Although commercial accountable care organization populations are healthy on average, some individuals might benefit from programs for high-risk patients to mitigate high levels of health care utilization.
Objectives: The study examined a commercial accountable care organization (ACO) population and then assessed the impact of an integrated care management program on medical spending and clinical event rates.
Study Design: Retrospective cohort study of high-risk individuals (n = 487) in a population of 365,413 individuals aged 18 to 64 years within the Mass General Brigham health system who were part of commercial ACO contracts with 3 large insurers between 2015 and 2019.
Methods: Using medical spending claims and other enrollment data, the study assessed the demographic and clinical characteristics, medical spending, and clinical event rates of patients in the ACO and its high-risk care management program. The study then examined the impact of the program using a staggered difference-in-difference design with individual-level fixed effects and compared outcomes of those who had entered the program with those of similar patients who had not entered.
Results: The commercially insured ACO population was healthy on average but included several hundred high-risk patients (n = 487). After adjustment, patients within the ACO’s integrated care management program for high-risk patients had lower monthly medical spending (by $1361 per person per month) as well as lower emergency department visit and hospitalization rates compared with similar patients who had yet to start the program. Accounting for early ACO departure decreased the magnitude of the program effects as expected.
Conclusions: Commercial ACO populations may be healthy on average but still include some high-risk patients. Identifying which patients might benefit from more intensive care management could be critical for reaping the potential savings.
Am J Manag Care. 2023;29(5):220-226. https://doi.org/10.37765/ajmc.2023.89355
Although commercial accountable care organization (ACO) populations might be healthy on average, some individuals might benefit from programs for high-risk patients. Identifying this subset of patients could be critical for reducing spending.
Medical spending growth remains a major threat to the sustainability of the health care system in the United States.1,2 Accountable care organization (ACO) contracts seek to better align payer and provider incentives. An estimated 40 million-plus Americans are part of an ACO, and 60% of these individuals are in a commercial ACO.3,4 The literature on altering spending growth for commercially insured, working-aged adults has yielded mixed results,5-11 with limited data on the extent to which commercial ACOs achieve financial benefits through improving health outcomes rather than discouraging utilization.
Much of the favorable ACO evidence with respect to cost savings has come from the traditional, fee-for-service Medicare program (eg, Medicare Pioneer ACO). The relevance of these Medicare ACO findings is questionable given program and population differences between traditional Medicare and those of commercial insurance plans.12 For example, employer-based commercial insurance plans cover working-aged persons and their dependents, who tend to be much younger and have lower spending and hospitalization rates compared with Medicare beneficiaries.
The ACO contract in theory both aligns the incentives of the provider organization with those of the payer and society writ large and shapes the incentive framework within which the organization makes investment and organizational decisions. In a well-designed contract, the ACO has incentives to improve the health of the patient population but fewer incentives to reduce spending by skimping on care or manipulating the risk pool. One additional challenge of commercial ACO contracts is that the patient population tends to be considerably younger and healthier than that of Medicare ACO contracts, which could affect the numbers of patients with high predicted levels of spending or the frequency of mutable sources of spending among these patients.
Indeed, the literature on the impact of commercial ACO contracts on spending growth has been mixed.5-11 Moreover, early studies have focused on individuals who are continuously enrolled in an ACO, who could differ from those who leave the ACO.5 Some ACOs also appear to have reduced spending by switching to lower-cost suppliers or providers; such price reductions alone may be difficult to sustain or unlikely to alter longer-term spending growth.13 Although an ACO contract applies to a defined population and creates incentives for the health care organization, the ACO also must account for the underlying levels and types of turnover within the population while determining investments in new programs or additional care processes.14 Any care management intervention initiated in the context of an ACO should then be assessed on an ongoing basis to determine whether health outcomes are improved while spending is mitigated, conditional on changes in the population. This is analogous to losses during follow-up in a clinical trial of a new investigation; differential losses across study arms create a serious competing risk, and substantial equal losses within the trial raise concerns about the relevance of the intervention and generalizability of the findings. To date, specific care intervention programs within commercial ACOs have not been evaluated. Our objective was to evaluate the effects of a care management program within the context of the ACO.
