The economic case for digital interventions for eating disorders among United States college students

Abstract

Eating disorders (EDs) are serious health problems affecting college students. This paper aimed to estimate the costs, in United States (US) dollars, of a stepped care model for online prevention and treatment among US college students to inform meaningful decisions regarding resource allocation and adoption of efficient care delivery models for EDs on college campuses.

Methods

Using a payer perspective, we estimated the costs of (1) delivering an online guided self-help (GSH) intervention to individuals with EDs, including the costs of “stepping up” the proportion expected to “fail”; (2) delivering an online preventive intervention compared to a “wait and treat” approach to individuals at ED risk; and (3) applying the stepped care model across a population of 1,000 students, compared to standard care.

Results

Combining results for online GSH and preventive interventions, we estimated a stepped care model would cost less and result in fewer individuals needing in-person psychotherapy (after receiving less-intensive intervention) compared to standard care, assuming everyone in need received intervention.

Conclusions

A stepped care model was estimated to achieve modest cost savings compared to standard care, but these estimates need to be tested with sensitivity analyses. Model assumptions highlight the complexities of cost calculations to inform resource allocation, and considerations for a disseminable delivery model are presented. Efforts are needed to systematically measure the costs and benefits of a stepped care model for EDs on college campuses, improve the precision and efficacy of ED interventions, and apply these calculations to non-US care systems with different cost structures.

Eating disorders (EDs) are serious health problems affecting college students. Nine percent of college students screen positive for ED symptoms,¹ and EDs often emerge during the traditional college years.^2,3 EDs are associated with high morbidity and healthcare costs,⁴ and can lead to declines in healthy lifestyle behaviors, academic performance, social functioning, and quality of life, as well as increased rates of suicide, medical leave, and dropout,^5–11 which negatively impact students’ future health trajectories^12,13 and economic potential.¹⁴

Given the serious and impairing effects of EDs, we believe all individuals in need of treatment should be able to receive accessible, affordable, and evidence-based care. Indeed, even in healthcare systems with limited and constrained resources [like that of the United States (US)], there is cost-offset to addressing mental health issues.^15–17 Although health services are commonly available on US college campuses, the process by which students seek care to address mental health problems is fraught with barriers,^18–21 as shown in Figure 1. Many students do not present to or receive care^1,22,23 given these barriers, which can delay treatment and lead to poorer prognosis and greater relapse rates.²⁴ Discrepancies between access to and demands for care suggest the need for novel care delivery models that optimize resource delivery while conserving costs.

Figure 1.

Theoretical Model of the System of Care on College Campuses

An ideal model for ED service delivery would combine interventions that reduce symptoms among those with clinical problems with interventions that decrease likelihood of disorder onset among those at risk for an ED and would deliver the majority of services using low-cost, scalable resources like online/mobile screening tools and interventions. Digital technologies can improve mental health care on college campuses by overcoming barriers and addressing the treatment gap to ensure care is delivered to the substantial subset of individuals who do not present for treatment.²⁵ In doing so, online/mobile interventions have potential to offset in-person clinical demands, increase access and deliver care to more people, decrease stigma, enhance precision to tailor care and improve outcomes, and reduce costs.²⁶

We have proposed a stepped care model for screening and delivering interventions to college students.^27,28 This model uses online screening to detect individuals at risk for or with an ED, and offers them: (a) an online self-help universal preventive intervention (for those at low ED risk); (b) an online self-help selective preventive intervention (for those at high ED risk); (c) an online guided self-help (GSH) intervention (for those with a subclinical or clinical ED, with the exception of those with full-syndrome anorexia nervosa); or (d) a referral to in-person care (for those with anorexia nervosa or medical concerns warranting more intensive intervention). Individuals who do not show symptom reduction are directed to more intensive intervention. This model has been proposed as an Idea Worth Researching.²⁷ We are testing parts of this model and its cost-effectiveness in a large-scale randomized controlled trial in the US using interventions we have developed, which we refer to as the Healthy Body Image (HBI) Program; however, this stepped care approach can be applied and used with other tools for ED screening and intervention.²⁹

Although research supports the efficacy of online ED screening and intervention,^30–32 there is scant data on the costs of these tools^33–35 or on estimates of the cost-effectiveness of stepped care models for EDs relative to standard care.^36,37 Thus, we aimed to estimate the costs of implementing a stepped care model for EDs, such as HBI, among US college students. Such work can inform meaningful decisions regarding the allocation of resources and the adoption of efficient care delivery models for addressing EDs on college campuses.³⁸

