed as “fried chicken, including nuggets and tenders”. The doughnut category included “doughnuts, cakes, pies and pastries,” and snack chips “potato chips, corn chips, cheese crackers and tortilla chips.” We assumed that all these categories comprised primarily of deep-fried foods. There were some exceptions, such as listing for “cheese crackers”; however, the contribution of these single items within its general category is likely to be small [23,25]. Maternal medical records were reviewed to collect detailed clinical information. As part of the routine antenatal follow-up of all women at participating clinics, a 50-g, 1-hr oral glucose challenge test was administered between 24 and 28 weeks of gestation to screen for GDM. Women who failed the screening test [glucose ≥ 7.8 mmol/L (≥ 140 mg/dL)] completed a diagnostic 100-g, 3-hr oral glucose tolerance test within two weeks of the screening test. According to ADA 2004 guidelines, women were diagnosed with GDM if two or more 100-g, 3-hr oral glucose tolerance test levels exceeded the following criteria: fasting ≥ 5.3 mmol/L (≥ 95 mg/dL); 1-hr ≥ 10.0 mmol/L (≥ 180 mg/dL); 2-hr ≥ 8.6 mmol/L (≥ 155 mg/dL); 3-hr ≥ 7.8 mmol/L (≥ 140 mg/dL) [26].

Statistical Analysis

We examined frequency distributions of maternal socio-demographic, reproductive, medical and dietary characteristics according to categories of total fried food and specific fried foods. We grouped total fried food consumption into three categories (<1 serving/week; 1–3 servings/week and >3 servings/week). For individual fried foods, we defined three frequency categories as follows: none; ≤1 serving/month; >1 servings/month. We then assessed GDM risk in relation to total periconceptional fried food intake and also relative to intake of individual fried food items (i.e., fried fish, fried chicken, fried potatoes, snack chips and doughnuts). To estimate relative risks (RRs) and 95% confidence intervals (95% CIs) for GDM, we fitted generalized linear models with a log-link function, Poisson family (a “log Poisson” regression model), and robust standard errors. This model allows estimation of RRs for prospective studies with binary outcome data [27]. To assess confounding, we entered covariates into each model one at a time and compared adjusted and unadjusted RRs. Final models included covariates that altered unadjusted RRs by at least 10% and those that were identified a priori as potential confounders. The following covariates were a priori considered as potential confounders based on their hypothesized role in the relationships between GDM and fried food consumption: maternal age (continuous), race/ethnicity (Non-Hispanic White, African American, Asian, Other), multiparity (yes/no), gravidity, post high school education (yes/no), smoking during pregnancy (never smoker, former or current smoker), no exercise during pregnancy (yes/no), prenatal vitamin intake (yes/no), pre-pregnancy overweight status as defined using body mass index (BMI) (<25, ≥25 kg/m²) and dietary covariates. Two multivariable models were constructed. Model 1 included maternal age, race/ethnicity, educational attainment, cigarette smoking, physical activity, family history of diabetes, total energy intake, intake of alcohol, coffee, sugar-sweetened beverages, red and processed meat, dietary magnesium and vitamin D. Model 2, included all the previously listed covariates and pre-pregnancy overweight status. In multivariable analyses, we evaluated linear trends for GDM risk by treating total or specific fried food intake as continuous variables after assigning a score to each category: none, ≤1 serving/month, >1 servings/month. All analyses were performed using Stata 12.0 statistical software (StataCorp, College Station, Texas, USA). All reported p-values are two-sided and deemed significant at α=0.05.

Results

Selected sociodemographic and lifestyle characteristics of the study cohort are summarized in Table 1. The mean age of study participants was 33 years and mean gestational age at delivery was 39 weeks. The majority of women were married (92.5%), White (86.8%), never-smokers (73.62%), and stayed physically active during pregnancy (88.5%). Women who reported higher consumption of total fried foods were younger, less likely to be nulliparous and White, and more likely to be current smokers (Table 1). Additionally, they tended to be less educated and more likely to be physically inactive during pregnancy. Mean pre-pregnancy BMI increased across categories of total fried food consumption. Women with higher total fried food consumption were more likely to consume more red and processed meats, saturated fat and drink more sugar-sweetened beverages. We observed lower intake of fruit and vegetables and total protein with increasing total fried foods intake (Table 1). Among the various types of fried foods, fried potatoes were consumed more frequently (more than one serving per month in 68.3% of participants) followed by snack chips and doughnuts (more than one serving per month in 67.8% and 42.3%, respectively) (Table 1). Consumption of fried fish was reported more frequently (>1 servings/month) than fried chicken (14.5 % and 12.1 %, respectively) (Table 1).

