Skip main navigation

Trends in 30- and 90-Day Readmission Rates for Heart Failure

Originally publishedhttps://doi.org/10.1161/CIRCHEARTFAILURE.121.008335Circulation: Heart Failure. 2021;14:e008335

Abstract

Background:

The impact of hospital readmission reduction program (HRRP) on heart failure (HF) outcomes has been debated. Limited data exist regarding trends of HF readmission rates beyond 30 days from all-payer sources. The aim of this study was to investigate temporal trends of 30- and 90-day HF readmissions rates from 2010 to 2017 in patients from all-payer sources.

Methods:

The National Readmission Database was utilized to identify HF hospitalizations between 2010 and 2017. In the primary analysis, a linear trend in 30-day and 90-day readmissions from 2010 to 2017 was assessed. While in the secondary analysis, a change in aggregated 30- and 90-day all-cause and HF-specific readmissions pre-HRRP penalty phase (2010–2012) and post-HRRP penalties (2013–2017) was compared. Subgroup analyses were performed based on (1) Medicare versus non-Medicare insurance, (2) low versus high HF volume, and (3) HF with reduced versus preserved ejection fraction (heart failure with reduced ejection fraction and heart failure with preserved ejection fraction). Multiple logistic and adjusted linear regression analyses were performed for annual trends.

Results:

A total of 6 669 313 index HF hospitalizations for 30-day, and 5 077 949 index HF hospitalizations for 90-day readmission, were included. Of these, 1 213 402 (18.2%) encounters had a readmission within 30 days, and 1 585 445 (31.2%) encounters had a readmission within 90 days. Between 2010 and 2017, both 30 and 90 days adjusted HF-specific and all-cause readmissions increased (8.1% to 8.7%, P trend 0.04, and 18.3% to 19.9%, P trend <0.001 for 30-day and 14.8% to 16.0% and 30.9% to 34.6% for 90-day, P trend <0.001 for both, respectively). Readmission rates were higher during the post-HRRP penalty period compared with pre-HRRP penalty phase (all-cause readmission 30 days: 18.6% versus 17.5%, P<0.001, all-cause readmission 90 days: 32.0% versus 29.9%, P<0.001) across all subgroups except among the low-volume hospitals.

Conclusions:

The rates of adjusted HF-specific and all-cause 30- and 90-day readmissions have increased from 2010 to 2017. Readmissions rates were higher during the HRRP phase across all subgroups except the low-volume hospitals.

What Is New

  • Between 2010 and 2017, both 30- and 90-days adjusted heart failure specific and all-cause readmissions increased following an index heart failure hospitalization.

  • Specifically, readmission rates were higher during the post–hospital readmission reduction program penalty period (2013–2017) compared with prehospital readmission reduction program penalty phase (2010–2012).

  • Among low-volume hospitals, the readmissions were significantly lower during the post–hospital readmission reduction program penalty period compared with the pre–hospital readmission reduction program penalty period.

What are the Clinical Implications?

  • In this nationally representative sample of the US population, an increase in readmissions following index heart failure hospitalization from 2014 is worrisome.

  • Future studies should confirm these results to improve heart failure surveillance policies.

Introduction

Nearly 1 in 4 heart failure (HF) patients are readmitted within 30 days of discharge and approximately half are readmitted within 6 months.1–4 It has been suggested that about one quarter of HF readmissions may be preventable.4 As such, researchers and policy makers have placed significant focus on efforts to decrease readmissions and avoid excessive healthcare spending. In 2009, the Centers for Medicare and Medicaid Services started public reporting of hospital-level 30-day risk adjusted readmission rates for acute myocardial infarction, HF, and pneumonia. In 2010, the Congress added a financial incentive to decrease readmission rates for these target conditions and introduced the Hospital Readmission Reduction Program (HRRP) under which hospitals with higher than expected readmission rates were penalized up to 3% of their annual Medicare reimbursements.5

Many studies have reported reduction in 30-day readmission rates after the implementation of HRRP, suggesting that the program was a success in improving care and outcomes.5–7 Some studies have called into question the overall effectiveness of HRRP, as the improvement in readmission rates had started even before HRRP,8,9 and has mostly been stagnant since the penalty phase of implementation in late 2012.10–12 Moreover, concerns regarding unintended consequences have been raised, including higher postdischarge mortality, inappropriate triage for emergency care, and inability of safety-net hospitals to provide care for lower socioeconomic status populations.13

Previous studies on HF readmission and impact of HRRP have focused mainly on Medicare-eligible patients and 30-day readmissions.14,15 It is well known that the vulnerable phase after HF hospitalization extends well beyond 30 days as readmissions rates remain as high as 30% within 60 and 90 days after discharge.16,17 The National Readmission Database (NRD), one of the largest publicly available databases, includes patients from all-payer sources, provides an excellent opportunity to assess such trends. Prior studies utilizing NRD have focused largely on etiologies and predictors of HF readmissions for specific years. We conducted this study to investigate temporal linear trends of 30- and 90-day HF readmissions rates from 2010 to 2017 in patients from all-payer sources. We hypothesized that 30- and 90-day readmissions declined over a period from 2010 to 2017.