Methods

Cost Analysis

This study was modeled on work evaluating interventions for generalized anxiety disorder among college students.³⁹ Our approach considers costs from the payer perspective (incurred by both the university and healthcare system) and excludes participant costs, because this paper serves as an estimation exercise of the costs relevant to a US university care delivery setting (or other relevant healthcare delivery settings) implementing an online stepped care model. Costs were calculated in US dollars for the year 2016 (with previous data inflated to 2016 price levels using the Consumer Price Index).⁴⁰

Although this paper focuses on costs, it is also helpful to estimate how many individuals could be successfully treated with online technologies, given the goal of optimizing the allocation of resources. For each model, we estimated the number of individuals who would avoid in-person psychotherapy (i.e., standard care on college campuses) using online preventive and/or GSH interventions, applied across a population of 1,000 individuals.

Costs of Models for Online Guided Self-Help Intervention

Online GSH interventions have been tested for EDs.^30–32 Our calculations used a conservative approach based on the premise that individuals would be “stepped up” to more-intensive care if they “fail” a less intensive intervention. Accordingly, we accounted for both (a) the cost of the GSH intervention, and (b) the cost of in-person treatment for those who “fail” the intervention, multiplied by the proportion expected to “fail” the GSH intervention (defined as 1 – the probability of success of GSH intervention).

Following Kanuri and colleagues’ approach,³⁹ we used success rate difference (SRD) to account for intervention success. SRD is an effect size metric with clinical application.⁴¹ For binary outcomes, SRD is the inverse of the number needed to treat (NNT). SRDs used in this paper were based on calculations from Cohen’s d and NNT.⁴¹ When reported Cohen’s d was between two integers, we used the more conservative (i.e., lower) SRD. Our equation for reducing ED symptoms with online GSH intervention was:

$$=[(\$\text{GSH}\text{intervention})+(1-{\text{SRD}}_{\text{GSH}\text{intervention}})(\$\text{In-person}\text{psychotherapy})]$$

Costs of Models for Online Preventive Intervention

Online preventive interventions have also been tested for individuals at risk for or with some ED symptomatology.^31,32,42 Preventive interventions are ideal for implementation if delivering a preventive intervention to all high-risk individuals costs less than treating the subset who go on to develop an ED (i.e., a “wait and treat” approach). Our equation comparing preventive intervention to “wait and treat” was:

$$\begin{array}{l}=[(\$\text{Preventive}\text{intervention})+(1-{\text{SRD}}_{\text{Preventive}\text{intervention}})(\$\text{GSH}\text{intervention})]\\ +(1-{\text{SRD}}_{\text{GSH}\text{intervention}})(\text{rate}\text{of}\text{ED}\text{onset}\text{among}\text{high-risk})(\$\text{Treaat}\text{one}\text{case})\end{array}$$

compared to

$$=[(\text{rate}\text{of}\text{ED}\text{onset}\text{among}\text{high-risk})(\$\text{Treat}\text{one}\text{case})].$$

Application of the Stepped Care Model across a Population of College Students

We calculated the costs of applying the stepped care model across a population of 1,000 college students compared to standard care on college campuses (i.e., providing in-person treatment to those with EDs). This calculation accounted for prevalence and reach associated with scaling this model. Our equation comparing the stepped care model to standard care was:

$$\begin{array}{l}=[(\text{Prevalence}\text{of}\text{anorexia}\text{nervosa})(\$\text{In-person}\text{psychotherapy})+(\text{Prevalence}\text{of}\text{EDs}\\ \text{excluding}\text{anorexia}\text{nervosa})(\mathrm{\sum }\$\text{for}\text{GSH}\text{intervention})+(\text{Prevalence}\text{of}\text{high}\text{risk})(\mathrm{\sum }\$\text{for}\\ \text{Preventive}\text{intervention})]\end{array}$$

compared to

$$\begin{array}{l}=[(\text{Prevalence}\text{of}\text{EDs})(\$\text{In-person}\text{psychotherapy})+(\text{rate}\text{of}\text{ED}\text{onset}\text{among}\text{high}\\ \text{risk})(\text{Prevalence}\text{of}\text{high}\text{risk})(\$\text{In-person}\text{psychotherapy})].\end{array}$$

Results

Table 1 shows the parameters used in the models presented below.

Table 1.