Table 1.

Participant characteristics according to reported fried food consumption in early pregnancy, Seattle and Tacoma, WA, Omega Cohort Study

	Total Fried Food				Fried Fish				Fried Chicken					Fried Potatoes				Snack Chips				Doughnuts
	<1 serving/ week	1–3 servings/ week	>3 servings/ week	P for trend	None	≤ 1 serving/ month	>1 servings/ month	P for trend	None	≤ 1 serving/ month	>1 servings/ month	P for trend		None	≤ 1 serving/ month	>1 servings/ month	P Value	None	≤ 1 serving/ month	>1 servings/ month	P Value	None	≤ 1 serving/ month	>1 servings/ month	P Value
Subjects													Subjects
N	793	1572	1049		2,212	708	494		2348	653	413		N	438	646	2,330		727	374	2,313		1,288	682	1,444
%	23.2	46.1	30.7		64.8	20.7	14.47		68.78	19.13	12.10		%	12.83	18.92	68.25		21.3	11.0	67.8		37.7	20.0	42.3
Maternal Age(y)	33.4±4.0	32.9±4.2	32.2±4.6	<0.001	32.9±4.3	32.8±4.5	32.5±4.3	0.085	33.1±4.13	32.5±4.42	31.8±5.0	<0.001	Maternal Age(y)	33.7±4.41	33.3±4.22	32.5±4.3	<0.001	33.2±4.11	33.0±4.44	32.7±4.36	0.001	33.0±4.3	32.9±4.2	32.6±4.4	0.042
Pre - pregnancy BMI (kg/m2)	22.8 ±4.1	23.5 ±4.7	23.9 ±4.8	<0.001	23.3±4.4	23.6±4.7	24.2±5.31	<0.001	23.1±4.23	23.8±4.74	25.0±5.92	<0.001	Pre - pregnancy BMI (kg/m2)	22.9±4.5	22.9±4.0	23.7±4.8	<0.001	23.0 4 ±4.3	23.6±4.6 1	23.6±4.70	0.014	23.1±4.3	23.4±4.3	23.8±4.9	<0.001
Gestational Age at delivery (w)	39.0±1.8	38.9 ±2.0	38.8±2.0	0.112	38.9±1.96	38.9±1.90	38.8±1.96	0.478	38.9±2.0	38.8±1.60	38.6±2.19	0.005	Gestational Age at delivery (w)	39.0±1.7 8	38.7±2.29	38.9±1.8 8	0.795	38.9±1.72	39.0±1.89	38. 9 ±2.03	0.467	38.9±1.91	38.8±2.15	38.8±1.90	0.210
White (%)	88.9	86.8	85.1	0.013	89.7	84.9	76.5	<0.001	91.23	80.9	70.7	<0.001	White (%)	88.81	88.24	86.0	0.062	82.94	88.5	87.7	0.009	88.7	86.7	85.11	0.060
Post - High School Education (%)	98.1	97.3	95.7	0.010	96.8	97.1	97.5	0.719	97.7	97.1	92.42	<0.001	Post - High School Education (%)	97.9	97.31	96.7	0.378	97.9	98.6	96.4	0.021	97.6	97.7	96.01	0.025
Married (%)	92.2	93.9	90.6	0.006	93.04	92.5	89.9	0.055	93.5	92.04	87.41	<0.001	Married (%)	92.7	93.34	92.2	0.605	92.2	92.0	92.7	0.844	92.24	94.0	92.0	0.236
Nulliparous (%)	66.1	61.4	61.0	0.047	62.84	59.8	64.0	0.244	65.7	56.51	52.8	<0.001	Nulliparous (%)	71.23	61.8	60.9	<0.001	62. 6 0	63.90	62.04	0.780	65.7	61.88	59.63	0.005
Smoked during pregnancy (%)	3.53	5.12	7.67	0.002	4.66	7.31	6.95	0.039	4.78	5.88	9.39	0.001	Smoked during pregnancy (%)	4.09	3.33	6.42	0.014	4.96	5.03	5.80	0.504	4.41	3.64	7.44	<0.001
Physically activity (%)	90.4	89.2	86.1	0.011	88.3	89.53	88.2	0.652	89.4	87.1	85.9	0.069	Physically activity (%)	89.7	89.13	88.12	0.575	89.55	89.9	87.96	0.370	89.11	89.2	87.7	0.427
Family history of diabetes (%)	10.8	13.7	15.3	0.022	12.97	12.57	17.41	0.024	11.88	18.22	15.50	<0.001	Family history of diabetes (%)	11.2	11.61	14.51	0.050	12.93	14.44	13.58	0.782	12.66	14.1	14.1	0.507
