Relationships among Alcohol Drinking Patterns, Macronutrient Composition, and Caloric Intake: National Health and Nutrition Examination Survey 2017–2018

Abstract

Background

Excessive alcohol consumption is associated with poor diet. Mixed reports in literature, so far, emphasize on the detailed understanding of relationships between diet composition and binge drinking at different drinking thresholds.

Objective

We examined the association of alcohol consumption thresholds with macronutrient composition, caloric intake and anthropometric measures from the NHANES 2017–2018 dataset.

Methods

A total of 2320 participants’ data were analyzed. Energy and nutrient content from daily food and beverage intake were assessed via two dietary recall interviews. Physical examination and Alcohol Use Questionnaire including details about lifetime and current usage patterns were obtained. Correlations were evaluated using the Rao-Scott F Adjusted Chi-square statistic and Wald F-test. Sample-weighted multiple linear regression models were built to analyze the associations among volume of alcohol consumed, weight history and macronutrient intake.

Results

Waist circumference was significantly higher in 0– < 4 drinks/episode (low-quantity) drinkers than 4–7 drinks/episode (medium-quantity) and 8–11 drinks/episode (high-quantity) drinkers. High-quantity drinkers consumed significantly more kilocalories (2569.91) compared with low-quantity drinkers (2106.73). Low-quantity drinkers consumed more energy from carbohydrate and fat than medium and high-quantity drinkers. Very high-quantity drinkers (12+ drinks/episode) consumed less fiber (12.81 g) than low-quantity drinkers (16.67 g).

Conclusions

We observed an association between high alcohol intake and differences in eating habits and body composition. The findings suggest a need to compare more specific drinking patterns and their impact on nutrient intake. Although some results conflicted with previous studies, the mechanisms underlying alcohol’s effect on ingestive and digestive metabolic pathways are still unclear and require further investigation.

INTRODUCTION

Excessive alcohol consumption is associated with poor diet (Addolorato et al., 2000) (Colditz et al., 1991). Binge drinking, consumption of five alcoholic drinks for men and four alcoholic drinks for women, is common in the United States (Kanny et al., 2020). Binge drinking constitutes raising a person’s blood alcohol level to at least 0.08% within 2 h. Binge drinking can contribute to cancer and cardiovascular disease (Xi et al., 2017; Poole et al., 2019). However, the influence of alcohol on dietary patterns is unclear.

Alcohol (ethanol) (29 kJ/g) and fat (38 kJ/g) are more energy-dense than carbohydrates and protein (17 kJ/g) (Brandhagen et al., 2012). Moderate alcohol intake levels may promote increased intake of high-fat, energy-dense foods, leading to increased adiposity. For instance, some studies have reported positive associations between alcohol drinking and waist circumference (WC) (Dallongeville et al., 1998; Tolstrup et al., 2005). On the other hand, energy intake from higher chronic alcohol consumption may disrupt proper macronutrient intake ratios, by reducing healthy carbohydrate and protein intake (Westerterp et al., 2004). This effect may be seen at extremely high alcohol intake levels. For example, Addolorato et al. reported that chronic drinkers exhibited reduced body weight and suffered from weight loss, temporal fat loss, peculiar body composition and malnourishment (Addolorato et al., 1997; Addolorato et al., 2000). Reduced body fat has been linked to loss of interest in food (French et al., 2010) or decreased absorption and metabolism of nutrients (Lieber, 2000). These differing trends may result from alcohol consumption volumes.

The mixed reports in the literature, to date, require more detailed understanding of the relationships between diet composition and binge drinking at different thresholds (i.e. number of drinks). Previous studies have used less specific thresholds when categorizing alcohol consumption (e.g. classification of binge versus not binge drinking). Thus, the dose effects of alcohol consumption and diet composition have not been well elucidated. In this analysis, we investigated the association of different alcohol consumption thresholds with macronutrient composition, caloric intake and anthropometric measures from the National Health and Nutrition Examination Survey (NHANES) 2017–2018 dataset.

METHODS

Study design

We analyzed questionnaire data from the annual NHANES, which investigated several health and nutrition-related factors of the US population, in a multi-wave cross-sectional study (Chen et al., 2020). This analysis examined demographic, dietary measures, alcohol use and anthropometric data collected from 2017 to 2018.

