ABSTRACT
Background
Previous studies have shown mixed results on the association between carbohydrate intake and insomnia. However, any influence that refined carbohydrates have on risk of insomnia is likely commensurate with their relative contribution to the overall diet, so studies are needed that measure overall dietary glycemic index (GI), glycemic load, and intakes of specific types of carbohydrates.
Objective
We hypothesized that higher GI and glycemic load would be associated with greater odds of insomnia prevalence and incidence.
Methods
This was a prospective cohort study with postmenopausal women who participated in the Women's Health Initiative Observational Study, investigating the relations of GI, glycemic load, other carbohydrate measures (added sugars, starch, total carbohydrate), dietary fiber, and specific carbohydrate-containing foods (whole grains, nonwhole/refined grains, nonjuice fruits, vegetables, dairy products) with odds of insomnia at baseline (between 1994 and 1998; n = 77,860) and after 3 y of follow-up (between 1997 and 2001; n = 53,069).
Results
In cross-sectional and longitudinal analyses, higher dietary GI was associated with increasing odds of prevalent (fifth compared with first quintile OR: 1.11; CI: 1.05, 1.16; P-trend = 0.0014) and incident (fifth compared with first quintile OR: 1.16; CI: 1.08, 1.25; P-trend < 0.0001) insomnia in fully adjusted models. Higher intakes of dietary added sugars, starch, and nonwhole/refined grains were each associated with higher odds of incident insomnia. By contrast, higher nonjuice fruit and vegetable intakes were significantly associated with lower odds of incident insomnia. Also, higher intakes of dietary fiber, whole grains, nonjuice fruit, and vegetables were significantly associated with lower odds of prevalent insomnia.
Conclusions
The results suggest that high-GI diets could be a risk factor for insomnia in postmenopausal women. Substitution of high-GI foods with minimally processed, whole, fiber-rich carbohydrates should be evaluated as potential treatments of, and primary preventive measures for, insomnia in postmenopausal women.
Keywords: insomnia, glycemic index, glycemic load, epidemiology, postmenopausal women
Introduction
Insomnia has high personal, societal, and financial costs and disproportionately affects women. It is associated with automobile accidents, problems with work performance, decreased quality of life, and increased medical and psychiatric comorbidities (1). The annual expenses from insomnia due to direct medical costs, lost productivity, accidents, and insomnia-related depression and alcohol abuse have been estimated to be $150–175 billion in 2016 dollars (2).
Cognitive behavioral therapy and pharmacotherapy are widely accepted treatments of insomnia, yet psychotherapy can be time-consuming and expensive and pharmacotherapy with sedative hypnotics can increase the risk of adverse cognitive and psychomotor events and daytime fatigue (3). It is therefore beneficial to identify novel risk factors for insomnia that could suggest straightforward and low-cost interventions with fewer potential iatrogenic effects.
The relation between diet and sleep is an emerging area of inquiry, with macronutrients and carbohydrate intake having recently been explored (4); however, previous studies have yielded inconsistent findings. Two cross-sectional studies found low carbohydrate intakes in subjects with insomnia symptoms (5, 6), whereas another cross-sectional study of middle-aged female Japanese workers found increased consumption of confectionary to be significantly associated with poor sleep quality (7). There was a significant trend toward worse sleep quality with increasing carbohydrate intake. Further, participants with poor sleep quality had the highest carbohydrate intake and consumed more confectionary and less rice. The authors therefore hypothesized that both the total amount of carbohydrate and the glycemic index (GI) of the carbohydrates consumed could affect sleep quality. However, because these studies were cross-sectional, and given that the relation between carbohydrate consumption and insomnia could be bidirectional, prospective studies are needed to further examine this association.
Any influence that refined carbohydrates have on insomnia is likely commensurate with the proportion of the overall diet they constitute, so dietary GI and glycemic load are good measures to evaluate the quality and quantity of carbohydrate intake. An experimental study in young 18- to 35-y-old men with random assignment to a high-GI carbohydrate meal 4 h before bedtime showed a significantly shortened sleep onset latency compared with a low-GI meal (8). A limitation of this experimental study was its small sample size and short duration. Longitudinal epidemiologic studies can enhance causal inference by establishing the temporal order of variables, incorporating longer durations, and including larger sample sizes that allow for the control of more factors. To our knowledge, there are no current prospective epidemiologic studies examining the relation between GI and glycemic load and subsequent insomnia.
The Women's Health Initiative (WHI) Observational Study of postmenopausal women provides a unique opportunity to explore the association between dietary GI and glycemic load and insomnia. We hypothesized that higher GI and glycemic load were associated with greater odds of prevalent and incident insomnia in a large, well-characterized sample of postmenopausal women who were followed longitudinally.
Methods
The WHI Observational Study includes a socioeconomically and racially/ethnically diverse cohort of 93,676 postmenopausal women between the ages of 50 and 79 y from 40 clinical centers in 24 states and the District of Columbia recruited between 1 September, 1994 and 31 December, 1998 (9). Women were excluded if they did not plan on residing in the area for ≥3 y, had a life expectancy of <3 y, or suffered from substance abuse, mental illness, or dementia. We conducted cross-sectional analyses so our results could be compared with other cross-sectional studies (5–7). Because the relation between diet and insomnia could be bidirectional, we also conducted longitudinal analyses, excluding baseline insomnia, to better explore a potential cause–effect relation. The cross-sectional analyses for this study included women who completed the food questionnaire and Women's Health Initiative Insomnia Rating Scale (WHIIRS) at baseline (n = 77,860). Persons with insomnia at baseline were excluded from the longitudinal analyses, which included women who completed the food questionnaire at baseline and the WHIIRS after 3 y of follow-up (n = 53,069). Supplemental Figure 1 shows a flow diagram of the study participants. All participants provided informed consent with documents approved by institutional review boards at all 40 study sites across the United States. This study was approved by the institutional review board of Columbia University/New York State Psychiatric Institute.
Ascertainment of dietary variables
At baseline, participants completed a 145-item FFQ designed for the WHI. The dietary variables were computed from mean daily intake of foods and beverages reported on the WHI FFQ. Data from 113 women screened for participation in the WHI were used to test the reliability and validity of the WHI FFQ (10). The mean energy-adjusted correlation coefficient between 30 nutrient estimates from the FFQ and the means from four 24-h dietary recalls and a 4-d food record was 0.50. The energy-adjusted de-attenuated correlation coefficient between carbohydrate intake estimated by the WHI FFQ and 8 d of dietary intake was 0.63 and for fiber 0.65. Test–retest reliability for the nutrient intake estimates between the first and second administrations of the WHI FFQ was high, with mean intraclass correlation coefficients of 0.76.
GI was defined as an index of the postprandial glucose response of a food, compared with an equal amount of carbohydrate (50 g) from a reference food, typically glucose or white bread (11, 12). The GI variable in the WHI was applied to available carbohydrate (total carbohydrate less dietary fiber) with glucose being used as the reference food. The GI of a specific food is equal to the blood glucose incremental AUC of the test food for a given time postconsumption divided by the blood glucose incremental AUC of the reference food multiplied by 100. Dietary GI is considered a measure of the quality of carbohydrate-based foods in the overall diet and is estimated as the weighted mean (with weights based on the total carbohydrate content per serving consumed) of the GI values of all carbohydrate foods consumed during the dietary period. The glycemic load of a food is equal to its GI multiplied by the total grams of carbohydrate per serving divided by 100. Dietary glycemic load is estimated as the sum of the glycemic loads of all carbohydrate foods consumed during the dietary period unadjusted for energy intake. The methodology used to construct the GI and glycemic load database for the WHI was detailed previously (11).
Although the exposures of interest were GI and glycemic load, we also examined other measures of carbohydrate consumption computed from mean daily intake of foods and beverages reported on the WHI FFQ, including dietary added sugar, total sugars, starch, and total carbohydrate. The USDA's MyPyramid Equivalents Database 2.0 (13) was used to assess added sugars used as ingredients in processed and prepared foods such as cakes, breads, sodas, jellies, chocolates, and ice cream, and sugars eaten separately or added to foods at the table. Examples of added sugars include white sugar, brown sugar, raw sugar, corn syrup, corn syrup solids, high-fructose corn syrup, malt syrup, maple syrup, pancake syrup, fructose sweetener, honey, molasses, anhydrous dextrose, and dextrin. Added sugars do not include naturally occurring sugars such as fructose in fruit or lactose in milk, unless the sugar is added to the food item. The percentage total sugar content of foods is used to estimate gram equivalents of added sugars in caloric sweeteners. The number of added sugars equivalents for each sweetener ingredient in a multi-ingredient food is calculated using the recipe retention factor method (14) and then totaled: total sugars in the sweetener ingredient in 100 g of a multi-ingredient food (grams) = [weight of sweetener ingredient in 100 grams of food × (total sugars in 100 g ingredient/100) × (% retention)]/100% + (%moisture change) + (%fat change).
