Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Higher Global Diet Quality Score Is Inversely Associated with Risk of Type 2 Diabetes in US Women

Higher Global Diet Quality Score Is Inversely Associated with Risk of Type 2 Diabetes in US Women The Journal of Nutrition Supplement Higher Global Diet Quality Score Is Inversely Associated with Risk of Type 2 Diabetes in US Women 1,2 2 2,3 2 4 Teresa T Fung, Yanping Li, Shilpa N Bhupathiraju, Sabri Bromage, Carolina Batis, Michelle 3,5 3,5 2,3 6 2,3 D Holmes, Meir Stampfer, Frank B Hu, Megan Deitchler, and Walter C Willett 1 2 Department of Nutrition, Simmons University, Boston, MA, USA; Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA; Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; CONACYT—Health and Nutrition Research Center, National Institute of Public Health, Cuernavaca, Mexico; 5 6 Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; and Intake—Center for Dietary Assessment, FHI Solutions, Washington, DC, USA ABSTRACT Background: We have developed a diet quality metric intended for global use. To assess its utility in high-income settings, an evaluation of its ability to predict chronic disease is needed. Objectives: We aimed to prospectively examine the ability of the Global Diet Quality Score (GDQS) to predict the risk of type 2 diabetes in the United States, examine potential differences of association by age, and compare the GDQS with other diet quality scores. Methods: Health, lifestyle, and diet information was collected from women (n = 88,520) in the Nurses’ Health Study II aged 27–44 y at baseline through repeated questionnaires between 1991 and 2017. The overall GDQS consists of 25 food groups. Points are awarded for higher intake of healthy groups and lower intake of unhealthy groups (maximum of 49 points). Multivariable HRs were computed for confirmed type 2 diabetes using proportional hazards models. We also compared the GDQS with the Minimum Diet Diversity score for Women (MDD-W) and the Alternate Healthy Eating Index-2010 (AHEI-2010). Results: We ascertained 6305 incident cases of type 2 diabetes during follow-up. We observed a lower risk of diabetes with higher GDQS; the multivariable HR comparing extreme quintiles of the GDQS was 0.83 (95% CI: 0.76, 0.91; P- trend < 0.001). The magnitude of association was similar between women aged <50 y and those aged ≥50 y. An inverse association was observed with lower intake of unhealthy components (HR comparing extreme quintiles of the unhealthy submetric: 0.76; 95% CI: 0.69, 0.84; P-trend < 0.001) but was not with the healthy submetric. The inverse association for each 1-SD increase in the GDQS (HR: 0.93; 95% CI: 0.91, 0.96) was stronger (P < 0.001) than for the MDD-W (HR: 1.00; 95% CI: 0.94, 1.04) but was slightly weaker (P = 0.03) than for the AHEI-2010 (HR: 0.91; 95% CI: 0.88, 0.94). Conclusions: A higher GDQS was inversely associated with type 2 diabetes risk in US women of reproductive age or older, mainly from lower intake of unhealthy foods. The GDQS performed nearly as well as the AHEI-2010. J Nutr 2021;151:168S–175S. Keywords: diet quality, diabetes, epidemiology, women, nutrition Introduction several chronic diseases, including cardiovascular disease and diabetes (1). Several diet quality indices have been developed and evaluated To apply these diet quality indices in clinical and public for their association with risk of chronic diseases (1). These health settings to guide individual dietary choices and public indices typically were based on recommendations for a healthy health surveillance, the metric must be simple and quick to diet (2–4) or reflections of regional dietary habits ( 5–7). Many administer. In addition, a metric that is valid and practical for include nutrient components and therefore require the use of use across different parts of the world and different economic a food composition database (2–4), or a scoring algorithm that development levels would have the additional advantage of is based on population-specific intake levels ( 5, 8). Evidence enabling global comparisons. Therefore, indices that involve a from prospective studies is consistent that adherence to food composition database or use population-specific scoring these diet quality indices is associated with a lower risk of would be difficult to implement across regions. To circumvent The Author(s) 2021. Published by Oxford University Press on behalf of the American Society for Nutrition. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Manuscript received February 27, 2021. Initial review completed March 29, 2021. Revision accepted May 25, 2021. 168S First published online October 1, 2021; doi: https://doi.org/10.1093/jn/nxab195. these limitations, we previously developed the Prime Diet Methods Quality Score (PDQS) that only consists of food groups. It is Participants inversely associated with cardiovascular disease and gestational The Nurses’ Health Study II (NHS II) is an ongoing prospective cohort diabetes in US men and women (9, 10). study that is comprised of 116,430 US female Registered Nurses To provide a metric that is usable in regions where between 25 and 42 y old at inception in 1989 (16). Information on nutritional adequacy is a concern, we have further modified lifestyle practices and incidence of type 2 diabetes was collected every 2 y the PDQS and tested it for association with nutritional by self-reported questionnaires. Diet was assessed every 4 y beginning in markers relevant to middle- and lower-income countries. Our 1991 using a validated FFQ. Women with diabetes, gestational diabetes, cancer, or cardiovascular disease or who died before the rst fi dietary final metric, the Global Diet Quality Score (GDQS), uses a assessment were excluded. We also excluded those who did not complete combination of healthy and unhealthy food groups. It has additional questionnaires beyond baseline and those who reported reasonable correlation with measures of nutrient adequacy (11). implausible energy intakes (<500 or >3500 kcal/d) at baseline. If a Because the GDQS has several differences from the PDQS, participant reported being pregnant in a questionnaire period, person- we assessed its utility in a higher-income setting by testing its time during that 2-y period was excluded. A total of 88,520 women ability to predict the risk of type 2 diabetes in US women. were included in this analysis and loss to follow-up was ∼10% during We chose type 2 diabetes because the incidence is increasing the study period. This study was approved by the institutional review globally, with a projected increase from >400 million affected in boards of Brigham and Women’s Hospital and the Harvard TH Chan 2019 to ∼700 million by 2045 (12). In the United States, it was School of Public Health. estimated that 12% of adult women and 14% of adult men were living with diabetes in 2013–2016 (13). Although a plethora of Diet assessment medications are available (14), there is no cure in most cases A validated semiquantitative FFQ was self-administered every 4 y, each and successful management requires adequate compliance and containing ∼135 items (17). For each food item, a standard portion regular access to health care (15). Therefore, prevention through size was provided with 9 intake frequency choices ranging from “never lifestyle, and especially diet, continues to be an important or less than once per month” to “≥6 times per day.” The GDQS approach. In this analysis, we prospectively examined the asso- was modified based on the PDQS ( 9) to capture food groups that would reflect nutrient adequacy and predict major noncommunicable ciation between the GDQS and the risk of type 2 diabetes among diseases in both lower- and high-income countries globally. It consists US women, and explored potential differences in association by of 16 healthy food groups (dark green leafy vegetables, cruciferous age. To understand the function of the GDQS, we also explored vegetables, deep orange vegetables, other vegetables, deep orange fruits, how the healthy and unhealthy components would drive any ob- deep orange tubers, citrus fruits, other fruits, legumes, nuts and seeds, served association. We hypothesized that the overall GDQS and poultry and game meats, fish and shellfish, whole grains, liquid oils, low the healthy components (GDQS+ submetric) would be inversely fat dairy, eggs) and 7 unhealthy food groups (white roots and tubers, associated with diabetes risk, whereas lower intake of the un- processed meats, refined grains and baked goods, sugar-sweetened healthy components (GDQS− submetric) would have an inverse beverages, sweets and ice cream, juices, purchased deep fried foods) association. For the GDQS to be a useful nutrition metric to (Supplemental Table 1). Intake of each food group was classified into predict noncommunicable diseases, it must also perform at least <1/wk, 1 to <4/wk, and ≥4/wk. For healthy food groups, points between 0 and 4 were given to each level of intake depending on the similarly as other established diet quality indices. Therefore, food group. For unhealthy food groups, 2, 1, and 0 points were given we also compared it with the Minimum Diet Diversity score for the same 3 intake levels so lower intake would receive more points. for Women (MDD-W) and the Alternate Healthy Eating Index- In addition to the aforementioned food groups, the GDQS also has a red 2010 (AHEI-2010) for prediction of type 2 diabetes. meat group and a full-fat dairy group with different scoring to account for their contribution to nutrient adequacy in low- to middle-income Funding for the research was provided by FHI Solutions, recipient of a Bill countries. Red meat was given 0, 1, and 0 points for intake of the same & Melinda Gates Foundation grant to support Intake—Center for Dietary 3 intake levels as for the other unhealthy food groups, and full-fat dairy Assessment and by NIH grant U01 CA176726. was given 0, 1, 2, and 0 points for intake of <1/wk, 1 to <4/wk, ≥4/wk Author disclosures: TTF is an Associate Editor for the Journal of Nutrition and to <3/d, and ≥3/d, respectively. The full GDQS has 25 food groups and played no role in the Journal’s evaluation of the manuscript. All other authors a score range of 0–49 points, with higher points representing a healthier report no conflicts of interest. diet. The healthy portion of the GDQS (GDQS+) has a range of 0–32. The content is solely the responsibility of the authors and does not necessarily For the purpose of this analysis, we included red meat and full-fat dairy represent the official views of the NIH. as part of the unhealthy portion (GDQS−), which has a range of 0–17, Supplemental Tables 1–5 are available from the “Supplementary data” link in with a higher score representing lower intake of unhealthy foods and the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn. hence healthier food choices. Published in a supplement to The Journal of Nutrition. Publication costs for this To compare the GDQS with other established diet quality indices, supplement were funded by the Bill & Melinda Gates Foundation in conjunction we also computed the AHEI-2010 (2) and the MDD-W (18) for each with FHI Solutions, recipient of a Bill & Melinda Gates Foundation grant to