|
Trends in 30-Day Readmission Rates for Medicare and Non-Medicare Patients in the Era of the Affordable Care Act
Abstract
BACKGROUND: Temporal changes in the readmission rates for patient groups and conditions that were not directly under the purview of the Hospital Readmissions Reduction Program (HRRP) can help assess whether efforts to lower readmissions extended beyond targeted patients and conditions. METHODS: Using the Nationwide Readmissions Database (2010-2015), we assessed trends in all-cause readmission rates for 1 of the 3 HRRP conditions (acute myocardial infarction, heart failure, pneumonia) or conditions not targeted by the HRRP in age-insurance groups defined by age group (65 years or <65 years) and payer (Medicare, Medicaid, or private insurance). RESULTS: In the group aged 65 years, readmission rates for those covered by Medicare, Medicaid, and private insurance decreased annually for acute myocardial infarction (risk-adjusted odds ratio [OR; 95% confidence interval] among Medicare patients, 0.94 [0.94-0.95], among Medicaid patients, 0.93 [0.90-0.97], and among patients with private-insurance, 0.95 [0.93-0.97]); heart failure (ORs, 0.96 [0.96-0.97], 0.96 [0.94-0.98], and 0.97 [0.96-0.99], for the 3 payers, respectively), and pneumonia (ORs, 0.96 [0.96-0.97), 0.94 [0.92-0.96], and 0.96 [0.95-0.97], respectively). Readmission rates also decreased in the group aged <65 years for acute myocardial infarction (ORs: Medicare 0.97 [0.96-0.98], Medicaid 0.94 [0.92-0.95], and private insurance 0.93 [0.92-0.94]), heart failure (ORs, 0.98 [0.97-0.98]: 0.96 [0.96-0.97], and 0.97 [0.95-0.98], for the 3 payers, respectively), and pneumonia (ORs, 0.98 [0.97-0.99], 0.98 [0.97-0.99], and 0.98 [0.97-1.00], respectively). Further, readmission rates decreased significantly for non-target conditions.
CONCLUSIONS: There appears to be a systematic improvement in readmission rates for patient groups beyond the population of fee-for-service, older, Medicare beneficiaries included in the HRRP.
Trends in 30-Day Readmission Rates for Medicare and Non-Medicare Patients in the Era of the Affordable Care Act BACKGROUND The Hospital Readmissions Reduction Program (HRRP) has been associated with substantial reductions in readmission within 30 days of discharge among fee-for-service Medicare beneficiaries aged 65 years who are hospitalized with acute myocardial infarction, heart failure, or pneumonia-the target population for this program. 1, 2 There have been suggestions that hospitals might have pursued reductions in readmissions through efforts mainly directed toward Medicare beneficiaries aged 65 years without pursuing systematic improvements in the care of patients. 3 Other reports have suggested an inconsistent cross-sectional association between hospital-level readmission rates for Medicare beneficiaries for conditions covered under the HRRP compared with other patient groups. 4, 5 However, an assessment of the temporal association between the HRRP's introduction and changes in readmissions for patient groups other than the Medicare beneficiaries targeted in the program is essential to assess how rates of readmission have evolved in an era with emphasis on readmission reduction for patients who were not directly being targeted for quality improvement nationally.
Accordingly, we used the Healthcare Cost and Utilization Project's Nationwide Readmissions Database (NRD), a nationally representative all-payer database, for 2010-2015 to assess temporal trends in 30-day readmission rates for the 3 HRRP target conditions (acute myocardial infarction, heart failure, and pneumonia) and other conditions not targeted by the HRRP, across age-insurance groups.
We used the NRD for the years 2010-2015. 6 The NRD is a nationally representative, all-payer dataset that has been constructed using discharge-level data for all hospitalizations from the Healthcare Cost and Utilization Project's State Inpatient Databases of geographically dispersed participating states (18-27 states during 2010-2015) . The sample includes nearly half of all US hospitalizations each year. In 2015, for example, the NRD included 56.6% of hospitalizations in the United States, representing all hospitalizations for 57.8% of the total population. The NRD uses a year-specific, patient-level identifier that allows tracking of patients across hospitalizations in a state within a calendar year.
