Assessing the use of prescription drugs and dietary supplements in obese respondents in the National Health and Nutrition Examination Survey

Laura A Barrett; Aiwen Xing; Julia Sheffler; Elizabeth Steidley; Terrence J Adam; Rui Zhang; Zhe He

doi:10.1371/journal.pone.0269241

. 2022 Jun 3;17(6):e0269241. doi: 10.1371/journal.pone.0269241

Assessing the use of prescription drugs and dietary supplements in obese respondents in the National Health and Nutrition Examination Survey

Laura A Barrett ¹, Aiwen Xing ², Julia Sheffler ³, Elizabeth Steidley ¹, Terrence J Adam ⁴, Rui Zhang ⁴, Zhe He ^1,^3,^*

Editor: Jingjing Qian⁵

PMCID: PMC9165812 PMID: 35657782

Abstract

Introduction

Obesity is a common disease and a known risk factor for many other conditions such as hypertension, type 2 diabetes, and cancer. Treatment options for obesity include lifestyle changes, pharmacotherapy, and surgical interventions such as bariatric surgery. In this study, we examine the use of prescription drugs and dietary supplements by the individuals with obesity.

Methods

We conducted a cross-sectional analysis of the National Health and Nutrition Examination Survey (NHANES) data 2003–2018. We used multivariate logistic regression to analyze the correlations of demographics and obesity status with the use of prescription drugs and dietary supplement use. We also built machine learning models to classify prescription drug and dietary supplement use using demographic data and obesity status.

Results

Individuals with obesity are more likely to take cardiovascular agents (OR = 2.095, 95% CI 1.989–2.207) and metabolic agents (OR = 1.658, 95% CI 1.573–1.748) than individuals without obesity. Gender, age, race, poverty income ratio, and insurance status are significantly correlated with dietary supplement use. The best performing model for classifying prescription drug use had the accuracy of 74.3% and the AUROC of 0.82. The best performing model for classifying dietary supplement use had the accuracy of 65.3% and the AUROC of 0.71.

Conclusions

This study can inform clinical practice and patient education of the use of prescription drugs and dietary supplements and their correlation with obesity.

Introduction

As a major health and economic crisis affecting the modern world, much progress has been made in identifying and developing strategies for preventing and treating obesity. Currently, treatment options include lifestyle changes, pharmacotherapy, and surgical interventions (e.g., intragastric balloons and bariatric surgery) [1]. In terms of pharmacotherapy, there are five approved prescription drugs (RXD) (orlistat, 1999; phentermine/topiramate, 2012; liraglutide, 2014; naltrexone/bupropion, 2014; and semaglutide, 2021) that can be prescribed for weight loss [2]. All but orlistat, which reduces the absorption of fat, work by helping the individual to limit caloric intake [3]. There are also four RXD that are similar to amphetamines that can be used short-term (phendimetrazine, diethylpropion, phentermine, and benzphetamine) [4]. There have been three other well-known RXD that were approved for use and then removed from the market. The first one is fenfluramine/phentermine (fen-phen) which was discontinued in 1997 because fenfluramine was shown to cause cardiac issues [5]. The second one is sibutramine, which was withdrawn in 2010 due to an increased risk of stroke and myocardial infarction [6]. The third one is lorcaserin, which was withdrawn in February 2020 after a clinical trial showed an increased occurrence of cancers [7]. Due to the cost of pharmacotherapy and surgical interventions, as well as other reasons, dietary supplements (DS) are often used as a cost-sensitive and easily accessible, albeit less scientifically supported, alternative treatment of obesity [8,9].

Individuals with obesity face an increased risk of chronic diseases, namely depression, type 2 diabetes, cardiovascular disease, and many cancers including those of the colon, breast, kidney, and pancreas [10]. Many of these conditions require pharmaceutical intervention as part of the treatment plan and individuals with obesity often use RXD to manage these conditions. Overall, RXD use in the United States has increased [9]. This increase is partly influenced by the development of new RXD, the expansion of RXD coverage by insurance companies, and increased rates of chronic conditions such as obesity [9]. The greatest increase in RXD use has been in those used for treating conditions found to be associated with obesity, specifically antihypertensives, antihyperlipidemic, antidiabetics, and antidepressants [11].

Recent studies have researched obesity in relation to specific drugs or drug types [12–16]. There have also been recent studies that examined various aspects of obesity such as childhood obesity [16,17], obesity and hours spent at work [18], exposure to certain pollutants or chemicals [19–21], trends in obesity [22], and obesity and waist circumference [23]. However, there have not been any studies that look at overall RXD use in individuals with obesity. Being able to see this bird’s eye view of this relationship is important because understanding the patterns of RXD use among people with obesity, who often have other chronic conditions, can inform both clinical practice and research [9]. This is challenging because there are cross-over issues between RXD, their side effect of weight gain, and their therapeutic effect on obesity and its comorbidities. For example, certain blood-glucose-lowering RXD and psychotropics may lead to unintended weight gain [24]. In this project, we aim to gain an in-depth understanding of both the relationship between obesity and RXD use, as well as the correlations between specific RXD and DS use in individuals with obesity. We also aim to understand if demographic variables and obesity status can assist with classifying an individual’s likelihood of using any RXD or DS, not just specific RXD or DS.

Materials and methods

The National Center for Health Statistics of CDC has been conducting the National Health and Nutrition Examination Survey (NHANES) as a continuous cross-sectional health survey [25]. It samples the non-institutionalized population of the United States with a stratified multistage probability model and releases results of a set of health surveys, medical examinations, a physical, and laboratory test every two years. Its rigorous quality control ensures high-quality data collection and national representativeness. The NHANES data have been used in many public health and epidemiology studies [26–32].

Demographic, physical examination, prescription drug (RXD) and dietary supplement (DS) use [8], and health insurance information were extracted from NHANES for survey years 2003–2018 (8 survey cycles). The 16-year sample weight (the number of people in the US population that a sample in the combined sample can represent) was calculated according to the analytical guideline of NHANES [25]. The obese group is defined as: (1) BMI ≥ 30 kg/m² [33], (2) age ≥ 18 [34]. From the original NHANES data, 2689 respondents with no BMI and 45 respondents with no RXD use information were removed from the dataset. S1 Table in the supplementary material lists the NHANES file, the NHANES variable names, the associated questions, and how they are referred to in this paper.

Data analysis

Basic characteristics

A profile for each group was created that included sex, age, race, annual household income, and health insurance status.

Statistical analysis

We conducted multivariate logistic regression analyses to access: 1) the associations between using RXD/DS and variables of interests (i.e., demographic characteristics, poverty income ratio, insurance status, and obesity status) and 2) the associations between taking specific types of RXD and obesity status. Weighted multivariate logistic regression analyses were used to obtain odds ratios (OR) and 95% CIs with 16-year sample weight. All interested variables were introduced in the model first then backward elimination with a threshold of p = 0.05 was applied to eliminate variables. We kept only the variables that were significant in the initial model in the final model. The significance level was set as 0.05. All statistical analyses were performed by SAS software (SAS Institute Inc), version 9.4.

We performed two separate logistic regression analyses. 1) Usage based on the specific number of RXD/DS that the individual used was the dependent variable. Demographic characteristics were included as independent variables to examine whether taking a specific number of RXD/DS was significantly associated with demographic characteristics within non-obese or obese groups separately. This regression evaluated covariates down to the two groups. 2) Obesity status and demographic characteristics were included as independent variables to test their associations with taking a specific number of RXD/DS within the whole population. In the second analysis, the dichotomous dependent variable was whether participants were prescribed the specific types of RXD; obesity status was set as independent variable with reference as non-obese group.

Classification modeling

We used Weka [35] to evaluate different machine learning models for classifying whether a respondent used one or more RXD or DS, respectively. In the first round of the modelling, the variables included age group, sex, BMI category, race, annual household income, and insurance status. As we are also interested in seeing whether DS use would help classify RDX use and vice versa, in the second round of modeling, DS use or RXD use was added as a variable for classifying the use of the other type. A third round of modelling was done using the poverty income ratio (PIR: a ratio of family income to poverty threshold) in place of the annual household income. Lastly, we further evaluated if machine learning models were able to classify how many RXD were used. For this round, we created four groups of RDX count (i.e., 0, 1–2, 3–5, >5). We used feature selection based on correlation (“CorrelationAttributeEval” in Weka) to rank the importance of the variables. We evaluated four major machine learning algorithms including Naïve Bayes, Logistic Regression, SMO (Weka’s implementation of Support Vector Machine), and Random Forest. Deep learning techniques were not employed because of the small number of variables in this dataset. The data was preprocessed to make all numerical data nominal. 10-fold cross validation was employed. In each fold, 90% of the data was used for training and 10% of the data was used for testing. The models were compared using overall accuracy, precision, recall, F1-score, and AUROC.

Results

Basic characteristics

Table 1 shows the basic characteristics of the two groups regarding their RXD use. There are a few differences based on obesity status and demographics. In the non-obese group 52.59% of people report taking 1 or more RXD. The obese group has a higher reported use at 63.84%. In both groups, females report higher use than males. In addition, RXD use increases with age in both groups. Race also plays a role in reported RXD use in both groups; Non-Hispanic Whites have the highest percentages of use while Mexican Americans show the lowest percentages. Lastly, in both groups, those with health insurance reported higher RXD use than those that reported not having insurance.

Table 1. Basic characteristics of the study population.