Dietary fiber and specific carbohydrate-containing foods (whole grains, nonwhole/refined grains, nonjuice fruits, vegetables, and dairy products) were also considered. Dietary fiber and specific types and sources of carbohydrates were categorized into quintiles for analyses.
Ascertainment of insomnia
The presence of insomnia was measured at baseline and at 3-y follow-up using participant responses to the WHIIRS, a 5-item instrument that has been found to be a reliable and valid measure of perceived insomnia symptoms. The construct validity of the WHIIRS was supported by successfully detecting self-reported sleep disturbance differences in women taking hormone therapy compared with those taking a placebo as well as in groups known to differ in severity of their vasomotor symptoms (15). The WHIIRS was found to have a highly stable factor structure with no major differences across age, race, and ethnicity groups (16). A score of ≥9 indicates a high risk of insomnia and the need for further clinical evaluation (17). We refer to a score of ≥9 on the WHIIRS as insomnia throughout the article.
Covariates
We chose as covariates for our multivariable models variables theorized to vary by diet and by insomnia and which therefore could act as confounders or mediators of the relation between the dietary variables and insomnia. Covariates in the analyses included age (5-y interval); race/ethnicity (American Indian/Alaskan Native, Asian/Pacific Islander, black, Hispanic, white not of Hispanic origin, missing); education (less than high school graduate, high school graduate, some college, college graduate, postgraduate, missing); annual income (<$10,000, $10K–19,999, $20K–34,999, $35K–49,999, $50K–74,999, $75K–99,999, $100K–149,999, ≥$150,000, don't know, missing); live alone (no, yes, missing); live with husband or partner (no, yes, missing); live with children (no, yes, missing); smoking status (never smoked, past smoker, current smoker, missing); alcohol intake (nondrinking, past drinker, <1 drink/mo, <1 drink/wk, 1 to <7 drinks/wk, ≥7 drinks/wk, missing); caffeine intake (quintiles); stressful life events (quintiles, missing); social support (quintiles, missing); depression (no, yes, missing); physical activity measured in metabolic equivalent (MET)-hours per week [<3, 3 to <9, 9 to <18, 18 to <27, ≥27 MET-h/wk (1 MET = 58.2 W/m2), missing]; BMI (in kg/m2) (underweight: <18.5; normal: 18.5–24.9; overweight: 25.0–29.9; obesity I: 30.0–34.9; obesity II: 35.0–39.9; obesity III: ≥40; missing); diabetes (no, yes, missing); hypertension (no, yes, missing); myocardial infarction (no, yes, missing); cardiovascular disease (no, yes, missing); asthma (no, yes, missing); overactive thyroid (no, yes, missing); bodily pain (none, very mild, mild, moderate, severe, missing); hot flashes (none, mild, moderate, severe, missing); hormone replacement therapy (no, yes, missing); and snoring (not in past 4 wk, less than once a week, 1 or 2 times/wk, 3 or 4 times/wk, ≥5 times/wk, don't know, missing).
Statistical analyses
Chi-square tests for categorical variables and t tests for continuous variables were used to evaluate differences by insomnia and by GI quintiles. Multivariable logistic regression was used to calculate ORs to examine the cross-sectional relation between the dietary variables and insomnia prevalence. Persons with insomnia at baseline were excluded from the longitudinal analyses. The longitudinal relation between dietary variables and incident insomnia after 3 y was examined using multivariable logistic regression to calculate energy-adjusted ORs. We adjusted for total energy intake because it can be associated with disease risk owing to differences in physical activity, body size, and metabolic efficiency. In addition, specific nutrients contribute toward total energy intake and individuals who consume more total energy also tend to eat more specific nutrients. The energy partition model (18) was used to adjust for energy consumption (Model 1) in analyses of dietary GI, glycemic load, added sugars, total sugars, starch, and total carbohydrate. Glycemic load, added sugars, total sugars, starch, and total carbohydrate are macronutrients that are highly collinear with total calories. The energy partition model allows the examination of the separate effects of calories and macronutrients by subtracting the macronutrient calories from the total calories, therefore creating a variable that is less highly correlated with the macronutrient. In the energy partition model, the primary nutrient coefficient is included as 1 term and energy from other nutrients is included as a second term in the multivariate models. It was not possible to precisely measure the number of calories in each increment of fiber, whole grains, nonwhole/refined grains, nonjuice fruit, vegetables, and dairy products, which would have been required to use the energy partition model. Also, these foods make up a smaller proportion of the overall diet and therefore are not highly collinear with total calories. The nutrient density model (19) was therefore used to adjust for energy consumption (Model 1) in analyses of fiber, whole grains, nonwhole/refined grains, nonjuice fruit, vegetables, and dairy products. In the nutrient density model, nutrient densities are computed by dividing nutrient values by total caloric intake and then included with total caloric intake in the multivariable models. To explore potential attenuation in the ORs from controlling for different types of covariates, we included variables in Model 2 that could act as confounders and included variables in Model 3 that could act as mediators. Covariates in the first multivariable-adjusted model (Model 2) included age, race/ethnicity, education, income, smoking, alcohol, caffeine, stressful life events, social support, overactive thyroid, bodily pain, hormone replacement therapy, and snoring. The final adjusted model (Model 3) included the variables in Model 2, plus depression, physical activity, BMI, diabetes, hypertension, myocardial infarction, cardiovascular disease, asthma, and hot flashes. Stratified analyses were conducted to explore interaction by physical activity. Tests for linear trend were performed by modeling a numeric value (−2, −1, 0, 1, 2) for each dietary quintile category.
Results are reported as ORs and 95% CIs. P values < 0.05 were considered statistically significant. Statistical analyses were conducted using SAS statistical software version 9.4 (SAS Institute).
Results
The baseline characteristics for women in the WHI Observational Study population according to their GI quintile, based on available carbohydrate, and incidence of insomnia after 3 y of follow-up are shown in Tables 1 and 2. Higher GI was associated with younger age, higher BMI, and less physical activity. Higher GI was also associated with African ancestry, lower education, lower income, not living alone, living with husband or partner, living with children, smoking, abstinence from alcohol, caffeine consumption, more stressful life events, less social support, depression, hypertension, myocardial infarction, overactive thyroid, bodily pain, hot flashes, not receiving hormone replacement therapy, and snoring. Insomnia was associated with higher GI quintiles, higher BMI, and less physical activity. Insomnia was also associated with white race/ethnicity, lower education, lower income, living with husband or partner, past and current smoking, more stressful life events, less social support, depression, diabetes, hypertension, myocardial infarction, cardiovascular disease, asthma, overactive thyroid, bodily pain, hot flashes, hormone replacement therapy, and snoring.
TABLE 1.
Baseline characteristics by dietary GI quintile and incidence of insomnia 3 y later, continuous variables1
| Dietary GI quintiles | Incident insomnia 3 y later | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Baseline characteristics | 1 (Low) | 2 | 3 | 4 | 5 (High) | P 2 | Yes | No | P 2 |
| Mean insomnia WHIIRS | 5.064 ± 3.66 | 5.263 ± 3.70 | 5.296 ± 3.63 | 5.413 ± 3.73 | 5.501 ± 3.80 | <0.0001 | 11.278 ± 2.36 | 3.957 ± 2.39 | <0.0001 |
| Age, y | 63.77 ± 7.3 | 63.56 ± 7.3 | 63.48 ± 7.3 | 63.32 ± 7.3 | 62.63 ± 7.2 | <0.0001 | 63.39 ± 7.3 | 63.35 ± 7.3 | 0.5837 |
| BMI, kg/m2 | 26.41 ± 5.3 | 26.46 ± 5.4 | 26.78 ± 5.5 | 27.13 ± 5.7 | 27.77 ± 6.1 | <0.0001 | 27.36 ± 5.9 | 26.81 ± 5.6 | <0.0001 |
| Physical activity,MET-h/wk | 17.72 ± 16.0 | 16.06 ± 14.8 | 14.71 ± 14.3 | 13.27 ± 13.7 | 11.10 ± 13.0 | <0.0001 | 13.57 ± 14.1 | 14.80 ± 14.7 | <0.0001 |
Values are means ± SDs unless stated otherwise. GI, glycemic index; MET, metabolic equivalent; WHIIRS, Women's Health Initiative Insomnia Rating Scale.
Differences by insomnia and GI quintiles were tested with t tests for continuous variables.
TABLE 2.