Parameters used in Calculations for the Models

VARIABLE	VALUE	SOURCE
Models for Online Guided Self-Help Intervention
Cost of online guided self-help intervention	$90.63	35,37
Cost of in-person CBT	$1,641.15	36
Effect of online guided self-help intervention
Reducing binge eating	d = 0.44 (SRD = 0.223)	31
Reducing purging	d = 0.43 (SRD = 0.223)	31
Reducing global ED psychopathology	d = 0.54 (SRD = 0.276)	31
Reducing “bulimic symptoms”	d = 0.27 (SRD = 0.112)	32
Reducing purging	d = 0.30 (SRD = 0.168)	32
Models for Online Preventive Intervention
Cost of online guided preventive intervention	$26	44
Cost of online unguided preventive intervention	$0	45
Effect of online guided preventive intervention
Preventing ED onset among high risk	NNT = 15 (SRD = 0.067)	44
Preventing ED onset among highest risk subgroup	NNT = 5 (SRD = 0.200)	44
Effect of online unguided preventive intervention	d = 0.33 (SRD = 0.168)	45
Incidence of EDs
ED onset among high risk	31%	44
ED onset among highest risk subgroup	42%	44
Application of the Stepped Care Model
Rates of EDs from campus screening
Anorexia nervosa	0.7%	28
EDs excluding anorexia nervosa	4.7%	28
High risk for an ED	31.8%	28
Low risk for an ED	62.9%	28
Cost of in-person enhanced CBT for anorexia nervosa	$4,109.56	46

Note: Costs are presented in 2016 US dollars.⁴⁰ CBT = cognitive behavioral therapy; NNT = number needed to treat; SRD = success rate difference⁴¹; ED = eating disorder.

Costs of Models for Online Guided Self-Help Intervention

To our knowledge, only one study has calculated the costs of an online GSH intervention;³⁵ although this study was not limited to college students, we used these costs as no comparable data for college students were available. In this evaluation, average costs were €53 and €107 per participant for interventions with low- and high-intensity therapist support, respectively.³⁵ Converted to 2016 US dollars and averaged, we assumed a cost of $90.63 for online GSH intervention. To justify this estimate, we compared the cost for online to in-person GSH intervention. We hypothesized online intervention would cost less than in-person, as in-person sessions with a participant would be expected to last longer than asynchronous online communication to a participant. Indeed, in-person GSH intervention was determined as $199.93 (converted to 2016 US dollars from $167).³⁷ Thus, we proceeded with $90.63 for online GSH intervention.

We used the cost of $1,641.15 (converted to 2016 US dollars from $1,328) calculated for in-person cognitive behavioral therapy.³⁶ Although this cost was based on treatment for individuals with bulimia nervosa in the US, we assumed this cost would hold true for individuals with binge eating disorder and subclinical EDs, and if limited to college students.

We evaluated online GSH interventions based on their efficacy in reducing clinically-significant ED outcomes. We derived efficacy values from a meta-analysis,³¹ which reported post-intervention effect sizes of 0.44, 0.43, and 0.54 for reducing binge eating, purging, and global ED psychopathology, respectively, using online interventions versus waitlist control conditions in individuals with bulimia nervosa.³¹ These effect sizes equated to respective SRDs of 0.223, 0.223, and 0.276.⁴¹

Using our equation, it would cost [($90.63) + (1–0.223)($1,641.15)] = $1,365.80 to reduce binge eating or purging, and [($90.63) + (1–0.276)($1,641.15)] = $1,278.82 to reduce global ED psychopathology. This estimate shows that online GSH intervention is less expensive to deliver than in-person psychotherapy (considered first-line care for bulimia nervosa and binge eating disorder^24,43), to achieve changes in clinically-significant outcomes, even when accounting for the rate of failure of the less-intensive GSH intervention. Applied across 1,000 individuals with EDs, 223 individuals would not need in-person psychotherapy to address binge eating or purging (or 276 individuals to address global ED psychopathology) after receiving GSH.