Participants

A total of 9254 male and female participants from the NHANES 2017–2018 survey were recruited from all 50 states, and the District of Columbia. Information about the overall NHANES sample are further described here (Chen et al., 2020). This study included individuals over 18 years in age who responded to demographic, alcohol use, dietary questionnaires and anthropometric examinations (Table 1). Participants were excluded from the current sample if they were pregnant at the time of examination, breastfeeding, had cancer in the past year (medical conditions questionnaire) or had missing dietary recall data. There were 6934 participants excluded, leaving 2320 participants to be included in the present analysis (Fig. 1).

Table 1.

demographic characteristics by dose

	Total	Low-quantity drinker	4/5 to 7 Drinks Medium-quantity drinker	8 to 11 Drinks High-quantity drinker	12+ Drinks Very high-quantity drinker	P-Value
N in millions (% of sample)	151.79	117.05 (77.11%)	24.82 (16.35%)	5.19 (3.42%)	4.73 (3.12%)
Unweighted N	2320	1858 (80.09%)	327 (14.09%)	67 (2.89%)	68 (2.93%)
Age, y, mean ± SE	44.43 ± 0.71	45.62 ± 0.78	40.95 ± 1.12**	40.40 ± 2.34	37.61 ± 3.03*	<0.01
Gender, N (%)						<0.01
Female	77.38 (50.98%)	63.87 (54.56%)	11.98 (48.28%)	0.90 (17.43%)	0.63 (13.30%)
Male	74.41 (49.02%)	53.19 (45.44%)	12.84 (51.72%)	4.29 (82.57%)	4.11 (86.70%)
Race, N (%)						0.24
Non-Hispanic White	97.11 (63.98%)	74.63 (63.76%)	16.09 (64.83%)	3.41 (65.64%)	2.98 (63.00%)
Non-Hispanic Black	16.51 (10.88%)	12.49 (10.67%)	3.43 (13.84%)	0.29 (5.52%)	0.30 (6.29%)
Asian	7.03 (4.63%)	6.31 (5.39%)	0.47 (1.89%)	0.12 (2.26%)	0.13 (2.73%)
Mexican American	13.83 (9.11%)	10.47 (8.95%)	2.51 (10.12%)	0.52 (10.03%)	0.32 (6.80%)
Other Hispanic	9.88 (6.51%)	7.58 (6.47%)	1.37 (5.52%)	0.55 (10.59%)	0.38 (8.06%)
Other race	7.45 (4.91%)	5.57 (4.76%)	0.95 (3.81%)	0.31 (5.96%)	0.62 (13.11%)
Family PIR, Mean ± SE	3.21 ± 0.07	3.26 ± 0.07	3.18 ± 0.18	2.84 ± 0.34	2.45 ± 0.40	0.09
BMI, kg/m2, mean ± SE	29.67 ± 0.30	29.92 ± 0.35	28.74 ± 0.62	30.35 ± 1.17	27.76 ± 1.11	0.11
WC (cm)	100.04 ± 0.79	100.51 ± 0.84	97.83 ± 1.65	102.72 ± 3.02	96.91 ± 2.90	0.14
Smoking status						<0.01
Non-smoker	91.76 (60.45%)	76.55 (65.40%)	11.86 (47.81%)	1.99 (38.29%)	1.35 (28.57%)
Former	34.43 (22.69%)	25.84 (22.08%)	7.23 (29.13%)	0.94 (18.04%)	0.43 (9.02%)
Current	25.60 (16.87%)	14.66 (12.52%)	5.72 (23.06%)	2.27 (43.66%)	2.95 (62.41%)
Sedentary minutes (per day)	349.25 ± 8.88	350.05 ± 9.11	362.45 ± 26.99	313.47 ± 29.01	301.07 ± 37.33	0.50
Bike/walk minutes (per day)	41.43 ± 4.33	45.02 ± 6.01	32.52 ± 4.86	39.71 ± 9.20	28.28 ± 7.10	0.28
Moderate activity minutes (per day)	146.91 ± 6.54	145.73 ± 7.09	129.71 ± 15.24	173.89 ± 64.27	227.55 ± 61.08	0.55
Vigorous activity minutes (per day)	179.75 ± 13.56	165.44 ± 11.61	152.97 ± 27.88	283.48 ± 94.77	349.66 ± 127.13	0.22

Note. Mean ± SE; ^* = P < 0.05; ^**P < 0.01 for no/infrequent drinker as reference

Figure 1.

