Medication Nonadherence, Mental Health, Opioid Use, and Inpatient and Emergency Department Use in Super-Utilizers

Nonadherence to essential chronic medications and mental health diagnosis were associated with higher hospitalizations and emergency department use among Medicare super-utilizers.

ABSTRACT

Objectives: To examine whether mental health conditions, opioid use, and medication nonadherence are associated with inpatient and emergency department (ED) use among Medicare super-utilizers from medically underserved areas.

Study Design: Retrospective panel study.

Methods: The study included Medicare super-utilizers (≥3 hospitalizations or ≥2 hospitalizations with ≥2 ED visits in 6 months) served by a health system in a medically underserved area in the South from February 2013 to December 2014 with at least 1 filled prescription for hypertension, type 2 diabetes, cardiovascular, and/or chronic obstructive pulmonary disease/asthma medications. We used random effects negative binomial models to assess whether mental health diagnosis, opioid use, and medication nonadherence were associated with preventable and overall hospitalizations and ED visits stratified by age (18-64 vs ≥65 years).

Results: Overall chronic disease medication nonadherence was associated with more frequent hospitalizations and ED visits for both younger (hospitalizations: incidence rate ratio [IRR], 1.31; 95% CI, 1.16-1.47; ED visits: IRR, 1.33; 95% CI, 1.14-1.55) and older (hospitalizations: IRR, 1.34; 95% CI, 1.20-1.49; ED visits: IRR, 1.18; 95% CI, 1.02-1.38) beneficiaries. Mental health diagnosis was significantly associated with higher hospitalizations and ED visits among both age groups. Although associations between opioid medication use and inpatient and ED use were inconsistent and not significant in most cases, we found that 7 or more days’ supply of opioids was associated with lower preventable hospitalizations in Medicare beneficiaries 65 years or older.

Conclusions: The study findings highlight the importance of improving medication adherence and addressing behavioral health needs in Medicare super-utilizers.

Am J Manag Care. 2020;26(3):e98-e103. https://doi.org/10.37765/ajmc.2020.42642

Takeaway Points

Nonadherence to essential chronic medications and mental health diagnosis were associated with higher hospitalizations and emergency department use among Medicare super-utilizers.

  • This study builds on previous research conducted among Medicare patients by focusing on a more vulnerable Medicare population from a medically underserved metropolitan area in the South with patients who are predominantly African American, younger, disabled, and super-utilizing.
  • This study has important implications for policy makers and institutions involved in improving the quality of care and reducing costs for super-utilizers.
  • The study findings highlight the importance of improving medication adherence and addressing behavioral health needs in Medicare super-utilizers.

Optimal adherence to essential prescription medications is critical for the treatment of chronic conditions.1-3 Medication nonadherence is a major barrier to achieving treatment goals, especially among high-risk patient populations.4 These include so-called super-utilizers—patients with disproportionately high inpatient and emergency department (ED) use—who generally have multiple chronic conditions and use multiple concurrent medications to treat these conditions.5-8 The cost of caring for super-utilizers is high, with estimates that they represent only 3% to 5% of the US population but account for 30% to 50% of total spending.5,9

Findings of previous studies conducted among Medicare beneficiaries have shown that optimal medication adherence is associated with lower inpatient and ED use.10,11 Evidence shows that depression is associated with higher incidence of hospitalizations among patients who are high utilizers of medical care, and our previous studies have highlighted high rates of chronic pain and opioid use in this vulnerable population.12-14 However, the impact of these factors on healthcare utilization among medically underserved communities is not well understood. This study builds on previous research by focusing on more vulnerable Medicare beneficiaries from a medically underserved metropolitan area in the South who are predominantly African American, younger, and disabled.13

This study was conducted using baseline data from the SafeMed Program, a care transitions program focusing on medication management and funded by the CMS Health Care Innovation Awards (HCIA).6 SafeMed targeted super-utilizers from medically underserved areas (MUAs) with ambulatory care—sensitive chronic conditions (hypertension, type 2 diabetes, congestive heart failure [CHF], coronary artery disease [CAD], chronic obstructive pulmonary disease [COPD], or asthma) for which outpatient care improvements can reduce inpatient utilization.15-18 Using longitudinal 2-year Medicare data for the period prior to the SafeMed intervention, we examined whether medication nonadherence, mental health diagnosis, and opioid medication use were associated with inpatient and ED use in super-utilizers. We also examined whether tobacco use disorder, polypharmacy, number of prescribers, and access to outpatient care were associated with hospitalizations and ED visits in this vulnerable population.

