All costs were inflation adjusted to 2013 dollars. Cost sharing was calculated by estimating regression models that adjusted for the diagnosis related group weights of the admissions. All inpatient hospitalizations were analyzed within the regression models, including those with $0 out-of-pocket spending.
All costs were inflation adjusted to 2013 dollars. Cost sharing was calculated by estimating regression models that adjusted for the diagnosis related group weights of the admissions. All inpatient hospitalizations were analyzed within the regression models, including those with $0 cost sharing.
All costs were inflation adjusted to 2013 dollars. Cost sharing was calculated by estimating regression models that adjusted for the diagnosis related group weights of the admissions. The overall category represents the adjusted mean cost sharing across all diagnoses or procedures. All relevant inpatient hospitalizations were analyzed within the regression models, including those with $0 cost sharing. AMI indicates acute myocardial infarction; CABG, coronary artery bypass graft.
eTable 1.ICD-9-CM Codes Associated With Broad Diagnosis- and Procedure-Related Categories Generated by the AHRQ Clinical Classification Software
eTable 2. Adjusted Mean Cost Sharing per Inpatient Hospitalization Using First Hospitalization Only, 2009-2013
eTable 3. Proportion of Total Hospital Payments Paid Out-of-Pocket by Patients
eTable 4. Adjusted Mean Cost Sharing per Inpatient Hospitalization for Hospitalizations With Cost Sharing of Greater than $0, 2009-2013
eTable 5. Adjusted Mean Cost Sharing per Inpatient Hospitalization, High Market Share States Only
Customize your JAMA Network experience by selecting one or more topics from the list below.
Adrion ER, Ryan AM, Seltzer AC, Chen LM, Ayanian JZ, Nallamothu BK. Out-of-Pocket Spending for Hospitalizations Among Nonelderly Adults. JAMA Intern Med. 2016;176(9):1325–1332. doi:10.1001/jamainternmed.2016.3663
Patients’ out-of-pocket spending for major health care expenses, such as inpatient care, may result in substantial financial distress. Limited contemporary data exist on out-of-pocket spending among nonelderly adults.
To evaluate out-of-pocket spending associated with hospitalizations and to assess how this spending varied over time and by patient characteristics, region, and type of insurance.
Design, Setting, and Participants
A retrospective analysis of medical claims for 7.3 million hospitalizations using 2009-2013 data from Aetna, UnitedHealthcare, and Humana insurance companies representing approximately 50 million members was performed. Out-of-pocket spending was evaluated by age, sex, type of insurance, region, and principal diagnosis or procedure for hospitalized adults aged 18 to 64 years who were enrolled in employer-sponsored and individual-market health insurance plans from January 1, 2009, to December 31, 2013. The study was conducted between July 1, 2015, and March 1, 2016.
Main Outcomes and Measures
Primary outcomes were total out-of-pocket spending and spending attributed to deductibles, copayments, and coinsurance for all hospitalizations. Other outcomes included out-of-pocket spending associated with 7 commonly occurring inpatient diagnoses and procedures: acute myocardial infarction, live birth, pneumonia, appendicitis, coronary artery bypass graft, total knee arthroplasty, and spinal fusion.
From 2009 to 2013, total cost sharing per inpatient hospitalization increased by 37%, from $738 in 2009 (95% CI, $736-$740) to $1013 in 2013 (95% CI, $1011-$1016), after adjusting for inflation and case-mix differences. This rise was driven primarily by increases in the amount applied to deductibles, which grew by 86% from $145 in 2009 (95% CI, $144-$146) to $270 in 2013 (95% CI, $269-$271), and by increases in coinsurance, which grew by 33% over the study period from $518 in 2009 (95% CI, $516-$520) to $688 in 2013 (95% CI, $686-$690). In 2013, total cost sharing was highest for enrollees in individual market plans ($1875 per hospitalization; 95% CI, $1867-$1883) and consumer-directed health plans ($1219; 95% CI, $1216-$1223). Cost sharing varied substantially across regions, diagnoses, and procedures.
Conclusions and Relevance
Mean out-of-pocket spending among commercially insured adults exceeded $1000 per inpatient hospitalization in 2013. Wide variability in out-of-pocket spending merits greater attention from policymakers.