The NRD includes clinical and demographic variables for each hospitalization and information on the primary insurance payer for each hospitalization. To ensure the uniformity of coding across different data sources, the payer in NRD is classified into broad insurance groups of Medicare, Medicaid, private insurance, self-pay, no charge, other, and missing or invalid. For this study, we included the patients with Medicare, Medicaid, or private insurance as the payers of the hospitalization; Medicare and Medicaid both include fee-for-service and managed care patients, and private insurance includes commercial insurance providers. Given the variability in reporting across participating states, only 1 payer is associated with each person in the NRD. In patients with more than 1 source of insurance coverage, only the insurance program expected to reimburse the hospital for the clinical encounter (the primary payer) is included in the NRD.
We included all hospitalizations among adults (18 years) with a primary discharge diagnosis of acute myocardial infarction, heart failure, and pneumonia, identified using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes for discharges between January 2010 and September 2015, and the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes for discharges between October and December 2015. These definitions are consistent with those used by the Centers for Medicare & Medicaid Services (CMS) for its readmission metrics for the respective conditions. [7] [8] [9] We also identified nontarget conditions, representing hospitalizations for conditions that were not subject to financial penalties under the HRRP. To define index hospitalizations for nontarget conditions, we excluded the hospitalizations for the 3 target conditions (acute myocardial infarction, heart failure, and pneumonia). We also excluded hospitalizations for chronic obstructive pulmonary disease and hip and knee arthroplasty because these types were included in the HRRP toward the end of the study period. 10 The ICD-9-CM and ICD-10-CM codes used to identify each of these conditions are included in Supplementary Tables 1-10 (available online) . We stratified all hospitalizations into age-insurance subgroups defined by age (65 years or <65 years) and insurance payer (Medicare, Medicaid, or private insurance). For each of the 4 conditions (acute myocardial infarction, heart failure, pneumonia, and nontarget hospitalizations), there were 6 age-insurance groups based on age and insurance payer: 65 years with Medicare, Medicaid, or
Between 2010 and 2015, all-cause 30day readmission rates for acute myocardial infarction, heart failure, and pneumonia, the three conditions originally targeted under the Hospital Readmissions Reduction Program (HRRP), decreased significantly across three payer groups of Medicare, Medicaid and Private Insurance. For hospitalizations not targeted by HRRP, there was a modest decrease in readmission rates.
private insurance, and <65 years with Medicare, Medicaid, or private insurance.
The outcome of interest was all-cause 30-day readmission. Readmission was defined as any hospitalization within 30 days of the discharge after an index hospitalization to the same or a different hospital within the state. A hospitalization classified as a readmission was not an index hospitalization for a subsequent readmission event. Data in the NRD are restricted to calendar years, without an ability to track patients across years. Therefore, as recommended by the Agency for Healthcare Research and Quality, we used data for 11 months (January-November) to allow 30-day follow-up for all patients for each year in the analyses. 11
We defined an index hospitalization as one in which patients were discharged alive and did not leave the hospital against medical advice. Further, hospitalizations with missing information on either the date of admission or hospital length of stay were excluded because that information was required for the assessment of postdischarge, 30-day events. Multiple index admissions were possible for each patient, regardless of the time elapsed between 2 hospitalizations.
To account for the complex survey design of the NRD, we used survey-specific methodology with hospital as cluster, NRD stratum as strata, and discharge-level weights as weight to obtain weighted nationwide, annual 30-day readmission rates and further evaluate the risk-adjusted annual trends in readmission. [11] [12] [13] For risk adjustment, we created a patientlevel survey logistic regression model with 30-day readmission as the outcome and risk factors that are part of the risk adjustment in the CMS publicly reported measure for acute myocardial infarction, heart failure, and pneumonia (age, gender, and comorbidities) based on the secondary diagnoses in the index hospitalization as independent variables. For nontarget conditions, we used the set of risk factors that are included in the CMS hospital-wide readmission measure. 14 In contrast to the CMS measures, where risk factors are defined based on the diagnoses in the index hospitalization as well as a preceding 12-month period before the index event, the comorbidities used in our survey logistic regression were defined in the index event. To allow assessment of risk-adjusted odds of readmission over years, we assessed calendar year as a dummy coded continuous variable in the model, allowing for an assessment of odds of annual changes in readmission stratified by the 6 age-insurance groups. To further evaluate the differences in annual trends of readmission between older Medicare and other patient groups, we tested the interaction between the calendar year and age-payer patient groups with Medicare patients aged 65 years as the reference group.
All analyses were conducted using SAS 9.4 software (SAS Institute, Cary, NC), and the level of significance was set at 0.05. The study was exempted by the institutional review board at Yale University (New Haven, Conn) because the data were de-identified. The data are available publicly through the Agency for Healthcare Research and Quality.