participant. The AHEI-2010 consists of 11 food and nutrient groups. support Intake—Center for Dietary Assessment. The opinions expressed in this High points are given for higher intakes of healthy groups (vegetables, publication are those of the authors and are not attributable to the sponsors whole fruits, nuts and legumes, whole grains, polyunsaturated fat, or the publisher, Editor, or Editorial Board of The Journal of Nutrition.The and long-chain n–3 fatty acids) and lower intakes of unhealthy Supplement Coordinators for the supplement publication were Megan Deitchler, groups (red and processed meats, sugar-sweetened beverages and fruit Intake—Center for Dietary Assessment at FHI Solutions, Washington, DC; and juice, trans fat, and sodium). Points are also given for moderate Sabri Bromage, Harvard TH Chan School of Public Health, Boston, MA. The GDQS research initiative was launched by Intake – Center for Dietary intake of alcohol. Each component ranges from 0 to 10 points Assessment. The research was led by Harvard T.H. Chan School of Public Health, with the total possible score ranging from 0 to 110 points. It has Department of Nutrition and carried out in collaboration with researchers at the previously been shown to be inversely associated with diabetes risk in National Public Health Institute (INSP), Mexico. Funding for the research was women (2). provided by FHI Solutions, recipient of a Bill & Melinda Gates Foundation grant The MDD-W, originally developed as a proxy indicator for nutrient to support Intake – Center for Dietary Assessment. adequacy, consists of 10 food groups: grains and starchy vegetables, Address correspondence to TTF (e-mail: [email protected]). pulses, nuts and seeds, dairy, animal flesh, eggs, dark green leafy Abbreviations used: AHEI-2010, Alternate Healthy Eating Index-2010; GDQS, vegetables, vitamin A–rich vegetables and fruits, other vegetables, and Global Diet Quality Score; MDD-W, Minimum Diet Diversity score for Women; other fruits (18). The scoring method for the original MDD-W is based MET, metabolic equivalent hour; PDQS, Prime Diet Quality Score. GDQS and diabetes risk 169S on intake collected by 24-h recall. To adapt it for the FFQ, we assigned the analysis by BMI status and physical activity. To examine the potential influence of pregnancy on the association between the 1 point for each food group with intake ≥1 serving/d and 0 for less (9). The MDD-W has a range of 0–10 points. GDQS and diabetes, we ran regression models separately for women based on pregnancy history, and among ever-pregnant women by history of gestational diabetes. Tests for 2-way interaction between Outcome assessment the GDQS and each of the stratified factors were conducted using the Incident type 2 diabetes was rst fi reported through the biennial ques- likelihood ratio test comparing regression models with and without an tionnaires and confirmed with a validated supplemental questionnaire interaction term. Analysis was conducted using SAS version 9.4 (SAS based on National Diabetes Data Group criteria. This included ≥1of Institute Inc.). the following: ≥1 classic symptom (excessive thirst, polyuria or frequent To compare the strength of association between the GDQS and the urination, weight loss, hunger), fasting plasma glucose concentrations AHEI-2010 and MDD-W, we standardized each score and modeled each ≥7.8 mmol/L, or random plasma glucose concentrations ≥11.1 mmol/L 1 SD of the scores in the same model. Differences in the regression (19). In the case of a lack of symptoms, diabetes was considered coefficients were compared using the Wald test. confirmed with ≥2 elevated plasma glucose concentrations on different occasions (fasting plasma glucose concentrations ≥7.8 mmol/L, random concentrations ≥11.1 mmol/L, and/or 2-h blood glucose concentrations ≥11.1 mmol/L during oral-glucose-tolerance testing); or treatment Results with hypoglycemic medications (insulin or oral hypoglycemic agent). For cases reported after 1998, criteria from the American Diabetes In ≤26 y of follow-up, we ascertained 6305 incident cases of Association were used in which the threshold for fasting plasma glucose type 2 diabetes, of which 2266 were women younger than 50 changed from 7.8 mmol/L to 7.0 mmol/L (20). The supplemental y old and 4039 were women ≥50 y old. Women with a higher questionnaire was validated by a review of medical reports (21). In GDQS tended to be leaner, more physically active, less likely a random sample of 62 cases in the Nurses’ Health Study that were to be current smokers, and consumed more alcohol and coffee confirmed by the supplementary questionnaire, 61 (98%) cases were (Table 1). reconfirmed after medical records were reviewed by an endocrinologist We observed a lower risk of diabetes with higher GDQS blinded to the supplementary questionnaire. (multivariable HR comparing extreme quintiles: 0.83; 95% CI: 0.76, 0.91; P-trend < 0.001) (Table 2). The association Covariate assessment for women age <50 y was 0.85 (95% CI: 0.73, 0.98; P- Information on age, race, and height was collected at cohort inception. trend < 0.001) and for age ≥50 y was 0.82 (95% CI: 0.74, Body weight, cigarette smoking (including the number of cigarettes 0.91, P-trend < 0.001) with no significant interaction. We also per day), physical activity, menopausal status and postmenopausal hormone use, oral contraceptive use, family history of diabetes, history separately examined the submetrics of the GDQS representing of hypercholesterolemia, and high blood pressure were collected in each healthy (GDQS+) and unhealthy (GDQS−) food components. biennial questionnaire. BMI (in kg/m ) was calculated using height These 2 submetrics were only weakly correlated (Spearman r collected at baseline and weight reported at each questionnaire cycle. =−0.06, P < 0.001). The healthy components of the GDQS Alcohol intake and supplemental vitamin and mineral use were collected (GDQS+) were not associated with diabetes risk (Table 3). On with FFQs. the other hand, higher GDQS−, which represents lower intake of the unhealthy components, showed an inverse association Statistical analysis (multivariable HR comparing extreme quintiles: 0.76; 95% CI: For this analysis, follow-up duration in person-years was calculated 0.69, 0.84; P-trend < 0.001). There was no apparent difference from the date of return of the 1991 questionnaire to the date of diabetes in association by age. Spline regression did not detect significant diagnosis, last questionnaire returned, or 30 June, 2017. We computed departure from linearity for the overall GDQS, GDQS+, cumulative averages of diet quality scores to reduce within-person or GDQS− (data not shown). In the sensitivity analysis in variation and represent long-term intake (22). We used time-dependent which we excluded the egg component from the GDQS+, Cox proportional hazards regression models to compute HRs of type the null association persisted in the remaining portion of the 2 diabetes for quintiles of the GDQS, GDQS+, and GDQS−. Eggs are GDQS+. included in the GDQS+ because of their protein and vitamin content, but they also contain substantial amounts of cholesterol. Hence, we in The GDQS was inversely associated with diabetes in both addition computed an alternate GDQS+ without the egg component for women ever or never pregnant (Supplemental Table 2). Al- sensitivity analysis. We tested for the proportional hazards assumption though the magnitude of association did not differ substantially by including an interaction term of GDQS and age (which reflects for pregnancy history, the trend appeared to be more consistent time) and used the likelihood ratio test. The P value for the chi- for never-pregnant women (P-interaction = 0.06). Among square distribution was >0.05, hence it did not show a violation of the women who had been pregnant, an inverse association with proportional hazards assumption. the GDQS was only observed for those without a history All models were adjusted by age (mo) at the start of follow-up of gestational diabetes (multivariable HR comparing extreme for each woman and the calendar year of each questionnaire cycle. quintiles: 0.83; 95% CI: 0.75, 0.91; P-trend < 0.001). We Multivariable models were adjusted for race (white/nonwhite), family also stratified the analysis by BMI and physical activity history of diabetes, smoking (never, past, 1–14 cigarettes/d, 15–24 cigarettes/d, ≥25 cigarettes/d), alcohol intake (none, <5g/d, 5 to (Supplemental Table 3). The inverse association was significant <10 g/d, ≥10 g/d), energy intake (quintiles), coffee intake (continuous), regardless of BMI status; however, it was stronger among physical activity [<3 metabolic equivalent hours (METs)/wk, 3 to leaner women (P-interaction < 0.001). On the other hand, <9METs/wk, 9 to <18 METs/wk, 18 to <27 METs/wk, ≥27 although the association between the GDQS and diabetes METs/wk], BMI (<23, 23 to < 25, 25 to <30, 30 to <35, ≥35), appeared stronger among those with physical activity above multivitamin use (yes/no), menopausal status and menopausal hormone the median, the P value for interaction did not reach statistical therapy (premenopausal, no hormone use, past use, current use), oral significance. contraceptive use (never, past, current), history of hypertension at We also compared the magnitude of association of the GDQS baseline, and history of hyperlipidemia at baseline. We used restricted with 2 other diet quality scores: the AHEI-2010 and MDD-W. cubic spline regression to assess potential nonlinear association. To The Spearman correlation coefficient between the GDQS and access potential differential association of the GDQS with diabetes by age, we conducted analyses stratified by age. We also stratified the AHEI-2010 was 0.74 (P < 0.001); it was 0.64 (P < 0.001) 170S Supplement 1 TABLE 1 Age-standardized baseline characteristics by quintiles of GDQS in the Nurses’ Health Study II Q1 Q2 Q3 Q4 Q5 BMI 24.8 ± 5.8 24.6 ± 5.5 24.4 ± 5.1 24.3 ± 5.0 24.2 ± 4.8 Physical activity, METs 14.5 ± 21.2 17.8 ± 24.1 20.4 ± 26.2 23.5 ± 28.7 29.1 ± 34.0 Current smoker, % 18 14 12 11 9 GDQS 14.3 ± 2.2 18.7 ± 0.9 21.5 ± 0.8 24.4 ± 0.9 28.8 ± 2.2 Unhealthy GDQS components 7.1 ± 2.3 8.2 ± 2.4 8.6 ± 2.4 9.1 ± 2.4 10.1 ± 2.2 Healthy GDQS components 7.3 ± 2.8 10.7 ± 2.5 12.9 ± 2.5 15.3 ± 2.4 18.7 ± 2.7 MDD-W 3.0 ± 1.3 3.6 ± 1.3 4.1 ± 1.3 4.6 ± 1.2 5.4 ± 1.2 AHEI-2010 37.8 ± 7.6 43.7 ± 7.7 47.8 ± 7.9 52.0 ± 8.3 58.8 ± 8.8 Energy intake, kcal/d 1641 ± 536 1689 ± 537 1743 ± 532 1831 ± 530 1990 ± 529 Fiber, g/d 14.3 ± 3.6 16.5 ± 4.0 18.2 ± 4.8 20.0 ± 5.2 22.7 ± 5.8 Alcohol, g/d 2.4 ± 5.7 3.0 ± 6.3 3.3 ± 6.2 3.5 ± 6.1 3.9 ± 6.5 Processed meats, servings/d 0.31 ± 0.33 0.26 ± 0.28 0.22 ± 0.25 0.19 ± 0.23 0.15 ± 0.20 Red meats, servings/d 0.67 ± 0.43 0.60 ± 0.41 0.55 ± 0.38 0.52 ± 0.37 0.44 ± 0.34 Vegetables, servings/d 1.8 ± 1.0 2.5 ± 1.3 3.0 ± 1.5 3.8 ± 1.7 5.1 ± 2.4 Fruit, servings/d 1.2 ± 1.0 1.5 ± 1.1 1.8 ± 1.2 2.1 ± 1.3 2.6 ± 1.6 Nuts and seeds, servings/d 0.04 ± 0.08 0.05 ± 0.10 0.06 ± 0.11 0.07 ± 0.16 0.11 ± 0.21 Legumes, servings/d 0.16 ± 0.16 0.20 ± 0.18 0.24 ± 0.23 0.29 ± 0.26 0.41 ± 0.35 Coffee, servings/d 1.4 ± 1.7 1.5 ± 1.7 1.6 ± 1.7 1.7 ± 1.7 1.8 ± 1.7 n = 88,520. Values are means ± SDs unless otherwise indicated. AHEI-2010, Alternate