Health insurance policies have changed dramatically in recent years, with patients increasingly responsible for a growing share of their health care expenditures through out-of-pocket spending. Proponents argue that this growth has the potential to reduce overuse and inappropriate use of health care.1 However, increased out-of-pocket spending can also impede access to appropriate care and affect treatment choices, and high levels of cost sharing may affect quality of life, particularly for low-income patients.2-7 Cost sharing may also reflect important competitive dynamics between insurers and health care professionals, especially given recent, dramatic changes in the health care market.8,9
Many estimates of out-of-pocket expenses incurred by insured patients for medical care rely on patient-reported data.10,11 These estimates may be biased and unreliable as a result of patients’ limited ability to recall the costs associated with medical conditions that they experienced several months or years earlier. Furthermore, survey-based estimates do not identify the specific features of insurance (eg, deductibles, copayments, and coinsurance) through which out-of-pocket spending may be changing. Understanding these issues is critical for policymakers. Recent estimates10,12 suggest that more than $300 billion is now spent out of pocket by patients on health care each year and, in 2015, nearly 80% of enrollees in employer-sponsored plans were subject to deductibles, the mean for which exceeded $1300 for single coverage.
Understanding the level and distribution of out-of-pocket spending associated with inpatient care is critical for both patients and policymakers. The costs associated with hospitalization can be high relative to the costs of other health care services, and for many conditions requiring inpatient care, patients are unable to choose among hospitals and services for less costly care. We therefore designed this study to address the following research questions: How much are patients paying out of pocket for inpatient care in the United States? Has this amount changed over time? What features of commercial insurance are associated with recent changes in out-of-pocket spending for hospitalizations? What patient and plan characteristics are associated with out-of-pocket spending for hospitalizations?
Question How much are insured patients paying out of pocket for inpatient care, and does that amount vary over time or by patient characteristics, region, or type of insurance?
Findings In this study of medical claims for 7.3 million hospitalizations from 2009 to 2013, total cost sharing associated with inpatient hospitalizations grew by over 37%. Our findings indicate a trend toward fewer plans requiring copayments at the point of service and more plans requiring higher coinsurance and deductibles after care is delivered.
Meaning Out-of-pocket spending is substantial for inpatient hospitalizations, even among insured individuals.
We used data from a large commercial health insurance claims database compiled by the Health Care Cost Institute (HCCI) (http://www.healthcostinstitute.org). The HCCI database includes all health care claims for approximately 50 million privately insured individuals from all 50 states who receive health care coverage from Aetna, Humana, and UnitedHealthcare. For this study, we used inpatient claims and associated member enrollment data to assess total cost sharing, deductibles, copayments, and coinsurance associated with inpatient care for patients aged 18 to 64 years who were enrolled in employer-sponsored group or individual market health insurance plans between January 1, 2009, and December 31, 2013.
The study was considered exempt by the institutional review board of the University of Michigan and was conducted between July 1, 2015, and March 1, 2016. Data were deidentified.
We restricted claims to those generated from inpatient hospitalizations. We excluded all sets of claims for sex-specific primary diagnoses and procedures that were attributed to a member of the wrong sex and all claims for which the length of stay was less than 1 day. We also excluded all claims for which the aggregate insurer-allowed amount, total cost sharing, copayment, coinsurance, or amount applied to the deductible was less than zero, as well as duplicate claims for services provided to the same individual on the same day. To adjust for statistical outliers, we capped the maximum length of stay at the 99th percentile (684 days). By applying these criteria, 85 557 hospitalizations (1.1%) were excluded. Our final analytic sample included data for 7 316 763 hospitalizations between 2009 and 2013.
