Cost-Effectiveness of Pharmacotherapy for the Treatment of Obesity in Adolescents

Key Points Question Compared with lifestyle counseling alone, is liraglutide, mid-dose phentermine and topiramate, top-dose phentermine and topiramate, or semaglutide adjunct to lifestyle counseling cost-effective in treating obesity among adolescents in a simulated cohort? Findings In this economic evaluation including 100 000 simulated adolescents, no pharmacotherapy was estimated to be cost-effective after 13 months and 2 years of treatment. After 5 years, top-dose phentermine and topiramate was projected to be cost-effective, with an incremental cost-effectiveness ratio of $56 876 per quality-adjusted life year gained. Meaning This economic evaluation found that top-dose phentermine and topiramate adjunct to lifestyle counseling was cost-effective for the treatment of obesity in adolescent patients after 5 years.


Introduction
3][4][5] The prevalence of adolescent obesity in the US has increased for nearly 2 decades, currently affecting more than 1 in 5 adolescents. 60][11] In 2023, the American Academy of Pediatrics (AAP) recommended antiobesity medications (AOMs) be offered as an adjunct to intensive health behavior and lifestyle counseling in adolescents with obesity ages 12 years and older. 7gnificant placebo-subtracted BMI reductions from baseline to approximately 1 year have been seen with pharmacologic treatment of obesity in clinical trials of adolescents ages 12 years to younger than 18 years, with reductions of 4.6% reported for liraglutide, 12 8.1% reported for mid-dose phentermine and topiramate (ie, 7.5 mg phentermine and 46 mg topiramate daily), 13  10.4% reported for top-dose phentermine and topiramate (ie, 15 mg phentermine and 92 mg topiramate daily), 13 and 16.7% reported for semaglutide. 14Subsequently, the US Food and Drug Administration (FDA) has approved liraglutide, phentermine and topiramate, and semaglutide for treatment of obesity in adolescents ages 12 years and older.However, these are proprietary medications and have annual costs ranging from $1100 to $15 000.Estimating the cost-effectiveness of these adjunctive AOMs may help guide uptake and pricing.
The purpose of this study was to estimate the costs, quality-adjusted life-years (QALYs), and cost-effectiveness of lifestyle counseling alone and adjunct to liraglutide, mid-dose phentermine and topiramate, top-dose phentermine and topiramate, or semaglutide to treat US adolescents with obesity over 13 months and estimate projections out to five years.

Methods
This economic evaluation used computer-based modeling and is not considered human participants research at Columbia University; therefore, neither institutional review board approval nor informed consent were sought.Reporting of this study followed the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guideline.The Criteria for Health Economic Quality Evaluation tool was referenced for methods and reporting quality. 15

Model Overview
We adapted a previously published patient-level microsimulation model for the treatment of obesity in adults to assess the cost-effectiveness of lifestyle counseling alone and adjunct to liraglutide (3 mg daily), mid-dose phentermine and topiramate, top-dose phentermine and topiramate, or semaglutide (2.4 mg weekly) in adolescents with obesity. 16The model simulated a hypothetical cohort of 100 000 adolescents with baseline characteristics similar to participants in clinical trials: age 15 years, 58 000 (58%) female, and BMI of 37. [12][13][14] While clinical trials stated their eligibility as BMI at the 95th percentile or higher according to sex-and age-specific growth charts, the average participant had severe obesity, with a mean BMI of 120% or more of the 95th percentile and/or BMI  35. 11 Cost-effectiveness was assessed at 13 months (ie, the most common treatment period in clinical trials), 2 years, and 5 years (eMethods in Supplement 1).We did not make projections beyond 5 years due to lack of long-term data.

