This paper utilizes latent class analysis to identify subgroups of complex conditions and of super-utilizers among health center patients to inform clinically tailored efforts.
Objectives: Existing literature indicates that multimorbidity, mental health (MH) conditions, substance use disorders (SUDs), and social determinants of health are hallmarks of high-need, high-cost patients. Health Resources and Services Administration–funded health centers (HCs) provide care to nearly 30 million patients, but data on their patients’ complexity and utilization patterns are limited. We identified subgroups of HC patients based on latent concepts of complexity and utilization.
Study Design: We used cross-sectional national data from the 2014 Health Center Patient Survey and latent class analyses to identify distinct and homogenous groups of complex high-utilizing patients aged 18 to 64 years.
Methods: We included indicators of chronic conditions (CCs), MH, SUD risk, and health behavior to measure complexity. We used number of outpatient and emergency department visits in the past year to measure utilization.
Results: HC patients were separated in 9 distinct groups based on 3 complexity latent classes (MH, multiple CCs, and low risk) and 3 utilization classes (low, high, and superutilizers). Conditions associated with each subgroup differed. The highest prevalence of bipolar disorder (45%) and high SUD risk (6%) was observed among MH superutilizers, whereas the highest prevalence of cardiovascular disease (48%) and obesity (96%) was seen among CC superutilizers. Most MH superutilizer patients concurrently had MH conditions and obesity and were smokers, but most CC superutilizer patients concurrently had hypertension, obesity, and cardiovascular disease.
Conclusions: Our examination of complexity and utilization indicated distinct HC patient populations. Managing the care of each group may require different targeted intervention approaches such as multidisciplinary care teams that include MH providers or specialists.
Am J Manag Care. 2022;28(2):66-72. https://doi.org/10.37765/ajmc.2022.88751
We identified 9 latent subgroups that represented the health status and utilization profiles of complex patients who use Health Resources and Services Administration–funded health centers. The findings demonstrate the importance of understanding health behaviors, physical and mental health conditions, and utilization patterns to develop effective interventions for complex, high-utilizing patients.
A small proportion of individuals incur the majority of health care spending in the United States.1 The quest to reduce and contain the escalating costs of care in the United States has generated extensive research on identifying high-cost patients, the reasons for their high costs, and what can be done to reduce those costs. A significant number of studies have focused on high utilization of services, categorizing patients based on frequent emergency department (ED) visits and hospitalizations.2-6 Research aimed at distinguishing characteristics of high utilizers (HUs) is dominated by assessing complicating factors such as mental health (MH) conditions, substance use disorders (SUDs), and homelessness.2-4,7 This has led to using labels such as “high-need high-cost” and calls for effectively managing these patients’ needs to reduce costs and improve their health.2,8 Most research on these patients is focused on populations with specific forms of coverage and patients of specific health care providers that grapple with identifying ways to address their patients’ needs and reduce utilization and costs. For example, some studies have focused on specific populations of payers, such as Medicaid, private insurance, or Medicare, or patients of integrated delivery systems or specific hospitals.4,5,7,9-11
Information that characterizes complex HUs is limited for nearly 30 million low-income and uninsured patients served by Health Resources and Services Administration (HRSA)–funded health centers (HCs), also called community or federally qualified HCs.4,9,10,12 HCs may be funded under Community HC, Migrant HC, Health Care for the Homeless, or Public Housing Primary Care programs.
HCs are the cornerstone of the safety-net system of care in the United States and provide comprehensive and culturally competent primary care and social services regardless of the ability to pay at more than 13,000 delivery sites.12 Evidence shows that HC patients have lower rates of ED visits and hospitalizations compared with patients served at other primary care providers.13,14 Yet, HCs serve patients with a high burden of disease, including diabetes (15%), hypertension (28%), asthma (6%), overweight/obesity (23%), depression (50%), generalized anxiety (35%), and panic disorders (19%).12,15 Additionally, these conditions are complicated by high rates of high-risk behaviors such as smoking and SUDs.16,17 This high burden of disease is further complicated by the disadvantages of being low income and uninsured and is likely a barrier to improving outcomes of care.18,19 Thus, successful provision of effective primary care in HCs has a crucial role in the overall success of efforts to improve population health and reduce costs.
