糖心vlog

Key Points

Question听 What potential changes in health care utilization and expenditures would be associated with national implementation of medically tailored meal (MTM) access for patients with diet-related diseases and limited instrumental activities of daily living who have Medicaid, Medicare, or private insurance?

Findings听 This economic evaluation among 6鈥�309鈥�998 eligible US adults found that national implementation of MTMs for patients with diet-sensitive conditions and activity limitations could potentially be associated with 1.6 million averted hospitalizations and net cost savings of $13.6 billion annually from an insurer perspective.

Meaning听 These findings suggest that national MTM coverage would be associated with cost savings for a range of circumstances if appropriately targeted.

Importance听 Medically tailored meals (MTMs) are associated with lower health care utilization among patients with complex diet-related diseases but are not a covered benefit in Medicare or Medicaid. The potential impact of extending insurance coverage for MTMs nationally remains unknown.

Objective听 To estimate 1- and 10-year potential changes in annual hospitalizations, potential changes in annual health care expenditures, and overall policy cost-effectiveness associated with national MTM coverage for US patients with diet-related disease and limited instrumental activities of daily living who have Medicaid, Medicare, or private insurance.

Design, Setting, and Participants听 In this economic evaluation, conducted from January 2021 to February 2022, a nationally representative sample from the 2019 Medical Expenditure Panel Survey was used to create a population-level cohort policy simulation model that estimated changes in annual hospitalizations and health care expenditures associated with coverage of MTMs. Participants were 6鈥�309鈥�998 US adults aged 18 years or older who had Medicare, Medicaid, or private payer insurance and at least 1 diet-sensitive condition and 1 limitation in instrumental activities of daily living.

Interventions听 Ten nutritionally tailored MTMs per week for a mean of 8 months in each year of intervention.

Main Outcomes and Measures听 The main outcomes were total hospitalizations, program costs, health care expenditures, and net policy costs. One thousand Monte Carlo simulations for each of 10 years (2019-2028) jointly incorporated uncertainty in model inputs for effect sizes, hospitalizations, health care expenditures, and program costs.

Results听 At the 2019 baseline, an estimated 6鈥�309鈥�998 US adults were eligible to receive MTMs. Mean (SD) age was 68.1 (16.6) years; most were female (63.4%), were non-Hispanic White (66.7%), and had Medicare and/or Medicaid (76.5%). The most common eligibility diagnoses were cardiovascular diseases (70.6%), diabetes (44.9%), and cancer (37.2%). If all eligible individuals received MTMs, an estimated 1鈥�594鈥�000 hospitalizations (95% uncertainty interval [UI], 1鈥�297鈥�000-1鈥�912鈥�000) and $38.7 billion (95% UI, $24.9 billion to $53.9 billion) in health care expenditures could potentially be averted in 1 year. Program costs were $24.8 billion (95% UI, $23.1 billion to $26.8 billion), for an associated net savings of $13.6 billion (95% UI, $0.2 billion to $28.5 billion) from a health care perspective. In 2019 dollars, 10 years of the MTM intervention was anticipated to cost $298.7 billion (95% UI, $279.7 billion to $317.4 billion) and to potentially be associated with 18鈥�257鈥�000 averted hospitalizations (95% UI, 14鈥�690鈥�000-22鈥�109鈥�000) and reductions in health care expenditures of $484.5 billion (95% UI, $310.2 billion to $678.4 billion), for net savings of $185.1 billion (95% UI, $12.9 billion to $377.8 billion). Findings were robust in multiple sensitivity analyses.

Conclusions and Relevance听 The findings suggest that national implementation of MTMs for patients with diet-sensitive conditions and activity limitations could potentially be associated with approximately 1.6 million averted hospitalizations and net cost savings of $13.6 billion annually. The results may inform US state, federal, and private-payer interest in expanding insurance coverage for MTMs among patients with diet-related chronic illness.

