The Centers for Medicare and Medicaid Services (CMS) recently issued the final rule for Medicare Diabetes Prevention Program (MDPP) coverage beginning April 2018,1 offering an unprecedented opportunity to prevent diabetes among the estimated 48.3% of seniors with prediabetes.2 The MDPP is a structured group class based on the National Diabetes Prevention Program (NDPP), a widely disseminated, evidence-based lifestyle intervention.3 The NDPP aims to help participants prevent or delay onset of type 2 diabetes by achieving at least 5% weight loss.
The MDPP pay-for-performance methodology reimburses suppliers based on beneficiaries’ attendance and weight loss outcomes.1 Sustainable reimbursement rates are critical for policy impact, yet whether Medicare payments are sufficient to cover costs of service delivery is unknown, particularly for racial/ethnic minority and low-income individuals who experience disparately high prevalence of type 2 diabetes but lesser NDPP outcomes.2,4–6 We compare projected payments based on performance data of a diverse, underserved sample of Medicare beneficiaries to service delivery costs and explore resulting implications for MDPP sustainability.
Denver Health and Hospital Authority is a safety-net health care system that has provided the NDPP following standards established by the Centers for Disease Control and Prevention (CDC),7 and using the CDC’s publicly available curriculum.3 Sixteen weekly to biweekly sessions were held in months 1–6 and a minimum of 6 monthly sessions were held in months 7–12. NDPP participants were encouraged to attend as many sessions as possible and to lose at least 5% of their initial body weight. Sessions were conducted in neighborhood primary care clinics at a variety of days and times, including evenings and weekends, for convenience. Six yearlong NDPP classes were initially offered in March 2013. Thereafter, 2–4 new NDPP classes were launched approximately each quarter through September 2016, for a total of 45 NDPP classes completed by August 2017.
Five lay health educators completed CDC-compliant training to serve as NDPP coaches, including courses from the Diabetes Training and Technical Assistance Center at Emory University.8 They also received ongoing supervision and training from a certified diabetes educator and health psychologist. Coaches conducted recruitment, led a total of 22–25 sessions per yearlong NDPP class, conducted weekly outreach calls to support engagement and weight loss, and offered individual make up sessions. The program was staffed by an average of 2.3 full-time equivalent coaches between March 2013 and August 2017.
Eligible participants included adults identified as being at risk for developing diabetes (eg, A1c of 5.7–6.4, or a history of gestational diabetes). Recruitment methods were described previously.9 There were no financial incentives or costs to participate. In total, 1165 participants attended at least 1 NDPP session, of whom 213 were Medicare beneficiaries (18.3% of all participants across payer types).
Outcomes measured for these 213 Medicare beneficiaries included the percentage that achieved program attendance and weight loss goals specified by CMS, as well as the associated performance-based payment for achievement of each goal.1 As shown in Table 1, attendance was defined as the number and percentage of participants attending at least 1, 4, or 9 sessions in months 1–6, and at least 2 sessions in months 7–9 and months 10–12, respectively. Weight loss was defined as achievement of at least a 5% or 9% reduction in body weight over the course of the program.
The MDPP pay-for-performance methodology provides a minimum payment of $25 for each participant who attends at least 1 session. A cumulative total of up to $470 per beneficiary is provided for achievement of all performance goals across months 1–12 of the program.1 Up to $200 is further added for attendance and weight loss during maintenance sessions in months 13–24 (for a maximum per-beneficiary payment of $670); however, the NDPP delivery model established by the CDC prescribes only a single year of services and classes offered by Denver Health and Hospital Authority ended after 12 months of sessions. Thus we used $470 as the maximum per-participant payment in this analysis of MDPP sustainability.
We estimated per-capita costs for delivering the NDPP to all participants, accounting for personnel, program development and startup costs, supplies and other direct program delivery costs, and indirect costs (eg, facilities and general administrative expenses). Program development and startup costs were amortized over 5 years. Supplies included material provided to participants and other direct costs included transportation for coaches to deliver sessions at neighborhood clinics. Indirect costs were calculated as 34% of all direct costs using the institution’s federally negotiated indirect rate. We calculated average per-participant program delivery costs for all classes active during calendar years 2014–2016 and adjusted all cost to constant 2017 dollars using the Consumer Price Index for All Urban Consumers. The most relevant marginal cost of providing service is for offering an additional class; thus, we also estimated the cost of delivering each yearlong NDPP class accounting for personnel, program development, supplies, other direct cost, and indirect cost. We then calculated a second estimate of per-participant costs based on the maximum starting class size in our demonstration of NDPP delivery. The Colorado Multiple Institutional Review Board approved this program evaluation project.
