Heart failure (HF) is a major health care burden in older adults (Mozaffarian et al., 2016) and the leading cause of hospital readmission in the U.S. population (Medicare Payment Advisory Commission, 2016). In the United States, the Centers for Medicare & Medicaid Services developed the Hospital Readmission Reduction Program (HRRP) to decrease readmission rates (Schwartz et al., 2014). Thirty-day all-cause readmission rates are used to incentivize or penalize hospitals to improve their quality of care (Schwartz et al., 2014). Since the HRRP was initiated in 2012, the 30-day readmission rate for patients with HF decreased from 19.5% in 2010 to 17% in 2014 (Buntin, Burke, Hoaglin, & Blumenthal, 2011; Wasfy et al., 2017).
In a recent systematic review, Kansagara et al. (2011) reported that a potential decrease in 30-day readmission rates could be achieved if the predictive value of risk factors identified in electronic health record (EHR) systems were well utilized. Current risk prediction models for hospital readmission have limited credibility because data have been sourced mainly from administrative and general population datasets (Dharmarajan et al., 2017). It has been suggested that risk prediction models using data from EHR systems might lead to more accurate predictions and help identify previously unrecognized risk factors for readmission in hospitalized geriatric patients with HF.
According to a recent study by Dharmarajan et al. (2017), older age was associated with 30-day readmission rate among a Medicare population. Krumholz et al. (2000) suggested HF was a predictor of 6-month hospital readmission in older adults. However, little is known regarding which risk factors, other than age, may be associated with 30-day hospital readmission in older adults with acute HF.
The eTracker-HF is a health risk management software program built into EHR systems designed to identify risk factors for clinicians via the risk identifier function in the EHR. The program presents risk factors on a screenshot. Using the eTracker-HF, the current authors sought to evaluate potential risk factors and their association with 30-day hospital readmission in patients with HF from demographic, geriatric, cardiovascular, and discharge perspectives.
Study Site and Participants
The current study was performed over a 12-month period from November 1, 2013 to October 31, 2014. Study inclusion criteria were consecutively hospitalized older adults (age ≥65) who were admitted to a hospital floor or intensive care unit and discharged with a principal or secondary diagnosis of HF with respect to International Classification of Disease 9th Revision Clinical Modification diagnostic code assignments (402.01, 402.11, 402.91, 404.03, 404.13, 404.93, 404.11, 404.91, 404.3, 404.13, 404.93, or 428.x). Exclusion criteria were patients who were home health care recipients or under hospice care. A total of 167 participants were excluded from the study. Specific patient cases were excluded from analysis (n = 299) due to incorrect data input (n = 97), follow-up loss (n = 35), current home health care recipient (n = 114), hospice service use (n = 28), or transfer to another hospital (n = 25). During the study period, only initial admission cases were entered into the analysis when participants were admitted multiple times.
eTracker-HF, Early Hospital Readmission Risk Factor Identifier Tool
The Acute Care for Elders (ACE) Tracker software was developed to identify risk factors that may prevent undesired health outcomes, such as in-hospital falls and pressure ulcers, with the goal of improving overall quality of life of hospitalized patients (Malone et al., 2010; Malone, Yoo, & Goodwin, 2014). The ACE Tracker was previously validated for use as a risk identifier of several hospital quality indicators in hospitalized geriatric patients (Malone et al., 2010). The ACE Tracker identified risk factors and categorized them into four domains: demographic, geriatric, clinical, and discharge disposition. The real-time risk identifier function of the ACE Tracker prompted loading of the eTracker-HF software into the tracker to identify readmission risk factors in hospitalized geriatric patients with decompensated HF.
