Economic and equity evaluation of age restrictions on over-the-counter diet pills and muscle-building supplements

Abstract

Over-the-counter diet pills and muscle-building supplements are linked to increased eating disorder diagnoses, especially among youth. With limited regulatory oversight, minors may unknowingly consume harmful substances leading to other adverse effects. Massachusetts has proposed restricting sales to individuals under 18 years. However, concerns about health equity and unintended consequences arise when proposing new policies. We conducted a cost-effectiveness analysis of the proposed age-restriction policy compared to the status quo, focusing on 2 closed cohorts of males and females aged 0-17 years in Massachusetts over a 30-year time horizon. We evaluated the impact from both societal and health systems’ perspectives and further assessed equity implications by modeling 3 racial/ethnic subgroups. The policy is projected to prevent 57 034 eating disorder cases and over 46 000 additional adverse medical events (eg, liver injuries). It would yield 51 749 quality-adjusted life years and generate healthcare savings of $14 million and societal savings of $30 million annually. The Latine subpopulation would see the highest per capita health benefits followed by Black and White residents, respectively. Restricting the sale of these supplements to minors offers both health and economic benefits. These findings underscore the policy's effectiveness, fiscal responsibility, and positive equity impacts, providing confidence for policymakers and the public.

Introduction

Eating disorders (ED) account for over 10 000 deaths, 33 000 hospital admissions, and over $64 billion in costs to the economy annually in the United States.¹ These psychiatric conditions greatly influence individuals’ health-related quality of life. Considering that more than half receive their first diagnosis of eating disorders by the age of 18 years, primary prevention plays a crucial role in reducing their prevalence and alleviating the economic burdens they impose.¹

Over-the-counter (OTC) diet pills, which include the OTC drug orlistat and myriad weight-loss supplements, and muscle-building dietary supplements (MBS) face little regulation and are prospectively linked with eating disorder diagnoses and other adverse medical events (AEs), including organ failure, heart attacks, and the use of illegal muscle-building drugs such as anabolic-androgenic steroids (AAS).^2-5 Studies have suggested racial and ethnic disparities in the use of OTC diet pills, muscle-building supplements, and anabolic steroids, with some studies reporting disproportionately high prevalence among Black/African American and Latine/Hispanic individuals.^6-8 Disparities are likely driven in part by industry marketing practices that target racially minoritized populations.^9-11

Limited federal regulatory oversight of dietary supplements puts youth at risk of unknowingly consuming banned substances and hazardous chemicals.^2,12 Restructuring accessibility to OTC diet pills and muscle-building supplements via an age-restriction policy, therefore, has the potential to reduce the upward trend in ED among youth through young adulthood. Several US states, including Massachusetts (MA), have proposed age restrictions to ban the sale of OTC diet pills and muscle-building supplements to minors under 18 years.¹³ The proposed legislation aims to regulate these products purchased within the respective state to behind-the-counter status, a regulatory measure akin to those already implemented for harmful substances such as tobacco and pseudoephedrine. The state of New York recently passed a similar bill into law, which became effective in April 2024.¹⁴

Critiques of health economics¹⁵ call for a greater awareness of unintentional widening of health inequities by health interventions. Distributional cost-effectiveness analysis (DCEA) is a methodology used to assess the differential effects on various segments of a population.^16,17 This approach extends the traditional cost-effectiveness analysis (CEA) by considering not only the overall efficiency and effectiveness of an intervention but also how its costs and benefits are distributed among subgroups within the population.

We conducted a CEA of the age-restriction policy aimed at preventing ED and other adverse medical events among youth in MA. Subsequently, we performed an aggregate DCEA to explore health equity implications based on the key equity variable of race/ethnicity.

Study design and data sources

This study was conducted to estimate economic and equity implications of the proposed age-restriction law for the health and financial wellbeing of all who live in MA. We compared the costs and effectiveness with the existing status quo, where no age-restriction law is in place. This study adhered to the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) 2022.¹⁸

Logic model

The logic model informing this study involves 2 main pathways (Figure 1). The first pathway posits that the implementation of the age-restriction policy would lead to fewer youths buying and using OTC diet pills, resulting in a reduction in the incidence of ED and other AEs. The second pathway depicts fewer youth buying and using muscle-building supplements, resulting in a reduction in immediate AEs and a reduction in youth initiating use of AAS, which leads to additional AEs avoided.

Study population

Using U.S. Census data from 2023, 2 total population cohorts were created for individuals aged 0 to 17 years old among the MA population—1 for females and 1 for males (Table 1). It was important to account for gender differences in the use of OTC diet pills, MBS, and AAS. We modeled these groups as closed cohorts over a 30-year period. The comparator is the same population under the status quo, that is, no age-restriction policy is implemented.