There is wide variability in defining complexity in the literature, making it difficult to identify and treat such patients.8,20 Complex patients are frequently distinguished by presence of multiple chronic conditions (CCs), complicating MH diagnosis and SUDs, risky behaviors, social determinants of health (SDOH) such as unstable housing and employment, and environmental factors.8,21-23 Conceptualization of complexity has varied by health care discipline, and various conceptual frameworks have been developed to identify and describe complexity and quantify the extent of the relationship among complexity aspects.4,21,23-27
Joint identification of complex patients and HUs can be instrumental in effective treatment of these patients, which includes delivery of a coordinated combination of medical, MH, and social interventions.2,28-30 This moves away from focusing on a specific condition or disease or treating complex patients as homogenous groups.28,29 A deeper understanding is needed of the confluence of factors that lead to complexity and of the subsequent relationship of complexity with high utilization of services.31
In this context, we aimed to simultaneously characterize complexity and high utilization among HC patients aged 18 to 64 years by identifying distinct subgroups. We described each population by examining the prevalence of specific conditions within each group. We also identified co-occurrence of various indicators of complexity and HUs for patients within each group. This information would allow for a better understanding of which subgroups of HC patients need targeted and focused interventions.
Data and Sample
We used the 2014 Health Center Patient Survey, a nationally representative cross-sectional survey of HC patients that included information on patient demographics, health care utilization, health conditions, and behaviors. The 3-stage sampling design included the selection of 169 HC organizations, 520 HC sites within those organizations, and a random sample of HC patients who were eligible if they had visited the HC once in the past 12 months and were interviewed during their second appointment. A total of 7002 patient interviews were surveyed, representing 59.1% of all patients screened. Patients were excluded if they refused, discontinued the screening interview, or did not have a prior visit. The survey completion rate was 91.4%. We restricted our sample to patients aged 18 to 64 years to reduce significant variations in health conditions, utilization patterns, and insurance coverage between these adults and children or older adults.32 Our final analytic sample size totaled 5040.
We included 2 sets of variables to measure the latent concepts of complexity and high utilization. To measure complexity, we included health indicators, MH, SUD, and health behaviors guided by the literature.4,22 We used direct, objective health indicators including having hypertension, diabetes, asthma, cancer, stroke, chronic obstructive pulmonary disease (COPD), liver disease, weak/failing kidneys, and cardiovascular disease (CVD). We also included indicators of overweight/obesity status. Indicators of MH included depression, generalized anxiety, panic disorder, schizophrenia, and bipolar disorder. We also included an indicator of high risk of SUD based on the World Health Organization Alcohol, Smoking, and Substance Involvement Screening Test algorithm. We used current or former smoking (vs never) as a health behavior indicator.
To measure HUs, we used self-reported number of ED and outpatient visits in the past year. To help understand unique characteristics of latent classes, we examined demographics such as age, gender, race/ethnicity, education, and employment. We also examined health insurance and poverty. In addition, we examined subjective or less direct health indicators including self-assessed health status, any functional limitations (any instrumental activities of daily living), homelessness, and veteran status. Further detail on the constructions of these measures can be found in eAppendix Table 1 (available at ajmc.com).
We used latent class analysis (LCA) to simultaneously identify latent concepts of complexity and high utilization.33 LCA is an established methodology used to define subgroups of a population based on an investigator-selected list of observed characteristics and relates them to an unobserved latent variable indicated by class membership using structural equation modeling techniques (see LCA section of the eAppendix). The results identify mutually exclusive and parsimonious and interpretable classes of individuals who share some similarities but focus on the most notable attributes of each class. The model-generated highest posterior probabilities are used to attribute each individual to a class. This approach is preferred to conventional regression approaches.4,9,11
We conducted the LCA using Mplus 7 (Muthén & Muthén) and used weights to account for the survey sampling design. We identified the classes for each concept by fitting models with 2 to 4 classes and determined the best fit by assessing likelihood ratio (lower values indicate better fit), Akaike and Bayesian information criteria (lower values indicate better distinction between classes), and entropy statistic (higher values indicate accurate class identification given the number of classes).4 We took into account the sample size of each class and did not retain those with less than 3% of the sample. We included age and gender as primary confounders of complexity and health insurance status as the primary confounder of high utilization.