Variable	Non-obese				Obese
	0 RXD		1 or more RXD		0 RXD		1 or more RXD
	Weight	%	Weight	%	Weight	%	Weight	%
	Count	%	Count	%	Count	%	Count	%
Total	66033363	47.41%	73250493	52.59%	28449346	36.16%	50222655	63.84%
Total Count	14092	49.27%	14508	50.73%	6119	37.25%	10310	62.75%
Gender
Male	37570793	54.85%	30921089	45.15%	15263957	41.90%	21161327	58.10%
Male Count	7941	54.27%	6692	45.73%	3005	41.85%	4175	58.15%
Female	28462570	40.21%	42329405	59.79%	13185389	31.21%	29061328	68.79%
Female Count	6151	44.04%	7816	55.96%	3114	33.67%	6135	66.33%
Age
18–24	14166880	69.26%	6287886	30.74%	4828265	70.30%	2040083	29.70%
18–24 Count	3708	75.94%	1175	24.06%	1214	73.53%	437	26.47%
25–34	17173005	66.88%	8505063	33.12%	8170177	63.04%	4789193	36.96%
25–34 Count	3383	71.93%	1320	28.07%	1680	66.77%	836	33.23%
35–44	14370023	59.67%	9713144	40.33%	7134034	46.66%	8154368	53.34%
35–44 Count	2784	65.11%	1492	34.89%	1418	50.00%	1418	50.00%
45–54	11133113	44.23%	14038322	55.77%	4884217	30.47%	11146006	69.53%
45–54 Count	2009	48.96%	2094	51.04%	958	33.22%	1926	66.78%
55–64	6198064	31.45%	13511728	68.55%	2505342	17.60%	11727619	82.40%
55–64 Count	1316	33.16%	2653	66.84%	583	19.56%	2398	80.44%
65–74	2030697	15.24%	11290366	84.76%	714089	8.08%	8126671	91.92%
65–74 Count	565	17.04%	2750	82.96%	195	8.71%	2043	91.29%
75 and over	961580	8.85%	9903984	91.15%	213221	4.79%	4238715	95.21%
75 and over Count	327	9.76%	3024	90.24%	71	5.37%	1252	94.63%
Race
Mexican American	7643611	70.58%	3186177	29.42%	4773068	60.80%	3077063	39.20%
Mexican American Count	2879	64.39%	1592	35.61%	1616	52.95%	1436	47.05%
Other Hispanic	4645998	62.49%	2788237	37.51%	2239910	52.63%	2016083	47.37%
Other Hispanic Count	1366	54.90%	1122	45.10%	643	43.45%	837	56.55%
Non-Hispanic White	38690156	40.72%	56324147	59.28%	15007808	29.36%	36107896	70.64%
Non-Hispanic White Count	4696	37.94%	7680	62.06%	1758	27.01%	4750	72.99%
Non-Hispanic Black	7827436	58.04%	5659904	41.96%	4831848	41.64%	6772380	58.36%
Non-Hispanic Black Count	2953	54.23%	2492	45.77%	1695	38.15%	2748	61.85%
Other Race or Multi-Racial	7226162	57.73%	5292029	42.27%	1596713	41.52%	2249232	58.48%
Other Race or Multi-Racial Count	2198	57.54%	1622	42.46%	407	43.02%	539	56.98%
Household Income
$0 to $34,999	18871838	47.76%	20642494	52.24%	8786323	36.67%	15177268	63.33%
$0 to $34,999 Count	5322	48.16%	5728	51.84%	2460	36.45%	4289	63.55%
$35,000 to $74,999	19147850	48.23%	20556912	51.77%	9374135	37.16%	15854727	62.84%
35,000 to $74,999 Count	3850	50.25%	3812	49.75%	1849	39.01%	2891	60.99%
$75,000 and over	21861695	45.06%	26655556	54.94%	7853627	33.23%	15782826	66.77%
$75,000 and over Count	3352	47.98%	3634	52.02%	1193	35.04%	2212	64.96%
Poverty Income Ratio
0-.99	11985065	54.42%	10038503	45.58%	5239978	40.20%	7794072	59.80%
0-.99 Count	3443	55.31%	2782	44.69%	1520	40.90%	2196	59.10%
1–1.99	14166723	50.07%	14127988	49.93%	6774522	38.90%	10642021	61.10%
1–1.99 Count	3471	49.70%	3513	50.30%	1630	38.87%	2563	61.13%
2–2.99	9141944	46.30%	10604451	53.70%	4248535	35.37%	7763139	64.63%
2–2.99 Count	1810	46.98%	2043	53.02%	820	35.56%	1486	64.44%
3–3.99	7371681	47.32%	8205982	52.68%	3044609	33.21%	6121952	66.79%
3–3.99 Count	1233	47.31%	1373	52.69%	529	34.33%	1012	65.67%
4–4.99	5149047	43.13%	6790653	56.87%	2275579	31.94%	4847852	68.06%
4–4.99 Count	821	44.94%	1006	55.06%	335	32.06%	710	67.94%
5	10758790	39.64%	16379571	60.36%	3821273	31.85%	8178192	68.15%
5 Count	1558	41.74%	2175	58.26%	537	31.98%	1142	68.02%
Health Insurance ^*
Yes	46794083	41.29%	66549204	58.71%	19812159	30.28%	45614641	69.72%
Yes Count	9206	41.52%	12968	58.48%	3912	29.92%	9163	70.08%
No	19073757	74.28%	6604505	25.72%	8582725	65.52%	4516163	34.48%
No Count	4830	76.10%	1517	23.90%	2194	66.04%	1128	33.96%

Open in a new tab

*Overall insurance coverage was 82.0% covered and 17.8% not covered.

Specific RXD types

For the identified RXD types, we evaluated the association between the specific types of RXD use and obesity status. Table 2 shows the odds ratio between obese and non-obese group taking prescription drugs. While looking at the types of RXD, there were differences in use between the obese and non-obese groups. Cardiovascular agents are the most used RXD type in both groups. This is not surprising given the prevalence of cardiovascular diseases in the United States. Compared with those in the non-obese group, individuals with obesity are more likely to take cardiovascular agents (OR = 2.095, 95% CI 1.989–2.207) and metabolic agents (OR = 1.658, 95% CI 1.573–1.748). Fig 1(a) shows a comparison of the cardiovascular agents based on percent of the RXD used stratified by each age group and weight category (underweight, normal weight, overweight, and obese. Fig 1(b) shows a comparison of the metabolic agents used stratified by age group and weight category. In both types of RXD, the obese population uses more of the RXD in the younger age groups, starting clearly at 25–34. The decrease in usage by obese individuals in the 75+ group may, in part, reflect the potential mixed effects of obesity in old age. Studies indicate that people with higher than normal BMI have lower mortality in the 75+ age group, though this is dependent on many other variables [36–38].

Table 2. Odds ratios between obese and non-obese group taking prescription drugs.

Prescription drugs	Obese (%)	Non-Obese (%)	Odds ratios	95% Wald CI		Pr > ChiSq
Cardiovascular agents	37.11	21.01	2.095	1.989	2.207	< .0001
Metabolic agents	28.02	16.86	1.658	1.573	1.748	< .0001
Central nervous system agents	23.91	17.61	1.187	1.126	1.252	< .0001
Gastrointestinal agents	14.21	9.62	1.277	1.199	1.36	< .0001
Respiratory agent	9.46	6.59	1.214	1.128	1.306	< .0001
Psychotherapeutic agents	15.16	10.13	1.304	1.226	1.387	< .0001
Hormones/hormone modifiers	15.16	13.76	0.877	0.827	0.93	< .0001
Topical agents	5.77	4.7	1.011	0.925	1.105	0.8118
Anti-infectives	5.5	5.56	0.795	0.729	0.868	< .0001
Coagulation modifiers	4.41	2.69	1.372	1.233	1.526	< .0001
Antineoplastics	1.48	1.46	0.83	0.705	0.977	0.0248

Open in a new tab

Fig 1 — CVD is cardiovascular agents. MA is metabolic agents. (a) Comparison of CVD RXD used by different age groups and weight categories. (b) Comparison of MA RXD used by different age groups and weight categories.

Correlation analysis of RDX and DS Use with demographic characteristics

Examining the correlation between reported RDX use and demographic characteristics for the non-obese and obese groups, we found a few items of interest. S2 Table shows the results of this regression analysis. Male from both the non-obese and the obese group were significantly less likely than female to use RXD (OR_control = 0.53, 95% CI 0.498–0.564, OR_obese = 0.569, 95% CI 0.52–0.623). Compared with those individuals with obesity older than 75, adults younger than 54 were significantly less likely to use RXD. When controlling for all other variables, non-obese people covered by insurance were around 1.597 times (p < .0001) as likely to use RXD than those who did not have any insurance coverage. Specifically, individuals with obesity covered by Medicare were 3.657 times more likely to use RXD (p < .0001) than those with no Medicare covered.

When looking at the correlation between reported DS use and demographic characteristics for the non-obese and obese groups, we found a few interesting findings (S3 Table). Within both groups, Mexican American and Non-Hispanic Black were significantly less likely to take DS compared with Non-Hispanic White. Individuals with obesity covered by private insurance, Medicare, and other government insurance were significantly more likely to take DS, while individuals without obesity covered by insurance were 1.411 times as likely to take DS than those who were not covered by any insurance while holding other variables constant.

We were also interested in how obesity status and demographic characteristics associated with the use RXD or that of DS within the whole population group (Table 3). Female and older people were more likely to take RXD or DS. PIR is an interesting factor, as people with higher PIR were significantly more likely to take DS. For RXD use, only those with PIR from 1 to 2 were found to be statistically significant. This was also the case with the PIR showing that those with a PIR higher than 1 are more likely to take DS. The higher the PIR is, the higher odds people use DS. Individuals with obesity were more likely to take RXD (OR = 1.565, 95% CI 1.482–1.654) while less likely to take DS (OR = 0.785, 95% CI 0.746–0.825) compared with individuals without obesity.

Table 3. Reported prescription drug use/dietary supplements use by demographic characteristics and obesity status correlates.