Strategies
Lifestyle counseling consisted of counseling on healthy nutrition, including a diet aligned with dietary guidelines, and daily moderate-to-high intensity physical activity, with the inclusion of parents or guardians. 12Liraglutide and semaglutide are glucagon-like peptide 1 analogues that are administered via subcutaneous injection daily and weekly, respectively, when used for weight management. 12,14,36,37Phentermine and topiramate is a fixed-dose combination of immediaterelease phentermine and extended-release topiramate administered orally daily for weight management. 13,19,20AOMs were administered with lifestyle counseling.Bariatric surgery was not a comparator because it is typically considered for adolescents with a BMI greater than that of our modeled cohort. 38

BMI Change
A natural BMI trajectory estimating BMI change without treatment was derived from extended BMI-for-age growth charts from the CDC. 39At age 15 years, a BMI of 37.0 is greater than 120% of the 95th percentile for girls and boys.We assumed no change in the BMI-for-age percentile occurred, resulting in a BMI of 42.5 after 5 years without treatment.
Patients receiving treatment experienced BMI reduction relative to their natural BMI trajectory.
When an individual discontinued treatment, they regained BMI at an increased rate until they returned to their expected natural BMI.The number of months required to return to their natural BMI was derived from the liraglutide trial by Kelly et al 12 and was dependent on how long treatment was received (eMethods and eTable 2 in Supplement 1).3][14] Lifestyle counseling alone was based on the placebo group of the liraglutide trial by Kelly et al. 12 BMI reduction while receiving treatment was calibrated to reproduce the intention-to-treat values for relative change in BMI from baseline, accounting for the proportion of

Death from all-cause mortality
All patients began in the microsimulation model receiving treatment, either lifestyle counseling alone or adjunct to an antiobesity medication, ie, liraglutide, mid-dose phentermine and topiramate (7.5 mg phentermine and 46 mg topiramate), top-dose phentermine and topiramate (15 mg phentermine and 92 mg topiramate), or semaglutide.Every month of the simulation, patients could either continue receiving treatment or permanently discontinue treatment.Patients receiving lifestyle counseling adjunct to an antiobesity medication could experience an adverse event every month.Monthly change in body mass index was dependent on the specific treatment strategy, and all patients experienced weight gain after discontinuing any treatment.Death was determined from age-and sex-specific mortality rates.
3][14] Calibration was validated by comparing the model output of relative BMI change from baseline with reported intentionto-treat values.
Due to lack of long-term follow-up data, we made several assumptions to extrapolate BMI beyond the treatment period of the adolescent trials.For lifestyle counseling, individuals receiving treatment maintained the same relative BMI reduction compared with their natural BMI trajectory achieved at 13 months (eMethods in Supplement 1).9][20][21][22] In adults, weight changes from baseline to approximately 1 and 2 years of treatment were converted to BMI changes using mean baseline weight and BMI (eTable 4 in Supplement 1).We assumed the same relative BMI change from 1 to 2 years in adults applied to adolescents (eMethods and eTable 3 in Supplement 1).Patients who continued to receive treatment for at least 2 years did not experience further BMI reduction and maintained the same relative change compared with their natural BMI trajectory achieved at 2 years.

Treatment Discontinuation
9][20][21][22] All individuals who were still receiving treatment at 2 years were assumed to continue treatment until 5 years.Patients who discontinued use of AOMs also discontinued lifestyle counseling.

Adverse Events
3][14] The probability of AEs was not projected beyond the duration of the trials.However, most events occurred within the first several weeks of treatment.

Quality-of-Life Adjustments and Costs
The initial utility of severe obesity in adolescents was obtained from published literature. 23An increase in utility was incorporated per 1-unit decrease in BMI. 246][27] Utilities were derived from a patient perspective.
We adopted a health care sector and limited societal perspective (eMethods in Supplement 1).
An hour of time was valued to be the mean hourly wage from the US Bureau of Labor Statistics, $32. 40The cost of lifestyle counseling was estimated from a family-based intervention for pediatric obesity (eMethods in Supplement 1). 28Upper and lower bounds were estimated from similar interventions. 29,41The base case monthly cost of AOMs was determined from the Center for Medicare and Medicaid Services National Average Drug Acquisition Cost. 30Upper and lower bounds of estimates were the mean wholesale price from Micromedex Red Book and the lowest price from the Department of Veterans Affairs Federal Supply Schedule, respectively (eMethods in Supplement 1). 31,32A cost of 2 physician visits was added to the first month of AOM treatment to account for extra visits for initiation. 16,33,42Patients who experienced a nonsevere AE received the cost of 1 physician visit.Costs of severe AEs were estimated from the costs of severe drug AEs. 34tients who discontinued treatment did not accrue further costs.Total medical care costs for obesity were estimated from nationally representative data (eMethods in Supplement 1). 35All costs are reported in 2022 US dollars and were adjusted using the health care component of the Personal Consumption Expenditures price index. 43Model inputs are detailed in Table 1.Key model assumptions are summarized in eTable 6 in Supplement 1.