We described the resulting classes in Stata 15 (StataCorp) using demographics, health status, and utilization variables. We also used Gephi version 0.9.2 (Association Gephi) to visualize the interrelationships among various indicators within a class. eAppendix Figure 1 shows the prevalence of conditions (bubbles) and the co-occurrence of conditions (lines) where bigger bubbles show higher prevalence and thicker lines show a higher rate of co-occurrence. The visualization of the line thickness indicates the independent co-occurrence between conditions. We calculated the thickness of the lines by summing the sampling weights of dyads of conditions among all patients with these conditions and rescaling them from 0 to 1 to visually present the relationship between dyads.
Sample characteristics showed high rates of hypertension (39%), diabetes (20%), obesity (50%), depression (44%), anxiety (30%), and current or past smoking (47%) (eAppendix Table 2). Additional health status indicators showed many with fair/poor health (42%) and functional limitations (38%). The majority were Medicaid beneficiaries (54%), and many were not in the labor force (42%) or were employed (40%). Patients had a mean of 8.3 outpatient and 1.6 ED visits last year.
The LCA led to the identification of 3-class complexity and 3-class high utilization latent variables based on best fit and parsimony (Table 1). Figure 1 shows an MH conditions class (32%) dominated by highest prevalence of conditions such as depression (92%), general anxiety (85%), panic disorder (48%), bipolar disorder (39%), and schizophrenia (10%). This class also had the highest prevalence of smoking (70%). A second multiple CC class (27%) was dominated by highest prevalence of hypertension (80%), obesity (71%), diabetes (46%), and asthma (14%). The third class was the largest (41%) and dominated by lowest prevalence of all the examined conditions and was classified as low risk (LR). These individuals may still be at risk for developing more severe and complex conditions due to dominance of risk factors such as smoking, obesity, CVD, and depression.
The prevalence of indicators for the 3 classes of high utilization indicated low utilizers (LUs; 73%), HUs (24%), and superutilizers (SUs; 3%) (Table 2). Among LUs, the mean numbers of outpatient and ED visits last year were 3.8 and 1.2, respectively. HUs represented 24% of the sample and had a mean of 15.3 outpatient and 2.3 ED visits last year. SUs (3%) had a mean of 56.7 outpatient and 6.0 ED visits.
Comparing the SUs across the complexity classes shows that those in the MH class had the highest prevalence of depression (95%), bipolar disorder (45%), COPD (25%), cancer (14%), stroke (10%), and high SUD risk (6%) (Figure 2). Among the CC class, SUs had the highest rates of obesity (96%), being current or past smokers (83%), and CVD (48%). Among the LR class, SUs had high rates of being current or past smokers (66%), obesity (62%), depression (52%), and overweight (33%).
Comparing HUs, the MH class had the highest rates of general anxiety (88%) and asthma (30%); the CC class had the highest prevalence of hypertension (89%), liver conditions (15%), and weak/failing kidneys (10%); the LR class had the highest rate of overweight but was not distinguishable by high prevalence of specific chronic or MH conditions (Figure 3). Additional data on LU characteristics in each complexity class are available in eAppendix Figure 2.
Using the underlying LCA data, we displayed the co-occurrence of combinations of most prevalent conditions for individual patients in each group (eAppendix Figure 1). Within the MH SUs, most patients had MH conditions along with hypertension, were obese or overweight, and were current or past smokers. In contrast, within the small CC SU group, most patients had hypertension and CVD, were current or past smokers, and were overweight or obese. Within the small group of LR SUs, most patients were overweight or obese, but few had CCs. Complete prevalence of complex conditions, sociodemographic, financial, and health status indicators among the 9 latent classes are available in eAppendix Table 3.
Using a nationally representative sample of patients who obtain care services at HRSA-funded HCs, we identified 3 complexity and 3 high utilization latent classes among patients aged 18 to 64 years. The complexity classes indicated distinct populations of patients with a high prevalence of MH conditions, multiple CCs, or conditions and behaviors that might increase the risk of more complex conditions. The high utilization classes also identified distinct populations separated by gradation in use of outpatient and ED visits from low to high to super-HUs. The confluence of 3 latent classes identified 9 distinct classes with different profiles of health and utilization and common co-occurrence of specific chronic or MH conditions.