Healthy Eating Index-2010; GDQS, Global Diet Quality Score; MDD-W, Minimum Diet Diversity score for Women; MET, metabolic equivalent hour; Q, quintile. with the MDD-W. The AHEI-2010 was inversely associated Discussion with diabetes (multivariable HR comparing extreme quintiles: In this analysis, we observed an inverse association between a 0.62; 95% CI: 0.56, 0.68; P-trend < 0.001) and there was no diet quality score designed for global use and risk of type 2 dia- appreciable difference by age (Supplemental Table 4). However, betes among US women. The association appeared to be driven no association was observed with the MDD-W (Supplemental by lower intakes of unhealthy foods. The GDQS compared well Table 5). When we compared the association of the GDQS with with the AHEI-2010 which showed a strong inverse association diabetes pairwise with the AHEI-2010 and the MDD-W, the with diabetes in a cohort of middle-aged nurses (23). The lower association for each SD increase in the AHEI-2010 was slightly diabetes risk with a higher GDQS was similar between women stronger than for the GDQS (HR: 0.91 compared with 0.93, of reproductive age and those who were older. P for difference = 0.03) (Figure 1). On the other hand, the Prospective studies from the United States (24), Europe (6), association for the GDQS was clearly stronger than for the and Asia (25, 26) have shown adherence to healthy eating MDD-W (P for difference < 0.001). TABLE 2 HRs (95% CI) for type 2 diabetes according to quintiles of the Global Diet Quality Score in the Nurses’ Health Study II Q1 Q2 Q3 Q4 Q5 P-trend All women Median score 15.8 19.5 21.9 24.4 27.8 Cases, n 1647 1309 1262 1112 975 Person-years 365,779 364,667 365,382 373,363 364,174 Age- and kcal-adjusted 1 0.76 (0.71, 0.82) 0.71 (0.66, 0.76) 0.59 (0.55, 0.64) 0.48 (0.44, 0.52) <0.001 Multivariable 1 0.91 (0.84, 0.97) 0.94 (0.87, 1.01) 0.87 (0.80, 0.94) 0.83 (0.76, 0.91) <0.001 Women < age 50 y Median score 15.3 18.9 21.3 23.8 27.3 Cases, n 634 456 459 395 322 Person-years 210,566 202,881 198,185 201,222 184,898 Age- and kcal-adjusted 1 0.72 (0.64, 0.82) 0.73 (0.65, 0.83) 0.61 (0.53, 0.69) 0.50 (0.44, 0.58) <0.001 Multivariable 1 0.86 (0.76, 0.98) 1.00 (0.88, 1.13) 0.90 (0.79, 1.02) 0.85 (0.73, 0.98) 0.02 Women age ≥ 50 y Median score 16.7 20.3 22.8 25.0 28.1 Cases, n 1013 853 803 717 653 Person-years 155,214 161,786 167,196 172,140 179,276 Age- and kcal-adjusted 1 0.79 (0.72, 0.86) 0.70 (0.63, 0.76) 0.58 (0.53, 0.64) 0.47 (0.43, 0.52) <0.001 Multivariable 1 0.93 (0.85, 1.02) 0.91 (0.82, 1.00) 0.85 (0.77, 0.94) 0.82 (0.74, 0.91) <0.001 n = 88,520. Q, quintile. Adjusted for age, BMI, energy intake, smoking, family history of diabetes, oral contraceptive use, menopausal status and postmenopausal hormone use (“all women” analysis only), physical activity, alcohol intake, and multivitamin use. GDQS and diabetes risk 171S TABLE 3 HRs (95% CI) for type 2 diabetes according to quintiles of the healthy (GDQS+) and unhealthy (GDQS−) submetrics of the GDQS in the Nurses’ Health Study II Q1 Q2 Q3 Q4 Q5 P-trend GDQS+ submetric (max = 32) All women Median score 8.0 11.3 13.6 15.8 18.8 Cases, n 1441 1290 1188 1232 1154 Person-years 366,057 365,828 368,066 366,408 367,005 Age- and kcal-adjusted 1 0.83 (0.77, 0.90) 0.70 (0.64, 0.76) 0.67 (0.62, 0.73) 0.54 (0.49, 0.59) <0.001 Multivariable 1 1.00 (0.92, 1.08) 0.98 (0.90, 1.07) 1.05 (0.96, 1.14) 1.00 (0.91, 1.10) 0.86 Women < age 50 y Median score 7.5 10.8 13.2 15.4 18.5 Cases, n 554 459 403 443 407 Person-years 205,773 202,984 200,220 197,583 191,192 Age- and kcal-adjusted 1 0.79 (0.69, 0.89) 0.64 (0.56, 0.73) 0.67 (0.58, 0.76) 0.55 (0.47, 0.64) <0.001 Multivariable 1 0.96 (0.84, 1.09) 0.92 (0.80, 1.06) 1.04 (0.90, 1.20) 1.00 (0.85, 1.17) 0.97 Women age ≥ 50 y Median score 8.6 11.9 14.1 16.2 19.1 Cases, n 887 831 785 789 747 Person-years 160,284 162,845 167,846 168,825 175,813 Age- and kcal-adjusted 1 0.86 (0.78, 0.94) 0.74 (0.67, 0.82) 0.68 (0.61, 0.76) 0.54 (0.48, 0.60) <0.001 Multivariable 1 1.02 (0.92, 1.13) 1.03 (0.92, 1.14) 1.05 (0.94, 1.18) 1.01 (0.89, 1.14) 0.77 GDQS− submetric (max = 14) (high score = less unhealthy) All women Median score 5.5 7.2 8.5 9.6 11.0 Cases, n 1701 1446 1151 1050 957 Person-years 374,851 354,527 367,116 359,807 377,063 Age- and kcal-adjusted 1 0.84 (0.78, 0.90) 0.66 (0.61, 0.71) 0.56 (0.51, 0.61) 0.47 (0.43, 0.51) <0.001 Multivariable 1 0.96 (0.89, 1.04) 0.85 (0.78, 0.92) 0.80 (0.73, 0.88) 0.76 (0.69, 0.84) <0.001 Women < age 50 y Median score 5.3 7.0 8.0 9.5 11.0 Cases, n 661 517 394 375 319 Person-years 219,316 190,611 202,692 190,611 194,521 Age- and kcal-adjusted 1 0.86 (0.76, 0.97) 0.69 (0.60, 0.79) 0.61 (0.53, 0.70) 0.52 (0.44, 0.61) <0.001 Multivariable 1 0.96 (0.85, 1.09) 0.87 (0.76, 1.00) 0.83 (0.72, 0.97) 0.81 (0.68, 0.95) <0.001 Women age ≥ 50 y Median score 5.8 7.5 8.7 9.8 11.2 Cases, n 1040 929 757 675 638 Person-years 155,535 163,915 164,424 169,196 182,542 Age- and kcal-adjusted 1 0.82 (0.75, 0.90) 0.64 (0.58, 0.71) 0.54 (0.48, 0.60) 0.45 (0.40, 0.50) <0.001 Multivariable 1 0.96 (0.87, 1.05) 0.83 (0.75, 0.92) 0.78 (0.70, 0.88) 0.74 (0.65, 0.83) <0.001 n = 88,520. GDQS, Global Diet Quality Score; Q, quintile. Adjusted for age, BMI, energy intake, smoking, family history of diabetes, oral contraceptive use, menopausal status and postmenopausal hormone use (“all women” analysis only), physical activity, alcohol intake, multivitamin use, and mutually adjusted for the other submetric. guidelines, as reflected by higher diet quality indices, to be In our analysis, lower intakes of foods in the unhealthy sub- associated with lower risk of type 2 diabetes. Although different metric of the GDQS (GDQS−) were more strongly associated diet quality indices were used in these studies, such as the with a lower diabetes risk than was the healthy submetric of the Healthy Diet Score, the Healthy Eating Index, the Alternate GDQS (GDQS+). Among the foods in the GDQS−,high intakes Healthy Eating Index, and some form of Mediterranean diet of red and processed meats (27), refined grains ( 28), sugar- score, the common features among them were higher intakes sweetened beverages (28), and potatoes, especially as French of fruits, vegetables, whole grains, and lean protein and lower fries (29), have previously been shown to be directly associated intakes of red and processed meats, added sugar, and refined with higher risk of type 2 diabetes. In addition, fried foods grains. The number of components ranged from 6 in the Healthy have also been shown to increase risk of type 2 diabetes (30) Nordic Food Index (6) to 11 in the Alternate Healthy Eating or gestational diabetes (31) in US women. Fried foods may be Index (24). The GDQS features similar food groups, but in a risk factor for diabetes owing to the high energy content or more refined categories and hence a total of 25 food groups. We the increase in lipid oxidation products (32)and trans fat (33) have chosen the approach of using more specific food groups to created in the process of frying. Red and processed meat may be better specify nutrients, such as vitamin C and provitamin A involved in the pathogenesis of type 2 diabetes through inducing carotenoids that are nutrients of concern in some parts of the proinflammatory advanced glycation end products ( 34)and world. pancreatic injury due to oxidative stress from heme iron (35). 172S Supplement AHEI-2010 MDD-W = 0.03 FIGURE 1 Multivariable HR for a 1-SD increase of the GDQS, AHEI-2010, and MDD-W. Models were adjusted for age, BMI, energy intake, smoking, family history of diabetes, oral contraceptive use, menopausal status and postmenopausal hormone use (“all women” analysis only), physical activity, alcohol intake, and multivitamin use. Vertical lines represent 95% CIs. Chi-square test P values tested for significant differences in HR between the GDQS and AHEI-2010, and GDQS and MDD-W. AHEI-2010, Alternate Healthy Eating Index-2010; GDQS, Global Diet Quality Score; MDD-W, Minimum Diet Diversity score for Women. In addition, nitrites and nitrates in processed meats could for multiple confounders that were updated throughout follow- be precursors for the pro-oxidant peroxynitrate (36). Refined up, we cannot exclude the possibility of residual confounding. grains and sugar-sweetened beverages may contribute to weight In designing the GDQS, the metric has to be applicable to gain (37) and the high glycemic load has been associated with geographical regions with a wide range of economic resources diabetes risk (38). and nutrition challenges. Therefore, the score was constructed Healthy dietary patterns similar to the healthy submetric of to balance the needs to reflect nutrient adequacy and predict the GDQS (GDQS+) are inversely associated with diabetes (39). chronic disease risk. For that purpose, the red meat component However, a meta-analysis only found marginally significant which would normally be considered as unhealthy in high- inverse associations for individual food groups such as fruits, income countries was given 1 point for moderate intake and vegetables, and nuts (28). Our analysis also did not observe 0 for low or high intake, to recognize its value as a protein an inverse association of the GDQS+ with diabetes, even and iron source in lower-resource regions. Similarly, points were when the egg component, which has been associated with given for moderate consumption of full-fat dairy to recognize its diabetes risk in US studies (40), was removed. Although the value as a protein, calcium, and energy source, but we did not GDQS+ encompasses a number of healthy food groups and can award points for very high or no consumption. Also, the GDQS potentially detect joint association of these food groups, each promotes moderate consumption of poultry, fish, eggs, and low food group only has 3 levels of scoring. It is possible that only fat dairy. high intakes of specific foods or food groups are associated with Because the GDQS was not designed specifically to predict lower risk of diabetes and our scoring could not differentiate the risk of diabetes, it does not include coffee (42) and moderate these high intakes. On the other hand, the food groups in the alcohol consumption in the metric score (43), both of which unhealthy submetric might be more strongly associated with are inversely associated with type 2 diabetes risk. Nevertheless, diabetes than our scoring method was sufficient to detect. we were still able to observe a strong association with type 2 The strengths of this study include the large sample size and diabetes risk, and the GDQS performed well against 2 other long follow-up which allowed us to accrue a sufficient number diet quality scores. In particular, the GDQS is easier to use of cases to examine diabetes risk even among women of repro- than the AHEI-2010. The GDQS, however, reflects overall diet ductive age. The detailed and repeated assessment of lifestyle healthfulness and is not specifically aimed for the prevention of a and health information in the Nurses’ Health Study II allowed specific disease. As a result, a high GDQS does not represent the us to explore potential difference in risk by reproductive optimal dietary characteristics for the prevention of diabetes. history. On the other hand, lifestyle and diet information was In the current global drive to shift food consumption to obtained from self-report. Although the validity of the dietary be more plant focused for both human and planetary health questionnaire has been well documented (41), some degree of (44), the food groups chosen for the GDQS have implicit misclassification is inevitable. And although we have adjusted concordance with this goal. Out of the 17 healthy food groups GDQS and diabetes risk 173S HR to emphasize in the diet, only 4 were from animal origin. And 8. Fung TT, Chiuve SE, McCullough ML, Rexrode KM, Logroscino G, Hu FB. Adherence to a DASH-style diet and risk of coronary heart disease out of the 9 unhealthy food groups to minimize intake, 3 and stroke in women. Arch Intern Med 2008;168:713–20. were animal protein, and 1 (sweets and ice cream) often has 9. Fung TT, Isanaka S, Hu FB, Willett WC. International food group–based ingredients from animal origin. Therefore, a diet that scores high diet quality and risk of coronary heart disease in men and women. Am on theGDQSwouldtendtobecorrelatedwithdiets that are J Clin Nutr 2018;107:120–9. relatively more plant-based. 