Data Collection Process
The eTracker-HF tracked inpatient variables for every hospitalized older adult with HF. The software was used in 15 acute care hospitals of Aurora Health Care located in Wisconsin. Hospital bed size ranged from 125 to 965 certified beds. The eTracker-HF integrated patient-identifying information and risk factors from EHR systems for discharged patients who had HF as their principal or secondary diagnosis on initial admission. Risk factor information was presented through a screenshot, allowing clinicians to quickly recognize risk factors when logged into EHR systems. Research assistants transferred encrypted eTracker-HF data to Microsoft® Office Excel sheets every week. The institutional review board of Aurora Health Care approved the study.
Outcomes and Measurement
The measured outcomes were all-cause and HF readmission within 30 days after discharge date of index hospital admission. Readmission was defined as the first subsequent inpatient admission in the 15 participating hospitals. In total, 12 variables were assessed from inpatient data at the time of index hospital admission. Variables were grouped into four domains: demographic, geriatric, clinical, and discharge disposition. The demographic domain included age, gender, ethnicity, and education level. The geriatric domain included cognitive impairment, delirium, potentially inappropriate medications, and Katz Activities of Daily Living (ADL) Index score. The clinical domain included left ventricular ejection fraction (EF), advanced chronic kidney disease (CKD) stage, and history of myocardial infarction. Discharge disposition included home without home health, home with home health, or skilled nursing facility (SNF) placement.
Cognitive impairment was determined using medical histories. The Confusion Assessment Method (CAM) was used to detect delirium. The CAM is a valid tool for hospitalized older adults (Inouye et al., 1990). Potentially inappropriate medication use was defined using the Beers Criteria medication list (American Geriatrics Society 2015 Beers Criteria Update Expert Panel, 2015). The Katz ADL Index assessed physical function and has been validated as the most common measurement of physical function in hospitalized older adults (Buurman, van Munster, Korevaar, de Haan, & de Rooij, 2011). The Katz ADL Index ranks adequacy of performance in six functions (i.e., bathing, dressing, toileting, transferring, continence, and feeding). The Katz ADL Index was scored yes/no for independence in each of the six functions and was weighted with 2 as no impairment, 1 as mild impairment, and 0 as total impairment. A Katz ADL Index score of 0 represented total dependence of all six functions (minimum score) and 12 represented total independence of all six functions (maximum score). Patients were assessed with the Katz ADL Index within 24 hours of hospital admission. All data were obtained from the eTracker-HF after input by nursing staff who completed the assessments. The most recent EF data were defined within the past 12 months and grouped by EF ≥40 and EF <40. The most recent CKD data were defined and grouped based on estimated glomerular filtration rate grade 3 (59–30), grade 4 (29–15), and grade 5 (<15) (units in mL/min/1.73 m2).
Differences in characteristics were analyzed by readmission groups. One-way analysis of variance (ANOVA) and chi-square tests were performed for continuous and categorical variables, respectively. ANOVA tests were applied in two comparison steps. The first step was comparison between the all-cause readmission and no readmission groups. The second step was comparison between the HF readmission, readmission for a reason other than HF, and no readmission groups. Separate p values were computed within each comparison step. Bonferroni correction was applied as post-hoc analysis. Multivariate analysis was performed using all-cause and HF readmission as the outcome variables.
Spearman's rank correlations of the independent variables were used to determine whether any variables were highly correlated and therefore not recommended for inclusion in the same regression model. A high correlation was defined as a correlation coefficient >|0.7| (Tabachnick & Fidell, 2006). No high correlations were found among independent variables. A multilevel mixed-effect logistic regression model was used to account for correlation within clusters (Hubbard et al., 2010). Multilevel regression models included patient- and hospital-level data. Odds ratios and p values for each covariate were calculated after adjusting explanatory variables (i.e., age, gender, education, cognitive impairment, delirium, potentially inappropriate medication use, Katz ADL Index, EF <40, advanced CKD, previous myocardial infarction, and discharge disposition). Forward and backward stepwise selection methods were applied to ensure that the method used for the selection of covariates had no impact on the final multilevel regression model. p values between 0.10 and 0.20 were used for entry and retention in the model, respectively. The concordance index was examined for all-cause and HF readmission in multivariate analysis. All statistical analyses were performed using SAS statistical software version 9.4. All reported p values were two-tailed, and p < 0.05 was considered statistically significant.