Our classification of complex patients in this study is consistent with other studies that highlight the high rate of co-occurrence of MH conditions, but we also show a distinct separation between those who are primarily defined by high prevalence of MH conditions with some CCs vs those primarily defined by CCs complicated with depression.4,25,34,35 Our findings indicating co-occurrence of MH conditions with smoking status or of hypertension and diabetes with obesity have been shown in the literature, but other relationships such as high prevalence of MH conditions with obesity or hypertension are more emergent.36-39 Our utilization classes found that only 3% are SUs, supporting the existing literature that only a small proportion of the population are SUs.1 Of those characterized as SUs, our findings support previous contributions that the majority are driven by MH conditions, followed by CCs such as cancer or social risk factors.4,7 Our classification of patients with high utilization is also consistent with studies that distinguish SUs as those with many ED visits, but we examined the combined use of outpatient and ED visits and separated SUs from other HUs.4,7,40 Our joint classification of patients by complexity and high utilization is aligned with the concept of high-need, high-cost patients and provides insights into 9 homogenous and distinct groups of HC patients who require targeted interventions designed to address their unique health care needs.2
Our findings have implications based on complexity alone, as well as for complex patients who are HUs or SUs. The MH class is characterized clinically by a high prevalence of MH conditions along with other CVD risk factors including smoking, obesity, and hypertension. These findings suggest that these risk factors cannot be treated effectively without addressing MH issues, and, therefore, collaboration between primary care providers and psychiatrists and other MH specialists is most appropriate to treat this patient population.41 The CC class is also characterized by CVD risk factors including hypertension, obesity, and diabetes, and it has lower prevalence of most MH conditions. These risk factors differ somewhat from those of the MH class. This patient population would likely benefit from appropriate interventions to reduce cardiovascular events, including lifestyle modification and medications to control blood pressure and blood glucose level. The LR class is not dominated by chronic health or MH conditions but likely reflects those with risk of developing more severe and complex health conditions, such as diabetes or hypertension, based on their risk factors including obesity, smoking behavior, and depression. These patients would probably benefit most from intensive patient-centered primary and preventive care to arrest development of more serious conditions.
By focusing on complexity classes and SUs, more refined sets of interventions to avoid excessive service use emerge. SUs with MH complexity may require the presence of MH providers to manage patients’ depression and bipolar disorder; SUD staff in the care team to address high SUD risk; and significant coordination with oncologists, pulmonary specialists, and neurologists.7,42,43 In contrast, reducing excessive use among SUs with CCs may require a greater focus on smoking cessation efforts in the primary care setting, including nutritionists in care teams and coordination with cardiologists.27,28,30,44 Similarly, care of HU patients with CCs may require a greater focus on managing patients with hypertension and coordination with hepatologists and nephrologists to reduce avoidable use. Additionally, care of HU patients with MH complexity may require better management of general anxiety and asthma to reduce avoidable care use. Further research is required to better assess factors associated with high use of care by LR patients.
Our findings highlight the importance of providing effective care to HC patients and the potential complexity of this task, which is complicated by SDOH such as low income, lack of transportation, inability to take time off from jobs that do not offer paid time for medical care, and lack of insurance. The profile of these patients indicates the need for multidimensional approaches including multidisciplinary teams of providers and care coordination with specialists and other providers outside the HC. In addition, treatment of these patients requires multiple care steps and visits. These issues have led to HRSA initiatives to help maximize patients’ access to integrated care and promoting the patient-centered medical home model of care. HRSA has also made significant investments in behavioral health and primary care integration, including establishing a supplemental funding opportunity in 2019 to help increase behavioral health staffing and access in HCs.41 Our data indicate the importance of escalating these and other efforts to improve the health of these patients.
We used survey data for this study, which may be susceptible to measurement error and recall bias. In particular, utilization indicators are likely to be estimates at the higher ranges. The sample sizes of SUs among LR and CC classes were small. We lacked other important indicators of high utilization such as hospitalizations and expenditures. The sample was not large enough to test the fit of additional complexity and high-utilization classes and did not allow us to include other indicators of complexity such as severity. The number of conditions identified in the data were also not exhaustive of all diagnoses for each patient. In addition, our measure of complexity lacked the provider perspective, which would include as assessment of the severity of conditions and other observations of patients’ problems.22,25 We also lacked a more direct measure of actual SUD and relied on risk status. We lacked data on SDOH, receipt of social support services, and ability of patients to manage their condition, which are important indicators for utilization and health status.45 We lacked data on future utilization patterns to assess regression to the mean.
