[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.
Mean and Predicted Hospital-Level Quarterly Readmission Rates for Targeted and Nontargeted Conditions Over Time
Mean and Predicted Hospital-Level Quarterly Readmission Rates for Targeted and Nontargeted Conditions Over Time

Dotted lines indicated start and end of the performance period of the Hospital Readmissions Reduction Program.

Figure 2.
Mean and Predicted Hospital-Level Quarterly Observation Stay Rates for Targeted and Nontargeted Conditions Over Time
Mean and Predicted Hospital-Level Quarterly Observation Stay Rates for Targeted and Nontargeted Conditions Over Time

Dotted lines indicated start and end of the performance period of the Hospital Readmissions Reduction Program.

Table 1.  
Patient and Hospital Characteristics for Targeted (Total Knee Arthroplasty and Total Hip Arthroplasty) and Nontargeted (Abdominal Aortic Aneurysm Repair, Colectomy, and Lung Resection) Procedures Across Study Periods
Patient and Hospital Characteristics for Targeted (Total Knee Arthroplasty and Total Hip Arthroplasty) and Nontargeted (Abdominal Aortic Aneurysm Repair, Colectomy, and Lung Resection) Procedures Across Study Periods
Table 2.  
Risk-Adjusted Beneficiary-Level Readmission Rates for Each Procedure and Changes in Likelihood of Readmission Over Timea
Risk-Adjusted Beneficiary-Level Readmission Rates for Each Procedure and Changes in Likelihood of Readmission Over Timea
1.
Gonzalez  AA, Abdelsattar  ZM, Dimick  JB, Dev  S, Birkmeyer  JD, Ghaferi  AA.  Time-to-readmission and mortality after high-risk surgery.  Ann Surg. 2015;262(1):53-59.PubMedGoogle ScholarCrossref
2.
Varacallo  MA, Herzog  L, Toossi  N, Johanson  NA.  Ten-year trends and independent risk factors for unplanned readmission following elective total joint arthroplasty at a large urban academic hospital.  J Arthroplasty. 2016.PubMedGoogle Scholar
3.
Dharmarajan  K, Hsieh  AF, Lin  Z,  et al.  Diagnoses and timing of 30-day readmissions after hospitalization for heart failure, acute myocardial infarction, or pneumonia.  JAMA. 2013;309(4):355-363.PubMedGoogle ScholarCrossref
4.
Jencks  SF, Williams  MV, Coleman  EA.  Rehospitalizations among patients in the Medicare fee-for-service program [published correction appears in N Engl J Med. 2011;364(16):1582].  N Engl J Med. 2009;360(14):1418-1428.PubMedGoogle ScholarCrossref
5.
Dimick  JB, Ghaferi  AA.  Hospital readmission as a quality measure in surgery.  JAMA. 2015;313(5):512-513.PubMedGoogle ScholarCrossref
6.
Centers for Medicare & Medicaid Services. Readmissions Reduction Program (HRRP). https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Readmissions-Reduction-Program.html. Published 2016. Accessed March 11, 2017.
7.
Rau  J. Half of nation’s hospitals fail again to escape Medicare’s readmission penalties. http://khn.org/news/half-of-nations-hospitals-fail-again-to-escape-medicares-readmission-penalties/. Published August 3, 2015. Accessed March 15, 2017.
8.
Boccuti  C. Aiming for fewer hospital U-turns: the Medicare Hospital Readmission Reduction Program. http://kff.org/medicare/issue-brief/aiming-for-fewer-hospital-u-turns-the-medicare-hospital-readmission-reduction-program/. Published March 10, 2017. Accessed March 15, 2017.
9.
Zuckerman  RB, Sheingold  SH, Orav  EJ, Ruhter  J, Epstein  AM.  Readmissions, observation, and the Hospital Readmissions Reduction Program.  N Engl J Med. 2016;374(16):1543-1551.PubMedGoogle ScholarCrossref
10.
Desai  NR, Ross  JS, Kwon  JY,  et al.  Association between hospital penalty status under the Hospital Readmission Reduction Program and readmission rates for target and nontarget conditions.  JAMA. 2016;316(24):2647-2656.PubMedGoogle ScholarCrossref
11.
