American Hospital association september 2011 TrendWatch Examining the Drivers of Readmissions and Reducing Unnecessary Readmissions for Better Patient Care N early one-fifth of Medicare benefi- ciaries—roughly 2 million beneficiaries per year1—discharged from Rates of readmission occurring for any reason following hospitalization for one of three common conditions­ heart — all of which are crucial to appropriate risk adjustment of readmission rates—is still not fully understood. a hospital return within 30 days, accord- attack, heart failure, and pneumonia—are America’s hospitals are committed to ing to the Medicare Payment Advisory displayed.3 Most recently, in the Patient improving the safety and quality of care commission (MedPAC).2 Some of the Protection and Affordable Care Act (ACA), they deliver, and many are already work- readmissions are planned, some are Congress enacted the Hospital Readmis- ing to reduce avoidable readmissions. unplanned and others are unrelated to the sions Reduction Program (HRRP) under Innovative programs focus on improving initial reason the patient came to the which Medicare will penalize hospitals care transitions, bolstering post- hospital. Identifying and reducing avoid- for higher-than-expected rates of readmis- discharge monitoring and follow-up able readmissions will improve patient sions beginning in FY 2013.4 care, and strengthening linkages with safety, enhance quality of care, and lower Careful planning is warranted to ensure other community providers. Payment health care spending. That is why pol- that the HRRP achieves its dual aims rules should encourage hospitals to icymakers, consumers, hospital leaders of improving quality and reducing costs. invest in programs proven effective, and and the medical community are focused There are opportunities to achieve cost should avoid unintended adverse conse- increasingly on readmissions to hospitals. savings by reducing readmissions, but quences for other aspects of patient care. Policymakers are proposing incen- not all readmissions can or should be This TrendWatch examines recent tives to reduce hospital readmissions by avoided. Additionally, as CMS proceeds research on hospital readmissions, includ- publicly posting data on readmission rates with the HRRP, evidence is mounting ing the linkages between readmissions and and lowering payments to hospitals with that the link between readmissions and quality of care, and the various circum- high rates. First, in 2009, hospitals began quality of care is more complex than stances that may drive readmissions. It also voluntarily reporting hospital readmis- assumed. Further, the role of other fac- discusses the changes put in place by the sion rates to the Centers for Medicare tors—such as a patient’s demographic and ACA and highlights the considerations and & Medicaid Services (CMS) for public socioeconomic characteristics, social sup- additional research that are warranted as review on its website, Hospital Compare. port structure, and co-morbid conditions, policymakers implement the new HRRP . “  Patients who have chronic disease like heart failure are a vulnerable group. At the hospital, they receive 24-hour monitoring, so if there is any change, the doctors “” from the field and nurses can respond immediately. Yet when they are discharged home, they are pretty much on their own.” Sarwat I. Chaudhry, M.D., assistant professor, Yale School of Medicine34 Classification of Readmissions Can Help Identify Targets for Reduction The AHA, in consultation with clinicians, Unplanned readmissions related to the initial stay likely offer has developed a framework to help the best opportunity for savings and care improvements. policymakers and providers consider the different types of readmissions. Chart 1: A Framework for Classification of Readmissions While some readmissions might have been avoided if the patient received the right care at the right time, still others Related to Initial Unrelated to Initial may be unavoidable due to the natural Admission Admission progression of disease, accepted treatment protocol, or a patient’s preferences. Some A planned readmission A planned readmission readmissions are part of a planned course for which the reason for for which the reason for of treatment. The framework can aid Planned readmission is related readmission is not policymakers in designing a program for Readmission to the reason for the related to the reason for reducing readmissions that targets those initial admission. the initial admission. rehospitalizations that are less desirable and potentially avoidable. The framework contemplates four An unplanned An unplanned distinct types of readmissions: readmission for readmission for Unplanned which the reason for which the reason for 1) A planned readmission related to the readmission is related readmission is not Readmission initial admission, such as a series of che- to the reason for the related to the reason motherapy treatments or reconstructive initial admission. for the initial admission. surgery following removal of a body part. Source: American Hospital Association. 2) A planned readmission unrelated to the initial admission, such as readmission for removal of a lung tumor discovered the initial admission—on which AHA days of myocardial infarction for pre- during an admission for a heart attack. seeks to focus public policy efforts to vention of sudden cardiac death.5 In 3) An unplanned readmission unrelated reduce readmissions. Hospitals cannot this case, a patient admitted for a heart to the initial admission, such as read- influence the occurrence of unplanned, attack would have to be readmitted later mission for a fracture sustained in a car unrelated readmissions because they are for implantation of the ICD. Similarly, accident following an initial stay for an not predictable or preventable. if a patient experienced a significant appendectomy. Likewise, hospitals ought not to be side effect during outpatient admin- expected to eliminate planned read- istration of chemotherapy, then the 4) An unplanned readmission related to missions, as these are typically part of patient would typically be admitted the initial admission, such as readmission clinically appropriate treatment plans. for inpatient administration for all for a surgical site infection or adverse reac- For