Flowchart showing participant selection.

Measures

Demographics

Demographic questions included gender, age, race and ethnic background, socioeconomic status (SES), and education level. In-person interviews were conducted in the participant’s homes, via computer-assisted personal interview system (CAPI) system, by trained interviewers. Interviews were conducted in the preferred language of the participant (Prevention, 2020c).

Medical conditions

The medical conditions examination consisted of a broad range of questions about the participants’ health conditions and medical history. Responses were collected in the home via the CAPI system by trained interviewers. The data were then reviewed for completeness and consistency (Prevention, 2020d). The current analysis excluded participants who indicated a previous cancer diagnosis.

Mental health depression screener

The mental health depression screener was a nine-item depression screening instrument, also known as the Patient Health Questionnaire (PHQ) (Kroenke et al., 2001; Kroenke and Spitzer, 2002). This was administered to participants to determine frequency of depression symptoms over the last 2 weeks (Spitzer et al., 1999). For each symptom question, answers ranged from 0 to 3 with 0 corresponding with ‘not all;’ 1 with ‘several days’, 2 with ‘more than half the days’ and 3 with ‘nearly every day.’ Trained interviewers asked these questions in a private room at the Mobile Examination Center (MEC) via the CAPI system. A total score ranging from 0 to 27 was calculated from participants’ answers to determine depression severity (Prevention, 2020e).

Dietary data

The Dietary Examination consisted of two dietary recall interviews to assess energy and nutrient content from daily food and beverage intake. Dietary recall assessed all food consumed during the prior 24-h period. In addition, it collected data concerning salt use, fish and shellfish intake, and adherence to any specific diet. Interviewers also confirmed if the dietary recall was consistent with regular dietary patterns. The first interview was administered to all NHANES participants in the MEC. To estimate accurate measurements of food consumed, participants were given standard materials and other kitchen utensils for reference. These included a ruler, spoons, measuring cups and a food model guides with detailed pictures for food amounts. These materials were given to participants to take home for reference during the next interview. The second interview was held via telephone within the next 3–10 days, on a different weekday than the first interview.

Alcohol use

The Alcohol Use Questionnaire included details about lifetime and current usage patterns, and recorded details concerning number of drinks consumed monthly, daily, weekly and within a 2-h period. Details about alcohol types were not collected. Frequency and quantity responses from these questionnaires categorized people as binge or non-binge drinkers, according to NIAAA guidelines (Alcoholism). At the MEC, participants were interviewed by trained MEC personnel via the CAPI system, immediately after the dietary examination (Prevention, 2020a).

Smoking – cigarette use

The Cigarette Use Questionnaire was administered to characterize participants’ overall history of use and past 30-day use of traditional cigarettes, cigars, electronic nicotine methods and smokeless tobacco. Other details concerning cigarette type and brand were also collected. Participants were interviewed at home by trained administrators via the CAPI system (Prevention, 2020g). All participants were categorized as never, former or current cigarette smokers. Those who had not smoked 100 or more cigarettes in their lifetimes were considered non-smokers. Those who answered ‘yes’ to having smoked 100 or more cigarettes in their lifetimes and who had indicated ‘not at all’ to current cigarette use were classified as former smokers. Current smokers consisted of those who had smoked at least 100 cigarettes in their lifetimes and who currently smoke ‘every day’ or ‘some days’.

Physical activity

The Physical Activity Questionnaire for adults was a 16-item survey, based on the Global Physical Activity Questionnaire (Armstrong and Bull, 2006), that collected details about physical activity. Most questions asked about the number of minutes spent performing different levels of activity in day or week. Participants were interviewed at home by trained administrators via the CAPI system (Prevention, 2020f).