Total energy intake (kcal)	1484±505	1675 ±528	1980±548	<0.001	1678±541	1776±577	1856±593	< 0.001	1706±545	1704±573	1856±612	<0.001	Total energy intake (kcal)	1557±512	1617±529	1785±567	<0.001	1566±573	1657±543	1784±549	<0.001	1580±543	1677±504	1874±564	<0.001
% calories from total fat	27.5±6.80	30.5 ±5.53	33.2±5.4	<0.001	29.8±6.3	31.6±5.7	32.8±5.56	<0.001	57.0±22.6	60.2±23.5	68.1±25.7	<0.001	% calories from total fat	47.6±19.5	52.7±20.8	62.8±23.7	<0.001	49.4±22.5	54.8±20.7	62.6±23.2	<0.001	51.6±21.8	57.0±20.4	66.4±23.9	<0.001
% calories from satured fat	10.0±3.14	11.2 ±2.7	12.0±2.6	<0.001	10.9±2.94	11.5±2.79	11.6±2.64	<0.001	10.9±2.94	11.51±2.8	11.7±2.47	<0.001	% calories from satured fat	10.0±3.06	10. 7 ±2.95	11. 5 ±2.8	<0.001	10.3 3 ±3.2	10.9±2.79	11.5±2.75	<0.001	10.6±3.09	11.1±2.70	11.7±2.67	<0.001
% calories from protein	17.8±3.20	17.5 ±2.70	16.7±2.5	<0.001	17.3±2.95	17.3±2.40	17.4±2.53	0.503	17.3±2.8	17.32±2.7	17.40±2.6	0.843	% calories from protein	17.91±3.3	17.51±2.87	17.2±2.65	<0.001	17.8±3.16	17.84±2.7	17.1±2.65	<0.001	17.3±3.02	17.3±2.71	17.0±2.56	<0.001
% calories from carbohydrates	56.25±8.5	53.1 ±6.90	51.0±6.36	<0.001	54.14±7.5	52.1±6.62	50.6±6.8	<0.001	54.0±7.39	52.0±7.27	50.53±6.6	<0.001	% calories from carbohydrates	56.0±8. 60	54.62±7.55	52.3±6. 90	<0.001	55.3±8.61	53.4±7.40	52.5±6.81	<0.001	54.2±8.12	53.2±7.4	52.3±6.51	<0.001
Alcohol (g/day)*	0.60 ±2.8	0.82 ±3.83	0.87 ±3.5	0.123	0.73±3.68	0.79±3.0	1.02±3.48	0.125	0.86±3.86	0.71±2.75	0.48±2.27	0.036	Alcohol (g/day)*	0.65±2.52	0.61±3.32	0.86±3.71	0.118	0.77±3.86	0.78±3.3	0.79±3.43	0.882	0.74±3.75	0.87±3.56	0.78±3.25	0.790
Total fiber (g/day)*	18.12±7.7	18.3 ±7.50	19.1±7.01	0.004	18.7±7.44	18.3±7.31	18.2±7.4	0.111	19±7.47	17.5±7.08	17.3±7.2	<0.001	Total fiber (g/day)*	19.01±7.8	18.4±7. 20	18.5±7.40	0.280	18.2±7.95	18.6±7.65	18.62±7.2	0.186	18.01±7.7	18.5±7.36	19.0±7.12	0.001
Red and Processed meat (serv/day)	0.44±0.40	0.62 ±0.43	0.81±0.53	<0.001	0.59±0.47	0.71±0.46	0.77±0.52	<0.001	0.37±0.32	0.48±0.34	0.57±0.39	<0.001	Red and Processed meat (serv/day)	0.46±0.44	0.51±0.42	0.71±0.49	<0.001	0.53±0.46	0.59±0.46	0.68±0.48	<0.001	0.55±0.45	0.62±0.46	0.73±0.50	<0.001
Fruits and Vegetables (serv/day)	4.61±2.33	4.32 ±2.28	4.07±2.07	<0.001	4.39±2.26	4.18±2.19	4.13±2.18	0.005	4.46±2.24	4.0±2.14	4.0±2.32	<0.001	Fruits and Vegetables (serv/day)	4.90±2.42	4.50±2.24	4.15±2.18	<0.001	4. 50 ±2.48	4.46±2.03	4.23±2.19	0.003	4.43±2.41	4.32±2.10	4.20±2.15	0.005
Sugar Sweetened Beverages (serv/day)	0.07±0.26	0.11 ±0.25	0.23±0.44	<0.001	0.13±0.34	0.15±0.31	0.18±0.30	0.001	0.12±0.32	0.15±0.31	0.22±0.41	<0.001	Sugar Sweetened Beverages (serv/day)	0.07±0.33	0.09±0.26	0.17±0.34	<0.001	0.09±0.29	0.10±0.23	0.16±0.35	<0.001	0.11±0.28	0.13±0.40	0.17±0.33	<0.001

