Skip to main content
JAMA Network logoLink to JAMA Network
. 2026 Oct 13;9(10):e2638780. doi: 10.1001/jamanetworkopen.2026.38780

Out-of-Pocket Costs for Chronic Illness Management in High-Deductible Health Plans

Risha Gidwani 1,2,3,4,✉, Zachary Wagner 2,5, Steven M Asch 6, Veronica Yank 7, Lane Burgette 8
PMCID: PMC13662656  NIHMSID: NIHMS2205855  PMID: 42867861

This cohort study uses national commercial insurance claims to compare out-of-pocket costs for adults enrolled in high-deductible health plans vs plans without a high deductible to receive guideline-concordant care for common chronic conditions.

Key Points

Question

Among patients who received guideline-concordant chronic disease management when enrolled in high-deductible health plans (HDHPs) vs non-HDHPs, what is the difference in and magnitude of out-of-pocket (OOP) costs incurred?

Findings

In this cohort study of 539 590 adults with diabetes, hypertension, or both, estimated annual OOP costs to receive recommended care were 1.6 to 2.4 times as high in HDHPs compared with non-HDHPs, with significant differences of approximately $64 to $86 across conditions.

Meaning

Findings suggest that patients in an HDHP incurred significant but modest increases in OOP costs to receive chronic disease management; results quantify 1 potential financial mechanism underlying previously documented lower health care utilization for patients enrolled in HDHPs.

Abstract

Importance

High-deductible health plans (HDHPs) enroll a substantial share of privately insured Americans and are associated with reduced use of care, even care that is considered medically necessary. However, the magnitude of patient out-of-pocket (OOP) costs required to obtain evidence-based medical care when enrolled in HDHPs remains poorly understood.

Objective

To estimate patient OOP costs required to receive guideline-concordant care for common chronic conditions when enrolled in an HDHP vs a non-HDHP.

Design, Setting, and Participants

This retrospective cohort study used 2022 to 2023 national commercial insurance claims data. Participants included adults aged 18 to 64 years with employer-sponsored insurance and previously diagnosed diabetes, hypertension, or both. To enter the cohort, persons had to have received guideline-concordant care for their chronic illnesses. Data were analyzed from July 2025 to April 2026.

Exposure

Enrollment in an HDHP vs a non-HDHP.

Main Outcomes and Measures

The primary outcome was annual patient OOP costs to receive evidence-based care. Entropy balancing and generalized linear models with geographic fixed effects estimated the mean OOP costs of receiving minimum recommended medical care for diabetes, hypertension, or both in HDHPs and non-HDHPs. Estimates were weighted to ensure that observed costs were representative of the full sample of chronically ill persons.

Results

There were 539 590 adults receiving guideline-concordant care who entered the cohort. After entropy balancing, groups exhibited mean covariate balance, with males representing 57.9% of HDHP enrollees and 57.8% of non-HDHP enrollees, and similar age distributions across groups, with 45.4% of both HDHP and non-HDHP enrollees 55 to 64 years of age. The estimated annual OOP cost for individuals to receive guideline-concordant diabetes care was $153.91 (95% CI, $150.18-$157.64) in HDHPs vs $68.06 (95% CI, $66.34-$69.78) in non-HDHPs; $128.94 (95% CI, $127.25-$130.64) in HDHPs vs $53.89 (95% CI, $53.20-$54.58) in non-HDHPs for guideline-concordant hypertension care; and $164.18 (95% CI, $160.93-$167.44) in HDHPs vs $100.60 (95% CI, $98.80-$102.39) in non-HDHPs for guideline-concordant care for comorbid diabetes and hypertension (all P < .001).

Conclusions and Relevance

In this cohort study of a large, national sample of chronically ill adults with hypertension, diabetes, or both, annual patient costs were 1.6 to 2.4 times as high to receive recommended care in HDHPs vs non-HDHPs, although the magnitude of the difference in OOP costs was relatively small (range, $63.68 to $85.85). Findings provide novel evidence quantifying the magnitude of cost burden experienced by chronically ill patients under various insurance mechanisms and complement prior HDHP studies evaluating health care utilization.

Introduction

More than half of Americans 45 to 64 years of age are chronically ill and require regular access to health care.1,2 Access to routine care can be supported or impeded by insurance benefit design, especially through policies regarding patient cost-sharing. Approximately 42% of privately insured persons are enrolled in high-deductible health plans (HDHPs), which are characterized by low premiums but high cost-sharing. HDHP plans with Health Savings Accounts (HSAs) are subject to Internal Revenue Service regulations regarding deductibles; in 2023, their minimum deductible was $1500 for an individual and $3000 for family coverage.3 Chronic disease management is subject to this deductible. Evidence from multiple surveys indicates the majority of eligible HDHP enrollees do not accumulate sufficient HSA balances to offset their deductible, thus exposing patients to high after-tax cost-sharing at the point of care.4,5,6

Health care costs pose numerous challenges for patients, even for those of moderate to high income. More than one-third of Americans make serious trade-offs to be able to afford health care, including skipping meals and prolonging current prescriptions.7 Such trade-offs persist across the income spectrum, including for persons making up to $90 000 a year.7

Extensive quasi-experimental research shows that HDHP enrollment reduces use of care, including for high-value care such as disease management.8,9,10,11,12 Previous research conducted by members of our team showed that HDHP enrollment reduced use of evidence-based recommended medical care for chronically ill patients, such as 1 annual visit to a primary care physician for coronary artery disease management.13 That research added to existing strong evidence that health care costs, especially those engendered by HDHPs, pose a financial burden to patients and are associated with lower use of medical care, even that which is indicated or emergent.14,15,16,17,18,19

