Comparing the Study Populations in Dietary Supplement and Drug Clinical Trials for Metabolic Syndrome and Related Disorders

Abstract

Clinical trials are essential in exploring the safety and efficacy of a new intervention. However, restrictive eligibility criteria pose recruitment challenges that could prolong study durations and reduce study generalizability to the real-world population. The objective of this study is to compare the study populations of dietary supplement (DS) and drug trials on metabolic syndrome related conditions. Using the COMPACT database, we retrieved the DS and drug trials related to metabolic syndrome and performed aggregate analyses on the study populations with respect to various quantitative eligibility criteria. We also extracted and compared baseline characteristics, both quantitative and qualitative, of recruited patients in completed trials. We found similarities and differences in baseline characteristics of enrolled patients between drug and DS clinical trials on metabolic syndrome-related conditions. This comparative aggregate analysis is an initial step towards improving patient recruitment efficacy and population representativeness for clinical trials across conditions and intervention types.

Introduction

In the United States, the use of dietary supplements (DS), including vitamins, minerals, botanical extracts, and protein powders has dramatically increased in recent years. The 2017 Council for Responsible Nutrition (CRN) Consumer Survey shows that more than 170 millions (76%) U.S. adults are taking DS, and overall health and wellness are the main reason for DS use.¹ The Food and Drug Administration (FDA) and Federal Trade Commission (FTC) require “competent and reliable scientific evidence” to substantiate health claims of DS products about their benefits and safety.² Clinical trials are considered as the “gold standards” in clinical research and are executed to validate efficacy and safety of new medical, surgical, and behavioral interventions among humans.³ Thus, designing and conducting clinical trials is a vital step for generating scientific evidence related to the intended health claims of a new intervention, including DS products. Success of a clinical trial largely depends upon recruitment and retention of a study population of an adequate sample size. However, patient recruitment often encounters many barriers, including the lack of health literacy,⁴ limited awareness of clinical trials,⁵ age limitations,⁶ trials restraints,⁷ and restrictive inclusion and exclusion criteria⁵ among various other reasons. According to recent reports, recruitment difficulties caused delays from 1 to 6 months for 86% of clinical trials, with the remaining 14% experiencing even longer delays.⁸ Finding the targets for qualifying and enrolling patients can be a demanding process, both in terms of time and money. There are published studies that have investigated and compared clinical trials across diseases and intervention types, primarily for drug trials.⁹ However, no studies have specifically compared the study populations between trials on DS and drugs. Understanding study population design in completed and ongoing clinical trials between DS and drug intervention types could help us better understand the differences and similarities among the participant traits on key study features.

Among existing registries that provide easy access to information on publicly and privately funded clinical studies, ClinicalTrial.gov is the largest human clinical study registry holding registrations of over 289,000 research studies in all the 50 states of the United States and in 205 countries (as of November 2018).¹⁰ ClinicalTrials.gov was established by the U.S. National Library of Medicine (NLM) of the National Institutes of Health (NIH), in collaboration with the FDA. It is a comprehensive, reliable, one-stop information resource for wide range of users (e.g., patients, clinicians, researchers) on key elements pertaining to clinical trials registered to date. Specifically, under the DS category, there are 9,468 clinical trials on ClinicalTrials.gov, larger than any other World Health Organization (WHO) registry network members. Since all the clinical trials related data are readily accessible, this database offers great opportunities for knowledge reuse and retrospective analysis on trial design through extraction and analysis of relevant data, e.g., purpose of the study and study population (e.g., sample size, eligibility criteria).¹¹ Compared to journal publications that mainly report positive results from completed studies, ClinicalTrials.gov contains the study information for all kinds of studies, regardless of their statuses of completeness and final outcomes. Therefore, we opted to use ClinicalTrials.gov as the data source in order to conduct a comprehensive comparison between drug and DS trials.

Metabolic syndrome, also known as syndrome X, insulin resistance syndrome, or dysmetabolic syndrome, is a cluster of metabolic disorders contributing to major public health concerns both in the US and around the globe.¹² According to a recent study, one third of all US adults met criteria for metabolic syndrome.¹² Changes in life style, body weight reduction, healthy diet, and regular exercises, are considered the first line of defense against metabolic syndrome.¹³ A diet including DS is often considered to be healthier, more natural, and free of adverse effects as opposed to their drug counterparts. DS is more commonly used among people with at least one of the risk factors for metabolic syndrome and its related disorders.¹⁴ However, to our best knowledge, there is no study comparing the main aspects of clinical trials of DS vs. drugs, among metabolic syndrome-related medical conditions.