Our findings are primarily generalizable to patients of HCs, who may differ from low-income and uninsured patients receiving their care in other settings such as private providers or hospital-based outpatient care. Our study is applicable to adult patients aged 18 to 64 years and may not be generalizable to older adults, whose condition profiles are frequently dominated by low functional status and frailty, multiple CCs, and cognitive impairment. Older adults also are primarily covered by Medicare, and these differences lead to different patterns of care utilization and access.32,46
Our data provide a deeper understanding of the combined latent concepts of complexity and high utilization for nearly 30 million low-income and uninsured patients receiving care from HRSA-funded HCs.12 Our findings provide insights into factors associated with complexity and high utilization and how the care of these patients may be adequately or effectively managed. Improvement in health of these patients could reduce unnecessary outpatient and ED visits and garner savings at the federal level.
Author Affiliations: University of California, Los Angeles (UCLA) Center for Health Policy Research (NP, XC, CL, WZ), Los Angeles, CA; Department of Health Policy and Management, UCLA Fielding School of Public Health (NP, YT), Los Angeles, CA; Division of General Internal Medicine and Health Services Research, UCLA Geffen School of Medicine (YT), Los Angeles, CA; Office of Quality Improvement, Bureau of Primary Health Care, Health Resources and Services Administration (HH, BH, JB, AS), Rockville, MD.
Source of Funding: This article was funded by the US Department of Health and Human Services (HHS), Health Resources and Services Administration (HRSA), under contract number HHSH250201300023I. The views expressed in this publication are solely the opinions of the authors and do not necessarily reflect the official policies of HHS or HRSA, nor does mention of the department or agency names imply endorsement by the US government.
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 (NP, XC, YT, HH, BH, JB, AS); acquisition of data (XC, CL, WZ, HH); analysis and interpretation of data (NP, XC, YT, WZ, HH, BH, JB, AS); drafting of the manuscript (NP, CL, HH); critical revision of the manuscript for important intellectual content (NP, XC, YT, CL, WZ, HH, BH, AS); statistical analysis (XC, WZ); provision of patients or study materials (WZ); obtaining funding (NP); administrative, technical, or logistic support (CL); and supervision (NP, HH, JB, AS).
Address Correspondence to: Nadereh Pourat, PhD, MSPH, UCLA Center for Health Policy Research, 10960 Wilshire Blvd, Ste 1550, Los Angeles, CA 90024. Email: firstname.lastname@example.org.
1. Das LT, Abramson EL, Kaushal R. High-need, high-cost patients offer solutions for improving their care and reducing costs. NEJM Catalyst. February 5, 2019. Accessed August 21, 2019. https://catalyst.nejm.org/high-need-high-cost-patients-solutions/
2. Blumenthal D, Chernof B, Fulmer T, Lumpkin J, Selberg J. Caring for high-need, high-cost patients—an urgent priority. N Engl J Med. 2016;375(10):909-911. doi:10.1056/NEJMp1608511
3. Hayes SL, Salzberg CA, McCarthy D, et al. High-need, high-cost patients: who are they and how do they use health care? a population-based comparison of demographics, health care use, and expenditures. Issue Brief (Commonw Fund). 2016;26:1-14.