Data are Means ± SD or Percentage.

^*Energy adjusted (2,000 kcal/day). For continuous variables, P-value was calculated using the one-way ANOVA; for categorical variables, P-value was calculated using the Chi-square test.

A total of 169 incident GDM cases were identified in this cohort (4.96%). Total fried food intake (i.e., fried fish, fried chicken, fried potatoes, doughnuts and snack chips combined) was not statistically significantly associated with higher GDM risk. After adjustment, the RR of GDM were 1.0, 0.92 (95% CI 0.61, 1.38) and 1.26 (95% CI 0.79, 2.01) for <1 serving/week, 1–3 servings/week, >3 servings/week (P for trend=0.30) (Table 2). When we analyzed specific fried foods, the RR of GDM among women who consumed more than one serving per month of fried fish was 1.73 (95% CI 1.19, 2.52) compared with those who reported no consumption of fried fish (P for trend=0.013) after adjustment for age, race/ethnicity, educational attainment, smoking status, physical activity, family history of diabetes, alcohol, coffee, sugar-sweetened beverages, red and processed meat, total energy intake, and other dietary covariates (Table 3). Similarly, the adjusted RR of GDM among women who consumed fried chicken > 1 servings/month was 2.06 (95% CI 1.39, 3.05) compared with those who reported no fried chicken consumption (P for trend <0.001) (Table 3). These associations remained significant after additional adjustment for pre-pregnancy overweight status (Table 3). We assessed GDM risk in those women who consumed more than one serving per week of fried fish or fried chicken as compared with no intake of either type of fried food. The intake of fried fish [(RR and 95% CI: 2.66 (1.45, 4.87)] and fried chicken [(RR and 95% CI: 2.42 (1.14, 5.15)] was associated with increased GDM risk (Supplemental Table 1). Maternal consumption of fried potatoes, snack chips, and doughnuts did not appear to be associated with statistically significant increases in GDM risk (Table 3, Supplemental Table 1). We conducted a sensitivity analysis including vegetables and nuts intake in the logistic models, and these food items were not confounders in the association GDM risk and fried food intake (data not shown).

Table 2.

Relative risks (RR) and 95% confidence intervals (95% CI) for developing GDM according to total fried food consumption, Seattle and Tacoma, WA, Omega Cohort Study

	<1 serving/week	1–3 servings/week	>3 servings/week	P for trend
Total Fried Food consumption
N	793	1572	1049
Cases (%)	38 (4.79)	71 (4.52)	60 (5.72)
Energy Adjusted RR (95% CI)	1.00 (Ref)	0.97 (0.66, 1.40)	1.30 (0.85, 1.99)	0.209
1Adjusted RR (95% CI)	1.00 (Ref)	0.96 (0.64, 1.44)	1.32 (0.83, 2.11)	0.228
2Adjusted RR (95% CI)	1.00 (Ref)	0.92 (0.61, 1.38)	1.26 (0.79, 2.01)	0.300

¹Adjusted for daily energy intake, maternal age, race/ethnicity, educational attainment, family history of diabetes, physical activity, alcohol, coffee, sugar-sweetened beverages, red and processed meats, calcium, dietary magnesium and vitamin D intake.