Despite a suboptimal diet being a leading factor associated with morbidity and mortality in the US,¹ the health care system has traditionally had few tools to connect patients with chronic illness to nutrition services.² The availability of nutrition programs is rapidly growing due to interest among health care systems, payers, patients, and policy makers in 鈥渇ood is medicine鈥� interventions, such as medically tailored meals (MTMs), as potential tools to improve nutrition security, health outcomes, and health care utilization for these patients.^2,3 Medically tailored meals are fully prepared, nutritionally tailored, and generally home-delivered healthy meals for individuals living with advanced and costly diet-sensitive conditions, such as diabetes, heart failure, end-stage kidney disease, HIV infection, and cancer. Programs generally provide 10 weekly meals (lunch and dinner for 5 days per week) designed by a registered dietitian based on disease diagnosis and nutritional assessment.⁴ Medically tailored meal programs are often designed to treat individuals with lower income, food insecurity, and/or limitations in instrumental activities of daily living (IADLs) that make it difficult to prepare healthy meals.

In observational studies and pilot randomized clinical trials, patients receiving MTMs experienced better disease management and had fewer hospitalizations, emergency department admissions, nursing home visits, and lower health care expenditures compared with similar control patients.^5-10 Compared with no receipt of MTMs, MTM receipt has been associated with a 37% to 52% lower risk of hospitalization, 16% to 31% reduction in monthly health care expenditures,^5-8 and decreased net costs of approximately $2500 per patient-year after paying for meal costs.^5,6

Medically tailored meals, especially those evaluated in the literature, are generally provided by community-based organizations supported by grants, donations, and additional ad hoc restricted funding from home health care services benefits, Medicare Advantage programs, or state Section 1115 waivers allowing coverage of MTMs.^11,12 Medically tailored meals are not currently a covered benefit in Medicaid or Medicare. Bills at both state and federal levels have recently proposed expanded access to MTMs in Medicaid and Medicare, but these have not passed.^13,14 Availability of MTMs in federal programs depends on (1) whether regulatory flexibility or other special circumstances permit their inclusion and, if permitted, (2) whether participating private health care entities choose to cover them.^2,12

Given the limited coverage for MTMs nationally, this treatment is unavailable to most US individuals who might benefit. To our knowledge, no previous research has modeled expected changes in health care expenditures and hospitalizations if MTMs were covered nationally by health insurance for the population routinely served by existing MTM organizations. The objective of this study was to estimate 1- and 10-year associations of MTMs with hospitalizations, health care expenditures, and net costs among patients with diet-related diseases and limitations in IADLs who are covered by Medicaid, Medicare, and private insurance.

In this economic evaluation, the study sample was drawn from the 2019 Medical Expenditure Panel Survey (MEPS), a nationally representative survey of health care utilization and costs among noninstitutionalized US adults. The MEPS provides detailed individual-level data, including age, sex, race and ethnicity, family income, geographic region, IADL limitations, medical conditions, health care utilization, and costs. Race and ethnicity were assessed by self-report within the MEPS data set and were included to better understand the study sample and the potential implications expanded MTM coverage may have for health equity. This study was conducted from January 2021 to February 2022 and was approved by the Tufts Health Sciences Institutional Review Board, which deemed informed consent not applicable because the study was considered non鈥揾uman participants research based on use of publicly available, deidentified data. We followed the Consolidated Health Economic Evaluation Reporting Standards () reporting guideline.

We modeled eligible patients as adults aged 18 years or older who were covered by Medicare, Medicaid, or private payers and had 1 or more diet-sensitive conditions and 1 or more IADL limitations. Diet-sensitive conditions included diabetes, congestive heart failure, myocardial infarction, other heart disease, emphysema, and stroke, as defined as priority conditions in the MEPS 2019 Combined File,¹⁵ and nonmelanoma cancer, chronic kidney disease, and HIV infection, as defined by International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes in the MEPS 2019 Medical Conditions File.^16,17 These diagnoses were selected because they reflect the patient populations in previous MTM research^5-10,18-20 and are served by the national network of MTM providers.⁴ The IADL limitations were defined as a positive survey response to receiving help or supervision using the telephone, paying bills, taking medications, preparing light meals, doing laundry, or going shopping due to an impairment or health problem.¹⁵