example, clinical guidelines for subsequent courses of chemotherapy.6 tion to a medication. (Chart 1) implantation of an implantable cardiac These repeat admissions should not be It is this last group of readmis- defibrillator (ICD) do not recommend targeted as they are markers of sions—those unplanned but related to implantation of an ICD within 40 appropriate care. “” from the field “  I think that the message to patients and the general public is that they should be wary of seemingly simple measures of quality of care. One simple measure is not enough.” Eiran Z. Gorodeski, M.D., researcher and associate staff member in cardiovascular medicine, Cleveland Clinic, OH20 2 TRENDWATCH The ACA Hospital Readmissions Reduction Program The HRRP creates an incentive to reduce hospital readmissions by lowering Risk-adjusted readmission rates do not account for Medicare payment rates to hospitals some factors that may influence risk of readmission. showing greater-than-expected, or “excess,” readmissions. Beginning in Chart 2: Risk Adjustment Variables for 30-­ ay All-­ ause Risk Standardized d c FY 2013, 30-day readmission rates for Readmission Rate Following Pneumonia Hospitalization a hospital’s Medicare patients with heart attack (AMI), heart failure and pneu- Included in Risk Adjustment Not Included in Risk Adjustment monia will be compared to the expected Age Medicare eligibility status (e.g., aged, disabled) rate of readmissions, using risk adjust- Gender Dual eligibility (Medicaid) status or income ment to account for age, gender, medical History of CABG Frailty diagnosis and selected medical history. Condition Categories including: Social support structure In FY 2009, 2.5 percent of discharges History of infection Septicemia/shock Race or ethnicity from hospitals paid under the inpatient prospective payment system and included Cancer Diabetes Geographic region in Hospital Compare had a principal Hematological Gastrointestinal Limited English proficiency diagnosis of AMI, 5.7 percent had heart disorders disorders failure, and 4.2 percent had pneumonia.7 Malnutrition Dementia & senility The rate of excess readmissions for these Drug/alcohol abuse Psychiatric disorders three conditions will translate to a pay- Paraplegia, CHF & other ment reduction for the hospital for each paralysis, et al. heart disease Medicare admission, capped at 1 percent Stroke & vascular COPD & lung of Medicare payments in FY 2013 and disease disorders at 3 percent by FY 2015. The HHS Asthma Pneumonia Secretary may expand HRRP to additional ESRD or dialysis Renal failure Source: National Quality Forum. Measure # 0506. www.qualityforum.org. conditions beginning in FY 2015.8 Urinary tract infection Skin ulcers Note: CABG=coronary artery bypass graft; ESRD=end The Obama administration has also Vertebral fractures Other injuries stage renal disease; CHF=congestive heart failure; and COPD=chronic obstructive pulmonary disease. launched the Partnership for Patients: Better Care Lower Costs, a new public- private partnership that will help improve the risk adjustment mechanisms account Other challenges in risk adjustment the quality, safety and affordability of for these factors so that hospitals treat- exist around race, ethnicity and limited health care for all Americans.9 This pro- ing a more complex patient mix are not English proficiency. (Chart 2) gram will build on the Community-based unduly penalized for readmissions. The ACA specifies that certain read- Care Transition Program, a five-year However, the risk adjusters proposed for missions will be excluded from hospitals’ program created by the ACA in which use in the HRRP are imperfect; while readmission rates. These exclusions hospitals and community-based organiza- there are important factors that will be include “readmissions that are unrelated tions will work together to improve care nearly impossible to measure and account to the prior discharge (such as a planned transitions, including post-discharge for, such as social support structure, the readmission or transfer to another follow-up, and thus aim to reduce risk adjusters also fail to capture certain applicable hospital).”13 This language readmissions for high-risk Medicare important known factors, such as whether is consistent with AHA’s readmissions beneficiaries.10 Together these programs a beneficiary has “dual eligible” status. framework. However, the statute also will receive $1 billion in funding.11 Dual eligible beneficiaries are those who reflects the difficulty in distinguishing Many factors contribute to a hospital’s qualify for both Medicare and Medicaid. each of the four types of readmissions, as readmission rate, including patients’ These 9 million beneficiaries are the most it specifically addresses only one of the socioeconomic status, demographic fac- chronically ill in both programs and have four types discussed above. How these tors, co-morbid conditions, and access health care costs that are nearly five times exclusions are accounted for in regula- to social supports. It is important that those of other Medicare beneficiaries.12 tion remains a key policy concern. 3 Calculating the ACA Payment Reduction for Excess Readmissions The hospital readmissions reduction •  ,000 total admissions for the three 1 (0.02) * $5,000 * 50 expected read- program in the ACA directs the conditions missions = $5,000 Medicare program to recoup payments •  5,000 average base diagnosis-related $ However, the technical error in the made to hospitals for “excess” readmis- group (DRG) payment ACA’s language results in this hospital’s sions for patients with heart attack, payments being reduced much more • 51 actual readmissions heart failure or pneumonia by reducing than $5,000. According to the legisla- payment for each Medicare admission • 50 expected readmissions tive language, the excess readmissions moving forward. In this example, the hospital had ratio (0.02) and average DRG payment However, the AHA has identified a one more readmission than expected, ($5,000) would be multiplied by the technical error in the ACA language or one “excess” readmission. Since the hospital’s total number of admissions that leads to an overstatement of the average base DRG payment for the for the three conditions (1,000) for a amount of money hospitals must pay hospital is $5,000, its