In this study, using data from one of the largest commercial ACOs in the United States, we examined the characteristics of the commercial ACO population and a high-risk subgroup. We then examined the impact of an integrated care management program for the high-risk group on medical spending and clinical event rates while accounting for ACO departures.12,15,16 The analyses help address whether care improvements and investment in intensive care management programs represent a feasible strategy for reducing spending growth within commercial ACO populations.
The study population included all individuals within the Mass General Brigham (MGB) health system who received health insurance from any 1 of the 3 largest contracted commercial insurers. MGB is a large academic health system providing care in Massachusetts and surrounding states.17 The ACO structure and health care delivery system are the same across all ACOs, and the ACO contracts and incentives are largely similar across insurance types.
We included all individuals who entered the risk contract from 2015 to 2019. Individuals could join the ACO at any point during the year with a qualifying insurance change. After 2015, the commercial risk contracts included multiple plan types (eg, health maintenance organizations, preferred provider organization, point of service). For all risk contracts, the health plans provided complete medical claims for individuals within the risk population, including claims for care provided outside of the MGB system; however, there were no systematic claims data from before ACO entry.
To minimize the inclusion of patients with overlapping noncommercial insurance, we restricted our population to patients 64 years or younger. We then excluded pediatric patients (< 18 years) because MGB pediatric care management was in a nascent stage during the study period.
ACO’s Integrated Care Management Program for High-risk Patients
As part of a Medicare demonstration program starting in 2006, MGB created a primary care practice–based integrated care management program for high-risk Medicare beneficiaries, supplementing care through care managers.17 Details evaluating the Medicare program are available in published articles.12,18,19 MGB later extended the program to younger, commercially insured patients.
Since its inception, the program has provided care to more than 40,000 patients.20 The care managers perform the following main functions: (1) assess patients for unmet care needs; (2) outline and implement care plans; (3) educate patients regarding emergency department (ED) avoidance strategies (eg, urgent care or telehealth options); (4) connect patients with appropriate community resources; (5) connect with inpatient care managers to ensure safe discharge planning; (6) conduct follow-up visits post discharge; and (7) partner with primary care physicians to improve patient engagement in ambulatory care. The care managers are supported by ancillary staff, including pharmacists and community resource specialists. Previously, we assessed the impact of this program on Medicare ACO beneficiaries.12,18,19
Each year, the health system identified a subgroup of the ACO population that was at risk for high levels of spending and hospitalizations. Eligibility criteria for this study included the following: (1) coverage through a group health insurance plan offered by 1 of 3 main insurers in Massachusetts starting in 2015-2019 and (2) being 18 years or older but younger than 65 years at the time of ACO entry.
The integrated high-risk care management program sought to systematically identify patients who had substantial illness burden and high levels of predicted spending, which could be amenable to care improvements. Specifically, during the study period, the ACO would risk-stratify all patients once annually using a sophisticated algorithm to identify high-risk patients using variables such as presence of chronic medical conditions, health care utilization patterns, and risk of having increased spending. The aim was to identify patients likely to have high levels of future medical spending whose spending was potentially mutable given improvements in care management. Details on the identification of candidates are available in published articles.15,16
After generating this high-risk list, each patient’s primary care team reviewed the list to further assess mutability and fit for the program. As each primary care team completed their review, care managers received the names of patients who met the entry criteria for the integrated care management program; the care managers did not receive information from the identification process (ie, they did not receive individual patient spending risk level or mutability information). Care managers set up individual assessments with each patient to review their specific medical and social needs and to develop a management plan with the goal of assessing all eligible patients during the next year. We found no association between the order of the assessments and predicted spending levels, which is consistent with using a level playing field approach for starting the program among eligible patients. However, there were additional sources of variation, including capacity constraints such that individuals might not automatically be entered into the program even if eligible.