While the literature establishing the inverse relationship between HDHP-based cost sharing and health care utilization is robust, the magnitude of the costs that dissuade such care is largely unknown. Our work aims to fill that gap. Here, we estimated how much the typical patient would pay in total and out-of-pocket (OOP) costs if receiving care concordant with evidence-based practice guidelines when enrolled in an HDHP vs a non-HDHP, across the following clinical profiles: diabetes, hypertension, and diabetes with hypertension. These clinical profiles represent some of the most common chronic conditions in the country: 13% of US adults have diabetes and 35% of US adults have hypertension.1 Disease management for these conditions is crucial to prevent substantial morbidity or mortality from downstream cardiovascular events (myocardial infarction, heart failure, and stroke) or end-stage kidney disease.20,21 Notably, federal legislation has been introduced seeking to exempt chronic disease management services from HDHP deductibles.22

Methods

We conducted a retrospective cohort study using 2022 to 2023 national MarketScan commercial insurance claims data. MarketScan data contain approximately 25 million persons annually enrolled in employer-sponsored insurance, which covers 49% of the US population.4 This study was approved by the RAND institutional review board with a waiver of informed consent as all data were deidentified. The reporting of this study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

Our cohort consisted of persons aged 18 to 64 years enrolled in an HDHP with an HSA (treatment group) or a non-HDHP (control group) with 1 of our conditions of interest. HDHP with HSAs are federally legislated to subject virtually all chronic illness management to the deductible.3 Cohort members were allocated into 1 of 3 groups: diabetes, hypertension, or diabetes and hypertension. To enter the cohort, persons must have had (1) the chronic illness identified in 2022, (2) continuous enrollment in the same insurance plan from 2022 through 2023, (3) no secondary insurance, such as Medicare, that could contribute to paying OOP costs, and (4) recommended medical care received for their chronic illness. As there are more opportunities to contribute to a deductible in a family plan, models were run separately for persons with individual vs family plans.

Medicare care recommendations were extracted by 2 physician-investigators (S.M.A. and V.Y.) from evidence-based clinical practice guidelines and focused on those that could be operationalized in claims data (eTable 1 in Supplement 1).13 Recommendations spanned clinic visits, prescription drugs, and laboratory tests; only persons who met the relevant care requirements were included in the cohort (Table 1).

Table 1. Receipt of Recommended Medical Care, Per Disease Group and Health Care Plan.

Condition Total No. No. of unique individualsa No. (%) of participants who received recommended care
Clinic visits Laboratory tests Prescription drugs
All plans
Diabetes 119 658 91 275 85 996 (94.2) 33 662 (36.9) NAb
Hypertension 521 613 333 174 279 821 (84.0) 241 792 (72.6) NAb
Diabetes and hypertension 197 062 115 141 103 740 (90.1) 41 320 (35.9) 52 002 (45.2)
HDHP plans
Diabetes 19 730 14 511 13 842 (95.4) 5888 (40.6) NAb
Hypertension 82 841 51 625 44 200 (85.6) 38 641 (74.8) NAb
Diabetes and hypertension 28 023 15 384 14 220 (92.4) 6158 (40.0) 7645 (49.7)
Non-HDHP plans
Diabetes 99 928 76 764 72 154 (94.0) 27 774 (36.2) NAb
Hypertension 438 772 281 549 235 621 (83.7) 203 151 (72.2) NAb
Diabetes and hypertension 169 039 99 757 89 520 (89.7) 35 162 (35.2) 44 357 (44.5)

Abbreviations: HDHP, high-deductible health plan; NA, not applicable.

a

The same person could not be in both the diabetes group and the diabetes plus hypertension group; that is, the groups were mutually exclusive.

b

There were no evidence-based guidelines regarding prescriptions drugs for patients solely with diabetes or solely with hypertension that could be operationalized using claims data.

Outpatient Care

Guidelines recommend that persons with diabetes or hypertension have 1 clinic visit a year to a physician.23,24 Patients can satisfy this recommendation if they have an evaluation and management visit with a primary care physician (family practice, internal medicine, or geriatrics) or a specialist (endocrinology for patients with diabetes, cardiology for patients with hypertension). For this study, only the costs associated with the evaluation and management component of the visit were retained (eg, if a patient with diabetes had evaluation and management and foot ulcer debridement in the same visit, only the cost of the former was included). Retaining only the evaluation and management component of each visit reduced the contribution of illness severity to observed cost differences by excluding the costs of ancillary diagnostic or therapeutic services. Evaluation and management visits can take on 1 of 5 levels, each of which corresponds to a different price. The distribution of evaluation and management visits was highly similar across HDHP and non-HDHP enrollees (eTable 2 and eTable 3 in Supplement 1).

Laboratory Tests

Guidelines indicate evaluation of serum creatinine levels annually for persons with hypertension.25 Guidelines also denote that persons with diabetes should have tests for estimated glomerular filtration rate (eGFR) annually, urine albumin level annually, serum creatinine:eGFR ratio, and potassium level annually when receiving relevant prescription medications, and 2 evaluations of hemoglobin A1c levels at least 90 days apart.23 Any laboratory test that satisfied these requirements, be it an individual test or a panel of tests, was included in the analytic dataset. When a person had both hypertension and diabetes, only 1 serum creatinine test was required.