Intervention effectiveness

The effectiveness of the proposed policy on use of OTC diet pills and muscle-building supplements was estimated using parallel evidence from previous studies on age-restriction policies.^20,21

Health outcomes

We estimated the health outcomes of this study along 2 pathways (Figure 1). Outcomes along the OTC diet pill pathway included number of ED avoided, number of other (non-ED) AEs prevented post-OTC diet pill use, and the gains in quality-adjusted life years (QALYs). The outcomes resulting from the MBS pathway included the number of AEs avoided post-MBS use, AEs avoided post-AAS use, and QALYs saved.

For the OTC diet pill pathway, we calculated the expected total number of the 4 most prevalent eating disorders (anorexia nervosa, bulimia nervosa [BN], binge eating disorder [BED], and other specified feeding and eating disorder [OSFED]) among our cohorts after 30 years (Table 1 and Table S1). Since the entire population is covered by the intervention, the number of incremental ED avoided was estimated as the intervention effectiveness times this expected total. Our first sensitivity analysis included an additional modest reduction in severity for the children that started with an ED, but no OTC diet pill use, as a portion of them would not get access to OTC diet pills they otherwise might have, which could then improve their 30-year morbidity. The values for the distribution for this assumption was based on Ward et al.'s findings regarding persons with ED relapse (Table 1).²² Our second sensitivity analysis (estimating a reduced cost-effectiveness) included only 2 of the 4 ED, BN, and OSFED.

Further, we estimated the probability of AEs (other than ED) post-OTC diet pill over 30 years. In refining our search, the most frequent AEs post-OTC diet pill use were tachycardia, hypertension, gastrointestinal pain, depression, and liver injury (Table 1). In our sensitivity analysis estimating a greater cost-effectiveness, we expanded this list to add suicidality. Our second sensitivity analysis (estimating a reduced cost-effectiveness) removed the AEs related to males as that pathway contained the greatest amount of uncertainty.

For the MBS pathway, we estimated the overall anticipated incidence of AEs linked to the use of muscle-building supplements (immediate) and subsequent AAS after 30 years. Drawing on the work conducted by Ganson et al.,⁴³ we applied an overall risk for any AEs following the use of MBS, used Or and colleagues’ findings to refine our estimates to focus on risk of “severe” AEs, and bundled them into a single, weighted, incremental health outcome.¹² This outcome included 6 components: disability, death, life-threatening events, hospitalizations, emergency room visits, and/or interventions to prevent disability.¹²

Additionally, we included the probability of experiencing the 5 most common AEs post-AAS use: cardiovascular disease, male infertility, liver injury, anxiety, and depression (Table 1).^56-68 For the first sensitivity analysis, we expanded this list to 7 to include heart attack and female infertility.^52,72

We calculated the QALYs for each ED, considering their average duration^23-28 multiplied by the health-related quality of life scores from Long et al.²¹ Similarly, for each of the AEs we multiplied their average duration (Tables S2 and S3) by the disability weights obtained from the Global Burden of Disease study.³⁴

Costs and cost-effectiveness

For the OTC diet pill pathway, we derived our estimate of the total costs associated with ED from Long et al. (Table 1).²¹ The total costs encompassed expenses related to vendors, ED-specific healthcare costs in outpatient and inpatient settings, emergency department visits, diagnostic tests, transportation, as well as indirect costs such as caregivers’ opportunity costs. We multiplied this per-person total by our study population.

For all other AEs prevented (in both the OTC diet pill pathway and the MBS pathway), we estimated healthcare cost savings associated with the reduction of AEs resulting from decreased use of OTC diet pills, MBS, or AAS over the 30-year period (Table 1).

We then estimated the intervention costs as a per person per year amount to account for impacted vendors. Finally, we looked at the incremental cost-effectiveness ratio (ICER) by dividing the incremental costs by the incremental QALYs gained.

Statistical analysis

The CEA was conducted from both a societal perspective and a health systems perspective with a time horizon of 30 years, 2023-2053. The societal perspective included various indirect costs where available, for example, opportunity costs of family caregivers. All costs were adjusted to 2023 US dollars; all QALYs and dollars were discounted by the standard rate of 3%.

We used Monte Carlo simulation (10 000 iterations) to predict possible outcomes of an uncertain event. This allowed 80 inputs to vary simultaneously and probabilistically along a Pert-Beta distribution, which we chose for all the probabilistic inputs. Additional Monte Carlo simulations were run for each of the 2 sensitivity analyses and each of 3 subgroups for the aggregate DCEA. All analyses were conducted using Microsoft Excel and Palisade @Risk (https://www.palisade.com/risk/). All major input parameters used for this study are summarized in Table 1 (additional details in Tables S1-S3).