Variable	RXD use				DS use
Variable	Odds ratio	95% Wald CI		P value	Odds ratio	95% Wald CI		P value
Gender
Male	0.543	0.516	0.572	< .0001	0.574	0.548	0.602	< .0001
Female (reference)	1				1
Age Group
18–24	0.1	0.081	0.124	< .0001	0.256	0.219	0.301	< .0001
25–34	0.127	0.103	0.157	< .0001	0.349	0.299	0.408	< .0001
35–44	0.189	0.153	0.233	< .0001	0.397	0.341	0.463	< .0001
45–54	0.325	0.263	0.401	< .0001	0.534	0.459	0.621	< .0001
55–64	0.592	0.479	0.733	< .0001	0.802	0.689	0.933	0.0043
65–74	0.742	0.606	0.908	0.0039	0.885	0.779	1.006	0.0614
75 over (reference)	1				1
Race
Mexican American	0.442	0.4	0.487	< .0001	0.586	0.535	0.642	< .0001
Other Hispanic	0.536	0.476	0.605	< .0001	0.757	0.677	0.845	< .0001
Non-Hispanic White (reference)	1				1
Non-Hispanic Black	0.61	0.562	0.663	< .0001	0.589	0.545	0.636	< .0001
Other Race—Including Multi-Racial	0.542	0.489	0.6	< .0001	0.902	0.82	0.993	0.0345
Poverty Income Ratio
0–1 (reference)	1				1
1–2	0.913	0.842	0.989	0.0255	1.102	1.024	1.185	0.0097
2–3	0.972	0.888	1.064	0.5362	1.171	1.079	1.272	0.0002
3–4	0.914	0.83	1.006	0.066	1.325	1.213	1.448	< .0001
4–5	0.984	0.885	1.094	0.7621	1.511	1.371	1.666	< .0001
> = 5	1.058	0.968	1.157	0.2155	1.758	1.618	1.91	< .0001
Covered by Any Insurance
Yes	1.58	1.338	1.866	< .0001	1.21	1.069	1.369	0.0025
No (reference)	1				1
Covered by Private Insurance
Yes	1.38	1.181	1.612	< .0001	1.299	1.163	1.451	< .0001
No (reference)	1				1
Covered by Medicare
Yes	2.779	2.346	3.292	< .0001	1.245	1.097	1.413	0.0007
No (reference)	1				1
Covered by Medicaid
Yes	2.024	1.692	2.422	< .0001	0.819	0.714	0.94	0.0045
No (reference)	1
Covered by Other Government Insurance
Yes	1.935	1.634	2.291	< .0001	1.17	1.031	1.327	0.0149
No (reference)	1				1
Obesity
Yes	1.565	1.482	1.654	< .0001	0.785	0.746	0.825	< .0001
No (reference)	1				1

Open in a new tab

Fig 2 shows the correlation between age, BMI, and the number of RDX and DS used by both the obese group and non-obese group. Fig 2(a) illustrates the correlation between the average number of RXD used by a respondent and the age groups. Generally, the average number of RXD used by a respondent increased with an increase in age for both non-obese and obese groups. Regardless of age, individuals with obesity generally take more RXD than individuals without obesity. The distribution of the average number of RXD used shows a positive skewness distribution: the average number is greater than the median within each age group. Fig 2(b) illustrates the correlation between the average number of DS used and the age groups. Similarly, the average number of DS used generally increased with an increase in age for both groups, with people aging from 65 to 74 taking the highest number of DS (mean of non-obese vs. obese: 2.47 vs.2.09). However, the difference in average number of DS used between the obese and non-obese groups was not as clear as that between the number of RXD used and age group. Fig 2(c) illustrates the correlation between the number of RXD/DS used and BMI. BMI mainly clustered around 18 to 37 kg/m². People typically take higher numbers of RXD than DS. As the BMI categories from 83 to 87 kg/m² and from 128 to 132 kg/m² have only a few persons (three persons in 83 to 87 kg/m² category and one in 128 to 132 kg/m²) included in the sample, no bar shows in the figure, the points indicate the mean value among those groups.

Classification of RDX use and DS use using machine learning

S4–S8 Tables in the Supplementary Material show the detailed results of classifying DS and RXD use. Classification of DS use (binary variable) was not as accurate as classification of RXD use (binary variable). Results from the models run were similar for DS use with the highest overall accuracy being 64.6% and the AUROC at 0.7 (S4 Table). The results for RXD use were better, with the highest overall accuracy being 74.3% and the highest AUROC at 0.816 (S5 Table). To see if results would change with the addition of the variable DS use (for RXD use classification) and RXD use (for DS classification) the same machine learning algorithms were evaluated again. The overall accuracy and AUROC increased slightly in both DS models (S4 Table) and RXD models (S5 Table). This shows that adding the extra variable did not make a significant contribution to the classifications. Using the PIR in place of the annual household income also did not significantly change the results for classification of both DS use (S6 Table) and RDX use (S7 Table). For DS classification, RXD use, insurance status, and sex were the top three important features. For RXD classification, insurance status, DS use, and age were the top three important features. We further created four groups for RDX count (i.e., 0, 1–2, 3–5, >5) and classified samples into one of these groups, using the same models and variables. In this experiment (S8 Table), the best model is logistic regression with an AUROC of 0.76. Note that multiclass classification is inherently more challenging than binary classification.

Discussion

Individuals with obesity experience increased risk for developing chronic diseases across the lifespan, which are often managed using RXD. Overall, RXD use in the United States has increased [9], with the greatest increases seen in the treatment of conditions found to be associated with obesity, specifically antihypertensives, antihyperlipidemic, antidiabetics, and antidepressants [11]. Further, many individuals may use DS in addition to or instead of RXD. Given the increases in obesity, RXD use, and DS use, it is important to analyze trends in order to better characterize individuals RXD and DS use in relation to obesity.

In this study, we used NHANES data from 2003–2018 to examine RXD and DS use in relation to obesity status. We showed that demographics and obesity status do play a role in usage. In regard to demographic variables, we found that the obese group has a higher reported use of RXD at 63.68%. In both groups, females report higher use than males. In addition, RXD use increases with age in both groups. The difference in usage based on sex is explainable because females tend to have more consistent visits to medical practitioners and typically use more RXD in general, compared to males [39]. The increase in RDX use shown with age can be explained by increased prevalence in clinical comorbidities as the population gets older and is consistent with prior population studies [40]. The reason why the RDX usage decreases by obese individuals in the 75+ group may be that individuals whose obesity was associated with CVD earlier in life may have higher mortality rates. Another possibility is that higher BMI in older age may be protective, as previous research has suggested. Regardless, our findings further highlight that BMI is not as useful of a health parameter in older adults as it is in young and middle-aged adults, which is consistent with previous research [41]. The differences in race can be explained by the healthcare and insurance gap seen in minority races [42]. The increased use by those that are insured can be explained by an increased use of health care services and the associated increase coverage of RXD [43].

Regarding specific RXD types, cardiovascular agents and metabolic agents were used more by the obese group, while hormone/hormone modifiers and psychotherapeutic RXD usage was higher in the non-obese group. While increased use of cardiovascular and metabolic RXD was expected given the cardiometabolic comorbidities with obesity, higher usage of hormonal and psychotherapeutic RXD in non-obese individuals was surprising. Prior research has found higher rates of reproductive issues and increases in depression and other mental health disorders associated with obesity [44]; thus, we would expect higher use of related RXD. One explanation may be the association between high TSH and high BMI and low free-T4 and high BMI. These levels may not be outside of the “normal” range for these values but still cause an increase in BMI which could mean that RXD use would not necessarily be indicated [45]. In regard to psychotherapeutic RXD, it is possible that individuals with obesity may have unrecognized or undiagnosed mental health issues that are seen as medically related to obesity rather than mental health. For example, unhealthy coping mechanisms, such as increasing food intake or binging may be related to undiagnosed depressive symptomatology [46].

In looking at tracking the use of specific RXD prescribed for weight loss, there was a small proportion of the population that utilized these drugs. The problem with further study of these RXD is that many of them were approved outside of the 16-year survey data used in this project. Additionally, many of the drugs used for weight loss are also used for other purposes such as management of diabetes or as a general CNS stimulant. Based on the information available it was unclear why respondents used a specific RXD. This made it difficult to understand and analyze the use of these drugs.

When looking at both DS use and RXD use, the population with obesity was more likely to use RXD, but less likely to use DS compared with the non-obese population. This finding is consistent with a previous report that only 33.9% of adults use DS for weight loss [47]. Similar results were also obtained regardless of sex, age, and financial status [48]. However, we found additional novel predictors, such as insurance status. While higher RXD use in obese individuals was expected, lower use of DS was surprising. Given the low percentage of individuals who use DS for weight loss, it is possible that a majority of DS use is related to seeking other purported benefits. For example, there are a plethora of DS marketed toward enhancing brain function or enhancing physical function, which may be more likely to be consumed by either older or non-obese individuals who are already health-conscious.

The experiments with machine learning models showed that predictions of RXD and DS use could be improved. Future studies should examine alternative variables that may better predict RXD and DS use. For example, diagnostic information, out-of-pocket costs, and RXD coverage information may provide valuable additional information for understanding these relationships. Validated models of RXD and DS use would provide valuable information that may inform patient education. Further, physicians and other health care providers may be able to use this information to better understand patient trends and to make informed prescribing decisions. In particular, health care providers may benefit from being able to characterize individuals most likely to use DS, as these may go unreported otherwise.

Overall, our findings indicate that obesity is associated with higher RXD and lower DS use, but income and age are important demographic factors to consider in this relationship. Since it appears that the more used RXD types are associated with comorbid conditions related to obesity instead of treating obesity directly, there may be opportunities for better health education. This may include expanding education on the benefits of lifestyle changes to minimize both obesity and the impact of common comorbid conditions. Further, the lower observed use of DS in the obese population may offer an opportunity for patient education on DS like omega-3 fatty acids and Vitamin B/B-Complex, which are shown to benefit cardiovascular health, omega-3 fatty acids which are shown to benefit weight loss and to maintain blood sugar, and multivitamin/multiminerals which are shown to impact general overall health [49]. Understanding use patterns of DS in obese and non-obese individuals may also provide opportunities for educating patients about potentially harmful or ineffective DS, such as those targeting weight loss specifically [50].

Based on Figs 1 and 2(a), there seems to be a point in the 20–30 age range that would be ideal for addressing obesity proactively and aggressively before it escalates into having comorbidities that require RXD. Additionally, there is an opportunity for better education in those under 20 as studies show that obesity in childhood and adolescence can lead to obesity as an adult [51]. Education has shifted from education on treating obesity to education on prevention of obesity, given that losing small amounts of weight or maintaining a healthy weight are more effective than treating obesity once it has developed [52]. Future studies should examine how access to healthcare may further impact these relationships, especially in regards to accessing quality obesity prevention education.