Methods

Data Source

The data that support the findings of this study are available from the corresponding author upon reasonable request. A retrospective analysis using Healthcare Cost and Utilization Project NRD from Agency for Healthcare Research and Quality for 2010 to 2017 was conducted. The NRD is one of the largest publicly available databases in the US including patients from all-payer sources. It provides longitudinal information about patient’s initial hospitalization and subsequent readmissions in a calendar year. It contains weighted sample of hospitalizations in the United States and this can be used to derive national estimates of various hospitalizations directly. In 2017, NRD broadened to include 28 states with approximately 18 million discharges and accounted for 60% of the total US resident population. Each year NRD contains data from January 1 to December 31. Beyond these dates, patient or hospital data cannot be linked, but within a year, patients can be tracked using a unique linkage number. This study was exempted from institutional board review approval as NRD database contains deidentified patient information.

Sample Population

Patients who were admitted with a principal diagnosis of acute HF defined as the first admission from the start of the year who then had a readmission were included. Admissions were not considered as an index admission if the patients were <18 years old, transferred to another acute care hospital, or had missing data for age, sex, or mortality. In NRD, patient identifiers cannot be linked across years, hence patients who had an index hospitalization on December 1 or beyond for 30-day readmissions and October 1 or beyond for 90-day readmissions were excluded. Patients were assessed for 30-day or 90-day readmissions. Time to readmission was calculated by subtracting length of stay of index admissions from time between 2 admissions. Patients who had at least one readmission within 30 or 90 days for respective 30-day and 90-day readmission analysis were included. However, only the first readmission and respective index readmission were included in the final analysis. Planned readmissions within 30 or 90 days of discharge were excluded. To determine the cause of readmission, the International Classification of Diseases (ICD; ICD, Ninth Revision and ICD, Tenth Revision) codes assigned for HF were used appropriately (Methods in the Data Supplement).

Variables

NRD variables to identify patient demographics, including age, sex, median household income (income quartiles referred to patients as 1-low income, 2-middle income, 3-upper middle income, 4-high income) were used. Hospital-specific variables included bed size, teaching status, location, quartile of HF volume defined on the basis of admitting hospital annual HF volume as 5 to 38 (quartile 1), 39 to 77 (quartile 2), 78 to 122 (quartile 3), 123 to 457 (quartile 4). We did the stratification of hospitalizations based on hospital HF volume for each year separately. The HF hospitalizations from all the years belonging to a particular quartile based on HF volume were combined together for final analysis in this study. Comorbidities were identified using CM_ variables of ICD, Ninth Revision and ICD, Tenth Revision respective to years that were used in NRD. These clinical conditions likely originated before hospitalization and were not primary reason for admission. For assessing the severity of comorbid conditions, Charlson Comorbidity Index was used which provides a weighted index irrespective of the coding system utilized. Data relating to these variables were collected separately for 30-day and 90-day outcomes.

Exposure

The exposure of interest was time intervals in relation to HRRP. The study was divided into 2 phases: (1) pre-HRRP penalty period (HRRP announcement phase)—2010 to 2012 and (2) post-HRRP penalty period (HRRP penalties implementation phase)—2013 to 2017. NRD does not contain data before 2010, and therefore, the analysis was limited to the available period of NRD from 2010 to 2017. We acknowledge that HRRP was not announced until March 2010; however, Centers for Medicare and Medicaid Services reporting of readmission rates had already started in 2009 and it was likely that hospitals were aware of the program by 2010. Similarly, the HRRP penalty phase started in October 2012, but it is unlikely that inclusion of the last quartile of 2012 within the HRRP announcement phase would impact results significantly since (1) we excluded the month of December and months of October to December of each year for assessment of 30-day and 90-day outcomes, respectively, and (2) our results represent yearly trends rather than monthly.

Outcomes

In the primary analysis, a linear trend in 30-day and 90-day readmissions from 2010 to 2017 was assessed. While in the secondary analysis, a change in aggregated 30- and 90-day all-cause and HF-specific readmissions pre-HRRP penalty phase (2010–2012) and post-HRRP penalties (2013–2017) was compared. Subgroup analyses were performed based on (1) Medicare versus non-Medicare, (2) men versus women, (3) low versus high HF volume hospitals, and (4) HF with reduced versus preserved ejection fraction (heart failure with reduced ejection fraction and heart failure with preserved ejection fraction). All of these analyses assessed readmission rates at 30 and 90 days.