Baseline characteristics by dietary GI quintile and incidence of insomnia 3 y later, categorical variables1
| Dietary GI quintiles | Incident insomnia 3 y later | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Baseline characteristics | n | 1 (Low) | 2 | 3 | 4 | 5 (High) | P 2 | Yes | No | P 2 |
| Total n | 53,069 | 10,613 | 10,614 | 10,614 | 10,614 | 10,614 | 9783 | 43,286 | ||
| Insomnia n | 9783 | 1728 | 1924 | 1943 | 2065 | 2123 | <0.0001 | 9783 | 43,286 | |
| Median GI | 47.0 | 49.8 | 51.7 | 53.5 | 56.3 | |||||
| Race/ethnicity, % | <0.0001 | 0.0079 | ||||||||
| American Indian/Alaskan Native | 730 | 1.4 | 1.4 | 1.2 | 1.3 | 1.6 | 1.4 | 1.4 | ||
| Asian/Pacific Islander | 1653 | 2.3 | 3.1 | 3.8 | 3.7 | 2.6 | 2.5 | 3.3 | ||
| Black | 3474 | 4.7 | 4.3 | 5.4 | 6.8 | 11.6 | 6.3 | 6.6 | ||
| Hispanic | 1543 | 3.5 | 2.9 | 2.7 | 2.9 | 2.5 | 3.0 | 3.0 | ||
| White not of Hispanic origin | 45,528 | 87.7 | 87.9 | 86.8 | 85.1 | 81.4 | 86.5 | 85.6 | ||
| Missing race/ethnicity | 141 | 0.3 | 0.3 | 0.2 | 0.3 | 0.3 | 0.3 | 0.3 | ||
| Education ≥college grad, % | 24,350 | 52.8 | 50.4 | 47.9 | 43.2 | 35.2 | <0.0001 | 42.4 | 46.7 | <0.0001 |
| Annual income ≥$50,000, % | 21,941 | 48.0 | 45.7 | 43.8 | 41.1 | 36.7 | <0.0001 | 40.5 | 43.7 | <0.0001 |
| Live alone, % | 14,040 | 28.7 | 27.5 | 26.2 | 25.2 | 24.7 | <0.0001 | 25.8 | 26.6 | 0.1680 |
| Live with husband or partner, % | 33,934 | 62.8 | 64.1 | 64.7 | 64.8 | 63.4 | 0.0003 | 65.3 | 63.6 | 0.0060 |
| Live with children, % | 5403 | 8.3 | 8.9 | 9.6 | 11.5 | 12.7 | <0.0001 | 10.3 | 10.2 | 0.9117 |
| Smoking status, % | <0.0001 | <0.0001 | ||||||||
| Never smoked | 27,042 | 48.5 | 51.6 | 52.4 | 52.2 | 50.5 | 48.6 | 51.5 | ||
| Past smoker | 22,429 | 45.6 | 43.0 | 41.7 | 40.6 | 40.5 | 44.3 | 41.8 | ||
| Current smoker | 2993 | 4.6 | 4.4 | 5.1 | 6.2 | 7.8 | 5.8 | 5.6 | ||
| Missing smoking | 605 | 1.4 | 1.0 | 1.2 | 1.0 | 1.2 | 1.3 | 1.1 | ||
| Alcohol intake, % | <0.0001 | <0.0001 | ||||||||
| Nondrinker | 5428 | 7.9 | 8.2 | 9.7 | 11.1 | 14.1 | 9.7 | 10.3 | ||
| Past drinker | 9010 | 15.6 | 14.8 | 16.0 | 17.7 | 20.8 | 18.5 | 16.6 | ||
| <1 drink/mo | 6108 | 9.5 | 10.0 | 11.5 | 11.8 | 14.7 | 12.0 | 11.4 | ||
| <1 drink/wk | 10,820 | 19.3 | 20.8 | 20.4 | 21.1 | 20.3 | 20.4 | 20.4 | ||
| 1 to <7/wk | 14,393 | 28.7 | 30.7 | 29.0 | 26.6 | 20.7 | 26.1 | 27.4 | ||
| ≥7/wk | 7061 | 18.3 | 15.0 | 13.1 | 11.3 | 8.8 | 12.8 | 13.4 | ||
| Missing alcohol | 249 | 0.6 | 0.5 | 0.4 | 0.5 | 0.5 | 0.5 | 0.5 | ||
| Caffeine, % | <0.0001 | 0.2737 | ||||||||
| Quintile 1 | 10,613 | 21.5 | 19.8 | 19.6 | 19.5 | 19.5 | 20.0 | 20.0 | ||
| Quintile 2 | 10,614 | 20.5 | 20.6 | 20.2 | 20.0 | 18.7 | 20.1 | 20.0 | ||
| Quintile 3 | 10,615 | 23.8 | 21.0 | 18.9 | 18.2 | 18.2 | 19.7 | 20.1 | ||
| Quintile 4 | 10,613 | 13.5 | 19.3 | 21.7 | 22.4 | 23.0 | 20.7 | 19.8 | ||
| Quintile 5 | 10,614 | 20.7 | 19.3 | 19.5 | 20.0 | 20.5 | 19.5 | 20.1 | ||
| Stressful life events, % | <0.0001 | <0.0001 | ||||||||
| Quintile 1 | 15,361 | 29.8 | 29.6 | 29.4 | 28.5 | 27.5 | 23.6 | 30.2 | ||
| Quintile 2 | 6792 | 12.8 | 13.2 | 13.1 | 12.9 | 12.0 | 10.2 | 13.4 | ||
| Quintile 3 | 8902 | 16.2 | 17.0 | 17.0 | 16.9 | 16.7 | 15.9 | 17.0 | ||
| Quintile 4 | 11,684 | 21.8 | 22.2 | 21.4 | 22.3 | 22.3 | 23.6 | 21.7 | ||
| Quintile 5 | 9396 | 17.4 | 16.5 | 17.1 | 17.5 | 20.0 | 25.1 | 16.0 | ||
| Missing stressful life events | 934 | 1.9 | 1.6 | 2.0 | 1.9 | 1.4 | 1.7 | 1.8 | ||
| Social support, % | <0.0001 | <0.0001 | ||||||||
| Quintile 1 | 10,969 | 20.0 | 19.6 | 20.8 | 20.6 | 22.3 | 25.3 | 19.6 | ||
| Quintile 2 | 10,924 | 20.4 | 20.9 | 20.6 | 20.8 | 20.2 | 21.7 | 20.3 | ||
| Quintile 3 | 10,362 | 19.0 | 20.2 | 19.6 | 19.5 | 19.3 | 18.0 | 20.0 | ||
| Quintile 4 | 10,552 | 20.4 | 20.2 | 19.4 | 20.2 | 19.2 | 18.5 | 20.2 | ||
| Quintile 5 | 9169 | 18.2 | 16.9 | 17.7 | 16.8 | 16.7 | 14.5 | 17.9 | ||
| Missing social support | 1093 | 2.0 | 2.1 | 1.9 | 2.0 | 2.4 | 2.0 | 2.1 | ||
| Depression, % | 3285 | 5.3 | 5.6 | 6.3 | 6.4 | 7.5 | <0.0001 | 9.5 | 5.4 | <0.0001 |
| Diabetes, % | 2375 | 4.4 | 4.4 | 4.3 | 4.6 | 4.7 | 0.9285 | 4.9 | 4.4 | 0.0296 |
| Hypertension, % | 16,104 | 28.1 | 28.8 | 30.6 | 31.3 | 32.9 | <0.0001 | 32.9 | 30.0 | <0.0001 |
| Myocardial infarction, % | 1051 | 1.9 | 1.6 | 1.8 | 2.2 | 2.4 | 0.0001 | 2.5 | 1.9 | 0.0004 |
| Cardiovascular disease, % | 8951 | 16.5 | 16.7 | 17.0 | 17.0 | 17.3 | 0.2214 | 19.6 | 16.3 | <0.0001 |
| Asthma, % | 3829 | 7.4 | 7.0 | 7.2 | 7.0 | 7.6 | 0.4081 | 8.8 | 6.9 | <0.0001 |
| Overactive thyroid, % | 1447 | 2.4 | 2.6 | 2.7 | 2.8 | 3.2 | <0.0001 | 3.0 | 2.7 | <0.0001 |
| Bodily pain, % | <0.0001 | <0.0001 | ||||||||
| None | 11,894 | 25.1 | 23.0 | 22.3 | 21.2 | 20.4 | 13.7 | 24.4 | ||
| Very mild | 20,967 | 39.8 | 40.4 | 39.8 | 39.5 | 38.0 | 33.7 | 40.8 | ||
| Mild | 10,321 | 18.0 | 19.3 | 19.4 | 20.3 | 20.3 | 23.0 | 18.7 | ||
| Moderate | 8254 | 14.3 | 14.3 | 15.4 | 15.9 | 18.0 | 23.4 | 13.8 | ||
| Severe | 1534 | 2.7 | 2.8 | 2.9 | 2.9 | 3.2 | 6.2 | 2.2 | ||
| Missing bodily pain | 99 | 0.2 | 0.2 | 0.1 | 0.2 | 0.2 | 0.1 | 0.2 | ||
| Hot flashes, % | <0.0001 | <0.0001 | ||||||||
| None | 42,764 | 82.7 | 83.0 | 80.8 | 79.7 | 76.9 | 77.4 | 81.3 | ||
| Mild | 7889 | 13.5 | 13.7 | 14.7 | 15.3 | 17.1 | 16.5 | 14.5 | ||
| Moderate | 1892 | 2.9 | 2.7 | 3.5 | 3.9 | 4.8 | 4.6 | 3.3 | ||
| Severe | 381 | 0.6 | 0.5 | 0.7 | 0.7 | 1.0 | 1.2 | 0.6 | ||
| Missing hot flashes | 143 | 0.3 | 0.2 | 0.3 | 0.4 | 0.2 | 0.3 | 0.3 | ||
| Hormone replacement therapy, % | 26,509 | 51.1 | 52.3 | 50.5 | 49.2 | 46.7 | <0.0001 | 50.9 | 49.8 | <0.0001 |
| Snoring, % | <0.0001 | <0.0001 | ||||||||
| Not in past 4 wk | 12,712 | 26.8 | 24.7 | 24.4 | 23.0 | 20.8 | 22.5 | 24.3 | ||
| Less than once a week | 2722 | 5.6 | 5.1 | 4.9 | 5.0 | 5.0 | 5.2 | 5.1 | ||
| 1 or 2 times/wk | 3282 | 6.0 | 5.9 | 6.6 | 6.4 | 6.1 | 5.9 | 6.3 | ||
| 3 or 4 times/wk | 2574 | 4.3 | 4.9 | 4.8 | 5.2 | 5.1 | 5.1 | 4.8 | ||
| ≥5 times/wk | 4792 | 7.6 | 7.8 | 8.5 | 9.9 | 11.4 | 10.3 | 8.7 | ||
| Don't know | 26,806 | 49.3 | 51.4 | 50.6 | 50.2 | 51.1 | 50.6 | 50.5 | ||
| Missing snoring | 181 | 0.4 | 0.3 | 0.3 | 0.3 | 0.4 | 0.4 | 0.3 | ||
GI, glycemic index.