A total of 2,279 participants were analyzed. Table 1 presents all study factors based on readmission group as univariate (simple) analysis between each variable and hospital readmission. All-cause and HF 30-day readmission rates were 22.3% (n = 508) and 9.8% (n = 223), respectively. Older age, non-White ethnicity, delirium, physical impairment, EF <40, advanced CKD stage 4 or 5, and previous myocardial infarction were associated with all-cause and HF 30-day hospital readmission (all p < 0.001 except for previous myocardial infarction; p = 0.044 in all-cause readmission, p = 0.026 in HF readmission of previous myocardial infarction). In all-cause hospital readmission (22.3%), the causes of readmission were HF (9.8%), pneumonia (4.5%), heart disease (3.8%; coronary artery disease, arrhythmia, and cerebral vascular disease), acute renal failure (1.2%), and other (3%).
Participant Characteristics (N= 2,279)
Table 2 displays multi-level multivariate regressions of hospital readmission as well as the odds ratios and corresponding 95% confidence intervals of each variable. Of the four demographic variables assessed, older age and non-White ethnicity were associated with increased all-cause hospital readmission. Of the four geriatric variables assessed, delirium and physical impairment (as assessed by the Katz ADL Index) were associated with increased readmission rates for all-cause readmission. Of the three clinical variables assessed, EF < 40, advanced CKD (stage 4 or 5), and previous myocardial infarction were associated with increased all-cause readmission. Of the discharge disposition, home health care use was associated with lower all-cause readmission. Discharge to SNF was associated with increased risk of all-cause readmission. In HF readmission regressions, risk factors of all-cause readmission were associated with HF readmission except for discharge to SNF, which was not significant (p = 0.137). Similar to all-cause readmission, home health care use was associated with lower HF readmission. C-statistics (i.e., area under the ROC curve) ranged from 0.584 to 0.810 for all-cause readmission and 0.639 to 0.827 for HF readmission. Delirium and Katz ADL Index maintained the highest c-statistics (>0.75) in all-cause and HF readmission.
Multilevel Multivariate Regressions of Hospital Readmission
Twelve variables identified by the eTracker-HF were evaluated to determine which were associated with all-cause and HF 30-day readmission among hospitalized geriatric patients with decompensated HF. Old age, non-White race/ethnicity, advanced CKD, delirium, functional impairment, low EF, and previous myocardial infarction were associated with higher risk of all-cause and HF readmission. Home health care use was associated with lower risks of all-cause and HF readmission.
Similar to previous reports identifying risk factors of geriatric hospital readmission, older age was associated with all-cause and HF 30-day readmission (Dharmarajan et al., 2017; Kansagara et al., 2011; Krumholz et al., 2000). Higher all-cause and HF 30-day readmission rate in regard to minority race/ethnicity in the current analysis replicated the results from Joynt, Orav, and Jha (2011). Failure of transitional care after hospital discharge was suggested as the main explanation for racial disparity in the early hospital readmission gap. Further study is warranted to explain why racial disparity exists and how disparity can be resolved with interventions.
In previous studies, Patel et al. (2015) did not systemically exclude patients with delirium, and Hyunh et al. (2016), in a nationwide study from Australia, did not control for delirium. In the current study, prevalence of delirium was approximately 24% in the sampled patient population. Hyunh et al. (2016) also did not include discharge status in their multivariate regression analysis. Adjusting for delirium and discharge status is necessary because these factors might be associated with readmission risk in patients with HF.
Delirium and its lingering cognitive effects may persist for weeks to months beyond hospital discharge. A study by Uthamalingam, Gurm, Daley, Flynn, and Capodilupo, (2011) demonstrated an independent association between delirium and increased risk of 30-day readmission in hospitalized older adults. Delirium can result from non-compliance to medication, disruption of daily weight monitoring, inappropriate sodium intake, and delay in seeking medical attention. These factors may explain the increase in 30-day readmission (Uthamalingam et al., 2011).