Our outcomes were total cost sharing, deductibles, copayments, and coinsurance for inpatient hospitalizations. These calculations were based on inpatient facility claims (excluding physician claims) for all hospitalizations, including those for which out-of-pocket spending was $0. We also analyzed the out-of-pocket spending associated with 7 inpatient diagnoses and procedures, chosen because they were among the more commonly occurring primary inpatient diagnoses and procedures during the study period: acute myocardial infarction, live birth, pneumonia, appendicitis, coronary artery bypass graft, total knee arthroplasty, and spinal fusion.13,14 These broad diagnosis- and procedure-related categories were compiled from International Classification of Diseases, Ninth Revision, Clinical Modification codes using the Agency for Healthcare Research and Quality’s Clinical Classification Software (eTable 1 in the Supplement).15
For mean costs to be comparable across the study years, we controlled for inflation and case-mix differences over time. All costs were inflation adjusted to 2013 dollars.16 Predicted cost sharing was calculated from generalized linear regression models that adjusted for the diagnosis related group weights of each hospitalization and included year fixed effects.
We used Wilcoxon rank sum tests to assess whether there was a statistically significant difference in total cost sharing across several patient and health plan characteristics in 2009 and 2013. We also estimated generalized linear regression models to evaluate the association between total cost sharing and plan and patient characteristics in 2013. Explanatory variables included age, sex, region, whether the patient was enrolled in an individual market plan, whether the patient was enrolled in a health maintenance organization, and whether the patient was enrolled in a consumer-directed health plan. In the HCCI database, each insurer flagged the members they considered to be enrolled in a consumer-directed health plan, which is typically a plan with a high deductible that is paired with a health savings account, flexible spending account, or health reimbursement arrangement. Individual market and consumer-directed health plans are not mutually exclusive.
The generalized linear regression model analysis was performed using a log link and γ distribution, with Huber-White Sandwich estimators17,18 for the SEs. The modified Park test was used to determine variance structure. All analyses were performed using Stata, version 14 (StataCorp); the postestimation margins command was used to obtain predicted costs.
Patients with multiple hospitalizations are more likely to reach their deductible and/or out-of-pocket maximum, which would lead to lower adjusted mean out-of-pocket costs per hospitalization. To examine this possibility further, we performed a sensitivity analysis using a sample restricted to the first inpatient hospitalization for each patient per year. Additional sensitivity analyses examined changes in the proportion of care that required out-of-pocket spending, total hospital payments, and the percentage of total hospital payments paid out of pocket by the patient as key drivers of out-of-pocket spending.
From 2009 through 2013, a total of 5 989 855 of 67 537 107 nonelderly, commercially insured adults (8.9%) in the HCCI database experienced an inpatient hospitalization. Among the admitted patients, 5 057 604 (84.4%) had 1 hospitalization, 661 751 (11.1%) had 2 hospitalizations, and 270 500 (4.5%) had more than 2 hospitalizations in a given year. During the study period, 5 074 367 (69.4%) of inpatient hospitalizations included in the sample were associated with some form of cost sharing.
Adjusted total cost sharing per inpatient hospitalization with percentage of increase is reported in Table 1. The cost increased by 37% from $738 in 2009 (95% CI, $736-$740) to $1013 in 2013 (95% CI, $1011-$1016). Growth in total cost sharing was lower in individual-market and consumer-directed health plans, although both had higher overall levels of cost sharing. In 2013, total cost sharing per inpatient hospitalization was highest among patients enrolled in individual health insurance plans, with adjusted total cost sharing of $1875 (95% CI, $1867-$1883), and consumer-directed health plans, with adjusted total cost sharing of $1219 (95% CI, $1216-$1223). In a sensitivity analysis restricting the sample to the first hospitalization for patients in each year, adjusted mean out-of-pocket costs were even greater (eTable 2 in the Supplement). Our findings indicate that the growth in cost sharing was driven both by increases in total payments per hospitalization, which grew at an annual rate of 3.3% during the study period (from a mean of $13 654 in 2009 to $16 091 in 2013), and by increases in the percentage of total hospital payments paid out of pocket by the patient, which rose from a mean of 11.2% in 2009 (95% CI, 11.1%-11.2%) to 12.7% in 2013 (95% CI, 12.7%-12.7%) after inflation and case-mix adjustments (eTable 3 in the Supplement).