payments should penalty of $100,000. The calculation back to the Medicare program. As be reduced by $5,000—the payment mandated by the statute is: shown in the example, the formula amount for the one excess readmission. (0.02) * $5,000 * 1,000 total was intended to use a hospital’s num- To arrive at this amount, the calcu- admissions = $100,000 ber of expected readmissions for each lation is made by first determining condition as the basis for the calcula- Because the formula uses the total the ratio of excess readmissions by tion in order to calculate the payments number of admissions for the three subtracting 1 from the ratio of actual to associated with excess readmissions. conditions, the penalty for this hospital expected readmissions (((51/50) – 1) = Instead, the formula in the statute uses will be 20 times the amount of pay- 0.02). Next, that factor (0.02) is a hospital’s total number of admissions ment for excess readmissions. The AHA multiplied by the average DRG pay- for the conditions. This error, if left believes Congress intended to recoup ment ($5,000), and the number of unresolved, will inappropriately inflate money paid out for excess readmissions, expected readmissions (50) for a hospitals’ payment reductions. which the current formula fails to do. penalty of $5,000, correctly repre- For example, consider a hospital Therefore, the error in the legislative senting the dollar amount associated with the following statistics: language should be clarified to reflect with the “excess” readmissions. “expected readmissions,” instead of the total number of admissions. The Relationship Between Readmissions and Quality of Care Is Complex In the AHA framework, only one type indicate lower quality care. However, 30-day mortality rates.14 of readmission—a readmission that is growing evidence reveals that mortality A similar analysis using Hospital unplanned and related to the initial and readmissions may in fact be inversely Compare data conducted by the Greater admission—could indicate a lapse in associated with one another, calling into New York Hospital Association also con- providing the right care at the right time question the assumption that low read- cluded that mortality is inversely related within the hospital. Yet, payers and others mission rates are always desirable. to readmissions.15 (Chart 3) And com- commonly use global readmission rates, Researchers at Cleveland Clinic (OH) parison of better- and worse-than-expect- along with mortality rates, as indicators recently used Medicare data posted on the ed hospital referral regions (HRRs)16 with of the quality of care delivered to patients Hospital Compare website to study the link respect to 30-day mortality and readmis- during a hospital stay. between readmissions and mortality for sions shows that some hospitals with Conventional wisdom is that higher beneficiaries admitted for heart failure. better-than-expected mortality actually rates of either mortality or readmission They found that hospitals with higher had worse-than-expected readmissions.17 within 30 days of hospital discharge readmission rates actually had lower Another analysis compared hospitals 4 TRENDWATCH There appears to be an inverse relationship between mortality and readmissions. Chart 3: Percentile Rankings of Adjusted Mortality and 30-day Readmission Rates, States with the Lowest and Highest Adjusted Mortality Rates 100% Mortality Readmissions 90% 80% 70% Percentile Ranking 60% 50% 40% 30% 20% 10% 0% MA CT DC DE MN NJ IL OH MI PA CO OK NE MS WY WA NV NM NH OR ID AR VT Low Mortality States High Mortality States Source: Analysis by Greater New York Hospital Association, 2009. Note: Findings based on CMS’ Hospital Compare data released on July 7, 2009. Chart shows 11 states with lowest and 12 states with highest adjusted mortality rates. MD was omitted from the low mortality states because readmissions data were not available. ranked among the top 50 “America’s Best hospitals that performed better than measuring access, quality, costs, equity Hospitals” in cardiac surgery by U.S. News expected on both mortality and and the potential for children to lead & World Report and those not ranked. For readmissions.18 healthy lives. The likelihood of read- patients treated for heart failure, the study Similarly, a study of 39 children’s mission in the year following discharge found lower rates of mortality among hospitals in 24 states examined the rela- rose as the states’ performance ranking ranked hospitals but no difference in tionship between readmissions and the improved. States with the highest-per- readmission rates between ranked and state’s child health system performance, forming systems had significantly higher non-ranked hospitals. Interestingly, among as ranked by The Commonwealth Fund. readmission rates than states with the the top-ranked hospitals, there were no The ranking is based on 13 indicators lowest-performing systems.19 Readmission Rates May Be Ill Suited as Measures of Quality Experts do not all agree that readmis- readmissions are within or beyond of care that cloud the picture. For sion rates are valid as quality measures. the hospital’s control. For instance, example, patients hospitalized for A review of the literature on hospital patients may be readmitted as a result heart failure in the Veterans Affairs (VA) readmissions concluded that readmission of unpreventable progression of disease, Health Care System became sicker over rates are not useful indicators of quality planned follow-­ p care, their own u a four-­ ear study period, with greater y of care.21 In part, these measures do not preference for treatment timing, or an co-­ orbidities, yet mortality rates m typically distinguish among the different unrelated diagnosis or trauma. declined, in part due to more frequent types of readmissions arrayed in Chart Additionally, there are multiple factors use of recommended therapies.22 At the 1­­ — meaning they do not consider which such as patient characteristics and patterns same time, readmission rates climbed, 5 perhaps because sicker patients who care is that a readmission is an indicator evidence linking readmissions with the might have otherwise died remained alive, of an omission of needed care, or an error care patients received during the initial and thus were at greater risk for rehos- in the