4. Rinehart DJ, Oronce C, Durfee MJ, et al. Identifying subgroups of adult superutilizers in an urban safety-net system using latent class analysis: implications for clinical practice. Med Care. 2016;56(1):e1-e9. doi:10.1097/MLR.0000000000000628
5. Barnett ML, Hsu J, McWilliams JM. Patient characteristics and differences in hospital readmission rates. JAMA Intern Med. 2015;175(11):1803-1812. doi:10.1001/jamainternmed.2015.4660
6. LaCalle E, Rabin E. Frequent users of emergency departments: the myths, the data, and the policy implications. Ann Emerg Med. 2010;56(1):42-48. doi:10.1016/j.annemergmed.2010.01.032
7. 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
8. Manning E, Gagnon M. The complex patient: a concept clarification. Nurs Health Sci. 2017;19(1):13-21. doi:10.1111/nhs.12320
9. Davis AC, Shen E, Shah NR, et al. Segmentation of high-cost adults in an integrated healthcare system based on empirical clustering of acute and chronic conditions. J Gen Intern Med. 2018;33(12):2171-2179. doi:10.1007/s11606-018-4626-0
10. Buja A, Claus M, Perin L, et al. Multimorbidity patterns in high-need, high-cost elderly patients. PLoS One. 2018;13(12):e0208875. doi:10.1371/journal.pone.0208875
11. Whitson HE, Johnson KS, Sloane R, et al. Identifying patterns of multimorbidity in older Americans: application of latent class analysis. J Am Geriatr Soc. 2016;64(8):1668-1673. doi:10.1111/jgs.14201
12. National health center data. Health Resources and Services Administration. Accessed August 22, 2019. https://data.hrsa.gov/tools/data-reporting/program-data/national
13. Rust G, Baltrus P, Ye J, et al. Presence of a community health center and uninsured emergency department visit rates in rural counties. J Rural Health. 2009;25(1):8-16. doi:10.1111/j.1748-0361.2009.00193.x
14. Rothkopf J, Brookler K, Wadhwa S, Sajovetz M. Medicaid patients seen at federally qualified health centers use hospital services less than those seen by private providers. Health Aff (Millwood). 2011;30(7):1335-1342. doi:10.1377/hlthaff.2011.0066
15. Jones E, Lebrun-Harris LA, Sripipatana A, Ngo-Metzger Q. Access to mental health services among patients at health centers and factors associated with unmet needs. J Health Care Poor Underserved. 2014;25(1):425-436. doi:10.1353/hpu.2014.0056
16. Rodriguez HP, Glenn BA, Olmos TT, et al. Real-world implementation and outcomes of health behavior and mental health assessment. J Am Board Fam Med. 2014;27(3):356-366. doi:10.3122/jabfm.2014.03.130264
17. Lebrun-Harris LA, Fiore MC, Tomoyasu N, Ngo-Metzger Q. Cigarette smoking, desire to quit, and tobacco-related counseling among patients at adult health centers. Am J Public Health. 2015;105(1):180-188. doi:10.2105/AJPH.2013.301691
18. Ricketts TC III. Geography and disparities in health care. In: Swift EK, ed. Guidance for the National Healthcare Disparities Report. The National Academies Press; 2002.
19. Khullar D, Chokshi DA. Health, income, & poverty: where we are & what could help. Health Affairs. October 4, 2018. Accessed September 5, 2019. https://www.healthaffairs.org/do/10.1377/hpb20180817.901935/full/
20. Mount JK, Massanari RM, Teachman J. Patient care complexity as perceived by primary care physicians. Fam Syst Health. 2015;33(2):137-145. doi:10.1037/fsh0000122
21. Grembowski D, Schaefer J, Johnson KE, et al; AHRQ MCC Research Network. A conceptual model of the role of complexity in the care of patients with multiple chronic conditions. Med Care. 2014;52(suppl 3):S7-S14. doi:10.1097/MLR.0000000000000045
22. Loeb DF, Binswanger IA, Candrian C, Bayliss EA. Primary care physician insights into a typology of the complex patient in primary care. Ann Fam Med. 2015;13(5):451-455. doi:10.1370/afm.1840
23. Safford MM, Allison JJ, Kiefe CI. Patient complexity: more than comorbidity. the vector model of complexity. J Gen Intern Med. 2007;22(suppl 3):382-390. doi:10.1007/s11606-007-0307-0
24. Maciejewski ML, Powers BJ, Sanders LL, et al. The intersection of patient complexity, prescriber continuity and acute care utilization. J Gen Intern Med. 2014;29(4):594-601. doi:10.1007/s11606-013-2746-0
25. Loeb DF, Bayliss EA, Binswanger IA, Candrian C, deGruy FV. Primary care physician perceptions on caring for complex patients with medical and mental illness. J Gen Intern Med. 2012;27(8):945-952. doi:10.1007/s11606-012-2005-9
26. 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.
27. Jean-Baptiste D, O’Malley AS, Shah T. Population segmentation and targeting of health care resources: findings from a literature review. Mathematica working paper No. 58. December 1, 2017. Accessed August 30, 2019. https://mathematica.org/publications/population-segmentation-and-targeting-of-health-care-resources-findings-from-a-literature-review
28. Rudin RS, Gidengil CA, Predmore Z, Schneider EC, Sorace J, Hornstein R. Identifying and coordinating care for complex patients: findings from the leading edge of analytics and health information technology. Rand Health Q. 2017;6(3):2.