10. Gicevic S, Gaskins AJ, Fung TT, Rosner B, Tobias DK, Isanaka S, Willett Health metrics that have specific cutoffs are useful for risk WC. Evaluating pre-pregnancy dietary diversity vs. dietary quality scores as predictors of gestational diabetes and hypertensive disorders assessment and setting treatment targets. Clinically relevant of pregnancy. PLoS One 2018;13:e0195103. cutoffs can be identified if there are inflection points in the 11. Bromage S, Batis C, Bhupathiraju SN, Fawzi WW, Fung TT, Li relation of the GDQS and risk of diabetes. Cutoffs can also Y, Deitchler M, Angulo E, Birk N, Castellanos-Gutíerrez A, et al. be set by assigning a priori categories. However, this latter Development and validation of a novel food-based Global Diet Quality approach requires somewhat arbitrary decisions and also needs Score (GDQS). J Nutr 2021;151(Suppl 10):75S–92S. to consider other outcomes and diverse populations. In our 12. Saeedi P, Petersohn I, Salpea P, Malanda B, Karuranga S, Unwin N, results, there was no departure from linearity in the GDQS. Colagiuri S, Guariguata L, Motala AA, Ogurtsova K, et al. Global and regional diabetes prevalence estimates for 2019 and projections Because our results point toward a progressively lower risk for 2030 and 2045: results from the International Diabetes Federation of diabetes with higher GDQS, there is no strong premise to th Diabetes Atlas, 9 edition. Diabetes Res Clin Pract 2019;157:107843. support specific cutoffs for the GDQS in this cohort of US 13. CDC. National Diabetes Statistics Report, 2020. Atlanta, GA: CDC, US women. Department of Health and Human Services; 2020. In conclusion, the GDQS was inversely associated with type 14. American Diabetes Association. 9. Pharmacologic approaches to 2 diabetes in both reproductive-age and older women in a high- glycemic treatment: Standards of Medical Care in Diabetes — 2020. Diabetes Care 2020;43:S98–S110. income country. It performed well compared with the AHEI- 15. American Diabetes Association. 5. Facilitating behavior change and 2010 in predicting diabetes risk and our results showed that well-being to improve health outcomes: Standards of Medical Care in lower intake of unhealthy foods appeared to be more important Diabetes — 2020. Diabetes Care 2020;43:S48–65. than higher intake of healthy foods. Further testing of the 16. Bao Y, Bertoia ML, Lenart EB, Stampfer MJ, Willett WC, Speizer FE, GDQS in other populations is needed to confirm its usefulness Chavarro JE. Origin, methods, and evolution of the three Nurses’ Health in a broad range of populations to predict noncommunicable Studies. Am J Public Health 2016;106:1573–81. diseases. 17. Feskanich D, Rimm EB, Giovannucci EL, Colditz GA, Stampfer MJ, Litin LB, Willett WC. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. Acknowledgments J Am Diet Assoc 1993;93:790–6. The authors’ responsibilities were as follows—TTF, SNB, and 18. Women’s Dietary Diversity Project (WDDP) Study Group. Development SB: designed the research; TTF: analyzed the data, wrote the of a dichotomous indicator for population-level assessment of paper, and had primary responsibility for the final content; dietary diversity in women of reproductive age. Curr Dev Nutr WCW, MS, CB, MDH, FBH, YL, and MD: provided study 2017;1:cdn.117.001701. oversight and advice; and all authors: read and approved the 19. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes final manuscript. 1979;28:1039–57. 20. The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 1997;20:1183–97. References 21. Manson JE, Stampfer MJ, Colditz GA, Willett WC, Rosner B, Hennekens CH, Speizer FE, Rimm EB, Krolewski AS. Physical activity 1. Schulze MB, Martínez-González MA, Fung TT, Lichtenstein AH, and incidence of non-insulin-dependent diabetes mellitus in women. Forouhi NG. Food based dietary patterns and chronic disease Lancet 1991;338:774–8. prevention. BMJ 2018;361:k2396. 22. Hu FB, Stampfer MJ, Rimm E, Ascherio A, Rosner BA, Spiegelman D, 2. Chiuve SE, Fung TT, Rimm EB, Hu FB, McCullough ML, Wang M, Willett WC. Dietary fat and coronary heart disease: a comparison of Stampfer MJ, Willett WC. Alternative dietary indices both strongly approaches for adjusting for total energy intake and modeling repeated predict risk of chronic disease. J Nutr 2012;142:1009–18. dietary measurements. Am J Epidemiol 1999;149:531–40. 3. Kaluza J, Harris HR, Håkansson N, Wolk A. Adherence to the 23. Fung TT, McCullough M, van Dam RM, Hu FB. A prospective study WCRF/AICR 2018 recommendations for cancer prevention and risk of overall diet quality and risk of type 2 diabetes in women. Diabetes of cancer: prospective cohort studies of men and women. Br J Cancer Care 2007;30:1753–7. 2020;122:1562–70. 24. Cespedes EM, Hu FB, Tinker L, Rosner B, Redline S, Garcia L, Hingle 4. Kirkpatrick SI, Reedy J, Krebs-Smith SM, Pannucci TE, Subar AF, M, Van Horn L, Howard BV, Levitan EB, et al. Multiple healthful Wilson MM, Lerman JL, Tooze JA. Applications of the healthy dietary patterns and type 2 diabetes in the Women’s Health Initiative. eating index for surveillance, epidemiology, and intervention research: Am J Epidemiol 2016;183:622–33. considerations and caveats. J Acad Nutr Diet 2018;118:1603–21. 25. Chen G-C, Koh W-P, Neelakantan N, Yuan J-M, Qin L-Q, van Dam 5. Hernández-Ruiz A, García-Villanova B, Guerra Hernández EJ, Amiano RM. Diet quality indices and risk of type 2 diabetes mellitus: the P, Azpiri M, Molina-Montes E. Description of indexes based on the Singapore Chinese Health Study. Am J Epidemiol 2018;187:2651–61. adherence to the Mediterranean dietary pattern: a review. Nutr Hosp 26. Yu D, Zheng W, Cai H, Xiang Y-B, Li H, Gao Y-T, Shu X-O. Long-term 2015;32:1872–84. diet quality and risk of type 2 diabetes among urban Chinese adults. 6. Lacoppidan SA, Kyrø C, Loft S, Helnæs A, Christensen J, Hansen Diabetes Care 2018;41:723–30. CP, Dahm CC, Overvad K, Tjønneland A, Olsen A. Adherence to a 27. Micha R, Michas G, Mozaffarian D. Unprocessed red and processed healthy Nordic Food Index is associated with a lower risk of type-2 meats and risk of coronary artery disease and type 2 diabetes – diabetes—the Danish Diet, Cancer and Health cohort study. Nutrients an updated review of the evidence. Curr Atheroscler Rep 2012;14: 2015;7:8633–44. 515–24. 7. Puaschitz NG, Assmus J, Strand E, Karlsson T, Vinknes KJ, Lysne V, 28. Schwingshackl L, Hoffmann G, Lampousi A-M, Knüppel S, Iqbal K, Drevon CA, Tell GS, Dierkes J, Nygård O. Adherence to the Healthy Schwedhelm C, Bechthold A, Schlesinger S, Boeing H. Food groups and Nordic Food Index and the incidence of acute myocardial infarction risk of type 2 diabetes mellitus: a systematic review and meta-analysis and mortality among patients with stable angina pectoris. J Hum Nutr of prospective studies. Eur J Epidemiol 2017;32:363–75. Diet 2019;32:86–97. 174S Supplement 29. Schwingshackl L, Schwedhelm C, Hoffmann G, Boeing H. Potatoes and 38. Ludwig DS. The glycemic index: physiological mechanisms relating to risk of chronic disease: a systematic review and dose–response meta- obesity, diabetes, and cardiovascular disease. JAMA 2002;287:2414– analysis. Eur J Nutr 2019;58:2243–51. 23. 30. Cahill LE, Pan A, Chiuve SE, Sun Q, Willett WC, Hu FB, Rimm EB. 39. Jayedi A, Soltani S, Abdolshahi A, Shab-Bidar S. Healthy and unhealthy Fried-food consumption and risk of type 2 diabetes and coronary artery dietary patterns and the risk of chronic disease: an umbrella review of disease: a prospective study in 2 cohorts of US women and men. Am J meta-analyses of prospective cohort studies. Br J Nutr 2020;124:1133– Clin Nutr 2014;100:667–75. 44. 31. Osorio-Yáñez C, Gelaye B, Qiu C, Bao W, Cardenas A, Enquobahrie 40. Djoussé L, Khawaja OA, Gaziano JM. Egg consumption and risk of DA, Williams MA. Maternal intake of fried foods and risk of gestational type 2 diabetes: a meta-analysis of prospective studies. Am J Clin Nutr diabetes mellitus. Ann Epidemiol 2017;27:384–90.e1. 2016;103:474–80. 32. Gupta MK. Frying oils. In: Shahidi F, editor. Baileys’ industrial oil and 41. Yuan C, Spiegelman D, Rimm EB, Rosner BA, Stampfer MJ, Barnett fat products. Hoboken, NJ: John Wiley & Sons; 2005. JB, Chavarro JE, Subar AF, Sampson LK, Willett WC. Validity of a dietary questionnaire assessed by comparison with multiple 33. Yang M, Yang Y, Nie S, Xie M, Chen F. Analysis and formation of trans weighed dietary records or 24-hour recalls. Am J Epidemiol 2017;185: fatty acids in corn oil during the heating process. J Am Oil Chem Soc 570–84. 2012;89:859–67. 42. Carlström M, Larsson SC. Coffee consumption and reduced risk of 34. Uribarri J, Woodruff S, Goodman S, Cai W, Chen X, Pyzik R, Yong developing type 2 diabetes: a systematic review with meta-analysis. Nutr A, Striker GE, Vlassara H. Advanced glycation end products in foods Rev 2018;76:395–417. and a practical guide to their reduction in the diet. J Am Diet Assoc 2010;110:911–16.e12. 43. Knott C, Bell S, Britton A. Alcohol consumption and the risk of type 2 diabetes: a systematic review and dose-response meta-analysis of more 35. Rajpathak SN, Crandall JP, Wylie-Rosett J, Kabat GC, Rohan TE, Hu than 1.9 million individuals from 38 observational studies. Diabetes FB. The role of iron in type 2 diabetes in humans. Biochim Biophys Acta Care 2015;38:1804–12. 2009;1790:671–81. 36. Pacher P, Beckman JS, Liaudet L. Nitric oxide and peroxynitrite in 44. Willett W, Rockström J, Loken B, Springmann M, Lang T, Vermeulen health and disease. Physiol Rev 2007;87:315–424. S, Garnett T, Tilman D, DeClerck F, Wood A, et al. Food in the Anthropocene: the EAT–Lancet Commission on healthy diets from 37. Malik VS, Hu FB. Sweeteners and risk of obesity and type 2 diabetes: the sustainable food systems. Lancet 2019;393:447–92. role of sugar-sweetened beverages. Curr Diab Rep 2012;12:195–203. GDQS and diabetes risk 175S http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journal of Nutrition Pubmed Central

Higher Global Diet Quality Score Is Inversely Associated with Risk of Type 2 Diabetes in US Women

Download PDF
8 pages

Loading next page...