care given to a patient. Yet, as the hospital stay.24 Another study found that pitalization. Further, providers reported AHA framework describes, this theory is patients who were readmitted within 28 heightened monitoring of discharged unsupported for readmissions that are a days of discharge were no more likely patients, a factor that may have resulted in specified step in a treatment plan, and for to have received low-­ uality care than q more readmissions but saved more lives.23 unrelated, unplanned readmissions that patients without a readmission.25 One argument in favor of using could not have been anticipated. readmission rates to measure quality of A review of the literature found mixed Patient Characteristics and Health Conditions Play an Important Role in Readmissions A patient’s life circumstances (low-­ncome i Chronic Conditions Additional research has focused on or lack of social support) and individual Patients with co-­ orbidities are at height- m identifying the specific conditions that characteristics (co-­ orbid conditions m ened risk of rehospitalization. (Chart 4) are predictive of readmissions. A study or underlying disability) are all impor- An analysis of adults hospitalized in six of more than 6,800 general medicine tant factors in whether a patient will be states found that, the higher the num- patients in a large urban, university readmitted to the hospital. The risk ber of chronic conditions a patient had, medical center found six co-­ orbidities m adjustment methods used to calculate the greater the chance of readmission.27 to be associated with readmission: readmission rates for beneficiaries with Similarly, the likelihood of readmission congestive heart failure, renal disease, heart attack, heart failure or pneumonia was greater for patients with a higher cancer (with and without metastasis), do not account for all of these factors. severity of illness score.28 Another study weight loss and iron deficiency anemia.30 Thus, new financial penalties on hospitals of 37 U.S. children’s hospitals also found MedPAC also found that the readmis- may not make appropriate accommoda- that a higher rate of readmissions was sion rates for Medicare beneficiaries with tion for patients’ life circumstances that associated with higher prevalence of use end-­ tage renal disease (ESRD) are higher s could drive readmissions. of assistive technology such as a gastron- than average—31.6 percent of ESRD omy tube or cerebrospinal fluid shunt.29 patients are readmitted within 30 days The more chronic conditions a patient has, the greater likelihood of readmission. Chart 4: 30-­ ay Readmission Rate for Non-­ ual, Disabled Medicaid Beneficiaries d d by Number of Chronic Illness and Disability Categories 40% 35% 30-day Readmission Rate 30% 25% 20% 15% 10% 5% 0% New 0 1 2 3 4 5 6 7 8 9 10+ Enrollee Number of Categories Source: Gilmer T. and Hamblin, A. (December 2010). Hospital Readmissions among Medicaid Beneficiaries with Disabilities: Identifying Targets of Opportunity. New Jersey: Center for Health Care Strategies. Note: Number of chronic illnesses and disabilities measured using Chronic Illness and Disability Payment System (CDPS), a risk adjustment model used to adjust capitated payments to health plans that enroll Medicaid beneficiaries. Study included 941,208 Medicaid beneficiaries hospitalized from 2003-2005 in 50 states and DC. 6 TRENDWATCH as compared to only 16.9 percent of non-­ Medicaid beneficiaries are consistently at greater risk ESRD patients.31 of readmission than privately insured adults. Other studies have focused on patients admitted for a particular condition or Chart 5: Non-obstetric, Adult 30-day Readmission Rates by procedure. One such study found that Insurance Coverage and Number of Co-morbidities, 2007 patients readmitted after being hospital- ized with heart failure—one of the targets 14.2% of the HRRP—were more likely to have Medicaid Private Insurance diabetes, peripheral vascular disease, 11.3% stroke and coronary artery disease than 9.9% Readmission Rate 9.7% their counterparts who weren’t readmit- 8.2% 7.1% ted.32 An analysis of readmissions for 5.8% complications of coronary artery bypass 4.7% graft (CABG) surgery had similar conclu- sions. Six co-­ orbid conditions were m associated with readmission within 30 days of discharge: vascular disease, con- 0 1 2 3+ gestive heart failure, chronic obstructive Number of Co-morbidities pulmonary disease (COPD), diabetes, Source: Jiang, H.J., and Wier, L.M. (April 2010). All-cause Hospital Readmissions among Non-elderly Medicaid Patients, 2007. HCUP liver failure and dialysis.33 Statistical Brief #89. Rockville, MD: Agency for Healthcare Research and Quality. Depression Psychological conditions, such as socioeconomic factors have a significant minority serving hospitals.” The article depression, can adversely affect patient effect on readmissions.42 concludes that racial disparities in read- recovery and increase the likelihood of Race and ethnicity have been shown missions are related to both race and site readmission. Following hospitalization to be a predictor of readmissions but how of care, recognizing there also are factors for coronary artery disease, “distressed” this plays out is not completely under- beyond hospitals’ control that could patients were found to have significantly stood. One study of general medicine explain the findings.46 higher rates of readmission within six patients in a large urban, university medi- Language barriers lead to greater risk months than “non-­ istressed” patients.35 d cal center found that African American of readmission because patients and their Similarly, a Canadian study found that patients had a higher risk of readmission families are less likely to understand their heart attack patients who were depressed than patients of any other race.43 A dif- diagnosis or discharge instructions. were more likely to be readmitted in the ferent study of just Medicare beneficiaries Among more than 7,000 patients admit- year after discharge.36 The link between with heart failure found that minority ted to a northern California hospital, depression and readmission is not limited patients, other than African Americans, Latino and Chinese patients who did not to patients with cardiac conditions; a study had a higher risk of readmission.44 speak English