Between 2010 and 2015, there were 1,051,140 hospitalizations for acute myocardial infarction, 2,128,140 for heart failure, 2,067,240 for pneumonia, and 53,734,220 for nontarget conditions in the NRD, representing an estimated 2,364,371, 4,795,327, 4,900,012, and 121,093,299 hospitalizations nationally for acute myocardial infarction, heart failure, pneumonia, and nontarget conditions, respectively (Supplementary Figure 1 , available online). Overall, there were an estimated 349,139 readmissions nationally for acute myocardial infarction (14.8%), 1,111,593 for heart failure (23.2%), 817,431 for pneumonia (16.7%), and 16,290,748 for nontarget conditions (13.5%). The selected characteristics of index hospitalizations in the 6 patient age-insurance groups for acute myocardial infarction, heart failure, pneumonia, and nontarget conditions are included in Table 1 , and the characteristics of the patients hospitalized for target conditions are included in Supplementary Table 11 (available online) .
Over the 6 years of the study period (2010-2015), 30-day readmission rates for acute myocardial infarction decreased for all 6 age-insurance groups ( Figures 1A and 2A Similar to acute myocardial infarction, all-cause 30-day readmission rates for heart failure decreased in all age-insurance groups between 2010 and 2015, with a similar relative decline in all groups (P for calendar-year*age-insurance Figure 2 [available online]). There was a much smaller relative decrease in readmission rates for all insurance groups in the aged <65 years age group, relative to the trends in the reference group of those covered by Medicare and who were aged 65 years (P for calendar year*age-insurance interaction <.05 for all groups aged <65 years). Specifically, in this age group, readmission rates decreased from 20.1% in 2010 to 19.4% in 2015 for those covered by Medicare (risk-adjusted OR, 0.98; 95% CI, 0.97-0.99), from 18.2% to 17.4% for Medicaid patients (risk-adjusted OR, 0.98; 95% CI, 0.97-0.99), and from 11.0% to 10.9% for the privately insured (risk-adjusted OR, 0.98; 95% CI, 0.97-1.00).
Patients hospitalized for nontarget conditions who were covered by either Medicare or Medicaid were more frequently readmitted than patients who were privately insured; this was found in patients both aged >65 and <65 years. Further, there was a small decrease in observed readmission rates in most age-insurance groups ( Overall, the decrease in readmission rates was significantly higher for target conditions than for nontarget conditions (P < .001). A similar pattern of a larger decrease in readmission rates for target vs nontarget conditions was observed across age-insurance groups (P < .05 for all).
From 2010-2015, 30-day all-cause readmission rates for acute myocardial infarction, heart failure, and pneumonia declined across all age-insurance groups. Readmission rates decreased modestly for conditions not targeted by the HRRP in all age-payer groups, with larger declines among Medicare patients aged 65 years. These results are consistent with the hypothesis that interventions designed to reduce readmissions due to the HRRP were implemented broadly rather than exclusively applied to older Medicare fee-for-service beneficiaries.
Under the HRRP, hospitals are financially incentivized to lower excess readmissions among fee-for-service Medicare patients aged 65 years. Studies have shown that since the announcement of the HRRP, readmission rates in Medicare have improved for patients aged 65 years with conditions specifically targeted under the HRRP, exceeding the decrease in readmission rates for conditions that are not included in the program. 2, 15 Our study further extends our understanding of the trends of readmission rates across the non-Medicare population, including privately insured patients who also have value-based payment models with a consideration for readmissions.
Our findings may reflect that hospital-wide efforts were implemented to improve quality of care to decrease readmission rates. Some of the structural changes might have been spurred by CMS itself, which has initiated several programs aimed at reducing readmissions. One such program, Hospital Engagement Networks, was established by the CMS Partnership for Patients in 2011 and focuses on patient safety, particularly the prevention of patient harm after discharge. 16 The program operated through dissemination of information and the development of collaborations between hospitals. Hospital to Home is a nationwide initiative to develop resources for hospitals to improve transitions through discharge and reduce their readmission rates. 17 The program emphasized early postdischarge ambulatory care follow-up and patient education. Further, new programs are being initiated to address the burden of unplanned readmissions. The Patient Navigator Program of the American College of Cardiology aims to reduce avoidable hospital readmissions for patients discharged with acute myocardial infarction by supporting a culture of patient-centered care during the hospital stay and in the weeks following discharge. 18 Adoption of these programs may have led to much wider improvements in readmission rates through improved healthcare delivery, patient education, and improved follow-up, which were not limited to the patients covered under the HRRP directly. This information also represents critical feedback to health policy makers, since health policy interventions may have wider implications for patient health and wellness.