The growth in total cost sharing varied by the type of out-of-pocket spending (Figure 1). From 2009 to 2013, the mean amount of coinsurance associated with hospitalizations increased by 33% from $518 in 2009 (95% CI, $516-$520) to $688 in 2013 (95% CI, $686-$690). The amount applied to deductibles increased by 86% from $145 in 2009 (95% CI, $144-$146) to $270 in 2013 (95% CI, $269-$271). In contrast, the mean copayment associated with an inpatient treatment fell by 27% from $75 in 2009 (95% CI, $74-$75) to $55 in 2013 (95% CI, $55-$56). This decrease in mean was largely because the percentage of inpatient hospitalizations requiring a copayment decreased over time from 25.0% (389 479 of 1 556 756 hospitalizations) in 2009 to 16.8% (236 085 of 1 401 232) in 2013 as opposed to the amount of the copayments falling (eTable 4 in the Supplement). Conversely, the proportion of inpatient hospitalizations requiring coinsurance grew from 50.6% (from 787 719 of 1 556 756 hospitalizations) in 2009 to 55.4% (776 283 of 1 401 232) in 2013 and, among hospitalizations that required coinsurance, adjusted mean coinsurance grew 21.7% during the study period from $1020 in 2009 to $1242 in 2013. The proportion of hospitalizations involving payment toward a deductible grew from 14.8% (230 400 of 1 556 756 hospitalization) in 2009 to 21.4% (299 864 of 1 401 232) in 2013, and, among those hospitalizations, the adjusted mean amount applied to the deductible grew 28.8% from $981 in 2009 to $1264 in 2013.
Geographic variation in cost sharing was also substantial. Between 2009 and 2013, total cost sharing per inpatient hospitalization increased in every state, with the smallest increases occurring in Rhode Island, which increased from $474 in 2009 to $490 in 2013 (3%), and Montana, which increased from $995 in 2009 to $1115 in 2013 (12%). The largest increases occurred in Georgia, which rose from $658 in 2009 to $1097 in 2013 (67%); Louisiana, which increased from $737 in 2009 to $1103 in 2013 (50%); and Colorado, which increased from $790 in 2009 to $1182 in 2013 (50%). In 2013, the states with the highest adjusted total cost sharing per inpatient hospitalization were Utah ($1323 per hospitalization [95% CI, $1298-$1348]), Alaska ($1290 per hospitalization [95% CI, $1215-$1366]), and Oregon ($1236 per hospitalization [95% CI, $1209-$1263]) (Figure 2).
During the study period, cost sharing varied substantially across common medical diagnoses and procedures for inpatients (Figure 3). The cost sharing associated with a live birth increased by 63% during the study period but remained relatively low overall, with adjusted total cost sharing growing from $159 in 2009 (95% CI, $151-$168) to $259 in 2013 (95% CI, $245-$272). By comparison, out-of-pocket spending associated with emergent hospitalization increased from 2009 to 2013 by 37% for acute myocardial infarction (from $1158 [95% CI, $1135-$1182] to $1586 [95% CI, $1562-$1610]) and 40% for acute appendicitis (from $1079 [95% CI, $1060-$1098] to $1509 [95% CI, $1487-$1531]). Cost sharing associated with procedures was lower in general, especially for coronary artery bypass graft, which grew from $585 in 2009 (95% CI, $572-$599) to $862 in 2013 (95% CI, $849-$876), a 47% increase in total cost sharing.
Results from multivariable regression models showed that enrollment in individual-market plans was associated with 74.3% higher (95% CI, 72.6% to 76.0%; P < .001) or $728 (95% CI, $711 to $745; P < .001) greater total cost sharing per inpatient hospitalization compared with total cost sharing for patients enrolled in group plans (Table 2). Enrollment in consumer-directed health plans was associated with 34.7% higher (95% CI, 34.1% to 35.3%; P < .001) or $340 greater (95% CI, $335 to $346; P < .001) total cost sharing per inpatient hospitalization compared with patients enrolled in other plans. Holding other factors constant, total cost sharing was lowest in the New York/New Jersey region at −3.4% (95% CI, −5.1% to −1.8%; P < .001) or −$34 (95% CI, −$50 to −$17; P < .001) compared with the New England region and highest in the Southwest at 49.8% (95% CI, 48.1% to 51.4%; P < .001) or $488 (95% CI, $472 to $504; P < .001) higher than the New England region.