Prescription Drugs

As our analyses were limited to claims data, practice guidelines23 regarding prescription drugs that pertained to our cohort were for persons with both diabetes and hypertension, who are recommended to receive angiotensin-converting enzyme inhibitors or angiotensin receptor blockers. We generated prescription drug costs through evaluating days’ supply. If days’ supply crossed years, such that a person received a 90-day supply on December 1, we prorated for costs incurred only in that year.

As clinic visit, laboratory, and prescription drug data were obtained from claims associated with clinical practice, they reflected patients’ actual experiences in obtaining necessary medical care, including natural variations in seeking care from primary care physicians vs specialists, in use of brand name vs generic medications, and in care utilization throughout the year (ie, averaged across different stages of deductible accumulation). A flow diagram of included participants is shown in the eFigure in Supplement 1.

Outcomes

Our primary outcome was annual patient OOP costs, derived as the sum of deductible, coinsurance, and copayment incurred to receive all recommended medical care (outpatient, laboratory tests, and prescription drugs). Patient OOP costs were evaluated using generalized linear models with adjustments to make costs generalize to the full population (including individuals who did not use care) and to adjust for confounders that could bias our estimates. Our secondary outcomes were total costs (OOP costs plus cost paid by the insurer) and OOP costs as a share of total costs. Persons entered the subcohort evaluating the latter if they received all recommended medical care.

Statistical Analysis

We ran 6 sets of primary models: 1 per clinical group (diabetes, hypertension, or diabetes and hypertension) and 1 per individual vs family insurance plan. Models adjusted for age category and sex included fixed effects for each metropolitan statistical area (MSA) to adjust for geographic differences in price of care and used both entropy balancing weights and MSA weights. Separate from MSA fixed effects, MSA weights were used to ensure that our cohort of HDHP enrollees who received recommended medical care for their chronic illness was representative of the national MarketScan sample of HDHP enrollees with that chronic illness.

Our focus on OOP costs that were conditional on receiving recommended medical care protected this analysis against many types of selection bias. However, some selection concerns remained; for example, if a patient with diabetes had a spouse with heart failure, the spouse’s heart failure care would contribute to the family HDHP deductible, resulting in a lower marginal OOP cost for diabetes care. We therefore conducted entropy balancing (EB) to weight HDHP and non-HDHP groups with respect to background health care utilization for other conditions, age group, and sex (eTable 4 in Supplement 1).26 Specifically, we identified care utilization for Elixhauser comorbidities, collapsing those comorbidities into categories (eTable 5 in Supplement 1).

We used generalized linear models to evaluate both our primary outcome of OOP cost and our secondary outcome of total cost of receiving recommended medical care, using a gamma family and log link. Our other secondary outcome of OOP costs as a share of total costs was generated using ordinary least squares models, with the same covariate, EB, and MSA weighting specifications as aforementioned, with the exception that these models did not include an MSA fixed effect; this was not needed, as the outcome was a proportion rather than a dollar figure. Lastly, we examined tail behavior of OOP costs across HDHPs and non-HDHPs. To do so, we conducted quantile regression using the same covariate and EB and MSA weighting specifications as described for the primary outcome, evaluating cost values at every decile between the 10th and 90th percentiles. All tests of significance used a 2-sided α of .05.

Data were analyzed from July 2025 to April 2026. Data cleaning was conducted using SAS Enterprise Guide, version 8.3 (SAS Institute Inc); statistical models were run using STATA MP, version 19 (StataCorp LLC).

Results

Our cohort consisted of 539 590 persons, 91 275 with diabetes (17.8% enrolled in an HDHP), 333 174 with hypertension (63.3% enrolled in an HDHP), and 115 141 with diabetes and hypertension (18.9% enrolled in an HDHP). After entropy balancing, groups exhibited covariate balance (all P > .05), with both HDHP and non-HDHP groups having the following age distribution: 4.8% 18 to 34 years, 15.2% 35 to 44 years, 34.6% 45 to 54 years, and 45.4% 55 to 64 years. Males represented 57.9% of HDHP enrollees and 57.8% of non-HDHP enrollees (Table 2).

Table 2. Demographic and Clinical Characteristics.