 
/lp/pubmed-central/higher-global-diet-quality-score-is-inversely-associated-with-risk-of-0IxM07YCT5

References (56)

Publisher
Pubmed Central
Copyright
Copyright © The Author(s) on behalf of the American Society for Nutrition 2021.
ISSN
0022-3166
eISSN
1541-6100
DOI
10.1093/jn/nxab195
Publisher site
See Article on Publisher Site

Abstract

The Journal of Nutrition Supplement Higher Global Diet Quality Score Is Inversely Associated with Risk of Type 2 Diabetes in US Women 1,2 2 2,3 2 4 Teresa T Fung, Yanping Li, Shilpa N Bhupathiraju, Sabri Bromage, Carolina Batis, Michelle 3,5 3,5 2,3 6 2,3 D Holmes, Meir Stampfer, Frank B Hu, Megan Deitchler, and Walter C Willett 1 2 Department of Nutrition, Simmons University, Boston, MA, USA; Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA; Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA; CONACYT—Health and Nutrition Research Center, National Institute of Public Health, Cuernavaca, Mexico; 5 6 Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; and Intake—Center for Dietary Assessment, FHI Solutions, Washington, DC, USA ABSTRACT Background: We have developed a diet quality metric intended for global use. To assess its utility in high-income settings, an evaluation of its ability to predict chronic disease is needed. Objectives: We aimed to prospectively examine the ability of the Global Diet Quality Score (GDQS) to predict the risk of type 2 diabetes in the United States, examine potential differences of association by age, and compare the GDQS with other diet quality scores. Methods: Health, lifestyle, and diet information was collected from women (n = 88,520) in the Nurses’ Health Study II aged 27–44 y at baseline through repeated questionnaires between 1991 and 2017. The overall GDQS consists of 25 food groups. Points are awarded for higher intake of healthy groups and lower intake of unhealthy groups (maximum of 49 points). Multivariable HRs were computed for confirmed type 2 diabetes using proportional hazards models. We also compared the GDQS with the Minimum Diet Diversity score for Women (MDD-W) and the Alternate Healthy Eating Index-2010 (AHEI-2010). Results: We ascertained 6305 incident cases of type 2 diabetes during follow-up. We observed a lower risk of diabetes with higher GDQS; the multivariable HR comparing extreme quintiles of the GDQS was 0.83 (95% CI: 0.76, 0.91; P- trend < 0.001). The magnitude of association was similar between women aged <50 y and those aged ≥50 y. An inverse association was observed with lower intake of unhealthy components (HR comparing extreme quintiles of the unhealthy submetric: 0.76; 95% CI: 0.69, 0.84; P-trend < 0.001) but was not with the healthy submetric. The inverse association for each 1-SD increase in the GDQS (HR: 0.93; 95% CI: 0.91, 0.96) was stronger (P < 0.001) than for the MDD-W (HR: 1.00; 95% CI: 0.94, 1.04) but was slightly weaker (P = 0.03) than for the AHEI-2010 (HR: 0.91; 95% CI: 0.88, 0.94). Conclusions: A higher GDQS was inversely associated with type 2 diabetes risk in US women of reproductive age or older, mainly from lower intake of unhealthy foods. The GDQS performed nearly as well as the AHEI-2010. J Nutr 2021;151:168S–175S. Keywords: diet quality, diabetes, epidemiology, women, nutrition Introduction several chronic diseases, including cardiovascular disease and diabetes (1). Several diet quality indices have been developed and evaluated To apply these diet quality indices in clinical and public for their association with risk of chronic diseases (1). These health settings to guide individual dietary choices and public indices typically were based on recommendations for a healthy health surveillance, the metric must be simple and quick to diet (2–4) or reflections of regional dietary habits ( 5–7). Many administer. In addition, a metric that is valid and practical for include nutrient components and therefore require the use of use across different parts of the world and different economic a food composition database (2–4), or a scoring algorithm that development levels would have the additional advantage of is based on population-specific intake levels ( 5, 8). Evidence enabling global comparisons. Therefore, indices that involve a from prospective studies is consistent that adherence to food composition database or use population-specific scoring these diet quality indices is associated with a lower risk of would be difficult to implement across regions. To circumvent The Author(s) 2021. Published by Oxford University Press on behalf of the American Society for Nutrition. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Manuscript received February 27, 2021. Initial review completed March 29, 2021. Revision accepted May 25, 2021. 168S First published online October 1, 2021; doi: https://doi.org/10.1093/jn/nxab195. these limitations, we previously developed the Prime Diet Methods Quality Score (PDQS) that only consists of food groups. It is Participants inversely associated with cardiovascular disease and gestational The Nurses’ Health Study II (NHS II) is an ongoing prospective cohort diabetes in US men and women (9, 10). study that is comprised of 116,430 US female Registered Nurses To provide a metric that is usable in regions where between 25 and 42 y old at inception in 1989 (16). Information on nutritional adequacy is a concern, we have further modified lifestyle practices and incidence of type 2 diabetes was collected every 2 y the PDQS and tested it for association with nutritional by self-reported questionnaires. Diet was assessed every 4 y beginning in markers relevant to middle- and lower-income countries. Our 1991 using a validated FFQ. Women with diabetes, gestational diabetes, cancer, or cardiovascular disease or who died before the rst fi dietary final metric, the Global Diet Quality Score (GDQS), uses a assessment were excluded. We also excluded those who did not complete combination of healthy and unhealthy food groups. It has additional questionnaires beyond baseline and those who reported reasonable correlation with measures of nutrient adequacy (11). implausible energy intakes (<500 or >3500 kcal/d) at baseline. If a Because the GDQS has several differences from the PDQS, participant reported being pregnant in a questionnaire period, person- we assessed its utility in a higher-income setting by testing its time during that 2-y period was excluded. A total of 88,520 women ability to predict the risk of type 2 diabetes in US women. were included in this analysis and loss to follow-up was ∼10% during We chose type 2 diabetes because the incidence is increasing the study period. This study was approved by the institutional review globally, with a projected increase from >400 million affected in boards of Brigham and Women’s Hospital and the Harvard TH Chan 2019 to ∼700 million by 2045 (12). In the United States, it was School of Public Health. estimated that 12% of adult women and 14% of adult men were living with diabetes in 2013–2016 (13). Although a plethora of Diet assessment medications are available (14), there is no cure in most cases A validated semiquantitative FFQ was self-administered every 4 y, each and successful management requires adequate compliance and containing ∼135 items (17). For each food item, a standard portion regular access to health care (15). Therefore, prevention through size was provided with 9 intake frequency choices ranging from “never lifestyle, and especially diet, continues to be an important or less than once per month” to “≥6 times per day.” The GDQS approach. In this analysis, we prospectively examined the asso- was modified based on the PDQS ( 9) to capture food groups that would reflect nutrient adequacy and predict major noncommunicable ciation between the GDQS and the risk of type 2 diabetes among diseases in both lower- and high-income countries globally. It consists US women, and explored potential differences in association by of 16 healthy food groups (dark green leafy vegetables, cruciferous age. To understand the function of the GDQS, we also explored vegetables, deep orange vegetables, other vegetables, deep orange fruits, how the healthy and unhealthy components would drive any ob- deep orange tubers, citrus fruits, other fruits, legumes, nuts and seeds, served association. We hypothesized that the overall GDQS and poultry and game meats, fish and shellfish, whole grains, liquid oils, low the healthy components (GDQS+ submetric) would be inversely fat dairy, eggs) and 7 unhealthy food groups (white roots and tubers, associated with diabetes risk, whereas lower intake of the un- processed meats, refined grains and baked goods, sugar-sweetened healthy components (GDQS− submetric) would have an inverse beverages, sweets and ice cream, juices, purchased deep fried foods) association. For the GDQS to be a useful nutrition metric to (Supplemental Table 1). Intake of each food group was classified into predict noncommunicable diseases, it must also perform at least <1/wk, 1 to <4/wk, and ≥4/wk. For healthy food groups, points between 0 and 4 were given to each level of intake depending on the similarly as other established diet quality indices. Therefore, food group. For unhealthy food groups, 2, 1, and 0 points were given we also compared it with the Minimum Diet Diversity score for the same 3 intake levels so lower intake would receive more points. for Women (MDD-W) and the Alternate Healthy Eating Index- In addition to the aforementioned food groups, the GDQS also has a red 2010 (AHEI-2010) for prediction of type 2 diabetes. meat group and a full-fat dairy group with different scoring to account for their contribution to nutrient adequacy in low- to middle-income Funding for the research was provided by FHI Solutions, recipient of a Bill countries. Red meat was given 0, 1, and 0 points for intake of the same & Melinda Gates Foundation grant to support Intake—Center for Dietary 3 intake levels as for the other unhealthy food groups, and full-fat dairy Assessment and by NIH grant U01 CA176726. was given 0, 1, 2, and 0 points for intake of <1/wk, 1 to <4/wk, ≥4/wk Author disclosures: TTF is an Associate Editor for the Journal of Nutrition and to <3/d, and ≥3/d, respectively. The full GDQS has 25 food groups and played no role in the Journal’s evaluation of the manuscript. All other authors a score range of 0–49 points, with higher points representing a healthier report no conflicts of interest. diet. The healthy portion of the GDQS (GDQS+) has a range of 0–32. The content is solely the responsibility of the authors and does not necessarily For the purpose of this analysis, we included red meat and full-fat dairy represent the official views of the NIH. as part of the unhealthy portion (GDQS−), which has a range of 0–17, Supplemental Tables 1–5 are available from the “Supplementary data” link in with a higher score representing lower intake of unhealthy foods and the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn. hence healthier food choices. Published in a supplement to The Journal of Nutrition. Publication costs for this To compare the GDQS with other established diet quality indices, supplement were funded by the Bill & Melinda Gates Foundation in conjunction we also computed the AHEI-2010 (2) and the