The current study findings were consistent with those of previous reports that impaired physical function on admission is associated with hospital readmission (Hoyer et al., 2013). There was a positive correlation between impaired function and readmission rate in hospitalized older adults (Hoyer et al., 2013). A recent systematic review of sociodemographic, clinical, and hospital factors by Pedersen, Meyer, and Uhrenfeldt (2017) extended the current findings and identified that functional impairment is universally one of the strongest risk factors of early hospital readmission in older adults regardless of comorbidities and causes of index hospital admission. The current findings supported a dose-response relationship that showed that more impaired physical function was associated with higher odds ratios of all-cause and HF readmission.
A series of meta-analyses predicting early readmission of patients with HF identified CKD as a risk factor (Rahimi et al., 2014; Saito, Negishi, & Marwick, 2016). Advanced CKD can contribute to worse outcomes in patients with HF in a number of ways (Smith et al., 2013). For example, advanced CKD leads to HF via volume expansion. CKD also worsens HF via reduced renal perfusion (Smith et al., 2013).
The current study demonstrated an association between self-reported history of myocardial infarction and HF hospital readmission. Previous studies have reported similar results (Jhaveri et al., 2012; Mentz et al., 2013). Coronary artery disease may be associated with increased hospitalization risk due to subclinical ischemia with worsening of underlying HF (Jhaveri et al., 2012; Mentz et al., 2013).
HF is the second most common diagnosis in Medicare home health patients (Fortinsky, Madigan, Sheehan, Tullai-McGuinness, & Kleppinger, 2014). Using nationally representative data, O'Connor, Hanlon, Naylor, and Bowles (2015) reported that longer home health stays and more skilled nursing visits reduced hospitalization in Medicare patients. The current study findings also support the previously reported effect of home health discharge on reduced risk of hospital readmission in older adults with HF (O'Connor et al., 2015).
A possible direction for future research is to examine whether any interventions, in response to the findings from the risk identifier function among high-risk patients, have the potential to lower the risk of hospital readmission. For example, daily delirium screenings by interdisciplinary teams may help lower the risk of delirium occurrences in hospitalized older adults (Yoo et al., 2014). Further, pre/post intervention studies would help determine whether the eTracker-HF is associated with reduced delirium occurrences and lowered hospital readmission rates.
The current study has several strengths. In contrast to previously validated readmission models, the eTracker-HF can be applied to several specific older adult populations and diseases (Donzé et al., 2016). The eTracker-HF can also be easily adjusted to store data from EHRs. For example, previous myocardial infarction could be replaced by either previous cardiac catheterization or coronary bypass surgery.
Limitations and Future Research
The current study had several limitations. First, the study did not include analysis of pharmacological treatment. The authors had limited ability to assess the indication, type, strength, and duration of beta-blockers or angiotensin-converting enzyme inhibitors/angiotensin receptor blockers. The next step in future analyses will be to include pharmacological treatment information. Second, recall bias or interobserver bias may have occurred, as cognitive impairment and previous myocardial infarction data were self-reported or historically reported from EHRs rather than from bedside cognitive assessment or angiographic evidence. Interobserver variation between nurse reports of delirium and Katz ADL Index scores should be validated in future studies. Finally, demographic and clinical data were limited and did not fully correspond to other risk factors from the literature of known predictors of 30-day rehospitalization. Such risk factors include chronic pulmonary disease, anemia, health insurance type, and previous hospitalization within 6 months.
In addition to demographic and cardiovascular risk factors, geriatric syndromes such as delirium and physical function impairment were associated with 30-day hospital readmission. Discharge to home health care may reduce early readmission in high-risk patients.