Characteristic Before EB weighting After EB weighting
Participants, No. (%) P value Participants with HDHP, No. (%) HDHP, sum of weighted No. (%)a P value
HDHP Non-HDHP
Total No. 81 520 458 070 NA 81 520 177 062 NA
Age group, y
18-34 3948 (4.8) 20 798 (4.5) <.001 3948 (4.8) 8570 (4.8) >.99
35-44 12 365 (15.2) 61 004 (13.3) 12 365 (15.2) 26 860 (15.2)
45-54 28 178 (34.6) 150 859 (32.9) 28 178 (34.6) 61 210 (34.6)
55-64 37 029 (45.4) 225 409 (49.2) 37 029 (45.4) 80 422 (45.4)
Sex
Female 34 354 (42.1) 221 708 (48.4) <.001 34 354 (42.1) 74 667 (42.2) .91
Male 47 166 (57.9) 236 362 (51.6) 47 166 (57.9) 102 359 (57.8)
Clinical group
Diabetes only 14 511 (17.8) 76 764 (16.8) <.001 14 511 (17.8) 31 517 (17.8) >.99
Hypertension only 51 625 (63.3) 281 549 (61.5) 51 625 (63.3) 112 134 (63.3)
Diabetes and hypertension 15 384 (18.9) 99 757 (21.8) 15 384 (18.9) 33 412 (18.9)
With individual plan 22 065 (27.1) 170 132 (37.1) <.001 22 065 (27.1) 47 931 (27.1) >.99
Autoimmune disorders 781 (3.5) 6023 (3.5) <.001 781 (3.5) 1999 (4.2) .91
Blood disorders 2621 (11.9) 20 297 (11.9) <.001 2621 (11.9) 7448 (15.5) .75
Cancer 1611 (7.3) 12 420 (7.3) .40 1611 (7.3) 3648 (7.6) .98
Cardiovascular conditions 3078 (14.0) 23 904 (14.0) <.001 3078 (14.0) 7886 (16.4) .52
COPD 2050 (9.3) 15 822 (9.3) <.001 2050 (9.3) 5378 (11.2) .92
Liver disease 1792 (8.1) 13 662 (8.0) .63 1792 (8.1) 4065 (8.5) .46
Metabolic and endocrine conditions 14 011 (63.5) 107 813 (63.4) <.001 14 011 (63.5) 32 262 (67.3) .55
Mental health and substance abuse conditions 3221 (14.6) 25 316 (14.9) <.001 3222 (14.6) 7976 (16.6) .08
Neurological condition 838 (3.8) 6414 (3.8) .001 838 (3.8) 2095 (4.34) .79
Kidney condition 18 720 (84.8) 143 098 (84.1) <.001 18 720 (84.8) 40 986 (85.5) .95
Other condition 1145 (5.2) 8813 (5.2) .002 1145 (5.2) 2929 (6.1) .001
With family plan 59 455 (72.9) 28 7938 (62.9) <.001 59 455 (72.9) 129 132 (72.9) >.99
Autoimmune disorders 4156 (7.0) 20 213 (7.0) .24 4156 (7.0) 9362 (7.3) .69
Blood disorders 12 057 (20.3) 58826 (20.4) <.001 12 057 (20.8) 29 520 (22.9) .27
Cancer 6778 (11.4) 32 508 (113) .32 6778 (11.4) 14 799 (11.5) .30
Cardiovascular condition 11 493 (19.3) 56 062 (19.5) <.001 11 493 (19.3) 27 608 (21.4) .27
COPD 11 790 (19.8) 57 559 (20.0) <.001 11 790 (19.8) 27 428 (21.2) .24
Liver disease 7931 (13.3) 38 152 (13.3) <.001 7931 (13.3) 17 614 (13.6) .46
Metabolic and endocrine conditions 45 061 (75.8) 218 401 (75.9) .59 45 061 (75.8) 101 382 (78.5) .66
Mental health and substance abuse conditions 17 759 (29.9) 87 879 (30.5) <.001 17 759 (29.9) 39 385 (30.5) <.001
Neurological condition 5095 (8.6) 24 964 (8.7) .46 5985 (8.6) 11 544 (8.9) .26
Kidney condition 51 720 (87.0) 249 959 (86.8) .58 51 720 (87.0) 114 024 (88.3) .35
Other condition 3525 (5.9) 17 276 (6.0) .12 3526 (5.9) 8781 (6.8) .11

Abbreviations: COPD, chronic obstructive pulmonary disease; EB, entropy balanced; HDHP, high-deductible health plan; NA, not applicable.

a

The total effective sample size for the control group is ((Σ(i)wi)2)/(Σ(i)wi2)), where i indicates the individual and w indicates the weight. The number denoted in each level of the variable, such as 8570 for age 18 to 34 years, is derived as the proportion of each cell multiplied by the effective sample size of the non-HDHP observations.

Annual patient OOP costs to receive all recommended medical care were significantly higher in an HDHP vs a non-HDHP (Table 3). The estimated mean OOP cost to receive all recommended care for patients with diabetes in an individual plan was $153.91 in an HDHP (95% CI, $150.18-$157.64) vs $68.06 (95% CI, $66.34-$69.78) in a non-HDHP (commensurate values in a family plan were $150.42 [95% CI, $148.28-$152.57] in an HDHP vs $62.94 [95% CI, $61.85-$64.03] in a non-HDHP) (all P < .001). Patients with hypertension in an individual plan paid a mean OOP cost of $128.94 (95% CI, $127.25-$130.64) to receive recommended medical care in an HDHP vs $53.89 (95% CI, $53.20-$54.58) in a non-HDHP (values in a family plan were $120.35 [95% CI, $119.25-$121.45] in an HDHP vs $51.43 [95% CI, $50.97-$51.89] in a non-HDHP) (all P < .001). Patients with both diabetes and hypertension in an individual plan paid a mean of $164.18 (95% CI, $160.93-$167.44) OOP to receive recommended medical care in an HDHP vs $100.60 (95% CI, $98.80-$102.39) in a non-HDHP (values in a family plan were $155.45 [95% CI, $153.32-$157.58] in an HDHP vs $96.49 [95% CI, $95.28-$97.69] in a non-HDHP) (all P < .001).

Table 3. Estimated Total and OOP Costs to Receive Minimum Levels of Recommended Medical Care, Per Disease Group and Family vs Individual Insurance Plana.