Limitations

In most of the survey cycles used, there was no reason or diagnosis code associated with the report of RXD use. This means that we do not know why a respondent was taking a certain RXD. In addition, only primary category was used to type the various RXD. This means that if an RXD has multiple uses or off-label uses that cross-disease categories, it would not be evident in these results. Further, NHANES does not have information specifically geared towards delineating what may be considered a DS whose main purpose is weight loss. Another limitation is that NHANES is a cross-section survey. It is thus not possible to infer any casual relationships between the variables.

Future work

Recent releases of NHANES survey cycles (2013–2018) include diagnosis codes with the RXD information. A future project that contains diagnosis codes associated with RXD use would provide more insight. Even though national health surveys like NHANES provide enormous opportunities for answering many important health-related questions, they have not been used widely for patient education and informatics research except for a few studies of our own [53–55]. In future work, we plan to build informatics tools such as a data dashboard to visualize various types of analyses of the NHANES data to provide patients, policy makers, and health providers a way to explore RXD and DS use in the general population and certain population subgroups.

Conclusions

As obesity becomes a larger issue and the weight crisis in the United States becomes increasingly detrimental, more needs to be done to understand the overall health status associated with this population and how to educate the public about obesity, its comorbidities, and preventive measures that may help. Knowing how RXD and DS use are different from those without obesity is only the start. Further steps can be taken to understand why there are differences and how the underlying diseases and conditions can be pre-empted in this group. Further knowledge on the association between obesity, DS, and RXD can inform patient education with the help of informatics tools such as data dashboards. Developing models that help us understand the causes of and lifestyle changes needed to change obesity status should improve overall health in the United States.

Supporting information

S1 Table. NHANES variables.

(PDF)

Click here for additional data file.^{(89.4KB, pdf)}

S2 Table. Reported prescription drug use by demographic characteristics among obese and control group.

(PDF)

Click here for additional data file.^{(117.6KB, pdf)}

S3 Table. Reported dietary supplements use by demographic characteristics among obese and control group.

(PDF)

Click here for additional data file.^{(110.1KB, pdf)}

S4 Table. Performance of machine learning models for classifying DS use.

(PDF)

Click here for additional data file.^{(112.1KB, pdf)}

S5 Table. Performance of machine learning models for classifying RXD use.

(PDF)

Click here for additional data file.^{(111.7KB, pdf)}

S6 Table. Performance of machine learning models for classifying DS use using PIR.

(PDF)

Click here for additional data file.^{(112.8KB, pdf)}

S7 Table. Performance of machine learning models for classifying RXD use using PIR.

(PDF)

Click here for additional data file.^{(112.1KB, pdf)}

S8 Table. Performance of machine learning models for classifying RXD use into categories.

(PDF)

Click here for additional data file.^{(82.2KB, pdf)}

Data Availability

All the data files are available from the National Health and Nutrition Examination Survey (NHANES) database at https://www.cdc.gov/nchs/nhanes/index.htm.