- American Geriatrics Society 2015 Beers Criteria Update Expert Panel. (2015). American Geriatrics Society updated Beers criteria for potentially inappropriate medication use in older adults. Journal of the American Geriatrics Society, 63, 2227–2246. doi:10.1111/jgs.13702 [CrossRef]
- Buntin, M.B., Burke, M.F., Hoaglin, M.C. & Blumenthal, D. (2011). The benefits of health information technology: A review of the recent literature shows predominantly positive results. Health Affairs, 30, 464–471. doi:10.1377/hlthaff.2011.0178 [CrossRef]
- Buurman, B.M., van Munster, B.C., Korevaar, J.C., de Haan, R.J. & de Rooij, S.E. (2011). Variability in measuring (instrumental) activities of daily living functioning and functional decline in hospitalized older medical patients: A systematic review. Journal of Clinical Epidemiology, 64, 619–627. doi:10.1016/j.jclinepi.2010.07.005 [CrossRef]
- Dharmarajan, K., Hsieh, A., Dreyer, R.P., Welsh, J., Quin, L. & Krumholz, M.H. (2017). Relationship between age and trajectories of rehospitalization risk in older adults. Journal of the American Geriatrics Society, 65, 421–426. doi:10.1111/jgs.14583 [CrossRef]
- Donzé, J.D., Williams, M.V., Robinson, E.J., Zimlichman, E., Aujesky, D., Vasilevskis, E.E. & Schnipper, J.L. (2016). International validity of the HOSPITAL score to predict 30-day potentially avoidable hospital readmissions. JAMA Internal Medicine, 176, 496–502. doi:10.1001/jamainternmed.2015.8462 [CrossRef]
- Fortinsky, R.H., Madigan, E.A., Sheehan, T.J., Tullai-McGuinness, S. & Kleppinger, A. (2014). Risk factors for hospitalization in a national sample of Medicare home health care patients. Journal of Applied Gerontology, 33, 474–493. doi:10.1177/0733464812454007 [CrossRef]
- Hoyer, E.H., Needham, D.M., Miller, J., Deutschendorf, A., Friedman, M. & Brotman, D.J. (2013). Functional status impairment is associated with unplanned readmissions. Archives of Physical Medicine and Rehabilitation, 94, 1951–1958. doi:10.1016/j.apmr.2013.05.028 [CrossRef]
- Huynh, Q.L., Negishi, K., Blizzard, L., Saito, M., De Pasquale, C.G., Hare, J.L. & Marwick, T.H. (2016). Mild cognitive impairment predicts death and readmission within 30 days of discharge for heart failure. International Journal of Cardiology, 221, 212–217. doi:10.1016/j.ijcard.2016.07.074 [CrossRef]
- Hubbard, A.E., Ahern, J., Fleischer, N.L., Van der Laan, M., Lippman, S.A., Jewell, N. & Satariano, W.A. (2010). To GEE or not to GEE: Comparing population average and mixed models for estimating the associations between neighborhood risk factors and health. Epidemiology, 21, 467–474. doi:10.1097/EDE.0b013e3181caeb90 [CrossRef]
- Inouye, S.K., van Dyck, C.H., Alessi, C.A., Balkin, S., Siegal, A.P. & Horwitz, R.I. (1990). Clarifying confusion: The confusion assessment method. A new method for detection of delirium. Annals of Internal Medicine, 113, 941–948. doi:10.7326/0003-4819-113-12-941 [CrossRef]
- Jhaveri, R.R., Reynolds, H.R., Katz, S.D., Jeger, R., Zinka, E., Forman, S.A. & Hochman, J.S. (2012). Heart failure in post-MI patients with persistent IRA occlusion: Prevalence, risk factors, and the long-term effect of PCI in the Occluded Artery Trial. Journal of Cardiac Failure, 18, 813–821. doi:10.1016/j.cardfail.2012.10.012 [CrossRef]
- Joynt, K.E., Orav, E.J. & Jha, A.K. (2011). Thirty-day readmission rates for Medicare beneficiaries by race and site of care. JAMA, 305, 675–681. doi:10.1001/jama.2011.123 [CrossRef]