Statistical Analysis
All analyses were performed using Python statistical software version 3.8.8(Python Software Foundation).Data analysis was performed from April 2022 to July 2023.Our primary end points were quality-adjusted life years (QALYs), total costs (2022 USD), and incremental cost-effectiveness ratios (ICERs) (eMethods in Supplement 1).Future costs and QALYs were discounted at a rate of 3%.
We used a willingness-to-pay (WTP) threshold of $100 000 per QALY to determine cost-effectiveness. 44A strategy was considered preferred if it resulted in the greatest increase in QALYs while being cost-effective.Secondary end points included relative BMI change from baseline and relative BMI change from natural BMI trajectory.
To determine the impact of uncertainty of model inputs on results, we performed 1-way and probabilistic sensitivity analyses.One-way sensitivity analyses vary 1 parameter at a time across a range of values while keeping all other inputs at their base-case value (eMethods in Supplement 1).
Upper and lower bounds were obtained from minimum and maximum values present in databases or calculated from 95% CIs (eMethods in Supplement 1).
Probabilistic sensitivity analyses were performed by repeatedly sampling model inputs simultaneously from defined probabilistic distributions (eMethods in Supplement 1).We used γ distributions for costs and β distributions for all other parameters.The model was run 1000 times with a cohort of 100 000 patients.We determined the percentage of times each strategy was preferred over a range of different WTP thresholds.To assess uncertainty of mortality estimates, the model was run separately using BMI-specific life tables and unadjusted life tables to compare costeffectiveness results.a The parameter was not varied in 1-way sensitivity analyses and was kept as its base case value.
b The parameter was not varied in probabilistic sensitivity analyses and was kept as its base case value.
c BMI reduction while receiving treatment is the percentage reduction relative to baseline BMI.
d Calibrated to match the mean intention-to-treat percentage BMI reductions from baseline from clinical trials, accounting for treatment discontinuation (eTable 3 in Supplement 1).
e Treatment discontinuation after 24 months for semaglutide and after 25 months for lifestyle counseling, liraglutide, and phentermine and topiramate was assumed to be zero.Patients who were still receiving treatment at that time point were assumed to continue treatment until 5 years.
f Initial utility for severe obesity and the increase in utility for a 1-unit decrease in BMI were applied across all strategies.A decrease in utility from AEs was only applied to patients receiving AOMs.
g The cost of 2 physician visits were applied to each patient who began treatment with AOMs.The cost of 1 physician visit was applied to each patient who experienced a gastrointestinal or psychiatric AE while using AOMs.
Over 13 months, liraglutide was strictly dominated (ie, cost more and less effective) and mid-dose phentermine and topiramate was extendedly dominated (ie, less effective and higher cost per QALY) compared with top-dose phentermine and topiramate (

Sensitivity Analyses
At 13 months, lifestyle counseling remained the preferred strategy across all parameter ranges in 1-way sensitivity analyses.At 2 years, top-dose phentermine and topiramate became preferred when its cost was at its minimum value and utility of BMI reduction was at its maximum value.Mid-dose Abbreviations: QALY, quality-adjusted life-year; ICER, incremental cost-effectiveness ratio.
a ICER is calculated relative to the next least-costly, nondominated strategy.
b Strictly dominated indicates that the strategy resulted in higher costs and fewer QALYs compared with another strategy.
phentermine and topiramate was preferred at its minimum cost.At 5 years, top-dose phentermine and topiramate was the preferred strategy across most parameter ranges (Figure 2).Lifestyle therapy was preferred when the utility of BMI reduction was at its minimum value.
Using a WTP threshold of $100 000 per QALY gained, probabilistic sensitivity analysis estimated lifestyle counseling to be the preferred strategy in 100.0% of 1000 probabilistic iterations at 13 months and 81.3% of 1000 probabilistic iterations at 2 years (Figure 3 and eFigure 5 in Supplement 1).By 5 years, top-dose phentermine and topiramate was the preferred strategy in 84.3% of simulations, followed by lifestyle counseling at 10.7%, and mid-dose phentermine and topiramate at 5.0%.
Due to semaglutide's effectiveness in maintaining BMI reduction from baseline, we performed a threshold analysis to determine the price reduction needed for it to become the preferred treatment strategy.At 13 months, the monthly cost of semaglutide would need to be reduced by 97.5% (ie, cost $32 per month) for it to become cost-effective (eFigure 6 in Supplement 1).At 2 and 5 years, the monthly cost of semaglutide would need to be reduced 89.4% and 85.2%, respectively, to become the preferred strategy.Changes in mortality rates did not appreciably affect costeffectiveness results (eTable 8 in Supplement 1).