We assumed meals were medically tailored and provided only for the index patient (ie, no meals were provided for other household members, as is common in most MTM insurance contracts). In practice, MTM organizations often create 10 to 15 meal plans per day, with registered dietitian nutritionists tailoring ratios of macronutrients and micronutrients for specific diagnoses, incorporating optimal quantities of healthy food groups such as fruits and vegetables, accounting for dietary preferences such as vegetarian options, and providing options for individuals who have challenges chewing solid foods. The MTM dietary guidelines for common diagnoses are provided on the Food Is Medicine Coalition website.²¹

We conducted a PubMed search for studies that measured the association of MTMs with hospitalizations and/or health care expenditures in the US in the past 20 years and identified 5 studies.^5-8,22 Using these studies, we calculated the weighted mean duration of MTM receipt as 8 months of MTMs per patient per year (eTable 1 in the Supplement). We also conducted an original meta-analysis to estimate policy effect sizes by pooling findings from published interventional studies^5-8,22 using inverse variance-weighted meta-analysis with random effects. We conservatively assumed that benefits of MTMs would occur only in the year of MTM provision, with no sustained or carryover benefits into the following year.

Annual MTM program costs included clinical screening costs and meal costs (Table 1). Screening costs were based on 2020 registered dietitian facility Medicare reimbursement rates for an initial, 15-minute medical nutritional therapy session at approximately $30 per patient (ranged from $27.34 in Mississippi to $35.17 in Santa Clara County, California).²³ Meal costs were based on 2019 insurance contracts from 10 major MTM organizations and included nutritional tailoring, ingredients, and labor, administrative, and delivery costs. In our 1-year and 10-year models, we calculated net policy costs as the sum of changes in health care expenditures associated with MTM receipt and total MTM program costs.

For the 10-year policy model, we assessed historical trends in population size and annual health care expenditures among MTM-eligible patients from 2010 to 2019 using MEPS data. We adjusted each year鈥檚 expenditures to 2019 US dollars²⁴ using log-linear regression stratified by insurance status for the eligible population and used these inflation-adjusted trends to project the 2020 to 2028 expenditures and population size from the 2019 MEPS data. Given the specific patient population eligible for MTMs, we chose this empirical forecasting approach using MEPS data rather than incorporating expected trends for the general US population. For the 10-year model, we assumed all eligible participants received meals for 8 months per year in each year. We modeled an open cohort in which the population size for each year reflected people newly eligible entering and those who were no longer eligible leaving.

We used a population-level, cohort policy simulation model to estimate the change in hospitalizations and health care expenditures that might occur after implementing an MTM policy compared with the status quo (ie, no new MTM policy), programmed in R, version 4.1.2 (R Project for Statistical Computing). For the 1-year (2019) and 10-year (2019-2028) time horizons, model outputs included changes in annual inpatient hospitalizations and health care expenditures, MTM program costs, and net policy costs from the health care perspective, separately analyzed among Medicaid, Medicare, dual-eligible, and privately insured patients. Model inputs included annual hospitalizations and expenditures from MEPS, relative risks of hospitalizations and the percentage change in health care expenditures associated with MTM receipt, and MTM program costs (Table 1). Medical Expenditure Panel Survey weights were used to scale the sample to be nationally representative.

Simulations were repeated 1000 times in each year so that probabilistic analyses jointly incorporated uncertainty in model inputs for hospitalizations, health care expenditures, effect sizes, and screening and meal costs by drawing randomly from within the plausible range of each input during simulation. Estimates reflect the mean of 1000 Monte Carlo simulations, with the 95% uncertainty interval (UI) defined as the 2.5th percentile to the 97.5th percentile of the simulations. Ten-year estimates summed separate simulations for each year (2019-2028) and assumed all eligible individuals received meals for 8 months in each year, with 3% annual discounting of health care expenditures and MTM program costs.