MDD-W (18) for each with FHI Solutions, recipient of a Bill & Melinda Gates Foundation grant to participant. The AHEI-2010 consists of 11 food and nutrient groups. support Intake—Center for Dietary Assessment. The opinions expressed in this High points are given for higher intakes of healthy groups (vegetables, publication are those of the authors and are not attributable to the sponsors whole fruits, nuts and legumes, whole grains, polyunsaturated fat, or the publisher, Editor, or Editorial Board of The Journal of Nutrition.The and long-chain n–3 fatty acids) and lower intakes of unhealthy Supplement Coordinators for the supplement publication were Megan Deitchler, groups (red and processed meats, sugar-sweetened beverages and fruit Intake—Center for Dietary Assessment at FHI Solutions, Washington, DC; and juice, trans fat, and sodium). Points are also given for moderate Sabri Bromage, Harvard TH Chan School of Public Health, Boston, MA. The GDQS research initiative was launched by Intake – Center for Dietary intake of alcohol. Each component ranges from 0 to 10 points Assessment. The research was led by Harvard T.H. Chan School of Public Health, with the total possible score ranging from 0 to 110 points. It has Department of Nutrition and carried out in collaboration with researchers at the previously been shown to be inversely associated with diabetes risk in National Public Health Institute (INSP), Mexico. Funding for the research was women (2). provided by FHI Solutions, recipient of a Bill & Melinda Gates Foundation grant The MDD-W, originally developed as a proxy indicator for nutrient to support Intake – Center for Dietary Assessment. adequacy, consists of 10 food groups: grains and starchy vegetables, Address correspondence to TTF (e-mail: [email protected]). pulses, nuts and seeds, dairy, animal flesh, eggs, dark green leafy Abbreviations used: AHEI-2010, Alternate Healthy Eating Index-2010; GDQS, vegetables, vitamin A–rich vegetables and fruits, other vegetables, and Global Diet Quality Score; MDD-W, Minimum Diet Diversity score for Women; other fruits (18). The scoring method for the original MDD-W is based MET, metabolic equivalent hour; PDQS, Prime Diet Quality Score. GDQS and diabetes risk 169S on intake collected by 24-h recall. To adapt it for the FFQ, we assigned the analysis by BMI status and physical activity. To examine the potential influence of pregnancy on the association between the 1 point for each food group with intake ≥1 serving/d and 0 for less (9). The MDD-W has a range of 0–10 points. GDQS and diabetes, we ran regression models separately for women based on pregnancy history, and among ever-pregnant women by history of gestational diabetes. Tests for 2-way interaction between Outcome assessment the GDQS and each of the stratified factors were conducted using the Incident type 2 diabetes was rst fi reported through the biennial ques- likelihood ratio test comparing regression models with and without an tionnaires and confirmed with a validated supplemental questionnaire interaction term. Analysis was conducted using SAS version 9.4 (SAS based on National Diabetes Data Group criteria. This included ≥1of Institute Inc.). the following: ≥1 classic symptom (excessive thirst, polyuria or frequent To compare the strength of association between the GDQS and the urination, weight loss, hunger), fasting plasma glucose concentrations AHEI-2010 and MDD-W, we standardized each score and modeled each ≥7.8 mmol/L, or random plasma glucose concentrations ≥11.1 mmol/L 1 SD of the scores in the same model. Differences in the regression (19). In the case of a lack of symptoms, diabetes was considered coefficients were compared using the Wald test. confirmed with ≥2 elevated plasma glucose concentrations on different occasions (fasting plasma glucose concentrations ≥7.8 mmol/L, random concentrations ≥11.1 mmol/L, and/or 2-h blood glucose concentrations ≥11.1 mmol/L during oral-glucose-tolerance testing); or treatment Results with hypoglycemic medications (insulin or oral hypoglycemic agent). For cases reported after 1998, criteria from the American Diabetes In ≤26 y of follow-up, we ascertained 6305 incident cases of Association were used in which the threshold for fasting plasma glucose type 2 diabetes, of which 2266 were women younger than 50 changed from 7.8 mmol/L to 7.0 mmol/L (20). The supplemental y old and 4039 were women ≥50 y old. Women with a higher questionnaire was validated by a review of medical reports (21). In GDQS tended to be leaner, more physically active, less likely a random sample of 62 cases in the Nurses’ Health Study that were to be current smokers, and consumed more alcohol and coffee confirmed by the supplementary questionnaire, 61 (98%) cases were (Table 1). reconfirmed after medical records were reviewed by an endocrinologist We observed a lower risk of diabetes with higher GDQS blinded to the supplementary questionnaire. (multivariable HR comparing extreme quintiles: 0.83; 95% CI: 0.76, 0.91; P-trend < 0.001) (Table 2). The association Covariate assessment for women age <50 y was 0.85 (95% CI: 0.73, 0.98; P- Information on age, race, and height was collected at cohort inception. trend < 0.001) and for age ≥50 y was 0.82 (95% CI: 0.74, Body weight, cigarette smoking (including the number of cigarettes 0.91, P-trend < 0.001) with no significant interaction. We also per day), physical activity, menopausal status and postmenopausal hormone use, oral contraceptive use, family history of diabetes, history separately examined the submetrics of the GDQS representing of hypercholesterolemia, and high blood pressure were collected in each healthy (GDQS+) and unhealthy (GDQS−) food components. biennial questionnaire. BMI (in kg/m ) was calculated using height These 2 submetrics were only weakly correlated (Spearman r collected at baseline and weight reported at each questionnaire cycle. =−0.06, P < 0.001). The healthy components of the GDQS Alcohol intake and supplemental vitamin and mineral use were collected (GDQS+) were not associated with diabetes risk (Table 3). On with FFQs. the other hand, higher GDQS−, which represents lower intake of the unhealthy components, showed an inverse association Statistical analysis (multivariable HR comparing extreme quintiles: 0.76; 95% CI: For this analysis, follow-up duration in person-years was calculated 0.69, 0.84; P-trend < 0.001). There was no apparent difference from the date of return of the 1991 questionnaire to the date of diabetes in association by age. Spline regression did not detect significant diagnosis, last questionnaire returned, or 30 June, 2017. We computed departure from linearity for the overall GDQS, GDQS+, cumulative averages of diet quality scores to reduce within-person or GDQS− (data not shown). In the sensitivity analysis in variation and represent long-term intake (22). We used time-dependent which we excluded the egg component from the GDQS+, Cox proportional hazards regression models to compute HRs of type the null association persisted in the remaining portion of the 2 diabetes for quintiles of the GDQS, GDQS+, and GDQS−. Eggs are GDQS+. included in the GDQS+ because of their protein and vitamin content, but they also contain substantial amounts of cholesterol. Hence, we in The GDQS was inversely associated with diabetes in both addition computed an alternate GDQS+ without the egg component for women ever or never pregnant (Supplemental Table 2). Al- sensitivity analysis. We tested for the proportional hazards assumption though the magnitude of association did not differ substantially by including an interaction term of GDQS and age (which reflects for pregnancy history, the trend appeared to be more consistent time) and used the likelihood ratio test. The P value for the chi- for never-pregnant women (P-interaction = 0.06). Among square distribution was >0.05, hence it did not show a violation of the women who had been pregnant, an inverse association with proportional hazards assumption. the GDQS was only observed for those without a history All models were adjusted by age (mo) at the start of follow-up of gestational diabetes (multivariable HR comparing extreme for each woman and the calendar year of each questionnaire cycle. quintiles: 0.83; 95% CI: 0.75, 0.91; P-trend < 0.001). We Multivariable models were adjusted for race (white/nonwhite), family also stratified the analysis by BMI and physical activity history of diabetes, smoking (never, past, 1–14 cigarettes/d, 15–24 cigarettes/d, ≥25 cigarettes/d), alcohol intake (none, <5g/d, 5 to (Supplemental Table 3). The inverse association was significant <10 g/d, ≥10 g/d), energy intake (quintiles), coffee intake (continuous), regardless of BMI status; however, it was stronger among physical activity [<3 metabolic equivalent hours (METs)/wk, 3 to leaner women (P-interaction < 0.001). On the other hand, <9METs/wk, 9 to <18 METs/wk, 18 to <27 METs/wk, ≥27 although the association between the GDQS and diabetes METs/wk], BMI (<23, 23 to < 25, 25 to <30, 30 to <35, ≥35), appeared stronger among those with physical activity above multivitamin use (yes/no), menopausal status and menopausal hormone the median, the P value for interaction did not reach statistical therapy (premenopausal, no hormone use, past use, current use), oral significance. contraceptive use (never, past, current), history of hypertension at We also compared the magnitude of association of the GDQS baseline, and history of hyperlipidemia at baseline. We used restricted with 2 other diet quality scores: the AHEI-2010 and MDD-W. cubic spline regression to assess potential nonlinear association. To The Spearman correlation coefficient between the GDQS and access potential differential association of the GDQS with diabetes by age, we conducted analyses stratified by age. We also stratified the AHEI-2010 was 0.74 (P < 0.001); it was 0.64 (P < 0.001) 170S Supplement 1 TABLE 1 Age-standardized baseline characteristics by quintiles of GDQS in the Nurses’ Health Study II Q1 Q2 Q3 Q4 Q5 BMI 24.8 ± 5.8 24.6 ± 5.5 24.4 ± 5.1 24.3 ± 5.0 24.2 ± 4.8 Physical activity, METs 14.5 ± 21.2 17.8 ± 24.1 20.4 ± 26.2 23.5 ± 28.7 29.1 ± 34.0 Current smoker, % 18 14 12 11 9 GDQS 14.3 ± 2.2 18.7 ± 0.9 21.5 ± 0.8 24.4 ± 0.9 28.8 ± 2.2 Unhealthy GDQS components 7.1 ± 2.3 8.2 ± 2.4 8.6 ± 2.4 9.1 ± 2.4 10.1 ± 2.2 Healthy GDQS components 7.3 ± 2.8 10.7 ± 2.5 12.9 ± 2.5 15.3 ± 2.4 18.7 ± 2.7 MDD-W 3.0 ± 1.3 3.6 ± 1.3 4.1 ± 1.3 4.6 ± 1.2 5.4 ± 1.2 AHEI-2010 37.8 ± 7.6 43.7 ± 7.7 47.8 ± 7.9 52.0 ± 8.3 58.8 ± 8.8 Energy intake, kcal/d 1641 ± 536 1689 ± 537 1743 ± 532 1831 ± 530 1990 ± 529 Fiber, g/d 14.3 ± 3.6 16.5 ± 4.0 18.2 ± 4.8 20.0 ± 5.2 22.7 ± 5.8 Alcohol, g/d 2.4 ± 5.7 3.0 ± 6.3 3.3 ± 6.2 3.5 ± 6.1 3.9 ± 6.5 Processed meats, servings/d 0.31 ± 0.33 0.26 ± 0.28 0.22 ± 0.25 0.19 ± 0.23 0.15 ± 0.20 Red meats, servings/d 0.67 ± 0.43 0.60 ± 0.41 0.55 ± 0.38 0.52 ± 0.37 0.44 ± 0.34 Vegetables, servings/d 1.8 ± 1.0 2.5 ± 1.3 3.0 ± 1.5 3.8 ± 1.7 5.1 ± 2.4 Fruit, servings/d 1.2 ± 1.0 1.5 ± 1.1 1.8 ± 1.2 2.1 ± 1.3 2.6 ± 1.6 Nuts and seeds, servings/d 0.04 ± 0.08 0.05 ± 0.10 0.06 ± 0.11 0.07 ± 0.16 0.11 ± 0.21 Legumes, servings/d 0.16 ± 0.16 0.20 ± 0.18 0.24 ± 0.23 0.29 ± 0.26 0.41 ± 0.35 Coffee, servings/d 1.4 ± 1.7 1.5 ± 1.7 1.6 ± 1.7 1.7 ± 1.7 1.8 ± 1.7 n = 88,520. Values are means ± SDs unless otherwise indicated. AHEI-2010, Alternate Healthy Eating Index-2010; GDQS, Global Diet Quality Score; MDD-W, Minimum Diet Diversity score for Women; MET, metabolic equivalent hour; Q, quintile. with the MDD-W. The AHEI-2010 was inversely associated Discussion with diabetes (multivariable HR comparing extreme quintiles: In this analysis, we observed an inverse association between a 0.62; 95% CI: 0.56, 0.68; P-trend < 0.001) and there was no diet quality score designed for global use and risk of type 2 dia- appreciable difference by age (Supplemental Table 4). However, betes among US women. The association appeared to be driven no association was observed with the MDD-W (Supplemental by lower intakes of unhealthy foods. The GDQS compared well Table 5). When we compared the association of the GDQS with with the AHEI-2010 which showed a strong inverse association diabetes pairwise with the AHEI-2010 and the MDD-W, the with diabetes in a cohort of middle-aged nurses (23). The lower association for each SD increase in the AHEI-2010 was slightly diabetes risk with a higher GDQS was similar between women stronger than for the GDQS (HR: 0.91 compared with 0.93, of reproductive age and those who were older. P for difference = 0.03) (Figure 1). On the other hand, the Prospective studies from the United States (24), Europe (6), association for the GDQS was clearly stronger than for the and Asia (25, 26) have shown adherence to healthy eating MDD-W (P for difference < 0.001). TABLE 2 HRs (95% CI) for type 2 diabetes according to quintiles of the Global Diet Quality Score in the Nurses’ Health Study II Q1 Q2 Q3 Q4 Q5 P-trend All women Median score 15.8 19.5 21.9 24.4 27.8 Cases, n 1647 1309 1262 1112 975 Person-years 365,779 364,667 365,382 373,363 364,174 Age- and kcal-adjusted 1 0.76 (0.71, 0.82) 0.71 (0.66, 0.76) 0.59 (0.55, 0.64) 0.48 (0.44, 0.52) <0.001 Multivariable 1 0.91 (0.84, 0.97) 0.94 (0.87, 1.01) 0.87 (0.80, 0.94) 0.83 (0.76, 0.91) <0.001 Women < age 50 y Median score 15.3 18.9 21.3 23.8 27.3 Cases, n 634 456 459 395 322 Person-years 210,566 202,881 198,185 201,222 184,898 Age- and kcal-adjusted 1 0.72 (0.64, 0.82) 0.73 (0.65, 0.83) 0.61 (0.53, 0.69) 0.50 (0.44, 0.58) <0.001 Multivariable 1 0.86 (0.76, 0.98) 1.00 (0.88, 1.13) 0.90 (0.79, 1.02) 0.85 (0.73, 0.98) 0.02 Women age ≥ 50 y Median score 16.7 20.3 22.8 25.0 28.1 Cases, n 1013 853 803 717 653 Person-years 155,214 161,786 167,196 172,140 179,276 Age- and kcal-adjusted 1 0.79 (0.72, 0.86) 0.70 (0.63, 0.76) 0.58 (0.53, 0.64) 0.47 (0.43, 0.52) <0.001 Multivariable 1 0.93 (0.85, 1.02) 0.91 (0.82, 1.00) 0.85 (0.77, 0.94) 0.82 (0.74, 0.91) <0.001 n = 88,520. Q, quintile. Adjusted for age, BMI, energy intake, smoking, family history of diabetes, oral contraceptive use, menopausal status and postmenopausal hormone use (“all women” analysis only), physical activity, alcohol intake, and multivitamin use. GDQS and diabetes risk 171S TABLE 3 HRs (95% CI) for type 2 diabetes according to quintiles of the healthy (GDQS+) and unhealthy (GDQS−) submetrics of the GDQS in the Nurses’ Health Study II Q1 Q2 Q3 Q4 Q5 P-trend GDQS+ submetric (max = 32) All women Median score 8.0 11.3 13.6 15.8 18.8 Cases, n 1441 1290 1188 1232 1154 Person-years 366,057 365,828 368,066 366,408 367,005 Age- and kcal-adjusted 1 0.83 (0.77, 0.90) 0.70 (0.64, 0.76) 0.67 (0.62, 0.73) 0.54 (0.49, 0.59) <0.001 Multivariable 1 1.00 (0.92, 1.08) 0.98 (0.90, 1.07) 1.05 (0.96, 1.14) 1.00 (0.91, 1.10) 0.86 Women < age 50 y Median score 7.5 10.8 13.2 15.4 18.5 Cases, n 554 459 403 443 407 Person-years 205,773 202,984 200,220 197,583 191,192 Age- and kcal-adjusted 1 0.79 (0.69, 0.89) 0.64 (0.56, 0.73) 0.67 (0.58, 0.76) 0.55 (0.47, 0.64) <0.001 Multivariable 1 0.96 (0.84, 1.09) 0.92 (0.80, 1.06) 1.04 (0.90, 1.20) 1.00 (0.85, 1.17) 0.97 Women age ≥ 50 y Median score 8.6 11.9 14.1 16.2 19.1 Cases, n 887 831 785 789 747 Person-years 160,284 162,845 167,846 168,825 175,813 Age- and kcal-adjusted 1 0.86 (0.78, 0.94) 0.74 (0.67, 0.82) 0.68 (0.61, 0.76) 0.54 (0.48, 0.60) <0.001 Multivariable 1 1.02 (0.92, 1.13) 1.03 (0.92, 1.14) 1.05 (0.94, 1.18) 1.01 (0.89, 1.14) 0.77 GDQS− submetric (max = 14) (high score = less unhealthy) All women Median score 5.5 7.2 8.5 9.6 11.0 Cases, n 1701 1446 1151 1050 957 Person-years 374,851 354,527 367,116 359,807 377,063 Age- and kcal-adjusted 1 0.84 (0.78, 0.90) 0.66 (0.61, 0.71) 0.56 (0.51, 0.61) 0.47 (0.43, 0.51) <0.001 Multivariable 1 0.96 (0.89, 1.04) 0.85 (0.78, 0.92) 0.80 (0.73, 0.88) 0.76 (0.69, 0.84) <0.001 Women < age 50 y Median score 5.3 7.0 8.0 9.5 11.0 Cases, n 661 517 394 375 319 Person-years 219,316 190,611 202,692 190,611 194,521 Age- and kcal-adjusted 1 0.86 (0.76, 0.97) 0.69 (0.60, 0.79) 0.61 (0.53, 0.70) 0.52 (0.44, 0.61) <0.001 Multivariable 1 0.96 (0.85, 1.09) 0.87 (0.76, 1.00) 0.83 (0.72, 0.97) 0.81 (0.68, 0.95) <0.001 Women age ≥ 50 y Median score 5.8 7.5 8.7 9.8 11.2 Cases, n 1040 929 757 675 638 Person-years 155,535 163,915 164,424 169,196 182,542 Age- and kcal-adjusted 1 0.82 (0.75, 0.90) 0.64 (0.58, 0.71) 0.54 (0.48, 0.60) 0.45 (0.40, 0.50) <0.001 Multivariable 1 0.96 (0.87, 1.05) 0.83 (0.75, 0.92) 0.78 (0.70, 0.88) 0.74 (0.65, 0.83) <0.001 n = 88,520. GDQS, Global Diet Quality Score; Q, quintile. Adjusted for age, BMI, energy intake, smoking, family history of diabetes, oral contraceptive use, menopausal status and postmenopausal hormone use (“all women” analysis only), physical activity, alcohol intake, multivitamin use, and mutually adjusted for the other submetric. guidelines, as reflected by higher diet quality indices, to be In our analysis, lower intakes of foods in the unhealthy sub- associated with lower risk of type 2 diabetes. Although different metric of the GDQS (GDQS−) were more strongly associated diet quality indices were used in these studies, such as the with a lower diabetes risk than was the healthy submetric of the Healthy Diet Score, the Healthy Eating Index, the Alternate GDQS (GDQS+). Among the foods in the GDQS−,high intakes Healthy Eating Index, and some form of Mediterranean diet of red and processed meats (27), refined grains ( 28), sugar- score, the common features among them were higher intakes sweetened beverages (28), and potatoes, especially as French of fruits, vegetables, whole grains, and lean protein and lower fries (29), have previously been shown to be directly associated intakes of red and processed meats, added sugar, and refined with higher risk of type 2 diabetes. In addition, fried foods grains. The number of components ranged from 6 in the Healthy have also been shown to increase risk of type 2 diabetes (30) Nordic Food Index (6) to 11 in the Alternate Healthy Eating or gestational diabetes (31) in US women. Fried foods may be Index (24). The GDQS features similar food groups, but in a risk factor for diabetes owing to the high energy content or more refined categories and hence a total of 25 food groups. We the increase in lipid oxidation products (32)and trans fat (33) have chosen the approach of using more specific food groups to created in the process of frying. Red and processed meat may be better specify nutrients, such as vitamin C and provitamin A involved in the pathogenesis of type 2 diabetes through inducing carotenoids that are nutrients of concern in some parts of the proinflammatory advanced glycation end products ( 34)and world. pancreatic injury due to oxidative stress from heme iron (35). 172S Supplement AHEI-2010 MDD-W = 0.03 FIGURE 1 Multivariable HR for a 1-SD increase of the GDQS, AHEI-2010, and MDD-W. Models were adjusted for age, BMI, energy intake, smoking, family history of diabetes, oral contraceptive use, menopausal status and postmenopausal hormone use (“all women” analysis only), physical activity, alcohol intake, and multivitamin use. Vertical lines represent 95% CIs. Chi-square test P values tested for significant differences in HR between the GDQS and AHEI-2010, and GDQS and MDD-W. AHEI-2010, Alternate Healthy Eating Index-2010; GDQS, Global Diet Quality Score; MDD-W, Minimum Diet Diversity score for Women. In addition, nitrites and nitrates in processed meats could for multiple confounders that were updated throughout follow- be precursors for the pro-oxidant peroxynitrate (36). Refined up, we cannot exclude the possibility of residual confounding. grains and sugar-sweetened beverages may contribute to weight In designing the GDQS, the metric has to be applicable to gain (37) and the high glycemic load has been associated with geographical regions with a wide range of economic resources diabetes risk (38). and nutrition challenges. Therefore, the score was constructed Healthy dietary patterns similar to the healthy submetric of to balance the needs to reflect nutrient adequacy and predict the GDQS (GDQS+) are inversely associated with diabetes (39). chronic disease risk. For that purpose, the red meat component However, a meta-analysis only found marginally significant which would normally be considered as unhealthy in high- inverse associations for individual food groups such as fruits, income countries was given 1 point for moderate intake and vegetables, and nuts (28). Our analysis also did not observe 0 for low or high intake, to recognize its value as a protein an inverse association of the GDQS+ with diabetes, even and iron source in lower-resource regions. Similarly, points were when the egg component, which has been associated with given for moderate consumption of full-fat dairy to recognize its diabetes risk in US studies (40), was removed. Although the value as a protein, calcium, and energy source, but we did not GDQS+ encompasses a number of healthy food groups and can award points for very high or no consumption. Also, the GDQS potentially detect joint association of these food groups, each promotes moderate consumption of poultry, fish, eggs, and low food group only has 3 levels of scoring. It is possible that only fat dairy. high intakes of specific foods or food groups are associated with Because the GDQS was not designed specifically to predict lower risk of diabetes and our scoring could not differentiate the risk of diabetes, it does not include coffee (42) and moderate these high intakes. On the other hand, the food groups in the alcohol consumption in the metric score (43), both of which unhealthy submetric might be more strongly associated with are inversely associated with type 2 diabetes risk. Nevertheless, diabetes than our scoring method was sufficient to detect. we were still able to observe a strong association with type 2 The strengths of this study include the large sample size and diabetes risk, and the GDQS performed well against 2 other long follow-up which allowed us to accrue a sufficient number diet quality scores. In particular, the GDQS is easier to use of cases to examine diabetes risk even among women of repro- than the AHEI-2010. The GDQS, however, reflects overall diet ductive age. The detailed and repeated assessment of lifestyle healthfulness and is not specifically aimed for the prevention of a and health information in the Nurses’ Health Study II allowed specific disease. As a result, a high GDQS does not represent the us to explore potential difference in risk by reproductive optimal dietary characteristics for the prevention of diabetes. history. On the other hand, lifestyle and diet information was In the current global drive to shift food consumption to obtained from self-report. Although the validity of the dietary be more plant focused for both human and planetary health questionnaire has been well documented (41), some degree of (44), the food groups chosen for the GDQS have implicit misclassification is inevitable. And although we have adjusted concordance with this goal. Out of the 17 healthy food groups GDQS and diabetes risk 173S HR to emphasize in the diet, only 4 were from animal origin. And 8. Fung TT, Chiuve SE, McCullough ML, Rexrode KM, Logroscino G, Hu FB. Adherence to a DASH-style diet and risk of coronary heart disease out of the 9 unhealthy food groups to minimize intake, 3 and stroke in women. Arch Intern Med 2008;168:713–20. were animal protein, and 1 (sweets and ice cream) often has 9. Fung TT, Isanaka S, Hu FB, Willett WC. International food group–based ingredients from animal origin. Therefore, a diet that scores high diet quality and risk of coronary heart disease in men and women. Am on theGDQSwouldtendtobecorrelatedwithdiets that are J Clin Nutr 2018;107:120–9. relatively more plant-based. 