The objective of this study is to conduct a comparative analysis on the study populations of DS and drugs clinical trials, pertaining to study metadata, recruitment eligibility criteria, and baseline characteristics of enrolled patients for the metabolic syndrome and its related conditions including hypertension, Type 2 diabetes mellitus (T2DM), hyperlipidemia, and obesity/overweight, using the data collected from ClinicalTrials.gov. The knowledge gained from this study could facilitate study design in future studies through more efficient and effective participant selection. It would also increase the transparency of patient selection for clinical trials on the disease level and help understand the systematic biases in patient selection.

Background

The COMPACT database for ClinicalTrials.gov

To support the aggregate analysis of clinical trials, we have previously built the COMPACT database which includes structured meta-data and eligibility criteria of all the trials on ClinicalTrials.gov.¹¹ In the COMPACT database, the quantitative eligibility criteria with a permissible value range, e.g., “patients with HbA1c < 8%”, were parsed with a numeric expression parsing tool ValX.¹⁵ The structured expression [‘HbA1C’, “<”, “8”, “%”] consists of a numeric variable, a comparison operator, a threshold value, and a measurement unit. ValX not only structures the numeric expression, but also unifies the measurement units and converts exclusion criteria to inclusion criteria, enabling aggregate analysis of a variable across multiple studies. In the evaluation of Valx,¹⁵ the precision, recall, and F-measure for extracting numeric expressions with the quantitative feature “HbA1c” were 99.6%, 98.1%, 98.8% for Type 1 diabetes mellitus (T1DM) trials, and 98.8%, 96.9%, 97.8% for T2DM trials, respectively. The results of the corresponding measures for extracting numeric expressions with “glucose” were 97.3%, 94.8%, 96.1% for T1DM trials, and 92.3%, 92.3%, 92.3% for Type 2 diabetes trials, respectively. COMPACT contains structured metadata, readily analyzable eligibility criteria in the studies of ClinicalTrials.gov. Using COMPACT database as the backend, we developed a web-based tool VITTA,¹⁶ which allows its users to flexibly select a set of trials of the same disease and profile their study population with respect to one quantitative eligibility criterion at a time. The current version of the COMPACT database is based on all the clinical study summaries in ClinicalTrials.gov downloaded on September 26, 2016. To index the trials by medical conditions, we used the “Condition” table of the Aggregate Analysis of Clinical Trial (AACT) database of Duke University,¹⁷ which uses MeSH term to index trials.

Methods

Retrieving clinical trials for metabolic syndrome and extracting their structured eligibility criteria

Using the COMPACT database, we retrieved all the drug and DS clinical trials on hypertension, T2DM, hyperlipidemia, obesity/overweight, and metabolic syndrome. Note that we used the local COMPACT database, which has a newer data from ClinicalTrials.gov than the one used by the web-based VITTA tool (http://is.gd/VITTA). We extracted study population information with respect to structured quantitative eligibility criteria from these retrieved clinical trials using the COMPACT database.

Extracting baseline characteristics of enrolled patients in clinical trials

In ClinicalTrials.gov, we searched for completed DS and drug clinical trials on various metabolic syndrome-related conditions that have reported baseline characteristics of their enrolled patients. We only focused on trials with a primary purpose of prevention or treatment. Only obesity and hyperlipidemia have more than 10 such trials using the intervention of a drug or a DS, respectively. For these trials, we manually extracted the baseline characteristics of the enrolled patients. As only a small portion of trials reported results in ClinicalTrials.gov, the enrolled patients in this analysis represent merely a convenience sample. For each study, we extracted both quantitative and qualitative data, i.e., enrollment counts, race, mean and standard deviation (SD) for age, and gender. For hyperlipidemia trials, we also extracted the baseline for low-density lipoprotein cholesterol.

Data Analysis

In this work, we performed comparative analyses between drug trials and DS trials on the following three aspects.