We first examined the characteristics of the ACO population, tracking the year of entry and months of ACO membership. We then identified the patients who were eligible for the care management program and again tracked the timing of program entry and duration within the program. Participants could enter the ACO or care management program at any point during the study period, but they were required to be part of the ACO to enter the care management program.
Using medical spending claims and other enrollment data, we summarized demographic and clinical characteristics of patients in the ACO and in the high-risk care management program. We also calculated total medical spending, ED visit rates, and inpatient hospital admission rates per month for each individual. For the subgroup who entered the high-risk care management program, we described the patterns of these outcomes during the months preceding and following entry into the high-risk intervention program.
To assess the impact of the care management intervention, we exploited timing differences created by sequential ACO entry across years and annual risk evaluation (ie, compared high-risk participants who entered the program early vs later in time or a staggered difference-in-difference design with individual-level fixed effects as in prior analyses).12
Using linear regression models that included both individual-level and time fixed effects, we first compared outcomes among high-risk patients in the program with those who have yet to start the program. Because we sought to identify downstream clinical changes and hypothesized that such changes would require time to manifest themselves, we assessed program effects among those who were in the program 7 months or longer compared with those who had not yet started or were in the program for 6 months or less. Importantly, this approach attempts to account for secular trends by comparing participants who entered the high-risk program earlier with those who entered later, exploiting the staggered entry into the high-risk care program over time.
Several factors affected the program entry timing, including the timing of ACO entry, discrepancies between ACO entry and the annual risk assessments, practice-level lags in review of high-risk candidates, and capacity constraints of care managers within the program. To help assess the design validity, we compared the preprogram spending trends for those entering earlier vs later using multiple approaches for differentiating between early and later entry. We compared the preprogram spending trends for those entering the program early vs later in time; we did so both visually and by using a logistic regression model with an interaction term for calendar time and timing of entry. The trends were comparable.
We also used inverse probability weighting to adjust for potential differential attrition from the ACO that could influence estimates of the impact of care management within the ACO program, because in prior work we found substantial turnover in the commercial ACO population with differential attrition with respect to risk level, and we wanted to account for any factors related to leaving that might affect program entry or the probability of having unfavorable events. Specifically, we modeled the probability of staying in the ACO for 36 months or more or the maximum possible time an individual could have remained in the ACO if their ACO entry was later than January 2017. We calculated inverse probability weights by selecting baseline predictors of remaining in the ACO and employing stepwise backward selection, removing terms with P values greater than or equal to .2. The final model included age, ACO start year, chronic obstructive pulmonary disease (COPD), seizure disorders, plan type, and indicators for several physician groups within the system. We then calculated the inverse probability weight (IPW) based on 1 divided by the predicted probability of staying in the ACO.
We used the IPWs in linear, fixed-effect models evaluating monthly cost and event rates for the high-risk population, accounting for time period, time-varying medical condition indicators (diabetes, cardiorespiratory failure, cancer, asthma, depression or bipolar disorder, heart arrhythmias, COPD, seizure disorders, endocrine disorders), and a time-varying pregnancy indicator.
Finally, using the above analytic approaches, we also examined changes in accidental injuries, which should not be affected by the high-risk care program (ie, a falsification test). The Mass General Brigham Institutional Review Board approved this study.
Between 2015 and 2019, 365,413 adults entered the ACO. As expected, the overall ACO population was relatively young (mean age, 41 years), had low spending in their first month in the ACO (mean monthly spending, $476), and had few chronic diseases (Table21).
The ACO identified 487 individuals as being at high risk, and they entered the care management program for high-risk patients between 2015 and 2019. Compared with the broader ACO population, individuals in the high-risk program were older (mean age, 50 years), were more likely to have chronic medical and psychiatric conditions, and had higher medical spending in their month of ACO entry (mean spending, $3381) (Table21).
Participants entering over time had comparable spending and clinical event rates prior to program entry (eAppendix Figure 1 and eAppendix Table 1 [eAppendix available at ajmc.com]). In the months prior to program entry, the mean monthly spending amount was $3719; mean monthly ED visit rates were 102 per 1000 persons and hospital admission rates were 55 per 1000 persons. More than half (59%) remained in the ACO for at least 12 months after entering the care management program (eAppendix Figure 2).