- Kansagara, D., Englander, H., Salanitro, A., Kagen, D., Theobald, C., Freeman, M. & Kripalani, S. (2011). Risk prediction models for hospital readmission: A systematic review. JAMA, 306, 1688–1698. doi:10.1001/jama.2011.1515 [CrossRef]
- Krumholz, H.M., Chen, Y.T., Wang, Y., Vaccarino, V., Radford, M.J. & Horwitz, R.I. (2000). Predictors of readmission among elderly survivors of admission with heart failure. American Heart Journal, 139, 72–77. doi:10.1016/S0002-8703(00)90311-9 [CrossRef]
- Malone, M.L., Vollbrecht, M., Stephenson, J., Burke, L., Pagel, P. & Goodwin, J.S. (2010). Acute Care for Elders (ACE) tracker and e-Geriatrician: Methods to disseminate ACE concepts to hospitals with no geriatricians on staff. Journal of the American Geriatrics Society, 58, 161–167. doi:10.1111/j.1532-5415.2009.02624.x [CrossRef]
- Malone, M.L., Yoo, J.W. & Goodwin, J.S. (2014). An introduction to the acute care for elders. In Malone, M.L., Capezuti, E. & Palmer, R.M. (Eds.), Acute care for elders: A model for interdisciplinary care (pp. 3–5). New York, NY: Springer Science+Business Media.
- Medicare Payment Advisory Commission. (2016). Report to Congress: Medicare payment policy; Chapter 3: Hospital inpatient and out-patient services (p. 460). Washington, DC: Centers for Medicare & Medicaid Services.
- Mentz, R.J., Allen, B.D., Kwasny, M.J., Konstam, M.A., Udelson, J.E., Ambrosy, A.P. & Gheorghiade, M. (2013). Influence of documented history of coronary artery disease on outcomes in patients admitted for worsening heart failure with reduced ejection fraction in the EVEREST trial. European Journal of Heart Failure, 15, 61–68. doi:10.1093/eurjhf/hfs139 [CrossRef]
- Mozaffarian, D., Benjamin, E.J., Go, A.S., Arnett, D.K., Blaha, M.J., Cushman, M. & Turner, M.B. (2016). Heart disease and stroke statistics-2016 update: A report from the American Heart Association. Circulation, 133, e38–e360. doi:10.1161/CIR.0000000000000350 [CrossRef]
- O'Connor, M., Hanlon, A., Naylor, M.D. & Bowles, K.H. (2015). The impact of home health length of stay and number of skilled nursing visits on hospitalization among Medicare-reimbursed skilled home health beneficiaries. Research in Nursing & Health, 38, 257–267. doi:10.1002/nur.21665 [CrossRef]
- Patel, A., Parikh, R., Howell, E.H., Hsich, E., Landers, S.H. & Gorodeski, E.Z. (2015). Mini-cog performance: Novel marker of post discharge risk among patients hospitalized for heart failure. Circulation Heart Failure, 8, 8–16. doi:10.1161/CIRCHEARTFAILURE.114.001438 [CrossRef]
- Pedersen, M.K., Meyer, G. & Uhrenfeldt, L. (2017). Risk factors for acute care hospital readmission in older persons in western countries: A systematic review. JBI Database of Systematic Reviews & Implementation Reports, 15, 454–485. doi:10.11124/JBISRIR-2016-003267 [CrossRef]
- Rahimi, K., Bennett, D., Conrad, N., Williams, T.M., Basu, J., Dwight, J. & MacMahon, S. (2014). Risk prediction in patients with heart failure: A systematic review and analysis. JACC. Heart Failure, 2, 440–446. doi:10.1016/j.jchf.2014.04.008 [CrossRef]
- Saito, M., Negishi, K. & Marwick, T.H. (2016). Meta-analysis of risks for short-term readmission in patients with heart failure. American Journal of Cardiology, 117, 626–632. doi:10.1016/j.amjcard.2015.11.048 [CrossRef]
- Schwartz, J., Strait, K.M., Keshawarz, A., Vellanky, S.S., Reilly, E.M. & Curtis, J.P. (2014). Medicare hospital quality chart book: Performance report on outcome measures (p. 105). New Haven, CT: Yale University.