Group Estimated cost, mean (95% CI), $ P value
HDHP Non-HDHP Difference
OOP costsb
Diabetes, family 150.42 (148.28 to 152.57) 62.94 (61.85 to 64.03) 87.49 (85.19 to 89.78) <.001
Diabetes, individual 153.91 (150.18 to 157.64) 68.06 (66.34 to 69.78) 85.84 (81.90 to 89.79) <.001
Hypertension, family 120.35 (119.25 to 121.45) 51.43 (50.97 to 51.89) 68.92 (67.79 to 70.04) <.001
Hypertension, individual 128.94 (127.25 to 130.64) 53.89 (53.20 to 54.58) 75.06 (73.33 to 76.78) <.001
Diabetes and hypertension, family 155.45 (153.32 to 157.58) 96.49 (95.28 to 97.69) 58.96 (56.54 to 61.38) <.001
Diabetes and hypertension, individual 164.18 (160.93 to 167.44) 100.6 (98.80 to 102.39) 63.59 (59.94 to 67.23) <.001
Total OOP costsb
Diabetes, family 224.92 (222.50 to 227.34) 229.08 (226.99 to 231.17) −4.16 (−7.11 to −1.21) .006
Diabetes, individual 217.34 (213.46 to 221.23) 225.41 (222.24 to 228.58) −8.07 (−12.66 to −3.48) <.001
Hypertension, family 174.01 (172.97 to 175.04) 177.86 (177.01 to 178.71) −3.85 (−5.03 to −2.68) <.001
Hypertension, individual 173.62 (171.90 to 175.33) 176.94 (175.66 to 178.23) −3.33 (−5.13 to −1.52) <.001
Diabetes and hypertension, family 300.34 (296.21 to 304.48) 326.44 (322.39 to 330.49) −26.1 (−31.84 to −20.36) <.001
Diabetes and hypertension, individual 295.34 (288.77 to 301.9) 326.9 (321.88 to 331.92) −31.56 (−39.55 to −23.57) <.001

Abbreviations: HDHP, high-deductible health plan; OOP, out-of-pocket.

a

Full regression results can be found in eTables 6 and 7 in Supplement 1.

b

Values were estimated using marginal effects after running generalized linear models with a gamma family and a log link, using entropy balancing weights and metropolitan statistical area—specific weights, and using a lincom command to generate total OOP costs.

Quantile regression used to examine the tails of OOP costs showed at the low end that there was virtually no difference between HDHP and non-HDHP plans (Figure). This finding can likely be explained by OOP maximums being reached through treatment for other conditions or family members. The difference was about $100 at the median and the peak of the OOP distribution; in both the typical and worst-case scenarios, OOP costs were approximately $100 higher for HDHP enrollees. However, the difference was even higher around the 75th percentile mark, suggesting that the largest difference occurred when the HDHP OOP amount was above the median, but not extreme.

Figure. Line Graph Illustrating Cost Differences Across Spending Deciles Per Disease Group and Individual vs Family Health Care Plans.

Line chart of O O P cost difference by O O P cost decile across six groups. Single-panel line chart with six colored lines and a legend centered across the top. Horizontal axis label: O O P cost decile, with tick labels at approximately zero point zero five, zero point one zero, zero point one five, zero point two zero, zero point two five, zero point three zero, zero point four zero, zero point five zero, zero point six zero, zero point seven zero, zero point seven five, zero point eight zero, zero point eight five, zero point nine zero, and zero point nine five. Vertical axis label: O O P cost difference, dollar sign, with values from 0 to 160. Light gray horizontal gridlines span the plot. Legend entries with line colors: Individual diabetes and hypertension in orange; Family diabetes and no hypertension in magenta; Family diabetes and hypertension in taupe gray; Individual hypertension only in light teal; Individual diabetes and no hypertension in bright blue; Family hypertension only in dark teal. All lines begin near 0 at decile zero point zero five, rise through mid deciles, peak around decile zero point eight zero, then decline toward decile zero point nine five. Approximate peaks near decile zero point eight zero: magenta about 135; bright blue about 126; taupe gray about 124 to 125; orange about 119; light teal about 114; dark teal about 113. At decile zero point two five, values range from about 12 for dark teal to about 55 for orange and about 50 for bright blue. At decile zero point four zero, values cluster around 60 to 90, with dark teal near 60 and orange and bright blue near 90. At decile zero point nine five, endpoints are lower: light teal about 73, bright blue about 75, dark teal about 80, orange about 89, magenta about 94, and taupe gray about 101.

Annual out-of-pocket (OOP) costs are cost of care for high-deductible health plan enrollees vs enrollees without high-deductible health plans.

Although patients with HDHPs incurred higher OOP costs, their total costs to receive recommended medical care were slightly lower (Table 3). The mean difference in total costs for recommended medical care across HDHPs vs non-HDHPs ranged from a low of $3.33 (95% CI, $1.52-$5.13) for a patient with hypertension (P < .001) to a high of $31.56 (95% CI, $23.57-$39.55) for a person with diabetes and hypertension (P < .001). Total costs per category of care can be found in eTable 8 in Supplement 1.

For patients with diabetes alone or hypertension alone, annual OOP costs as a share of total costs were approximately twice as high in an HDHP compared with a non-HDHP (Table 4). A person with diabetes in an individual plan was responsible for paying a mean of 70.5% (95% CI, 69.1%-72.0%) (66.2% [95% CI, 65.2%-67.4%] for a family plan) of total costs OOP when in an HDHP and a mean of 35.8% (95% CI, 35.2%-37.0%) (32.6% [95% CI, 32.0%-33.1%] for a family plan) of total costs OOP when in a non-HDHP (all P < .001). A person with hypertension in an individual plan was responsible for paying a mean of 68.5% (95% CI, 68.0%-69.2%) (62.4% [95% CI, 62.1%-63.2%] for a family plan) of total costs OOP when in an HDHP and a mean of 36.1% (95% CI, 35.7%-37.0%) (32.7% [95% CI, 32.5%-33.4%] for a family plan) of total costs OOP when in a non-HDHP (all P < .001). The difference between plan types was more muted for persons with both diabetes and hypertension: a person in an individual plan was responsible for paying for a mean of 58.2% (95% CI, 57.9$-60.4%) (54.4% [95% CI, 53.2%-55.8%] for a family plan) of total costs OOP when in an HDHP and a mean of 41.9% (95% CI, 41.4%-43.1%) (40.2% [95% CI, 39.9$-41.3%] for a family plan) of total costs OOP if in a non-HDHP (all P < .001) (eTable 7 in Supplement 1).