Medical Spending in High-risk Patients
After adjustment and accounting for population losses through weighting, total monthly medical spending was $1361 less per month for those with at least 7 months in the program (95% CI, –$2542 to –$181) (Figure 121), which represents a 37% decrease in spending. No statistically significant differences in spending were associated with being in the program for 6 months or less compared with spending for those not yet in the program. As expected, ignoring the population losses yielded larger spending reductions (ie, larger effects when limiting the population to those continuously enrolled). Other sensitivity analyses yielded similar findings (eAppendix Table 2).
Clinical Event Rates
After adjustment and accounting for population losses, ED visit rates were 50 per 1000 persons lower per month (or 0.05 per person per month) for those with at least 7 months in the program (95% CI, –80 to –20 per 1000 persons) (Figure 221) compared with those who had not yet entered; the change represents a 47% decrease in ED visits. Hospitalization rates were 21 per 1000 persons lower per month for those with 7 months or more in the program (95% CI, –42 to –1 per 1000 persons) (Figure 221) compared with those who had not yet entered; this change represents a 39% decrease. As with spending, ignoring the population losses yielded larger outcome reductions. Other sensitivity analyses yielded similar findings (eAppendix Table 3). There also were no statistically significant changes with respect to accidental injuries (ie, the falsification test), as expected (eAppendix Table 4).
The commercially insured ACO population in this study was relatively young and healthy on average, as expected. The large population, however, included several hundred patients who were substantially sicker and for whom entry into an integrated care management program was associated with decreased spending and rates of unfavorable clinical events compared with outcomes for those who had not yet entered the program. No measurable changes in the outcomes occurred during the initial 6 months of being in the program, but there were sizable reductions after 7 months in the program. Many ACO members left, including high-risk patients, after entering the care management program, and the program-related changes were more modest after accounting for the ACO population losses. Prior work has demonstrated that ACO programs for Medicare patients yield significant reductions in clinical event rates and medical spending12; however, to our knowledge, this is one of the first studies demonstrating the feasibility of a targeted program within a real-world, commercially insured ACO population with attention to population stability and an intervention with prior data supporting its effectiveness (ie, a mechanism for the changes).
The number of high-risk patients within the program, however, was relatively small, representing a fraction of 1% of the overall ACO population. Nevertheless, the savings were substantial at the individual level, with a decrease of $1361 per person per month among those with at least 7 months in the program. Indeed, the additional care for the 327 individuals present in the high-risk program for at least 7 months resulted in an estimated aggregate savings of $445,000 per month or as much as $2.7 million over 6 months. One important contribution of this study is the demonstration that a sizable number (albeit a small percentage) of commercial ACO members have high levels of predicted spending and that a high-risk integrated care management program developed for Medicare ACO beneficiaries translates well to these high-risk commercial ACO patients.
These results are important given challenges with extending findings from the Medicare ACO literature to a commercially insured population. For example, commercially insured patients generally have fewer chronic medical conditions compared with Medicare beneficiaries; chronic conditions represent an attractive area for improving care given the greater predictability of spending and larger evidence base regarding approaches for reducing disease exacerbations compared with acute medical conditions.22 Moreover, because many commercial insurance plans employ a range of strategies to manage medical use whereas traditional Medicare arguably has few limits on use, it could be more difficult to find new opportunities to reduce spending among commercially insured vs Medicare populations.23-27 Nevertheless, there were substantial reductions in spending for high-risk patients as well as reductions in downstream unfavorable clinical event rates that likely contributed to the spending reductions; thus, it was possible to make progress with the commercially insured population and not only with the wind at one’s back such as within traditional Medicare.
The extent to which this type of intervention might generalize to more patients within the commercially insured population or extend to less sick individuals is unknown. At some point, there will be diminishing returns and more uncertain value to programs for those at high risk. Indeed, these findings highlight the importance of targeting those patients who would be most likely to benefit (ie, have clinically modifiable spending) and for whom the intervention offers a favorable value.