were significantly more of 142 internal medicine patients at a hos- Another recent study explored whether likely to be readmitted within 30 days pital in Australia also found that depression disparities in readmission rates are attrib- than English speakers.47 predicted higher rates of readmission.37 utable primarily to race itself or to the site Income and socioeconomic status of care, given that care for minorities is also play a role. One study of Medicare Demographic Factors concentrated in a relatively small number patients found that patients discharged Numerous studies have demonstrated the of hospitals. Overall, black Medicare from hospitals in counties with low influence of patient characteristics such patients had higher readmission rates than median income had higher readmission as age, gender,38 race,39 geographic region40 whites,45 and patients from what the rates than those discharged in counties and Medicaid coverage41 on the risk of authors call “minority-­ erving hospitals” s with high median income.48 Another readmission. There is not yet consensus (hospitals in the top decile of proportion found that Medicare beneficiaries receiv- on the most important predictive factors, of black Medicare patients) had higher ing Supplemental Security Income (SSI) but it is evident that demographic and readmission rates than those from “non-­ were more likely to be rehospitalized.49 7 Public insurance coverage also appears The effect of socioeconomic factors raises questions associated with greater risk of readmis- about using readmissions to measure quality. sion.50 Using hospital data from 10 states, one study found that non-­ bstetric, adult o Chart 6: Community Characteristics and Hospital Quality Medicaid patients had higher readmission Measures for a Suburban and an Urban Community rates than their privately insured counter- parts, nearly 11 percent compared to only 6 percent.51 (Chart 5) Fairfield, CT Bronx, NY Hospitals serving disproportionate Community Characteristics numbers of minority, low-­ncome, or i Estimated Population 895,030 1,391,903 otherwise vulnerable patients may have Median Household Income $80,020 $34,031 higher readmission rates than other hos- Persons Below Poverty Line 7% 27% pitals because of the risk factors of their Non-Hispanic White Population 70% 13% patient population. This scenario is illus- trated in Chart 6 in a comparison of the No English Spoken at Home (aged 5) 24% 53% Bronx, NY, to Fairfield, CT, two com- Bachelor Degree or Higher (aged 25) 40% 15% munities with very different demographic Hospital Quality Data profiles located in close proximity to each other. The Bronx has a median household Hospitals in County with Quality Data on CMS Hospital Compare Site 6 7 income less than half that in Fairfield, Hospitals with HF Discharge Instruction Rate Better than the U.S. Average 3 6 nearly triple the minority population, and Hospitals with HF Readmission Rate Significantly Worse than U.S. Average 0 6 lower levels of educational attainment. Yet, Hospitals with HF Mortality Rate Significantly Worse than U.S. Average 0 0 nearly all of the hospitals in the Bronx Source: Bhalla, R., and Kalkut, G. (2010). Could Medicare Readmission Policy Exacerbate Health Care System Inequality? performed better than the U.S. average Annals of Internal Medicine, 152(2), 114-117. on giving discharge instructions to heart Note: HF=Heart Failure. failure patients, while only half of the hospitals in Fairfield performed the same. all had worse rates than the U.S. average, may further disadvantage their patients, One might then expect hospitals in the while none of the hospitals in Fairfield perhaps exacerbating health disparities. Bronx to have better readmission rates for were below the national average.52 (Chart 6) heart failure patients; but in fact, nearly Penalizing hospitals in these circumstances Hospitals Are Testing Innovative Approaches to Reduce Readmissions Even when considering unplanned, which patients are at greatest risk of read- ments in enhanced discharge planning related readmissions, it is not feasible for mission, and thus should receive the most and follow-­ p care can help reduce u a hospital to prevent all such readmissions, attention or be targeted for intervention. readmissions. For example, one study and too low a rate might actually indicate A review of five statistical models intended examined the impact of early follow-­ p u poor quality care.54 The best way to to predict patient risk of readmission care (within seven days) on readmis- improve quality of care as it relates to found little consistency among patient sions of patients with heart failure. The readmissions is to focus on those rehos- characteristics­­—including demographic authors found that patients with an pitalizations that may be avoidable. variables and co-­ orbid conditions­ that m — initial admission in a hospital in the The challenge is that there are varying are significant predictors. Further work 56 lowest quartile of performance on early methods, with no agreed-­ pon stan- u is warranted to help hospitals better focus follow-­ p care had a greater risk of read- u dards, for identifying such potentially their efforts on those patients most likely mission than patients initially admitted preventable readmissions.55 to benefit. to hospitals with higher rates of early There is also little agreement on the However, there are some promising follow-­up care.57 characteristics that might best predict findings indicating that hospitals’ invest- Already, hospitals are moving forward 8 TRENDWATCH with efforts to reduce readmissions and Hospital efforts to enhance discharge planning and improve quality of care. Hospitals are follow-up care can help reduce readmissions. testing varying approaches, including partnering with post-­ cute care providers a Chart 7: Cumulative Hazard Rate* of Hospital Utilization and enhancing discharge planning and for 30 Days Post Discharge, Patients Receiving follow-­ p services. (Chart 7) u Usual Care vs. Patients Receiving Discharge Intervention Metro Health Hospital in Wyoming, MI initiated its Congestive Heart Failure 0.4 (CHF) readmissions program in August Cumulative Hazard Rate of 2010 and—only six months into the 0.3 program—cut its CHF readmission rate in half, from 15.5 percent for the first and 0.2 second quarters of 2010 to 7.4 percent Usual Care Intervention for the third and fourth quarters of 2010. 