Funding Statement

This study was supported in part by the National Institute on Aging (NIA) of the National Institutes of Health (NIH) under Award Number R21AG061431 (ZH); and the University of Florida Clinical and Translational Science Institute, which is supported in part by the NIH National Center for Advancing Translational Sciences under award number UL1TR001427 (ZH); and National Center for Complementary and Integrative Medicine (NCCIH) of NIH under Award Number R01AT009457 (RZ). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Ruban A, Stoenchev K, Ashrafian H, Teare J. Current treatments for obesity. Clin Med (Lond). 2019. May;19(3):205–12. doi: 10.7861/clinmedicine.19-3-205 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Yanovski SZ, Yanovski JA. Progress in pharmacotherapy for obesity. JAMA. 2021. Jul 13;326(2):129–30. doi: 10.1001/jama.2021.9486 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Gadde KM, Martin CK, Berthoud HR, Heymsfield SB. Obesity: pathophysiology and management. J Am Coll Cardiol. 2018. Jan 2;71(1):69–84. doi: 10.1016/j.jacc.2017.11.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Yanovski SZ, Yanovski JA. Long-term drug treatment for obesity: a systematic and clinical review. JAMA. 2014. Jan 1;311(1):74–86. doi: 10.1001/jama.2013.281361 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Li R, Serdula MK, Williamson DF, Bowman BA, Graham DJ, Green L. Dose-effect of fenfluramine use on the severity of valvular heart disease among fen-phen patients with valvulopathy. Int J Obes Relat Metab Disord. 1999. Sep;23(9):926–8. doi: 10.1038/sj.ijo.0801020 [DOI] [PubMed] [Google Scholar]
6.Li MF, Cheung BM. Rise and fall of anti-obesity drugs. World J Diabetes. 2011. Feb 15;2(2):19–23. doi: 10.4239/wjd.v2.i2.19 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Sharretts J, Galescu O, Gomatam S, Andraca-Carrera E, Hampp C, Yanoff L. Cancer risk associated with lorcaserin—the FDA’s review of the Camellia-Timi 61 trial. N Engl J Med. 2020. Sep 10;383(11):1000–2. doi: 10.1056/NEJMp2003873 [DOI] [PubMed] [Google Scholar]
8.Gahche JJ, Bailey RL, Potischman N, Ershow AG, Herrick KA, Ahluwalia N, et al. Federal monitoring of dietary supplement use in the resident, civilian, noninstitutionalized US population, National Health and Nutrition Examination Survey. J Nutr. 2018. Aug;148(Suppl 2):1436S–1444S. doi: 10.1093/jn/nxy093 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Kantor ED, Rehm CD, Haas JS, Chan AT, Giovannucci EL. Trends in prescription drug use among adults in the United States from 1999–2012. JAMA. 2015. Nov 3;314(17):1818–31. doi: 10.1001/jama.2015.13766 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Hruby A, Hu FB. The epidemiology of obesity: a big picture. Pharmacoeconomics. 2015. Jul;33(7):673–89. doi: 10.1007/s40273-014-0243-x [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Randhawa AK, Parikh JS, Kuk JL. Trends in medication use by body mass index and age between 1988 and 2012 in the United States. PLoS One. 2017. Sep 20;12(9). doi: 10.1371/journal.pone.0184089 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Stokes A, Lundberg DJ, Hempstead K, Berry KM, Baker JF, Preston SH. Obesity and incident prescription opioid use in the U.S., 2000–2015. Am J Prev Med. 2020. Jun;58(6):766–75. doi: 10.1016/j.amepre.2019.12.018 [DOI] [PubMed] [Google Scholar]
13.Khalil H, Ellwood L, Lord H, Fernandez R. Pharmacological treatment for obesity in adults: an umbrella review. Ann Pharmacother. 2020. Jul;54(7):691–705. doi: 10.1177/1060028019898912 [DOI] [PubMed] [Google Scholar]
14.Singh AK, Singh R. Pharmacotherapy in obesity: a systematic review and meta-analysis of randomized controlled trials of anti-obesity drugs. Expert Rev Clin Pharmacol. 2020. Jan;13(1):53–64. doi: 10.1080/17512433.2020.1698291 [DOI] [PubMed] [Google Scholar]
15.Squadrito F, Rottura M, Irrera N, Minutoli L, Bitto A, Barbieri MA, et al. Anti-obesity drug therapy in clinical practice: Evidence of a poor prescriptive attitude. Biomed Pharmacother. 2020. Aug;128. doi: 10.1016/j.biopha.2020.110320 [DOI] [PubMed] [Google Scholar]
16.Srivastava SB. Is there a prescription to treat pediatric obesity? Am J Lifestyle Med. 2020. Feb;14(1):36–9. doi: 10.1177/1559827619881095 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Uche UI, Suzuki S, Fulda KG, Zhou Z. Environment-wide association study on childhood obesity in the U.S. Environ Res. 2020. Dec;191. [DOI] [PubMed] [Google Scholar]
18.Doerrmann C, Oancea SC, Selya A. The association between hours spent at work and obesity status: results from NHANES 2015 to 2016. Am J Health Promot. 2020. May;34(4):359–65. doi: 10.1177/0890117119897189 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Liao S, Wu N, Gong D, Tang X, Yin T, Zhang H, et al. Association of aldehydes exposure with obesity in adults. Ecotoxicol Environ Saf. 2020. Sep 15;201. doi: 10.1016/j.ecoenv.2020.110785 [DOI] [PubMed] [Google Scholar]
20.Wu B, Jiang Y, Jin X, He L. Using three statistical methods to analyze the association between exposure to 9 compounds and obesity in children and adolescents: NHANES 2005–2010. Environ Health. 2020. Aug 31;19(1). doi: 10.1186/s12940-020-00642-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Zhang Y, Dong T, Hu W, Wang X, Xu B, Lin Z, et al. Association between exposure to a mixture of phenols, pesticides, and phthalates and obesity: Comparison of three statistical models. Environ Int. 2019. Feb;123:325–36. doi: 10.1016/j.envint.2018.11.076 [DOI] [PubMed] [Google Scholar]
22.Palmer MK, Toth PP. Trends in lipids, obesity, metabolic syndrome, and diabetes mellitus in the united states: an NHANES analysis (2003–2004 to 2013–2014). Obesity (Silver Spring). 2019. Feb;27(2):309–14. [DOI] [PubMed] [Google Scholar]
23.Kim D, Hou W, Wang F, Arcan C. Factors affecting obesity and waist circumference among us adults. Prev Chronic Dis. 2019. Jan 3;16. doi: 10.5888/pcd16.180220 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Hodish I. Insulin therapy, weight gain and prognosis. Diabetes, Obesity and Metabolism. 2018. Sep 1;20(9):2085–92. doi: 10.1111/dom.13367 [DOI] [PubMed] [Google Scholar]
25.NHANES Survey Methods and Analytic Guidelines [Internet]. [cited 2021 Mar 9]. https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx.
26.Gelfand A, Tangney CC. Analyses and interpretation of cannabis use among NHANES adults. European Journal of Preventive Cardiology. 2018. Jan 1;25(1):40–1. doi: 10.1177/2047487317736390 [DOI] [PubMed] [Google Scholar]
27.Yu X, Hao L, Crainiceanu C, Leroux A. Occupational determinants of physical activity at work: Evidence from wearable accelerometer in 2005–2006 NHANES. SSM Popul Health. 2022. Mar;17. doi: 10.1016/j.ssmph.2021.100989 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Casillas A, Liang LJ, Vassar S, Brown A. Culture and cognition-the association between acculturation and self-reported memory problems among middle-aged and older latinos in the National Health and Nutrition Examination Survey (NHANES), 1999 to 2014. J Gen Intern Med. 2022. Jan;37(1):258–60. doi: 10.1007/s11606-021-06608-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Perez-Lasierra JL, Casajus JA, González-Agüero A, Moreno-Franco B. Association of physical activity levels and prevalence of major degenerative diseases: Evidence from the national health and nutrition examination survey (NHANES) 1999–2018. Exp Gerontol. 2022. Feb;158. doi: 10.1016/j.exger.2021.111656 [DOI] [PubMed] [Google Scholar]
30.Wang L, Fu Z, Zheng J, Wang S, Ping Y, Gao B, et al. Exposure to perchlorate, nitrate and thiocyanate was associated with the prevalence of cardiovascular diseases. Ecotoxicol Environ Saf. 2022. Jan 6;230. doi: 10.1016/j.ecoenv.2022.113161 [DOI] [PubMed] [Google Scholar]
31.Dong L, Xie Y, Zou X. Association between sleep duration and depression in US adults: A cross-sectional study. J Affect Disord. 2022. Jan 1;296:183–8. doi: 10.1016/j.jad.2021.09.075 [DOI] [PubMed] [Google Scholar]
32.Xu Z, Gong R, Luo G, Wang M, Li D, Chen Y, et al. Association between vitamin D3 levels and insulin resistance: a large sample cross-sectional study. Sci Rep. 2022. Jan 7;12(1). doi: 10.1038/s41598-021-04109-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Centers for Disease Control and Prevention. Defining adult overweight and obesity [Internet]. Centers for Disease Control and Prevention. 2021 [cited 2022 Feb 3]. https://www.cdc.gov/obesity/adult/defining.html.
34.Suh E. Age of majority by state in 2022 [Internet]. Policygenius. 2021 [cited 2022 Feb 3]. https://www.policygenius.com/estate-planning/age-of-majority-by-state/.
35.Frank E, Hall M, Trigg L, Holmes G, Witten IH. Data mining in bioinformatics using Weka. Bioinformatics. 2004. Oct 12;20(15):2479–81. doi: 10.1093/bioinformatics/bth261 [DOI] [PubMed] [Google Scholar]
36.Stessman J, Jacobs JM, Ein-Mor E, Bursztyn M. Normal body mass index rather than obesity predicts greater mortality in elderly people: the Jerusalem longitudinal study. Journal of the American Geriatrics Society. 2009;57(12):2232–8. doi: 10.1111/j.1532-5415.2009.02567.x [DOI] [PubMed] [Google Scholar]
37.Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis. JAMA. 2013. Jan 2;309(1):71–82. doi: 10.1001/jama.2012.113905 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA. 2005. Apr 20;293(15):1861–7. doi: 10.1001/jama.293.15.1861 [DOI] [PubMed] [Google Scholar]
39.Ashman JJ, Hing E, Talwalkar A. Variation in physician office visit rates by patient characteristics and state, 2012. NCHS Data Brief. 2015. Sep;(212):1–8. [PubMed] [Google Scholar]
40.Martin CB, Hales CM, Gu Q, Ogden CL. Prescription drug use in the United States, 2015–2016 [Internet]. 2019 May [cited 2021 Aug 2]. Report No.: 334. http://www.cdc.gov/nchs/products/databriefs/db334.htm. [PubMed]
41.Batsis JA, Mackenzie TA, Bartels SJ, Sahakyan KR, Somers VK, Lopez-Jimenez F. Diagnostic accuracy of body mass index to identify obesity in older adults: NHANES 1999–2004. Int J Obes. 2016. May;40(5):761–7. doi: 10.1038/ijo.2015.243 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Kirby JB, Taliaferro G, Zuvekas SH. Explaining racial and ethnic disparities in health care. Med Care. 2006. May;44(5 Suppl):I64–72. doi: 10.1097/01.mlr.0000208195.83749.c3 [DOI] [PubMed] [Google Scholar]
43.Gu Q, Dillon CF, Burt VL. Prescription drug use continues to increase: U.S. prescription drug data for 2007–2008. National Center for Health Statistics, Centers for Disease Control and Prevention, U.S. Department of Health and Human Services; 2010 Sep. Report No.: 42.
44.de Wit L, Luppino F, van Straten A, Penninx B, Zitman F, Cuijpers P. Depression and obesity: a meta-analysis of community-based studies. Psychiatry Res. 2010. Jul 30;178(2):230–5. doi: 10.1016/j.psychres.2009.04.015 [DOI] [PubMed] [Google Scholar]
45.Knudsen N, Laurberg P, Rasmussen LB, Bülow I, Perrild H, Ovesen L, et al. Small differences in thyroid function may be important for body mass index and the occurrence of obesity in the population. J Clin Endocrinol Metab. 2005. Jul;90(7):4019–24. doi: 10.1210/jc.2004-2225 [DOI] [PubMed] [Google Scholar]
46.Wurtman RJ, Wurtman JJ. Brain serotonin, carbohydrate-craving, obesity and depression. Obesity Research. 1995;3(S4):477S–480S. [DOI] [PubMed] [Google Scholar]
47.Pillitteri JL, Shiffman S, Rohay JM, Harkins AM, Burton SL, Wadden TA. Use of dietary supplements for weight loss in the United States: Results of a national survey. Obesity. 2008;16(4):790–6. doi: 10.1038/oby.2007.136 [DOI] [PubMed] [Google Scholar]
48.Lubowiecki-Vikuk A, Król-Zielińska M, Kantanista A. Consumption of dietary supplements to support weight reduction in adults according to sociodemographic background, body mass index, waist-hip ratio, body fat and physical activity. Journal of Health, Population and Nutrition [Internet]. 2019. Nov 5 [cited 2021 Aug 2];38(1). Available from: doi: 10.1186/s41043-019-0191-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
49.He Z, Barrett LA, Rizvi R, Tang X, Payrovnaziri SN, Zhang R. Assessing the use and perception of dietary supplements among obese patients with National Health and Nutrition Examination Survey. AMIA Jt Summits Transl Sci Proc. 2020. May 30;2020:231–40. [PMC free article] [PubMed] [Google Scholar]
50.Office of Dietary Supplements—dietary supplements for weight loss [Internet]. [cited 2021 Aug 2]. https://ods.od.nih.gov/factsheets/WeightLoss-HealthProfessional/.
51.Simmonds M, Burch J, Llewellyn A, Griffiths C, Yang H, Owen C, et al. The use of measures of obesity in childhood for predicting obesity and the development of obesity-related diseases in adulthood: a systematic review and meta-analysis. Health Technol Assess. 2015. Jun;19(43). doi: 10.3310/hta19430 [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Gill T, King L, Caterson I. Obesity prevention: necessary and possible. A structured approach for effective planning. Proc Nutr Soc. 2005. May;64(2):255–61. doi: 10.1079/pns2005425 [DOI] [PubMed] [Google Scholar]
53.He Z, Wang S, Borhanian E, Weng C. Assessing the collective population representativeness of related type 2 diabetes trials by combining public data from clinicaltrials.gov and NHANES. Stud Health Technol Inform. 2015;216:569–73. [PMC free article] [PubMed] [Google Scholar]
54.He Z, Ryan P, Hoxha J, Wang S, Carini S, Sim I, et al. Multivariate analysis of the population representativeness of related clinical studies. J Biomed Inform. 2016. Apr;60:66–76. doi: 10.1016/j.jbi.2016.01.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
55.He Z, Charness N, Bian J, Hogan WR. Assessing the comorbidity gap between clinical studies and prevalence in elderly patient populations. IEEE EMBS Int Conf Biomed Health Inform. 2016. Feb;2016:136–9. doi: 10.1109/BHI.2016.7455853 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0269241.r001

Decision Letter 0

Jingjing Qian

3 Jan 2022

PONE-D-21-29526Assessing the Use of Prescription Drugs in Obese Respondents in the National Health and Nutrition Examination SurveyPLOS ONE

Dear Dr. He,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Feb 17 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Jingjing Qian

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf”

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

“This study was supported in part by the National Institute on Aging (NIA) of the National Institutes of Health (NIH) under Award Number R21AG061431; and the University of Florida Clinical and Translational Science Institute, which is supported in part by the NIH National Center for Advancing Translational Sciences under award number UL1TR001427; and National Center for Complementary and Integrative Medicine (NCCIH) of NIH under Award Number R01AT009457. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.”

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“This study was supported in part by the National Institute on Aging (NIA) of the National Institutes of Health (NIH) under Award Number R21AG061431 (ZH); and the University of Florida Clinical and Translational Science Institute, which is supported in part by the NIH National Center for Advancing Translational Sciences under award number UL1TR001427 (ZH); and National Center for Complementary and Integrative Medicine (NCCIH) of NIH under Award Number R01AT009457 (RZ). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors first conducted the secondary-data analyses based on selected variables from the NHANES (2003 ~ 2014) dataset, such as: demographic, physical examination, RXD, and insurance status to figure out the association between obesity status and usage of prescription drugs. Then, machine learning models have been utilized to predict the usage of prescription drugs for people with different backgrounds.

It is a good idea to develop a machine learning model based on the findings from a national-wide survey dataset. And here are my comments:

1. Please give reference for cut-off value on BMI and age (line: 84) and type of drugs on RXDs (line: 85).

2. Please give a clear description of the variables that have been used from NHANES. For example: does RXD represent the question (RXDUSE): “ have you taken or used any prescription medicines in the past month?” And what is DS corresponding to the variables in NHANES?

3. Since the average standard for annual income across states is different, please use the Poverty Income Ratio instead of Household Income, then run the regression and machine learning model to see if the conclusion (line: 167-168) still stands.

4. Line 51: “must”. Please give a reference if the author considers obese people must take RXDs to manage these conditions. Otherwise, without any disease symptoms, lifestyle changes, such as increasing the physics activities, dietary changes are more likely to be given by the clinic Doctors.