• (I)Metadata: We analyzed the number of clinical trials for each of the metabolic syndrome-related conditions, stratified by interventional type (DS vs. drugs), study type (interventional vs. observational), primary purpose (treatment vs. prevention) and endpoint classifications (safety, efficacy or both, i.e., safety/efficacy).
• (II)Study population with respect to frequently used quantitative eligibility criteria: We visualized the percentage of trials over permissible values of each of the quantitative eligibility criteria that are used by more than 15% of trials on the respective condition. For example, in obesity trials, body mass index (BMI) is one of the most frequently used eligibility criteria (used by 64.7% of obesity trials). Thus, for the variable BMI, we analyzed the percentage of DS trials on obesity that allow patients with BMI of each possible value between 12 and 60 kg/m². We performed a similar analysis for BMI in drug trials on obesity and compared the result with that of the DS trials. The comparison would reveal the difference in the study population between drug and DS trials with respect to BMI. Note that if an obesity trial does not use the BMI criterion, we assume it allows patients with any BMI value.
• (III)
  Baseline characteristics of enrolled patients: We compared the baseline characteristics of enrolled patients in the completed trials on obesity and hyperlipidemia. In ClinicalTrials.gov, for most baseline characteristics, the mean and standard deviation (SD) values were reported. We aggregated the mean and SD for each quantitative variable using the following formulas (adapted from¹⁸), where T is the number of studies:

  $$Weighted\_mean=\frac{{\displaystyle {\sum }_{i=1}^T(mea{n}_i*number\_participant{s}_i)}}{{\displaystyle {\sum }_{i=1}^{T}number\_participant{s}_i}}$$ (1)

  $$Weighted\_SD=\frac{{\displaystyle {\sum }_{i=1}^T(S{D}_i^2*(number\_participant{s}_i-1))}}{{\displaystyle {\sum }_{i=1}^T(number\_participant{s}_i-1)}}$$ (2)

Results

We analyzed clinical trials from three main perspectives: (I) yearly distribution of clinical trials for metabolic syndrome and its related conditions, (II) eligibility criteria and their permissible value ranges used in trials for each condition, and (III) the differences of the baseline measures of the enrolled patients between drug trials and DS trials. We provide the detailed results of these aforementioned perspectives as follows.

I. Metadata analysis of DS and drug trials

Table 1 shows the summary of the metadata of DS and drug trials on metabolic syndrome-related conditions with a start date between 1983 and 2016. For both DS and drug trials, there were substantially more interventional studies than observational studies. Across all the conditions, both DS and drug trials were more treatment driven. Regarding the endpoint classification, a higher proportion of drug trials focused on both safety and efficacy, whereas a higher proportion of DS trials focused only on efficacy. In contrast, only a small proportion of trials, especially for DS, focused only on safety.

Table 1.

Summary of metadata for dietary supplement trials and drug trials related to metabolic syndromes

Condition	Intervention name (n)	Study type		Primary Purpose			Endpoint Classification
		Interventional	Observational	Treatment	Prevention	Basic Science	Safety/ Efficacy	Efficacy	Safety
Hypertension	DS (n=89)	89 (100%)	0 (0%)	44 (49.4%)	24 (26.97%)	8 (8.99%)	13 (14.61%)	45 (50.56%)	3 (3.37%)
	Drug (n=1,158)	1072 (92.57%)	85 (7.34%)	906 (78.2%)	69 (5.96%)	23 (1.99%)	495 (42.75%)	264 (22.80%)	33 (2.85%)
T2DM	DS (n=177)	175 (98.87%)	2 (1.13%)	78 (44.07%)	39 (22.03%)	29 (16.38%)	25 (14.12%)	96 (54.24%)	1 (0.56%)
	Drug (n=2,548)	2368 (92.94%)	176 (6.91%)	2020 (79.28%)	57 (2.24%)	157 (6.16%)	1212 (47.57%)	433 (16.99%)	203 (7.97%)
Hyperlipidemia	DS (n=48)	47 (97.92%)	1 (2.08%)	22 (45.83%)	9 (18.75%)	6 (12.50%)	15 (31.25%)	20 (41.57%)	0 (0%)
	Drug (n=182)	177 (97.25%)	4 (2.20%)	146 (80.22%)	15 (8.24%)	6 (3.30%)	100 (54.95%)	35 (19.23%)	2 (1.10%)
Obesity	DS (n=445)	440 (98.88%)	4 (0.9%)	169 (37.98%)	125 (28.09%)	90 (20.22%)	76 (17.08%)	224 (50.34%)	4 (0.90%)
	Drug (n=734)	721 (98.23%)	12 (1.63%)	502 (68.39%)	53 (7.22%)	82 (11.17%)	260 (35.42%)	194 (26.43%)	31 (4.22%)
Metabolic syndrome	DS (n=137)	135 (98.54%)	2 (1.46%)	53 (38.69%)	47 (34.31%)	21 (15.33%)	27 (19.71%)	65 (47.45%)	1 (0.73%)
	Drug (n=200)	197 (98.50%)	3 (1.50%)	147 (73.50%)	15 (7.50%)	17 (8.50%)	77 (38.50%)	83 (41.50%)	5 (2.50%)
All trials	DS (n=5,961)	5857 (98.26%)	96 (1.61%)	2270 (38.08%)	1510 (25.33%)	871 (14.61%)	1238 (20.77%)	2645 (44.37%)	131 (2.20%)
	Drug (n =99,244)	95670 (96.40%)	3184 (3.21%)	77337 (77.9%)	5853 (5.9%)	3711 (3.74%)	42989 (43.32%)	21070 (21.23%)	8698 (8.76%)