0.1 Metro established a CHF unit staffed only by nurses with advanced training 0 in CHF care. The nurses on the CHF 0 5 10 15 20 25 30 floor developed education materials for Days After Discharge patients, including information on appropriate diet and self health care, and Source: Jack, B.W., et al. (2009). A Reengineered Hospital Discharge Program to Decrease Rehospitalization: A Randomized Trial. Annals of Internal Medicine, 150(3), 178-187. they review these materials with patients *The cumulative hazard rate is the cumulative number of hospital utilization events over total discharges over 30 days and illustrates during their stay in the hospital. Addition- how the risk of hospital utilization changes over time for each group. ally, clinical secretaries schedule primary Note: Hospital utilization is defined as a readmission or ED visit within 30 days. Intervention consisted of patient education, discharge planning, and follow-up phone call. care provider (PCP) appointments for Results statistically significant at p=0.004. each CHF patient before they leave the hospital and the PCP receives clear and concise information on the patient’s hospi- a hospital readmission and develop an care plans, identify any additional needs, talization, including what kind of follow-­ individualized care plan.58 and link patients to community services up care may be needed. Each patient is and providers to resolve any issues. scheduled to see his/her doctor within Rush University Medical Center in Social workers may continue to follow-­ seven days of discharge. Chicago, IL implemented its hospital-­ up for a week to a month, depending When patients require more than PCP wide Enhanced Discharge Planning on the patient’s needs. The majority of follow-­ p care, hospital case managers u Program (EDPP) randomized controlled the patients participating in the pro- enroll them in home health care before trial from June 2009 to March 2010 and gram require social worker intervention they leave the hospital. Case managers during that period patients participating post-­ ischarge. From June 2009 to d call each patient within 24-­ 8 hours of 4 in the program had 15 percent lower March 2010, EDPP identified problems discharge from the hospital to make sure 30-­ ay readmission rates, 24 percent d for 83 percent of program participants they are following discharge instructions lower 60-­ ay readmission rates and 23 d which did not emerge until after hospital and keeping their PCP appointments. percent lower 90-­ ay readmission rates d discharge for 74 percent of the patients. Metro planned to expand its CHF read- than patients not participating in the More than one follow-­ p call was needed u missions program to include pneumonia program. EDPP uses the hospital’s elec- for 254 of the 360, or 70.6 percent, of and acute myocardial infarction patients tronic medical record, which includes patients in this program.59 beginning in August 2011. Metro has clinical data as well as information on the also developed a high-­ isk readmission r patient’s social support structures, to The Agency for Healthcare Research and assessment tool based on information identify patients at risk for readmission. Quality is funding a project through the gathered from all patients upon admis- Social workers then telephone those AHA’s Health Research and Education sion which allows the hospital to identify patients after discharge to ensure they Trust to help hospitals adopt Project RED patients as moderate-­to high-­ isk for r are receiving the services detailed in their (Re-engineered Hospital Discharge), a 9 program developed by Boston University charge instructions, which are also shared understood the first time. Health First, that led to 30 percent fewer readmissions with the patient’s regular physician, and of Rockledge, FL, used Project RED to at Boston University Medical Center.60 follows up via phone within two days to reduce readmissions of congestive heart Hospitals use Project RED workflow identify and resolve any problems.61 failure patients in one unit by 29 percent, processes from the time a patient is Some hospitals are saving workforce compared to similar patients in units where admitted and throughout the stay to help resources by using Project RED’s “virtual Project RED was not deployed. Health the patient prepare for discharge. Each discharge advocate,” Louise, to deliver First also hired discharge advocates to assist patient is assigned a nurse “discharge discharge instructions to patients. Louise patients with home care and ensure they advocate” who assists the patient in speaks to the patient using the commu- see a doctor within seven days. One of the understanding his or her diagnosis, nication style of a nurse and her dialogue is noted barriers to wider use of these pro- arranges follow-­ p appointments, and u tailored for each patient based on the grams is the high level of provider invest- confirms medication plans. Upon information entered into a workstation. ment required to achieve savings that discharge, the discharge advocate provides Louise can answer questions and repeat accrues primarily to the health plans.62 each patient with a personalized set of dis- information the patient may not have Conclusion Readmissions can be categorized as plan- quality and save costs by reducing read- characteristics, including socioeconomic ned or unplanned, and related or missions, recent evidence suggests that factors, and hospitals’ particular circum- unrelated to the initial admission. Planned it is difficult to draw conclusions about stances and patient mix in determining readmissions and those unrelated to quality of patient care using data only financial penalties for readmissions. the initial admission either should not or on readmissions. Most confounding is Policymakers seeking statutory or cannot be prevented by hospitals. Thus, the evidence of an inverse relationship regulatory levers to reduce readmissions public policies should focus on readmis- between readmissions and mortality. should carefully weigh the potential sions that may be avoidable—those that Other factors, from patient character- for unintended adverse consequences. are unplanned and related to the initial istics and