Because some MTM contracts condition receipt on experiencing food insecurity, a secondary analysis restricted the patient population by adding food insecurity as an eligibility criterion for the 1-year model (using 2017 MEPS data, the most recent data set including food insecurity). Food insecurity is defined by the US Department of Agriculture as 鈥渢he limited or uncertain availability of nutritionally adequate and safe foods.鈥�²⁵ Multiple, 1-way sensitivity analyses assessed the robustness of findings to specific assumptions in the 1-year model, including estimating potential impacts for individuals with diabetes and congestive heart failure only, because these represent the most diet-sensitive conditions among our eligibility criteria. A third 1-year sensitivity analysis assumed only 50% of eligible patients received MTMs. A fourth sensitivity analysis estimated potential policy impact at the 2.5th 10th, 25th, 50th, 75th, 90th, and 97.5th percentiles of the effect size for change in health care expenditures associated with MTM receipt, holding constant all other inputs at their central estimate. Finally, we performed a threshold analysis to estimate the per-meal cost at which the overall 1-year net policy costs would break even (ie, be 0) and a separate threshold analysis to estimate the minimum change in health care expenditures associated with MTM receipt for the overall policy costs to break even, holding constant all other inputs at their central estimate. For the 10-year model, sensitivity analyses included alternative annual discounting rates of 0% and 5% and incorporating a second year of sustained reductions in health care expenditures among 15% of MTM recipients without requiring MTMs in the second year. All statistical analyses were done using R, version 4.1.2.

We estimated that in 2019, 6鈥�309鈥�998 US adults with Medicare, Medicaid, and private insurance would have been eligible to receive MTMs based on having at least 1 diet-sensitive disease and IADL limitation (Table 2). Mean (SD) patient age was 68.1 (16.6) years; 63.4% were female, 36.6% were male, 11.3% were Hispanic, 3.1% were non-Hispanic Asian, 14.2% were non-Hispanic Black, 66.7% were non-Hispanic White, and 4.7% were other (specific groups are unavailable because this category was precoded within the MEPS data set) or multiple races and ethnicities. Median ratio of household income to poverty level was 1.9 (IQR, 0.9-3.7), and 76.5% of patients were covered by Medicare and/or Medicaid. The most common eligibility diagnosis was cardiovascular diseases (70.6%), followed by diabetes (44.9%) and cancer (37.2%) (Table 2). Mean (SD) annual health care expenditures in 2019 were $31鈥�134 ($34鈥�749) per person, including a mean (SD) of 0.98 (1.68) emergency department visits and 0.54 (0.94) hospitalizations per person per year, consistent with the expected high severity of illness and health care utilization in this patient population.

In the meta-analysis of 5 previous studies,^5-8,22 MTM provision was associated with reductions of annual health care expenditures of 19.7% (95% CI, 6.9%-32.4%) and annual hospitalizations of 47.0% (31.7%-62.3%) compared with usual care (eTables 2 and 3 in the Supplement). The pooled mean (SD) per-meal cost was $9.30 ($0.64).

If all eligible individuals received MTMs, the estimated MTM program costs, including clinical screening and meals, would be $24.8 billion (95% UI, $23.1 billion to $26.8 billion), and an estimated 1鈥�594鈥�000 hospitalizations (95% UI, 1鈥�297鈥�000-1鈥�912鈥�000) and $38.7 billion (95% UI, $24.9 billion to $53.9 billion) in health care expenditures would potentially be averted in 1 year (Table 3). Most of the health care expenditure savings (77.0%) would occur in Medicare and Medicaid, totaling $29.8 billion (95% UI, $22.2 billion to $38.2 billion). Summed across all health care payers, the policy was estimated to potentially be associated with net cost savings of $13.6 billion (95% UI, $0.2 billion to $28.5 billion). By payer subsets, 1-year possible policy cost savings were estimated at $3.1 billion (95% UI, 鈭�$2.9 billion to $9.5 billion) for private payers, $3.4 billion (95% UI, 鈭�$5.4 billion to $12.1 billion) for Medicare, $1.7 billion (95% UI, 鈭�$1.1 billion to $5.1 billion) for Medicaid, and $5.9 billion (95% UI, 鈭�$1.9 billion to $14.1 billion) for dual-eligible individuals. In probabilistic analyses combining all insurance strata, the intervention was associated with net cost savings in more than 97% of the simulations (Figure).

