Jonathan P. Weiner, DrPH

 To assess the association of geriatric syndrome risk factors with postacute utilization among hospitalized Medicare patients (both Medicare Advantage [MA] and fee-for-service [FFS] cohorts) and to examine patterns of postacute care for MA and FFS cohorts with high geriatric syndrome risk.

Secondary data analysis using encounter-level data from the State Inpatient Databases (SID) of the Healthcare Cost and Utilization Project.

 The sample included 3.1 million Medicare hospitalizations from the Florida SID (2010 to 2014). We used multivariate linear regression to examine the impact of a geriatric syndrome risk measure, assessed as high risk, moderate risk, or nonrisk, on outcomes in MA and FFS cohorts. Outcome measures included postacute destination and inpatient utilization. We then examined if this risk measure was associated with differences in outcomes between MA and FFS cohorts.

Patients with high geriatric syndrome risk (in both MA and FFS cohorts) are less likely to be discharged to home or to home health care. They also have longer inpatient lengths of stay and higher inpatient costs. This risk measure also explains differences in postacute skilled nursing destination between MA and FFS cohorts.

 Geriatric syndrome risk factors not only play a role in postacute care and inpatient utilization in MA and FFS cohorts but also explain different utilizations between MA and FFS cohorts. This study’s results can be applied to guide discharge planning among a group of high-risk patients and evaluate alternative delivery models for this high-cost, high-need cohort.

Geriatric syndrome risk factors act as important predictors of health care utilization in Medicare patients.

Policy attention is needed to appropriately account for risk factors prevalent in aging older adults when evaluating delivery system models.

Hospitals and postacute facilities can utilize existing administrative data to effectively target high-risk patients.

Continued efforts to monitor how capitated and noncapitated systems differ in their care delivery of postacute services are needed.

Medicare postacute spending has become a key driver of geographic variation in per capita fee-for-service (FFS) Medicare spending, reaching $60 billion in 2015.

 Postacute care offers important rehabilitation and recuperation services for patients after an acute hospital stay. These services include care in home health and skilled nursing, inpatient rehabilitation, and long-term care facilities. Recent efforts have aimed to reduce spending and increase value in postacute settings (ie, skilled nursing value-based purchasing and bundled payments for care improvement initiatives). In the same time period of rising postacute care use, Medicare Advantage (MA), an alternative to FFS Medicare, has become increasingly important, as nearly one-third of Medicare beneficiaries are enrolled in MA. Capitated MA plans may benefit financially by carefully selecting and coordinating postdischarge care compared with payers who are not at financial risk.

 Despite health care services becoming increasingly bundled and paid according to the quality of care, it is unclear how these types of capitated systems best serve their high-risk patients in postacute services compared with FFS systems.

Studies have examined differential postacute use between MA and FFS patients, finding that MA patients utilize less postacute care and lower intensity of these services.

 found that MA patients are less cognitively impaired upon postacute admission than FFS patients and that after adjusting for ability to perform activities of daily living, pain, body mass index, and cognitive scores from the Minimum Data Set, MA patients are more likely to be discharged to the community successfully and have shorter skilled nursing facility (SNF) stays. Waxman et al

 found that MA beneficiaries use less home health overall than FFS beneficiaries, after adjusting for patient demographics and socioeconomic factors. Huckfeldt et al

 examined cohorts of patients with joint replacement, stroke, and heart failure and found that MA patients are less likely to be admitted to inpatient rehabilitation facilities and have shorter lengths of stay in SNFs than FFS patients, while finding no large differences in observable characteristics, such as comorbidities measured by the Elixhauser Index, between the 2 cohorts.

Of particular policy and clinical interest within the cohort of postacute care users are the frail elderly and high-cost and high-need populations with disabilities and multimorbid conditions.

 Fifty-one percent of patients discharged to SNFs, rehabilitation facilities, and long-term care facilities are 80 years and older, and 40% of Medicare hospitalizations in this age group end with a stay in these facilities.

 Much of the care delivered to this cohort of medically complex elderly patients is inadequate, particularly during care transitions.

 For example, care processes can break down during handoffs due to lack of reconciliation of medication regimens, preparation of the patient and caregiver, and completion of follow-up care.

 Recent research suggests that functional status and frailty factors outperform comorbidity in predicting outcomes of acute care.

 Studies that examine the role of functional and cognitive indicators in FFS Medicare use assessment-based data and find that these factors are prevalent and lead to higher postacute utilization and cost that cuts across clinical conditions.

 However, few studies examine the role of these frailty and other multimorbid risk factors captured in administrative claims data on health care utilization between a capitated system such as MA and an FFS system.

The motivations of this study are 2-fold. First, rather than using assessment data that are time and resource intensive to obtain, could risk identification of frailty, functional, and cognitive factors—collectively known as geriatric syndrome

—using existing administrative data sources predict postacute utilization in MA and FFS cohorts beyond what is explained by traditional measures of comorbidity? The study examines this question by using a previously validated geriatric syndrome risk measure captured in administrative data in predicting postacute utilization in hospitalized MA and FFS Medicare cohorts.