METHODS

Design and Setting

This study was a retrospective panel analysis of the baseline data for Medicare beneficiaries from a MUA served by a hospital system in Memphis, Tennessee, and meeting the SafeMed Program eligibility criteria during the enrollment period from February 2013 to December 2014.6 For each patient, we reviewed 2 years of data prior to SafeMed enrollment. The dates of this 2-year baseline period varied for each patient based on their enrollment date for the SafeMed Program. For instance, if a patient was enrolled in the SafeMed Program in February 2013, the 2-year evaluation period was from February 2011 to February 2013. The data for each patient were divided into four 6-month patient-periods, with the last 6-month period serving as the qualifying period for the SafeMed Program.6 The panel design enabled us to examine changes in hospitalizations, ED visits, and nonadherence over time. eAppendix Figure 1 (eAppendix available at ajmc.com) shows the study design.

Data

The study used Chronic Conditions Data Warehouse files including pharmacy and medical claims for all fee-for-service Medicare beneficiaries with parts A, B, and D coverage. Data on hospitalizations were obtained using Medicare Provider Analysis and Review files. Data on observation stays and ED visits were obtained using Medicare outpatient revenue center files. Part D drug event files were used to assess medication nonadherence, and diagnosis codes present in the inpatient, outpatient, or Part B claims were used to assess chronic conditions. Medicare beneficiary summary files (parts A, B, and D) were used to assess demographic factors. Patient identifier and claim identifier were used as linking variables.

Study Population

The final study sample (N = 1092) included adult super-utilizers (≥3 hospitalizations or ≥2 hospitalizations with ≥2 ED visits in 6 months) with a diagnosis of at least 1 of the previously mentioned ambulatory care—sensitive conditions, with continuous eligibility for Medicare parts A, B, and D and who had filled at least 1 of 17 type 2 diabetes, cardiovascular, or COPD/asthma drug classes during the 2-year period. The list of therapy classes is presented in eAppendix Table 1. The study sample was divided into 2 groups: (1) Medicare beneficiaries aged 18 to 64 years who are eligible due to their disability19 and (2) Medicare beneficiaries 65 years or older. The study excluded patients with insulin fills because adherence cannot be reliably calculated using claims data.20

Measures

The outcomes included overall and preventable inpatient stays and ED visits, measured for each 6-month period. Hospitalizations and ED visits for hypertension, type 2 diabetes, CHF, CAD, COPD, or asthma were considered preventable based on the Agency for Healthcare Research and Quality (AHRQ) Prevention Quality Indicators chronic composite measures.21 Preventable hospitalizations and ED visits were identified using the primary diagnosis documented during an encounter. Medication nonadherence was measured as interval-based average proportion of days covered (PDC).22,23 Using the PDC approach,22 patients were followed continuously from the first 6-month period in which they had 2 or more fills for a drug class through the fourth 6-month period. The numerator was number of days of medication supply, and the denominator was number of days during the interval from the index date to the last day of that period. We used the first day of medication fill as the index date only if patients had 2 or more fills for a drug class for the first time in a period. If patients filled 2 or more medications for a drug class in any of the previous periods, the denominator was 180 days. For the numerator, if the days’ supply extended into the next period, it contributed to both periods. We did not double-count the days when patients refilled their medications before the previous prescription ran out. We subtracted the inpatient days from the denominator. We calculated an overall average PDC at the patient level by adding the PDC for each drug class and then dividing by the number of drug classes for each patient. Nonadherence was defined as PDC less than 80%. Nonadherence was also measured by disease category for type 2 diabetes, cardiovascular, and COPD/asthma drug classes.

We identified patients with mental health diagnosis if they had depression/anxiety (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 296.2, 296.3, 298.0, 300, 309.1, or 311), bipolar disorder, schizophrenia, or other psychotic disorders (ICD-9-CM code 295, 297, 298, 293.81, 293.82, 296.0, 296.1, 296.4, 296.5, 296.6, 296.7, 296.8, or 296.9) in the current period or any prior period. Opioid medication use was defined by whether or not patients had 7 or more days’ supply of opioids in each 6-month period.