Anthropometric measures

These measures were used to categorize participants as underweight, normal weight, overweight and obese. Anthropometric data were accurately recorded by trained health technologists at the MEC. In this paper, we used participants’ height and weight to calculate body mass index (BMI; kg/m²). BMI was used to categorize participants as underweight, normal weight, overweight and obese. WC was also measured to assess adiposity (Prevention, 2020b).

Statistical analysis

Alcohol use classification

Participants were classified into one of four categorizations based on their self-reported number of drinks per episode of alcohol consumption with a frequency of at least once per month: 0– < 4 drinks/episode (low-quantity drinkers); 4–7 drinks/episode (medium-quantity drinkers); 8–11 drinks/episode (high-quantity drinkers); 12+ drinks/episode (very high-quantity drinkers). For example, if the participant had indicated that they had consumed 12+ drinks in a single sitting at least once per month over the course of a year, they were considered in the 12+ drinks/episode category. Participants who have had 4–7 drinks in a single sitting at least once a month over the past year were considered in the 4–7 drinks/episode category.

Data analysis

Data were cleaned according to NHANES recommendations (Centers for Disease Control and Prevention, 2013, 2019b). We employed the Balanced Repeated Replication method using 2-year sampling weights from the second 24-h dietary recall to create estimates (Centers for Disease Control and Prevention, 2020). We calculated frequency and descriptive statistics for the demographic variables. Correlations were evaluated using the Rao-Scott F Adjusted Chi-square statistic and Wald F-test. Sample-weighted multiple linear regression models were utilized to analyze the associations among volume of alcohol consumption, weight history and macronutrient intake. Model 1 demonstrated the effect of alcohol volume on each macronutrient consumption. Model 2 incorporated demographic variable adjustments for BMI, age, gender, race and ethnicity, and federal poverty to income ratio (PIR). Model 3 adjusted for cigarette smoking status. Data analyses were conducted with R version 3.6.3. Statistical significance was considered at a two-sided P-value of <0.05.

RESULTS

Demographic characteristics by alcohol use category are in Table 1. In this sample, 77.11% of participants consumed 0– < 4 drinks/episode, 16.35% drank 4–7 drinks/episode, 3.42% consumed 8–11 drinks/episode and 3.12% 12+ drinks/episode. Participants who drink 4–7 drinks/episode were older, on average, compared with high-quantity drinkers. A significantly higher percentage of high-quantity drinkers (82.57%) and very high-quantity drinkers (86.70%) were male. No significant racial differences in consumption were found. PIR was significantly inversely associated with the frequency of alcohol consumption per sitting. A significantly higher proportion of high-quantity drinkers were current smokers. No significant associations between alcohol consumption and physical activity, BMI or WC were found. WC was significantly higher in low-quantity drinkers compared with medium and high-quantity drinkers. The PHQ-9 scores and their corresponding depression classifications did not differ significantly by drinking status (Table 2).

Table 2.

Depression and weight history by alcohol intake dose

	Total	Low-quantity drinker	4/5 to 7 Drinks Medium-quantity drinker	8 to 11 Drinks High-quantity drinker	12+ Drinks Very high-quantity drinker	P-Value
PHQ-9 (Continuous)	3.01 (0.09)	2.97 ± 0.11	2.77 ± 0.18	3.86 ± 0.75	4.26 ± 0.66	0.14
PHQ-9 (Categorical)						0.16
None	44.58 (29.50%)	35.00 (30.06%)	7.23 (29.21%)	1.16 (22.47%)	1.18 (24.88%)
Mild	22.38 (14.81%)	15.67 (13.46%)	4.04 (16.32%)	1.14 (22.04%)	1.53 (32.32%)
Minimal	72.30 (47.84%)	56.37 (48.41%)	12.04 (48.62%)	2.41 (46.49%)	1.48 (31.29%)
Moderate	7.91 (5.23%)	6.38 (5.48%)	1.11 (4.47%)	0.13 (2.53%)	0.29 (6.10%)
Moderately severe	3.24 (2.15%)	2.48 (2.13%)	0.31 (1.24%)	0.34 (6.48%)	0.12 (2.57%)
Severe	0.71 (0.47%)	0.54 (0.47%)	0.04 (0.14%)	0 (0%)	0.13 (2.84%)