 The second is motivated by the question of whether capitated systems, such as MA, provide for a population with geriatric characteristics differently compared with noncapitated and non–risk-bearing systems, such as FFS. The study examines this by assessing whether these geriatric syndrome characteristics could explain some of the differences previously seen in postacute use between MA and FFS hospitalized cohorts.

The Healthcare Cost and Utilization Project (HCUP) State Inpatient Databases (SID) include all inpatient discharge abstracts across payers, including information on type of insurance payer and discharge destination. This study focuses on Florida hospitals from 2010 to 2014. The sample contains encounters of Medicare beneficiaries who had an inpatient hospital stay and who were subsequently discharged to home or to a postacute facility, defined as a discharge to skilled nursing, home health, inpatient rehabilitation, or a long-term care hospital. Encounters with other discharges were excluded (ie, left against medical advice, died, discharged to hospice care), as were encounters of patients who did not reside in Florida or were younger than 65 years and encounters from hospitals with fewer than 10 Medicare postacute discharges each year. Encounters in which the patient originated from a nursing home, was hospitalized, and was subsequently discharged to a nursing home could not be separated and excluded from the data. Encounters were determined to be covered by FFS Medicare or MA by primary payer type for the inpatient stay. In the Florida data, more than 45% of all inpatient encounters were attributable to Medicare patients, and 31% of these Medicare encounters were attributable to MA. Of Medicare encounters, 89% had discharges to home and postacute facilities, with hospice care as the next most prevalent discharge destination (4%) and inpatient death as the next most prevalent discharge status (3%). This study was approved by the institutional review board of Johns Hopkins University (#00008241).

 The geriatric syndrome risk measure was constructed from diagnosis codes using the 

International Classification of Diseases, Ninth Revision, Clinical Modification

 from the inpatient hospitalization and is an extension of the “frailty” marker in the Johns Hopkins University Adjusted Clinical Groups (ACG) System, a case-mix system based on diagnoses assigned by providers.

 The original measure was developed in the Medicare cohort to predict health care utilization using factors pertinent to activities of daily living and has been since adapted by clinicians and geriatricians to capture concepts prevalent in frail older adults and in adults with disabilities or other multimorbid and geriatric conditions.

 It is a modified version of capturing frailty in administrative claims data sets.

 It was not our goal to develop or validate a geriatric syndrome risk measure on a patient population using postacute care. Instead, it was our intent to explore the availability of information in administrative claims data in describing a wide range of geriatric risk factors relevant to older patients in the MA and FFS patient populations and to assess its potential predictive value in health care utilization between the 2 cohorts. We do so using a previously validated geriatric syndrome risk measure that was developed—based on the literature, past claims-based risk adjustment research, and input from geriatricians—to capture factors relevant for population health management and clinical care for an older population.

The geriatric syndrome risk measure includes the presence of particular cognitive and functional conditions: vision problems, urinary and fecal incontinence, difficulty walking, history of falls, pressure ulcers, dementia/cognitive impairment, lack of social support, weight loss, and malnutrition. A higher risk measure has been shown to predict higher health care utilization and outcomes beyond traditional comorbidity measures in cohorts of older adults,

 and when compared against a validated survey-based frailty measure on health utilization and functional outcomes, this measure performed as well if not better in performance metrics in a cohort of FFS Medicare patients.

 Further details of the definition of each condition, specific diagnosis codes, and its performance against a validated frailty measure can be found in prior work.

 Examples of diagnoses in each condition are included in 

We categorized the MA and FFS cohort into 3 categories based on the summation of the conditions and from previous use of the measure: nonrisk (meeting 0 conditions), moderate risk (meeting 1 condition), and high risk (meeting ≥ 2 conditions).

 The most prevalent conditions were dementia (10.6% of the sample) and falls (10.1%), with 1% of encounters having 3 or more conditions. A more detailed distribution of these factors in the current sample can be found in 

 Encounter-level patient demographic adjusters include age, race, and sex, derived from the HCUP hospital discharge file. Because we were interested in the association of geriatric syndrome risk factors and outcomes beyond a traditional comorbidity measure, we also adjusted for the Charlson Comorbidity Index (CCI) score, constructed from diagnosis codes of the inpatient hospitalization.

 The CCI was used because of its feasibility using only inpatient claims codes, compared with other comorbidity measures that utilize both outpatient and inpatient information.

Finally, socioeconomic factors from the Area Health Resources Files and the American Community Survey were used to merge to the patients’ county of residence. The number of SNFs, home health agencies, and physicians per capita were used as supply-side controls to control for potential access to postacute facilities and intensity of hospital care. Controls for socioeconomic status were also used and include county’s percentage of poverty, percentage of unemployment, and median income quartile, as well as an indicator for metropolitan status. These controls were used because MA plans tend to operate in urban counties with higher FFS spending.

