Abstract
For-profit companies addressing disparities in social determinants of health (SDOH), also known as SDOH Industry companies, often lack member-level claims data to evaluate their organizational interventions. Health-related quality of life (HRQOL) measures, such as the Centers for Disease Control and Prevention’s Healthy Days Measure, offer a unique proxy metric to evaluate impact. This retrospective study sought to explore the association between self-reported physically and mentally unhealthy days with health care costs among a Medicare Advantage (MA) population. A cross-sectional study of MA members receptive to a companion care program, and thus likely to have unmet social needs, was conducted. The analysis included members with recorded baseline unhealthy days and complete claims data (n = 2,354). Least squares regression analyses were performed to determine the relationship between baseline medical costs, physically unhealthy days, and mentally unhealthy days. A review of Major Diagnostic Categories (MDCs) was also included to elucidate the strength of the Healthy Days Measure as an indicator of the burden of health conditions. Each additional unhealthy day reported was associated with an increase in 30-day medical costs of $60 and $34 for physically and mentally unhealthy days, respectively. Unhealthy days and costs increased with an increasing number of MDCs. Compared with previous studies linking unhealthy days and health care expenditure, these data reveal the potential for even higher savings by reducing the number of unhealthy days in a high-risk population. This evidence supports using unhealthy days as a HRQOL measure and as an important tool for cost estimations.
Introduction
Health-related quality of life (HRQOL) is a multidimensional concept reflecting an individuals’ physical, mental, emotional, and social health status. Self-reported HRQOL is a useful indicator of the burden of diseases, disabilities, and health-related social needs (HRSN), which are additionally the result of social determinants of health (SDOH). 1,2 HRQOL data are a significant component of public health practice and policy, acting as a tool to track unmet health needs and supplement traditional measures of morbidity and mortality. 3 –5 Many entities, including the Centers for Disease Control and Prevention (CDC) and the World Health Organization, continue to prioritize tracking and improving HRQOL outcomes over time. 6,7
Beyond public health and government institutions, health plans and industry companies alike are making investments to address SDOH and improve HRQOL outcomes for various populations. 8 This has created the for-profit “SDOH industry,” which has grown substantially in recent years with $3.5 billion in funding and a $32.2 billion valuation as of 2023. 8,9 SDOH organizations are in a unique position to improve individual and population health, yet they often lack access to critical claims data, which can provide clear measurement of their impact. Without access to traditional data sources, self-reported proxy measures including HRQOL metrics that predict lagging indicators of health are necessary for the industry to understand their impact and implications on well-being. 10
One widely used HRQOL metric is the Healthy Days Measure, a key component of the CDC HRQOL questionnaire. This measure has demonstrated predictive power on morbidity and mortality and substantial literature links an increase in patient-reported number of unhealthy days to a higher number of chronic conditions and HRSNs. 1,11 –14 Since its inception, the Healthy Days Measure has been used as both an indicator of need and as an intervention outcome measure. 1,11,15 –17 There has even been recent traction by digital health companies to use unhealthy days as a key outcomes metric for evaluation of their behavioral health platforms. 18 This ameliorates the problematic trend of conventional clinical tools failing to sufficiently detect individuals with subclinical symptoms. Several studies have established the validity and reliability of the CDC HRQOL questionnaire for its use in measuring self-perceived health status in older adults and general noninstitutionalized adult populations. 1,11,19,20 Further adding to the validity of this tool, research has found that older adults with functional limitations report more physically and mentally unhealthy days than those without functional limitations. 15
However, limited results link the association between self-reported unhealthy days and health care costs. Two published studies estimated each additional unhealthy day costs the health care system an additional $8 to $16 per patient per month. 13,21 These studies evaluated broad Medicare Advantage (MA) populations, which may reflect an entire payor’s population of both high- and low-risk individuals. In addition, these studies used a methodology of grouping physically and mentally unhealthy days together, which does not allow for insight into the distinct individual contribution of physically or mentally unhealthy days on costs. 13,21
There is value to health plans and SDOH organizations to measure the association between unhealthy days and health care costs among populations that experience higher comorbidities and unmet HRSNs. Population health management strategies often target exclusively higher-risk populations. The rationale for this strategy is straightforward: higher-risk populations typically incur more health care costs, thus addressing this demographic provides the best opportunity to reduce health care costs.