^$A small number of studies were categorized as patient registry and expanded access and are not included in the analysis. ⊥ This includes all conditions associated with metabolic syndrome diagnosis, i.e., hypertension, T2DM, hyperlipidemia, obesity and metabolic syndrome.

Since there were substantially fewer studies started before year 2000, we plotted the data on number of studies after stratifying it by intervention type and study purpose (treatment vs. prevention), between February 2000 (when ClinicalTrials.gov was implemented) and September 2016 as shown in Figure 1. This would help us better understand the more recent trends in clinical trials for various conditions related to metabolic syndrome. The number of drugs trials exceeded DS trials for all the conditions under study. There is a large difference for hypertension and T2DM but a much smaller difference for metabolic syndrome and obesity. The number of drug trials showed a noticeable increase between years 2000-2008, followed by a downward trend especially for metabolic syndrome. In contrast to drugs trials, DS trials, specifically for metabolic syndrome and obesity, showed an upward trend. We also observed that DS peaks followed drug peaks, indicating DS research and development progress often follows drug R&D.

Figure 1.

Number of drug and DS clinical trials for various conditions related to metabolic symdrome between years 2000 and 2016 stratified by intervention type and study prupose.

II. Quantitative eligibility criteria and their permissible numerical ranges under each condition

We also performed a comparative analysis of the study popluation for drug and DS trials clinical trials under conditions associated with metabolic syndrome and for the same time period (1983-2016). We analyzed the percentage of trials over permissable value ranges of frequently used quantitative eligibility criteria (used by more than 15% of trials), as shown in Table 2 and Figure 2-4.

Table 2.

Top quantitative eligibility criteria used in clinical trials for the conditions included in our study.

Conditions	Eligibility criteria (% trials)
Hypertension	Age (100%)	Systolic (37.5%)	Diastolic (32.5%)	BMI (15.8%)
T2DM	Age (100%)	HbA1c (54.7%)	BMI (50.4 %)	Glucose (19.2%)
Hyperlipidemia	Age (100%)	BMI (29.2%)	Triglycerides (16.9%)	LDL Cholesterol (14.1%)
Obesity	Age (100%)	BMI (64.7%)	Weight (12.2%)	Glucose (11.1%)
Metabolic syndrome	Age (100%)	BMI (45.1%)	Glucose (42.6%)	Triglycerides (27.1%)

Figure 2.

Percentage of drugs and DS clinical trials conducted for hypertension, T2DM, hyperlipidemia, obesity and metabolic syndrome over (a) permissible age values and (b) permissible BMI values.

Figure 4.

Percentages of drug and DS trials conducted for (a) metabolic syndrome over permissible blood glucose levels, (b) hypertension over permissible systolic and diastolic blood pressure values, (c) hyperlipidemia over permissible triglyceride level, (d) metabolic syndrome over permissible triglyceride level.