This calculation assumes the published effects of intervention will hold with future implementation. To challenge this assumption, we varied this parameter to determine how changes in efficacy impact delivery costs. We used a second meta-analysis evaluating online interventions among individuals with a wider range of symptoms (i.e., symptomatic or with full-syndrome EDs).³² The meta-analysis showed an effect of d=0.27 for reducing “bulimic symptoms” and d=0.30 for changing purging frequency,³² equivalent to SRD=0.112 and 0.168, respectively.⁴¹ Using our equation, it would cost [($90.63) + (1–0.112)($1,641.15)] = $1,547.97 to reduce “bulimic symptoms” and [($90.63) + (1–0.168)($ 1,641.15)] = $1,456.07 to change purging frequency. These estimates continue to support that online GSH intervention, delivered via stepped care, costs less than providing in-person psychotherapy to impact ED symptoms, even at relatively low efficacy. Applied across 1,000 individuals with EDs, 112 individuals would not need in-person psychotherapy to address bulimic symptoms (or 168 individuals to address purging) after receiving GSH.

Costs of Models for Online Preventive Intervention

A 10-week online guided preventive intervention, Image and Mood, was evaluated among college-age women at high ED risk.⁴⁴ The intervention cost $26 per participant (assumed as 2016 US dollars). The intervention compared to waitlist control had an effect of NNT=15 for ED onset (i.e., there was a lower incidence of EDs at 2-year follow-up favoring the intervention, but this difference was not statistically significant), and the rate of ED onset over two years among those in the control condition was 31%.⁴⁴ Given that individuals in the control condition did not receive an intervention, they provided an estimate of the rate of ED onset in a “wait and treat” approach. The SRD (i.e., 1/NNT⁴¹) was 0.067. We evaluated the costs of delivering a prevention intervention to a population of individuals at ED risk compared to a “wait and treat” approach as:

$$\begin{array}{l}=[(\$26)+(1-0.067)(\$90.63)+(1-0.223)(31\%)(\$1,641.15)]=\$505.86,\text{compared}\text{to}\\ =[(31\%)(\$1,641.15)]=\$\mathrm{508.76.}\end{array}$$

Thus, delivering a prevention intervention to a population of individuals at ED risk costs less than a “wait and treat” approach, based on estimates from the population from which these data are drawn. Applied across 1,000 at-risk individuals, 241 individuals would require in-person psychotherapy (after receiving preventive and GSH interventions) in the prevention approach versus 310 individuals in “wait and treat.”

We also evaluated the moderator effect shown in the trial,⁴⁴ to consider the impact of directing resources at particularly vulnerable subpopulations. Among those at highest risk (i.e., with highest shape concerns), the intervention compared to waitlist control had an effect of NNT=5 (i.e., SRD=0.2) for ED onset. In this subgroup, 2-year ED onset rate for the control condition was 42%. For this comparison, among individuals at highest risk, the cost of prevention would be:

$$\begin{array}{l}=[(\$26)+(1-0.2)(\$90.63)+(1-0.223)(42\%)(\$1,641.15)]=\$634.08,\text{compared}\text{to}\\ =[(42\%)(\$1,641.15)]=\$689.28\text{for}“\text{wait}\text{and}\text{treat}.”\end{array}$$

Applied across 1,000 individuals, 327 individuals would require in-person psychotherapy (after receiving preventive and GSH interventions) in the prevention approach versus 420 individuals in “wait and treat.”

Finally, we varied the parameters to estimate the cost of delivering an unguided preventive intervention. For example, the preventive intervention, eBody Project, is an unguided online intervention that cost $0 to provide and had an effect of d=.33 compared to brochure control on ED symptoms.⁴⁵ Assuming the same 31% ED onset rate,⁴⁴ delivering a prevention intervention to a population of individuals at ED risk would cost $470.71 compared to $508.76 for “wait and treat.”

Application of the Stepped Care Model across a Population of College Students

We applied the stepped care model of online guided preventive and GSH interventions to 1,000 college students. Using prevalence rates from our previous work implementing an online screen on two US college campuses,²⁸ 0.7% of students screened positive for anorexia nervosa, 4.7% screened positive for EDs excluding anorexia nervosa, 31.8% screened as high risk for an ED, and 62.9% screened as low risk for an ED. Estimates for the sum cost of GSH intervention ($1,365.80) and prevention intervention ($505.86) were applied from the previous calculations. For individuals with anorexia nervosa, we used the cost from Egger and colleagues of $4,109.56 (converted to 2016 US dollars from €2,494) for delivering in-person enhanced cognitive behavioral therapy.⁴⁶

Comparing the approaches, the stepped care model would cost = [(7)($4,109.56) + (47)($1,365.80) + (318)($505.86)] = $253,823, compared to standard care = [(7)($4,109.56) + (47)($1,641.15) + (31%)(318)($1,641.15)] = $267,686. This shows cost-savings of $13,862.54 for the stepped care model among 1,000 students. With the stepped care approach, 37 (=47*0.777) individuals with EDs and 77 (=318*0.241) individuals at risk would need in-person psychotherapy, equating to 114 individuals total. Comparatively, in standard care, 146 (=47+99) individuals would need in-person psychotherapy.