Aggregate DCEA

Different subpopulations can reap the benefits or costs of a given intervention differently. After examining the available data and policy salience in the MA context, we chose the equity variable of race/ethnicity for our pragmatic, aggregate DCEA. Data allowed 3 subgroup categories: non-Hispanic Black/African American (henceforth Black), Latine/Hispanic (Latine), and non-Hispanic White (White). The analysis was conducted based on the conceptual framework of the “staircase of inequality” developed for this study (Figure S1). The “staircase” portion of the framework included the following “steps”: (1) need (users of OTC diet pills and MBS, prevalence of ED among minors), (2) receipt (of prevention, of treatment for ED), (3) short-term effects (costs/savings to society, ED prevented), and (4) lifetime health benefits (QALYs gained).

For each input, we investigated if the race/ethnicity variable would have a different value compared to the best estimates for the CEA (Table 1). Due to data limitations and the fact that all minors are subject to the intervention, we focused on calculating just the difference in QALYs and costs/savings between the intervention and the status quo resulting in a pragmatic aggregate DCEA.⁸¹

Results

Health gains

We found that over 30 years, the proposed age-restriction policy would prevent 57 034 (95% uncertainty interval, UI: 26 833-89 271) ED diagnoses among the MA population; 32 370 (95% UI: 15 113-51 525) among females and 24 664 (95% UI: 11 402-40 119) among males (Table 2, Table S4). Additionally, it would prevent 22 358 (95% UI: 2609-49 605) diet pill-related AEs and 24 201 (95% UI: 10 959-40 038) muscle-building supplement- and AAS-related AEs (Table 2, Table S4).

Costs

The policy implementation would not incur incremental costs but rather yield discounted societal savings of $898 million (95% UI: $511 million-$1.4 billion, Table 2) and a health systems’ savings of $445 million (95% UI: $123-$857 million) to MA over the 30-year timeframe. These savings translate into $30 million (95% UI: $17-$47 million) and $14 million (95% UI: $4-$29 million) per year from a societal and health systems’ perspective, respectively. The inputs that had the greatest influence on the final estimate was the success rate of the intervention, the number of female AAS users, and the proportion of the population using OTC diet pills (Figures S2 and S3).

Cost-effectiveness

The proposed age-restriction policy would yield a total gain of 51 749 QALYs (95% UI: 22 065-91 183), with an average of 0.039 (95% UI: 0.016-0.068) QALYs per person. The primary driver of these health gains was the intervention success rate, the number of female AAS users, and the average duration of BED, which is estimated to be longer than any other ED (Figure S4). No ICER is reported because we found the intervention to be cost-saving compared to the status quo.

Accounting for uncertainty

In the first sensitivity analysis, the cost-effectiveness of the intervention was increased by including 3 additional AEs, for which we found some evidence in the literature, and by adding modest reductions in morbidity among minors who started with an ED but not OTC diet pill use. Our results were consistent with the base cases (Table S5) and showed annual societal savings of $31 million (95% UI: $18-$49 million), annual health systems savings of $15 million (95% UI: $5-$30 million), and 52 549 (95% UI: 22 336-94 758) total QALYs gained.

The second sensitivity analysis reduced the cost-effectiveness of the intervention by limiting the types of ED prevented and eliminating the segment of the model with the greatest uncertainty, that is, all non-ED AEs post-OTC diet pill use among males (Table S6). Nevertheless, our findings remained robust with annual societal savings of $28 million (95% UI: $17-$43 million), annual health systems savings of $12 million (95% UI: $4-$24 million), and 11 875 (95% UI: 4609-22 461) total QALYs gained.

Aggregate distributional cost-effectiveness

While the overall findings of the aggregate DCEA paralleled those of our CEA findings, we also detected differences in economic and health gains among the 3 racial/ethnic subgroups (Figure 2). We found an average societal savings per person of $645, $694, and $773 for White, Latine, and Black MA residents, respectively. Moreover, we found societal health gains of 0.044 QALYs per person for the Latine subpopulation, 0.042 QALYs per person for the Black subpopulation, and 0.038 QALYs per person for the White subpopulation, suggesting a reduction in overall health inequity between Black and Latine residents and White residents.