Table 4. Proportion of Total Cost Paid OOP for Patients to Receive Minimum Levels of Recommended Medical Care, Per Disease Group and Family vs Individual Insurance Plana.

Group Proportion of total cost, mean (95% CI) P value
HDHP Non-HDHP
Diabetes, family 0.66 (0.65-0.67) 0.33 (0.32-0.33) <.001
Diabetes, individual 0.71 (0.69-0.72) 0.36 (0.35-0.37) <.001
Hypertension, family 0.62 (0.62-0.63) 0.33 (0.33-0.33) <.001
Hypertension, individual 0.69 (0.68-0.69) 0.36 (0.36-0.36) <.001
Diabetes and hypertension, family 0.54 (0.53-0.56) 0.40 (0.40-0.41) <.001
Diabetes and hypertension individual 0.58 (0.57-0.60) 0.42 (0.41-0.43) <.001

Abbreviations: HDHP, high-deductible health plan; OOP, out of pocket.

a

Full regression results can be found in eTable 9 in Supplement 1; estimated percentages per category of care can be found in eTable 10 in Supplement 1. Values were estimated using marginal effects from ordinary least squares models, using entropy balancing weights and using a lincom command to generate total proportion of costs paid OOP.

Discussion

There is a large body of work showing reduced use of important health services when patients enroll in HDHPs compared with non-HDHPs, but little is known about the size or difference in costs borne by such patients. The present cohort study helps to fill that gap in the literature. We found that the annual cost burden of receiving the minimum recommended medical care for 2 common chronic conditions (diabetes, hypertension, or both) was 1.6 to 2.4 times as high in HDHPs as in non-HDHPs. However, the absolute difference in the annual cost burden was typically less than $100. Prior work by members of our team found that use of recommended care for diabetes and hypertension was reduced by 4 to 6 percentage points as a result of HDHP enrollment.13 The findings of the present study, combined with the prior evidence base, suggest that even relatively minor increases in cost-sharing can result in important reductions in high-value health care use.11,13 This outcome exists even in relatively well-resourced populations, such as those with employer-sponsored insurance.

This work adds to the sparse literature evaluating the magnitude of cost-sharing experienced by chronically ill patients, especially in HDHPs. A large claims-based analysis found the mean annual OOP medication costs for patients with type 2 diabetes was $242,27 and the California Department of Health Care Access and Information identified median annual OOP costs of $171 for hypertension care and $145 for diabetes care; however, neither study disaggregated by type of commercial insurance (eg, HDHP).28 Using self-reported data from 2015 through 2018 Medical Expenditure Panel Survey, Park et al29 found that mean OOP costs for persons in HDHPs were $951 annually for a person with 1 chronic condition, although they did not disaggregate by clinical condition.

While the magnitude of annual OOP costs under an HDHP may seem small, it appears sufficiently large to dissuade use of recommended disease management.13 Other work also finds relatively small increases in OOP costs dissuade chronic illness management, which is then associated with greater use of costly emergency department visits or inpatient care or with greater mortality.30,31 In one study, a $10 to $50 increase in OOP costs for type 2 diabetes medications was associated with an increase of $342 in hospitalization costs.27 Another study found a $100 increase in patient cost-sharing for Medicare Part D medications was associated with a 14% increase in mortality.32 Doshi et al33 found that even a $10 OOP payment was associated with 10% of patients abandoning oral anticancer medications.

These findings suggest that small increases in up-front spending may come at the cost of foregone high-value care, with the potential for greater downstream costs and morbidity or mortality. Thus, cost-sharing for chronic illness management may be counterproductive as a value-based strategy. Policies promoting use of outpatient care at lower patient OOP cost, especially for patients with chronic illness or those that require regular monitoring, should be explored for their potential to be a cost-effective strategy for payers. Our findings suggest that policies to exempt chronic disease management from HDHP deductibles may improve access to necessary care.22

Limitations

Our study is a retrospective evaluation of claims data and was subject to certain limitations. We were limited in the guideline-concordant care we could evaluate; prescription drug treatment for diabetes alone or for hypertension alone depends on laboratory- or office-based measurements, which are unavailable in claims data. Our work thus presents conservative estimates; evaluating the full scope of guideline-concordance care would result in higher OOP costs. Lastly, as our research question evaluated the difference in OOP costs between HDHP and non-HDHP enrollees among patients who received recommended medical care, our cohort was intentionally restricted to individuals who received such care. Although we used MSA weighting to make our cohort more reflective of the broader chronically ill population, our findings may not generalize to individuals who did not receive recommended medical care.

Conclusions

This large, national cohort study evaluating the financial consequences to patients who received evidence-based chronic disease care found that annual OOP expenses for adults with hypertension, diabetes, or both were 1.6 to 2.4 times as high to receive recommended care in HDHPs as in non-HDHPs. This study provides some of the first estimates of the magnitude of costs experienced by patients enrolled in HDHPs. Given the sparsity of literature on the extent of costs borne by patients to receive necessary medical care, which stands in contrast to the robust evidence establishing the inverse relationship between cost-sharing in general terms and utilization of care, this study fills an important gap. The study’s findings suggest that relatively small increases in patient cost-sharing discourage necessary care and therefore exacerbate downstream health risks and costs. Policymakers and payers may wish to consider modifications to benefit design to better align financial incentives with clinical value.