Arguably, ACOs should create incentives for improving care and reducing expensive downstream events such as hospitalizations and not just reduce use or short-term costs. Achieving downstream savings from fewer disease exacerbations, however, would require both investments and time before benefits manifest themselves.28 In a working-aged population that is relatively young and healthy, the payoff time could be long and yield could be low, in part because of the low prevalence of serious chronic illnesses and because of limited follow-up time in this population for whom employment changes alter insurance coverage and likely result in ACO departure.
Within ACO contracts, health systems and providers could alter spending growth for the population in several other ways, including reducing access to care (eg, limiting the numbers of physicians available, restricting access to care) or finding lower-cost providers/suppliers.2,29
Employer-sponsored commercial insurance plans also could alter the benefit design (eg, raise patient cost sharing) to reduce spending growth, but patients often do not differentiate well between necessary and unnecessary care and more stringent constraints could lead to poor outcomes.23,30,31 Altering benefit designs and reducing coverage generosity also could be less attractive to employers when there is greater competition for labor.32
Aligning the benefit designs with the ACO contracts, however, could help ACOs better manage the health of the population, such as requiring less cost sharing for care obtained within the ACO compared with care obtained elsewhere. Similarly, linking cost sharing with specific ACO programs (eg, reductions in cost sharing when patients adhere to protocols for contacting care managers before presenting to EDs) or even ACO tenure could align patient incentives closer to those of payers and providers.
The study has several limitations. First, this study was conducted in a single state, Massachusetts, which was one of the earliest states to expand insurance coverage and reduce spending growth; therefore, these results may not generalize to newer ACOs or states with fewer prior efforts to target spending growth. Second, this study was conducted within a single health system and ACO, although individuals received insurance from multiple insurance companies. Therefore, these results might not be generalizable to other health systems or ACOs. Third, we were unable to observe the benefit design structure or provider network associated with each individual’s insurance plan, which might have affected care sought during the study period.
The commercial ACO population is relatively young and healthy on average, which poses challenges in designing programs that reduce spending growth through improving medical care. Nevertheless, within these large populations, there may be a subset of patients who benefit from targeted programs to lower unfavorable downstream events and spending. In this study, entering the ACO’s care management program appears to reduce relative spending among these high-risk patients, with measurable effects manifesting after several months in the program, as expected. These findings offer hope for reducing the growth of spending within commercially insured populations. Identifying which patients might benefit and paying particular attention to the duration that they remain in ACO contracts and care programs are critical for tacking upwind (ie, lowering spending growth within a population with modest average spending).
Author Affiliations: McLean Hospital, Harvard Medical School (NMB), Belmont, MA; Mongan Institute, Massachusetts General Hospital (NMB, MP, MW, CV, JH), Boston, MA; Department of Psychiatry, Massachusetts General Hospital (NMB), Boston, MA; Department of Psychiatry (NMB), Department of Medicine (MLM, AF, GSM, JH), and Department of Health Care Policy (JH), Harvard Medical School, Boston, MA; Population Health Management, Mass General Brigham (CV, MMV, MLM, LB, LJ, GSM), Somerville, MA; Department of Medicine, Brigham and Women’s Hospital (AF), Boston, MA.
Source of Funding: Dr Hsu received funding support from the National Institutes of Health (P50MH115846, P01AG032952).
Author Disclosures: Dr Benson volunteers on the Epic Behavioral Health Specialty Steering Board. Ms Price is paid by Mass General Brigham for her work as a data analyst. All remaining authors were employed by Mass General Brigham, which holds shared-risk contracts including those involving accountable care organizations. Dr Hsu has also received grants from the National Institutes of Health and Agency for Healthcare Research and Quality and has consulted for CHA, USC, Columbia University, AltaMed, RWJF, and Community Servings.