As expected, high-quantity drinkers consumed significantly more kilocalories (2569.91) compared with low-quantity drinkers (2106.73). After removing kilocalories consumed from alcohol, the relationship did not persist. High-quantity drinkers consumed a lower percentage of energy from fat intake (34.64%) compared with low-quantity drinkers (36.61%). High-quantity drinkers consumed a lower percentage of their energy from carbohydrates (42.11%, 41.68% and 41.94%) compared with low-quantity drinkers (46.57%). Very high-quantity drinkers consumed significantly fewer grams of dietary fiber (12.81 g) compared with low-quantity drinkers (16.67 g). Medium and high-quantity drinkers consumed a lower percentage of their energy from sugar (17.00% and 15.43%) compared with low-quantity drinkers (19.87%) (Table 3). Upon adjustments for demographic factors, these relationships did not hold.

Table 3.

Macronutrient by alcohol intake dose

	Total	Low-quantity drinker	4/5 to 7 Drinks Medium-quantity drinker	8 to 11 Drinks High-quantity drinker	12+ Drinks Very high-quantity drinker	P-Value
KiloCal	2143.4 ± 30.07	2106.73 ± 32.77	2183.81 ± 65.87	2569.91 ± 151.04*	2370.32 ± 199.72	0.03
KiloCal without alcohol	2063 ± 28.68	2059.56 ± 31.99	2017.52 ± 51.21	2339.36 ± 137.59	2084.5 ± 174.44	0.12
Protein %	15.98 ± 0.17	15.96 ± 0.14	16.47 ± 0.48	15.36 ± 0.87	14.67 ± 0.88	0.16
Protein (gm)	83.58 ± 1.43	82.10 ± 1.60	87.23 ± 2.07	98.39 ± 7.51	84.65 ± 7.59	0.10
Fat %	36.29 ± 0.30	36.61 ± 0.36	35.61 ± 0.63	34.64 ± 0.85*	33.78 ± 1.74	0.13
Fat (gm)	87.00 ± 1.37	86.61 ± 1.54	86.36 ± 2.03	99.46 ± 6.45	86.55 ± 7.40	0.26
Carb %	45.53 ± 0.36	46.57 ± 0.34	42.11 ± 0.96**	41.68 ± 1.21**	41.94 ± 1.50**	<0.01
Carb (gm)	243.31 ± 3.77	244.96 ± 4.19	229.49 ± 8.78	269.11 ± 16.85	246.92 ± 24.40	0.17
SFat %	11.77 ± 0.14	11.78 ± 0.16	11.85 ± 0.40	11.47 ± 0.26	11.78 ± 0.16	0.63
Saturated fat (gm)	28.22 ± 0.51	27.86 ± 0.54	28.75 ± 0.75	33.15 ± 2.35*	29.03 ± 2.23	0.19
Dietary fiber (gm)	16.53 ± 0.43	16.67 ± 0.48	16.40 ± 0.61	17.58 ± 1.40	12.81 ± 1.46*	0.11
Sugar (gm)	103.57 ± 2.15	105.98 ± 2.34	93.38 ± 6.14	102.81 ± 11.13	98.00 ± 14.23	0.38
Sugar %	19.14 ± 0.24	19.87 ± 0.28	17.00 ± 0.88*	15.43 ± 11.08**	16.31 ± 1.57	<0.01

Note. Mean ± SE; ^* = P < 0.05; ^**P < 0.01 for no/infrequent drinker as reference.

We also conducted similar analysis by combining high and very-high quantity alcohol drinking groups and compared with medium and low-quantity drinkers. However, we observed similar results as seen on separating the excessive drinkers into 8–11 drinks and 12+ drinks/episode. The one statistical difference we observed in this analysis which we did not observe in the previous analysis was the significantly higher kilocalorie intake in 8+ drinkers (2569.91 ± 151.04) compared with low-quantity drinkers (2106.73 ± 32.77; P = 0.02) (Supplemental Data S1).

DISCUSSION

Alcohol consumption has gradually increased over the last couple of decades and can contribute to the development of several comorbidities. This study aimed to determine how alcohol consumption may influence macronutrient intake, anthropometric measures and depression severity. We hypothesized that high quantity drinkers would exhibit higher food intake and dietary fat composition. This analysis specifically found differences between the drinking groups in the following categories: total daily calories, carbohydrate consumption and fat composition. Primarily, differences were reported for those who drank 8–11 alcoholic drinks/episode.