Supplement 1.

eTable 1. Recommended Clinical, Laboratory, and Pharmaceutical Care for Patients With Diabetes and Hypertension

eTable 2. Distribution of Evaluation and Management CPT Codes for Outpatient Care

eTable 3. Distribution of Evaluation and Management CPT Codes for Outpatient Care, by Specialist vs Primary Care Physician

eFigure. Study CONSORT Diagram

eTable 4. Loss of Sample Size From EB Weighting

eTable 5. Elixhauser Conditions and Categories

eTable 6. Full Regression Results, OOP Costs

eTable 7. Full Regression Results, Total Costs

eTable 8. Predicted Total Cost of Care, by Category of Care

eTable 9. Full Regression Results, OOP Costs as a Share of Total Costs

eTable 10. Predicted OOP Costs (95% CI) as a Share of Total Costs

Supplement 2.

Data Sharing Statement

References

  • 1.Watson KB, Wiltz JL, Nhim K, Kaufmann RB, Thomas CW, Greenlund KJ. Trends in multiple chronic conditions among US adults, by life stage, behavioral risk factor surveillance system, 2013-2023. Prev Chronic Dis. 2025;22:E15. doi: 10.5888/pcd22.240539 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Buttorff C, Ruder T, Bauman M. Multiple chronic conditions in the United States. RAND Corporation. Accessed May 14, 2026. https://www.rand.org/pubs/tools/TL221.html
  • 3.Part III. § 54.9831-1: Rev. Proc. 2022-24 Section 1. Internal Revenue Service. 2023. Accessed September 2, 2026. https://www.irs.gov/pub/irs-drop/rp-22-24.pdf
  • 4.Claxton G, Rae M, Winger A, et al. Employer health benefits, 2023 annual survey. 2023. KFF. Accessed 14 May 2026. https://files.kff.org/attachment/Employer-Health-Benefits-Survey-2023-Annual-Survey.pdf
  • 5.Kullgren JT, Cliff EQ, Krenz C, et al. Use of health savings accounts among US adults enrolled in high-deductible health plans. JAMA Netw Open. 2020;3(7):e2011014. doi: 10.1001/jamanetworkopen.2020.11014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Spiegel JF, Fronstin P. Trends in health savings account balances, contributions, distributions, and investments, 2011-2021. EBRI Issue Brief. 2023;(581):2023. [Google Scholar]
  • 7.Maese E. One-third of Americans cut back to cover healthcare expenses. GALLUP. Accessed 14 May, 2026. https://news.gallup.com/poll/702596/one-third-americans-cut-back-cover-healthcare-expenses.aspx?utm_source=NIHCM+Foundation&utm_campaign=7d4c2f66f2-NIHCM_EMAIL_CAMPAIGN_2025_12_11_COPY_01&utm_medium=email&utm_term=0_-8cf6f06966-489574410
  • 8.Jiang DH, Herrin J, Van Houten HK, McCoy RG. Evaluation of high-deductible health plans and acute glycemic complications among adults with diabetes. JAMA Netw Open. 2023;6(1):e2250602. doi: 10.1001/jamanetworkopen.2022.50602 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Brot-Goldberg ZC, Chandra A, Handel BR, Kolstad JT. What does a deductible do? the impact of cost-sharing on health care prices, quantities, and spending dynamics. Q J Econ. 2017;132(3):1261-1318. doi: 10.1093/qje/qjx013 [DOI] [Google Scholar]
  • 10.Shafer PR, Dusetzina SB, Sabik LM, Platts-Mills TF, Stearns SC, Trogdon JG. High deductible health plans and use of free preventive services under the Affordable Care Act. Inquiry: The Journal of Health Care Organization. Provision and Financing. 2023;60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Wharam JF, Zhang F, Wallace J, et al. Vulnerable and less vulnerable women in high-deductible health plans experienced delayed breast cancer care. Health Aff (Millwood). 2019;38(3):408-415. doi: 10.1377/hlthaff.2018.05026 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Lewey J, Gagne JJ, Franklin J, Lauffenburger JC, Brill G, Choudhry NK. Impact of high deductible health plans on cardiovascular medication adherence and health disparities. Circ Cardiovasc Qual Outcomes. 2018;11(11):e004632. doi: 10.1161/CIRCOUTCOMES.118.004632 [DOI] [PubMed] [Google Scholar]
  • 13.Gidwani R, Yank V, Asch SM, et al. High-deductible health plans and receipt of guideline-concordant care for adults with chronic illness. JAMA Netw Open. 2025;8(4):e258045. doi: 10.1001/jamanetworkopen.2025.8045 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Rook JM, Kelley-Quon LI, Jen HC, Gidwani R. Original study: association of high-deductible health plans and complicated appendicitis among children. Ann Surg. 2025. doi: 10.1097/SLA.0000000000006680 [DOI] [PubMed] [Google Scholar]
  • 15.Kozhimannil KB, Law MR, Blauer-Peterson C, Zhang F, Wharam JF. The impact of high-deductible health plans on men and women: an analysis of emergency department care. Med Care. 2013;51(8):639-645. doi: 10.1097/MLR.0b013e31829742d0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Galbraith AA, Ross-Degnan D, Soumerai SB, Rosenthal MB, Gay C, Lieu TA. Nearly half of families in high-deductible health plans whose members have chronic conditions face substantial financial burden. Health Aff (Millwood). 2011;30(2):322-331. doi: 10.1377/hlthaff.2010.0584 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Williams CP, Rocque GB, Caston NE, et al. Health insurance and financial hardship in cancer survivors during the COVID-19 pandemic. PLoS One. 2022;17(8):e0272740. doi: 10.1371/journal.pone.0272740 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Fusco N, Sils B, Graff JS, Kistler K, Ruiz K. Cost-sharing and adherence, clinical outcomes, health care utilization, and costs: a systematic literature review. J Manag Care Spec Pharm. 2023;29(1):4-16. doi: 10.18553/jmcp.2022.21270 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Eaddy MT, Cook CL, O’Day K, Burch SP, Cantrell CR. How patient cost-sharing trends affect adherence and outcomes: a literature review. P T. 2012;37(1):45-55. [PMC free article] [PubMed] [Google Scholar]