Statistical Analyses

Stratified weighted data were utilized using svy command to obtain nationwide estimates. In NRD, as the discharges are clustered within the hospitals, it can potentially lead to dependence in readmission outcomes because the patients discharged from a hospital tend to have a common set of resources which is different from that of another hospital facility. As per specific Healthcare Cost and Utilization Project recommendations, we performed all our analysis utilizing the Healthcare Cost and Utilization Project STATA survey data analysis packages which incorporate the NRD specific variables including hospital identifiers, stratum, and discharge weights to account for clustering and large survey-weighted data analysis to obtain statistical and variance calculations independent of individual hospital discharge characteristics. Continuous variables were compared using the Student t test and categorical variables were compared using the χ2 test. Multiple logistic and linear regression analysis was performed for yearly trends with year of admission as the independent variable along with readmission event as the dependent variable. Readmission rate was adjusted for age, sex, medical comorbidities in the form of Charlson comorbidity index, and income quartile of the study population over the years using these variables as independent variables. Charlson comorbidity index has been validated for statistical use in survey-weighted sample including National Readmission Database. Adjusted readmission rates were obtained using marginal effects following multiple regression analysis. Trends in readmission rate between subgroups were compared using a subgroup-year interaction term in a multivariate regression analysis with readmission event as the dependent variable. In addition, nonlinear analysis was also performed by comparing each year to the previous year’s readmission rate in a 2 by 2 comparison using nonlinear binary logistic regression adjusted for age, sex, income, and comorbidities. A 2-sided P<0.05 was considered to represent statistical significance. All analyses were performed using STATA, version 16 (StataCorp, TX).

Results

Demographic and Clinical Characteristics

From 2010 to 2017, of the 7 272 341 index hospitalizations for HF in the US, 6 669 313 index HF hospitalizations were included for 30-day readmission analysis, and 5 077 949 index HF hospitalizations were included for 90-day readmission analysis. Of these, 1 213 402 (18.2%) patients had a readmission within 30 days and 1 585 445 (31.2%) patients had a readmission within 90 days. Table 1 provides the characteristics of patients assessed for 30-day readmissions. The mean age was 71.5±12.1 years and 49% were females. More than half of the patients (63.3%) had ≥3 major medical comorbidities with Charlson Comorbidity Index score of ≥3. Over half of the hospitalizations were at large teaching hospitals. Medicare was the largest payer (77.6%). Table I in the Data Supplement summarizes the clinical and hospital characteristics of patients assessed for 90-day readmissions. Overall, these were similar to the 30-day readmission cohort.

Table 1. Baseline Characteristics of Patients Evaluated for 30-Day Readmissions

Baseline characteristicsTotal (n=6 669 313)
Age, mean in years72.1±12.1
Female, n (%)3 281 302 (49.2)
Uncomplicated hypertension, n (%)2 147 518 (32.2)
Complicated hypertension, n (%)2 047 479 (30.7)
Anemia, n (%)486 860 (7.3)
Obesity, n (%)1 347 201 (20.2)
Systolic heart failure, n (%)2 440 969 (36.6)
Diastolic heart failure, n (%)1 440 572 (21.6)
Atrial fibrillation/flutter, n (%)2 414 291 (36.2)
Mitral valve disease, n (%)740 293 (11.1)
Aortic valve disease, n (%)473 521 (7.1)
Other valvular heart disease, n (%)1 213 814 (18.2)
Prior myocardial infarction, n (%)927 034 (13.9)
Prior percutaneous coronary intervention, n (%)753 632 (11.3)
Prior coronary artery bypass grafting, n (%)1 027 074 (15.4)
Charlson comorbidity index, n (%)
 0173 402 (2.6)
 1960 381 (14.4)
 21 307 185 (19.6)
 ≥34 221 675 (63.3)
Insurance, n (%)
 Medicare5 175 387 (77.6)
 Medicaid606 907 (9.1)
 Private706 947 (10.6)
 Uninsured180 071 (2.7)
Teaching hospital status, n (%)3 501 389 (52.5)
Metropolitan hospital location, n (%)5 615 561 (84.2)
Income quartile, n (%)
 12 314 251 (34.7)
 21 734 021 (26.0)
 31 480 587 (22.2)
 41 140 452 (17.1)
Hospital bed size, n (%)
 Low (1–249)1 040 412 (15.6)
 Medium (250–449)1 740 690 (26.1)
 Large (≥450)3 888 209 (58.3)
Hospital volume quartile for heart failure, n (%)
 Q1–low volume133 386 (2.0)
 Q2566 891 (8.5)
 Q31 487 257 (22.3)
 Q4–high volume4 481 778 (67.2)

Readmissions

Between 2010 and 2017, the rates of adjusted HF and all-cause 30-day readmissions increased from 8.11% to 8.7% (P trend 0.04) and 18.3% to 19.9% (P trend <0.001), respectively (Figure 1). According to HRRP time intervals, the 30-day all-cause readmissions (17.5%–18.6%, P<0.001) and 30-day HF readmissions (7.7%–8.1%, P<0.001) were noted to increase during the post-HRRP penalty period. Similar trends were seen for HF and all-cause 90-day readmission rates increasing from 14.8% to 16.0% (P trend <0.001) and 30.9% to 34.6% (P trend <0.001), respectively. Similarly, the 90-day all-cause readmissions (29.9% to 32.0%, P<0.001) and 90-day HF-specific readmissions (14.2% to 15.0%, P<0.001) were also observed to increase during the post-HRRP penalty period (Table 2). On nonlinear trend analysis, from 2010 till 2013, the readmission rates showed a consistent downtrend. However, the year 2014 was the time point where the readmission rate showed a static trend where there was no significant difference in the readmission rate compared with that of the preceding year 2013. The year 2015 could be identified as an inflection point for increase in both 30-day and 90-day all-cause and HF-specific readmission rates (Table II in the Data Supplement).