 Family composition was also controlled by using the county’s percentage of adults older than 65 years who are married, as literature has suggested that family support may influence discharge destinations.

 Several outcome measures were constructed. Using discharge destination information from the SID, a binary indicator was first constructed for the probability of discharge home, equaling 1; otherwise, 0. Second, among encounters that were discharged to postacute care (skilled nursing, home health, inpatient rehabilitation, and long-term care hospital), the probability of discharge to each destination was also examined and coded as a binary measure equaling 1 for the target destination; otherwise, 0. Third, to measure the intensity of inpatient utilization prior to being discharged to postacute care, both the length of stay and total cost of the inpatient hospital stay were examined among encounters that were discharged to postacute care. The HCUP data report total inpatient facility charges, which can be converted to costs by multiplying the hospital’s cost-to-charge ratio obtained from HCUP. The cost-to-charge ratio is calculated annually for each hospital using information from the hospital’s Medicare Cost Reports.

A multivariate linear regression model was constructed to first estimate the predicted probability of discharge home within each MA and FFS cohort. Then, for encounters discharged to postacute destinations, we separately estimated 4 models using binary outcomes (for example, outcome for skilled nursing discharge was coded equaling 1; otherwise, 0). For these binary outcomes, the coefficients indicate the predicted probability of each outcome. Further, we separately estimated 2 models of inpatient utilization (length of stay and inpatient cost) among encounters discharged to the 4 postacute destinations. First, base models of demographic characteristics, CCI score, and socioeconomic status were run in predicting outcomes. Then, the geriatric syndrome risk measure was added to test if it was independently associated with utilization and outcomes.

To compare MA and FFS encounters, separate multivariate linear regressions were run to estimate differences in each outcome (discharge home, discharges to each of the 4 postacute destinations, and inpatient lengths of stay and cost) between the 2 cohorts. A binary indicator was set to 1 if the encounter was covered by MA and 0 if covered by FFS. All models adjusted for demographic characteristics, CCI score, geriatric syndrome risk, and socioeconomic factors. All models adjusted for the discharging hospital and the year of the hospital discharge, and models clustered error terms at the hospital level to account for correlation of model within hospitals.

A total of 3,098,369 inpatient Medicare encounters discharged to home or postacute facilities occurred during 2010-2014 in Florida. Fifty-two percent of these discharges were to a postacute facility. Thirty-one percent of all discharges to home and postacute care were covered by MA, whereas 27% of all discharges to postacute care were covered by MA (

). Of postacute discharges, patients with FFS coverage were older and more likely to be women and White. Additional clinical characteristics indicate that FFS patients were not sicker as measured by the CCI, but they were more likely to be categorized as geriatric syndrome high risk than MA discharges. Eleven percent of FFS patients discharged to postacute care were categorized as high risk compared with 8% of MA discharges to postacute care. Because skilled nursing and home health encompass the majority of discharges to postacute care, we focus our discussion on these 2 destinations.

Postacute Destination by Geriatric Syndrome Risk

Geriatric syndrome risk is a robust predictor of postacute destination (

). In base models, the CCI was not statistically associated with most postacute destinations (it was statistically associated in the FFS cohort discharged to home health and MA and FFS cohorts discharged to long-term care hospitals, but the magnitude of the association was very small; see 

). When the geriatric syndrome risk measure was added, it was independently and statistically associated with discharges to skilled nursing and home health. Among MA encounters discharged to postacute care, geriatric syndrome moderate-risk encounters were 19 percentage points (

 < .01) more likely and high-risk encounters were 28 percentage points (

 < .01) more likely to be discharged to skilled nursing than nonrisk encounters. The same pattern held true for FFS encounters. Among MA and FFS encounters discharged to postacute care, geriatric syndrome moderate-risk encounters were 20 percentage points (

 <.01) less likely and high-risk were 28 to 30 percentage points (

 < .01) less likely to be discharged to home health than nonrisk encounters. Inpatient rehabilitation and long-term care hospital results are shown in eAppendix Table 4, but we do not see large percentage-point differences among users of inpatient rehabilitation or long-term care hospitals.

Geriatric syndrome risk is also a robust predictor of inpatient utilization among encounters discharged to postacute care. In both MA and FFS cohorts (Figure 1), geriatric syndrome high-risk encounters subsequently discharged to any postacute destination stayed longer in the inpatient setting and cost more than nonrisk encounters (MA, 1.25 days and $1892 [

Geriatric syndrome high-risk encounters were also less likely to be discharged home (

). Among MA and FFS encounters, geriatric syndrome high-risk encounters were 35 to 40 percentage points (

 < .01) less likely and moderate-risk encounters were 24 to 26 percentage points (

 < .01) less likely to be discharged home compared with nonrisk encounters.