Centers for Medicare & Medicaid Services. Comprehensive Care for Joint Replacement (CJR) model. https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2015-Fact-sheets-items/2015-11-16.html. Published November 16, 2015. Accessed March 15, 2017.
12.
Hicks  CW, Bronsert  M, Hammermeister  KE,  et al.  Operative variables are better predictors of postdischarge infections and unplanned readmissions in vascular surgery patients than patient characteristics.  J Vasc Surg. 2017;65(4):1130-1141.PubMedGoogle ScholarCrossref
13.
Gani  F, Lucas  DJ, Kim  Y, Schneider  EB, Pawlik  TM.  Understanding variation in 30-day surgical readmission in the era of accountable care: effect of the patient, surgeon, and surgical subspecialties.  JAMA Surg. 2015;150(11):1042-1049.PubMedGoogle ScholarCrossref
14.
Kim  Y, Gani  F, Lucas  DJ,  et al.  Early versus late readmission after surgery among patients with employer-provided health insurance.  Ann Surg. 2015;262(3):502-511.PubMedGoogle ScholarCrossref
15.
Jenq  GY, Doyle  MM, Belton  BM, Herrin  J, Horwitz  LI.  Quasi-experimental evaluation of the effectiveness of a large-scale readmission reduction program.  JAMA Intern Med. 2016;176(5):681-690.PubMedGoogle ScholarCrossref
16.
Jack  BW, Chetty  VK, Anthony  D,  et al.  A reengineered hospital discharge program to decrease rehospitalization: a randomized trial.  Ann Intern Med. 2009;150(3):178-187.PubMedGoogle ScholarCrossref
17.
Hamar  B, Rula  EY, Wells  AR, Coberley  C, Pope  JE, Varga  D.  Impact of a scalable care transitions program for readmission avoidance.  Am J Manag Care. 2016;22(1):28-34.PubMedGoogle Scholar
18.
Feigenbaum  P, Neuwirth  E, Trowbridge  L,  et al.  Factors contributing to all-cause 30-day readmissions: a structured case series across 18 hospitals.  Med Care. 2012;50(7):599-605.PubMedGoogle ScholarCrossref
19.
Tsai  TC, Joynt  KE, Orav  EJ, Gawande  AA, Jha  AK.  Variation in surgical-readmission rates and quality of hospital care.  N Engl J Med. 2013;369(12):1134-1142.PubMedGoogle ScholarCrossref
20.
Brooke  BS, Goodney  PP, Kraiss  LW, Gottlieb  DJ, Samore  MH, Finlayson  SR.  Readmission destination and risk of mortality after major surgery: an observational cohort study.  Lancet. 2015;386(9996):884-895.PubMedGoogle ScholarCrossref
21.
Tsai  TC, Orav  EJ, Jha  AK.  Care fragmentation in the postdischarge period: surgical readmissions, distance of travel, and postoperative mortality.  JAMA Surg. 2015;150(1):59-64.PubMedGoogle ScholarCrossref
22.
Barrett  ML, Fingar  KR, Owens  PL, Stocks  C, Steiner  C, Sheng  M. Identifying observation services in the Healthcare Cost and Utilization Project (HCUP) state databases. HCUP Methods Series Report 2015-05. US Agency for Healthcare Research and Quality. http://www.hcup-us.ahrq.gov/reports/methods/methods.jsp. Published September 1, 2015. Accessed March 28, 2017.
23.
Feng  Z, Wright  B, Mor  V.  Sharp rise in Medicare enrollees being held in hospitals for observation raises concerns about causes and consequences.  Health Aff (Millwood). 2012;31(6):1251-1259.PubMedGoogle ScholarCrossref
24.
Stone  J, Hoffman  GJ; Congressional Research Service. Medicare hospital readmissions: issues, policy options and PPACA. http://www.hhnmag.com/ext/resources/inc-hhn/pdfs/resources/opencrs-R40972_20100921.pdf. Published September 21, 2010. Accessed March 28, 2017.
25.
Pope  GC, Kautter  J, Ellis  RP,  et al.  Risk adjustment of Medicare capitation payments using the CMS-HCC model.  Health Care Financ Rev. 2004;25(4):119-141.PubMedGoogle Scholar
26.
Norton  EC, Bieler  GS, Ennett  ST, Zarkin  GA.  Analysis of prevention program effectiveness with clustered data using generalized estimating equations.  J Consult Clin Psychol. 1996;64(5):919-926.PubMedGoogle ScholarCrossref
27.