²Further adjusted for pre-pregnancy overweight status.

Table 3.

Relative risks (RR) and 95% confidence intervals (95% CI) for developing GDM according to type of deep-fried foods, Seattle and Tacoma, WA, Omega Cohort Study

	None	≤1 serving/month	>1 servings/month	P - trend
Fried fish
N	2,212	708	494
Cases (%)	97 (4.39)	32 (4.52)	40 (8.10)
Energy Adjusted RR (95% CI)	1.00 (Ref)	1.05 (0.71, 1.55)	1.89 (1.32, 2.71)	0.003
1 Adjusted RR (95% CI)	1.00 (Ref)	1.01 (0.67, 1.54)	1.73 (1.19, 2.52)	0.013
2 Adjusted RR (95% CI)	1.00 (Ref)	0.99 (0.66, 1.49)	1.68 (1.16, 2.45)	0.019
Fried chicken
N	2, 348	653	413
Cases (%)	94 (4.0)	40 (6.13)	35 (8.5)
Energy Adjusted RR (95% CI)	1.00 (Ref)	1.53 (1.07, 2.19)	2.16 (1.48, 3.14)	<0.001
1 Adjusted RR (95% CI)	1.00 (Ref)	1.48 (1.02, 2.16)	2.06 (1.39, 3.05)	<0.001
2 Adjusted RR (95% CI)	1.00 (Ref)	1.44 (0.98, 2.09)	1.81 (1.22, 2.70)	0.002
Fried potatoes
N	438	646	2330
Cases (%)	24 (5.48)	32 (4.95)	113 (4.85)
Energy Adjusted RR (95% CI)	1.00 (Ref)	0.91 (0.54, 1.52)	0.90 (0.59, 1.39)	0.683
1 Adjusted RR (95% CI)	1.00 (Ref)	0.97 (0.56, 1.67)	0.93 (0.58, 1.47)	0.710
2 Adjusted RR (95% CI)	1.00 (Ref)	1.02 (0.59, 1.75)	0.93 (0.58, 1.47)	0.662
Snack chips
N	727	374	2313
Cases (%)	38 (5.23)	22 (5.88)	109 (4.71)
Energy Adjusted RR (95% CI)	1.00 (Ref)	1.13 (0.68, 1.90)	0.92 (0.63, 1.33)	0.548
1 Adjusted RR (95% CI)	1.00 (Ref)	1.16 (0.67, 2.01)	1.04 (0.71, 1.54)	0.897
2 Adjusted RR (95% CI)	1.00 (Ref)	1.08 (0.62, 1.87)	1.03 (0.70, 1.52)	0.894
Doughnuts
N	1,288	682	1444
Cases (%)	64 (4.97)	37 (5.43)	68 (4.71)
Energy Adjusted RR (95% CI)	1.00 (Ref)	1.10 (0.74, 1.64)	0.97 (0.69, 1.37)	0.868
1 Adjusted RR (95% CI)	1.00 (Ref)	1.06 (0.70, 1.56)	0.85 (0.59, 1.22)	0.363
2 Adjusted RR (95% CI)	1.00 (Ref)	1.04 (0.69, 1.55)	0.83 (0.58, 1.19)	0.306

¹Adjusted for daily energy intake, maternal age, race/ethnicity, educational attainment, family history of diabetes, physical activity, alcohol, coffee, sugar-sweetened beverages, red and processed meats, total fiber intake, fatty fish, calcium, dietary magnesium and vitamin D intake.

²Further adjusted for pre-pregnancy overweight status.

Discussion

In this large pregnancy cohort, we found that increasing periconceptional (i.e., pre-pregnancy and early pregnancy) fried fish and fried chicken consumption is associated with increased risk of GDM, even after adjustment for other GDM risk factors such as maternal age, pre-pregnancy overweight status, family history of diabetes, physical activity during pregnancy, and other dietary characteristics. However, total fried food consumption was not significantly associated with increased GDM risk.