patients’ life circumstances Payment penalties intended to shrink read- admission, such as a surgical site infec- to the nature of post-­ ospital care, also h mission rates could exacerbate inequities tion. Additionally, the AHA believes seem to matter in determining the rate and leave hospitals with fewer resources to the technical error in the ACA should be of readmissions. make needed investments in improving clarified in order to recoup an appropri- More research is needed on the driv- patient care. Further, misaligned policies ate amount from hospitals with excess ers of readmissions. Such insight will could direct hospitals to reduce readmis- readmissions. be integral to developing risk adjusters sions that are appropriate for safe patient While the ACA aims to improve that appropriately account for patient care and may actually save lives. Policy Questions •  ow can the HRRP account for patients’ life circumstances H •  hat are the best approaches for disseminating informa- W and socioeconomic factors in calculating expected and actual tion about programs proven to be effective in reducing readmission rates? readmissions? •  ow can regulators best focus the ACA’s HRRP on H •  ow can regulators anticipate and avoid unintended H unplanned, related readmissions—those that hospitals are adverse consequences for patients and providers in imposing best able to prevent? financial penalties for excess readmissions? •  ow can policymakers effectively encourage hospitals and H •  hat additional research is warranted to ensure appropriate W other providers to continue to design and implement inno- risk adjustment of readmission rates for the HRRP? vative approaches to reduce readmissions? 10 Endnotes 1. Estimate derived from total inpatient volume (inpatient PPS discharges) obtained from of Medicare Data. Clinical Cardiology, 32(1), 47-52. Centers for Medicare & Medicaid Services. Medicare Inpatient Hospital Dashboard, 33. Hannan, E.L., et al. (2003). Predictors of Readmission for Complications of Coronary Inpatient Prospective Payment System. Data updated March 2011 for discharges from Artery Bypass Graft Surgery. JAMA, 290(6), 773-780. 10/2006-12/2010. 34. Krupa, C. (6 December 2010). No Benefit from Telemonitoring Heart Patients. American 2. Medicare Payment Advisory Commission. (June 2007). Payment Policy for Inpatient Medical News. http://www.ama-assn.org/amednews/2010/12/06/prsb1206.htm. Readmissions. Report to the Congress: Promoting Greater Efficiency in Medicare. 35. Allison, T.G., et al. (1995). Medical and Economic Costs of Psychologic Distress in Washington, DC. Patients with Coronary Artery Disease. Mayo Clinic Proceedings, 70, 734-742. 3. See: http://www.hospitalcompare.hhs.gov/. 36. Frasure-Smith, N., et al. (2000). Depression and Health-care Costs During the First Year 4. The Affordable Care Act is the combination of the Patient Protection and Affordable Care Following Myocardial Infarction. Journal of Psychosomatic Research, 48, 471-478. Act (PPACA), P.L. 111-148, enacted on March 23, 2010, and the Health Care and 37. Mudge, A.M., et al. (2010). Recurrent Readmissions in Medical Patients: A Prospective Education Reconciliation Act of 2010 (HCERA), P.L. 111-152, enacted on March 30, Study. Journal of Hospital Medicine, online. 2010. See: Section 3025. 38. Hannan, E.L., et al. (2003). Predictors of Readmission for Complications of Coronary 5. Steinbeck, G., et al. (2009). Defibrillator Implantation Early after Myocardial Infarction. Artery Bypass Graft Surgery. JAMA, 290(6), 773-780, and Jiang, H.J., and Wier, L.M. New England Journal of Medicine, 361(15), 1427-1436. (April 2010). All-cause Hospital Readmissions among Non-elderly Medicaid Patients, 6. Dollinger, M. (1996). Guidelines for Hospitalization for Chemotherapy. The Oncologist, 2007. HCUP Statistical Brief #89. Rockville, MD: Agency for Healthcare Research and 1, 107-111. Quality. 7. American Hospital Association analysis of the FY 2011 IPPS Impact File and Hospital 39. Allaudeen, N., et al. (2010). Redefining Readmission Risk Factors for General Medicine Compare, April 2011. Patients. Journal of Hospital Medicine, online, and Berry, J.G., et al. (2011). Hospital 8. The Affordable Care Act, Public Laws 111-148 & 111-152. Utilization and Characteristics of Patients Experiencing Recurrent Readmissions within 9. See: http://www.healthcare.gov/center/programs/partnership. Children’s Hospitals. JAMA, 305(7), 682-690. 10. The Affordable Care Act, Public Laws 111-148 & 111-152. 40. Aranda, J.M., et al. (2009). Current Trends in Heart Failure Readmission Rates: Analysis of Medicare Data. Clinical Cardiology, 32(1), 47-52. 11. U.S. Department of Health and Human Services. (12 April 2011). News Release. http://www.hhs.gov/news/press/2011pres/04/20110412a.html. 41. Jiang, H.J., and Wier, L.M. (April 2010). All-cause Hospital Readmissions among Non- elderly Medicaid Patients, 2007. HCUP Statistical Brief #89. Rockville, MD: Agency for 12. Center for Health Care Strategies, Inc. (July 2009). Policy Brief: Supporting Integrated Healthcare Research and Quality; Allaudeen, N., et al. (2010). Redefining Readmission Care for Dual Eligibles. Hamilton, NJ. Risk Factors for General Medicine Patients. Journal of Hospital Medicine, online; and 13. The Affordable Care Act, Public Laws 111-148 & 111-152. Berry, J.G., et al. (2011). Hospital Utilization and Characteristics of Patients Experiencing 14. Gorodeski, E.Z., et al. (2010). Are All Readmissions Bad Readmissions? New England Recurrent Readmissions within Children’s Hospitals. JAMA, 305(7), 682-690. Journal of Medicine, 363(3), 297-298. 42. Jencks, S.F., et al. (2009). Rehospitalizations among Patients in the Medicare Fee-for- 15. Analysis by Greater New York Hospital Association, 2009. Service Program. New England Journal of Medicine, 360(14), 1418-1428. 16. Hospital referral regions (HRRs) are regional market areas for tertiary medical care. Each 43. Allaudeen, N., et al. (2010). Redefining Readmission Risk Factors for General Medicine HRR contains at least one hospital that performs major cardiovascular procedures and Patients. Journal of Hospital Medicine, online. neurosurgery. See www.dartmouthatlas.org. 44. Hannan, E.L., et al. (2003). Predictors of Readmission for Complications of Coronary 17. Bernheim, S.M., et al. (2010). National Patterns of Risk-standardized Mortality and Artery Bypass Graft Surgery. JAMA, 290(6), 773-780. Readmission for Acute Myocardial Infarction and Heart Failure: Update on Publicly 45. All non-black patients were categorized as white. Reported Outcomes Measures Based on the 2010 Release. Circulation Cardiovascular 46. Joynt, K.E., et al. (2011). Thirty-day Readmission Rates for Medicare Beneficiaries by Quality and Outcomes, 3, 459-467. Race and Site of Care. JAMA, 305(7), 675-681. 