Dorsey  K, Grady JN, Horwitz  LI, et al. 2016 All-Cause Hospital-Wide Measure Updates and Specifications Report: Hospital-Level 30-Day Risk-Standardized Readmission Measure. Version 5.0. New Haven, CT: Yale New Haven Health Services Corp/Center for Outcomes Research & Evaluation; March 2016. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Measure-Methodology.html. Accessed October 11, 2017.
28.
Wagner  AK, Soumerai  SB, Zhang  F, Ross-Degnan  D.  Segmented regression analysis of interrupted time series studies in medication use research.  J Clin Pharm Ther. 2002;27(4):299-309.PubMedGoogle ScholarCrossref
29.
Penfold  RB, Zhang  F.  Use of interrupted time series analysis in evaluating health care quality improvements.  Acad Pediatr. 2013;13(6)(suppl):S38-S44.PubMedGoogle ScholarCrossref
30.
Dorsey  K, Grady JN, Desai  N,  et al. 2016 Procedure-Specific Measures Updates and Specifications Report: Hospital-Level 30-Day Risk-Standardized Readmission Measures. Version 5.0. New Haven, CT: Yale New Haven Health Services Corp/Center for Outcomes Research & Evaluation; March 2016. https://www.qualitynet.org/dcs/ContentServer?cid=1228774371008&pagename=QnetPublic%2FPage%2FQnetTier4&c=Page. Accessed October 11, 2017.
31.
Benbassat  J, Taragin  M.  Hospital readmissions as a measure of quality of health care: advantages and limitations.  Arch Intern Med. 2000;160(8):1074-1081.PubMedGoogle ScholarCrossref
32.
Greenblatt  DY, Weber  SM, O’Connor  ES, LoConte  NK, Liou  JI, Smith  MA.  Readmission after colectomy for cancer predicts one-year mortality.  Ann Surg. 2010;251(4):659-669.PubMedGoogle ScholarCrossref
33.
Lucas  DJ, Haider  A, Haut  E,  et al.  Assessing readmission after general, vascular, and thoracic surgery using ACS-NSQIP.  Ann Surg. 2013;258(3):430-439.PubMedGoogle ScholarCrossref
34.
Centers for Medicare & Medicaid Services. Hospital Compare datasets. https://data.medicare.gov/data/hospital-compare. Updated July 26, 2017. Accessed June 6, 2017.
35.
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(5):324-331.PubMedGoogle ScholarCrossref
36.
McGarry  BE, Blankley  AA, Li  Y.  The impact of the Medicare Hospital Readmission Reduction Program in New York State.  Med Care. 2016;54(2):162-171.PubMedGoogle ScholarCrossref
37.
Joynt  KE, Figueroa  JE, Oray  J, Jha  AK.  Opinions on the Hospital Readmission Reduction Program: results of a national survey of hospital leaders.  Am J Manag Care. 2016;22(8):e287-e294.PubMedGoogle Scholar
38.
Ellimoottil  C, Ryan  AM, Hou  H, Dupree  JM, Hallstrom  B, Miller  DC.  Implications of the definition of an episode of care used in the Comprehensive Care for Joint Replacement model.  JAMA Surg. 2017;152(1):49-54.PubMedGoogle ScholarCrossref
39.
Krell  RW, Girotti  ME, Fritze  D, Campbell  DA, Hendren  S.  Hospital readmissions after colectomy: a population-based study.  J Am Coll Surg. 2013;217(6):1070-1079.PubMedGoogle ScholarCrossref
40.
Orr  NT, El-Maraghi  S, Korosec  RL, Davenport  DL, Xenos  ES.  Cost analysis of vascular readmissions after common vascular procedures.  J Vasc Surg. 2015;62(5):1281-1287.e1.PubMedGoogle ScholarCrossref
41.
Ricciardi  BF, Oi  KK, Daines  SB, Lee  YY, Joseph  AD, Westrich  GH.  Patient and perioperative variables affecting 30-day readmission for surgical complications after hip and knee arthroplasties: a matched cohort study.  J Arthroplasty. 2017;32(4):1074-1079.PubMedGoogle Scholar
42.
Venkatesh  AK, Wang  C, Ross  JS,  et al.  Hospital use of observation stays: cross-sectional study of the impact on readmission rates.  Med Care. 2016;54(12):1070-1077.PubMedGoogle ScholarCrossref
43.
Centers for Medicare & Medicaid Services. Accountable care organizations (ACOs): general information. https://innovation.cms.gov/initiatives/aco/. Updated September 5, 2017. Accessed June 6, 2017.
Original Investigation
March 2018