5. Line 85: 1937/32 is the unweighted sample size for respondents, however, in Table 1, the weighted sample sizes are given. So please be consistent. I would suggest using the unweighted sample size in Table 1 to describe the basic characteristics of the study population.

6. The quality of Figure 2 needs to be improved.

7. Line 79: please consider citing more recent work, since the plural “studies” is used.

Reviewer #2: This study of using National Health and Nutrition Examination Survey (NHANES) data to assess the use of prescription drugs in obese respondents is very interesting, but there are still some concerns for this study, which are shown below.

1. What’s the purpose and importance (or impact) to predict an individual’s likelihood of using any RXD or DS, instead of specific RXD or DS? In other words, how would clinicians and patients benefit from the high or low likelihoods of using any RXD or DS, as the possibility of needing any RXD or DS instead of specific RXD or DS does not deliver clear information for both clinicians and patients. In addition, since NHANES data also provide detailed specific types for DS, why did the authors only report the specific types of RXD, but not specific types for DS?

2. The NHANES data may not be an appropriate data for prediction purpose. As the NHANES is a cross-section survey data, we do not know whether the outcome of using any RXD or DS was before or after the obesity status. In this case, if most individuals already had RXD or DS before the obesity, the prediction model constructed based on the predictors that obtained after the outcome (using RXD or DS) will not provide valid prediction performance to apply for in the real-world settings.

3. As mentioned by authors, the most recent release of NHANES survey cycle is from 2017-2018, then why did this study only use the data from 2003-2014, excluding 2015-2018? In addition, one published study by Randhawa etc. also used NHANES data from 1988-2012 to examine the medication use by body mass index and age. Compared with Randhawa’s study, what new information and scientific findings could be added and need to be discussed?

4. The methods section lacks many details. For example, how did this study identify the use of specific RXD types, DS use, BMI results, demographic and insurance information from NHANES data?

5. The authors mentioned that they performed two separate logistic regression analyses, but in the paper, only the results from the second logistic regression were shown in table 3. The results from the first logistic regression may need to be added in the supplementary materials.

6. Result section, in table one, the authors only provided the weighted data, what about the unweighted raw sample size?

7. Result section, lines 145-148, “the decrease in usage by obese individuals in the 75+ group may, in part, reflect the potential mixed effects of obesity in old age, where those with a high BMI have a lower mortality…” It is hard to understand this statement for the relationship between the lower mortality and decreased CVD drugs as well as metabolic agents. If the authors meant older patients with high BMI were less likely to develop CVD or other diseases to have better survival, then how to explain increased trend of RXD and DS use was observed in Figure 2c where high BMI (≥30) had clearly higher counts of RXD or DS?

8. Result section, lines 201-202, the largest maximum number of DS used seems among people with BMI from 23-27 kg/m2, not 18 to 22 kg/m2 based on Figure 2c.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: You Lu

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2022 Jun 3;17(6):e0269241. doi: 10.1371/journal.pone.0269241.r002

Author response to Decision Letter 0

15 Feb 2022

Response to Reviews

We would like to sincerely thank the reviewers for their constructive feedback to our manuscript. We address their comments point by point as follows:

It is a good idea to develop a machine learning model based on the findings from a national-wide survey dataset. And here are my comments:

1. Please give reference for cut-off value on BMI and age (line: 84) and type of drugs on RXDs (line: 85).

Response: We added the references to the cut-off value on BMI and age. Note that we only included adult samples in this study. In this study, because the goal was to analyze the RXD use, we had to make sure that there is reported drug use information (no empty value for drug use) for these samples.

Response: A new Table S1 has been added to the Supplementary Materials to give the clear description of the variables used in NHANES.

Response: We extracted the PIR (poverty income ratio) from NHANES and re-ran both the regression and machine learning models using the PIR in place of the annual household income.

Response: We revised this sentence as “Many of these conditions require pharmaceutical intervention as part of the treatment plan and individuals with obesity often use RXDs to manage these conditions.”

Response: Table 1 has been updated to include both the weighted and unweighted sample size.

6. The quality of Figure 2 needs to be improved.

Response: High-resolution figure can be downloaded from the PDF file.

7. Line 79: please consider citing more recent work, since the plural “studies” is used.

Response: We have added more recent work using NHANES.

26. Gelfand A, Tangney CC. Analyses and interpretation of cannabis use among NHANES adults. European Journal of Preventive Cardiology. 2018 Jan 1;25(1):40–1.

27. Yu X, Hao L, Crainiceanu C, Leroux A. Occupational determinants of physical activity at work: Evidence from wearable accelerometer in 2005-2006 NHANES. SSM Popul Health. 2022 Mar;17.

28. Casillas A, Liang L-J, Vassar S, Brown A. Culture and cognition-the association between acculturation and self-reported memory problems among middle-aged and older latinos in the National Health and Nutrition Examination Survey (NHANES), 1999 to 2014. J Gen Intern Med. 2022 Jan;37(1):258–60.

29. Perez-Lasierra JL, Casajus JA, González-Agüero A, Moreno-Franco B. Association of physical activity levels and prevalence of major degenerative diseases: Evidence from the national health and nutrition examination survey (NHANES) 1999-2018. Exp Gerontol. 2022 Feb;158.

30. Wang L, Fu Z, Zheng J, Wang S, Ping Y, Gao B, et al. Exposure to perchlorate, nitrate and thiocyanate was associated with the prevalence of cardiovascular diseases. Ecotoxicol Environ Saf. 2022 Jan 6;230.

31. Dong L, Xie Y, Zou X. Association between sleep duration and depression in US adults: A cross-sectional study. J Affect Disord. 2022 Jan 1;296:183–8.

32. Xu Z, Gong R, Luo G, Wang M, Li D, Chen Y, et al. Association between vitamin D3 levels and insulin resistance: a large sample cross-sectional study. Sci Rep. 2022 Jan 7;12(1).

Response: We have changed the predictions to classifications. This can assist in informing both the patient and the practitioner about the likelihood of using RXDs based on demographics and BMI. Previous cited work covers the specifics of DS use in the population. The other reviewer commended us to use machine learning to classify DS and RXD use (“It is a good idea to develop a machine learning model based on the findings from a national-wide survey dataset.”) We revised the paper to justify the importance of classifying the likelihood of using RXD or DS as: “For example, diagnostic information, out-of-pocket costs, and RXD coverage information may provide valuable additional information for understanding these relationships. Validated models of RXD and DS use would provide valuable information that may inform patient education. Further, physicians and other health care providers may be able to use this information to better understand patient trends and to make informed prescribing decisions. In particular, health care providers may benefit from being able to characterize individuals most likely to use DS, as these may go unreported otherwise.”

Response: We have changed the predictions to classifications. Hence, we are not trying to predict RXD use but classify where someone is compared to others in the same group. We do not go into the causality of taking specific medications in this paper or how long someone has taken a specific RXD. We do mention that the use of specific RXD may be the result of being obese or may be the cause of increased weight gain leading to obesity.

Response: Randhawa et al. looked at changes in medication use over time (1988-2012) to see if obesity or age were factors. They focused significantly on the different drug categories. They focused more on the causality of the increase in RXD usage from 1988 to 2012. We used the data collectively instead of based on collection cycle. We examined multiple factors beyond just age and obesity status. While we touch on the different categories, overall, that was not the focus of the study. We did not seek to address the cause of the increase in RXD use, we were more interested in the comparison of the control group versus the non-control group over the different variables and the connection to the use of DS.

4. The methods section lacks many details. For example, how did this study identify the use of specific RXD types, DS use, BMI results, demographic and insurance information from NHANES data?

Response: A table indicating what fields from NHANES were used has been added to the Supplementary Materials Table S1.

Response: The result of the first regression can be found in Table S2 and Table S3 in the supplementary material. The result of the second regression can be found in Table 3 in the main text.

6. Result section, in table one, the authors only provided the weighted data, what about the unweighted raw sample size?

Response: Table 1 has been updated to include both the weighted and unweighted amounts.

Response: This section has been revised as “(In Results Section Page 10) The decrease in usage by obese individuals in the 75+ group may, in part, reflect the potential mixed effects of obesity in old age. Studies indicate that higher than normal BMI can have a lower mortality in the 75+ age group, though this is dependent on many other variables.37–39 ….... (In Discussion Section Page 18) The reason why the RDX usage decreases by obese individuals in the 75+ group may be that individuals whose obesity was associated with CVD earlier in life may have higher mortality rates. Another possibility is that higher BMI in older age may be protective, as previous research has suggested. Regardless, our findings further highlight that BMI is not as useful of a health parameter in older adults as it is in young and middle-aged adults, which is consistent with previous research.42”

8. Result section, lines 201-202, the largest maximum number of DS used seems among people with BMI from 23-27 kg/m2, not 18 to 22 kg/m2 based on Figure 2c.

Response: According to the raw data, the largest maximum number of DS used is among people with BMI from 18-22 kg/m2.

Attachment

Submitted filename: PLOS One Response to Reviewers.pdf

Click here for additional data file.^{(130.9KB, pdf)}

PLoS One. doi: 10.1371/journal.pone.0269241.r003

Decision Letter 1

Jingjing Qian

16 Mar 2022

PONE-D-21-29526R1Assessing the Use of Prescription Drugs in Obese Respondents in the National Health and Nutrition Examination SurveyPLOS ONE

Dear Dr. He,

Please submit your revised manuscript by Apr 30 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

We look forward to receiving your revised manuscript.

Kind regards,

Jingjing Qian

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments:

Dear authors, we appreciate your effort in addressing reviewers' comments. However, please directly respond to reviewer #2's comment regarding not adding newer NHANES 2015-2018 data in your manuscript. The current analysis using the 2003-2014 NHANES data is dated and might compromise implications of your findings.