- Smith, D.H., Thorp, M.L., Gurwitz, J.H., McManus, D.D., Goldberg, R.J., Allen, L.A. & Go, A.S. (2013). Chronic kidney disease and outcomes in heart failure with preserved versus reduced ejection fraction: The Cardiovascular Research Network PRESERVE Study. Circulation: Cardiovascular Quality and Outcomes, 6, 333–342. doi:10.1161/CIRCOUTCOMES.113.000221 [CrossRef]
- Tabachnick, B.G. & Fidell, L.S. (2006). Using multivariate statistics (5th ed.). Needham Heights, MA: Allyn & Bacon.
- Uthamalingam, S., Gurm, G.S., Daley, M., Flynn, J. & Capodilupo, R. (2011). Usefulness of acute delirium as a predictor of adverse outcomes in patients >65 years of age with acute decompensated heart failure. American Journal of Cardiology, 108, 402–408. doi:10.1016/j.amjcard.2011.03.059 [CrossRef]
- Wasfy, J.H., Zigler, C.M., Choirat, C., Wang, Y., Dominici, F. & Yeh, R.W. (2017). Readmission rates after passage of the hospital readmissions reduction program: A pre-post analysis. Annals of Internal Medicine, 166, 324–331. doi:10.7326/M16-0185 [CrossRef]
- Yoo, J.W., Seol, H., Kim, S.J., Yang, J.M., Ryu, W.S., Min, T.D. & Kim, S. (2014). Effects of hospitalist-directed interdisciplinary medicine floor service on hospital outcomes for seniors with acute medical illness. Geriatrics Gerontology International, 71, 71–77. doi:10.1111/ggi.12056 [CrossRef]
Participant Characteristics (N= 2,279)
|Variable||n(%)||p Valuea (All-Cause/HF or Other)|
|Total||All-Cause Readmission||HF Readmission||Readmission Other Than HF||No Readmission|
|Number of patients||2,279 (100)||508 (22.3)||223 (9.8)||285 (12.5)||1,771 (77.7)|
| Age (years)||<0.001/<0.001|
| 65 to 74||919 (40.3)||188 (37.0)||731 (41.8)||80 (35.9)||108 (37.9)|
| 75 to 84||813 (35.7)||201 (39.6)||612 (34.0)||87 (39.0)||114 (40.0)|
| ≥85||547 (24.0)||119 (23.4)||428 (24.2)||56 (25.1)||63 (22.1)|
| Male||1,084 (47.6)||238 (46.9)||846 (47.7)||105 (47.0)||133 (46.7)||0.714/0.579|
| White||1,406 (61.7)||279 (54.9)||1,127 (63.6)||125 (56.1)||154 (54.0)|
| Other||873 (38.3)||229 (45.1)||644 (36.4)||98 (43.9)||131 (46.0)|
| Up to middle school||724 (31.8)||166 (32.7)||558 (31.5)||69 (30.1)||97 (34.0)|
| High school or higher||1,555 (68.2)||342 (67.3)||1,213 (58.5)||154 (59.9)||188 (66.0)|
| Cognitive impairment||496 (21.8)||110 (21.7)||386 (21.8)||46 (20.6)||64 (22.5)||0.521/0.339|
| Delirium||542 (23.8)||154 (30.3)||388 (21.9)||71 (31.8)||83 (29.1)||<0.001/<0.001|
| Potentially inappropriate medication use||993 (43.6)||215 (43.3)||778 (43.9)||100 (44.8)||115 (40.4)||0.682/0.513|