  • 20.Vaughan AS, Coronado F, Casper M, Loustalot F, Wright JS. County-level trends in hypertension-related cardiovascular disease mortality-United States, 2000 to 2019. J Am Heart Assoc. 2022;11(7):e024785. doi: 10.1161/JAHA.121.024785 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Gregg EW, Cheng YJ, Srinivasan M, et al. Trends in cause-specific mortality among adults with and without diagnosed diabetes in the USA: an epidemiological analysis of linked national survey and vital statistics data. Lancet. 2018;391(10138):2430-2440. doi: 10.1016/S0140-6736(18)30314-3 [DOI] [PubMed] [Google Scholar]
  • 22.S.655 — 118th Congress (2023-2024) – Chronic Disease Management Act of 2023. Congress.gov. Accessed May 15, 2026. https://www.congress.gov/bill/118th-congress/senate-bill/655/text
  • 23.American Diabetes Association . Standards of medical care in diabetes--2015: summary of revisions. Diabetes Care. 2015;38(suppl):S4. doi: 10.2337/dc15-S003 [DOI] [PubMed] [Google Scholar]
  • 24.Chobanian AV, Bakris GL, Black HR, et al. ; Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. National Heart, Lung, and Blood Institute; National High Blood Pressure Education Program Coordinating Committee . The seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure. Hypertension. 2003;42(6):1206-1252. doi: 10.1161/01.HYP.0000107251.49515.c2 [DOI] [PubMed] [Google Scholar]
  • 25.National Clinical Guideline Centre (UK) . Hypertension: the clinical management of primary hypertension in adults: update of clinical guidelines 18 and 34. Royal College of Physicians. 2011. PMID:22855971. [PubMed] [Google Scholar]
  • 26.Hainmueller J. Entropy balancing for causal effects: a multivariate reweighting method to produce balanced samples in observational studies. Political Analysis. 2012;20(1):25-46. doi: 10.1093/pan/mpr025 [DOI] [Google Scholar]
  • 27.Thornton Snider J, Seabury S, Lopez J, McKenzie S, Goldman DP, Goldman DP. Impact of type 2 diabetes medication cost sharing on patient outcomes and health plan costs. Am J Manag Care. 2016;22(6):433-440. [PubMed] [Google Scholar]
  • 28.Health Care Payments Data (HPD): HCAI. California Department of Health Care Access and Information . Accessed July 16, 2025. https://hcai.ca.gov/data/cost-transparency/healthcare-payments/
  • 29.Park KE, Saluja S, Kaplan CM. Alleviating financial hardships associated with high-deductible health plans for adults with chronic conditions through health savings accounts. J Gen Intern Med. 2023;38(7):1593-1598. doi: 10.1007/s11606-022-07985-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Thomas A, Valero-Elizondo J, Khera R, et al. Forgone medical care associated with increased health care costs among the U.S. heart failure population. JACC Heart Fail. 2021;9(10):710-719. doi: 10.1016/j.jchf.2021.05.010 [DOI] [PubMed] [Google Scholar]
  • 31.McCoy RG, Swarna KS, Jiang DH, et al. Enrollment in high-deductible health plans and incident diabetes complications. JAMA Netw Open. 2024;7(3):e243394. doi: 10.1001/jamanetworkopen.2024.3394 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Chandra A, Flack E, Obermeyer Z. The Health Costs of Cost-Sharing. National Bureau of Economic Research. Accessed May 14, 2026. https://www.nber.org/papers/w28439 [DOI] [PMC free article] [PubMed]
  • 33.Doshi JA, Li P, Huo H, Pettit AR, Armstrong KA. Association of patient out-of-pocket costs with prescription abandonment and delay in fills of novel oral anticancer agents. J Clin Oncol. 2018;36(5):476-482. doi: 10.1200/JCO.2017.74.5091 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1.

eTable 1. Recommended Clinical, Laboratory, and Pharmaceutical Care for Patients With Diabetes and Hypertension

eTable 2. Distribution of Evaluation and Management CPT Codes for Outpatient Care

eTable 3. Distribution of Evaluation and Management CPT Codes for Outpatient Care, by Specialist vs Primary Care Physician

eFigure. Study CONSORT Diagram

eTable 4. Loss of Sample Size From EB Weighting

eTable 5. Elixhauser Conditions and Categories

eTable 6. Full Regression Results, OOP Costs

eTable 7. Full Regression Results, Total Costs

eTable 8. Predicted Total Cost of Care, by Category of Care

eTable 9. Full Regression Results, OOP Costs as a Share of Total Costs

eTable 10. Predicted OOP Costs (95% CI) as a Share of Total Costs

Supplement 2.

Data Sharing Statement


Articles from JAMA Network Open are provided here courtesy of American Medical Association

RESOURCES