When adding food insecurity as an eligibility criterion, eligibility decreased from 6鈥�309鈥�998 to 1鈥�887鈥�681 individuals. Compared with the original population, the smaller population was younger and more likely to be Hispanic or non-Hispanic Black, have lower household income, and be dual-eligible for Medicare and Medicaid (Table 2). Baseline annual emergency department admissions, hospitalizations, and health care expenditures were slightly greater than for the base-case population, although these differences were not statistically significant. In the smaller population, our model estimated that implementation of MTMs would potentially be associated with 506鈥�000 averted hospitalizations (95% UI, 398鈥�000-654鈥�000) and $13.0 billion (95% UI,鈥�$7.9 billion to $18.9 billion) averted health care expenditures in 1 year, with a net policy cost savings of $5.5 billion (95% UI, $0.7 billion to $11.1 billion) (Table 3).

A total of 2鈥�830鈥�507 individuals in the primary population (44.9%) had diabetes. In this population, our model estimated that MTMs could potentially be associated with 701鈥�000 averted hospitalizations (95% UI, 524鈥�000-911鈥�000) and $19.3 billion (95% UI, $12.2 billion to $27.3 billion) averted health care expenditures in 1 year, with a net policy cost savings of $10.9 billion (95% UI, $3.6 billion to $18.8 billion). A total of 1鈥�695鈥�294 individuals had congestive heart failure (26.8%), and in this population, our model estimated that MTM receipt was associated with a potential reduction of 530鈥�000 hospitalizations (95% UI, 373鈥�000-705鈥�000) and $10.9 billion (95% UI, $6.3 billion to $15.6 billion) health care expenditures in 1 year, with a net policy cost savings of $5.8 billion (95% UI, $1.3 billion to $10.6 billion). The 3 secondary populations were anticipated to have between 139% and 173% greater per capita net policy cost savings than the primary population (eTable 4 in the Supplement).

Assuming only 50% coverage of eligible participants in the primary population, the policy was estimated to potentially be associated with 798鈥�000 averted hospitalizations (95% UI, 648鈥�500-956鈥�000) and $19.4 billion (95% UI, $12.5 billion to $26.9 billion) averted health care expenditures, with a net policy cost savings of $6.90 billion (95% UI, $0.10 billion to $14.20 billion) in 1 year. For the 1-year policy to be cost neutral rather than cost saving, the estimated per-meal cost would need to double from $9.30 to $18.89, and the change in health care expenditures associated with MTM receipt would need to decrease by more than one-third, from 19.7% to 12.7%. Across the uncertainty range from the 2.5th to 97.5th percentile of the effect size for change in health care expenditures associated with MTM receipt, most scenarios would be anticipated to be cost saving, with the potential for significantly greater benefits than are reported in our primary findings (eFigure in the Supplement). However, it remains possible that the policy may have a net positive cost if the true effect of MTM receipt on change in health care expenditures is below the 14th percentile of our effect size uncertainty range.

Based on observed national trends with 2019 as the base, we assumed that from 2020 to 2028, the eligible patient population would increase annually by 1.0% among privately insured individuals, 2.1% among those in Medicare, 3.0% among those in Medicaid, and 5.7% among dual-eligible individuals and that per-patient, inflation-adjusted health care expenditures would increase by 1.5% annually among privately insured individuals, 1.7% among those in Medicare, 3.5% among those in Medicaid, and 3.9% among dual-eligible individuals. In 2019 dollars, 10 years of the MTM intervention, in which the target population received MTMs for 8 months per year in each of the 10 years modeled, was estimated to potentially cost $298.7 billion (95% UI, $279.7 billion to $317.4 billion) and to be associated with reductions in hospitalizations of 18鈥�257鈥�000 (95% UI, 14鈥�690鈥�000-22鈥�109鈥�000) (Table 4) and in health care expenditures of $484.5 billion (95% UI, $310.2 billion to $678.4 billion) (eTable 5 in the Supplement). The net cost savings from an insurer perspective would be $185.1 billion (95% UI, $12.9 billion to $377.8 billion) if the target population received MTMs for 8 months per year in each of the 10 years modeled (Table 4).