Discussion
In this economic evaluation, we used a simulation model to examine the cost-effectiveness of 3 AOMs adjunct to lifestyle counseling to treat adolescents with obesity.AOMs were not estimated to be cost-effective compared with lifestyle counseling alone at 13 months and 2 years.At 5 years, top-dose phentermine and topiramate was projected to be the preferred strategy, with an ICER of Model parameters were independently varied across a set range while all other parameters were held constant at their mean value.The 10 model parameters with the largest effect on the incremental net monetary benefit relative to lifestyle counseling at a willingness-to-pay threshold of $100 000 per quality-adjusted life-year are shown.Parameters with the largest effect are located at the top, and parameters with the smallest effect are shown at the bottom.The incremental net monetary benefit represents the monetary value of an intervention for a given WTP threshold and is calculated as incremental quality-adjusted life-years × willingness-to-pay − incremental costs.An incremental net monetary benefit greater than $0 indicates the strategy is cost-effective compared with lifestyle counseling.The horizontal bars represent the range of incremental net monetary benefit when changing the value of each model parameter, and the color indicates which strategy was preferred.A change in the preferred strategy is shown with a colored vertical bar at the end of the horizontal bar.+ and − at the end of the horizontal bars denote that the incremental net monetary benefit was calculated at the maximum and minimum value of the parameter, respectively.A tornado diagram was not generated for a time horizon of 13 months and 2 years, as the preferred strategy in most of the parameter ranges examined was lifestyle counseling.
$56 876 per QALY gained vs lifestyle counseling.Our results were sensitive to utility of BMI reduction and BMI loss associated with lifestyle counseling and top-dose phentermine and topiramate.
Semaglutide resulted in the greatest gain in QALYs but was not projected to be cost-effective due to its high monthly cost, which would need to be reduced more than 85% to become cost-effective.
Several analyses have evaluated the cost-effectiveness of school-based programs, 45,46 policy changes, 47 or bariatric surgery 24,48 for the prevention or treatment of obesity in children and The dashed line indicates the base case willingness-to-pay threshold of $100 000 per quality-adjusted life-year (QALY) gained.
adolescents.However, less is known about the cost-effectiveness of AOMs in this population.To our knowledge, this is the first analysis to compare liraglutide, phentermine and topiramate, and semaglutide for the treatment of obesity in adolescents.Adult studies have generally demonstrated the cost-effectiveness of AOMs over short (ie, Յ5 years) 16,42,49 and long (ie, Ն30 years) [50][51][52] time horizons with some exceptions. 531][52] All models estimated semaglutide to have the greatest gain in QALYs, and 2 models projected phentermine and topiramate to be the cost-effective strategy, 50,52 similar to our findings.
The goal of obesity treatment in adolescents is to improve quality of life and reduce the risk of future chronic disease. 11Physical health benefits with AOMs may not be observed for years, but quality of life improvements may be seen more quickly.Adolescents with obesity experience weight stigma, especially in school settings. 54Therefore, short-term analyses may be useful in this school-age population as short-term weight loss was still associated with increased QALYs in our model.With the recent release of guidelines from the AAP, which recommend the use of AOMs for adolescents ages 12 years or older, 7 our analysis is timely and underscores the need to compare the effectiveness and cost-effectiveness of AOMs.
Though semaglutide was the only strategy estimated to have a BMI reduction relative to baseline at 5 years, its monthly price of nearly $1300 resulted in an ICER well above our WTP threshold.The high cost of AOMs, especially out-of-pocket, is a major barrier that may discourage patients from receiving or adhering to the aggressive treatment that the AAP recommends. 7Most state Medicaid plans do not cover AOMs, and private insurance often has strict criteria for authorization. 55,56Given that obesity is a chronic disease, mounting AOM costs over time may prevent use of these medications.