Table 2. Average 30-Day and 90-Day Readmission Pre-HRRP and Post-HRRP Period

Pre-HRRP (2010–2012)Post-HRRP (2013–2017)Pre-HRRP (2010–2012)Post-HRRP (2013–2017)Interaction odds ratio (95% CI), P value
30-d95% CI30-d95% CIP valueInteraction odds ratio (95% CI), P value90-d95% CI90-d95% CIP value
Average all-cause readmission, %17.517.4–17.718.618.5–18.7<0.00129.929.7–30.132.031.8–32.2<0.001
Average HF-specific readmission rate, %7.77.6–7.88.18.0–8.1<0.00114.214.0–14.415.014.9–15.1<0.001
All-cause readmission–subgroup differences
 All-cause readmission males, %17.517.3–17.718.718.6–18.8<0.0011.00 (0.99–1.01), 0.6029.429.2–29.731.731.6–31.9<0.0011.01 (0.99–1.02), 0.60
 All-cause readmission females, %17.517.4–17.718.518.3–18.6<0.00130.430.1–30.632.332.1–32.5<0.001
 All-cause readmission Medicare, %17.917.8–18.118.918.8–19.0<0.0010.98 (0.97–0.99), <0.00130.630.4–30.932.632.4–32.8<0.0010.99 (0.98–1.00), 0.06
 All-cause readmission non-Medicare, %16.216.0–16.417.717.5–17.9<0.00127.527.2–27.830.029.7–30.2<0.001
 All-cause readmission high volume quartile, %17.617.4–17.919.018.9–19.1<0.0011.03 (1.01–1.05), <0.00130.129.8–30.432.732.5–32.9<0.0011.03 (1.01–1.05), <0.001
 All-cause readmission low-volume quartile, %14.614.1–15.212.712.3–13.2<0.00125.124.3–25.922.221.6–22.9<0.001
 All-cause readmission HFrEF, %17.817.6–18.018.118.0–18.20.0101.00 (0.99–1.00), 0.5030.129.8–30.331.231.0–31.4<0.0011.00 (0.99–1.01), 0.40
 All-cause readmission HFpEF, %17.617.3–17.817.116.9–17.20.00230.530.2–30.829.629.4–29.9<0.001
HF-specific readmission–subgroup differences
 HF-specific readmission males, %8.28.0–8.38.58.4–8.6<0.0011.00 (0.99–1.01), 0.8014.614.4–14.815.515.4–15.6<0.0011.00 (0.99–1.00), 0.80
 HF-specific readmission females, %7.37.2–7.47.67.5–7.70.01013.713.6–13.914.514.4–14.6<0.001
 HF-specific readmission Medicare, %7.67.5–7.77.97.8–8.00.0020.98 (0.98–0.99), <0.00114.214.0–14.414.914.7–15.0<0.0010.99 (0.98–0.99), 0.07
 HF-specific readmission non-Medicare, %7.97.8–8.18.68.5–8.7<0.00114.113.9–14.315.515.3–15.7<0.001
 HF-specific readmission high volume quartile, %7.87.6–7.88.38.2–8.4<0.0011.03 (1.01–1.06), <0.00114.414.2–14.615.515.4–15.6<0.0011.03 (1.02–1.05), <0.001
 HF-specific readmission low-volume quartile, %6.15.7–6.45.25.0–5.4<0.00110.710.2–11.29.49.0–9.8<0.001
 HF-specific readmission HFrEF, %8.88.7–8.98.88.7–8.90.8001.00 (0.99–1.01), 0.5016.015.8–16.216.416.2–16.60.0101.01 (0.99–1.00), 0.30
 HF-specific readmission HFpEF, %6.56.4–6.86.46.3–6.50.05012.412.3–12.712.312.1–12.50.070

HF indicates heart failure; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; and HRRP, Hospital Readmission Reduction Program.

Figure 1.

Figure 1. Trends in 30-d and 90-d all-cause and heart failure (HF) specific readmissions. Between 2010 and 2017, the rates of adjusted HF-specific and all-cause 30- and 90-d readmissions increased. Blue and red lines represent 30-d and 90-d all-cause readmission rates, respectively, whereas orange and yellow lines represent 30-d and 90-d HF readmission rates, respectively.

Subgroup Analyses

Increasing trend in readmission rate among both Medicare and non-Medicare patients was observed (Figure 2). Similarly, increasing rates of readmission were observed in both men and women in similar proportions at 30 days (18.1%–19.9% versus 18.6%–20.2%, P interaction=0.6) and 90 days (30.1%–33.9% versus 31.8%–35.3%, P interaction=0.6; Figure I in the Data Supplement). Whereas the low-volume hospitals tended to have relative decline in the readmission rate from 2010 to 2017, high volume hospitals showed a larger increase in readmission rate at 30 days (15.7%–15.0%, P trend=0.001 versus 18.5%–20.2%, P trend <0.001, P interaction <0.001) and 90 days (27.1%–26.2%, P trend <0.001 versus 31.1%–35.1%, P trend <0.001, P interaction <0.001; Figure II in the Data Supplement). The overall study did not show any significant differences in the trends of readmission rates for heart failure with reduced ejection fraction versus heart failure with preserved ejection fraction across the study at 30 days (absolute difference, 1.0% versus 1.6%, P interaction=0.5) and 90 days (2.2% versus 3.7%, P interaction=0.4; Figure III in the Data Supplement). In relation to HRRP, both the 30-day and 90-day (all-cause and HF-specific) readmissions were higher during the post-HRRP penalty period compared with the pre-HRRP penalty period across all prespecified subgroups except among low-volume hospitals in which the readmissions were significantly lower during the post-HRRP penalty period (all-cause 30 days: 12.7% versus 14.6%, P<0.001, all-cause 90 days: 22.2% versus 25.1%, P<0.001; Table 2).