18. Mulvey, G.K., et al. (2009). Mortality and Readmission for Patients with Heart Failure 47. Karliner, L.S., et al. (2010). Influence of Language Barriers on Outcomes of Hospital among U.S. News & World Report’s Top Heart Hospitals. Circulation Cardiovascular Care for General Medicine Inpatients. Journal of Hospital Medicine, 5(5), 276-282. Quality and Outcomes, 2, 558-565. 48. Joynt, K.E., and Jha, A.K. (2011). Who Has Higher Readmission Rates for Heart Failure, 19. Feudtner, C., et al. (2010). State-level Child Health System Performance and the and Why? Implications for Efforts to Improve Care Using Financial Incentives. Circulation Likelihood of Readmission to Children’s Hospitals. The Journal of Pediatrics, 157(1), Cardiovascular Quality and Outcomes, 4, 53-59. 98-102. 49. Jencks, S.F., et al. (2009). Rehospitalizations among Patients in the Medicare Fee-for- 20. Pallarito, K. (14 July 2010). High Readmission Rates May Not Mean Worse Hospital Service Program. New England Journal of Medicine, 360(14), 1418-1428. Care. HealthDay News. http://consumer.healthday.com/Article.asp?AID=641139. 50. Berry, J.G., et al. (2011). Hospital Utilization and Characteristics of Patients Experiencing 21. Benbassat, J., and Taragin, M. (2000). Hospital Readmissions as a Measure of Quality Recurrent Readmissions within Children’s Hospitals. JAMA, 305(7), 682-690. of Health Care: Advantages and Limitations. Archives of Internal Medicine, 160,1074- 1081. 51. Jiang, H.J., and Wier, L.M. (April 2010). All-cause Hospital Readmissions among Non- elderly Medicaid Patients, 2007. HCUP Statistical Brief #89. Rockville, MD: Agency for 22. Heidenreich, P.A., et al. (2010). Divergent Trends in Survival and Readmission Following Healthcare Research and Quality. a Hospitalization for Heart Failure in the Veterans Affairs Health Care System 2002 to 2006. Journal of the American College of Cardiology, 56(5), 362-368. 52. Bhalla, R., and Kalkut, G. (2010). Could Medicare Readmission Policy Exacerbate Health Care System Inequality? Annals of Internal Medicine, 152(2), 114-117. 23. O’Connor, C.M., and Fiuzat, M. (2010). Is Rehospitalization After Heart Failure Admission a Marker of Poor Quality? Time for Re-Evaluation. Journal of the American 53. Six Key Take-aways from STAAR on Reducing Readmissions. (December 2009). College of Cardiology, 56(5), 369-371. Leadership e-newsletter. Healthcare Financial Management Association. http://www.hfma.org/Templates/InteriorMaster.aspx?id=18651. 24. Ashton, C.M., et al. (1997). The Association between the Quality of Inpatient Care and Early Readmission: A Meta-analysis of the Evidence. Medical Care, 35(10), 1044-1059. 54. Minott, J. (September 2008). Reducing Hospital Readmissions. Washington, DC: Academy Health. 25. Hayward, R.A., et al. (1993). An Evaluation of Generic Screens for Poor Quality of Hospital Care on a General Medicine Service. Medical Care, 31(5), 394-402. 55. Medicare Payment Advisory Commission. (June 2007). Payment Policy for Inpatient Note: The article notes that, for the assessment of quality, the reviewers were asked Readmissions. Report to the Congress: Promoting Greater Efficiency in Medicare. to concentrate on the quality of the process of care, to evaluate whether they felt that Washington, DC. the care received was appropriate, regardless of whether the patient had a good or bad 56. Ross, J.S., et al. (2008). Statistical Models and Patient Predictors of Readmission for outcome. Heart Failure: A Systematic Review. Archives of Internal Medicine, 168(13), 1371- 26. Behal, R. Quality: Beyond the Numbers. Rush University Medical Center. 1386. http://www.rush.edu/rumc/page-1277738436397.html. 57. Hernandez, A.F., et al. (2010). Relationship between Early Physician Follow-up and 30- 27. Friedman, B., et al. (Winter 2008/2009). Costly Hospital Readmissions and Complex Day Readmission Among Medicare Beneficiaries Hospitalized for Heart Failure. JAMA, Chronic Illness. Inquiry, 45, 408-421. 303(17), 1716-1722. 28. Friedman, B., et al. (Winter 2008/2009). Costly Hospital Readmissions and Complex 58. American Hospital Association personal communication with Cyd deWaha, Quality Chronic Illness. Inquiry, 45, 408-421. Coordinator, Metro Health Hospital, Wyoming, MI. 29. Berry, J.G., et al. (2011). Hospital Utilization and Characteristics of Patients Experiencing 59. American Hospital Association personal communication with Gayle Shier, MSW, Program Recurrent Readmissions within Children’s Hospitals. JAMA, 305(7), 682-690. Coordinator, Older Adult Programs, Rush University Medical Center, Chicago, IL. 30. Allaudeen, N., et al. (2010). Redefining Readmission Risk Factors for General Medicine 60. Jack, B.W., et al. (2009). A Reengineered Hospital Discharge Program to Decrease Patients. Journal of Hospital Medicine, online. Rehospitalization: A Randomized Trial. Annals of Internal Medicine, 150(3), 178-187. 31. Medicare Payment Advisory Commission. (June 2007). Payment Policy for Inpatient 61. Landro, L. (2011, June 7). Don’t Come Back, Hospitals Say. The Wall Street Journal, Readmissions. Report to the Congress: Promoting Greater Efficiency in Medicare. pp. D3. Washington, DC. 62. Landro, L. (2011, June 7). Don’t Come Back, Hospitals Say. The Wall Street Journal, 32. Aranda, J.M., et al. (2009). Current Trends in Heart Failure Readmission Rates: Analysis pp. D3. TrendWatch, produced by the American Hospital American Hospital Association Avalere Health LLC Association, highlights important trends in the Liberty Place, Suite 700 1350 Connecticut Avenue, NW hospital and health care field. Avalere Health supplies 325 Seventh Street, NW Suite 900 research and analytic support. Washington, DC 20004-2802 Washington, DC 20036 TrendWatch—September 2011 202.638.1100 202.207.1300 Copyright © 2011 by the American Hospital Association. www.aha.org www.avalerehealth.net All Rights Reserved