Obesity rates in the United States have reached unprecedented levels, with nearly 42% of adults now classified as obese[1]—a trend that presents profound implications for the health care system. Among Medicare-aged individuals, obesity is closely linked to chronic conditions such as diabetes, cardiovascular disease and musculoskeletal disorders, all of which drive up health care utilization and spending. As policymakers, payers and providers grapple with rising costs, understanding the financial impact of obesity management has never been more critical.
At the same time, the health care landscape is rapidly evolving with the growing availability of weight-loss medications such as Glucagon-like-Peptide-1 (GLP-1) receptor agonists and the expanding role of care management vendors focused on improving outcomes while reducing costs. While these interventions hold promise, their true effect on total cost of care remains uncertain. Do investments in obesity treatment and care management lead to long-term cost savings, or do they introduce new financial burdens to the system? This article aims to explore these questions by using detailed claims and eligibility data for traditional Medicare to analyze the relationship between obesity levels in Medicare-aged members and health care expenditures. We also evaluate the impact over time as members increase or decrease in diagnosed levels of obesity.
Data and Methodology
To evaluate the cost-of-care implications for Medicare-aged members with varying obesity levels, we conducted a longitudinal analysis using limited data set (LDS) medical claims data spanning 2021–2023. Pharmacy claims were not included in this analysis. This approach ensured consistency in tracking individuals over time while applying specific inclusion and exclusion criteria to refine the study population.
We included only non–end-stage renal disease and nonhospice members who aged into Medicare and maintained continuous enrollment in Medicare for 12 months in each of the three study years. Obesity status was determined using ICD-10 Z68 codes, which denote body mass index (BMI) categories. Members without any corresponding diagnosis were not included in the study. Members were segmented according to standard diagnosis groupings of BMI as follows:
- Healthy: Z6820–Z6824
- Overweight: Z6825–Z6829
- Class I obesity: Z6830–Z6834
- Class II obesity: Z6835–Z6839
- Class III obesity: Z6840–Z6845
Members were assigned cohorts according to the most frequent diagnosis they received in a calendar year. For example, if a member had three diagnoses for Class I obesity and one diagnosis for Class II obesity, then they were assigned to Class I obesity. If after applying this logic an individual had a tie in an assigned cohort, then they were removed from the study because we had insufficient confidence as to which cohort to assign them. These cases were infrequent.
We excluded the following ICD codes from classifying overweight because they were either too broad or were pediatric-specific: Z681, Z6851, Z6852, Z6583 and Z684.
Regarding claims, we relied on the allowed costs from LDS without adjustment unless otherwise stated—for example, there are no adjustments for trend, comorbidities or potential effects of COVID-19 on health care utilization and costs. The “Service Category Breakout” section includes allowed results that are normalized using risk scores from the v24 risk model. For results that show cost breakouts by major service category, we relied on Wakely’s proprietary grouper model to assign claims into service category.
Results
Our analysis examined medical allowed per member per month (PMPM) costs across Medicare-aged beneficiaries segmented by obesity severity, as classified using ICD-10 diagnosis codes. The results show a consistent and intuitive pattern: individuals with higher levels of obesity incur greater health care costs, and transitions in obesity status are closely associated with shifts in spending—though not always as expected.
Consistent Severity Analysis
To analyze the impact of obesity on health care costs, we organized the populations within the dataset into various cohorts by severity of obesity. Note that the definitions for each of the obesity cohorts are based on standard groupings of BMI diagnosis codes within the detailed claims as defined in the preceding Data and Methodology section. For this first analysis, we only evaluated the costs for members who did not cross between cohorts during the time frame of the study. For example, the Class I obesity row includes only individuals who had a corresponding diagnosis of Class I obesity in all three years. Table 1 shows the results of this comparison. The results are in line with those we generally expect to see—higher medical costs among higher severity cohorts. The full differential in costs between healthy and Class III obesity ranges from $418 in 2021 up to $567 in 2023. We also observed a two-year trend in the 30% to 32% range among the cohorts.
Table 1
Members with Consistent Severity
| Severity Cohort | Members | Medical Allowed PMPM | 2-yr Trend | ||
|---|---|---|---|---|---|
| 2021 | 2022 | 2023 | |||
| Healthy | 17,772 | $846 | $911 | $1,105 | 31% |
| Overweight | 28,941 | $812 | $874 | $1,056 | 30% |
| Class I obesity | 21,223 | $963 | $1,011 | $1,249 | 30% |
| Class II obesity | 9,475 | $1,007 | $1,087 | $1,318 | 31% |
| Class III obesity | 9,957 | $1,264 | $1,319 | $1,672 | 32% |
While the analysis uses standard BMI categories via ICD-10 codes, the minimal cost difference between the healthy and overweight cohorts in Table 1 suggests that BMI thresholds may not fully capture the onset of cost-driving comorbidities. This opens the door to exploring more nuanced metrics—such as waist circumference, fat distribution or even metabolically healthy obesity statuses—as potential cost predictors. However, this would require more detailed medical records compared to the claims data used in the present study.
Increasing Severity Analysis
We next evaluated members whose obesity status worsened over the study period. In general, we expected to see that as beneficiaries increased in obesity severity, their costs would increase; conversely as they decreased in severity, costs would decrease. We also expected that costs would materialize between the respective initial and final cohort costs from Table 1. For example, a member increasing in severity from overweight to Class I might have costs between $1,056 and $1,249 by 2023.
Table 2 shows results for beneficiaries increasing one degree of severity at any point between 2021 and 2023. We observed that each cohort had a higher starting cost than the corresponding cohort from Table 1 (e.g., $1,178 versus $846 for the healthy cohort). The ending costs varied for each cohort whether they were higher or lower than their Table 1 counterparts, with all but Class III obesity being higher. Generally, these results seem to imply that there may be higher comorbidities in these populations that could be contributing to the weight gain, which warrants further analysis to confirm. The Class III obesity result is lower, but that may be due to statistical variance since there are fewer members in that cohort.
It’s also noteworthy that while the starting costs are higher in Table 2, the costs don’t trend as high as the cohorts in Table 1. This may imply that while weight gain may be associated with other comorbidities, the weight gain itself doesn’t immediately cause significant complications that contribute to higher medical expenditures.
Table 2
Members with Increasing Severity
| Severity Cohort | Members | Medical Allowed PMPM | 2-yr Trend | |||
|---|---|---|---|---|---|---|
| 2021 | 2023 | 2021 | 2022 | 2023 | ||
| Healthy | Overweight | 2,854 | $1,178 | $1,041 | $1,122 | −5% |
| Overweight | Class I obesity | 3,637 | $1,231 | $1,058 | $1,275 | 4% |
| Class I obesity | Class II obesity | 2,815 | $1,429 | $1,342 | $1,487 | 4% |
| Class II obesity | Class III obesity | 1,519 | $1,471 | $1,329 | $1,603 | 9% |
Decreasing Severity Analysis
We also assessed individuals whose obesity severity improved over time. Table 3 shows results for beneficiaries with decreasing severity between 2021 and 2023. Similar to the cohorts in Table 2, we observed that the majority of comparison groups had starting and ending costs materially higher than the stable severity cohorts in Table 1. However, we observed a much more significant increase in the ending costs compared to those in Tables 1 and 2, particularly in less severe cohorts. Just like there are comorbidities that contribute to weight gain, there are comorbidities that contribute to weight loss, which may explain the higher costs compared to those in Table 1.
Table 3
Members with Decreasing Severity
| Severity Cohort | Members | Medical Allowed PMPM | 2-yr Trend | |||
|---|---|---|---|---|---|---|
| 2021 | 2023 | 2021 | 2022 | 2023 | ||
| Overweight | Healthy | 5,316 | $929 | $1,304 | $1,679 | 81% |
| Class I obesity | Overweight | 6,218 | $1,033 | $1,308 | $1,655 | 60% |
| Class II obesity | Class I obesity | 4,416 | $1,150 | $1,363 | $1,698 | 48% |
| Class III obesity | Class II obesity | 2,560 | $1,263 | $1,617 | $1,993 | 58% |
Given the results, we felt it was important to further segment the populations in Table 3 based on the year they decreased in severity. We found that there were more favorable cost outcomes for members who reduced obesity severity in Year 2 and maintained the reduced severity through Year 3, as shown in Table 4. This implies that sustained efforts to reduce weight and keep it off may produce cost savings over time, but we may need a longer time horizon to observe savings. It’s unclear based on the parameters of this study if the elevated costs are due to comorbidities contributing to weight loss or if members are more engaged in their health care and thus utilizing more costly medical interventions.
Table 4
Members with Decreasing Severity (Year 2 Decrease)
| Severity Cohort | Members | Medical Allowed PMPM | 2-yr Trend | ||||
|---|---|---|---|---|---|---|---|
| 2021 | 2022 | 2023 | 2021 | 2022 | 2023 | ||
| Overweight | Healthy | Healthy | 2,480 | $977 | $1,403 | $1,564 | 60% |
| Class I | Overweight | Overweight | 2,978 | $1,128 | $1,372 | $1,517 | 34% |
| Class II | Class I | Class I | 2,255 | $1,211 | $1,410 | $1,519 | 25% |
| Class III | Class II | Class II | 1,132 | $1,363 | $1,595 | $1,811 | 33% |
Gender
From Table 4, we’ve broken out the cohort of those shifting from Class I obesity to overweight. Within the cohort that is transitioning and maintaining that classification, Table 5 shows that females had lower baseline costs in 2021 but experienced a higher percentage increase in PMPMs over the two-year period compared to males (38% versus 30%). This divergence may reflect differences in health care engagement, utilization patterns or the clinical trajectory of weight loss interventions between males and females.
Table 5
Gender Breakout for Class I-Overweight-Overweight Cohort
| Gender | Members | Medical Allowed PMPM | 2-yr Trend | ||
|---|---|---|---|---|---|
| 2021 | 2022 | 2023 | |||
| Male | 1,312 | $1,253 | $1,535 | $1,634 | 30% |
| Female | 1,666 | $1,030 | $1,243 | $1,424 | 38% |
| Total | 2,978 | $1,128 | $1,372 | $1,517 | 34% |
Service Category Breakout
In addition to the gender breakout, we also reviewed major service category–level results to see if there were any particular service categories contributing to the change in costs. As shown in Table 6, while total PMPMs rose 34% for the selected Class I-overweight-overweight cohort, the largest growth occurred in inpatient facility costs (60%) and the “other” category (90%). This suggests that reductions in obesity severity may coincide with increased acute care episodes or ancillary services—potentially due to heightened clinical engagement or underlying health conditions.
Table 6
Service Category Breakout for Class I-Overweight-Overweight Cohort
| Service Category | Medical Allowed PMPM | 2-yr Trend | ||
|---|---|---|---|---|
| 2021 | 2022 | 2023 | ||
| Total cost of care | $1,128 | $1,372 | $1,517 | 34% |
| Total inpatient facility | $315 | $448 | $504 | 60% |
| Total outpatient facility | $241 | $287 | $341 | 41% |
| Total professional | $505 | $544 | $545 | 8% |
| Total other | $67 | $94 | $127 | 90% |
In Table 7, we normalized the allowed costs using raw risk scores from the v28 risk score model. This revealed a mostly flat trend (0%) despite significant growth in the unadjusted allowed. Professional service spending dropped 19%, while inpatient costs rose 19%. Assuming that risk score is a perfect predictor of morbidity, this further suggests that the weight loss is most likely caused by medical intervention or else we wouldn’t observe the large increases in inpatient costs coupled with decreases in professional costs. This is corroborated by the fact that the inpatient trend is flat from 2022 to 2023, suggesting that the increase in cost is incurred as a one-time expense in the first year of weight reduction. Additionally, the fact that the normalized costs are flat from 2021 to 2023 could suggest either that this population has morbidity that is outpacing trend or that reducing the severity of obesity has a relatively immediate savings impact that is obscured by comorbidities.
Table 7
Normalized Service Category Breakout for Class I-Overweight-Overweight Cohort
| Service Category | Medical Allowed PMPM | 2-yr Trend | ||
|---|---|---|---|---|
| 2021 | 2022 | 2023 | ||
| Total cost of care | $1,170 | $1,196 | $1,176 | 0% |
| Total inpatient facility | $327 | $390 | $390 | 19% |
| Total outpatient facility | $250 | $250 | $265 | 6% |
| Total professional | $523 | $474 | $422 | −19% |
| Total other | $69 | $82 | $99 | 42% |
Summary
This analysis confirms a direct relationship between obesity severity and health care costs for Medicare-aged beneficiaries, with stable Class III obesity members incurring over $500 more in PMPM than stable healthy individuals by 2023. Members transitioning between severity categories experienced the sharpest cost changes, signaling the importance of these inflection points.
Weight gain correlated with higher starting PMPMs but moderate trend growth, which together implies there are existing comorbidities but the weight gain contributes slowly to further cost increases. Perhaps counterintuitively, weight reduction often brought steeper cost increases, especially in less severe cohorts. This is possibly driven by increased health care engagement or adverse events. Sustained intentional weight loss produced more favorable outcomes, including normalized cost declines, suggesting long-term savings potential.
For the broader health care system, these findings spotlight obesity, not just as a clinical concern, but as a driver of long-term system sustainability. Stratifying risk by severity of obesity can highlight the opportunities to improve outcomes and generate savings over time if the outcomes are sustained. Benefit designs that encourage engagement without relying solely on short-term cost signals—and that factor in the risk-adjusted trajectory of care—are more likely to achieve durable impact.
Caveats
As with any modeling or analytical effort, several caveats should be noted to provide appropriate context for the findings presented. The following caveats outline important limitations and assumptions underlying this analysis.
- The analysis relies on claims data from LDS, including diagnosis reporting. The rigor and accuracy of diagnosis reporting may vary geographically and may not always be reliably populated. Similarly, diagnoses for BMI may be underreported unless the visit is directly associated with the individual’s weight or potentially for an annual wellness visit.
- BMI diagnoses, while generally accepted, may not be an accurate clinical measure of obesity.
- Diagnoses may be inconsistent, such as a member being assigned both Class II obesity and Class III obesity within the same year. In these instances we relied on the diagnosis mapping with the highest levels of frequency within the calendar year. Individuals with equal frequency between two or more groupings were excluded.
- COVID-19 may play a role in health care costs and utilization and diagnosis patterns as the data spans 2021–2023.
- We assigned claims to categories based on our mapping of revenue codes, bill type codes, Healthcare Common Procedure Coding System (HCPCS) and so on. This could cause differences if someone were to replicate this study.
- We do not know the underlying causes of weight loss in the data—are these members losing weight due to medical intervention (such as lap band surgery or pharmaceuticals), or is the weight loss caused by lifestyle changes such as reduced caloric intake coupled with increased exercise? Unintentional weight loss attributable to underlying illness (e.g., cancer or endocrine disorders) may also bias the results. This type of weight loss is more common among individuals aged 65 and older. Assessing such considerations was beyond the scope of this analysis. As such, we cannot conclude that the results presented are to be expected in the case of widespread change in benefit offerings such as for coverage of GLP-1s.
- In addition to the preceding analyses, we also considered reviewing cases where there were more drastic changes in obesity status (e.g., shifting from Class III to Class I) or where members shifted back and forth between a cohort (e.g., overweight to Class I to overweight) but did not find sufficient membership in these situations to ensure the results would be reliable. Expanding the analysis to 100% of Medicare data may allow for further analysis of these cohorts and may be worthwhile to understand cost impacts of medical or pharmaceutical interventions.
Conclusion
This study reinforces the clear financial implications of obesity across varying levels of severity in a Medicare-aged population. A consistent correlation emerges between obesity severity and total medical allowed PMPM costs, with higher BMI categories consistently associated with increased expenditures. Notably, changes in obesity status, whether increases or decreases, also correspond to shifts in health care costs, although not always in immediately predictable ways.
Importantly, members transitioning to higher obesity classes exhibited both elevated starting costs and more moderate trends over time, potentially signaling underlying comorbidities that complicate outcomes. Conversely, members reducing severity, especially those with prolonged weight loss, often demonstrated lower or even negative cost trends, particularly among the highest severity cohort. The stark PMPM differences are clearest in the Class III cohort, both when increasing (44% trend) and decreasing (–1% trend in the prolonged weight loss subgroup). This supports the hypothesis that interventions targeting the most severe obesity cases could produce the most immediate and measurable economic impact. However, the reliability of these results is constrained by limitations in diagnosis coding, membership thresholds and unmeasured variables such as the cause of weight change.
As the health care industry evaluates the role of emerging treatments such as GLP-1s and holistic care management programs, these findings suggest meaningful opportunities to improve both health outcomes and system sustainability. Future research should expand data sources, refine methodologies and incorporate clinical indicators to more clearly isolate the impact of interventions and guide sound policy decisions.
This article is provided for informational and educational purposes only. Neither the Society of Actuaries nor the respective authors’ employers make any endorsement, representation or guarantee with regard to any content, and disclaim any liability in connection with the use or misuse of any information provided herein. This article should not be construed as professional or financial advice. Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries or the respective authors’ employers.
Robert Lang, FSA, MAAA, is a consulting actuary at Wakely Consulting Group. He can be reached at Robert.Lang@wakely.com.
Chase Peterson is an actuarial analyst at Wakely Consulting Group. He can be reached at Chase.Peterson@wakely.com.
Ben Cruz, ASA, MAAA, is a consulting actuary at Wakely Consulting Group. He can be reached at Ben.Cruz@wakely.com.
Endnote
[1] The prevalence of obesity among US adults 20 and over was 41.9% from 2017–March 2020. Bryan Stierman, Joseph Afful, Margaret D. Carroll et al., “National Health and Nutrition Examination Survey 2017–March 2020 Prepandemic Data Files—Development of Files and Prevalence Estimates for Selected Health Outcomes,” National Health Statistics Reports no. 158 (June 14, 2021):158, https://stacks.cdc.gov/view/cdc/106273.