There were several differences in macronutrient composition between the high quantity drinkers and low-quantity drinkers. Mainly, diets from high quantity drinkers consisted of a lower percentage of both carbohydrates than that of low-quantity drinkers; this remained significant even after demographic adjustment. In addition, the highest quantity drinkers consumed significantly fewer grams of dietary fiber than low-quantity drinkers, which may reflect poorer diet quality. This finding is critical because fiber intake is important to sustain proper digestion. These results are consistent with Colditz et al. who found that increases in alcohol consumption were linked with decreases in carbohydrate, particularly sugar, intake (Colditz et al., 1991). Our hypothesis is that ethanol’s effects on gut hormones and brain circuitry may decrease appetite for carbohydrate-rich and sweet foods.

Participants who drank 8–11 alcoholic drinks/episode consumed 460 more kilocalories than low-quantity drinkers. Although this observation was not significant after alcohol calorie and demographic adjustment, this trend has been seen in several studies (Colditz et al., 1991; Yeomans, 2004; Gee, 2006; Schrieks et al., 2015). Interestingly, although not statistically significant, energy intake values that excluded calories consumed from alcohol revealed that medium quantity drinkers consumed around 300 more calories from food than the other groups. This trend may reflect our hypothesis on alcohol’s tendency to increase energy dense food intake. In a randomized control trial, Schreiks et al. found that alcohol consumption 45 min before an ad libitum meal increased food intake (Schrieks et al., 2015). Authors and a previous review suggest that this may be increases in fat preference and intake. Yet, there is little evidence detailing an underlying mechanism, and this stimulatory effect is evident outside of acute alcohol intake (Yeomans et al., 2003).

Some studies have found that alcohol consumption is associated with increases in fat preference and intake (Schrieks et al., 2015). This may result from disinhibited eating and/or the need to sooth an irritated stomach after a drinking episode. Palatable, energy-dense foods have the potential to assuage the sickness that manifests after alcohol consumption (Sirohi et al., 2017). Yet, this analysis reports the opposite trend; participants who drank 8–11 alcoholic drinks/episode had a lower portion of energy from dietary fat. In addition, participants who drank 8–11 alcoholic drinks/episode consumed more grams of protein than the other groups. Although this difference was not statistically significant, this may indicate lifestyle differences between the drinking groups. Ultimately, the differences in macronutrient consumption found in this analysis may reflect alcohol’s influence on food intake, appetite and preferences. Even so, more research is needed to fully understand the underlying mechanisms.

Our study dataset comprised of roughly equal number of males and females, while majority of participants identified as white. Younger individuals consumed more alcohol, and this is consistent with literature that has found that young adults are developmentally and socially higher at risk for binge drinking (Krieger et al., 2018). Likewise, the family PIR was negatively associated with alcohol consumption. SES can strongly influence alcohol use and related outcomes. In a meta-analysis of 15 studies capturing data on 133 million people worldwide, poor economic condition was found to be an important indicator of alcohol dependence (Collins, 2016). There is also literature suggesting a higher mortality risk in lower income individuals with alcohol use disorder (Woon Kwak et al., 2017). Although the mechanism is not fully understood, this may be due to everyday stresses of poverty that contribute to substance abuse disorders and higher cortisol levels in low-income individuals.

One advantage of this analysis is its inclusion of physical activity data. Details on physical activity can shed light on participants’ lifestyle habits. Higher quantity drinkers self-reported less sedentary behavior and more moderate and vigorous physical activity than the low-quantity drinkers. Both groups spent 3–4 h in moderate physical activity and 5–6 h of vigorous physical activity per day. This may indicate that these participants had extremely active occupations or lifestyle. This increased physical activity time may be associated with the energy intake results. Intense physical activity demands higher energy consumption to maintain energy balance. Individuals who drank more heavily may have compensated for extra calories consumed with exercise, and this may account for the lack of differences we found in weight status.