Differences by depression and GI quintiles were tested with chi-square (χ2) tests for categorical variables.
Table 3 shows the results from the cross-sectional multivariable analyses at the baseline visit (n = 77,860). In energy-adjusted results (Model 1), participants whose dietary GIs were in the third, fourth, and fifth quintiles reported significantly more insomnia than the participants in the first quintile. The inclusion of variables in Models 2 and 3 progressively attenuated the associations, resulting in only the fifth quintile for GI being significantly associated with insomnia. The test for trend was statistically significant for all 3 models. As the consumption of dietary added sugars and nonwhole/refined grains increased, the likelihood of experiencing co-occurring insomnia was greater with the trends being statistically significant. The odds of insomnia prevalence were lower with higher consumption of dietary fiber, whole grains, nonjuice fruit, and vegetables and the trends were statistically significant. There were no associations of glycemic load, dietary total sugars, starch, total carbohydrate, or dairy products with insomnia prevalence. Physical activity was not found to act as an effect modifier in stratified analyses.
TABLE 3.
Adjusted ORs and 95% CIs from logistic regression analyses of insomnia prevalence according to quintiles of energy-adjusted GI, glycemic load, carbohydrate consumption, fiber, whole grains, nonwhole/refined grains, nonjuice fruit, vegetables, and dairy products1
| Prevalent insomnia at baseline, n | Prevalent insomnia at baseline, OR (95% CI) | ||||
|---|---|---|---|---|---|
| Median | Model 1 | Model 2 | Model 3 | ||
| Dietary GI | |||||
| Quintile 1 | 47.1 | 4478 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 49.9 | 4667 | 1.05 (1.00, 1.10) | 1.04 (0.99, 1.09) | 1.03 (0.98, 1.09) |
| Quintile 3 | 51.7 | 4785 | 1.08 (1.03, 1.13) | 1.06 (1.01, 1.11) | 1.04 (0.99, 1.10) |
| Quintile 4 | 53.6 | 4835 | 1.09 (1.04, 1.14) | 1.05 (0.99, 1.10) | 1.02 (0.97, 1.08) |
| Quintile 5 | 56.4 | 5257 | 1.23 (1.17, 1.29) | 1.14 (1.08, 1.20) | 1.11 (1.05, 1.16) |
| P-trend | P < 0.0001 | P < 0.0001 | P = 0.0014 | ||
| Dietary glycemic load, g/d | |||||
| Quintile 1 | 53.8 | 4647 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 74.9 | 4611 | 0.94 (0.90, 0.99) | 0.98 (0.93, 1.03) | 0.99 (0.94, 1.04) |
| Quintile 3 | 92.3 | 4817 | 0.96 (0.91, 1.01) | 1.02 (0.97, 1.08) | 1.03 (0.98, 1.09) |
| Quintile 4 | 112.0 | 4848 | 0.93 (0.88, 0.98) | 0.99 (0.94, 1.05) | 1.01 (0.95, 1.06) |
| Quintile 5 | 147.5 | 5099 | 0.94 (0.89, 0.99) | 1.00 (0.95, 1.07) | 1.01 (0.95, 1.07) |
| P-trend | P = 0.0272 | P = 0.7506 | P = 0.5603 | ||
| Dietary added sugar, g | |||||
| Quintile 1 | 17.9 | 4436 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 29.0 | 4709 | 1.06 (1.01, 1.12) | 1.07 (1.01, 1.12) | 1.07 (1.02, 1.13) |
| Quintile 3 | 39.6 | 4815 | 1.07 (1.02, 1.13) | 1.08 (1.03, 1.14) | 1.09 (1.03, 1.15) |
| Quintile 4 | 53.1 | 4882 | 1.07 (1.01, 1.13) | 1.07 (1.01, 1.13) | 1.07 (1.01, 1.13) |
| Quintile 5 | 80.5 | 5180 | 1.13 (1.07, 1.20) | 1.10 (1.04, 1.17) | 1.09 (1.03, 1.16) |
| P-trend | P < 0.0001 | P = 0.0051 | P = 0.0182 | ||
| Dietary total sugars, g | |||||
| Quintile 1 | 50.4 | 4795 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 73.8 | 4784 | 0.96 (0.91, 1.01) | 0.99 (0.94, 1.04) | 1.00 (0.95, 1.05) |
| Quintile 3 | 93.4 | 4735 | 0.91 (0.87, 0.96) | 0.96 (0.91, 1.01) | 0.97 (0.92, 1.02) |
| Quintile 4 | 116.0 | 4749 | 0.88 (0.84, 0.93) | 0.94 (0.89, 0.99) | 0.95 (0.90, 1.01) |
| Quintile 5 | 156.1 | 4959 | 0.89 (0.84, 0.94) | 0.94 (0.89, 0.99) | 0.95 (0.90, 1.01) |
| P-trend | P < 0.0001 | P = 0.0075 | P = 0.0277 | ||
| Dietary starch, g | |||||
| Quintile 1 | 38.4 | 4586 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 56.1 | 4627 | 0.97 (0.92, 1.02) | 1.00 (0.95, 1.05) | 1.01 (0.95, 1.06) |
| Quintile 3 | 70.8 | 4775 | 0.98 (0.93, 1.03) | 1.02 (0.97, 1.08) | 1.03 (0.97, 1.09) |
| Quintile 4 | 88.5 | 4950 | 0.99 (0.93, 1.04) | 1.05 (0.99, 1.11) | 1.06 (1.00, 1.12) |
| Quintile 5 | 120.3 | 5084 | 0.97 (0.91, 1.03) | 1.04 (0.97, 1.11) | 1.04 (0.98, 1.11) |
| P-trend | P = 0.5039 | P = 0.1206 | P = 0.0831 | ||
| Dietary total carbohydrate, g | |||||
| Quintile 1 | 115.0 | 4677 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 158.5 | 4734 | 0.96 (0.92, 1.01) | 1.00 (0.95, 1.06) | 1.01 (0.96, 1.06) |
| Quintile 3 | 193.7 | 4736 | 0.92 (0.88, 0.97) | 0.99 (0.94, 1.05) | 1.00 (0.95, 1.06) |
| Quintile 4 | 233.6 | 4882 | 0.92 (0.88, 0.97) | 1.00 (0.95, 1.06) | 1.02 (0.96, 1.08) |
| Quintile 5 | 303.6 | 4993 | 0.89 (0.84, 0.94) | 0.97 (0.92, 1.03) | 0.99 (0.93, 1.05) |
| P-trend | P < 0.0001 | P = 0.4432 | P = 0.8248 | ||
| Fiber, g | |||||
| Quintile 1 | 10.6 | 5404 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 13.7 | 5045 | 0.91 (0.87, 0.96) | 0.96 (0.92, 1.01) | 0.97 (0.92, 1.02) |