Figure 2.

Figure 2. Trends in 30-d and 90-d readmissions in Medicare vs non-Medicare patients. Increasing trend in readmission rate among both Medicare beneficiaries and non-Medicare patient population was observed; however, the absolute increase in readmission rate was higher among the non-Medicare group compared with Medicare group. Blue and red lines represent 30-d and 90-d all-cause readmission rates among Medicare patients, respectively, whereas orange and yellow lines represent 30-d and 90-d all-cause readmission rates in non-Medicare patients, respectively.

Discussion

Our study has several important findings. First, between the periods of announcement and implementation of HRRP penalties, an increase in rate of 30- and 90-day HF-specific and all-cause readmissions was observed. Second, among low-volume hospitals, the readmissions were significantly lower during the post-HRRP penalty period compared with the pre-HRRP penalty period. NRD includes data from 28 states accounting for 58.2% of all US hospitalizations, and because it also includes patients from all-payer sources, it is an important database to evaluate trends in HF readmissions.

Several studies of Medicare beneficiaries have suggested that implementation of HRRP has been associated with decrease in HF readmissions. Specifically, Gupta et al14 (20%–18.4%) and Wadhera et al18 (21.6%–20.9%) reported a decline in 30-day readmissions after HRRP implementation phase. Furthermore, Desai et al11 pointed a parallel reduction in readmissions for target and nontarget conditions post-HRRP till 2014. Our study found similar results through 2014 with an absolute decrease in 30-day HF and all-cause readmissions of 0.5% and 1.4%, respectively, suggesting that previously identified early reductions in readmissions extended to a broader population of Medicare and non-Medicare patients. However, beyond 2014 we observed a fluctuating trend with uptick in readmissions in 2015 and 2017. In contrast, the Medicare Payment Advisory Commission report noted an overall consistent 0.6% per year decline in 30-day HF readmissions in the 2012 to 2016 HRRP penalty phase; although similar to our report, the Medicare Payment Advisory Commission report did find an increase in all-cause readmissions for the year 2015.5

It is unclear why these variations were noted but it is important to emphasize that number of discharges increased from ≈15 million in 2014 to ≈18 million in 2015 as 6 more states were included in the NRD. However, the number of discharges has remained the same, ≈18 million, for the past three years, thus overall sample size does not explain these trends. Of note, Hsuan et al19 reported an increased probability of readmission for patients presenting to emergency department (ED) for nontargeted conditions versus decrease in readmission for targeted conditions such as HF. Similarly, during HRRP penalty phase (2012–2015), Wadhera et al20 noted an increase in ED visits and observation stays for targeted conditions with increase in revisits exceeding the decrease in readmissions during this period. Hence, it can be speculated that fluctuations in ED triage trends can be a possible explanation of the variations in readmission observed in the later years. However, this should be considered hypothesis-generating.

Studies have suggested that the vulnerable phase after an index HF hospitalization lasts 2 to 3 months.16,17 In a study of Medicare beneficiaries, Kilgore et al21 noted HF-related readmission rates of 8.4%, 13.4%, and 16.7% for 30-day, 60-day, and 90-day readmissions, respectively. In our study, the 30- and 90-day HF readmission rates were 7.8% and 14.2% in HRRP announcement phase which then increased to 8.0% and 14.9% in the HRRP penalty phase, respectively. These findings suggest that the risk of HF readmission is progressive and extends well beyond the 30-day cutoff. Because 30-day readmissions may be more directly impacted by acute care during index hospitalization, assessment of outcomes at 90 days may be more helpful in evaluating the role of ambulatory management and care coordination. Accordingly, Centers for Medicare and Medicaid Services has recently announced a voluntary 90-day episode payment model for acute myocardial infarction.22 Such an approach may also be beneficial for HF patients, as it may prompt hospitals to invest in more comprehensive transitional care programs that will provide long-term benefit and will be able to decrease HF readmission across diagnosis rather than just after a principal HF hospitalization.