Association of the Hospital Readmissions Reduction Program With Surgical Readmissions

Author Affiliations
  • 1Dow Health Services Research Division, Department of Urology, University of Michigan, Ann Arbor
  • 2Center for Clinical Management Research, Veterans Affairs Ann Arbor Healthcare System, Ann Arbor, Michigan
  • 3Department of Health Management and Policy, University of Michigan, Ann Arbor
  • 4Department of Economics, University of Michigan, Ann Arbor
  • 5National Bureau of Economic Research, Cambridge, Massachusetts
  • 6Center for Healthcare Outcomes and Policy, Department of Surgery, University of Michigan, Ann Arbor
  • 7Surgical Innovation Editor, JAMA Surgery
  • 8Kidney Epidemiology Cost Center, University of Michigan, Ann Arbor
JAMA Surg. 2018;153(3):243-250. doi:10.1001/jamasurg.2017.4585
Key Points

Question  What is the association of the Hospital Readmissions Reduction Program with readmissions after surgical procedures targeted by the policy and other major procedures with historically high readmission rates not under its purview?

Findings  In this cohort study of 672 135 Medicare beneficiaries treated at 2773 hospitals, we found that readmissions decreased significantly for all procedures during the study period. Corresponding with the performance period of the policy, readmissions after targeted procedures decreased significantly faster compared with 2 earlier periods, while readmission rates for nontargeted procedures stabilized.

Meaning  The Hospital Readmissions Reduction Program may decrease readmissions for targeted procedures; however, there were no associated spillover effects for common nontargeted procedures.

Abstract

Importance  Readmissions after surgery lead to poor patient outcomes and increased costs. The Hospital Readmissions Reduction Program (HRRP) penalizes hospitals with excess readmissions after specified medical and surgical discharges.

Objective  To evaluate the association of the HRRP with readmissions after major joint surgery (targeted) and procedures with historically high rates not under its purview (nontargeted).

Design, Setting, and Participants  In this population-based analysis using a 20% Medicare sample, a retrospective cohort study was performed of patients undergoing one of 5 major surgical procedures between January 1, 2006, and November 30, 2014. The study included 507 663 patients with targeted (total knee arthroplasty and total hip arthroplasty) and 164 472 patients with nontargeted (abdominal aortic aneurysm repair, colectomy, and lung resection) procedures performed at 2773 hospitals.

Exposure  Implementation of the HRRP policy.

Main Outcomes and Measures  Hospital-level 30-day risk-adjusted rates of readmission and observation stays were calculated using multivariable logistic regression models. Changes in these rates were analyzed for 3 distinct periods (prepolicy [January 1, 2006, to June 30, 2010], performance [July 1, 2010, to June 30, 2013], and penalty [July 1, 2013, to November 30, 2014]) corresponding to the HRRP implementation timeline for major joint surgery using interrupted time series.

Results  Among 672 135 Medicare beneficiaries 66 years or older treated at 2773 hospitals, readmissions for all procedures decreased significantly over the study period. Readmission rates after targeted procedures decreased faster during the performance period (slope, −0.060; 95% CI, −0.079 to −0.041) compared with the prepolicy period (slope, −0.012; 95% CI, −0.027 to 0.034) (P < .002). For the nontargeted procedures, readmission rates were decreasing during the prepolicy period (slope, −0.200; 95% CI, −0.240 to −0.160) but stabilized during the performance period (slope, 0.008; 95% CI, −0.049 to 0.066 (P < .001). The use of observation stays increased slightly, accounting for 11% of the decrease in readmissions.