Age and BMI: Figure 2 shows the percentage of clinical trials for hypertension, T2DM, hyperlipidemia, obesity/overweight and metabolic syndromes over permissible values of (a) age, and (b) BMI. X-axis represents the value of the variable. Y-axis represents the percentage of trials for a condition that allow patients with a certain value of the variable. We stratified the analysis by condition and intervention types (drugs vs. DS). Patients between 17-40 years of age are generally more acceptable in drug trials corresponding to all conditions and DS trials for only hypertension and obesity. In other words, more DS trials allow patients with age range between 40-65 than those between 17-40, whereas there is no noticeable difference between these two age groups in drug trials, except obesity. For both DS trials and drug trials, the number of trials that allow older adults (>= 65 years old) decreases quickly as the age increases. For BMI, except for obesity, a higher percentage of drug trials allow patients with BMI between 18-45 kg/m². Patients with BMI 18-25 kg/m² are less accepted for trials on obesity than for other conditions.

Hemoglobin A1c (HbA1c) and blood glucose levels: Figure 3(a) and 3(b) show the percentages of trials conducted for T2DM over permissible HbA1c values (%) and blood glucose levels (mg/dL), respectively. For HbA1c, DS trials are more lenient for HbA1c values <= 8% or >= 12% than drug trials. For glucose, a slightly lower percentage of DS trials allows patients with glucose <= 100 mg/dL than drug trials. Figure 4(a) shows the percentage of trials conducted for metabolic syndrome over permissible blood glucose levels. For glucose, not much difference was observed between DS and drug trials.

Figure 3.

Percentages of drug and DS clinical trials conducted for T2DM over permissible (a) HbA1c and (b) glucose levels.

Systolic and diastolic blood pressure (BP): Figure 4(b) shows the percentage of trials conducted for hypertension over permissible systolic BP and diastolic BP. The curves for the DS trials are slightly shifted to the left of those for the drug trials while keeping the similar shapes, indicating that DS trials focused on patients who had a slightly lower blood pressure and were thus healthier than those in drug trials. We also observed that, drug trials as compared to DS trials, considered a broader range of clinical severity, i.e., higher diastolic (120 vs. 110 mm Hg) and systolic (200 vs. 180 mm Hg) BP values.

Blood triglycerides levels: Figure 4(c) shows the percentage of trials for hyperlipidemia over permissible triglyceride values (mg/dL). The two curves are similar, indicating that the same study population is being sought for drug and DS trials, with respect to triglyceride values. Figure 4(d) shows the percentage of trials for metabolic syndrome over permissible triglyceride values. The trends of curves for drug and DS trials are quite similar, indicating that trials on both types of interventions have the similar study population w.r.t. triglyceride levels.

III. Comparison of baseline characteristics of enrolled patients between drug and DS trials

Table 3 gives the comparison of the baseline information of the recruited patients in drug and DS trials for obesity and hyperlipidemia. Overall, the patients recruited for obesity trials are younger than hyperlipidemia trials. For obesity trials, more female patients were recruited than male patients. For hyperlipidemia trials, more male patients were recruited than female patients. For both obesity and hyperlipidemia trials, the patients enrolled in drug trials are older than those in DS trials with statistical significance (two-sample t-test P < 0.0001), indicating that older adults might have been underrepresented in these DS trials.

Table 3.

Baseline characteristics of the recruited patients in drug and DS trials for obesity and hyperlipidemia

Type of Trials		Obesity trials		Hyperlipidemia trials
Baseline Characteristics		Drug trials (# of trials)	DS trials (# of trials)	Drug trials (# of trials)	DS trials (# of trials)
Age		49.31 ± 9.86 (n=94)	39.02 ± 9.56 (n=16)	59.18 ± 10.02 (n=146)	52.33 ± 7.95 (n=5)
Gender	Female	66.5% (n=109)	69.6% (n=19)	41.5% (n=175)	42.1% (n=6)
	Male	34.9% (n=109)	30.4% (n=19)	58.4% (n=175)	57.9% (n=6)
LDL (mg/dl)		--	--	129.64 ± 36.90 (n=55)	143.40 ± 27.83 (n=2)

Table 4 shows the comparison of the races of the recruited patients in DS and drug trials for obesity and hyperlipidemia. For obesity, the DS trials recruited a higher proportion of black or African American than in drug trials (chi-square test P < 0.05). For hyperlipidemia, black or African American are much underrepresented in drug trials.