| Quintile 3 | 15.9 | 4754 | 0.85 (0.81, 0.89) | 0.92 (0.88, 0.97) | 0.93 (0.88, 0.98) |
| Quintile 4 | 18.2 | 4575 | 0.81 (0.77, 0.85) | 0.90 (0.86, 0.95) | 0.92 (0.87, 0.97) |
| Quintile 5 | 21.1 | 4244 | 0.74 (0.71, 0.78) | 0.84 (0.80, 0.89) | 0.87 (0.82, 0.92) |
| P-trend | P < 0.0001 | P < 0.0001 | P < 0.0001 | ||
| Whole grains, oz | |||||
| Quintile 1 | 0.24 | 5095 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 0.68 | 4945 | 0.95 (0.90, 0.99) | 0.98 (0.93, 1.03) | 0.99 (0.94, 1.04) |
| Quintile 3 | 1.06 | 4811 | 0.91 (0.87, 0.96) | 0.96 (0.92, 1.01) | 0.98 (0.93, 1.03) |
| Quintile 4 | 1.47 | 4633 | 0.87 (0.83, 0.92) | 0.93 (0.88, 0.98) | 0.95 (0.90, 1.00) |
| Quintile 5 | 2.29 | 4538 | 0.85 (0.81, 0.89) | 0.91 (0.86, 0.95) | 0.93 (0.88, 0.98) |
| P-trend | P < 0.0001 | P < 0.0001 | P = 0.0010 | ||
| Nonwhole/refined grains, oz | |||||
| Quintile 1 | 1.73 | 4572 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 2.64 | 4762 | 1.04 (0.99, 1.10) | 1.06 (1.00, 1.11) | 1.06 (1.01, 1.12) |
| Quintile 3 | 3.29 | 4831 | 1.06 (1.01, 1.11) | 1.08 (1.03, 1.14) | 1.08 (1.03, 1.13) |
| Quintile 4 | 3.97 | 4801 | 1.04 (1.00, 1.10) | 1.07 (1.02, 1.13) | 1.07 (1.01, 1.12) |
| Quintile 5 | 5.34 | 5056 | 1.12 (1.07, 1.18) | 1.17 (1.11, 1.23) | 1.16 (1.11, 1.23) |
| P-trend | P < 0.0001 | P < 0.0001 | P < 0.0001 | ||
| Nonjuice fruit, cups | |||||
| Quintile 1 | 0.54 | 5470 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 1.06 | 4968 | 0.88 (0.84, 0.92) | 0.92 (0.88, 0.97) | 0.93 (0.89, 0.98) |
| Quintile 3 | 1.59 | 4747 | 0.83 (0.79, 0.87) | 0.91 (0.86, 0.95) | 0.92 (0.87, 0.97) |
| Quintile 4 | 2.07 | 4521 | 0.79 (0.75, 0.83) | 0.87 (0.83, 0.92) | 0.89 (0.84, 0.93) |
| Quintile 5 | 2.74 | 4316 | 0.74 (0.70, 0.78) | 0.83 (0.79, 0.87) | 0.85 (0.80, 0.89) |
| P-trend | P < 0.0001 | P < 0.0001 | P < 0.0001 | ||
| Vegetables, cups | |||||
| Quintile 1 | 0.81 | 5370 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 1.20 | 4911 | 0.88 (0.84, 0.92) | 0.93 (0.88, 0.97) | 0.94 (0.90, 0.99) |
| Quintile 3 | 1.48 | 4803 | 0.82 (0.78, 0.86) | 0.89 (0.85, 0.94) | 0.91 (0.86, 0.95) |
| Quintile 4 | 1.76 | 4516 | 0.79 (0.76, 0.83) | 0.88 (0.84, 0.93) | 0.90 (0.86, 0.95) |
| Quintile 5 | 2.22 | 4422 | 0.76 (0.72, 0.80) | 0.85 (0.81, 0.90) | 0.88 (0.83, 0.93) |
| P-trend | P < 0.0001 | P < 0.0001 | P < 0.0001 | ||
| Dairy products, cups | |||||
| Quintile 1 | 0.49 | 4901 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 1.03 | 4845 | 0.96 (0.92, 1.01) | 0.96 (0.92, 1.01) | 0.98 (0.93, 1.03) |
| Quintile 3 | 1.46 | 4804 | 0.95 (0.90, 1.00) | 0.97 (0.92, 1.01) | 0.99 (0.94, 1.04) |
| Quintile 4 | 1.97 | 4812 | 0.95 (0.91, 1.00) | 0.97 (0.92, 1.02) | 0.99 (0.94, 1.04) |
| Quintile 5 | 3.17 | 4660 | 0.91 (0.86, 0.95) | 0.94 (0.89, 0.99) | 0.96 (0.92, 1.02) |
| P-trend | P = 0.0001 | P = 0.0432 | P = 0.3147 | ||
1Model 1 adjusted for energy. Glycemic load was based on available carbohydrate. Model 2 adjusted for the variables in Model 1 plus age, race/ethnicity, education, income, smoking, alcohol, caffeine, stressful life events, social support, overactive thyroid, bodily pain, hormone replacement therapy, and snoring. Model 3 adjusted for the variables in Model 2 plus depression, physical activity, BMI, diabetes, hypertension, myocardial infarction, cardiovascular disease, asthma, and hot flashes. 1 oz = 28.3495 g; 1 cup = 236.588 mL. GI, glycemic index; oz, ounces.
The results from the 3-y longitudinal multivariable analyses (n = 53,069) are shown in Table 4. Subjects with higher dietary GI and glycemic load and higher consumption of added sugars, starch, and nonwhole/refined grains were more likely to have insomnia after 3 y in energy-adjusted analyses (Model 1). Although the results were slightly attenuated by the variables in Models 2 and 3, women in the higher dietary GI, added sugars, and starch quintiles remained significantly more likely to have insomnia after 3 y with significant tests for linear trend across all models. There was a significant linear trend toward higher insomnia incidence for higher glycemic load. Higher consumption of fiber, whole grains, nonjuice fruit, and vegetables were significantly associated with decreased odds of insomnia incidence in Model 1, but these associations were attenuated with the inclusion of the covariates in Models 2 and 3, with only the fifth quintiles for nonfruit juice and vegetables remaining significant in fully adjusted models. There was a significant linear trend toward lower insomnia incidence for increasing nonjuice fruit consumption in all 3 models. There were no associations of dietary total sugar, total carbohydrate, whole grains, or dairy products with insomnia incidence in fully adjusted models. Effect modification by physical activity was not found in stratified analyses.
TABLE 4.