Advancing the understanding of the association between unhealthy days and health care costs can elucidate this measure as a tool to assess SDOH interventions and the companies that offer them. The authors sought to build on this evidence by focusing on a cohort of MA members who are receptive to an SDOH industry company’s companion care program and thus likely to have unmet social needs. The objective of this study was to determine the association of self-reported unhealthy days, both physically and mentally, with health care costs in this MA population.
Methods
Study population
A companion care company pairs older adults, families, and other underserved people with a vetted individual (‘‘Pal’’) to provide companionship and assistance with everyday tasks. These tasks include grocery shopping, transportation to doctors’ appointments, assistance with picking up prescriptions, help around the home, and managing escalations for high-risk unmet social needs or changes in clinical conditions. Pals are screened, trained, and hired by the companion care company and are often from the same communities that they serve. Visits can be done in person or telephonically.
A regional Medicare Advantage organization (MAO), located in Northeast Ohio, partnered with the companion care company to provide companion care free of charge to roughly 23,000 MA members. Participants had free access to a certain number of companion care hours each calendar year according to their health plan’s package with the service (ranging from 60 to 100 hours per year). From January through December 2021, the companion care company called all members for enrollment into the companion care program (CCP). There were no specific eligibility criteria, all MA members had access and could enroll in the CCP if interested. During the intake enrollment call, members were surveyed using the CDC’s Healthy Days Measure and were asked to report physically unhealthy days (“Now thinking about your physical health, which includes physical illness and injury, for how many days during the past 30 days was your physical health not good?”; range: 0–30 days) and mentally unhealthy days (“Now thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?”; range: 0–30 days). 1
This study was reviewed by Advarra’s Institutional Review Board (IRB) (Pro00067236) and deemed to be exempt from IRB oversight in accordance with the US Department of Health and Human Services regulations at 45 CFR 46.
Data collection
The MAO provided administrative claims data and enrollment data, and the companion care company provided baseline survey data. A national actuarial firm consolidated the data and matched records. Members were included in the analysis if they had recorded baseline physically unhealthy days data, recorded baseline mentally unhealthy days data, and complete claims data for the 30-day period prior to enrollment with the CCP (n = 2,354). Due to the lag in claims data, members who joined the CCP on or after 10/1/2021 were not included as this did not allow for complete claims data. Members with incomplete member identification information were also excluded from the analysis (n = 4). Baseline medical costs were calculated using allowed medical costs for the 30-day period before the member’s enrollment with the CCP and including the date of enrollment with the CCP.
Data analysis
Least squares regression analyses were performed to determine the relationship between baseline medical costs, physically unhealthy days, and mentally unhealthy days. Baseline medical costs were modeled as a function of physically unhealthy days and mentally unhealthy days in separate univariate regression models as well as together in a multiple-variable regression model.
To understand the burden of health conditions and their relationship with unhealthy days and costs, this study used information about Major Diagnostic Categories (MDCs) from claims data. MDCs are formed by dividing all possible principal diagnoses (from International Classification of Diseases 10th Revision) into 25 mutually exclusive diagnosis areas. For this subanalysis, the authors reviewed MDCs occurring between 2018 and 2020, before the CCP enrollment year of 2021, to substantiate the incidence of a condition that could drive unhealthy days before members self-reported their unhealthy days. This analysis included the subset of members with at least one MDC before the CCP enrollment year and complete CCP enrollment data (n = 2,059).
There are no generally accepted cut points for analyzing medical costs. In determining cut points used for outcome reporting, the costs were broken up into categories that represent different levels of overall health care resource consumption ranging from negligible ($0 to $500), low (>$500 to $2,000), below average (>$2,000 to $10,000), above average (>$10,000 to $50,000), and very high (>$50,000 to $95,000).