WC and BMI were similar amongst all groups. Both WC and BMI were slightly higher in medium and low-quantity drinkers than in low and high-quantity drinkers. In a prospective cohort study conducted by Tolstrup et al., increased drinking frequency in women was inversely associated with WC changes (Tolstrup et al., 2008). Their finding supports our results, indicating an association between drinking patterns and abdominal obesity (Tolstrup et al., 2005; Tolstrup et al., 2008). Interestingly, we observed similar WC among the medium and low-quantity drinkers, which presumably accounts for the balanced calorie intake in these groups. The high WC and BMI amongst the medium-quantity group may be linked with their relatively high total daily intake, as noted in Table 3.

Limitations

In our dataset, there was a lower number of high-quantity drinkers, as compared with participants in the other groups, which may have influenced the subgroup comparison with several variables. Even though the analysis was weighted, some of our cell sizes were generally small. This analysis excluded individuals with substance use disorder. These individuals may represent the excessive drinking patterns that can affect dietary intake; hence, in future studies it may be beneficial to include this population. Among NHANES participants who were asked the alcohol questions (~5500 out of around 9000), there were only 108 people who did not have dietary recall data and who had complete responses to our included demographic variables. We acknowledge that the exclusion of 6934 participants’ data in total (following our inclusion/exclusion criteria) could have possibly created nonresponse bias: the sample who completed the questionnaires used in this analysis may differ in important ways from those who did not complete the questionnaire. Unfortunately, due to missing variables, it is hard to characterize these differences. We assume the comparisons would have yielded more meaningful results if we had more details about macronutrient composition based on alcohol and non-alcohol diet. Moreover, for a more precise estimation of weight change among alcoholics, it is very crucial to know the type of alcohol beverage consumed by them. The information related to the precise amount of calorie intake is influenced by the type of alcohol consumed and we acknowledge the lack of this data as a limitation of our study. Therefore, considering the impact of these limitations in the present dataset on the results obtained, we hope future studies would take into consideration these factors to reach at a more conclusive result on the association of alcohol intake and weight gain.

CONCLUSION

Excessive alcohol consumption has several negative implications on an individual’s health including increasing risk for cancer, cardiovascular disease, lung damage, metabolic dysfunction and other comorbidities. This analysis utilized the NHANES 2017–2018 data on anthropometric measures, alcohol and food intake. Our findings within the current dataset are supported by previous literature and show that high alcohol intake is associated with differences in eating habits and body composition. Primarily, alcohol consumption may contribute to reductions in proper carbohydrate intake, especially dietary fiber, which is crucial for optimal digestive health. The results we obtained emphasize the need to compare more specific drinking patterns and their impact on nutrient intake. Although some results conflict with previous studies, the mechanisms underlying alcohol’s effect on ingestive and digestive metabolic pathways are still unclear and require further investigation. The recent and ongoing COVID-19 pandemic continues to affect drinking patterns and eating behavior. Hence, it is important to continue enhancing and investing in resources that can educate diverse communities about the long-term effects of alcohol use and dietary composition.

STATEMENT OF AUTHORS’ CONTRIBUTIONS TO MANUSCRIPT

A.M.C.: Conceptualization; methodology; software; analysis; investigation; data curation; writing; supervision; project administration. Y.Z.: Methodology; software; analysis; investigation; data curation; writing. B.B., C.MD., K.A.: Conceptualization; investigation; writing. P.V.J.: Conceptualization; methodology; investigation; writing; supervision; project administration.

DATA AVAILABILITY

Data described in the manuscript and code book are publicly and freely available without restriction at https://wwwn.cdc.gov/nchs/nhanes/continuousnhanes/default.aspx?BeginYear=2017. Analytic code is available by contacting the corresponding author (P.V.J.).

FUNDING

Dr Joseph is supported by the National Institute of Alcohol Abuse and Alcoholism under award number, Z01AA000135 and the National Institute of Nursing Research, the NIH Office of Workforce Diversity, National Institutes of Health Distinguished Scholar and the Rockefeller University Heilbrun. AMC was supported, in part, by the National Institute of Nursing Research of the National Institutes of Health [grant number K23NR017209].

CONFLICT OF INTEREST STATEMENT

AMC reports grants from Eli Lilly and Company and WW International, Inc., outside the submitted work. The other authors have no conflicts of interest to disclose.