To our knowledge, this study is the first national assessment of out-of-pocket spending for inpatient hospitalizations. From 2009 to 2013, total cost sharing associated with hospitalizations grew by more than 37% from $738 to $1013. This increase represents an annual growth rate of 6.5% compared with the 5.1% growth in health insurance premiums reported during this same period.19 To put our results into a broader context, this 6.5% annual growth in cost sharing for hospitalizations coincided with a notable slowing in the growth of overall health care spending, which rose at a rate of just 2.9% from 2009 to 2013.20
We found that the growth in cost sharing was driven primarily by increases in the amount applied to patients’ deductibles, which rose by 86%, and by increases in coinsurance, which grew by 33% during the study period, rather than by copayments. Our findings indicate a trend toward fewer plans requiring copayments at the point of service and more plans requiring higher coinsurance and deductibles after care is delivered.
Recent benefit surveys in the general population support our findings that deductibles and coinsurance are on the rise. The Kaiser Family Foundation21 reported that, in 2015, 81% of enrollees in employer-sponsored plans had annual deductibles averaging $1300, and 65% of the enrollees were subject to coinsurance requirements for inpatient hospitalizations. In addition, 72% of silver plans offered through the federal health insurance marketplace included coinsurance requirements for inpatient hospitalizations, with an average coinsurance rate of 26% of the cost of care.
Our findings are particularly notable given recent research suggesting that most Americans lack a basic understanding of the different forms of cost sharing associated with medical care. In their 2013 survey of enrollees in employer-sponsored health insurance plans, Lowenstein et al22 found that only 11% of the respondents could correctly estimate the cost sharing associated with a hospitalization when given information regarding the cost of the admission and plan benefits, and only 14% of the respondents were able to provide correct answers to a set of 4 multiple-choice questions relating to deductibles, copayments, coinsurance, and out-of-pocket maximums. Moreover, other research23 has found no evidence that patients search for the lowest-cost health care provider or services when faced with high deductibles. Increasing out-of-pocket spending coupled with the complexities of cost sharing mechanisms means that many poorly informed patients may face substantial financial risk when they are hospitalized. High and rising patient cost sharing also has implications for hospitals since any unpaid out-of-pocket costs add to hospitals’ uncompensated care burden. Insurers, employers, and policymakers could play a more active role in educating enrollees about the cost-sharing requirements of their plans. Health care professionals could work to improve their communication with patients around the issue of out-of-pocket costs as well.5,24
Our results also indicate that a large part of the variation in cost sharing for inpatient hospitalizations appears to be driven by the type of insurance product, with enrollment in individual market and consumer-directed health plans associated with significantly higher cost sharing for inpatient hospitalizations during the study period. Increasing cost sharing may also reflect a recent elevation in insurer and provider consolidation,25 which may limit competition and increase costs for beneficiaries.9 Future research should assess the market dynamics underlying patient cost sharing.
Finally, we found that cost sharing varied substantially across common medical diagnoses and procedures. Out-of-pocket spending related to hospitalization for acute myocardial infarction and acute appendicitis rose significantly and exceeded $1500 in 2013. Cost sharing associated with procedures was lower in general, especially for coronary artery bypass graft. These differences could be due to the fact that patients with emergent conditions may be less likely to have met their deductible or out-of-pocket spending maximum before hospitalization than are patients undergoing a procedure such as coronary artery bypass graft, which may involve more outpatient care and spending before hospitalization. It is also possible that negotiated rates and/or insurer pricing structure varied across services, especially as use of procedures changed over time.
Our analyses have some limitations. Although the HCCI database contains a large amount of rich data for analysis, it does not include information on specific benefit package features, such as premiums, total deductibles, or out-of-pocket maximums. We were also unable to definitively identify out-of-network care, which can contribute substantially to out-of-pocket spending. We did not assess the costs of physicians’ services during hospitalizations, costs after hospitalizations, or indirect costs (eg, lost productivity and transportation). Thus, our analyses captured only 1 major component of out-of-pocket costs. The full extent of cost sharing associated with hospitalizations is likely greater than our estimates.