An important point of criticism of the HRRP is that it can disproportionately penalize hospitals that serve patients with lower socioeconomic status and income.13 This context raises the importance of our analysis as NRD includes patients from all-payer sources and has been validated in multiple studies to be a nationally representative sample. When Medicare versus non-Medicare patients were compared in our analysis, P interaction was statistically significant for 30-day readmission rates (<0.001) and almost reached significance for 90-day readmission rates (0.06). However, the curves were mostly parallel to each other indicating that any change that HRRP may have prompted at a system-level would have affected both groups equally. Of more concern is the absolute increase in rates between Medicare versus non-Medicare patients of 3.3% versus 3.5% for 30-day and 6.0% versus 5.8% for 90-day readmissions from 2010 to 2017. These fluctuations in readmission rates combined with unequal penalization of hospitals taking care of patients who are already financially vulnerable can be devastating for the entire safety-net. Keeping this in view, starting in October 2018, Centers for Medicare and Medicaid Services instituted risk adjustment to make the model more lenient towards hospital taking care of patients with lower socioeconomic status.5

Our study has several important clinical and policy implications. First, we observed a fluctuating trend with overall increase in readmissions during the latter years of the study period. It is important to note that increased readmissions even in 0.1% of the patients in NRD constitute>10 000 patients. Current evidence suggests an inverse relationship between readmissions and ED visits.19,20,23 It is possible that a similar trend was applicable in our study. Therefore, we agree with the expert opinion of broadening the readmission measure to include all postdischarge acute care consisting of rehospitalization, ED visits and observation stays.13 Second, the hospital volume for HF hospitalizations should be considered while assessing the adjusted excess readmission ratios for a particular hospital. Third, similar to Medicare Payment Advisory Commission and other reports, we assessed readmissions after an index acute HF hospitalization.5,14,16 Studies have shown comparable readmission and mortality rates for both primary and comorbid HF hospitalization.10,24 Hence, assessing readmissions after any HF hospitalization (ie, either primary or secondary diagnosis) may be appropriate. Fourth, although results have been variable, some HF transitional interventions such as outpatient multidisciplinary HF clinics, telemonitoring, and home visiting programs have shown decreased readmissions and mortality at 3 or 6 months with little or no effect on 30-day readmissions.25–27 Hence, considering readmissions at 90 days may provide hospitals with the necessary time to implement, assess, and improve these measures.

There are several limitations in our study. First, this is an observational analysis of NRD administrative data, and thus our findings cannot establish causation. Residual measured or unmeasured confounding may have influenced these findings Second, data regarding postdischarge clinic or specialist follow-up were not available, which may have impacted readmission mortality rates after index hospitalization. Third, NRD excludes interstate hospitalizations and does not link patient data across years, which may have affected readmission estimates. Fourth, use of ICD codes for administrative data analysis is subject to potential error. Of note, although the coding changed from ICD, Ninth Revision to ICD, Tenth Revision in 2015, it is unlikely that it would have substantially affected the proportion of readmissions. Although the change in ICD coding for HF hospitalizations could change the total number of hospitalizations included in our analysis, it is unlikely to affect the readmission rates as they are calculated as the proportion of the index hospitalizations tracked over the calendar year for readmissions. Similarly, the addition of more states to the NRD increases the national representativeness of the sample and should not affect the proportion of readmissions calculated.

Finally, NRD only collects data on in-hospital mortality, and out-of-hospital death data are not available. Thus, although prior works have evaluated the association of HRRP with postdischarge mortality, such assessment was not possible in the current study.

Conclusions

In conclusion, in this analysis of patients hospitalized for HF across all-payer sources, rates of adjusted HF-specific and all-cause 30- and 90-day readmissions increased from 2010 to 2017. Both 30- and 90-day readmissions rates were higher during the HRRP penalty phase compared with the pre-HRRP announcement phase across all subgroups, with the exception of low-volume hospitals.

Nonstandard Abbreviations and Acronyms

HF

heart failure

HRRP

Hospital Readmission Reduction Program

ICD

International Classification of Diseases

NRD

National Readmission Database

Supplemental Materials

Methods

Tables I–II

Figures I–III

Disclosures Dr Greene has received a Heart Failure Society of America/Emergency Medicine Foundation Acute Heart Failure Young Investigator Award funded by Novartis; receives research support from the American Heart Association, Amgen, AstraZeneca, Bristol Myers Squibb, Merck, and Novartis; has served on advisory boards for Amgen and Cytokinetics and serves as a consultant for Amgen and Merck. Dr Ahmad reports support from Amgen, cytokinetics, research support from AZ, BMS, Amgen. Dr Anker reports receiving fees from Bayer, Boehringer Ingelheim, Cardiac Dimension, Impulse Dynamics, Novartis, Servier, St. Jude Medical, and Vifor Pharma, and grant support from Abbott Vascular and Vifor Pharma. Dr Fonarow reports consulting for Abbott, Amgen, AstraZeneca, Bayer, CHF Solutions, Edwards, Janssen, Medtronic, Merck, Novartis. Dr Butler declares that he serves as a consultant for Abbott, Adrenomed, Amgen, Array, Astra Zeneca, Bayer, Boehringer Ingelheim, Bristol Myers Squib, CVRx, G3 Pharmaceutical, Impulse Dynamics, Innolife, Janssen, LivaNova, Luitpold, Medtronic, Merck, Novartis, NovoNordisk, Relypsa, Roche, V-Wave Limited, and Vifor. The other authors report no conflicts.

Footnotes

*M.S. Khan and J. Sreenivasan contributed equally.