Other Factors

Baseline factors included age, gender, race, dual Medicare-Medicaid eligibility, and low-income cost-sharing subsidy (LICS) receipt. Because the income threshold for receiving a LICS is higher than the threshold for Tennessee Medicaid, we created a variable with 3 groups: non—dual-eligible, non-LICS; non–dual-eligible, LICS; and dual-eligible, LICS. Other factors included tobacco use disorder, cancer, Charlson Comorbidity Index score,24 1 or more physician office visits, unique medications filled, and unique prescribers.

Statistical Analysis

For multivariate analyses, observations from the four 6-month periods were pooled using patient-period as the unit of analysis. Random effects negative binomial models were used as multivariate models. Along with medication adherence and opioid use (time-varying factors), we were interested in mental health diagnosis that did not vary much with time. Therefore, we used random effects models. Additionally, negative binomial models were used, given that the utilization outcomes were overdispersed count variables.25 Separate models for beneficiaries aged 18 to 64 years and those 65 or older were estimated to understand associations specific to these Medicare groups. Lagged effects of all independent factors were included in the models to account for potential endogeneity. Thus, mental health, opioid use, medication nonadherence, and all other covariates were measured in a period before the outcomes were measured. Multivariate analyses were conducted to examine the association between nonadherence and inpatient and ED utilization by disease category.

Sensitivity Analysis

Realizing that medication switching between drug classes may be common after hospitalizations, we measured adherence using an alternative method. Patients were included in the 6-month period if they had (1) at least 2 fills for a drug class for the first time in the period or (2) at least 1 fill in the period if they previously had at least 2 fills for that drug class.

RESULTS

Table 1 shows that among 1092 study subjects, nonelderly disabled beneficiaries were more likely to be African Americans, be low-income dual-eligible, and have higher rates of mental health diagnosis compared with older Medicare beneficiaries (P <.05 for all). Among both age groups, the most common primary discharge diagnoses were for COPD, CHF, chronic kidney disease (CKD), pneumonia, and hypertensive CKD (data not shown). Opioid fill rates were high in all 4 study periods but were highest in the fourth 6-month period when all patients met the super-utilizer definition, with 64.1% of elderly and 75.8% of nonelderly Medicare patients filling 1 or more opioid medications. Additionally, overall nonadherence rates increased from the first period to the fourth period (eAppendix Table 2).

Table 2 shows the multivariate associations among elderly Medicare beneficiaries. Nonadherent patients had significantly higher overall and preventable inpatient and ED visits compared with adherent patients. Mental health diagnosis was significantly associated with higher overall inpatient and overall and preventable ED visits, whereas 7 or more days’ supply of opioids was significantly associated with lower preventable hospitalizations.

Table 3 displays the same results among nonelderly Medicare beneficiaries. As observed with older beneficiaries, nonadherence and mental health diagnosis were also associated with significantly higher hospitalizations and ED visits in this younger age group. Additionally, tobacco use disorder was significantly associated with higher hospitalizations and ED visits.

eAppendix Table 3 shows that inpatient and ED visits were significantly higher for nonadherent patients compared with adherent patients for all disease categories. In sensitivity analyses using the alternative PDC definition, we found similar patterns of nonadherence over time (eAppendix Figure 2 [A and B]). Additionally, multivariate results were similar in terms of significance and directionality to our main analyses (eAppendix Table 4; eAppendix Table 5).

DISCUSSION

In our retrospective study of super-utilizing Medicare beneficiaries, we found higher hospitalizations and ED visits among nonadherent patients. Similar associations were found across the 3 disease categories and in both age groups. These findings are consistent with those of a previous study that found that Medicare patients with COPD who were adherent to their chronic medications experienced fewer hospitalizations.11 Our work extends these findings to super-utilizing patients who were more disadvantaged than average elderly Medicare patients: More were dual-eligible, disabled, with multiple comorbidities, and nonadherent to their essential chronic disease medications.13

This analysis reinforced previous findings that super-utilization is dynamic and changes with time.5,14 Less than 20% of nonelderly and less than 10% of elderly Medicare beneficiaries were super-utilizers in the first three 6-month periods (eAppendix Table 2). The current study also demonstrates that nonadherence worsens over time as patients approach super-utilizer status (eAppendix Figure 2 [A and B]). This suggests that it may be helpful to identify and address medication nonadherence earlier among patients with ambulatory care—sensitive conditions who are at risk of becoming super-utilizers.