Applying either 5.0% discounting or no discounting of costs, $441.2 billion (95% UI, $282.7 billion to $617.7 billion) or $558.4 billion (95% UI, $357.3 billion to $782.1 billion), respectively, in health care expenditures would potentially be adverted (eTable 5 in the Supplement). The net 10-year cost savings were $143.7 billion (95 % UI, 鈭�$11.8 billion to $319.4 billion) for 5.0% discounting and $260.7 billion (95% UI, $62.7 billion to $481.5 billion) for no discounting (Table 4). Assuming benefits of MTMs persisted into a second year for 15% of individuals (and with 3% discounting), the estimated 10-year potential net cost savings were $231.5 billion (95% UI, $41.2 billion to $441.4 billion).

Combining nationally representative data on patient eligibility and health care utilization with evidence from interventional studies of MTMs, our simulation model estimated that full national coverage of MTMs in Medicare, Medicaid, and private insurance for patients with both a diet-sensitive condition and an IADL limitation would be associated with meaningful reductions in annual hospitalizations and health care expenditures. Among 6鈥�309鈥�998 eligible recipients, MTMs could possibly be associated with a reduction of 1鈥�594鈥�000 hospitalizations annually. Furthermore, after accounting for the costs of identifying and referring patients to MTM organizations and providing 10 weekly meals for a mean of 8 months annually, the policy was anticipated to be associated with a net savings of $13.6 billion over 1 year and $185.1 billion over 10 years. These findings were robust to a range of sensitivity and scenario analyses.

The provision of prepared meals through health care first arose as a palliative measure for patients with AIDS under the Ryan White Comprehensive AIDS Resources Emergency Act of 1990, at a time when few effective treatments for HIV infection existed. The potential utility of MTMs in clinical care is now supported by interventional studies observing improved diet quality, food security, and disease management when patients with diet-sensitive conditions receive MTMs.^8-10,20,26 Medically tailored meals have been associated with reduced depressive symptoms and fewer dilemmas among paying for food, health care, or prescriptions.²⁰ Receipt of MTMs is also associated with improved disease management. For example, among patients with HIV infection receiving MTMs, antiretroviral therapy adherence increased, and among patients with diabetes, diabetes self-management improved.^20,26 These prior findings suggest that MTMs may be associated with improved health through several pathways, including improved nutrition, improved food security, better financial well-being, reduced stress and anxiety, and improved medication adherence and self-management.

Other research supports additional mechanisms by which MTMs are associated with reduced hospitalizations and lower health care expenditures. For example, patients with advanced cirrhosis and ascites required fewer weekly paracenteses and experienced improved ascites-specific quality of life after 3 months of MTM receipt.⁹ Among patients with recent hospitalization for heart failure, 1 month of MTM receipt was associated with improved clinical symptom and quality-of-life scores on the Kansas City Cardiomyopathy Questionnaire.⁸ In a recent randomized clinical trial, 600 patients hospitalized with chronic heart failure were assigned to receive either usual hospital meals or medically tailored meal plans, nutritional counseling, and if necessary, supplemental intravenous nutrition. The tailored nutritional support led to a 56% reduction in mortality at 30 days.²⁷ Although meals were provided in the hospital rather than delivered at home, this trial supports the benefits of comprehensive, tailored nutritional support for patients with chronic illness who are at high risk of hospitalization.