Authorship Information: Concept and design (NMB, MMV, AF, LB, GSM, JH); acquisition of data (LB, GSM); analysis and interpretation of data (NMB, MP, MW, MLM, AF, LB, JH); drafting of the manuscript (NMB, MP, MMV, MLM); critical revision of the manuscript for important intellectual content (MW, CV, MMV, MLM, AF, LJ, GSM, JH); statistical analysis (NMB, MP); provision of patients or study materials (CV, AF); obtaining funding (LJ, GSM, JH); administrative, technical, or logistic support (MW, CV, MMV, MLM, GSM); and supervision (LJ, JH).
Address Correspondence to: Nicole M. Benson, MD, MBI, McLean Hospital, Harvard Medical School, 115 Mill St, Belmont, MA 02478. Email: email@example.com.
1. Montero A, Kearney A, Hamel L, Brodie M. Americans’ challenges with health care costs. Kaiser Family Foundation. Updated July 14, 2022. Accessed April 4, 2023. https://www.kff.org/health-costs/issue-brief/americans-challenges-with-health-care-costs/
2. Kramer W. Stairway to hell (with apologies to Led Zeppelin). Health Affairs. November 7, 2018. Accessed November 30, 2021. https://www.healthaffairs.org/do/10.1377/forefront.20181105.590047/
3. Muhlestein D, Saunders RS, Richards R, McClellan MB. Recent progress in the value journey: growth of ACOs and value-based payment models in 2018. Health Affairs. August 14, 2018. Accessed December 21, 2020. https://www.healthaffairs.org/do/10.1377/hblog20180810.481968/full/
4. Muhlestein D, Bleser WK, Saunders RS, Richards R, Singletary E, McClellan MB. Spread of ACOs and value-based payment models in 2019: gauging the impact of Pathways to Success. Health Affairs. October 21, 2019. Accessed November 13, 2021. https://www.healthaffairs.org/do/10.1377/forefront.20191020.962600/
5. Song Z, Ji Y, Safran DG, Chernew ME. Health care spending, utilization, and quality 8 years into global payment. N Engl J Med. 2019;381(3):252-263. doi:10.1056/NEJMsa1813621
6. Donohue JM, Barry CL, Stuart EA, et al. Effects of global payment and accountable care on medication treatment for alcohol and opioid use disorders. J Addict Med. 2018;12(1):11-18. doi:10.1097/ADM.0000000000000368
7. Song Z, Safran DG, Landon BE, et al. Health care spending and quality in year 1 of the alternative quality contract. N Engl J Med. 2011;365(10):909-918. doi:10.1056/NEJMsa1101416
8. Song Z, Rose S, Safran DG, Landon BE, Day MP, Chernew ME. Changes in health care spending and quality 4 years into global payment. N Engl J Med. 2014;371(18):1704-1714. doi:10.1056/NEJMsa1404026
9. Barry CL, Stuart EA, Donohue JM, et al. The early impact of the ‘Alternative Quality Contract’ on mental health service use and spending in Massachusetts. Health Aff (Millwood). 2015;34(12):2077-2085. doi:10.1377/hlthaff.2015.0685
10. Kaufman BG, Spivack BS, Stearns SC, Song PH, O’Brien EC. Impact of accountable care organizations on utilization, care, and outcomes: a systematic review. Med Care Res Rev. 2019;76(3):255-290. doi:10.1177/1077558717745916
11. Zhang H, Cowling DW, Graham JM, Taylor E. Five-year impact of a commercial accountable care organization on health care spending, utilization, and quality of care. Med Care. 2019;57(11):845-854. doi:10.1097/MLR.0000000000001179
12. Hsu J, Price M, Vogeli C, et al. Bending the spending curve by altering care delivery patterns: the role of care management within a Pioneer ACO. Health Aff (Millwood). 2017;36(5):876-884. doi:10.1377/hlthaff.2016.0922
13. Song Z. Accountable care organizations in the U.S. health care system. J Clin Outcomes Manag. 2014;21(8):364-371.
14. Benson NM, Price M, Vogeli C, et al. Population turnover and leakage in commercial ACOs. Am J Manag Care. 2023;29(4):e104-e110. doi:10.37765/ajmc.2023.89350
15. Haime V, Hong C, Mandel L, et al. Clinician considerations when selecting high-risk patients for care management. Am J Manag Care. 2015;21(10):e576-e582.