Similar to previous studies,5,26 we found high rates of mental health conditions and tobacco use disorders, especially among nonelderly Medicare patients. Mental health diagnosis was significantly associated with high inpatient and ED utilization among both age groups, whereas tobacco use disorder was a significant risk factor for high inpatient and ED utilization among nonelderly beneficiaries. These findings suggest that Medicare super-utilizers may benefit from psychosocial interventions that provide assistance with mental health, tobacco cessation, and social support.

Results of this study show high opioid use in these super-utilizers. A substantial amount of opioid utilization is likely driven by the high prevalence of chronic pain in super-utilizers, as documented in a previous study.14 Efforts to address opioid use among super-utilizers may include alternative strategies for managing chronic pain. Although associations between opioid medication use and inpatient and ED use were inconsistent and not significant in most cases, we found that receipt of 7 or more days’ supply of opioids was associated with lower preventable hospitalizations in elderly Medicare beneficiaries. This finding suggests that adequately treated pain may lead to lower hospitalizations in elderly Medicare patients. Policies that encourage limiting use of opioid medications to short courses of fewer than 7 days may be beneficial if complemented with proven alternative methods of chronic pain management.

There were weak but significant associations between outcomes and other factors including physician office visits, polypharmacy, and number of prescribers. Although having 1 or more physician office visits was associated with lower hospitalizations, it was significantly associated with higher ED visits among elderly Medicare beneficiaries. Polypharmacy was protective, whereas having a higher number of prescribers was associated with higher utilization among nonelderly Medicare beneficiaries. These findings should be interpreted with caution given the weak associations for these factors.

This study has important implications for policy makers and institutions involved in improving care and reducing healthcare costs for super-utilizers. Our study findings suggest that early medication adherence support is needed to address medication nonadherence. Hospitals and payers may evaluate programs aimed at improving medication adherence among this population. Furthermore, our findings of an association between mental health diagnosis and utilization highlight the importance of mental health care and social support. Given the complex medical and social needs of Medicare super-utilizers, future studies should examine programs that address these barriers. Moreover, disabled nonelderly Medicare patients, who have higher rates of comorbidities, mental illness, tobacco use disorder, and opioid use compared with elderly Medicare beneficiaries, merit particular targeting and social support.

Limitations

The PDC measure is an indirect method that may not accurately capture medication use and is not accurate for measuring insulin adherence. Excluding patients on insulin limits the generalizability to insulin users. The study did not account for medication switching between therapy classes. However, examining medication adherence using an alternate method that did not follow patients continuously over time allowed for checking the reliability of the main adherence measure. Additionally, pharmacy claims data capture only medication fills and not prescriptions written by the provider. Evidence suggests that patients receiving 90-day supplies and prescriptions through mail-order pharmacy have higher PDC.27 Although we did not have information on mail-order pharmacy in the prescription drug event file, we found that only about 10% of drug claims were 90-day fills. The variable Pharmacy Service Type Code indicating the type of pharmacy that dispensed the prescription was not available. Because mental health conditions are often underdiagnosed,28,29 our definition based on diagnosis might not reflect the true prevalence in super-utilizers. The study findings may be generalizable only to similar populations and settings across the country because they are based on super-utilizers living in Memphis, Tennessee. Additionally, because this study included fee-for-service Medicare beneficiaries, the results may not be generalizable to beneficiaries on Medicare Advantage plans. Given their high inpatient and ED utilization and high rates of comorbidities, our patient population was much sicker than beneficiaries on Medicare Advantage plans. Although we adjusted for comorbidities and other factors that affect utilization, other unobserved factors that were not available in claims data may not have been accounted for in this study. Because we used data from the CMS/HCIA-funded SafeMed Program, we had access to data for only super-utilizing patients who were potentially eligible for this program. This may limit the generalizability of this study. Finally, although we defined hospitalizations and ED visits for ambulatory care—sensitive conditions as preventable using the AHRQ definition, not all hospitalizations for these conditions are preventable.