Participating Medicare beneficiaries (N=213) were 64.3% female, 40.6% Latino, 31.6% non-Hispanic black, and 26.9% non-Hispanic white. In total, 70% of beneficiaries were low-income (below 133% of federal poverty level), including a majority of individuals in all racial/ethnic groups. Mean age was 59.9 (SD=10.7). The average number of sessions attended by participating Medicare beneficiaries was 8.6 (SD=7.5) and their mean weight loss was 1.8% (SD=4.3). Table 1 shows achievement of MDPP pay-for-performance milestones and associated payments for participating Medicare beneficiaries. The total average per-beneficiary payment is $138.52 (interquartile range=162.50). Only 4.7% of beneficiaries achieved all milestones associated with the maximum $470 payment.
Average per-capita program delivery costs were estimated at $800 for participants in 2014–2016 classes, which included $431 in expenses for coaching staff (53.9% of total cost), $107 for program management personnel (13.4%), $17 for data collection and reporting staff to meet CDC measurement requirements (2.1%), $5 for travel of staff to primary care clinics (0.1%), $25 for supplies and other direct program costs (3.1%), $12 for program development and startup costs (1.5%), and $203 for indirect expenses (25.4%). We also estimated the cost of delivering each yearlong NDPP class at $18,092, which included $9838 in expenses for coaching staff, $2354 for program management personnel, $382 for data collection and reporting staff, $108 for travel, $560 for supplies and other direct program costs, $261 for program development and startup costs, and $4589 for indirect expenses. The maximum starting class size in our demonstration of the NDPP was 42 participants (M=25.9) across all payer types. Assuming the maximum number of participants per class yields a second per-participant cost estimate of $431.
As summarized in Table 2, the average per-beneficiary payment is $661 lower than the average per-participant program delivery cost. Assuming every NDPP class started with our demonstrated maximum of 42 participants, there is a $292 gap between the per-participant cost and average per-beneficiary payment.
Our experience serving a diverse and predominantly low-income group of Medicare beneficiaries in the NDPP indicates that the current CMS performance-based payments are financially unsustainable. Other suppliers may incur a smaller gap and replication of findings is needed. Nonetheless, the pay-for-performance methodology seems unlikely to adequately reimburse suppliers overall given other reports on cost and participant performance to date.
Standard NDPP costs have not been definitively established and likely vary depending on factors including wages, setting, number and frequency of classes, and participants served. A 2015 systematic review found a median cost of $417, but without a comprehensive valuation of all expenses among included studies.10 CMS further noted that multiple commenters reported NDPP delivery costs as “greater than $500 per beneficiary,” exceeding the $470 payment cap for months 1–12.1 Cost of providing the additional year of MDPP services is also unknown. Furthermore, costs may be greater when delivering the MDPP to diverse and low-income populations, including more staff resources for recruitment, ensuring understanding of the curriculum, assisting with transportation needs, and completing make up sessions. Although a consistently large starting class size may reduce the per-participant cost and ensuing gap, a relatively modest class size of 10–20 NDPP participants has been considered optimal.11,12
A pay-for-performance model could promote more cost-efficient delivery and better participant outcomes than we obtained, yet known limitations in NDPP effectiveness make optimal performance improbable. For example, achieving 5% weight loss largely drives performance payments, but only 35.5% of NDPP participants have achieved this goal nationwide.5 Moreover, it is now well-documented that NDPP outcomes are disparate for racial/ethnic minority participants and non-Hispanic whites who are low-income,5,6,13 which would yield lesser performance-based payments to suppliers serving these populations. For example, among non-Hispanic white participants who complete the NDPP, those who are low-income achieve only a quarter of the weight loss as their higher-income counterparts.6
Numerous MDPP suppliers are needed to reach all Medicare beneficiaries with prediabetes, yet gaps between service costs and anticipated payments may deter new suppliers from offering the MDPP and challenge the financial stability of existing providers. Moreover, health disparities may widen as suppliers serving diverse, underserved populations will likely receive especially low payments, further threatening their MDPP delivery. As a result, lack of MDPP access seems most likely for diverse, low-income beneficiaries who have the highest risks and greatest need for services.