Postacute Destination by Source of Coverage

Geriatric syndrome risk explains some of the differences in postacute destination between MA and FFS discharges (

). When adjusting only for a traditional measure of comorbidity, MA encounters were more likely to be discharged home than FFS (8 percentage points; 

 < .01) and, among postacute discharges, less likely to be discharged to skilled nursing (2 percentage points; 

 < .01) than FFS. When geriatric syndrome risk is added, the difference between MA and FFS in the probability of discharge to home falls to 6 percentage points from 8 percentage points, a relative reduction of 4%. Notably, when geriatric syndrome risk is added to the model, there is no statistically significant difference between MA and FFS in the probability of discharge to skilled nursing. After adjusting for geriatric syndrome risk, inpatient cost differences between MA and FFS encounters discharged to postacute care decrease to $585 from $610.

We find that geriatric syndrome risk factors that can be measured using existing administrative data are associated with higher postacute utilization among hospitalized Medicare patients. That is, inpatient encounters with high geriatric syndrome risk are more likely to have longer and more expensive inpatient stays, less likely to be discharged home, and more likely to use high-intensity postacute destinations (ie, skilled nursing vs home health). We do not see large differences in inpatient rehabilitation or long-term care hospital utilization, potentially because of low prevalence in discharges to these 2 destinations. Further, we find that after controlling for geriatric syndrome risk factors, unadjusted utilization differences between MA and FFS Medicare patient discharges become insignificant or attenuated. After accounting for geriatric syndrome risk, there were no differences in the use of skilled nursing among MA and FFS inpatient encounters.

The added value of the factors that make up the geriatric syndrome risk measure indicates that these factors may be important to incorporate into risk adjustment for payment or quality reporting, especially when comparing across different payment programs. This study demonstrates that geriatric syndrome risk factors, as conceptualized here, can be captured using existing administrative data, are predictive of utilization, and do narrow differences across payment models when assessing utilization and costs. It appears that, in some instances, the MA program utilizes postacute care differentially from FFS Medicare for their high-risk enrollees, but it is unclear whether these differences are efficient for the Medicare program as a whole. Risk-adjusted capitation or value-based payment policies that do not currently incorporate these types of factors may be penalizing providers who care for cohorts with disproportionately higher levels of geriatric syndrome risk. Moreover, quality reporting that does not account for these measures may also be under- or overestimating the quality of care provided. For example, research has demonstrated that high-cost patients (ie, long-term/short-term nursing home care and home health care users) enrolled in MA plans have an increased rate of leaving MA and joining FFS Medicare.

 Coupled with the fact that federal policies have expanded benefit packages aimed at older adults with chronic conditions,

 accurately assessing how capitated systems best serve their high-risk patients in postacute services will be an important policy question as enrollment in MA increases and health care services are increasingly delivered outside the inpatient setting.

Further, these findings suggest that such risk adjustment tools could potentially guide a first stage of discharge planning, before relying on time-intensive clinician-completed assessments. Hospitals and postacute facilities could focus on these high-risk patients during the transition period to ensure care for specialized needs or prioritize information transfer across facilities. Health care providers and systems could choose the set of tools that best suits their needs in identifying patients at risk when, for example, comprehensive assessments cannot be done in times of emergencies or nursing home evacuations.

There are several limitations of our study to be noted. First-time postacute care, long-term care, or readmission to such a long-term facility cannot be differentiated in the HCUP data set. For example, if a patient was admitted to the hospital from a nursing home and then readmitted to the nursing home, we are unable to exclude them from our postacute analyses. Second, MA plans include institutional special needs plans (ISNPs) that may especially target frail elderly patients. We were unable to distinguish these plans from the data set. In 2014, of 1.3 million Medicare Advantage enrollees, 2706 beneficiaries (< 1%) were enrolled in ISNP plans in Florida, indicating that they represent a very small fraction of beneficiaries.

 Third, previous literature has suggested that availability of and access to postacute facilities and providers is an important factor in explaining utilization levels. Although we do not have information on the choice of postacute facilities available to each patient at the time of discharge or the network size of postacute providers contracted with MA plans, we do attempt to control for the supply of each type of postacute provider in the patients’ county of residence. Further, the analysis is done at the encounter level, so we are unable to control for the frequency of unique patients being hospitalized.

The geriatric syndrome risk measure was also calculated here using only inpatient data, the quality of which may vary from hospital to hospital (ie, a hospital with a unit for acute care for the elderly may assess frailty and other geriatric syndromes differently compared with hospitals without such units); further modeling should examine the addition of ambulatory data and other types of frailty risk factors developing in the literature.

 Previous research using a survey-based assessment for patients admitted to the hospital has found higher prevalence of geriatric syndromes; for example, bladder incontinence was present in 44% of a cohort of 500 patients 70 years and older who were admitted to the hospital.

 However, our prevalence derived from inpatient claims is in line with previous research demonstrating that 8% of a large FFS Medicare cohort and 11% of a large cohort of older Americans can be categorized as frail through inpatient and outpatient claims.