Lastly, in order to understand the socioeconomic status of this population and the potential generalizability to other MA populations, members were classified as having low income if they received “Extra Help” through the Medicare Part D Low-Income Subsidy program. These enrollees were identified using enrollment data.
Results
A total of 2,354 unique members were included in the analysis (median age 79 years; 63% female) (Table 1). On average, members reported 7.9 (SD, 10.7) and 4.5 (SD, 8.2) physically and mentally unhealthy days over 30 days, respectively. Baseline 30-day medical costs were highly varied, ranging from $0 to $94,289. Mean baseline 30-day medical costs were $1,428. When cost outliers (30-day cost exceeding $50,000, n = 3) were removed, the mean 30-day cost per member was reduced to $1,320. As a point of reference, in 2021, the average monthly expenditure for a MA plan member was $1,092 according to the 2022 Medicare Trustees Report. 22 When evaluating the socioeconomic status of this population, 11.3% of the study population were eligible for the Medicare Low-Income Subsidy. For comparison, 6.4% of the entire health plan population was eligible for the Low-Income Subsidy, suggesting that those who have low income are roughly twice as likely to engage with the CCP.
Sample Demographics a
A small subset of members (n = 12) had missing age and sex data as reported.
CDC, Centers for Disease Control and Prevention.
Table 2 provides a summary of the number of members, average baseline 30-day medical costs per member, and average unhealthy days reported. A relationship between unhealthy days and medical costs was determined: as medical costs increase, so do reported physically and mentally unhealthy days. In separate univariate regression analysis, baseline costs increased $70.62 (95% CI: 51.18 to 90.06; P < 0.001; adjusted R-squared = 0.021) for every additional physically unhealthy day reported and $65.21 (95% CI: 39.59 to 90.82; P < 0.001; adjusted R-squared = 0.01) for every additional mentally unhealthy day reported. In a multiple variable regression analysis, a change in one physically unhealthy day was associated with a $60.38 change (95% CI: 39.21 to 81.55; P < 0.001) in baseline cost, and a change in one mentally unhealthy day was associated with $33.72 change (95% CI: 5.98 to 61.46; P < 0.02) in baseline cost; adjusted R-squared = 0.023.
Summary of Baseline Medical Costs and Unhealthy Days
Table 3 provides a distribution summary of the number of distinct MDCs, average baseline 30-day medical costs per member, and average unhealthy days reported for the subset of the population who had a MDC and complete CCP enrollment information. As distinct MDCs increase, physically unhealthy days, mentally unhealthy days, and costs all monotonically increase.
Summary of Number of MDCs, Baseline Unhealthy Days, and Baseline Costs a
n = 2059 members had at least one MDC before enrollment and complete companion care program enrollment data.
MDC, Major Diagnostic Categories.
Discussion
This study found that each additional unhealthy day reported was associated with an increase in 30-day medical costs of $60 and $34 for physically and mentally unhealthy days, respectively. These findings enhance ongoing work to understand HRQOL measures that can serve as leading indicators to better identify members with poorer health outcomes, higher risk, and associated increasing health care costs. However, compared with the previous studies linking unhealthy days and health care expenditure (additional $8 to $16 per member per month), 13,21 these data reveal the potential for even higher savings (additional $60 and $34 for physically and mentally unhealthy days, respectively, per member per month) by reducing the number of unhealthy days in a higher risk population.
Previous studies drew upon a broad population of MA members and reported each unhealthy day added $8 to $16 in average medical costs per patient per month. 13,21 In these cost analyses, authors used CDC methodology to calculate total unhealthy days, combining both physically and mentally unhealthy days together with 30-day cap. Reporting outcomes for only the combined unhealthy days obscures the distinct health and financial impact of each perceived component. The model used in the current study allows for a deeper understanding of this impact. This study highlights that older adults who are likely to have unmet social needs may be more indicative of the potential savings by addressing the MA population in need.