We modeled MTM provision to patients with severe comorbidities and limitations to independence, and these findings should not be generalized to a healthier population or to a less intensive nutrition intervention. We do not anticipate similar results for individuals with the same diagnoses who do not have limited IADLs. Our main analysis was based on full coverage of all eligible individuals to provide a best-case policy scenario. In practice, it would take time for MTM services to scale and serve all eligible patients. Thus, our scenario analysis of 50% coverage is an alternative benchmark for comparison. Scaling of MTMs may be associated with increased program efficiency and reduced costs, leading to greater net savings. Conversely, scaling may be associated with lower nutritional quality or tailoring of meals, reducing efficacy. These possibilities need to be evaluated with empirical research. Nonetheless, the robustness of our results suggests that national MTM coverage may be cost saving under a range of circumstances if appropriately targeted, including adding food insecurity as a criterion for eligibility or focusing only on patients with diabetes or congestive heart failure. Sensitivity analyses suggest that further targeting to these higher-need populations may be associated with greater per capita cost savings but lower population-level cost savings because fewer people are eligible. Finally, the primary goal of MTMs is to provide high-quality medical care for patients with chronic illness, and the observed reductions in hospitalizations and expenditures do not incorporate potential additional benefits in patient-related quality of life, disease progression, caregiver well-being, and population-level health equity.

This investigation leveraged national data and interventional findings to estimate the health and economic impacts of MTM expansion within Medicaid, Medicare, and private insurance. The findings support policy expansion of access to MTMs by adopting health policy reforms. Several states are currently piloting expanded MTM access, including a $6 million pilot in California for patients with heart failure²⁸; a $149 million pilot of flexible services in Massachusetts that covers nutrition and housing programs including MTMs among patients with Medicaid²⁹; and a similar $650 million Section 1115 waiver in North Carolina鈥檚 Medicaid program that allows payment for MTMs.³⁰ Multiple private payers are experimenting with MTMs through charitable donations and grants.⁴ Kaiser Permanente, the largest health maintenance organization in the US, is undertaking a large clinical trial of MTMs for chronically ill patients.³¹ At the federal level, the Medically Tailored Home-Delivered Meals Demonstration Pilot Act of 2021 has been introduced to direct Medicare to implement and evaluate MTMs, and similar bills have recently been introduced at the state level.^13,14 As of 2020, Medicare Advantage plans may also choose to provide MTMs to certain beneficiaries. Despite this accelerating use of MTMs, access depends on buy-in from state-level Medicaid administrators or managed care plan leadership and is therefore limited by geographic region and/or insurance carriers. The current limitations in national coverage of MTMs present an opportunity to improve health if future policies expand MTM access, with an additional opportunity to improve health equity if such policies prioritize patients with low income and/or food insecurity. Our findings support the use of MTM programs and the need for their timely implementation, scaling, and evaluation in both public and private health care.

The COVID-19 pandemic has further highlighted the need to invest in prevention and treatment of nutrition-sensitive chronic conditions, with a focus on health equity. The US Centers for Disease Control and Prevention has documented diet-related conditions including diabetes, cardiovascular diseases, chronic kidney disease, cancers, and obesity as leading risk factors for COVID-19 hospitalizations and deaths.³² Our research suggests that expanding MTM access should be considered as a health care strategy to improve care for patients with diet-related health conditions.

This study has strengths. We incorporated national data on eligible patients and health care utilization and expenditures, increasing generalizability of the findings. Effect sizes were derived from interventional studies of MTMs and MTM program costs from insurance contracts between MTM providers and health care systems. Patient eligibility criteria were consistent with prior research studies and existing MTM programs. Our policy model included probabilistic sensitivity analyses across 1000 Monte Carlo simulations to jointly incorporate uncertainty and report a range of plausible outcomes, and additional 1-way sensitivity analyses tested the effects of specific assumptions on the results. We estimated both 1-year and 10-year outcomes, providing a range of clinically relevant and policy-relevant time horizons.