 Health care providers and systems need to find the right balance of utilizing time- and resource-intensive assessment instruments vs less resource-intensive data-driven techniques to identify patients at risk.

We have attempted to identify and adjust for measurable characteristics between MA and FFS discharges. Although we control for a range of characteristics pertinent to older adults, it is possible that we have not adjusted for all underlying differences across the 2 cohorts. In particular, with regard to the admitted inpatient stay cohorts, prior research has suggested that MA enrollees are less likely to be admitted to the inpatient setting in the first place, suggesting that those who are less healthy may be captured within our MA inpatient sample.

 Finally, the current data are also limited to Florida during 2010-2014; although it is a large state with a significant proportion of US Medicare beneficiaries, we cannot fully generalize these results to the full Medicare program across the nation.

Despite these limitations, this study expands our existing knowledge of the role of nontraditional comorbidity factors on health care utilization and outcomes and on how MA utilizes postacute services compared with FFS Medicare. We analyzed data on a particularly vulnerable cohort of hospitalized patients with high geriatric syndrome risk. This study suggests that high-risk hospitalized patients utilize postacute services differently. As services are increasingly bundled and paid according to the quality of care they provide, it is important to optimally match high-risk patients to appropriate postacute services. Further research is needed to evaluate the efficiency of the MA program on postacute service delivery compared with that of the FFS program, especially among a cohort of high-risk Medicare beneficiaries.


 Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health (SSW, JPW), Baltimore, MD; Department of Medicine, Johns Hopkins School of Medicine (MB), Baltimore, MD.

 This project was supported in part by grant number T32HS000029 from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.

 The manuscript references measures that are part of the Adjusted Clinical Groups (ACG) System. Johns Hopkins University holds the copyright to the ACG System and receives royalties from its distribution. Dr Weiner is a member of a group of researchers who develop and maintain the ACG System with support from Johns Hopkins University. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (SSW, MB, JPW); acquisition of data (SSW); analysis and interpretation of data (SSW); drafting of the manuscript (SSW); critical revision of the manuscript for important intellectual content (SSW, MB, JPW); statistical analysis (SSW); and supervision (SSW, MB, JPW).

 Shannon S. Wu, PhD, Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, 624 N Broadway, Room 428, Baltimore, MD 21205. Email: swu50@jhu.edu.

1. Hussey PS, Huckfeldt P, Hirshman S, Mehrotra A. Hospital and regional variation in Medicare payment for inpatient episodes of care. 

. 2015;175(6):1056-1057. doi:10.1001/jamainternmed.2015.0674

2. Newhouse JP, Garber AM. Geographic variation in Medicare services. 

. 2013;368(16):1465-1468. doi:10.1056/NEJMp1302981

3. Meyers DJ, Mor V, Rahman M. Medicare Advantage enrollees more likely to enter lower-quality nursing homes compared to fee-for-service enrollees. 

. 2018;37(1):78-85. doi:10.1377/hlthaff.2017.0714

4. Huckfeldt PJ, Escarce JJ, Rabideau B, Karaca-Mandic P, Sood N. Less intense postacute care, better outcomes for enrollees in Medicare Advantage than those in fee-for-service. 

. 2017;36(1):91-100. doi:10.1377/hlthaff.2016.1027

5. Kumar A, Rahman M, Trivedi AN, Resnik L, Gozalo P, Mor V. Comparing post-acute rehabilitation use, length of stay, and outcomes experienced by Medicare fee-for-service and Medicare Advantage beneficiaries with hip fracture in the United States: a secondary analysis of administrative data. 

. 2018;15(6):e1002592. doi:10.1371/journal.pmed.1002592

6. Waxman DA, Min L, Setodji CM, Hanson M, Wenger NS, Ganz DA. Does Medicare Advantage enrollment affect home healthcare use? 

7. Long P, Abrams M, Milstein A, et al, eds. 

Effective Care for High-Need Patients: Opportunities for Improving Outcomes, Value, and Health

8. Joynt KE, Figueroa JF, Beaulieu N, Wild RC, Orav EJ, Jha AK. Segmenting high-cost Medicare patients into potentially actionable cohorts. 

. 2017;5(1-2):62-67. doi:10.1016/j.hjdsi.2016.11.002

9. Burke RE, Juarez-Colunga E, Levy C, Prochazka AV, Coleman EA, Ginde AA. Patient and hospitalization characteristics associated with increased postacute care facility discharges from US hospitals. 

. 2015;53(6):492-500. doi:10.1097/MLR.0000000000000359

10. Coleman EA. Falling through the cracks: challenges and opportunities for improving transitional care for persons with continuous complex care needs. 

. 2003;51(4):549-555. doi:10.1046/j.1532-5415.2003.51185.x

11. Coleman EA, Berenson RA. Lost in transition: challenges and opportunities for improving the quality of transitional care. 

. 2004;141(7):533-536. doi:10.7326/0003-4819-141-7-200410050-00009

12. Coleman EA, Min SJ, Chomiak A, Kramer AM. Posthospital care transitions: patterns, complications, and risk identification. 