Adjusted ORs and 95% CIs from logistic regression analyses of insomnia incidence according to quintiles of energy-adjusted GI, glycemic load, carbohydrate consumption, fiber, whole grains, nonwhole/refined grains, nonjuice fruit, vegetables, and dairy products1
| Incident insomnia after 3 y, n | Incident insomnia 3 y later, OR (95% CI) | ||||
|---|---|---|---|---|---|
| Median | Model 1 | Model 2 | Model 3 | ||
| Dietary GI | |||||
| Quintile 1 | 47.0 | 1728 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 49.8 | 1924 | 1.13 (1.05, 1.21) | 1.13 (1.05, 1.21) | 1.13 (1.05, 1.21) |
| Quintile 3 | 51.7 | 1943 | 1.14 (1.06, 1.22) | 1.12 (1.04, 1.20) | 1.11 (1.03, 1.19) |
| Quintile 4 | 53.5 | 2065 | 1.22 (1.14, 1.31) | 1.19 (1.11, 1.28) | 1.18 (1.09, 1.27) |
| Quintile 5 | 56.3 | 2123 | 1.26 (1.18, 1.36) | 1.18 (1.10, 1.28) | 1.16 (1.08, 1.25) |
| P-trend | P < 0.0001 | P < 0.0001 | P < 0.0001 | ||
| Dietary glycemic load, g/d | |||||
| Quintile 1 | 53.6 | 1809 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 74.5 | 1895 | 1.04 (0.97, 1.12) | 1.07 (1.00, 1.16) | 1.08 (1.00, 1.16) |
| Quintile 3 | 91.7 | 1950 | 1.06 (0.98, 1.14) | 1.10 (1.02, 1.19) | 1.10 (1.02, 1.19) |
| Quintile 4 | 111.3 | 2065 | 1.11 (1.03, 1.20) | 1.14 (1.05, 1.24) | 1.14 (1.05, 1.24) |
| Quintile 5 | 146.5 | 2064 | 1.07 (0.98, 1.16) | 1.10 (1.01, 1.19) | 1.09 (1.00, 1.19) |
| P-trend | P = 0.0365 | P = 0.0148 | P = 0.0223 | ||
| Dietary added sugar, g | |||||
| Quintile 1 | 17.7 | 1765 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 28.6 | 1837 | 1.04 (0.97, 1.12) | 1.03 (0.95, 1.11) | 1.03 (0.95, 1.11) |
| Quintile 3 | 39.1 | 1988 | 1.14 (1.05, 1.22) | 1.13 (1.05, 1.22) | 1.12 (1.04, 1.22) |
| Quintile 4 | 52.5 | 2050 | 1.17 (1.09, 1.27) | 1.15 (1.06, 1.24) | 1.13 (1.04, 1.22) |
| Quintile 5 | 79.2 | 2143 | 1.23 (1.14, 1.33) | 1.16 (1.07, 1.26) | 1.13 (1.04, 1.24) |
| P-trend | P < 0.0001 | P < 0.0001 | P = 0.0007 | ||
| Dietary total sugars, g | |||||
| Quintile 1 | 50.5 | 1882 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 73.6 | 1948 | 1.01 (0.94, 1.09) | 1.04 (0.96, 1.12) | 1.05 (0.97, 1.13) |
| Quintile 3 | 93.3 | 1959 | 0.99 (0.92, 1.07) | 1.03 (0.96, 1.11) | 1.03 (0.96, 1.12) |
| Quintile 4 | 115.6 | 1984 | 0.99 (0.92, 1.06) | 1.03 (0.95, 1.11) | 1.03 (0.95, 1.11) |
| Quintile 5 | 155.4 | 2010 | 0.97 (0.90, 1.05) | 1.00 (0.92, 1.08) | 0.99 (0.91, 1.08) |
| P-trend | P = 0.3203 | P = 0.8450 | P = 0.7027 | ||
| Dietary starch, g | |||||
| Quintile 1 | 38.1 | 1807 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 55.6 | 1896 | 1.04 (0.96, 1.12) | 1.07 (0.99, 1.16) | 1.08 (1.00, 1.17) |
| Quintile 3 | 70.3 | 1944 | 1.05 (0.97, 1.14) | 1.09 (1.00, 1.18) | 1.09 (1.01, 1.18) |
| Quintile 4 | 87.8 | 2039 | 1.09 (1.01, 1.18) | 1.13 (1.03, 1.23) | 1.13 (1.04, 1.23) |
| Quintile 5 | 119.3 | 2097 | 1.11 (1.01, 1.21) | 1.14 (1.04, 1.25) | 1.15 (1.05, 1.27) |
| P-trend | P = 0.0167 | P = 0.0064 | P = 0.0043 | ||
| Dietary total carbohydrate, g | |||||
| Quintile 1 | 114.7 | 1839 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 157.9 | 1946 | 1.05 (0.97, 1.12) | 1.08 (1.00, 1.16) | 1.08 (1.00, 1.16) |
| Quintile 3 | 192.9 | 1928 | 1.01 (0.94, 1.09) | 1.06 (0.98, 1.14) | 1.06 (0.98, 1.14) |
| Quintile 4 | 232.8 | 2036 | 1.06 (0.98, 1.14) | 1.10 (1.02, 1.19) | 1.10 (1.02, 1.19) |
| Quintile 5 | 302.7 | 2034 | 1.01 (0.93, 1.09) | 1.05 (0.97, 1.15) | 1.05 (0.96, 1.14) |
| P-trend | P = 0.7533 | P = 0.2319 | P = 0.2724 | ||
| Fiber, g | |||||
| Quintile 1 | 10.6 | 2141 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 13.7 | 2052 | 0.96 (0.90, 1.03) | 1.00 (0.93, 1.07) | 1.01 (0.94, 1.09) |
| Quintile 3 | 16.0 | 1921 | 0.89 (0.83, 0.96) | 0.96 (0.90, 1.03) | 0.98 (0.91, 1.05) |
| Quintile 4 | 18.2 | 1899 | 0.89 (0.83, 0.95) | 0.98 (0.91, 1.05) | 1.00 (0.92, 1.07) |
| Quintile 5 | 21.0 | 1770 | 0.83 (0.77, 0.89) | 0.91 (0.85, 0.99) | 0.94 (0.87, 1.01) |
| P-trend | P < 0.0001 | P = 0.0221 | P = 0.1119 | ||
| Whole grains, oz | |||||
| Quintile 1 | 0.25 | 2020 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 0.69 | 2049 | 1.01 (0.94, 1.08) | 1.03 (0.96, 1.11) | 1.05 (0.97, 1.12) |
| Quintile 3 | 1.07 | 1951 | 0.95 (0.89, 1.02) | 0.98 (0.91, 1.05) | 1.00 (0.93, 1.08) |
| Quintile 4 | 1.47 | 1868 | 0.91 (0.85, 0.97) | 0.96 (0.89, 1.03) | 0.97 (0.91, 1.05) |
| Quintile 5 | 2.30 | 1895 | 0.93 (0.87, 0.99) | 0.97 (0.91, 1.05) | 1.00 (0.93, 1.07) |
| P-trend | P = 0.0014 | P = 0.1146 | P = 0.3442 | ||
| Nonwhole/refined grains, oz | |||||
| Quintile 1 | 1.70 | 1820 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 2.62 | 1944 | 1.07 (1.00, 1.15) | 1.08 (1.01, 1.16) | 1.09 (1.01, 1.17) |
| Quintile 3 | 3.25 | 1960 | 1.07 (1.00, 1.15) | 1.09 (1.01, 1.17) | 1.09 (1.02, 1.18) |
| Quintile 4 | 3.93 | 1976 | 1.08 (1.01, 1.16) | 1.08 (1.00, 1.16) | 1.08 (1.01, 1.17) |
| Quintile 5 | 5.26 | 2083 | 1.15 (1.08, 1.24) | 1.17 (1.08, 1.25) | 1.16 (1.08, 1.25) |
| P-trend | P = 0.0002 | P = 0.0003 | P = 0.0007 | ||
| Nonjuice fruit, cups | |||||
| Quintile 1 | 0.56 | 2180 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 1.09 | 2034 | 0.93 (0.87, 0.99) | 0.97 (0.91, 1.04) | 0.98 (0.92, 1.06) |
| Quintile 3 | 1.61 | 1905 | 0.86 (0.80, 0.92) | 0.94 (0.87, 1.00) | 0.95 (0.88, 1.02) |
| Quintile 4 | 2.09 | 1922 | 0.88 (0.82, 0.94) | 0.97 (0.90, 1.04) | 0.98 (0.91, 1.06) |
| Quintile 5 | 2.75 | 1742 | 0.80 (0.74, 0.86) | 0.88 (0.82, 0.95) | 0.90 (0.83, 0.97) |
| P-trend | P < 0.0001 | P = 0.0049 | P = 0.0199 | ||
| Vegetables, cups | |||||
| Quintile 1 | 0.82 | 2111 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 1.20 | 1983 | 0.93 (0.87, 1.00) | 0.98 (0.91, 1.05) | 0.99 (0.92, 1.06) |
| Quintile 3 | 1.48 | 2032 | 0.97 (0.91, 1.04) | 1.03 (0.96, 1.11) | 1.05 (0.98, 1.13) |
| Quintile 4 | 1.77 | 1911 | 0.91 (0.85, 0.98) | 0.98 (0.91, 1.05) | 0.99 (0.92, 1.07) |
| Quintile 5 | 2.21 | 1746 | 0.83 (0.78, 0.90) | 0.90 (0.83, 0.97) | 0.91 (0.84, 0.99) |
| P-trend | P < 0.0001 | P = 0.0127 | P = 0.0556 | ||
| Dairy products, cups | |||||
| Quintile 1 | 0.48 | 1982 | 1.00 | 1.00 | 1.00 |
| Quintile 2 | 1.02 | 2012 | 1.00 (0.94, 1.07) | 1.01 (0.94, 1.09) | 1.02 (0.95, 1.10) |
| Quintile 3 | 1.45 | 1951 | 0.96 (0.90, 1.03) | 0.98 (0.91, 1.06) | 1.00 (0.93, 1.08) |
| Quintile 4 | 1.97 | 1951 | 0.96 (0.89, 1.03) | 0.97 (0.90, 1.04) | 0.99 (0.92, 1.07) |
| Quintile 5 | 3.16 | 1887 | 0.92 (0.86, 0.99) | 0.96 (0.89, 1.03) | 0.98 (0.91, 1.06) |
| P-trend | P = 0.0090 | P = 0.1070 | P = 0.3866 | ||
1Model 1 adjusted for energy. Glycemic load was based on available carbohydrate. Model 2 adjusted for the variables in Model 1 plus age, race/ethnicity, education, income, smoking, alcohol, caffeine, stressful life events, social support, overactive thyroid, bodily pain, hormone replacement therapy, and snoring. Model 3 adjusted for the variables in Model 2 plus depression, physical activity, BMI, diabetes, hypertension, myocardial infarction, cardiovascular disease, asthma, and hot flashes. 1 oz = 28.3495 g; 1 cup = 236.588 mL. GI, glycemic index; oz, ounces.