Limitations
The results of our analysis should be considered in the context of several limitations.We did not include all FDA-approved AOMs for adolescents.We excluded orlistat, since it is currently rarely used due to AEs and limited tolerability. 7While lifestyle counseling often has high attrition rates, it was included as a comparator because it is the foundational approach for BMI reduction. 7Our cost estimate for lifestyle counseling may not have represented the intensity of counseling provided in the adolescent clinical trials, since the number of contact hours in the trials was not reported.We used a limited societal perspective that did not include all indirect costs.We assumed patients receiving no treatment maintained their BMI-for-age percentile and did not incorporate secular trends or individual heterogeneity in BMI trajectories.Our model assumed BMI regain occurred at a constant rate and may not have accounted for nonlinear associations of costs and mortality with BMI.
Because model inputs were estimated from multiple sources, sensitivity analyses were used to address input uncertainty and showed that model results remained robust.Due to lack of long-term adolescent data, we extrapolated BMI changes and treatment adherence from adult data and assumed no treatment discontinuation or further BMI reduction after 2 years of treatment.While there is evidence that adults discontinue AOMs within 2 years, 51,57 we modeled a maximum of 5 years of treatment, since obesity requires long-term treatment. 8We may have underestimated the benefit of AOMs, since obesity-related comorbidities may be prevented or delayed in adulthood. 7However, due to limited research with long-term follow-up, it is unknown whether AOMs lead to sustained BMI reduction, and it is unclear how comorbidities develop through childhood to adulthood. 7Modeled patients were based on participants in clinical trials and were not a nationally representative sample.

Figure 2 .
Figure 2. One-Way Sensitivity Analyses Over 5-Year Time Horizon

Future clinical trials should
enroll participants who are representative of the US population, including low-income; racial, ethnic, gender, or sexual minority; and other underrepresented patients and should report differential outcomes based on patient subgroups.The ideal follow-up time of future trials is at least 10 to 20 years to capture benefits of disease prevention.

Table 1 .
Microsimulation Model Inputs Cost-Effectiveness of Pharmacotherapy for the Treatment of Obesity in Adolescents

Table 2
and eFigure 4 in Supplement 1).Top-dose phentermine and topiramate was not cost-effective, with an ICER of $317 010 per QALY gained vs lifestyle counseling.At 2 years, liraglutide and mid-dose phentermine and topiramate were strictly dominated by top-dose phentermine and topiramate.The ICER of top-dose phentermine and topiramate vs lifestyle counseling decreased to $138 045 per QALY gained.By 5 years, top-dose phentermine and topiramate became the preferred strategy, with an ICER of $56 876 per QALY gained vs lifestyle counseling.Over each time horizon, semaglutide was projected to accumulate the most QALYs.However, the ICERs for semaglutide vs top-dose phentermine and topiramate were well above our WTP threshold, ranging from $1.1 to $3.0 million per QALY gained.

Table 2 .
Cost-Effectiveness Results Over Each Time Horizon

SUPPLEMENT 2. Data Sharing Statement
BMI-Specific All-Cause Annual Mortality Rates Among Males and Females eTable 2. Number of Months to Return to Natural BMI Based on Treatment Length eTable 3. Relative Change in BMI From Baseline Calibration Targets eTable 4. Relative Change in BMI From Baseline in Adult Clinical Trials eTable 5. Proportion of Patients Who Continued Treatment in Adult Clinical Trials eTable 6. Key Model Assumptions eTable 7. Model Output of Relative BMI Change From Baseline eTable 8. Cost-Effectiveness Results Over Each Time Horizon Using Non-BMI Specific Life Tables eFigure 1. Relative BMI Change From Baseline Over a 5-Year Time Horizon eFigure 2. BMI Over a 5-Year Time Horizon eFigure 3. Relative BMI Change From Natural BMI Trajectory Over 5-Year Time Horizon eFigure 4. Cost-Effectiveness Planes eFigure 5. Incremental Cost-Effectiveness Scatterplots eFigure 6. Threshold Analysis for Monthly Cost of Semaglutide eReferences.