Discussion

We estimated the costs of implementing a stepped care model for online ED prevention and treatment among US college students. Results showed that the cost of online GSH and online preventive intervention was less than the alternative, standard approach. Combined, the stepped care model—applied across a population of 1,000 students—was estimated to cost less and result in fewer individuals needing in-person psychotherapy (after receiving less-intensive interventions) compared to standard care on college campuses, although the reliability of these estimates needs to be tested with sensitivity analyses. Taken together, this work has important implications for discussions regarding allocation of resources and adoption of efficient care delivery models for EDs on college campuses.

We showed that implementing the stepped care model would yield an estimated cost savings of $13,862.54 compared to standard care for 1,000 college students. However, these results are based on estimated, not actual, costs. One assumption of our results is that intervention effects from efficacy testing will hold with future implementation, which may not prove true.⁴⁷ In evaluating the model across a population of students, assumptions about costs may not generalize to other countries (e.g., with universal healthcare), and assumptions about prevalence rates/uptake into treatment may not replicate. For example, one study showed that individuals who screened positive for an ED were more likely to enroll in an online intervention than individuals at high or low ED risk.²¹

Additionally, our model assumed that individuals with anorexia nervosa would receive outpatient psychotherapy only. Outpatient treatment may be appropriate for some individuals with anorexia nervosa, whereas higher (more costly) levels of care may be indicated at times for others. Country or healthcare system practice guidelines may also impact the types of care utilized for treating anorexia nervosa; for example, in Germany, inpatient treatment is recommended initially as treatment for anorexia nervosa.⁴⁸ Indeed, our model is limited in that we do not estimate the costs associated with higher levels of care for anorexia nervosa or other EDs (e.g., hospitalization, which bears high costs), medical expenditures that can result from EDs, or indirect costs associated with lost productivity, all of which increase total costs,^46,49 meaning our calculation is an underestimate. We also do not account for the cost savings of avoiding excess medical utilization through prevention and early intervention associated with the stepped care model.

Actual costs could be lower for other reasons. Though digital technologies can theoretically offset problems with dropout from care (see Figure 1), many students decline clinical services offered through screens,^18,50 and online and in-person GSH interventions have high drop-out rates.⁵¹ Students in a “wait and treat” approach might also not seek treatment. In these instances, costs are reduced at the expense of effectiveness. It is possible students will be unwilling to enroll in a higher level of care after “failing” a less-intensive intervention; alternatively, tailored recommendations, reduced stigma, and greater understanding of the illness/need for treatment may increase the likelihood students engage in subsequent care.

An additional limitation is that we do not include the cost of screening. Siphoning individuals to appropriate intervention requires an efficient screen, particularly when aiming to reach an at-risk population. True costs must account for screen implementation. For example, universal online screening—which has been shown to yield high completion rates²⁸—could be automated through pre-programmed campus-wide email alerts, which would incur minimal costs for staff time to program. Minimizing staff burden is important, as one study showed that secondary high schools were less willing to agree to advertise for an online ED intervention if they had to engage in intensive dissemination strategies (although more intensive, expensive strategies yielded highest uptake).³⁴ Costs must also account for the screen’s precision. An imperfect screen could increase costs by yielding incorrectly identified cases (to whom interventions would be offered when care may not be needed) or missed cases (for whom more intensive services may be later required due to missed earlier intervention that would have been offered sooner if those cases had been detected).⁵²

Taken together, our results highlight the need for ongoing research on cost-effectiveness, precision of ED symptom identification, and intervention efficacy.⁵³ Improved efficacy of lower-intensity interventions would yield greater cost savings within stepped care models and expand the reach of individuals receiving care.

Considerations for a Disseminable Delivery Model

For this system to be disseminated, a delivery model must be established. Costs include hosting and maintaining the online intervention on a web-based platform (e.g., maintaining security and technological functioning, implementing technological updates) and providing online coaching to students. We calculated cost estimates from the payer perspective of both a university and healthcare system. Thus, for this delivery model, a health insurance plan would need to either (a) develop online screening and interventions, or (b) contract these services to a vendor. Companies may differ in their fee structure; for example, they may charge a per-student rate with unlimited enrollment capacity, or charge a fixed price for a capped amount of students and a supplemental per-student rate for additional users.