Discussion
This large prospective study among postmenopausal women demonstrated that progressively higher dietary GI was associated with increased insomnia incidence over 3 y after adjusting for demographic, behavioral, lifestyle, psychosocial, and medical factors. Added sugars, but not total sugars or total carbohydrates, were associated with insomnia incidence. This may be partly explained by the observation that added sugars, caloric sweeteners not naturally found in foods, typically have higher GIs, whereas total dietary sugars comprise an amalgam of various types of sugar and sugar from different food sources. The food source of a sugar influences the GI, with higher fiber content of a food slowing the metabolism of carbohydrates and lowering the GI (20). Increased consumption of nonjuice fruit was associated with a lower prevalence and incidence of insomnia. The GIs of fruits vary, but of the 7 most commonly consumed fruits in the United States, 4 have low GIs (≤55 on the glucose reference scale) (apples, strawberries, oranges, peaches), 2 have intermediate GIs (56–69) (ripe bananas and grapes), and only 1 has a high GI (≥70) (watermelon) (21, 22). A relatively low GI could be one of the characteristics of fruit that contribute to the association with a lower prevalence and incidence of insomnia.
Cross-sectional analyses did not demonstrate a relation between the consumption of starch and the prevalence of insomnia, but prospective analyses showed that individuals with higher starch consumption were progressively and significantly more likely to have developed insomnia incidence after 3 y. Although starches are complex carbohydrates, sources such as refined white bread and boiled potatoes have high GIs, whereas other fiber-rich sources including legumes, nuts/seeds, and yams generally have low GIs. When we examined different types of starches according to fiber content, we found that progressively higher consumption of whole grains was associated with lower insomnia prevalence, whereas the opposite was true for nonwhole/refined grains, with progressively higher consumption associated with higher insomnia prevalence and incidence. Increased vegetable consumption was also associated with decreased insomnia prevalence and incidence. The most commonly consumed vegetable by far in the United States is potatoes (21), with most varieties and methods of cooking resulting in high GIs (23), whereas the next 6 most commonly consumed vegetables (onions, tomatoes, head lettuce, romaine and leaf lettuce, bell peppers, and cucumbers) all have low GIs (21).
Our results could be viewed as surprising because carbohydrate intake has been shown to affect concentrations of tryptophan, a precursor for both serotonin and melatonin. Tryptophan competes with larger amino acids for the same transport system to cross the blood–brain barrier (24). Carbohydrate intake promotes insulin release, which stimulates the uptake of the competing larger amino acids into muscle tissue, increasing the ratio of tryptophan to the other amino acids in plasma, allowing tryptophan access to the transport system to cross the blood–brain barrier and contribute to serotonin synthesis (25). However, for this process to occur, the meal must contain only carbohydrate and be consumed without any protein remaining in the stomach. If the meal contains as little as 5% protein, then tryptophan concentrations will not increase; if the meal contains as little as 2.5% protein, the increase in tryptophan will be blunted (26). Sweet foods such as milk chocolate, sweetened yogurts, ice cream, and egg-based cakes and pastries contain enough protein to block any increase in tryptophan. Even foods such as potatoes, bread, and rice can have enough protein to blunt or prevent increases in tryptophan concentrations. Serotonin is a precursor for melatonin, but even if serotonin concentrations increase, any resultant effect upon melatonin is dependent upon the presence of darkness because sunlight and artificial light inhibit melatonin synthesis (27).
A plausible mechanism by which a high-GI diet may increase the risk of insomnia is through acute spikes and troughs in blood glucose. GI and glycemic load have been shown to provide physiologically valid estimates of postprandial glycemia and insulin demand in healthy individuals (28). Postprandial hyperglycemia from high dietary glycemic load and resultant compensatory hyperinsulinemia can lower plasma glucose to concentrations that compromise brain glucose, ∼70 mg/dL (3.8 mmol/L) (29), triggering secretion of autonomic counter-regulatory hormones such as adrenaline, cortisol, glucagon, and growth hormone (30). Symptoms of counter-regulatory hormone responses can include heart palpitations, tremor, cold sweats, paresthesia, anxiety, irritability, and hunger (31). Hypoglycemia has been shown to produce arousal from sleep and substantially reduce sleep efficiency in nondiabetic adults (32, 33). High blood sugar from carbohydrate consumption can initially make one drowsy, helping one to fall asleep (4), but the compensatory hyperinsulinemia and counter-regulatory hormone responses can awaken one from sleep (32, 33). Higher-GI diets have also been shown to stimulate inflammatory immune responses (34), which could function to increase the risk of insomnia through anti-inflammatory cytokines that inhibit sleep (35). Added sugars could also negatively affect sleep quality by compromising the intestinal microbiome. Higher consumption of added sugars can contribute to intestinal dysbiosis, a maladaptive microbiota imbalance that can profoundly affect multiple aspects of sleep (36).
Possible limitations of our study include the measurement of dietary exposures from FFQs instead of dietary biomarkers or food records and the assessment of our outcome of insomnia from self-reported symptoms as opposed to objective clinical diagnosis. The exact nutrient amounts for each food were not analytically measured, so some of the nutrient values were estimated or imputed rather than being exact analytic values from a laboratory assay. For example, 26–50% of the values for the variable “dietary added sugars” are estimated or imputed. Estimates were generally based on a similar food, another form of the same food, a known nutrient value associated with the missing value, or recipes or formulations from manufacturers. Although we would expect any misclassification of exposure or outcome to be random, resulting in nondifferential misclassification which typically leads to bias toward the null hypothesis (37), we cannot rule out the possibility that bias, particularly food recall bias, could be systematic and related to variables such as BMI, age, or ethnicity. Sleep deprivation from insomnia could also induce carbohydrate cravings, so reverse causation could have contributed to our results in the cross-sectional analyses (38). There is also a potential for residual confounding from unmeasured confounders and the possibility of false positives with multiple statistical tests. Because the variables included in Model 3 are theorized to be mediators of the relation between the dietary variables and insomnia, any resultant attenuation from their inclusion does not necessarily imply confounding, but could be consistent with some of these variables lying along the causal pathway. The participants’ eating habits may not be representative of those common now, almost 20 y later. Finally, our study sample was confined to postmenopausal women, limiting the generalizability of our findings to other populations.
The results from this study suggest that a high-GI diet could be a risk factor for insomnia in postmenopausal women, whereas dietary fiber, nonjuice fruit, and vegetables reduce its risk. If high-GI diets increase the risk of insomnia, then dietary interventions that promote the consumption of whole unprocessed carbohydrates that are high in fiber and have low GIs could serve as potential treatments of, and primary preventive measures for, insomnia in postmenopausal women. Randomized controlled trials examining dietary patterns in relation to insomnia are needed to clarify these findings.
Supplementary Material
ACKNOWLEDGEMENTS
We acknowledge the following WHI investigators: Program Office (National Heart, Lung, and Blood Institute, Bethesda, MD): Jacques Rossouw, Shari Ludlam, Dale Burwen, Joan McGowan, Leslie Ford, and Nancy Geller; Clinical Coordinating Center (Fred Hutchinson Cancer Research Center, Seattle, WA): Garnet Anderson, Ross Prentice, Andrea LaCroix, and Charles Kooperberg; Investigators and Academic Centers: JoAnn E Manson (Brigham and Women's Hospital, Harvard Medical School, Boston, MA); Barbara V Howard (MedStar Health Research Institute/Howard University, Washington, DC); Marcia L Stefanick (Stanford Prevention Research Center, Stanford, CA); Rebecca Jackson (The Ohio State University, Columbus, OH); Cynthia A Thomson (University of Arizona, Tucson/Phoenix, AZ); Jean Wactawski-Wende (University at Buffalo, Buffalo, NY); Marian Limacher (University of Florida, Gainesville/Jacksonville, FL); Robert Wallace (University of Iowa, Iowa City/Davenport, IA); Lewis Kuller (University of Pittsburgh, Pittsburgh, PA); and Sally Shumaker (Wake Forest University School of Medicine, Winston-Salem, NC).
The authors’ responsibilities were as follows—JEG and LH: designed the research; JEG: analyzed the data and had primary responsibility for the final content; and all authors: wrote the paper and read and approved the final manuscript. All authors report no conflicts of interest.
Notes
Supported by NIH National Heart, Lung, and Blood Institute (NHLBI) grant R01HL128226 (to M-PS-O). The WHI program is funded by the NHLBI, NIH, US Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C.
Supplemental Figure 1 is available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Data described in the article, code book, and analytic code will be made available upon request pending application and approval.
Abbreviations used: GI, glycemic index; MET, metabolic equivalent; WHI, Women's Health Initiative; WHIIRS, Women's Health Initiative Insomnia Rating Scale.
