Schizophrenia, chronic pain, and multiple medical diagnoses at enrollment into a large managed Medicaid system are associated with long-term high health care utilization.
Objectives: Prior studies have had difficulty predicting which patients will have persistent high utilization past 1 year within the Medicaid population. The objective of this study was to examine the medical diagnoses at the time of enrollment of patients with persistent high health care utilization over 24 months following enrollment in Medicaid managed care in a large integrated care setting.
Study Design: Retrospective cohort study in a large integrated managed health care system.
Methods: We identified a cohort of high utilizers (top 5% of health care costs in 2014) and extracted their electronic health record data (2014-2016). Differences in baseline characteristics of high utilizers and the general Medicaid population were determined using bivariate analysis. We used multivariable regression to determine the independent association between medical comorbidities and demographics with persistent high health care utilization over the 2 years following enrollment.
Results: Compared with the general Medicaid managed care enrollee population, schizophrenia was the only mental health diagnosis at the time of enrollment associated with persistent high health care utilization (risk ratio [RR], 1.50; 95% CI, 1.20-1.86). Additional characteristics associated with persistent high utilization included age between 31 and 50 years (RR, 1.20; 95% CI, 1.02-1.41), dual enrollment in Medicaid and Medicare (RR, 1.26; 95% CI, 1.09-1.45), chronic pain diagnoses (RR, 1.26; 95% CI, 1.04-1.53), and multimorbidity (RR, 1.43; 95% CI, 1.25-1.63).
Conclusions: Among adults newly enrolled in Medicaid managed care, certain diagnoses noted at the time of enrollment into the plan are associated with persistent high health care utilization over the first 2 years, suggesting that targeting early supportive case management to these individuals could optimize care and reduce health care costs.
Am J Manag Care. 2021;27(8):340-344. https://doi.org/10.37765/ajmc.2021.88725
Diagnoses of mental health conditions, substance use, and other comorbid medical conditions at enrollment for Medicaid patients can predict high utilization over the first year. Among high utilizers, only patients with schizophrenia, chronic pain, and comorbid medical illness showed significantly increased odds of long-term high utilization over the second year. Although these data are limited to a single health care system, they provide new evidence regarding Medicaid enrollees in managed care at risk for long-term high health care spending within 2 years of enrollment.
Health care spending in the United States has been accelerating over the past decade, with Medicaid contributing to a significant portion.1 In 2018, Medicaid represented 16% of total health care spending in the United States, more than half of which was attributable to Medicaid managed care plans.1 Coordinated supportive health care services have been shown to reduce use of inpatient hospital resources among Medicaid enrollees, suggesting that these patients have otherwise unmet health needs due to lack of coordination of care.2 However, Medicaid spending continues to grow despite increasing enrollment of Medicaid enrollees into comprehensive managed care plans.3 Individuals with serious mental illness (SMI) may benefit from this shift into managed care organizations following the passage of the Affordable Care Act.4
Lack of mental health integration may be a critical barrier to optimizing health care quality and containing spending among Medicaid managed care enrollees with mental illness.5 The lack of coordination between primary care and behavioral health services, exacerbated by carve-outs (contracted agreements with special care providers) of funding for mental health care, can result in missed opportunities and suboptimal treatment strategies.5 On average, Medicaid spending is 4 times higher for enrollees with SMI compared with spending for individuals without it.4 The highest Medicaid utilizers (“high utilizers”) are those who account for the top 5% of expenditures in any given year and are more likely to have SMI and chronic conditions such as major depression, schizophrenia, and bipolar disorder.6-8
Initial observational studies of intensive case management for high utilizers have shown benefits, but these benefits disappear when studied in randomized controlled trials.9 This lack of effectiveness may be explained by the temporal variability in utilization. As a consequence, for most Medicaid enrollees, prior periods of excessive utilization may be a poor predictor of high utilization in subsequent years.6-8 In addition, population-wide care for this vulnerable population is rarely focused on specific medical conditions.9 Identifying vulnerable Medicaid enrollees who are at risk of persistent high health care utilization after 1 year, particularly in managed care settings, can help better identify at-risk individuals in population management.
In this retrospective cohort study, we leverage the rich, longitudinal health care data from a managed care integrated health care system in California to examine the association between specific patient characteristics and persistent high health care utilization among adults in Medicaid managed care who were high utilizers in their first year of enrollment. Leveraging a stable, longitudinal cohort of managed Medicaid recipients, we hypothesized that certain patient comorbidities can be identified as risk factors for persistent high health care utilization.