To our knowledge, this is the first study that evaluated relationships between the types of fried food consumption in the periconceptional period and incident GDM risk. In a previous study examining the association of pre-pregnancy fried food consumption and risk of GDM, no information on the types of individual fried foods was available [7]. Among men and non-pregnant women, findings from studies of fish intake and type 2 diabetes had conflicting results that vary by geographic areas [28], type of fish consumption [29,30] and cooking method [30]. Our results are consistent with some, but not all, previous studies assessing type 2 diabetes and fried fish intake in men and non-pregnant women. For example, in a population-based cohort of Swedish men, fried fish intake (≥6 servings/month vs. ≤1 servings/month) was significantly associated with higher risk of type 2 diabetes, (hazard ratio=1.14 ; 95% CI: 1.03, 1.31) [31]. In another study, African American women who consumed fried fish 1 to 3 times per month had 21 % higher risk of type 2 diabetes compared to women that never consumed fried fish (IRR: 1.21; 95% CI: 1.07, 1.37) [8]. However, Patel and coworkers (2009), in the European Prospective Investigation of Cancer [EPIC]-Norfolk, found no increase in risk of type 2 diabetes among men and women who consumed fried fish [(≥ 1 vs. <1 portion/week) (OR=0.91; 95% CI: 0.75, 1.10)] after adjustment for confounders [30].

In our study cohort, fried chicken intake was associated with increased GDM risk even after we controlled for a number of confounders. Consistent with our results, previous studies have documented associations of fried chicken consumption with higher risk of type 2 diabetes in African American women [(≥2 times per week vs. none) (IRR=1.68; 95%CI: 1.36, 2.08)] [8]. Furthermore, Native Americans diagnosed with diabetes were more likely to report higher intake of fried chicken and fried fish in the Inter-Tribal Heart Project –a cross-sectional study [32].

Although biological mechanisms underlying the observed association of fried food consumption with GDM risk are unknown, some authors have hypothesized that increased formation of trans-fatty acids [33] and advanced glycoxidation end products (AGEs) during the frying process [34] may play a role [35,36]. This thesis is supported by evidence from experimental studies showing reduced insulin sensitivity with increased intake of trans fatty acids [37]. Furthermore, investigators have shown that AGEs contribute to cellular oxidative damage and chronic systemic inflammation [38,39] which have been implicated in insulin resistance, pancreatic cell damage and diabetes [36,40].

Potential benefits of fish consumption in relation to insulin resistance and type 2 diabetes have been suggested to be mediated through long-chain n-3 fatty acids, which have inhibitory effects on inflammatory pathways and activation of peroxisome proliferator-activated receptors (PPARs) [41]. However, frying may lead to degradation of long-chain n-3 fatty acids and increase in other fatty acids including trans-fatty acids [42]. Of note, fried fish is usually lean fish (i.e. bass, cod, flounder), and contains lower amounts of n-3 fatty acids than oily fish (i.e. salmon, trout, and herring) [43]. Additionally, environmental contaminants such as polychlorinated biphenyls (PCBs) and methylmercury (MeHg) present in fish might increase the risk of diabetes [ 44, 45].

The present study has several strengths and limitations. We used a large well-characterized cohort of pregnant women. Additionally, structured interviews, medical record abstractions, and the validated semi-quantitative FFQ provide high-quality data to evaluate the relationships and adjust for potential confounding factors. Estimating the risk of GDM in relation to specific types of fried foods commonly consumed during pregnancy is a key strength of our study, a consequence of the detailed dietary information we collected. Limitations of our study include the lack of information about serving size or frying methods (e.g., deep or pan frying; fresh or reused oil; duration and temperature, types of frying oil [i.e. olive, corn, vegetable, etc.]), and the intake of fried foods at home and away from home. Even though we adjusted our analysis for unhealthy lifestyle and dietary habits, we could not exclude the possibility of residual confounding as in all observational studies.

Conclusions

Our findings suggest that higher levels of maternal periconceptional fried fish and fried chicken, are associated with increased GDM risk. Additional studies are needed to confirm these findings and elucidate whether the association we observed between fried food consumption and risk of GDM are attributable to fried food intake away from home or unhealthy lifestyles.

Regardless, from a clinical and public health perspective, we have identified a risk factor for GDM that may be readily modifiable by lifestyle or cooking choices that lead to the consumption of less fried foods, especially fried chicken and fried fish.