In addition, I suggest you to further revise the title and/or abstract to reflect the objective of the paper. Your objective was to examine the use of prescription drugs and dietary supplements by the individuals with obesity, but the current title and abstract (results section) only focus on use of prescription drugs. The aspect of dietary supplements is missing in title and abstract (results section). Please also double check the results in the abstract to reflect the most up to date data based on the revision. Thanks.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

6. Review Comments to the Author

Reviewer #1: (No Response)

Reviewer #2: The authors addressed almost all comments, except for comment 3 which was not fully addressed. The authors may need to update the analysis including the most recent data cycles (2015-2018) which are available now. In addition, the Figure 2C is hard to tell the statement "the largest maximum number of DS used is among people with BMI from 18-22 kg/m2", so this figure may need some modification and improvement.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

PLoS One. 2022 Jun 3;17(6):e0269241. doi: 10.1371/journal.pone.0269241.r004

Author response to Decision Letter 1

25 Apr 2022

Thanks for your feedback! We have addressed the comments point-by-point as follows.

Journal Requirements:

***** Response: We have reviewed the reference list to make sure it is complete and correct.

Additional Editor Comments:

***** Response: We have added the data of NHANES survey cycles 2015-2016, 2017-2018 in the revised version. We have re-done all the analyses and updated the results accordingly.

***** Response: We revised the title and abstract to reflect the objective of the paper.

***** Response: We have added the data of NHANES survey cycles 2015-2016, 2017-2018 in the revised version. We have re-done all the analyses and updated the results accordingly. For better clarity, we removed the two sentences “The largest maximum number of DS used (24) can be found among people with BMI from 18 to 22 kg/m2. The highest average number of RXD used can be found among people with BMI from 63 to 67 kg/m2.”

Attachment

Submitted filename: Response to Review_PLOSONE.docx

Click here for additional data file.^{(14.1KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0269241.r005

Decision Letter 2

Jingjing Qian

18 May 2022

Assessing the Use of Prescription Drugs and Dietary Supplements in Obese Respondents in the National Health and Nutrition Examination Survey

PONE-D-21-29526R2

Dear Dr. He,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Jingjing Qian

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Thank you for further revising your manuscript by incorporating new years of data.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #2: (No Response)

**********

6. Review Comments to the Author

Reviewer #2: The authors may need to double check the submitted Table S2 in their supporting information. The numbers are not consistent with the text in the manuscript 'RXD ( = 0.53, 95% CI 0.498-0.564, = 0.569, 95% CI 0.52-0.623).' It seems the authors still used the old table S2, as the numbers are same as the numbers deleted in the track-changing version.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

PLoS One. doi: 10.1371/journal.pone.0269241.r006

Acceptance letter

Jingjing Qian

23 May 2022

PONE-D-21-29526R2

Assessing the Use of Prescription Drugs and Dietary Supplements in Obese Respondents in the National Health and Nutrition Examination Survey

Dear Dr. He:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Jingjing Qian

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Table. NHANES variables.

(PDF)

Click here for additional data file.^{(89.4KB, pdf)}

S2 Table. Reported prescription drug use by demographic characteristics among obese and control group.

(PDF)

Click here for additional data file.^{(117.6KB, pdf)}

S3 Table. Reported dietary supplements use by demographic characteristics among obese and control group.

(PDF)

Click here for additional data file.^{(110.1KB, pdf)}

S4 Table. Performance of machine learning models for classifying DS use.

(PDF)

Click here for additional data file.^{(112.1KB, pdf)}

S5 Table. Performance of machine learning models for classifying RXD use.

(PDF)

Click here for additional data file.^{(111.7KB, pdf)}

S6 Table. Performance of machine learning models for classifying DS use using PIR.

(PDF)

Click here for additional data file.^{(112.8KB, pdf)}

S7 Table. Performance of machine learning models for classifying RXD use using PIR.

(PDF)

Click here for additional data file.^{(112.1KB, pdf)}

S8 Table. Performance of machine learning models for classifying RXD use into categories.

(PDF)

Click here for additional data file.^{(82.2KB, pdf)}

Attachment

Submitted filename: PLOS One Response to Reviewers.pdf

Click here for additional data file.^{(130.9KB, pdf)}

Attachment

Submitted filename: Response to Review_PLOSONE.docx

Click here for additional data file.^{(14.1KB, docx)}

Data Availability Statement

All the data files are available from the National Health and Nutrition Examination Survey (NHANES) database at https://www.cdc.gov/nchs/nhanes/index.htm.

[pone.0269241.ref001] 1.Ruban A, Stoenchev K, Ashrafian H, Teare J. Current treatments for obesity. Clin Med (Lond). 2019. May;19(3):205–12. doi: 10.7861/clinmedicine.19-3-205 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref002] 2.Yanovski SZ, Yanovski JA. Progress in pharmacotherapy for obesity. JAMA. 2021. Jul 13;326(2):129–30. doi: 10.1001/jama.2021.9486 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref003] 3.Gadde KM, Martin CK, Berthoud HR, Heymsfield SB. Obesity: pathophysiology and management. J Am Coll Cardiol. 2018. Jan 2;71(1):69–84. doi: 10.1016/j.jacc.2017.11.011 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref004] 4.Yanovski SZ, Yanovski JA. Long-term drug treatment for obesity: a systematic and clinical review. JAMA. 2014. Jan 1;311(1):74–86. doi: 10.1001/jama.2013.281361 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref005] 5.Li R, Serdula MK, Williamson DF, Bowman BA, Graham DJ, Green L. Dose-effect of fenfluramine use on the severity of valvular heart disease among fen-phen patients with valvulopathy. Int J Obes Relat Metab Disord. 1999. Sep;23(9):926–8. doi: 10.1038/sj.ijo.0801020 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref006] 6.Li MF, Cheung BM. Rise and fall of anti-obesity drugs. World J Diabetes. 2011. Feb 15;2(2):19–23. doi: 10.4239/wjd.v2.i2.19 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref007] 7.Sharretts J, Galescu O, Gomatam S, Andraca-Carrera E, Hampp C, Yanoff L. Cancer risk associated with lorcaserin—the FDA’s review of the Camellia-Timi 61 trial. N Engl J Med. 2020. Sep 10;383(11):1000–2. doi: 10.1056/NEJMp2003873 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref008] 8.Gahche JJ, Bailey RL, Potischman N, Ershow AG, Herrick KA, Ahluwalia N, et al. Federal monitoring of dietary supplement use in the resident, civilian, noninstitutionalized US population, National Health and Nutrition Examination Survey. J Nutr. 2018. Aug;148(Suppl 2):1436S–1444S. doi: 10.1093/jn/nxy093 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref009] 9.Kantor ED, Rehm CD, Haas JS, Chan AT, Giovannucci EL. Trends in prescription drug use among adults in the United States from 1999–2012. JAMA. 2015. Nov 3;314(17):1818–31. doi: 10.1001/jama.2015.13766 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref010] 10.Hruby A, Hu FB. The epidemiology of obesity: a big picture. Pharmacoeconomics. 2015. Jul;33(7):673–89. doi: 10.1007/s40273-014-0243-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref011] 11.Randhawa AK, Parikh JS, Kuk JL. Trends in medication use by body mass index and age between 1988 and 2012 in the United States. PLoS One. 2017. Sep 20;12(9). doi: 10.1371/journal.pone.0184089 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref012] 12.Stokes A, Lundberg DJ, Hempstead K, Berry KM, Baker JF, Preston SH. Obesity and incident prescription opioid use in the U.S., 2000–2015. Am J Prev Med. 2020. Jun;58(6):766–75. doi: 10.1016/j.amepre.2019.12.018 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref013] 13.Khalil H, Ellwood L, Lord H, Fernandez R. Pharmacological treatment for obesity in adults: an umbrella review. Ann Pharmacother. 2020. Jul;54(7):691–705. doi: 10.1177/1060028019898912 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref014] 14.Singh AK, Singh R. Pharmacotherapy in obesity: a systematic review and meta-analysis of randomized controlled trials of anti-obesity drugs. Expert Rev Clin Pharmacol. 2020. Jan;13(1):53–64. doi: 10.1080/17512433.2020.1698291 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref015] 15.Squadrito F, Rottura M, Irrera N, Minutoli L, Bitto A, Barbieri MA, et al. Anti-obesity drug therapy in clinical practice: Evidence of a poor prescriptive attitude. Biomed Pharmacother. 2020. Aug;128. doi: 10.1016/j.biopha.2020.110320 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref016] 16.Srivastava SB. Is there a prescription to treat pediatric obesity? Am J Lifestyle Med. 2020. Feb;14(1):36–9. doi: 10.1177/1559827619881095 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref017] 17.Uche UI, Suzuki S, Fulda KG, Zhou Z. Environment-wide association study on childhood obesity in the U.S. Environ Res. 2020. Dec;191. [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref018] 18.Doerrmann C, Oancea SC, Selya A. The association between hours spent at work and obesity status: results from NHANES 2015 to 2016. Am J Health Promot. 2020. May;34(4):359–65. doi: 10.1177/0890117119897189 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref019] 19.Liao S, Wu N, Gong D, Tang X, Yin T, Zhang H, et al. Association of aldehydes exposure with obesity in adults. Ecotoxicol Environ Saf. 2020. Sep 15;201. doi: 10.1016/j.ecoenv.2020.110785 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref020] 20.Wu B, Jiang Y, Jin X, He L. Using three statistical methods to analyze the association between exposure to 9 compounds and obesity in children and adolescents: NHANES 2005–2010. Environ Health. 2020. Aug 31;19(1). doi: 10.1186/s12940-020-00642-6 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref021] 21.Zhang Y, Dong T, Hu W, Wang X, Xu B, Lin Z, et al. Association between exposure to a mixture of phenols, pesticides, and phthalates and obesity: Comparison of three statistical models. Environ Int. 2019. Feb;123:325–36. doi: 10.1016/j.envint.2018.11.076 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref022] 22.Palmer MK, Toth PP. Trends in lipids, obesity, metabolic syndrome, and diabetes mellitus in the united states: an NHANES analysis (2003–2004 to 2013–2014). Obesity (Silver Spring). 2019. Feb;27(2):309–14. [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref023] 23.Kim D, Hou W, Wang F, Arcan C. Factors affecting obesity and waist circumference among us adults. Prev Chronic Dis. 2019. Jan 3;16. doi: 10.5888/pcd16.180220 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref024] 24.Hodish I. Insulin therapy, weight gain and prognosis. Diabetes, Obesity and Metabolism. 2018. Sep 1;20(9):2085–92. doi: 10.1111/dom.13367 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref025] 25.NHANES Survey Methods and Analytic Guidelines [Internet]. [cited 2021 Mar 9]. https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx.