References
- 1. Leger D, Bayon V. Societal costs of insomnia. Sleep Med Rev. 2010;14(6):379–89. [DOI] [PubMed] [Google Scholar]
- 2. Reynolds SA, Ebben MR. The cost of insomnia and the benefit of increased access to evidence-based treatment: cognitive behavioral therapy for insomnia. Sleep Med Clin. 2017;12:39–46. [DOI] [PubMed] [Google Scholar]
- 3. Glass J, Lanctot KL, Herrmann N, Sproule BA, Busto UE. Sedative hypnotics in older people with insomnia: meta-analysis of risks and benefits. BMJ. 2005;331(7526):1169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. St-Onge MP, Mikic A, Pietrolungo CE. Effects of diet on sleep quality. Adv Nutr. 2016;7;938–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Tanaka E, Yatsuya H, Uemura M, Murata C, Otsuka R, Toyoshima H, Tamakoshi K, Sasaki S, Kawaguchi I, Aoyama A. Associations of protein, fat, and carbohydrate intakes with insomnia symptoms among middle-aged Japanese workers. J Epidemiol. 2013;23:132–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Tan X, Alen M, Cheng SM, Mikkola TM, Enthunen J, Lyytikainen A, Wiklund P, Cong F, Saarinen A, Tarkka L et al.. Associations of disordered sleep with body fat distribution, physical activity and diet among overweight middle-aged men. J Sleep Res. 2015;24:414–24. [DOI] [PubMed] [Google Scholar]
- 7. Katagiri R, Asakura K, Kobayashi S, Suga H, Sasaki S. Low intake of vegetables, high intake of confectionary, and unhealthy eating habits are associated with poor sleep quality among middle-aged female Japanese workers. J Occup Health. 2014;56(5):359–68. [DOI] [PubMed] [Google Scholar]
- 8. Afaghi A, O'Connor H, Chow CM. High glycemic-index carbohydrate meals shorten sleep onset. Am J Clin Nutr. 2007;85:426–30. [DOI] [PubMed] [Google Scholar]
- 9. Hays J, Hunt JR, Hubbell FA, Anderson GL, Limacher M, Allen C, Rossouw JE. The Women's Health Initiative recruitment methods and results. Ann Epidemiol. 2003;13:S18–77. [DOI] [PubMed] [Google Scholar]
- 10. Patterson RE, Kristal AR, Tinker LF, Carter RA, Bolton MP, Agurs-Collins T. Measurement characteristics of the Women's Health Initiative food frequency questionnaire. Ann Epidemiol. 1999;9:178–87. [DOI] [PubMed] [Google Scholar]
- 11. Neuhouser ML, Tinker LF, Thomson C, Caan B, Van Horn L, Snetselaar L, Parker LM, Patterson RE, Robinson-O'Brien R, Beresford SAA et al.. Development of a glycemic index database for food frequency questionnaires used in epidemiologic studies. J Nutr. 2006;136:1604–9. [DOI] [PubMed] [Google Scholar]
- 12. Wolever TM, Jenkins DJ, Jenkins AL, Josse RG. The glycemic index: methodology and clinical implications. Am J Clin Nutr. 1991;54:846–54. [DOI] [PubMed] [Google Scholar]
- 13. Bowman SA, Friday JE, Moshfegh A. MyPyramid Equivalents Database, 2.0 for USDA survey foods, 2003–2004. [Internet] Beltsville, MD: Food Surveys Research Group, Beltsville Human Nutrition Research Center, Agricultural Research Service, USDA; 2008; [date cited: November 8, 2019]. Available from: http://www.ars.usda.gov/ba/bhnrc/fsrg. [Google Scholar]
- 14. Powers PM, Hoover LW. Calculating the nutrient composition of recipes with computers. J Am Diet Assoc. 1989;89(2):224–32. [PubMed] [Google Scholar]
- 15. Levine DW, Dailey ME, Rockhill B, Tipping D, Naughton MJ, Shumaker SA. Validation of the Women's Health Initiative Insomnia Rating Scale in a multicenter controlled clinical trial. Psychosom Med. 2005;67(1):98–104. [DOI] [PubMed] [Google Scholar]
- 16. Levine DW, Kaplan RM, Kripke DF, Bowen DJ, Naughton MJ, Shumaker SA. Factor structure and measurement invariance of the Women's Health Initiative Insomnia Rating Scale. Psychol Assess. 2003;15(2):123–36. [DOI] [PubMed] [Google Scholar]
- 17. Levine DW, Kripke DF, Kaplan RM, Lewis MA, Naughton MJ, Bowen DJ, Shumaker SA. Reliability and validity of the Women's Health Initiative Insomnia Rating Scale. Psychol Assess. 2003;15(2):137–48. [DOI] [PubMed] [Google Scholar]
- 18. Howe GR, Miller AB, Jain M. Re: “Total energy intake: implications for epidemiologic analyses.”. Am J Epidemiol. 1986;124:157–9. [DOI] [PubMed] [Google Scholar]
- 19. Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol. 1986;124:17–27. [DOI] [PubMed] [Google Scholar]
- 20. Jenkins DJ, Jenkins AL. Dietary fiber and the glycemic response. Proc Soc Exp Biol Med. 1985;180(3):422–31. [DOI] [PubMed] [Google Scholar]
- 21. USDA, Economic Research Service. Food availability per capita data system. [Internet]. Updated July 31, 2019. Available from: http://www.ers.usda.gov/date-products/food-availability-per-capita-data-system/.
- 22. Atkinson FS, Foster-Powell R, Brand-Miller JC. International tables of glycemic index and glycemic load values: 2008. Diabetes Care. 2008;31:2281–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Fernandes G, Velangi A, Wolever TM. Glycemic index of potatoes commonly consumed in North America. J Am Diet Assoc. 2005;105:557–62. [DOI] [PubMed] [Google Scholar]
- 24. Oldendorf LA, Szabo J. Amino acid assignment to one of three blood-brain barrier amino acid carriers. Am J Physiol. 1976;230:94–8. [DOI] [PubMed] [Google Scholar]
- 25. Wurtman RJ, Hefti F, Melamed E. Precursor control of neurotransmitter synthesis. Pharmacol Rev. 1980;32:315–35. [PubMed] [Google Scholar]
- 26. Yokogoshi H, Wurtman RJ. Meal composition and plasma amino acid ratio: effect of various proteins or carbohydrates and of various protein concentrations. Metabolism. 1986;35:837–42. [DOI] [PubMed] [Google Scholar]
- 27. Claustrat B, Brun J, Chazot G. The basic physiology and pathophysiology of melatonin. Sleep Med Rev. 2005;9:11–24. [DOI] [PubMed] [Google Scholar]
- 28. Bao J, Atkinson F, Petocz P, Willett WC, Brand-Miller JC. Prediction of postprandial glycemia and insulinemia in lean, young, healthy adults: glycemic load compared with carbohydrate content alone. Am J Clin Nutr. 2011;93:984–96. [DOI] [PubMed] [Google Scholar]
- 29. Seaquist ER, Anderson J, Childs B, Cryer P, Dagogo-Jack S, Fish L, Heller SR, Rodriguez H, Rosenzweig J, Vigersky R. Hypoglycemia and diabetes: a report of a workgroup of the American Diabetes Association and The Endocrine Society. Diabetes Care. 2013;36:1384–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Ludwig DS. The glycemic index: physiological mechanisms relating to obesity, diabetes, and cardiovascular disease. JAMA. 2002;287:2414–23. [DOI] [PubMed] [Google Scholar]
- 31. Alsahli M, Gerich JE. Hypoglycemia. Endocrin Metab Clin. 2013;42:657–76. [DOI] [PubMed] [Google Scholar]
- 32. Banarer S, Cryer PE. Sleep-related hypoglycemia-associated autonomic failure in type I diabetes: reduced awakening from sleep during hypoglycemia. Diabetes. 2003;52:1195–203. [DOI] [PubMed] [Google Scholar]
- 33. Gais S, Born J, Peters A, Schultes B, Heindl B, Fehm HL, Kern W. Hypoglycemia counterregulation during sleep. Sleep. 2003;26:55–9. [DOI] [PubMed] [Google Scholar]
- 34. Kim Y, Chen J, Wirth MD, Shivappa N, Hebert JR. Lower dietary inflammatory index scores are associated with lower glycemic index scores among college students. Nutrients. 2018;10(2):182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Krueger JM, Majide JA. Humoral links between sleep and the immune system. Ann N Y Acad Sci. 2003;992:9–20. [DOI] [PubMed] [Google Scholar]
- 36. Opp MR, Toth LA. Neural-immune interactions in the regulation of sleep. Front Biosci. 2003;8:d768–79. [DOI] [PubMed] [Google Scholar]
- 37. Rothman KJ, Greenland S. Modern epidemiology. 2nd ed Philadelphia, PA: Lippincott Williams & Wilkins; 1998. [Google Scholar]
- 38. Yang C, Schnepp J, Tucker RM. Increased hunger, food cravings, food reward, and portion size selection after sleep curtailment in women without obesity. Nutrients. 2019;11(3):663–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.