Setting and Participants
This is a retrospective cohort study using electronic health record data from Kaiser Permanente Northern California (KPNC), a large multispecialty health care system serving nearly 4.4 million people, including more than 300,000 Medicaid patients.10 We included nonelderly adults (aged 18-64 years) who were newly enrolled (no enrollment in the prior 12 months) in Medicaid managed care between June and December 2014 (n = 28,209). Previously, cost-based definitions have been used when using Medicaid claims data to define high utilizers.11 We identified the subset of short-term high utilizers (n = 1414), individuals who were in the top 5% of utilization costs per patient during their first 12 months of enrollment (June 2014-December 2015). Costs calculated included all inpatient and outpatient costs, including drug costs; however, costs of inpatient psychiatric hospitalizations were excluded. Two individuals were excluded because they were not enrolled in KPNC Medicaid managed care for at least a month in their first and second year of enrollment. Patients were all enrolled in KPNC insurance during the time of the study but did have churning between Medicaid and commercial KPNC insurance. Persistent high utilizers were defined as the subset of patients from the first-year cohort who had persistent high utilization in the second year after their enrollment (June 2015-December 2016) (eAppendix Figures 1 and 2 [eAppendix available at ajmc.com]).
Outcomes of Interest
Covariate information was collected from data generated during the first 12 months of enrollment, prior to the outcome of interest. We identified several factors that may be associated with high health care utilization risk, such as demographics (age, gender, race, English as first language, neighborhood median income), insurance coverage (dual enrollment in Medicaid and Medicare, months enrolled), prior primary care health services use, and comorbid condition diagnoses as defined by International Classification of Diseases, Ninth Revision and Tenth Revision codes12 (severe depression, anxiety, bipolar disorder, schizophrenia, chronic pain, active cancer, alcohol abuse, and drug abuse). We also included a measure of chronic comorbid conditions using the Charlson Comorbidity Index (CCI).13
Bivariate analyses were conducted using χ2 tests to assess differences in the high-utilizer and non–high-utilizer groups over the first year. Multivariable Poisson regression, with a robust variance estimator, was used to determine the association between SMI and high utilization (top 5% of utilization during the year of concern) over the first and second years, controlling for demographics, clinical characteristics, and past utilization in 2 separate models. All variables were included in the 2 models regardless of statistical significance in univariate models. The analyses were conducted in SAS version 9.4 (SAS Institute), and the study was approved by the KPNC Institutional Review Board.
We identified 1428 adults who were in the top 5% of spenders during their first year of Medicaid managed care enrollment, of whom 14 patients died over the second year of enrollment, leaving 1414 patients in the second-year cohort (Table 1). The median length of enrollment per year was 365 days for both high utilizers and non–high utilizers over the first year after enrollment. During the first year after enrollment, patients with a CCI score greater than 1 and a diagnosis of any SMI, chronic pain, active cancer, or substance use disorder were significantly more likely to be high utilizers (Table 2). Within the cohort of the first-year high-utilizer population, only 667 patients (47%) had persistent high utilization in the second year. Only certain patient characteristics were noted to be associated with continued risk for being a persistent high utilizer in the second year after their enrollment (Table 1). Controlling for other factors, individuals with schizophrenia had higher risk of being persistent high utilizers into the second year (risk ratio [RR], 1.50; 95% CI, 1.20-1.86) relative to those without any mental health diagnosis (Table 2). In contrast, anxiety, depression, comorbid anxiety and depression, and bipolar disorder were not associated with persistent high utilization.
Other factors associated with persistent high utilization included age between 31 and 50 years (reference: aged 18-30 years; RR, 1.20; 95% CI, 1.02-1.41), dual enrollment in Medicaid and Medicare (RR, 1.26; 95% CI, 1.09-1.45), having a chronic pain diagnosis (RR, 1.26; 95% CI, 1.04-1.53), and having a CCI score of 2 or more (RR, 1.43; 95% CI, 1.25-1.63). Enrollment in Medicaid managed care for less than a full 12 months in the first year was associated with lower risk of persistent high utilization (RR, 0.69; 95% CI, 0.49-0.97) (Table 2). Patients with the following characteristics were less likely to be high utilizers in the second year despite being high utilizers in the first year after enrollment: female sex (RR, 0.79; 95% CI, 0.70-0.88) and obesity relative to a body mass index in normal or overweight range (RR, 0.87; 95% CI, 0.77-0.97).