. 2004;39(5):1449-1465. doi:10.1111/j.1475-6773.2004.00298.x

13. Greysen SR, Stijacic Cenzer I, Auerbach AD, Covinsky KE. Functional impairment and hospital readmission in Medicare seniors. 

. 2015;175(4):559-565. doi:10.1001/jamainternmed.2014.7756

14. Johnston KJ, Wen H, Hockenberry JM, Joynt Maddox KE. Association between patient cognitive and functional status and Medicare total annual cost of care: implications for value-based payment. 

. 2018;178(11):1489-1497. doi:10.1001/jamainternmed.2018.4143

15. Meddings J, Reichert H, Smith SN, et al. The impact of disability and social determinants of health on condition-specific readmissions beyond Medicare risk adjustments: a cohort study. 

. 2017;32(1):71-80. doi:10.1007/s11606-016-3869-x

16. Shih SL, Gerrard P, Goldstein R, et al. Functional status outperforms comorbidities in predicting acute care readmissions in medically complex patients. 

. 2015;30(11):1688-1695. doi:10.1007/s11606-015-3350-2

17. Greysen SR, Stijacic Cenzer I, Boscardin WJ, Covinsky KE. Functional impairment: an unmeasured marker of Medicare costs for postacute care of older adults. 

. 2017;65(9):1996-2002. doi:10.1111/jgs.14955

18. Burke RE, Whitfield EA, Hittle D, et al. Hospital readmission from post-acute care facilities: risk factors, timing, and outcomes. 

. 2016;17(3):249-255. doi:10.1016/j.jamda.2015.11.005

19. Middleton A, Graham JE, Lin YL, et al. Motor and cognitive functional status are associated with 30-day unplanned rehospitalization following post-acute care in Medicare fee-for-service beneficiaries. 

. 2016;31(12):1427-1434. doi:10.1007/s11606-016-3704-4

20. Kumar A, Karmarkar A, Downer B, et al. Current risk adjustment and comorbidity index underperforms in predicting post-acute utilization and hospital readmissions after joint replacements: implications for Comprehensive Care for Joint Replacement model. 

. 2017;69(11):1668-1675. doi:10.1002/acr.23195

21. Kim DH, Schneeweiss S, Glynn RJ, Lipsitz LA, Rockwood K, Avorn J. Measuring frailty in Medicare data: development and validation of a claims-based frailty index. 

. 2018;73(7):980-987. doi:10.1093/gerona/glx229

22. Joynt Maddox KE, Orav EJ, Zheng J, Epstein AM. How do frail Medicare beneficiaries fare under bundled payments? 

. 2019;67(11):2245-2253. doi:10.1111/jgs.16147

23. Lakhan P, Jones M, Wilson A, Courtney M, Hirdes J, Gray LC. A prospective cohort study of geriatric syndromes among older medical patients admitted to acute care hospitals. 

. 2011;59(11):2001-2008. doi:10.1111/j.1532-5415.2011.03663.x

24. Inouye SK, Studenski S, Tinetti ME, Kuchel GA. Geriatric syndromes: clinical, research, and policy implications of a core geriatric concept. 

. 2007;55(5):780-791. doi:10.1111/j.1532-5415.2007.01156.x

25. Wu S, Mulcahy J, Kasper JD, Kan HJ, Weiner JP. Comparing survey-based frailty assessment to Medicare claims in predicting health outcomes and utilization in Medicare beneficiaries. 

. Published online May 31, 2019. doi:10.1177/0898264319851995

26. The Johns Hopkins ACG System. Accessed May 20, 2018. https://www.hopkinsacg.org/

27. Weiner JP, Dobson A, Maxwell SL, Coleman K, Starfield B, Anderson GF. Risk-adjusted Medicare capitation rates using ambulatory and inpatient diagnoses. 

28. Bandeen-Roche K, Seplaki CL, Huang J, et al. Frailty in older adults: a nationally representative profile in the United States. 

2015;70(11):1427-1434. doi:10.1093/gerona/glv133

29. Kan HJ, Kharrazi H, Leff B, et al. Defining and assessing geriatric risk factors and associated health care utilization among older adults using claims and electronic health records. 

. 2018;56(3):233-239. doi:10.1097/MLR.0000000000000865

30. Anzaldi LJ, Davison A, Boyd CM, Leff B, Kharrazi H. Comparing clinician descriptions of frailty and geriatric syndromes using electronic health records: a retrospective cohort study. 

. 2017;17(1):248. doi:10.1186/s12877-017-0645-7

31. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with 

. 1992;45(6):613-619. doi:10.1016/0895-4356(92)90133-8

32. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. 

. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8

33. Nicholas LH, Wu S. Do Medicare Advantage rebates reduce enrollees’ out-of-pocket spending? 

. 2020;77(5):474-482. doi:10.1177/1077558718807847

34. Kane RL, Lin WC, Blewett LA. Geographic variation in the use of post-acute care. 

. 2002;37(3):667-682. doi:10.1111/1475-6773.00043

35. Meyers DJ, Belanger E, Joyce N, McHugh J, Rahman M, Mor V. Analysis of drivers of disenrollment and plan switching among Medicare Advantage beneficiaries. 

. 2019;179(4):524-532. doi:10.1001/jamainternmed.2018.7639

36. Rahman M, Keohane L, Trivedi AN, Mor V. High-cost patients had substantial rates of leaving Medicare Advantage and joining traditional Medicare. 

. 2015;34(10):1675-1681. doi:10.1377/hlthaff.2015.0272

37. Goldberg EM, Trivedi AN, Mor V, Jung HY, Rahman M. Favorable risk selection in Medicare Advantage: trends in mortality and plan exits among nursing home beneficiaries. 

. 2017;74(6):736-749. doi:10.1177/1077558716662565

38. Bunis D. Medicare Advantage plans adding benefits for 2020. News release. AARP. September 4, 2019. Accessed March 23, 2020. https://www.aarp.org/health/medicare-insurance/info-2019/medicare-advantage-expanded-benefits.html

39. Segal JB, Chang HY, Du Y, Walston JD, Carlson MC, Varadhan R. Development of a claims-based frailty indicator anchored to a well-established frailty phenotype. 

. 2017;55(7):716-722. doi:10.1097/MLR.0000000000000729

40. Medicare Advantage: Special Needs Plan (SNP) enrollment, by SNP type. Kaiser Family Foundation. Accessed March 23, 2020. https://www.kff.org/medicare/state-indicator/snp-enrollment-by-snp-type

41. Afendulis CC, Chernew ME, Kessler DP. The effect of Medicare Advantage on hospital admissions and mortality. 

. 2017;3(2):254-279. doi:10.1162/AJHE_a_00074

Articles

Geriatric syndrome risk factors play a role in understanding postacute destination within and between Medicare Advantage and fee-for-service Medicare cohorts.

Geriatric Syndrome Risk Factors Among Hospitalized Postacute Medicare Patients

 Analyses of emergency department (ED) use require visit classification algorithms based on administrative data. Our objectives were to present an expanded and revised version of an existing algorithm and to use this tool to characterize patterns of ED use across US hospitals and within a large sample of health plan enrollees.

 Observational study using National Hospital Ambulatory Medical Care Survey ED public use files and hospital billing data for a health plan cohort.

 Our Johns Hopkins University (JHU) team classified many uncategorized diagnosis codes into existing New York University Emergency Department Algorithm (NYU-EDA) categories and added 3 severity levels to the injury category. We termed this new algorithm the NYU/JHU-EDA. We then compared visit distributions across these 2 algorithms and 2 other previous revised versions of the NYU-EDA using our 2 data sources.

 Applying the newly developed NYU/JHU-EDA, we classified 99% of visits. Based on our analyses, it is evident that an even greater number of US ED visits than categorized by the NYU-EDA are nonemergent. For the first time, we provide a more complete picture of the level of severity among patients treated for injuries within US hospital EDs, with about 86% of such visits being nonsevere. Also, both the original and updated classification tools suggest that, of the 38% of ED visits that are clinically emergent, the majority either do not require ED resources or could have been avoided with better primary care.

 The updated NYU/JHU-EDA taxonomy appears to offer cogent retrospective inferences about population-level ED utilization.

There is renewed interest in understanding emergency department (ED) use patterns in populations, both because of increased use associated with healthcare reform and as private payers seek to stem their rising ED spending.

To assess the appropriateness of ED use at the population level, validated classification methods that use available administrative data will be required.

Our analysis using an updated classification suggests that an even greater number of ED visits than previously categorized are nonemergent.

Health plans and other organizations might use ED visit classification algorithms to gain an understanding about how populations make use of hospital services.

The New York University Emergency Department Algorithm (NYU-EDA) is widely used to classify emergency department (ED) visits.

 This measurement tool’s development occurred in the late 1990s and was based on 5700 ED discharge abstracts from 6 hospitals in the Bronx, a borough of New York City. The NYU-EDA probabilistically classified 659 diagnosis codes from the 

). The original NYU-EDA mapped only about 5% of all 

 diagnosis codes. We propose an algorithm that remedies this shortfall and classifies nearly all ED visits.

The NYU-EDA has been applied in health services research studies to identify emergent visits that required ED care.

 Several studies have focused on nonemergent and primary care—treatable ED visits and evaluated emergent and nonemergent utilization patterns to assess the impact of healthcare reforms.

 Estimates of proportions of nonemergent visits have ranged between 17% and 49%. One study looked at primary care—sensitive (PCS) visits (ie, emergent visits that are potentially avoidable and nonemergent and primary care–treatable visits) and found that up to 50% of ED visits were PCS in a statewide all-payer claims database with 92% of ED visits classified.