Table 4.

Races of the recruited patients in drug and DS trials for obesity and hyperlipidemia

Race	American Indian/Alaska Native	Asian	Native Hawaiian/Other Pacific Islander	Black or African American	White	More than one race	Unknown/ Not Reported
Trial set (Number of Trials)
Obesity DS trials (n=2)	0%	0%	0%	35.71%	64.29%	0%	0%
Obesity drug trials (n=33)	1.25%	1.12%	0.27%	18.59%	74.19%	0.44%	3.58%
Hyperlipidemia drug trials (n=60)	0.18%	11.80%	0.16%	4.82%	78.76%	0.62%	4.39%

Discussion

In this study, we analyzed the data extracted from ClinicalTrials.gov, both for DSs and drugs, to understand the ongoing trends, permissible value ranges of quantitative eligibility criteria as well as characteristics of enrolled patients. It provides a number of pertinent contributions to our understanding of metabolic syndrome-related clinical trials including a broad picture of study distribution, inclusion criteria, enrollment and subject characteristics both for DS and drug studies. This information can provide researchers and clinicians a high-level understanding of DS and drug trials to help identify research gaps and inform future study planning. Substantial work has been done around aggregate meta-analysis on existing data of drug trials, both at macro and micro levels, in order to better understand and estimate an overall effect across similar studies.¹⁹ Such studies are mainly conducted on conditions with high prevalence in the U.S. and around the world, e.g., T2DM,²⁰ oncology,²¹ and fall prevention.²² For the macro level analysis, elements were extracted from various trial data resources (e.g., registries, journal publications, systematic reviews) and analyzed for the change in study characteristics over time and other important elements such as randomization.²³ Such descriptive analysis is also performed at the micro level on other key study elements such as study design, eligibility criteria, endpoints, and adverse events.²⁴

This study provides important findings about DS and drug clinical trials conducted on highly prevalent metabolic syndrome-related conditions. These conditions increase the risk of more serious conditions, e.g., coronary artery disease, stroke, chronic kidney disease.¹² Overall, substantially fewer trials were found to evaluate the effectiveness of DS than drugs. Both DS and drug trials focused more on treatment than prevention. A substantial number of preventive DS trials were conducted for obesity, which has been recognized as a disease by the American Medical Association in 2013.²⁵ This is partly due to the false belief that DSs are “natural” products with no or relatively fewer severe adverse effects than their drug counterparts.^26,27 We also observed that a higher proportion of drug trials focused on both safety and efficacy, whereas a higher proportion of DS trials focused only on efficacy. This is in part as a result of the DS regulatory policy not only in the U.S. but also globally with lower requirements for pre- marketing approval.²⁸ More aligned regulatory requirements for DS approval as usual drug approval may help to enhance our understanding of DS safety and efficacy.

We observed a consistent increase in trial registration after the implementation of ClinicalTrials.gov in February 2000, followed by other international efforts, such as the policies of the International Committee of Medical Journal Editors (ICMJE) and WHO, Food and Drug Administration Amendments Act (FDAAA), Declaration of Helsinki (DoH) and NPRM Notice of Proposed Rulemaking (NPRM).²⁹ After 2007, there was a gradual decline of trial registration except for DS trials on obesity and metabolic syndrome, possibly due to the changes in eligibility criteria needed to make the diagnoses of metabolic syndrome after 2008³⁰ or introductions of international regulatory actions, which posed greater scrutiny on weight loss studies due to safety issues.³¹ The notable upward trend observed for trials on obesity and metabolic syndromes may be due to a greater focus on these conditions with growing prevalence and public demand for therapies after withdrawals of some previously available medications.³²