CONCLUSIONS

Overall medication nonadherence was associated with more hospitalizations and ED use among Medicare super-utilizers, who are patients with disproportionately high inpatient and ED use. We found similar associations for type 2 diabetes, cardiovascular, and COPD/asthma disease categories. Mental health diagnosis was associated with higher inpatient and ED use among both age groups. The associations between opioid medication use and inpatient and ED use were inconsistent and not significant in most cases. The exception was 7 or more days’ supply of opioid medications, which was significantly associated with lower preventable hospitalizations among elderly Medicare super-utilizers. The study findings suggest the importance of early monitoring and addressing of medication nonadherence and behavioral health needs among Medicare super-utilizers. More research is needed to examine programs that target medication nonadherence and behavioral health among Medicare super-utilizers.Author Affiliations: Center for Health System Improvement (SS, JEB), Department of Preventive Medicine (JYW, JEB), and Department of Medicine (JEB), College of Medicine, University of Tennessee Health Science Center, Memphis, TN; Department of Health Policy & Management, Rollins School of Public Health, Emory University (IG), Atlanta, GA; Department of Clinical Pharmacy and Translational Science, College of Pharmacy, University of Tennessee Health Science Center (JG), Nashville, TN.

Source of Funding: The project was supported by Funding Opportunity Number 1C1CMS331067-01-00 from CMS Center for Medicare and Medicaid Innovation. Contents of this study are solely the responsibility of the authors. This project was also supported by the Pharmaceutical Research and Manufacturers of America Foundation (PhRMA). The content is solely the responsibility of the authors and does not necessarily represent the official views of PhRMA.

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 (SS, JYW, JEB); acquisition of data (SS, JEB); analysis and interpretation of data (SS, IG, JYW, JG, JEB); drafting of the manuscript (SS, JG, JEB); critical revision of the manuscript for important intellectual content (SS, IG, JG, JEB); statistical analysis (SS, JYW); obtaining funding (SS, JG, JEB); administrative, technical, or logistic support (JEB); and supervision (IG, JG, JEB).

Address Correspondence to: Satya Surbhi, PhD, University of Tennessee Health Science Center, 956 Court Ave, Coleman D224-A, Memphis, TN 38163. Email: ssurbhi@uthsc.edu.REFERENCES

1. Ho PM, Bryson CL, Rumsfeld JS. Medication adherence: its importance in cardiovascular outcomes. Circulation. 2009;119(23):3028-3035. doi: 10.1161/CIRCULATIONAHA.108.768986.

2. Brown MT, Bussell JK. Medication adherence: WHO cares? Mayo Clin Proc. 2011;86(4):304-314. doi: 10.4065/mcp.2010.0575.

3. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med. 2005;353(5):487-497. doi: 10.1056/NEJMra050100.

4. Williams A, Manias E, Walker R. Interventions to improve medication adherence in people with multiple chronic conditions: a systematic review. J Adv Nurs. 2008;63(2):132-143. doi: 10.1111/j.1365-2648.2008.04656.x.

5. 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.

6. Surbhi S, Munshi KD, Bell PC, Bailey JE. Drug therapy problems and medication discrepancies during care transitions in super-utilizers. J Am Pharm Assoc (2003). 2016;56(6):633-642.e1. doi: 10.1016/j.japh.2016.07.004.

7. Pines JM, Asplin BR, Kaji AH, et al. Frequent users of emergency department services: gaps in knowledge and a proposed research agenda. Acad Emerg Med. 2011;18(6):e64-e69. doi: 10.1111/j.1553-2712.2011.01086.x.

8. Kanzaria HK, Niedzwiecki MJ, Montoy JC, Raven MC, Hsia RY. Persistent frequent emergency department use: core group exhibits extreme levels of use for more than a decade. Health Aff (Millwood). 2017;36(10):1720-1728. doi: 10.1377/hlthaff.2017.0658.

9. Riley GF. Long-term trends in the concentration of Medicare spending. Health Aff (Millwood). 2007;26(3):808-816. doi: 10.1377/hlthaff.26.3.808.

10. Jha AK, Aubert RE, Yao J, Teagarden JR, Epstein RS. Greater adherence to diabetes drugs is linked to less hospital use and could save nearly $5 billion annually. Health Aff (Millwood). 2012;31(8):1836-1846. doi: 10.1377/hlthaff.2011.1198.

11. Simoni-Wastila L, Wei YJ, Qian J, et al. Association of chronic obstructive pulmonary disease maintenance medication adherence with all-cause hospitalization and spending in a Medicare population. Am J Geriatr Pharmacother. 2012;10(3):201-210. doi: 10.1016/j.amjopharm.2012.04.002.