Discussion

We found that an age-restriction policy aiming to ban the sales of OTC diet pills and muscle-building supplements could prevent over 57 000 ED diagnoses among the MA population and over 46 500 other AEs over a 30-year time horizon. Accounting for the annual intervention costs still resulted in a finding of societal cost savings of over $898 million ($30 million per year). From a health systems-specific perspective, we still found the intervention to be cost-saving, that is, $14 million per year and $425 million over a 30-year time horizon. Moreover, we found evidence that the policy would modestly decrease health inequity as the Latine subpopulation gained the most QALYs per person, followed by the Black and White subpopulations, respectively. Thus, this age-restriction policy is strongly preferable to the comparator of the status quo.

While traditional cost-effectiveness analyses often exclude startup costs, policymakers frequently ask, “How much will this new policy cost?” A previous study examined startup opportunity costs for MA related to this policy.⁸² The authors valued these activities at less than $48 000 in total,⁸² suggesting that this intervention would be cost-saving even when initial startup costs are included. Furthermore, our DCEA findings suggested that the proposed age-restriction policy will reduce health inequity, meaning that the policy is likely to increase both population health and financial wellbeing while simultaneously reducing health disparities. Indeed, New York passed a very similar bill into law in 2023.⁸³

Our findings support the effectiveness of age-restriction policies in protecting the health of minors, aligning with the well-established evidence from the Tobacco 21 law and the National Minimum Drinking Age Act of 1984.^84,85 Age-restriction laws have the potential to safeguard youth health and reduce long-term adverse health effects. For instance, a study comparing the long-term health effects of increasing cigaret excise taxes and raising the legal smoking age to 21 years demonstrated the effectiveness of such laws, significantly reducing youth smoking prevalence by 13%.⁸⁶ This reduction would also lead to a decrease in adult smoking prevalence and a cumulative gain of 109 million QALYs.⁸⁶

Although our 30-year time frame resulted in additional reductions to the QALYs and cost savings through discounting, we aimed to match the time frame of key citations.^21,22 Additionally, evidence from age-restriction polices for tobacco usage suggested that a longer view is worth considering.⁸⁶ Finally, there is evidence showing that most ED are diagnosed by age 40.²²

The economic and health gains, along with equity considerations highlighted in this study, underscore the need for effective public health measures to safeguard the wellbeing of youth. While our analysis focused specifically on the cost-effectiveness of the age-restriction law on ED diagnoses and other AEs associated with the use of OTC diet pills, MBS, and AAS, it also suggested broader implications, particularly in terms of increasing public awareness of the negative impact of these products on youth health and wellbeing.

Limitations

The study has some limitations. First, due to data availability, we relied on parallel evidence from the tobacco-control literature for our estimate of intervention effectiveness. Second, due to limited race/ethnicity-specific data, our aggregate DCEA focused on only the 3 largest subgroups. That left an unknown change in equity for approximately 8% of the population. However, given the overwhelmingly positive findings for the base case CEA, and our success in analyzing the health gains for the White and Black subgroups, which are often found at the extreme positive and extreme negative end of health equity distributions, respectively, we anticipate future findings will remain consistent.

Third, data limitations also affected our ability to explore other equity categories of salience, which include urbanicity, socioeconomic status, gender identity, and sexual orientation. New data sources and studies are required to gain further health inequity insights related to this public health policy.

Fourth, our societal opportunity costs for the aggregate DCEA are assumed to be zero based on expected light enforcement actions. Following the tobacco model, cities and towns may decide independently to employ underage “secret shoppers.”⁸⁷ This implies flexible funds available to municipalities for enforcement, which will be rare. Otherwise, enforcement will be via the passive pathway of tips called into the Attorney General's office, and no additional infrastructure needs to receive those calls are anticipated.

Even without new state enforcement allocations, we expect the policy to be at least modestly effective for several reasons. The act of moving products behind counters or into locked cabinets will reduce accessibility, and the illegality of purchasing such products is enough to deter some would-be underaged buyers. Further, the success rate of the intervention we used was modest (26%) and the sensitivity analyses produced consistent results with our base case.

Conclusions and implications

Our findings indicated that restricting the sale of OTC diet pills and muscle-building supplements to minors will produce both health gains and reduce societal direct and indirect costs. Additionally, we estimated the Latine population would experience the highest per capita health benefits, followed by the Black and White populations. As a result, policymakers and the public may have increased confidence in the effectiveness, fiscal responsibility, and positive equity impacts of this prevention policy.

Acknowledgments

All authors report no conflict of interest. Additionally, the authors wish to thank Dr. Michael W. Long for his invaluable feedback on earlier drafts, as well as Mikayla Ferg, Maddie Glenn, and Marriam Zaky for assistance with the manuscript preparation.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

Funding was provided in part by the Strategic Training Initiative for the Prevention of Eating Disorders.

Notes

Author notes