The HCCI database does not include information from all commercial health insurers, and those that are included do not have uniform penetration across all states; thus, the state-level analyses should be interpreted with caution. However, restricting our sample to states where HCCI-contributing insurers had a high market share did not yield markedly different results (eTable 5 in the Supplement). Our analyses focused on cost sharing for nonelderly adults enrolled in commercial health insurance plans and did not address cost sharing among uninsured persons, children, elderly adults, or those enrolled in Medicare or Medicaid. Other studies26 have evaluated cost sharing in these populations and shown similar concerns regarding rising out-of-pocket spending.
Finally, our study period did not extend far enough to capture the implementation of several provisions of the Affordable Care Act of 2010 (ACA) that affect benefit design and have important implications for out-of-pocket spending moving forward. Beyond the cost-sharing implications of extending health insurance coverage to millions of previously uninsured Americans, the ACA also allowed many individuals who previously could purchase insurance coverage only through the individual market to take advantage of risk pooling and premium subsidies through participation in the health insurance marketplaces. Thus, our results are still notable for policymakers since they show that, immediately before these changes resulting from the ACA, enrollment in individual market plans was associated with the highest levels of cost sharing for hospitalizations.
Despite these limitations, our findings are particularly important in light of the provisions introduced by the ACA that could affect out-of-pocket spending. These provisions include cost-sharing subsidies for individuals and families with incomes below 400% of the federal poverty level. However, these subsidies apply only to coverage purchased through a state or federal marketplace and thus do not apply to health care benefits obtained through an employer, which represent most of the plans we studied using the HCCI database.27
Another key ACA provision was the introduction of an out-of-pocket spending maximum as part of the essential health benefits package. In 2016, maximum out-of-pocket spending for coverage purchased through the health insurance marketplace is capped at $6850 for individuals and $13 700 for families.28 More than 29 000 inpatient hospitalizations in our sample involved cost sharing in excess of $6850. Out-of-pocket maximums have the potential to reduce the burden of high amounts of cost sharing. However, premiums, cost sharing for care received out of network, and cost sharing for care that is not part of the essential health benefits package do not necessarily count toward these caps, thereby limiting the impact of this provision. Finally, the future implementation of the “Cadillac tax” on high-cost health plans, which Congress recently delayed from 2018 to 2020, is likely to result in more costs being shifted to patients.
Out-of-pocket spending is significant for inpatient hospitalizations, even among insured individuals, and grew substantially from 2009 through 2013. Variations in cost sharing highlight important features of insurance policies associated with a higher burden of out-of-pocket spending. With an estimated 85% of all commercial health insurance benefit packages requiring coinsurance for inpatient hospitalizations in addition to meeting an annual deductible,12 cost sharing for inpatient hospitalizations remains an important, if often overlooked, area for policy reform.
Corresponding Author: Emily R. Adrion, PhD, MSc, Center for Healthcare Outcomes and Policy, University of Michigan Medical School, 2800 Plymouth Rd, NCRC Bldg 14, Room G100-34, Ann Arbor, MI 48109 (firstname.lastname@example.org).
Accepted for Publication: May 24, 2016.
Published Online: June 27, 2016. doi:10.1001/jamainternmed.2016.3663.
Author Contributions: Dr Adrion had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Adrion, Ryan, Seltzer, Ayanian, Nallamothu.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Adrion, Ryan, Seltzer.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Adrion, Ryan.
Obtained funding: Adrion, Ryan, Ayanian.
Administrative, technical, or material support: Adrion, Ryan, Seltzer, Nallamothu.
Study supervision: Adrion, Ryan, Seltzer.
Conflict of Interest Disclosures: None reported.
Funding/Support: National Research Service Award T32 HS000053-24 from the Agency for Healthcare Research and Quality supported Dr Adrion’s effort on this project. Funding to access data of the Health Care Cost Institute was provided by the Institute for Healthcare Policy and Innovation at the University of Michigan. Dr Chen receives support from the Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation. She also receives funding from National Institute on Aging grant P01AG019783 and the Blue Cross Blue Shield of Michigan Foundation.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Previous Presentation: This study was presented at the 2016 Annual Research Meeting of AcademyHealth; June 27, 2016; Boston, Massachusetts.