10. Gicevic S, Gaskins AJ, Fung TT, Rosner B, Tobias DK, Isanaka S, Willett Health metrics that have specific cutoffs are useful for risk WC. Evaluating pre-pregnancy dietary diversity vs. dietary quality scores as predictors of gestational diabetes and hypertensive disorders assessment and setting treatment targets. Clinically relevant of pregnancy. PLoS One 2018;13:e0195103. cutoffs can be identified if there are inflection points in the 11. Bromage S, Batis C, Bhupathiraju SN, Fawzi WW, Fung TT, Li relation of the GDQS and risk of diabetes. Cutoffs can also Y, Deitchler M, Angulo E, Birk N, Castellanos-Gutíerrez A, et al. be set by assigning a priori categories. However, this latter Development and validation of a novel food-based Global Diet Quality approach requires somewhat arbitrary decisions and also needs Score (GDQS). J Nutr 2021;151(Suppl 10):75S–92S. to consider other outcomes and diverse populations. In our 12. Saeedi P, Petersohn I, Salpea P, Malanda B, Karuranga S, Unwin N, results, there was no departure from linearity in the GDQS. Colagiuri S, Guariguata L, Motala AA, Ogurtsova K, et al. Global and regional diabetes prevalence estimates for 2019 and projections Because our results point toward a progressively lower risk for 2030 and 2045: results from the International Diabetes Federation of diabetes with higher GDQS, there is no strong premise to th Diabetes Atlas, 9 edition. Diabetes Res Clin Pract 2019;157:107843. support specific cutoffs for the GDQS in this cohort of US 13. CDC. National Diabetes Statistics Report, 2020. Atlanta, GA: CDC, US women. Department of Health and Human Services; 2020. In conclusion, the GDQS was inversely associated with type 14. American Diabetes Association. 9. Pharmacologic approaches to 2 diabetes in both reproductive-age and older women in a high- glycemic treatment: Standards of Medical Care in Diabetes — 2020. Diabetes Care 2020;43:S98–S110. income country. It performed well compared with the AHEI- 15. American Diabetes Association. 5. Facilitating behavior change and 2010 in predicting diabetes risk and our results showed that well-being to improve health outcomes: Standards of Medical Care in lower intake of unhealthy foods appeared to be more important Diabetes — 2020. Diabetes Care 2020;43:S48–65. than higher intake of healthy foods. Further testing of the 16. Bao Y, Bertoia ML, Lenart EB, Stampfer MJ, Willett WC, Speizer FE, GDQS in other populations is needed to confirm its usefulness Chavarro JE. Origin, methods, and evolution of the three Nurses’ Health in a broad range of populations to predict noncommunicable Studies. Am J Public Health 2016;106:1573–81. diseases. 17. Feskanich D, Rimm EB, Giovannucci EL, Colditz GA, Stampfer MJ, Litin LB, Willett WC. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. Acknowledgments J Am Diet Assoc 1993;93:790–6. The authors’ responsibilities were as follows—TTF, SNB, and 18. Women’s Dietary Diversity Project (WDDP) Study Group. Development SB: designed the research; TTF: analyzed the data, wrote the of a dichotomous indicator for population-level assessment of paper, and had primary responsibility for the final content; dietary diversity in women of reproductive age. Curr Dev Nutr WCW, MS, CB, MDH, FBH, YL, and MD: provided study 2017;1:cdn.117.001701. oversight and advice; and all authors: read and approved the 19. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes final manuscript. 1979;28:1039–57. 20. The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 1997;20:1183–97. References 21. Manson JE, Stampfer MJ, Colditz GA, Willett WC, Rosner B, Hennekens CH, Speizer FE, Rimm EB, Krolewski AS. Physical activity 1. Schulze MB, Martínez-González MA, Fung TT, Lichtenstein AH, and incidence of non-insulin-dependent diabetes mellitus in women. Forouhi NG. Food based dietary patterns and chronic disease Lancet 1991;338:774–8. prevention. BMJ 2018;361:k2396. 22. Hu FB, Stampfer MJ, Rimm E, Ascherio A, Rosner BA, Spiegelman D, 2. Chiuve SE, Fung TT, Rimm EB, Hu FB, McCullough ML, Wang M, Willett WC. Dietary fat and coronary heart disease: a comparison of Stampfer MJ, Willett WC. Alternative dietary indices both strongly approaches for adjusting for total energy intake and modeling repeated predict risk of chronic disease. J Nutr 2012;142:1009–18. dietary measurements. Am J Epidemiol 1999;149:531–40. 3. Kaluza J, Harris HR, Håkansson N, Wolk A. Adherence to the 23. Fung TT, McCullough M, van Dam RM, Hu FB. A prospective study WCRF/AICR 2018 recommendations for cancer prevention and risk of overall diet quality and risk of type 2 diabetes in women. Diabetes of cancer: prospective cohort studies of men and women. Br J Cancer Care 2007;30:1753–7. 2020;122:1562–70. 24. Cespedes EM, Hu FB, Tinker L, Rosner B, Redline S, Garcia L, Hingle 4. Kirkpatrick SI, Reedy J, Krebs-Smith SM, Pannucci TE, Subar AF, M, Van Horn L, Howard BV, Levitan EB, et al. Multiple healthful Wilson MM, Lerman JL, Tooze JA. Applications of the healthy dietary patterns and type 2 diabetes in the Women’s Health Initiative. eating index for surveillance, epidemiology, and intervention research: Am J Epidemiol 2016;183:622–33. considerations and caveats. J Acad Nutr Diet 2018;118:1603–21. 25. Chen G-C, Koh W-P, Neelakantan N, Yuan J-M, Qin L-Q, van Dam 5. Hernández-Ruiz A, García-Villanova B, Guerra Hernández EJ, Amiano RM. Diet quality indices and risk of type 2 diabetes mellitus: the P, Azpiri M, Molina-Montes E. Description of indexes based on the Singapore Chinese Health Study. Am J Epidemiol 2018;187:2651–61. adherence to the Mediterranean dietary pattern: a review. Nutr Hosp 26. Yu D, Zheng W, Cai H, Xiang Y-B, Li H, Gao Y-T, Shu X-O. Long-term 2015;32:1872–84. diet quality and risk of type 2 diabetes among urban Chinese adults. 6. Lacoppidan SA, Kyrø C, Loft S, Helnæs A, Christensen J, Hansen Diabetes Care 2018;41:723–30. CP, Dahm CC, Overvad K, Tjønneland A, Olsen A. Adherence to a 27. Micha R, Michas G, Mozaffarian D. Unprocessed red and processed healthy Nordic Food Index is associated with a lower risk of type-2 meats and risk of coronary artery disease and type 2 diabetes – diabetes—the Danish Diet, Cancer and Health cohort study. Nutrients an updated review of the evidence. Curr Atheroscler Rep 2012;14: 2015;7:8633–44. 515–24. 7. Puaschitz NG, Assmus J, Strand E, Karlsson T, Vinknes KJ, Lysne V, 28. Schwingshackl L, Hoffmann G, Lampousi A-M, Knüppel S, Iqbal K, Drevon CA, Tell GS, Dierkes J, Nygård O. Adherence to the Healthy Schwedhelm C, Bechthold A, Schlesinger S, Boeing H. Food groups and Nordic Food Index and the incidence of acute myocardial infarction risk of type 2 diabetes mellitus: a systematic review and meta-analysis and mortality among patients with stable angina pectoris. J Hum Nutr of prospective studies. Eur J Epidemiol 2017;32:363–75. Diet 2019;32:86–97. 174S Supplement 29. Schwingshackl L, Schwedhelm C, Hoffmann G, Boeing H. Potatoes and 38. Ludwig DS. The glycemic index: physiological mechanisms relating to risk of chronic disease: a systematic review and dose–response meta- obesity, diabetes, and cardiovascular disease. JAMA 2002;287:2414– analysis. Eur J Nutr 2019;58:2243–51. 23. 30. Cahill LE, Pan A, Chiuve SE, Sun Q, Willett WC, Hu FB, Rimm EB. 39. Jayedi A, Soltani S, Abdolshahi A, Shab-Bidar S. Healthy and unhealthy Fried-food consumption and risk of type 2 diabetes and coronary artery dietary patterns and the risk of chronic disease: an umbrella review of disease: a prospective study in 2 cohorts of US women and men. Am J meta-analyses of prospective cohort studies. Br J Nutr 2020;124:1133– Clin Nutr 2014;100:667–75. 44. 31. Osorio-Yáñez C, Gelaye B, Qiu C, Bao W, Cardenas A, Enquobahrie 40. Djoussé L, Khawaja OA, Gaziano JM. Egg consumption and risk of DA, Williams MA. Maternal intake of fried foods and risk of gestational type 2 diabetes: a meta-analysis of prospective studies. Am J Clin Nutr diabetes mellitus. Ann Epidemiol 2017;27:384–90.e1. 2016;103:474–80. 32. Gupta MK. Frying oils. In: Shahidi F, editor. Baileys’ industrial oil and 41. Yuan C, Spiegelman D, Rimm EB, Rosner BA, Stampfer MJ, Barnett fat products. Hoboken, NJ: John Wiley & Sons; 2005. JB, Chavarro JE, Subar AF, Sampson LK, Willett WC. Validity of a dietary questionnaire assessed by comparison with multiple 33. Yang M, Yang Y, Nie S, Xie M, Chen F. Analysis and formation of trans weighed dietary records or 24-hour recalls. Am J Epidemiol 2017;185: fatty acids in corn oil during the heating process. J Am Oil Chem Soc 570–84. 2012;89:859–67. 42. Carlström M, Larsson SC. Coffee consumption and reduced risk of 34. Uribarri J, Woodruff S, Goodman S, Cai W, Chen X, Pyzik R, Yong developing type 2 diabetes: a systematic review with meta-analysis. Nutr A, Striker GE, Vlassara H. Advanced glycation end products in foods Rev 2018;76:395–417. and a practical guide to their reduction in the diet. J Am Diet Assoc 2010;110:911–16.e12. 43. Knott C, Bell S, Britton A. Alcohol consumption and the risk of type 2 diabetes: a systematic review and dose-response meta-analysis of more 35. Rajpathak SN, Crandall JP, Wylie-Rosett J, Kabat GC, Rohan TE, Hu than 1.9 million individuals from 38 observational studies. Diabetes FB. The role of iron in type 2 diabetes in humans. Biochim Biophys Acta Care 2015;38:1804–12. 2009;1790:671–81. 36. Pacher P, Beckman JS, Liaudet L. Nitric oxide and peroxynitrite in 44. Willett W, Rockström J, Loken B, Springmann M, Lang T, Vermeulen health and disease. Physiol Rev 2007;87:315–424. S, Garnett T, Tilman D, DeClerck F, Wood A, et al. Food in the Anthropocene: the EAT–Lancet Commission on healthy diets from 37. Malik VS, Hu FB. Sweeteners and risk of obesity and type 2 diabetes: the sustainable food systems. Lancet 2019;393:447–92. role of sugar-sweetened beverages. Curr Diab Rep 2012;12:195–203. GDQS and diabetes risk 175S

Journal

The Journal of Nutrition – Pubmed Central

Published: Oct 23, 2021

There are no references for this article.