[pone.0269241.ref026] 26.Gelfand A, Tangney CC. Analyses and interpretation of cannabis use among NHANES adults. European Journal of Preventive Cardiology. 2018. Jan 1;25(1):40–1. doi: 10.1177/2047487317736390 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref027] 27.Yu X, Hao L, Crainiceanu C, Leroux A. Occupational determinants of physical activity at work: Evidence from wearable accelerometer in 2005–2006 NHANES. SSM Popul Health. 2022. Mar;17. doi: 10.1016/j.ssmph.2021.100989 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref028] 28.Casillas A, Liang LJ, Vassar S, Brown A. Culture and cognition-the association between acculturation and self-reported memory problems among middle-aged and older latinos in the National Health and Nutrition Examination Survey (NHANES), 1999 to 2014. J Gen Intern Med. 2022. Jan;37(1):258–60. doi: 10.1007/s11606-021-06608-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref029] 29.Perez-Lasierra JL, Casajus JA, González-Agüero A, Moreno-Franco B. Association of physical activity levels and prevalence of major degenerative diseases: Evidence from the national health and nutrition examination survey (NHANES) 1999–2018. Exp Gerontol. 2022. Feb;158. doi: 10.1016/j.exger.2021.111656 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref030] 30.Wang L, Fu Z, Zheng J, Wang S, Ping Y, Gao B, et al. Exposure to perchlorate, nitrate and thiocyanate was associated with the prevalence of cardiovascular diseases. Ecotoxicol Environ Saf. 2022. Jan 6;230. doi: 10.1016/j.ecoenv.2022.113161 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref031] 31.Dong L, Xie Y, Zou X. Association between sleep duration and depression in US adults: A cross-sectional study. J Affect Disord. 2022. Jan 1;296:183–8. doi: 10.1016/j.jad.2021.09.075 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref032] 32.Xu Z, Gong R, Luo G, Wang M, Li D, Chen Y, et al. Association between vitamin D3 levels and insulin resistance: a large sample cross-sectional study. Sci Rep. 2022. Jan 7;12(1). doi: 10.1038/s41598-021-04109-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref033] 33.Centers for Disease Control and Prevention. Defining adult overweight and obesity [Internet]. Centers for Disease Control and Prevention. 2021 [cited 2022 Feb 3]. https://www.cdc.gov/obesity/adult/defining.html.

[pone.0269241.ref034] 34.Suh E. Age of majority by state in 2022 [Internet]. Policygenius. 2021 [cited 2022 Feb 3]. https://www.policygenius.com/estate-planning/age-of-majority-by-state/.

[pone.0269241.ref035] 35.Frank E, Hall M, Trigg L, Holmes G, Witten IH. Data mining in bioinformatics using Weka. Bioinformatics. 2004. Oct 12;20(15):2479–81. doi: 10.1093/bioinformatics/bth261 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref036] 36.Stessman J, Jacobs JM, Ein-Mor E, Bursztyn M. Normal body mass index rather than obesity predicts greater mortality in elderly people: the Jerusalem longitudinal study. Journal of the American Geriatrics Society. 2009;57(12):2232–8. doi: 10.1111/j.1532-5415.2009.02567.x [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref037] 37.Flegal KM, Kit BK, Orpana H, Graubard BI. Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis. JAMA. 2013. Jan 2;309(1):71–82. doi: 10.1001/jama.2012.113905 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref038] 38.Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity. JAMA. 2005. Apr 20;293(15):1861–7. doi: 10.1001/jama.293.15.1861 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref039] 39.Ashman JJ, Hing E, Talwalkar A. Variation in physician office visit rates by patient characteristics and state, 2012. NCHS Data Brief. 2015. Sep;(212):1–8. [PubMed] [Google Scholar]

[pone.0269241.ref040] 40.Martin CB, Hales CM, Gu Q, Ogden CL. Prescription drug use in the United States, 2015–2016 [Internet]. 2019 May [cited 2021 Aug 2]. Report No.: 334. http://www.cdc.gov/nchs/products/databriefs/db334.htm. [PubMed]

[pone.0269241.ref041] 41.Batsis JA, Mackenzie TA, Bartels SJ, Sahakyan KR, Somers VK, Lopez-Jimenez F. Diagnostic accuracy of body mass index to identify obesity in older adults: NHANES 1999–2004. Int J Obes. 2016. May;40(5):761–7. doi: 10.1038/ijo.2015.243 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref042] 42.Kirby JB, Taliaferro G, Zuvekas SH. Explaining racial and ethnic disparities in health care. Med Care. 2006. May;44(5 Suppl):I64–72. doi: 10.1097/01.mlr.0000208195.83749.c3 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref043] 43.Gu Q, Dillon CF, Burt VL. Prescription drug use continues to increase: U.S. prescription drug data for 2007–2008. National Center for Health Statistics, Centers for Disease Control and Prevention, U.S. Department of Health and Human Services; 2010 Sep. Report No.: 42.

[pone.0269241.ref044] 44.de Wit L, Luppino F, van Straten A, Penninx B, Zitman F, Cuijpers P. Depression and obesity: a meta-analysis of community-based studies. Psychiatry Res. 2010. Jul 30;178(2):230–5. doi: 10.1016/j.psychres.2009.04.015 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref045] 45.Knudsen N, Laurberg P, Rasmussen LB, Bülow I, Perrild H, Ovesen L, et al. Small differences in thyroid function may be important for body mass index and the occurrence of obesity in the population. J Clin Endocrinol Metab. 2005. Jul;90(7):4019–24. doi: 10.1210/jc.2004-2225 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref046] 46.Wurtman RJ, Wurtman JJ. Brain serotonin, carbohydrate-craving, obesity and depression. Obesity Research. 1995;3(S4):477S–480S. [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref047] 47.Pillitteri JL, Shiffman S, Rohay JM, Harkins AM, Burton SL, Wadden TA. Use of dietary supplements for weight loss in the United States: Results of a national survey. Obesity. 2008;16(4):790–6. doi: 10.1038/oby.2007.136 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref048] 48.Lubowiecki-Vikuk A, Król-Zielińska M, Kantanista A. Consumption of dietary supplements to support weight reduction in adults according to sociodemographic background, body mass index, waist-hip ratio, body fat and physical activity. Journal of Health, Population and Nutrition [Internet]. 2019. Nov 5 [cited 2021 Aug 2];38(1). Available from: doi: 10.1186/s41043-019-0191-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref049] 49.He Z, Barrett LA, Rizvi R, Tang X, Payrovnaziri SN, Zhang R. Assessing the use and perception of dietary supplements among obese patients with National Health and Nutrition Examination Survey. AMIA Jt Summits Transl Sci Proc. 2020. May 30;2020:231–40. [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref050] 50.Office of Dietary Supplements—dietary supplements for weight loss [Internet]. [cited 2021 Aug 2]. https://ods.od.nih.gov/factsheets/WeightLoss-HealthProfessional/.

[pone.0269241.ref051] 51.Simmonds M, Burch J, Llewellyn A, Griffiths C, Yang H, Owen C, et al. The use of measures of obesity in childhood for predicting obesity and the development of obesity-related diseases in adulthood: a systematic review and meta-analysis. Health Technol Assess. 2015. Jun;19(43). doi: 10.3310/hta19430 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref052] 52.Gill T, King L, Caterson I. Obesity prevention: necessary and possible. A structured approach for effective planning. Proc Nutr Soc. 2005. May;64(2):255–61. doi: 10.1079/pns2005425 [DOI] [PubMed] [Google Scholar]

[pone.0269241.ref053] 53.He Z, Wang S, Borhanian E, Weng C. Assessing the collective population representativeness of related type 2 diabetes trials by combining public data from clinicaltrials.gov and NHANES. Stud Health Technol Inform. 2015;216:569–73. [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref054] 54.He Z, Ryan P, Hoxha J, Wang S, Carini S, Sim I, et al. Multivariate analysis of the population representativeness of related clinical studies. J Biomed Inform. 2016. Apr;60:66–76. doi: 10.1016/j.jbi.2016.01.007 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269241.ref055] 55.He Z, Charness N, Bian J, Hogan WR. Assessing the comorbidity gap between clinical studies and prevalence in elderly patient populations. IEEE EMBS Int Conf Biomed Health Inform. 2016. Feb;2016:136–9. doi: 10.1109/BHI.2016.7455853 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Assessing the use of prescription drugs and dietary supplements in obese respondents in the National Health and Nutrition Examination Survey

Laura A Barrett

Aiwen Xing

Julia Sheffler

Elizabeth Steidley

Terrence J Adam

Rui Zhang

Zhe He

Roles

Abstract

Introduction

Methods

Results

Conclusions

Introduction

Materials and methods

Data analysis

Basic characteristics

Statistical analysis

Classification modeling

Results

Basic characteristics

Table 1. Basic characteristics of the study population.

Specific RXD types

Table 2. Odds ratios between obese and non-obese group taking prescription drugs.

Fig 1. Comparison of RXD used by age group and weight category.

Correlation analysis of RDX and DS Use with demographic characteristics

Table 3. Reported prescription drug use/dietary supplements use by demographic characteristics and obesity status correlates.

Fig 2. (a) The correlation between the number of RXD used and age; (b) The correlation between the number of DS used and age; (c) the number of RXD/DS used compared to BMI.

Classification of RDX use and DS use using machine learning

Discussion

Limitations

Future work

Conclusions

Supporting information

Data Availability

Funding Statement

References

Decision Letter 0

Jingjing Qian

Roles

Author response to Decision Letter 0

Decision Letter 1

Jingjing Qian

Roles

Author response to Decision Letter 1

Decision Letter 2

Jingjing Qian

Roles

Acceptance letter

Jingjing Qian

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases