In total, 139,276 individuals fulfilled all inclusion criteria, of which 126,052 had commercial or Medicare insurance. At index, 10,196 (8.09%) individuals had normal BMI, 37,610 (29.8%) were classified as overweight, 36,161 (28.7%) had Class 1 obesity, 22,389 (17.8%) had Class 2 obesity, and 19,696 (15.6%) had Class 3 obesity (Table 1).

Table 1 summarizes the baseline characteristics of individuals. The mean (SD) age was higher in individuals with normal BMI (67.7 [10.9] years) than in those with overweight or other obesity classes (range: 59.2 [8.5] to 65.4 [10.4] years). More than half of the study population across all five cohorts was female, and approximately 80% of the study population was White (Table 1). The mean baseline BMI range was 23.5–45.7 kg/m2 across the different BMI classifications. Among people with obesity or overweight, a numerical trend of multiple obesity-related complications and higher mean (SD) CCI scores was observed in individuals with higher BMI than in those with lower BMI (CCI scores: normal BMI: 1.3 [1.8]; overweight: 1.1 [1.6]; Class 1 obesity: 1.2 [1.7]; Class 2 obesity: 1.3 [1.7]; and Class 3 obesity: 1.5 [1.8]). Hypertension and dyslipidemia were the most prevalent comorbidities across all cohorts, with a direct increase in prevalence with higher BMI. Similarly, the prevalence of type 2 diabetes (T2D), atherosclerotic cardiovascular disease (ASCVD), obstructive sleep apnea (OSA), metabolic dysfunction-associated fatty liver disease (MAFLD), and metabolic dysfunction-associated steatohepatitis (MASH) was higher in individuals with higher obesity classes (Table 1).

Supplementary Tables 3, 4, and 5 present demographic and clinical characteristics among commercial-, Medicare-, and Medicaid-insured individuals, respectively. The demographic characteristics of the commercial- and Medicare-insured individuals were similar to those of the commercial- or Medicare-combined cohort. Notably, more than two-thirds of the Medicare-insured population across all cohorts were women (Supplementary Table 4). The prevalence of comorbidities in the commercial- and Medicare-insured cohorts was also similar to that in the commercial- or Medicare-combined cohort, in which higher obesity classes had higher prevalence of hypertension, dyslipidemia, T2D, ASCVD, OSA, MAFLD, and MASH than the lower obesity classes.

After adjustment for baseline and demographic factors, individuals with BMI ≥ 35 kg/m2 consistently incurred higher mean annualized healthcare costs than those in lower BMI categories through the end of follow-up (Fig. 1). Mean (SD) total healthcare costs were highest in individuals with Class 3 and 2 obesity ($37,698 [$34,239]) and $33,676 [$29,627], respectively) compared with those in individuals with normal BMI ($32,402 [$32,290]; Fig. 1). Similar trends were observed in total pharmacy costs, with mean (SD) costs of $15,040 ($41,754) and $11,088 ($33,879) in individuals with Class 3 and Class 2 obesity, respectively. In contrast, individuals with normal BMI had lower total pharmacy costs ($9327 [32,165]). Total medical costs also followed a similar pattern, with mean (SD) costs of $25,493 ($16,927) and $24,075 ($15,540) in individuals with Class 3 and Class 2 obesity, respectively, compared with $23,821 ($17,474) in those with normal BMI.

Adjusted all-cause costs among commercial- or Medicare-insured participants among people with MTS-OA stratified by BMI class. Note: All-cause healthcare costs are the sum of all-cause medical and all-cause pharmacy costs. Adjusted all-cause and KOA-related healthcare costs were estimated using a generalized linear model with gamma distribution and a log-link function. Models adjusted for age, sex, payor type, baseline CCI score, and corresponding baseline cost (i.e. adjusted baseline total costs for total cost outcome, adjusted baseline medical costs for medical cost outcome). Adjusted costs were winsorized at 1%. Costs in the database were CPI-adjusted for inflation to 2024 via a standard cost variable, and therefore, no additional inflation adjustment transformations were carried out. Normal: ≥ 18.5 to < 25 kg/m2. Overweight: ≥ 25 to < 30 kg/m2. Class 1 obesity: ≥ 30 to < 35 kg/m2. Class 2 obesity: ≥ 35 to < 40 kg/m2. Class 3 obesity: ≥ 40 kg/m2. BMI body mass index, CCI Charlson Comorbidity Index, CPI Consumer Price Index, KOA knee osteoarthritis, MTS-OA moderate-to-severe osteoarthritis, SD standard deviation, USD US Dollars

Similar to all-cause costs, individuals with BMI ≥ 35 kg/m2 incurred greater mean annualized KOA-related total, pharmacy, and medical costs than those with BMI < 35 kg/m2 through the end of follow-up (Fig. 2). Mean (SD) total healthcare costs were highest in individuals with Class 2 and Class 3 obesity ($4900 [$596] and $4886 [$583], respectively) compared with those in individuals with normal BMI ($3299 [$443]; Fig. 2). The largest contributor to KOA-related total costs was medical costs, followed by pharmacy costs. Mean (SD) total medical costs were highest in individuals with Class 2 and 3 obesity ($5077 [$808] and $4898 [$775], respectively). In contrast, individuals with normal BMI had lower total medical costs ($3581 [$655]) (Fig. 2). KOA-related mean (SD) pharmacy costs depicted a progressive increase with increasing BMI across all the cohorts (normal BMI: $393 [$191]; overweight: $433 [$196]; Class 1 obesity: $451 [$197]; Class 2 obesity: $508 [$217]; and Class 3 obesity: $627 [$259]; Fig. 2).

Adjusted KOA-related costs among people with MTS-OA stratified by BMI class among commercial- or Medicare-insured participants. Note: The total KOA-related costs included KOA-related medical, pharmacy, and surgical costs. Adjusted all-cause and KOA-related healthcare costs were estimated using a generalized linear model with gamma distribution and a log-link function. Models adjusted for age, sex, payor type, baseline CCI score, and corresponding baseline cost (i.e. adjusted baseline total costs for total cost outcome, adjusted baseline medical costs for medical cost outcome). Adjusted costs were winsorized at 1%. Costs in the database were CPI-adjusted for inflation to 2024 via a standard cost variable, and therefore, no additional inflation adjustment transformations were carried out. Normal: ≥ 18.5 to < 25 kg/m2. Overweight: ≥ 25 to < 30 kg/m2. Class 1 obesity: ≥ 30 to < 35 kg/m2. Class 2 obesity: ≥ 35 to < 40 kg/m2. Class 3 obesity: ≥ 40 kg/m2. BMI body mass index, CCI Charlson Comorbidity Index, CPI Consumer Price Index, KOA knee osteoarthritis, MTS-OA moderate-to-severe osteoarthritis, SD standard deviation, USD US Dollars

The overall frequency of all-cause and KOA-related healthcare visits in the first year post-index was comparable across different BMI classes (Table 2). However, the number of all-cause outpatient visits was greater among individuals with BMI ≥ 35 kg/m2 (range 20.0–20.3) than among those with BMI < 35 kg/m2 (range 18.6–19.8). A progressive increase in KOA-related outpatient visits was observed with higher BMI classes (normal: 1.5 [2.3]; overweight: 1.6 [2.4]; Class 1 obesity: 1.7 [2.5]; Class 2 obesity: 1.8 [2.5]; and Class 3 obesity: 1.8 [2.4]; Table 2).

Adjusted costs and HCRU data for commercial-, Medicare-, and Medicaid-insured individuals are shown in Supplementary Figs. 1–3 and Supplementary Tables 6–8. Overall, the HCRU and costs in the commercial- and Medicare-insured individuals were similar to those in the combined cohort. The largest contributor to all-cause and KOA-related total costs was medical costs, followed by pharmacy costs, in commercial- or Medicare-insured individuals. Individuals with BMI ≥ 35 kg/m2 incurred greater knee-OA-related pharmacy, medical, and total costs than those with BMI < 35 kg/m2.

Across all BMI categories, people with overweight or obesity reported high use of non-steroidal anti-inflammatory medications (NSAIDs), opioids, non-tramadol opioids, injectable corticosteroids, and acetaminophen, with several other pain relief medications also commonly used (Fig. 3). The use of opioids, NSAIDs, acetaminophen, antidepressants, and gabapentin/pregabalin was higher among individuals with BMI ≥ 35 kg/m2 (Class 2 and Class 3 obesity) than among those with BMI < 35 kg/m2. Overall, the use of obesity management treatments, including obesity management medications (OMMs)/glucagon-like peptide-1 receptor agonists (GLP-1 RAs), phentermine, orlistat, and bariatric surgery, was low across all BMI classes. However, the usage was higher among individuals with BMI ≥ 35 kg/m2 than among those with BMI < 35 kg/m2 (Fig. 3). Few people with KOA and obesity underwent bariatric surgery, with most cases in the Class 3 obesity cohort (pre-index period: 0.6%; post-index period: 1.7%; Supplementary Table 9). Treatment patterns data stratified by insurance type are provided in Supplementary Tables 10–12.

Medication use 12-months post-index among people with MTS-OA stratified by BMI class among commercial- or Medicare-insured participants. Use of medications and medical services was based on claims data. Normal: ≥ 18.5 to < 25 kg/m2. Overweight: ≥ 25 to < 30 kg/m2. Class 1 obesity: ≥ 30 to < 35 kg/m2. Class 2 obesity: ≥ 35 to < 40 kg/m2. Class 3 obesity: ≥ 40 kg/m2. OMM/GLP-1 RA included participants with ≥ 1 claim for OMM or GLP-1 RA medication; OMM included those with ≥ 1 claim for OMM only; GLP-1 RA included those with ≥ 1 claim for GLP-1 RA medication indicated for either weight management or T2D. AOM antiobesity medications, BMI body mass index, GLP-1 RA glucagon-like peptide-1 receptor agonist, MTS-OA moderate-to-severe osteoarthritis pain, NSAID non-steroidal anti-inflammatory drug, OMM obesity management medication, OW overweight, T2D type 2 diabetes