Variability in permissible value ranges for commonly used eligibility criteria, primarily age, BMI, systolic/diastolic BP and HbA1c was observed between DS and drugs trials. In particular, for diabetes and hypertension, the key physiologic inclusion parameters were less clinically constrained for DS studies than for drug studies. Visualizing them can help us better understand the similarities and differences between DS and drugs trials and help guide future clinical study design. Changes in the definitions in the clinical guidelines over time and differences between professional societies may in part affect the disease identification as well as the treatment goals.³³ With respect to the age for study population of the trials for the metabolic syndromes, we found a more even distribution of drug trials for ages ranging from 17 to 65 years, whereas the primary study population of dietary supplement trials was more aggregated around a population of 40-65 years old, presumably with an increasing health awareness, development of clinical disease states and a greater interest in disease prevention. For both DS trials and drug trials, the number of trials that allow older adults (>= 65 years old) decreases quickly as the age increases, possible being a precautionary measure to prevent any drug interactions among older populations with complex comorbidities, age- related physiological changes and other health issues. This also shows a systematic underrepresentation of older adults in both types of trials. Children are also underrepresented in trials, possibly due to the ethical, legal, and regulatory complexities in the enrollment, consent, and appropriate access of children of minor parents to clinical research.³⁴ A higher percentage of obesity trials on both DS and drugs allows patients with BMI >= 35 kg/m², mainly Class 2 (BMI 35 - 40 kg/m²) and Class 3 (BMI >= 40 kg/m²).³⁵

Baseline characteristics of enrolled patients on obesity and hyperlipidemia were compared on completed intervention trials for dietary supplements vs. drugs. Younger participants were recruited in dietary supplement trials vs. drug trials for both obesity (drug: 49.31 ± 9.86 vs. DS: 39.02 ± 9.56) and hyperlipidemia (drug: 59.18 ± 10.02 vs. DS: 52.33 ± 7.95). Interestingly, hyperlipidemia trials had more middle-aged male participants in spite of no well- established gender predilection in terms of hyperlipidemia prevalence and perception across high-risk age groups, e.g., postmenopausal women and diabetics. However, this gender trend may reflect the higher male cardiovascular disease risk. The underrepresentation of older adults in these DS trials are likely due to the insufficient evidence on DS safety, adverse effects and interactions especially among older patients/consumers. However, for obesity, more female patients were enrolled in obesity trials (drug trials: 66.5% vs. DS trials: 69.6%) than in hyperlipidemia trials (drug trials: 41.5% vs. DS trials: 42.1%). These results correlate with the higher prevalence of obesity among women than men. Women are more conscious about their weight and more likely to report interference of their body image with everyday social activities and greater participation in weight loss exercises.³⁶ The comparison of the racial makeup of recruited patients in DS and drug trials for obesity and hyperlipidemia showed substantial underrepresentation of black or African American as compared to whites despite the high prevalence of both conditions among them.³⁷

This study has certain limitations. First, we have not accessed the data on the outcomes of these trials nor the design issues of the trials. Second, we used the “Condition” table of the AACT database.¹⁷ There might be a small number of condition indexing errors. For the aggregate analysis of the trial metadata and eligibility criteria, some trials may not have been primarily conducted to treat or prevent the respective condition but merely test the patients with that condition. Nevertheless, we manually reviewed a sample of trials and found this case to be rare in our dataset, thus unlikely to affect our findings. Third, we may have missed a number of trials that started before the regulatory changes of trial registration took place. Fourth, there might be some discrepancies between the results reported in ClinicalTrials.gov and journal publications.³⁸

Conclusions

In this study, we performed a comparative analysis on metabolic syndrome-related DS and drug clinical trials documented on ClinicalTrials.gov regarding their overall trends, eligibility criteria and baseline characteristics of actual enrolled patient. In general, we observed a marked increase in number of clinical trials (both drugs and DS) from 2000 to 2007 followed by a gradual decline after 2008. The number of drugs trials, primarily therapeutic, exceeds DS trials for conditions like hypertension, T2DM and hyperlipidemia, whereas DS trials, both preventive and therapeutic, are more prevalent for obesity and metabolic syndrome reflecting a high and growing prevalence. Between drug and DS trials, some quantitative eligibility criteria (e.g., age, BMI, A1c, blood pressure) show variations, while others are much similar (e.g., blood glucose, triglycerides). Relatively younger participants were enrolled in DS than in drug trials in completed obesity and hyperlipidemia trials. Obesity trials recruited more young and female subjects, compared to hyperlipidemia. Both drug and DS trials show substantial underrepresentation of black or African American subjects. Further efforts in this area would help us understand disease-specific study populations of DS and drug trials for evidence-based patient selection design, leading to more efficient participant recruitment, a higher chance of successful trial execution, and a higher study generalizability.