As evidenced by the MDC analysis, as the burden of health conditions increases, unhealthy days and costs also increase. Chronic conditions are a driving factor of HRQOL and these results support published studies that found higher unhealthy days to be associated with a diagnosis of chronic conditions. 13,23,24 This signals the strength of the Healthy Days Measure as a proxy for the burden of chronic diseases and other health conditions. A potential future analysis, and an additional strategy to validate the Healthy Days Measure, would be to compare physically and mentally unhealthy days with Medicare Health Outcomes Survey (HOS) data.
These findings should be interpreted with an awareness of several limitations. While the cross-sectional study design is useful to establish preliminary evidence, future longitudinal studies are necessary to better understand if there is a significant causal and temporal relationship between unhealthy days and medical costs. Future studies should evaluate the longitudinal change in medical costs and unhealthy days. The results may incur the issue of selection bias as CCP members self-select into the program by picking up the enrollment phone call and answering the enrollment survey questions.
However, this study aims to evaluate members who are likely at greater risk for unmet social needs. Future studies that evaluate higher-risk members may validate the estimates from this analysis. Since members enrolled, and thus were surveyed, at various points in time throughout the year, the current study cannot control for any differences in survey data or medical costs due to seasonality or other time-based efforts by providers or the health plan. In addition, these data were collected while the COVID-19 pandemic was still occurring and potential implications of this are likely multifactorial in nature. Although the CDC’s Healthy Days Measure is a validated and reliable HRQOL metric, the self-reported nature of the tool still lends itself to the typical biases present in patient-reported outcomes. Further, because both the Healthy Days Measure and medical costs in this analysis evaluated a 30-day period, findings might not be generalizable beyond this time frame. Lastly, the study population was comprised of MA members from one Ohio-based health plan, thus findings might not be generalizable to other plan types or regions.
Conclusions
HRQOL measures, like the CDC’s Healthy Days Measure, are a potential opportunity for SDOH industry companies who lack member-level claims data to stratify their population’s risk level and evaluate their organizational interventions. A better understanding of the relationship between self-reported unhealthy days and medical costs has important implications for SDOH organizations that may use the Healthy Days Measure as a proxy for its impact on health outcomes and associated costs. This evidence supports using unhealthy days as a HRQOL measure and shows its strength as an important tool for cost estimations. Improvements in HRQOL metrics, like those undertaken by the companion care company, could lead to higher savings than those originally identified in previous studies. 13,21 What does this say about the true value of SDOH intervention companies? You can only know what you measure, and this evidence paves the way for more investigation and discussion. Industry alignment on SDOH outcome measures and efforts to increase utilization of the Healthy Days Measure is critical for patients and providers alike.
Footnotes
Acknowledgment
The authors wish to thank Anne Armao and Kerri Towsley for their partnership, which made this study possible, An-Vy Hoang and Andrew Marshall for analytic support, and Susan Pantely, FSA, MAAA for actuarial expertise.
Authors’ Contributions
K.C.M.: Conceptualization, investigation, writing—original draft, project administration. E.T.R.: Conceptualization, investigation, writing—original draft. J.R.: Data Curation, formal analysis. Z.N.G.: Conceptualization, writing—original draft. H.S.F.: Formal analysis. P.N.: Formal analysis. D.B.N.: Conceptualization, writing—review and editing.
Role of the Funder/Sponsor
The authors received no specific funding for this work.
Data Availability Statement
The datasets generated and analyzed during this study are not publicly available due to the nature of HIPAA agreements with clients.
Author Disclosure Statement
K.C.M. reports a relationship with Papa Inc. that includes: employment and equity or stocks. E.T.R. reports a relationship with Papa Inc. that includes: employment during the time of this study. J.R., H.S.F., and P.N. reports a relationship with Papa Inc. that includes: consulting or advisory.
Funding Information
The authors received no specific funding for this work.