The Data Supplement is available at https://www.ahajournals.org/doi/suppl/10.1161/CIRCHEARTFAILURE.121.008335.

For Sources of Funding and Disclosures, see page 457.

Correspondence to: Javed Butler, MD, MPH, MBA, Department of Medicine, University of Mississippi Medical Center, 2500 N State St, Jackson, MS 39216. Email

References

  • 1. Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Delling FN, et al. American heart association council on epidemiology and prevention statistics committee and stroke statistics subcommittee. heart disease and stroke statistics-2020 update: a report from the American Heart Association.Circulation. 2020; 141:e139–e596. doi: 10.1161/CIR.0000000000000757. Epub 2020 Jan 29.LinkGoogle Scholar
  • 2. Heidenreich PA, Albert NM, Allen LA, Bluemke DA, Butler J, Fonarow GC, Ikonomidis JS, Khavjou O, Konstam MA, Maddox TM, et al; American Heart Association Advocacy Coordinating Committee; Council on Arteriosclerosis, Thrombosis and Vascular Biology; Council on Cardiovascular Radiology and Intervention; Council on Clinical Cardiology; Council on Epidemiology and Prevention; Stroke Council. Forecasting the impact of heart failure in the United States: a policy statement from the American Heart Association.Circ Heart Fail. 2013; 6:606–619. doi: 10.1161/HHF.0b013e318291329aLinkGoogle Scholar
  • 3. Yancy CW, Jessup M, Bozkurt B, Butler J, Casey DE, Drazner MH, Fonarow GC, Geraci SA, Horwich T, Januzzi JL, et al. American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines.Circulation. 2013; 128:e240–327. doi: 10.1161/CIR.0b013e31829e8776.LinkGoogle Scholar
  • 4. van Walraven C, Jennings A, Forster AJ. A meta-analysis of hospital 30-day avoidable readmission rates.J Eval Clin Pract. 2012; 18:1211–1218. doi: 10.1111/j.1365-2753.2011.01773.xCrossrefMedlineGoogle Scholar
  • 5. Centers for Medicare and Medicaid Services. Readmissions Reduction Program.http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Readmissions-Reduction-Program.html. Accessed 10 Dec 2018.Google Scholar
  • 6. Suter LG, Li SX, Grady JN, Lin Z, Wang Y, Bhat KR, Turkmani D, Spivack SB, Lindenauer PK, Merrill AR, et al. National patterns of risk-standardized mortality and readmission after hospitalization for acute myocardial infarction, heart failure, and pneumonia: update on publicly reported outcomes measures based on the 2013 release.J Gen Intern Med. 2014; 29:1333–1340. doi: 10.1007/s11606-014-2862-5CrossrefMedlineGoogle Scholar
  • 7. Wasfy JH, Zigler CM, Choirat C, Wang Y, Dominici F, Yeh RW. Readmission rates after passage of the hospital readmissions reduction program: a pre-post analysis.Ann Intern Med. 2017; 166:324–331. doi: 10.7326/M16-0185CrossrefMedlineGoogle Scholar
  • 8. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program.N Engl J Med. 2009; 360:1418–1428. doi: 10.1056/NEJMsa0803563CrossrefMedlineGoogle Scholar
  • 9. Akintoye E, Briasoulis A, Egbe A, Dunlay SM, Kushwaha S, Levine D, Afonso L, Mozaffarian D, Weinberger J. National trends in admission and in-hospital mortality of patients with heart failure in the United States (2001-2014).J Am Heart Assoc. 2017; 6:e006955. doi: 10.1161/JAHA.117.006955.LinkGoogle Scholar
  • 10. Blecker S, Herrin J, Li L, Yu H, Grady JN, Horwitz LI. Trends in hospital readmission of medicare-covered patients with heart failure.J Am Coll Cardiol. 2019; 73:1004–1012. doi: 10.1016/j.jacc.2018.12.040CrossrefMedlineGoogle Scholar
  • 11. Desai NR, Ross JS, Kwon JY, Herrin J, Dharmarajan K, Bernheim SM, Krumholz HM, Horwitz LI. Association between hospital penalty status under the hospital readmission reduction program and readmission rates for target and nontarget conditions.JAMA. 2016; 316:2647–2656. doi: 10.1001/jama.2016.18533CrossrefMedlineGoogle Scholar
  • 12. Zuckerman RB, Sheingold SH, Orav EJ, Ruhter J, Epstein AM. Readmissions, Observation, and the Hospital Readmissions Reduction Program.N Engl J Med. 2016; 374:1543–1551. doi: 10.1056/NEJMsa1513024CrossrefMedlineGoogle Scholar
  • 13. Psotka MA, Fonarow GC, Allen LA, Joynt Maddox KE, Fiuzat M, Heidenreich P, Hernandez AF, Konstam MA, Yancy CW, O’Connor CM. The hospital readmissions reduction program: nationwide perspectives and recommendations: a JACC: heart failure position paper.JACC Heart Fail. 2020; 8:1–11. doi: 10.1016/j.jchf.2019.07.012CrossrefMedlineGoogle Scholar