29. Newcomer SR, Steiner JF, Bayliss EA. Identifying subgroups of complex patients with cluster analysis. Am J Manag Care. 2011;17(8):e324-e332.
30. Berry-Millett R, Bodenheimer TS. Care management of patients with complex health care needs. Synth Proj Res Synth Rep. 2009;(19):52372.
31. Porter ME, Pabo EA, Lee TH. Redesigning primary care: a strategic vision to improve value by organizing around patients’ needs. Health Aff (Millwood). 2013;32(3):516-525. doi:10.1377/hlthaff.2012.0961
32. Sun R, Karaca Z, Wong HS. Trends in hospital emergency department visits by age and payer, 2016-2015. Healthcare Cost and Utilization Project statistical brief No. 238. March 2018. Accessed August 22, 2019. https://www.hcup-us.ahrq.gov/reports/statbriefs/sb238-Emergency-Department-Age-Payer-2006-2015.jsp
33. Formann AK, Kohlmann T. Latent class analysis in medical research. Stat Methods Med Res. 1996;5(2):179-211. doi:10.1177/096228029600500205
34. Bower P, Gilbody S, Richards D, Fletcher J, Sutton A. Collaborative care for depression in primary care. making sense of a complex intervention: systematic review and meta-regression. Br J Psychiatry. 2006;189:484-493. doi:10.1192/bjp.bp.106.023655
35. Unützer J, Park M. Strategies to improve the management of depression in primary care. Prim Care. 2012;39(2):415-431. doi:10.1016/j.pop.2012.03.010
36. Tobacco use and quitting among individuals with behavioral health conditions. CDC. Accessed September 5, 2019. https://www.cdc.gov/tobacco/disparities/mental-illness-substance-use/index.htm
37. de Boer IH, Bangalore S, Benetos A, et al. Diabetes and hypertension: a position statement by the American Diabetes Association. Diabetes Care. 2017;40(9):1273-1284. doi:10.2337/dci17-0026
38. Rozario SS, Masho SW. The associations between mental health status, hypertension, and hospital inpatient visits in women in the United States. Am J Hypertens. 2018;31(7):804-810. doi:10.1093/ajh/hpy065
39. Allison DB, Newcomer JW, Dunn AL, et al. Obesity among those with mental disorders: a National Institute of Mental Health meeting report. Am J Prev Med. 2009;36(4):341-350. doi:10.1016/j.amepre.2008.11.020
40. Gawande A. The hot spotters. The New Yorker. January 16, 2011. Accessed August 30, 2019. https://www.newyorker.com/magazine/2011/01/24/the-hot-spotters
41. FY 2019 integrated behavioral health services (IBHS) supplemental funding (HRSA-19-100). Health Resources and Services Administration. Accessed December 10, 2019. https://bphc.hrsa.gov/program-opportunities/funding-opportunities/behavioral-health
42. Vaida B. For super-utilizers, integrated care offers a new path. Health Aff (Millwood). 2017;36(3):394-397. doi:10.1377/hlthaff.2017.0112
43. Lynn J, Straube BM, Bell KM, Jencks SF, Kambic RT. Using population segmentation to provide better health care for all: the “Bridges to Health” model. Milbank Q. 2007;85(2):185-208, 209-212. doi:10.1111/j.1468-0009.2007.00483.x
44. Berry LL, Rock BL, Smith Houskamp B, Brueggeman J, Tucker L. Care coordination for patients with complex health profiles in inpatient and outpatient settings. Mayo Clin Proc. 2013;88(2):184-194. doi:10.1016/j.mayocp.2012.10.016
45. Gallagher NA, Fox D, Dawson C, Williams BC. Improving care transitions: complex high-utilizing patient experiences guide reform. Am J Manag Care. 2017;23(10):e347-e352.
46. Aldridge MD, Kelley AS. Appendix E: epidemiology of serious illness and high utilization of health care. In: Institute of Medicine. Dying in America: Improving Quality and Honoring Individual Preferences Near the End of Life. The National Academies Press; 2015:487-532.