The findings of the study should be interpreted in light of certain limitations. First, we used serial cross-sectional data and assessed secular trends. Therefore, the findings cannot be interpreted to represent a direct causal relationship between health policy implementation and changes in readmission outcomes. Second, hospitalizations are included in the NRD at the time of discharge and cannot be tracked across years. To assess 30-day readmission rates, data from December were excluded. However, this exclusion was consistent across study years and has been recommended by the Agency for Healthcare Research and Quality. 19 Third, we are unable to specifically account for the competing risk of posthospitalization mortality. However, our analysis compares temporal trends in specific insurance groups, and therefore trends may be consistent across study years. Fourth, changing trends may be subject to directional changes in coding practices. However, it is unclear if such coding changes would disproportionately affect certain patient groups. Fifth, insurance categories defined in the study rely on those reported as the primary payer in the NRD. Moreover, individuals reported as covered under Medicare include both fee-for-service and managed care beneficiaries. Therefore, trends may vary if payers are defined differently. Last, the hospitals cannot be tracked across years, precluding an assessment of hospital-specific readmission rates.
There was a significant decline in readmission rates for the 3 conditions targeted by the HRRP across age and payer groups. Readmission rates also declined modestly for conditions not targeted by the HRRP. These patterns are consistent with the hypothesis that implementation of the HRRP was associated with systematic changes in the care of patients and reduced readmission risk beyond the HRRP's target population of fee-for-service Medicare beneficiaries. Benign hypertensive heart disease with heart failure 402.91
Unspecified hypertensive heart disease with heart failure 404.01
Hypertensive heart and chronic kidney disease, malignant, with heart failure and with chronic kidney disease stage I through stage IV, or unspecified 404.03
Hypertensive heart and chronic kidney disease, malignant, with heart failure and with chronic kidney disease stage V or end stage renal disease 404. 11 Hypertensive heart and chronic kidney disease, benign, with heart failure and with chronic kidney disease stage I through stage IV, or unspecified 404. 13 Hypertensive heart and chronic kidney disease, benign, with heart failure and chronic kidney disease stage V or end stage renal disease 404.91
Hypertensive heart and chronic kidney disease, unspecified, with heart failure and with chronic kidney disease stage I through stage IV, or unspecified 404.93
Hypertensive heart and chronic kidney disease, unspecified, with heart failure and chronic kidney disease stage V or end stage renal disease 428.0
Congestive heart failure, unspecified 428. Hypertensive heart and chronic kidney disease with heart failure and stage 1 through stage 4 chronic kidney disease, or unspecified chronic kidney disease I13. 2 Hypertensive heart and chronic kidney disease with heart failure and with stage 5 chronic kidney disease, or end stage renal disease I50. 1 Left ventricular failure I50.20
Unspecified systolic (congestive) heart failure I50.21
Acute systolic (congestive) heart failure I50.22
Chronic systolic (congestive) heart failure I50.23
Acute on chronic systolic (congestive) heart failure I50. 30 Unspecified diastolic (congestive) heart failure I50.31
Acute diastolic (congestive) heart failure I50.32
Chronic diastolic (congestive) heart failure I50.33
Acute on chronic diastolic (congestive) heart failure I50.40
Unspecified combined systolic (congestive) and diastolic (congestive) heart failure I50.41
Acute combined systolic (congestive) and diastolic (congestive) heart failure I50.42
Chronic combined systolic (congestive) and diastolic (congestive) heart failure I50.43
Acute on chronic combined systolic (congestive) and diastolic (congestive) heart failure I50. 9 Heart failure, unspecified 1331.e3
The American Journal of Medicine, Vol 131, No 11, November 2018 Supplementary Pneumonia due to other streptococci J15. 5 Pneumonia due to Escherichia coli J15. 6 Pneumonia due to other aerobic Gram-negative bacteria J15. 7 Pneumonia due to Mycoplasma pneumoniae J15. 8 Pneumonia due to other specified bacteria J15. 9 Unspecified bacterial pneumonia J16.0
Chlamydial pneumonia J16. 8 Pneumonia due to other specified infectious organisms J18.0 Bronchopneumonia, unspecified organism J18. 1 Lobar pneumonia, unspecified organism J18. 8 Other pneumonia, unspecified organism J18. 9 Pneumonia, unspecified organism J69.0
Pneumonitis due to inhalation of food and vomit Supplementary
|