Dissemination models must account for costs to the university for the time a health/counseling center clinician might spend liaising with the vendor or engaging in implementation activities to promote the screen and/or interventions, which are not included in our models. In a system in which a university bears healthcare costs and can recoup savings, the cost savings of the stepped care model could be applied towards a staff member’s salary for this purpose. For example, in health homes, costs saved from addressing mental health problems can offset costs of integrating behavioral health specialists.⁵⁴ However, in a system with private or nationalized healthcare, a university must consider whether resources are available and can be allocated.

It may also be advantageous to advocate for policies that continue to expand insurance coverage for online GSH and preventive interventions.

Future Research and Policy Efforts

Results from this work spur several areas for future study. First, a formal economic evaluation is warranted on the stepped care model compared to standard care. Several factors should be included to strengthen the analyses and policy relevance. First, the analysis should take a societal approach to account for costs relevant to multiple stakeholders, such as students, in addition to payers. As noted above, analyses should account for costs of higher levels of care and indirect costs, and should evaluate the reliability of the results via sensitivity analyses. Analyses should also test whether this approach leads to improved academic performance and reduced medical leave/dropout, which has potential for cost savings to a university and cost benefit to society (e.g., increased productivity). Future models should account for cost savings associated with preventing excess medical utilization due to prevention and early intervention, but also consider future costs that may be incurred by improving students’ health.⁵⁵ At the same time, economic analyses may require a time horizon that is relevant to a university, as US students commonly earn their undergraduate degree within four years. Therefore, benefits beyond the traditional college years may be unappealing university stakeholders who desire a return on investment while students remain part of their campus community.

Second, prospective analyses are needed to characterize changes in costs that may occur over time with the use of a stepped care model. Technological innovations could enable more features to become automated, resulting in decreased delivery costs. For example, a fully-automated online intervention with varying levels of interventionist support had higher probabilities of achieving a net benefit across a range of willingness-to-pay thresholds compared to a waiting list among individuals with ED psychopathology.^35,56 Ongoing implementation may also lead to changes in costs over time. For example, online coaches may become more efficient, such as by using a library of feedback messages or clinical management dashboards, which could allow more students to access care. However, there are costs to adapt online technologies to keep pace with modern advances.

Third, despite our findings of cost savings for the stepped care model, we should continue to identify other strategies that reduce costs (e.g., models that more precisely direct resources to vulnerable or high-utilizing populations, development of more effective interventions). Testing the model among populations of users would generate data for subgroup analyses that might identify, for instance, early responders who do not need additional intervention, reducing costs. Improving engagement also warrants attention, as premature dropout may have downstream costs. Finally, we can draw on strategies from global mental health research, such as using non-specialist providers for task-shifting care delivery.⁵⁷ For example, lay providers have demonstrated efficacy in delivering in-person preventive interventions⁵⁸ and GSH interventions,^59,60 and could be trained to deliver online coaching.

Fourth, our cost modeling exercise suggests economic costs favor prevention and early intervention using a stepped care model. Cost is a key consideration in the translation from research to policy;^38,61 however, enacting real change also needs political will.³⁸ We must advocate for mental health as an economic and social priority.⁶² More conversations are needed between university stakeholders, the ED research community, and technology partners to integrate stepped care models onto college campuses. If we identify targeted research questions that remain unanswered, we can design and implement strategic studies^38,61 that will derive the necessary evidence base to ensure college students at risk for or with an ED receive appropriate care.

Concluding Comments

At the start of this paper, we asserted our belief that all individuals in need of treatment should be able to receive accessible, affordable, and evidence-based care. However, the ED field must consider whether belief in the right to treatment extends to the right to prevention. In our stepped care model, we include prevention to reduce both the incidence and prevalence of EDs, as prevention and early intervention are imperative for halting the progression of symptoms. Indeed, left untreated, individuals with EDs may necessitate higher levels of care, which incur the highest intervention costs. Our estimates showed that a stepped care approach that includes both online prevention and treatment would result in modest cost savings and fewer people needing in-person treatment compared to standard care. However, future efforts are needed to systematically measure the costs and benefits of a stepped care model on college campuses, test these parameters using sensitivity analyses, and develop and evaluate less expensive models.