CMS noted that they “may consider proposing additional payment policies for the MDPP expanded model in the future,”1 which our findings suggest is needed to ensure access and sustainability. Potential solutions include revising the payment methodology to account for social risk factors and demographic differences in outcomes, and offering higher performance-based payments and total payment cap to cover per-participant costs. A risk-adjusted model accounting for demographic differences in outcomes may help eliminate disparities while emphasizing performance goals. For example, black participants lose nearly half as much weight as white participants in the NDPP (medians of 2.2% vs. 4.1%),5 such that doubling performance payments for black beneficiaries may be necessary. To achieve a sustainable MDPP that promotes access and cost-efficient services, a nationally representative study of beneficiary performance and comprehensive supplier costs seems needed toward refining payment policies.
Promising return-on-investment results support the actuarial establishment of increased payments for MDPP suppliers. Specifically, a recent report calculated an average reduction in Medicare Part A and B expenditures of $278 per quarter, or $1112 per member per year, over the 3 years following NDPP enrollment relative to a comparison group of beneficiaries.14 Thus, as shown in Table 3, our full cost of $800 per participant would be recouped within a year with savings of $312 and additional savings in the following 2 years, for a cumulative return on investment of 2.96.
The high economic and societal costs of diabetes and the importance of reducing health disparities suggest that financially sustainable payments are necessary to ensure benefit access, while yielding cost savings for Medicare. At the same time, further program improvement is needed to make the program maximally effective for all participating beneficiaries, including those with social risk factors. Potential strategies to improve program outcomes for diverse and low-income participants may include case management to support greater attendance, which is generally known to improve weight loss in the NDPP.5 In addition to enhanced cultural tailoring of the NDPP, developing low-income adaptations have also been suggested, which may include skills-based training in food preparation on a budget and increasing access to resources like food banks and recreation centers.6 More recently, the addition of “presessions” that focus on increasing knowledge and motivation, as well as problem-solving barriers to attendance before the NDPP has been shown to improve outcomes.15
In conclusion, new Medicare coverage of the NDPP presents a historical opportunity to achieve the quadruple aim of improved population health, reduced cost, enhanced patient experiences, and better provider experiences. However, the performance-based reimbursement structure seems to yield relatively low payment rates that are particularly disadvantageous for diverse, low-income populations. Our findings suggest that CMS will need to refine the MDPP performance payment methodology to ensure benefit access and sustainability, as well as reduce diabetes-related health disparities.
2. Centers for Disease Control and Prevention (CDC). National Diabetes
Statistics Report, 2017: Estimates of Diabetes
and its Burden in the United States. Atlanta, GA: CDC, US Department of Health and Human Services; 2017.
4. Agency for Healthcare Research and Quality. National Healthcare Quality Report, 2012. Rockville, MD: US Department of Health and Human Services; 2012.
5. Ely EK, Gruss SM, Luman ET, et al. A national effort to prevent type 2 diabetes
: participant-level evaluation of CDC’s National Diabetes
Prevention Program. Diabetes
6. Ritchie N, Sauder K, Phimphasone-Brady P, et al. Rethinking the National Diabetes
Prevention Program for low-income whites. Diabetes
9. Ritchie N, Swigert T. Establishing an effective primary care provider referral network for the National Diabetes
Prevention Program. AADE Pract. 2016;4:20–25.
10. Li R, Qu S, Zhang P, et al. Economic evaluation of combined diet and physical activity promotion programs to prevent type 2 diabetes
among persons at increased risk: a systematic review for the Community Preventive Services Task Force. Ann Intern Med. 2015;163:452–460.
13. Ritchie ND, Christoe-Frazier L, McFann KK, et al. Effect of the National Diabetes
Prevention Program on weight loss for English-and Spanish-speaking Latinos. Am J Health Promot. 2018;32:812–815.
14. Alva ML, Hoerger TJ, Jeyaraman R, et al. Impact of the YMCA of the USA diabetes
prevention program on Medicare
spending and utilization. Health Aff. 2017;36:417–424.
15. Ritchie ND, Kaufmann PG, Gritz RM, et al. Pre-sessions to the National Diabetes
Prevention Program may be a promising strategy to improve attendance and weight loss outcomes. Am J Health Promot. 2018;1:0890117118786195.