16. Vogeli C, Spirt J, Brand R, et al. Implementing a hybrid approach to select patients for care management: variations across practices. Am J Manag Care. 2016;22(5):358-365.
17. Bakshi S, Schiavoni KH, Carlson LC, et al. The essential role of population health during and beyond COVID-19. Am J Manag Care. 2021;27(3):123-128. doi:10.37765/ajmc.2021.88511
18. Hsu J, Price M, Spirt J, et al. Patient population loss at a large Pioneer accountable care organization and implications for refining the program. Health Aff (Millwood). 2016;35(3):422-430. doi:10.1377/hlthaff.2015.0805
19. Hsu J, Vogeli C, Price M, et al. Substantial physician turnover and beneficiary ‘churn’ in a large Medicare Pioneer ACO. Health Aff (Millwood). 2017;36(4):640-648. doi:10.1377/hlthaff.2016.1107
20. Population health. Mass General Brigham. Accessed November 3, 2021. https://www.massgeneralbrigham.org/en/research-and-innovation/centers-and-programs/population-health
21. Kautter J, Pope GC, Ingber M, et al. The HHS-HCC risk adjustment model for individual and small group markets under the Affordable Care Act. Medicare Medicaid Res Rev. 2014;4(3). doi:10.5600/mmrr2014-004-03-a03
22. Weinberger M, Oddone EZ, Henderson WG. Does increased access to primary care reduce hospital readmissions? Veterans Affairs Cooperative Study Group on Primary Care and Hospital Readmission. N Engl J Med. 1996;334(22):1441-1447. doi:10.1056/NEJM199605303342206
23. Claxton G, Damico A, Rae M, Young G, McDermott D, Whitmore H. Health benefits in 2020: premiums in employer-sponsored plans grow 4 percent; employers consider responses to pandemic. Health Aff (Millwood). 2020;39(11):2018-2028. doi:10.1377/hlthaff.2020.01569
24. Frost A, Barrette E, Kennedy K, Brennan N. Health care spending under employer-sponsored insurance: a 10-year retrospective. Health Aff (Millwood). 2018;37(10):1623-1631. doi:10.1377/hlthaff.2018.0481
25. Agarwal R, Mazurenko O, Menachemi N. High-deductible health plans reduce health care cost and utilization, including use of needed preventive services. Health Aff (Millwood). 2017;36(10):1762-1768. doi:10.1377/hlthaff.2017.0610
26. 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
27. Wharam JF, Zhang F, Eggleston EM, Lu CY, Soumerai S, Ross-Degnan D. Diabetes outpatient care and acute complications before and after high-deductible insurance enrollment: a Natural Experiment for Translation in Diabetes (NEXT-D) study. JAMA Intern Med. 2017;177(3):358-368. doi:10.1001/jamainternmed.2016.8411
28. Thompson A, Guthrie B, Payne K. Using the payoff time in decision-analytic models: a case study for using statins in primary prevention. Med Decis Making. 2017;37(7):759-769. doi:10.1177/0272989X17700846
29. Chernew ME, Newhouse J. Health care spending growth. In: Pauly MV, McGuire TG, Barros PP, et al, eds. Handbook of Health Economics. Elsevier; 2011:1-43.
30. Hsu J, Price M, Huang J, et al. Unintended consequences of caps on Medicare drug benefits. N Engl J Med. 2006;354(22):2349-2359. doi:10.1056/NEJMsa054436
31. Reed ME, Graetz I, Fung V, Newhouse JP, Hsu J. In consumer-directed health plans, a majority of patients were unaware of free or low-cost preventive care. Health Aff (Millwood). 2012;31(12):2641-2648. doi:10.1377/hlthaff.2012.0059
32. Morath E, Ip G. Tight labor market returns the upper hand to American workers. Wall Street Journal. June 20, 2021. Accessed November 30, 2021. https://www.wsj.com/articles/tight-labor-market-returns-the-upper-hand-