| Katz ADL Indexb||<0.001/<0.001|
| 0 to 5||427 (18.7)||116 (22.8)||311 (17.6)||53 (23.8)||63 (22.1)|
| 6 to 9||825 (36.2)||180 (35.5)||645 (36.4)||80 (36.0)||100 (35.1)|
| 10 to 12||1,027 (45.1)||212 (41.7)||815 (46.0)||90 (40.3)||122 (42.8)|
| Ejection fraction < 40%||945 (41.5)||235 (46.2)||710 (30.1)||109 (49.0)||126 (44.2)||<0.001/<0.001|
| Advanced CKD||661 (29.0)||176 (34.6)||485 (27.4)||70 (31.4)||106 (37.2)||<0.001/<0.001|
| Stage 3||353 (14.9)||92 (18.1)||261 (14.7)||36 (16.1)||56 (19.6)|
| Stage 4 or 5||308 (13.5)||84 (16.5)||224 (12.6)||34 (15.2)||50 (17.5)|
| Previous myocardial infarction||1,010 (44.4)||251 (49.4)||759 (42.9)||117 (52.4)||134 (47.0)||0.044/0.026|
| Discharge disposition|
| Discharge home, no home health||1,252 (54.9)||258 (50.8)||994 (56.1)||116 (52.0)||115 (40.4)||0.054/<0.001|
| Home health referral||540 (23.7)||139 (27.4)||401 (22.7)||58 (26.0)||98 (34.4)|
| SNF||487 (21.4)||111 (21.8)||376 (21.2)||49 (22.0)||72 (25.2)|
Multilevel Multivariate Regressions of Hospital Readmission
|Variable||Odds Ratio [95% Confidence Interval]||p Value||Odds Ratio [95% Confidence Interval]||p Value|
| Age [reference = 65 to 74]|
| 75 to 84||2.01 [1.06, 6.85]||0.002||2.25 [1.13, 7.04]||<0.001|
| ≥85||4.72 [1.28, 11.07]||<0.001||2.49 [1.07, 8.49]||<0.001|
| Male||1.06 [0.58, 3.30]||0.647||1.13 [0.75, 2.37]||0.219|
| Race, Other [reference = White]||2.26 [1.07,3.98]||<0.001||2.73 [1.20, 4.61]||<0.001|
| Education [reference = high school or higher]||0.94 [0.61, 3.02]||0.508||1.50 [0.86, 4.39]||0.162|
| Cognitive impairment||1.19 [0.74, 3.29]||0.264||1.14 [0.70, 5.02]||0.411|
| Delirium||2.54 [1.19, 4.78]||<0.001||3.48 [1.27, 6.35]||<0.001|
| Potentially inappropriate medication use||1.11 [0.63, 3.60]||0.407||0.90 [0.58, 3.14]||0.383|
| Katz ADL Index [reference = score 10 to 12]|
| 0 to 5||4.60 [1.32, 9.03]||<0.001||3.52 [1.21, 8.29]||<0.001|
| 6 to 9||3.37 [1.20, 11.28]||<0.001||3.24 [1.17, 10.75]||<0.001|
| Ejection fraction < 40||1.71 [1.18, 3.46]||0.005||2.27 [1.13, 8.31]||<0.001|
| Advanced CKD|
| Stage 3||1.54 [1.05, 3.62]||0.013||2.73 [1.40, 4.36]||<0.001|
| Stage 4 or 5||2.06 [1.07, 5.73]||0.004||3.30 [1.28, 7.10]||<0.001|
| Previous myocardial infarction||3.03 [1.48, 4.47]||<0.001||3.57 [1.72, 5.94]||<0.001|
| Discharge disposition [reference = discharge home, no home health]|
| Home health referral||0.64 [0.15, 0.97]||0.006||0.41 [0.20, 0.96]||<0.001|
| SNF||2.75 [1.04, 8.02]||0.029||0.82 [0.24, 6.40]||0.137|