Evidence for the validity of the NYU-EDA has grown over 2 decades. Emergent visits were associated with total charges and increased likelihood of death and inpatient hospitalization directly from the ED and within 30 days from a previous visit.

 However, researchers and emergency medicine clinicians have cautioned against using visit classifications based solely on discharge diagnoses for interventions aimed at reducing unnecessary visits or for denying payment. First, underlying differences in morbidity and access to care may, to some degree, account for utilization patterns that would be detected by an ED visit classification algorithm. Second, there are reasons for visits on the individual level that may be appropriate for ED utilization, which can differ from discharge diagnoses that categorize the encounter as nonemergent. For example, patients who are experiencing chest pain and come to the ED for evaluation are not necessarily inappropriately using the ED. ED visit classifications are useful tools for understanding the healthcare needs of populations, not the medical needs of individual patients.

A team of Johns Hopkins University (JHU) emergency medicine physicians and health services researchers has further updated and expanded the NYU-EDA using their best clinical judgment and diagnosis aggregations from the Adjusted Clinical Groups (ACG) System.

 In this revised JHU version of the NYU-EDA (or NYU/JHU-EDA for short) we undertook 3 significant modifications and improvements to the original version and updates undertaken by other teams. First, rather than assigning 

 codes probabilistically, we classify each ED visit into 1 of 11 categories. Second, rather than placing all injuries into 1 category, we subcategorize injuries into 3 severity levels: nonsevere injuries, severe injuries, and severe injuries that are likely to require inpatient admissions. Third, we significantly expand the classification of 

In this article, we describe the updated NYU/JHU-EDA, and, using data from a federal survey of US hospital EDs and a large claims database from multiple health plans, we compare results of our revised tool with the original NYU-EDA and 2 earlier modifications developed by Johnston et al and Ballard et al.

The first objective of this article is to offer a description and first-stage assessment of our ED classification algorithm. The second goal is to use this methodology to offer an account of use patterns of American EDs based on a representative sample of patients visiting hospital EDs and a large national sample of health plan enrollees. In addition to describing our new measurement tool, our analysis adds to the literature on how Americans use EDs and will offer insights into how health plans and other organizations might use classification algorithms to gain an understanding of how populations make use of hospital EDs.

Review of Previous Approaches for Classifying ED Visits

The original NYU-EDA first classifies common primary ED discharge diagnoses as having varying probabilities of falling into each of the 4 following categories: (1) nonemergent; (2) emergent, primary care treatable; (3) emergent, ED care needed, and preventable or avoidable with timely and effective ambulatory care; and (4) emergent, ED care needed, and not preventable.

 The original NYU system categorizes certain diagnoses separately and directly into 5 additional categories: injuries, psychiatric conditions, alcohol related, drug related, or unclassified.

The adaptation by Ballard et al sums the NYU-EDA probabilities for nonemergent and emergent primary care—treatable visits and compares this sum with the total probability of the emergent, ED care needed categories.

 Depending on the larger of the 2 resultant likelihoods, visits are classified as nonemergent or emergent, or as intermediate when there is an equal probability of being nonemergent or emergent. The Ballard et al method classifies visits into 1 of 8 categories, which have been shown to be a good predictor of subsequent hospitalization and death within 30 days of an ED visit.

 After 2001, the NYU-EDA was not updated, and newly added diagnosis codes were not classified. In 2018, Johnston and colleagues identified new codes that are “nested” within previously classified diagnoses and applied the original probabilistic weights to these codes.

 New diagnoses that remained unclassified were “bridged” to already weighted codes using 

-based condition groupings from the Agency for Healthcare Research and Quality Clinical Classification System.

 Instances in which a new diagnosis mapped to several codes with different weights were resolved in favor of a code most likely to represent an unavoidable emergent visit.

 Because assigned weights sum to 1, both the original NYU-EDA and the update by Johnston et al describe a collection of ED visits by averaging weights.

We took a different approach to update, enhance, and expand the NYU-EDA method. We did not use probabilities but rather assigned primary discharge diagnoses to single classes and uniquely classified each ED visit. For codes that had been previously included in the original NYU-EDA, we based our updated assignments on the category with the highest probability. We resolved cases of equally high probabilities among multiple categories by giving preference to the emergent, ED care needed category.

To build our revised methodology, we combined ED encounter data from the National Hospital Ambulatory Medical Care Survey (NHAMCS) from the period of 2009 to 2013. We used data from the 2014 survey for validation.

An updated emergency visit classification tool enables managers to make valid inferences about levels of appropriateness of emergency department utilization and healthcare needs within a population.

Jonathan P. Weiner, DrPH

Articles

Geriatric Syndrome Risk Factors Among Hospitalized Postacute Medicare Patients

A Revised Classification Algorithm for Assessing Emergency Department Visit Severity of Populations

Assessing Markers From Ambulatory Laboratory Tests for Predicting High-Risk Patients

Encouraging Value-Based Insurance Designs in State Health Insurance Exchanges