Only a subset of the high-utilizer population has persistent increased utilization in the second year of enrollment into managed Medicaid. In this managed integrated care setting, schizophrenia was associated with persistent health care utilization among adult Medicaid managed care enrollees, whereas other mental health diagnoses were not associated with it. Previous predictive modeling studies have also identified schizophrenia as a risk factor for short-term health care utilization.14 In addition, this study is one of the first to identify differences in risk of persistent high utilization by type of mental illness. The common use of carve-outs of mental health care delivery services from managed Medicaid as seen in the population studied could be leading not only to the exacerbation of poor care for their mental health diagnoses, but also to overall increased medical costs for the managed care systems that are treating other medical problems not directly associated with their mental illness.
This study identified additional risk for persistent high health care utilization in clinically complex subgroups. The finding of chronic pain being associated with persistent high utilization builds on prior evidence regarding its association with acute health care utilization. A study in an integrated health system has shown an association with high acute use of hospital services associated with diagnoses of chronic pain.15 The finding of higher risk among middle-aged (31-50 years) and male enrollees was not anticipated. However, it may reflect the changing demographics of the Medicaid population due to the Affordable Care Act enrollment of newly eligible middle-aged adults and men who might have had no previous health care coverage, resulting in high levels of unmet health care needs.16 In addition, pregnancy-related costs in women could be associated with high utilization over the first year of enrollment that decreases in the second year.
Two other factors were associated with lower risk of persistent high health care utilization in this population. Individuals enrolled in Medicaid managed care for less than 12 months had lower risk of persistent high utilization relative to individuals with a full 12 months of coverage. Lower rates of enrollment may reflect access to other types of health insurance and spending rather than gaps in access to medical care services. However, our finding of a protective effect of obesity was unexpected. It is possible that body mass index is correlated with some unmeasured factors or that obese individuals may be targeted for health system interventions at the time of enrollment that positively affect health services use in the following months. Further research is needed to refute or confirm these findings. In addition, use of any primary care was not associated with lower risk of persistent high utilization. Although prior utilization might be an indicator of health care needs, we expected that increased access to preventive care services would reduce the need for health care services over time. Additional research to understand how these patients are using primary care might identify potential missed opportunities.
Strengths and Limitations
Strengths of this study include the ability to follow newly enrolled Medicaid managed care enrollees for 2 years following entry. In addition, we can use information collected in the first year of enrollment to predict persistent high health care utilization. However, this study also has several limitations. First, this study uses cost data generated within the health care system to identify high utilizers. However, due to lack of integration in payment and care delivery related to carve-outs, information about behavioral health services incurred outside of this system is missing. Consequently, individuals with high behavioral health costs incurred by out-of-system contractors could have had lower measured costs. The findings in this study suggest associations but do not necessarily prove causation. The findings may not be generalized to all Medicaid managed care populations, because KPNC’s Medicaid population may not be representative of the national population. However, the findings may be highly relevant to other integrated care delivery systems that are increasingly providing care to vulnerable Medicaid enrollees.
Overall, this study focuses only on the characteristics associated with high utilization in the 2 years following enrollment when there can be pent-up health care demand. The driving forces behind high utilization in a more established managed Medicaid population can differ from those found to be associated with high utilization after enrollment here. Therefore, the findings in this study can be useful in targeting patients early in their Medicaid enrollment but may not apply to longer-term managed Medicaid patients.
Leveraging rich, longitudinal electronic health record data, we examined the association between patient characteristics at enrollment and the risk of persistent high health care utilization among adults newly enrolled in Medicaid managed care. Diagnoses of schizophrenia, chronic pain, and multimorbidity were found to be associated with persistent high health care utilization into the second year following enrollment. Targeting at-risk patients for intensive case management during the first year of managed care enrollment may be one important strategy for reducing costs and increasing the value of care for Medicaid enrollees.
The authors would like to acknowledge Anna Bavykina, MD, The Permanente Medical Group, for her guidance regarding the health care delivery context. The authors would also like to acknowledge Sarita Mohanty, MD, MPH, MBA, National Medicaid at Kaiser Permanente, for her comments on the preliminary findings.
Author Affiliations: Department of Hospital Based Specialty, Kaiser Permanente Oakland Medical Center (SB), Oakland, CA; Division of Research, Kaiser Permanente Northern California (AA, PK, ASA), Oakland, CA.
Source of Funding: This work was funded by the Kaiser Permanente Northern California Community Benefit grants program and The Permanente Medical Group via an administrative grant from the Health Care Delivery and Policy Section at Kaiser Permanente Division of Research.
Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (SB, PK, ASA); acquisition of data (SB, AA, PK, ASA); analysis and interpretation of data (SB, AA, ASA); drafting of the manuscript (SB); critical revision of the manuscript for important intellectual content (SB, AA, ASA); statistical analysis (SB, AA, PK); obtaining funding (ASA); administrative, technical, or logistic support (SB); and supervision (ASA).
Address Correspondence to: Somalee Banerjee, MD, MPH, Kaiser Permanente Oakland Medical Center, 3600 Broadway, Oakland, CA 94611. Email: Somalee.Banerjee@kp.org.
1. Hartman M, Martin AB, Benson J, Catlin A; National Health Expenditure Accounts Team. National health care spending in 2018: growth driven by accelerations in Medicare and private insurance spending. Health Aff (Millwood). 2020;39(1):8-17. doi:10.1377/hlthaff.2019.01451
2. Kelley L, Capp R, Carmona JF, et al. Patient navigation to reduce emergency department (ED) utilization among Medicaid insured, frequent ED users: a randomized controlled trial. J Emerg Med. 2020;58(6):967-977. doi:10.1016/j.jemermed.2019.12.001
3. Garfield R, Hinton E, Cornachione E, Hall C. Medicaid managed care plans and access to care: results from the Kaiser Family Foundation 2017 survey of Medicaid managed care plans. Kaiser Family Foundation. March 5, 2018. Accessed November 20, 2020. https://www.kff.org/medicaid/report/medicaid-managed-care-plans-and-access-to-care-results-from-the-kaiser-family-foundation-2017-survey-of-medicaid-managed-care-plans/
4. Hall JP, LaPierre TA, Kurth NK. Medicaid managed care: issues for enrollees with serious mental illness. Am J Manag Care. 2019;25(9):450-456.
5. Tuck K, Smith E. Behavioral health coverage in Medicaid Managed Care. Institute for Medicaid Innovation. April 2019. Accessed January 27, 2020. https://www.medicaidinnovation.org/_images/content/2019-IMI-Behavioral_Health_in_Medicaid-Report.pdf
6. Harris LJ, Graetz I, Podila PSB, Wan J, Waters TM, Bailey JE. Characteristics of hospital and emergency care super-utilizers with multiple chronic conditions. J Emerg Med. 2016;50(4):e203-e214. doi:10.1016/j.jemermed.2015.09.002
7. Rentas KG, Buckley L, Wiest D, Bruno CA. Characteristics and behavioral health needs of patients with patterns of high hospital use: implications for primary care providers. BMC Health Serv Res. 2019;19(1):81. doi:10.1186/s12913-019-3894-7
8. Kangovi S, Barg FK, Carter T, Long JA, Shannon R, Grande D. Understanding why patients of low socioeconomic status prefer hospitals over ambulatory care. Health Aff (Millwood). 2013;32(7):1196-1203. doi:10.1377/hlthaff.2012.0825
9. Johnson TL, Rinehart DJ, Durfee J, et al. For many patients who use large amounts of health care services, the need is intense yet temporary. Health Aff (Millwood). 2015;34(8):1312-1319. doi:10.1377/hlthaff.2014.1186
10. Escobar GJ, Plimier C, Greene JD, Liu V, Kipnis P. Multiyear rehospitalization rates and hospital outcomes in an integrated health care system. JAMA Netw Open. 2019;2(12):e1916769. doi:10.1001/jamanetworkopen.2019.16769
11. Mann C. CMCS informational bulletin: targeting Medicaid super-utilizers to decrease costs and improve quality. Medicaid.gov. July 24, 2013. Accessed January 10, 2021. https://www.medicaid.gov/federal-policy-guidance/downloads/cib-07-24-2013.pdf
12. World Health Organization. International Statistical Classification of Diseases and Related Health Problems. World Health Organization; 2004.
13. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8
14. Ziring J, Gogia S, Newton-Dame R, Singer J, Chokshi DA. An all-payer risk model for super-utilization in a large safety net system. J Gen Intern Med. 2018;33(5):596-598. doi:10.1007/s11606-018-4346-5
15. Romanelli RJ, Shah SN, Ikeda L, et al. Patient characteristics and healthcare utilization of a chronic pain population within an integrated healthcare system. Am J Manag Care. 2017;23(2):e50-e56.
16. Mazurenko O, Balio CP, Agarwal R, Carroll AE, Menachemi N. The effects of Medicaid expansion under the ACA: a systematic review. Health Aff (Millwood). 2018;37(6):944-950. doi:10.1377/hlthaff.2017.1491