The study also has limitations. Although all included MTM studies were interventional, most were observational, with carefully constructed rather than randomized groups. However, existing randomized clinical trials of MTMs that assessed other health outcomes^8-10,27 support the estimated benefits of interventions shown in observational studies. The published literature was insufficient to incorporate other outcomes that may be impacted by MTMs, such as emergency department admissions, nursing home admissions, and patient quality of life. Published MTM evaluations^5,6 report up to 2 years of intervention, and our 10-year estimates could be either overestimates or underestimates of cost savings, depending on whether MTM efficacy strengthens or wanes over time. The eligible population analyzed represents an open cohort, with some newly entering eligible patients and others exiting due to mortality or loss of eligibility. In addition, our base-case 10-year analysis assumed no carryover benefits of receiving MTMs; thus, policy benefits may have been underestimated if some MTM recipients had lasting improvements in health after 1 year of intervention.

Our simulation model of a nationally representative MTM-eligible population with diet-sensitive conditions and IADL limitations estimated that coverage for MTMs in Medicare, Medicaid, and private insurance could potentially be associated with approximately 1.6 million averted hospitalizations and net savings of $13.6 billion in health care costs in the first year. These findings may inform increasing state, federal, and private-payer interest in implementing 鈥渇ood is medicine鈥� interventions such as MTMs to address diet-related chronic illness in the US.

Accepted for Publication: August 30, 2022.

Published: October 17, 2022. doi:10.1001/jamanetworkopen.2022.36898

Open Access: This is an open access article distributed under the terms of the CC-BY License. 漏 2022 Hager K et al. 糖心vlog Open.

Correction: This article was corrected on November 21, 2022, to fix a typo in the first sentence of the Discussion section.

Corresponding Author: Kurt Hager, MS, Friedman School of Nutrition Science and Policy, Tufts University, 150 Harrison Ave, Boston, MA, 02111 (kurt.hager@tufts.edu).

Author Contributions: Mr Hager had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Hager, Cudhea, Wong, Berkowitz, Downer, Mozaffarian.

Acquisition, analysis, or interpretation of data: Hager, Cudhea, Wong, Lauren, Mozaffarian.

Drafting of the manuscript: Hager, Wong.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Hager, Cudhea, Lauren.

Obtained funding: Mozaffarian.

Supervision: Wong, Mozaffarian.

Conflict of Interest Disclosures: Mr Hager reported receiving personal fees from the Aspen Institute outside the submitted work. Dr Berkowitz reported receiving grants from the National Institutes of Health (NIH) and Blue Cross and Blue Shield of North Carolina and personal fees from the Aspen Institute, The Rockefeller Foundation, Kaiser Permanente, and the Gretchen Swanson Center for Nutrition outside the submitted work. Mrs Downer reported receiving grants from the Tufts University Friedman School of Nutrition Science and Policy during the conduct of the study, being employed by the Harvard Law School during the drafting and submission of this work, and receiving personal fees from the Aspen Institute outside the submitted work. Dr Mozaffarian reported receiving grants from the NIH, the Gates Foundation, The Rockefeller Foundation, Vail Institute for Global Research, and Kaiser Permanente; receiving personal fees from Acasti Pharma, Barilla, Danone, and Motif FoodWorks; serving on scientific advisory boards for Beren Therapeutics, Brightseed, Calibrate, DiscernDx, Elysium Health, Filtricine, Foodome, HumanCo, January Inc, Perfect Day, Tiny Organics, DayTwo, and Season Health; having stock ownership in Calibrate and HumanCo; and receiving royalties from UpToDate outside the submitted work. No other disclosures were reported.

Funding/Support: This research was supported by grant 2R01HL115189-06A1 from the National Heart, Lung, and Blood Institute, NIH (Mr Hager, Dr Cudhea, Dr Wong, Mrs Downer, Ms Lauren, and Dr Mozaffarian), by the Horowitz Foundation for Social Policy (Mr Hager), and by grant K23DK109200 from the National Institute of Diabetes and Digestive and Kidney Diseases, NIH (Dr Berkowitz).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: The Food Is Medicine Coalition implemented a survey of member organizations, which allowed us to have accurate cost data for this analysis.