  • 14. Gupta A, Allen LA, Bhatt DL, Cox M, DeVore AD, Heidenreich PA, Hernandez AF, Peterson ED, Matsouaka RA, Yancy CW, et al. Association of the hospital readmissions reduction program implementation with readmission and mortality outcomes in heart failure.JAMA Cardiol. 2018; 3:44–53. doi: 10.1001/jamacardio.2017.4265CrossrefMedlineGoogle Scholar
  • 15. Kumbhani DJ, Fonarow GC, Heidenreich PA, Schulte PJ, Lu D, Hernandez A, Yancy C, Bhatt DL. Association between hospital volume, processes of care, and outcomes in patients admitted with heart failure: insights from get with the guidelines-heart failure.Circulation. 2018; 137:1661–1670. doi: 10.1161/CIRCULATIONAHA.117.028077.LinkGoogle Scholar
  • 16. Gracia E, Singh P, Collins S, Chioncel O, Pang P, Butler J. The vulnerable phase of heart failure.Am J Ther. 2018; 25:e456–e464. doi: 10.1097/MJT.0000000000000794CrossrefMedlineGoogle Scholar
  • 17. Greene SJ, Fonarow GC, Vaduganathan M, Khan SS, Butler J, Gheorghiade M. The vulnerable phase after hospitalization for heart failure.Nat Rev Cardiol. 2015; 12:220–229. doi: 10.1038/nrcardio.2015.14CrossrefMedlineGoogle Scholar
  • 18. Wadhera RK, Joynt Maddox KE, Wasfy JH, Haneuse S, Shen C, Yeh RW. Association of the hospital readmissions reduction program with mortality among medicare beneficiaries hospitalized for heart failure, acute myocardial infarction, and pneumonia.JAMA. 2018; 320:2542–2552. doi: 10.1001/jama.2018.19232CrossrefMedlineGoogle Scholar
  • 19. Hsuan C, Carr BG, Hsia RY, Hoffman GJ. Assessment of hospital readmissions from the emergency department after implementation of medicare’s hospital readmissions reduction program.JAMA Netw Open. 2020; 3:e203857. doi: 10.1001/jamanetworkopen.2020.3857CrossrefMedlineGoogle Scholar
  • 20. Wadhera RK, Joynt Maddox KE, Kazi DS, Shen C, Yeh RW. Hospital revisits within 30 days after discharge for medical conditions targeted by the hospital readmissions reduction program in the United States: national retrospective analysis.BMJ. 2019; 366:l4563. doi: 10.1136/bmj.l4563CrossrefMedlineGoogle Scholar
  • 21. Kilgore M, Patel HK, Kielhorn A, Maya JF, Sharma P. Economic burden of hospitalizations of medicare beneficiaries with heart failure.Risk Manag Healthc Policy. 2017; 10:63–70. doi: 10.2147/RMHP.S130341CrossrefMedlineGoogle Scholar
  • 22. Centers for Medicare & Medicaid Services (CMS).Medicare program; advancing care coordination through Episode Payment Models (EPMs); cardiac rehabilitation incentive payment model; and changes to the comprehensive care for joint replacement model.Fed Regist. 2017; 82:180–651.MedlineGoogle Scholar
  • 23. Khera R, Wang Y, Bernheim SM, Lin Z, Krumholz HM. Post-discharge acute care and outcomes following readmission reduction initiatives: national retrospective cohort study of Medicare beneficiaries in the United States.BMJ. 2020; 368:l6831. doi: 10.1136/bmj.l6831CrossrefMedlineGoogle Scholar
  • 24. Jackson SL, Tong X, King RJ, Loustalot F, Hong Y, Ritchey MD. National burden of heart failure events in the United States, 2006 to 2014.Circ Heart Fail. 2018; 11:e004873. doi: 10.1161/CIRCHEARTFAILURE.117.004873LinkGoogle Scholar
  • 25. Feltner C, Jones CD, Cené CW, Zheng ZJ, Sueta CA, Coker-Schwimmer EJ, Arvanitis M, Lohr KN, Middleton JC, Jonas DE. Transitional care interventions to prevent readmissions for persons with heart failure: a systematic review and meta-analysis.Ann Intern Med. 2014; 160:774–784. doi: 10.7326/M14-0083CrossrefMedlineGoogle Scholar
  • 26. Van Spall HGC, Lee SF, Xie F, Oz UE, Perez R, Mitoff PR, Maingi M, Tjandrawidjaja MC, Heffernan M, Zia MI, et al. Effect of patient-centered transitional care services on clinical outcomes in patients hospitalized for heart failure: the PACT-HF randomized clinical trial.JAMA. 2019; 321:753–761. doi: 10.1001/jama.2019.0710CrossrefMedlineGoogle Scholar
  • 27. Pandor A, Gomersall T, Stevens JW, Wang J, Al-Mohammad A, Bakhai A, Cleland JG, Cowie MR, Wong R. Remote monitoring after recent hospital discharge in patients with heart failure: a systematic review and network meta-analysis.Heart. 2013; 99:1717–1726. doi: 10.1136/heartjnl-2013-303811CrossrefMedlineGoogle Scholar