12. Pearson SD, Katzelnick DJ, Simon GE, Manning WG, Helstad CP, Henk HJ. Depression among high utilizers of medical care. J Gen Intern Med. 1999;14(8):461-468. doi: 10.1046/j.1525-1497.1999.06278.x.

13. Surbhi S, Graetz I, Wan JY, Gatwood J, Bailey JE. The effect of opioid use and mental illness on chronic disease medication adherence in superutilizers. J Manag Care Spec Pharm. 2018;24(3):198-207. doi: 10.18553/jmcp.2018.24.3.198.

14. Harris LJ, Graetz I, Podila PS, 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.

15. Bindman AB, Chattopadhyay A, Osmond DH, Huen W, Bacchetti P. The impact of Medicaid managed care on hospitalizations for ambulatory care sensitive conditions. Health Serv Res. 2005;40(1):19-38. doi: 10.1111/j.1475-6773.2005.00340.x.

16. Wolff JL, Starfield B, Anderson G. Prevalence, expenditures, and complications of multiple chronic conditions in the elderly. Arch Intern Med. 2002;162(20):2269-2276. doi: 10.1001/archinte.162.20.2269.

17. Bindman AB, Chattopadhyay A, Auerback GM. Interruptions in Medicaid coverage and risk for hospitalization for ambulatory care-sensitive conditions. Ann Intern Med. 2008;149(12):854-860. doi: 10.7326/0003-4819-149-12-200812160-00004.

18. Parekh AK, Goodman RA, Gordon C, Koh HK; HHS Interagency Workgroup on Multiple Chronic Conditions. Managing multiple chronic conditions: a strategic framework for improving health outcomes and quality of life. Public Health Rep. 2011;126(4):460-471. doi: 10.1177/003335491112600403.

19. Medicare for people under 65. Center for Medicare Advocacy website. medicareadvocacy.org/under-65-project. Accessed May 8, 2019.

20. Frequently asked questions. Pharmacy Quality Alliance website. pqaalliance.org/measure-use-faqs. Accessed February 12, 2020.

21. Prevention Quality chronic composite technical specifications: Prevention Quality Indicators #92 (PQI #92). Agency for Health Research and Quality website. qualityindicators.ahrq.gov/Downloads/Modules/PQI/V45/TechSpecs/PQI%2092%20Prevention%20Quality%20Chronic%20Composite.pdf. Accessed February 12, 2020.

22. Choudhry NK, Shrank WH, Levin RL, et al. Measuring concurrent adherence to multiple related medications. Am J Manag Care. 2009;15(7):457-464.

23. Chu LH, Kawatkar A, Gu A. A SAS macro program to calculate medication adherence rate for single and multiple medication use. Lex Jansen SAS Conference Proceedings website. lexjansen.com/wuss/2011/hoc/Papers_Chu_L_74886.pdf. Accessed February 12, 2020.

24. Concept: Charlson Comorbidity Index. University of Manitoba Centre for Health Policy website. mchp-appserv.cpe.umanitoba.ca/viewConcept.php?conceptID=1098. Updated January 17, 2019. Accessed May 7, 2019.

25. Negative binomial regression: STATA data analysis examples. UCLA Institute for Digital Research & Education Statistical Consulting website. stats.idre.ucla.edu/stata/dae/negative-binomial-regression. Accessed May 7, 2019.

26. Jiang HJ, Weiss AJ, Barrett ML, Sheng M. Characteristics of hospital stays for super-utilizers by payer, 2012. Healthcare Cost and Utilization Project website. hcup-us.ahrq.gov/reports/statbriefs/sb190-Hospital-Stays-Super-Utilizers-Payer-2012.jsp. Published May 2015. Accessed February 12, 2020.

27. Schmittdiel JA, Nichols GA, Dyer W, Steiner JF, Karter AJ, Raebel MA. Health care system-level factors associated with performance on Medicare STAR adherence metrics in a large, integrated delivery system. Med Care. 2015;53(4):332-337. doi: 10.1097/MLR.0000000000000328.

28. Allan CE, Valkanova V, Ebmeier KP. Depression in older people is underdiagnosed. Practitioner. 2014;258(1771):19-22, 2-3.

29. Kasper S. Anxiety disorders: under-diagnosed and insufficiently treated. Int J Psychiatry Clin Pract. 2006;10(suppl 1):3-9. doi: 10.1080/13651500600552297.