Conclusions and Relevance  The HRRP effectively decreased readmissions for targeted procedures. There were no associated spillover effects for common nontargeted procedures. A better understanding of differences in the association of the policy with medical and surgical discharges will be necessary to further enhance its generalizability.

Introduction

Reducing hospital readmissions is a priority for both policymakers and clinical leaders. Although some readmissions may be unavoidable, unnecessary readmissions are emblematic of poor quality1-5 and are estimated to cost the Centers for Medicare & Medicaid Services (CMS) up to $17 billion annually.4 In efforts to promote better care coordination and reduce wasteful spending, the CMS began penalizing hospitals with readmission rates greater than the national average after discharges for acute myocardial infarction, pneumonia, and congestive heart failure through the Hospital Readmissions Reduction Program (HRRP).6 Penalties, which can be as high as 3% of Medicare payments,7 were levied against three-quarters of hospitals in 2016, totaling more than $1.4 billion.8 Prior studies9,10 have shown that the policy was successful at reducing readmissions for discharges of targeted and, to a lesser degree, nontargeted medical conditions.

While the HRRP has made early inroads into reducing readmissions after medical discharges, its ability to do so for targeted surgical procedures is unclear. In 2014, the policy was expanded to include major joint surgery (elective total knee arthroplasty and total hip arthroplasty), which is common in the elderly and, at $7 billion annually, constitutes a substantial expense to Medicare.11 However, in contrast to readmissions after medical discharges, readmissions for patients after elective surgery are more likely a consequence of procedure-related complications12-14 as opposed to failure of care coordination. Systems-level interventions, including discharge navigators, care transition programs, and medication reconciliation by pharmacists,15-17 that have proved effective at reducing readmissions for patients with chronic medical diseases may not be generalizable to those recovering from surgery. As such, the reasons for readmission differ from those for medical conditions, as does the degree to which they are preventable.18,19

For these reasons, we evaluated the association of the HRRP with readmission rates after major joint surgery. To assess possible spillover effects, we also investigated the extent to which the policy was associated with readmissions after nontargeted procedures. These nontargeted procedures are common among the elderly and are associated with a high risk of readmission.1,20,21

Methods
Study Population

We performed a retrospective cohort study of fee-for-service Medicare beneficiaries undergoing one of 5 major surgical procedures between January 1, 2006, and November 30, 2014, using a 20% national sample. We selected the 2 major joint procedures targeted by the HRRP (total knee arthroplasty and total hip arthroplasty) and 3 nontargeted procedures (abdominal aortic aneurysm repair, colectomy, and lung resection) (eTable 1 in the Supplement). The nontargeted procedures were chosen because they are common in the elderly, costly, and associated with high readmission rates, implying that they have considerable potential to improve from systems-based efforts implemented in response to the HRRP. The use of these procedures provides insight into possible spillover effects of the policy. Coronary artery bypass graft was purposely omitted from our evaluation because it became targeted by the policy in the latter portion of our study period. We included patients 66 years or older to facilitate risk adjustment. Only those with continuous enrollment in both Medicare Part A and B from 1 year before each procedure through 30 days after the discharge were included. Patients participating in Medicare Advantage plans were excluded due to the absence of complete claims. Consistent with the policy guidelines, we further limited our study population to patients undergoing each procedure in hospitals with a minimum volume of 25 procedures annually.6 The study protocol was judged to be exempt from approval by the institutional review board of the University of Michigan, and participant informed consent was not required.

Outcomes

Consistent with the HRRP policy, our primary outcome was hospital-level readmissions within 30 days of discharge from a procedure. We also measured hospital-level rates of observation stays to determine if potential changes in readmission resulted from greater use of this classification (ie, not penalized by the policy) as opposed to reducing overall returns to the hospital. We identified observation stays occurring within 30 days of discharge and not resulting in a subsequent admission (patients admitted after an observation stay are billed only for the admission in Medicare claims) using the center code 0762 for “observation room” revenue.22,23 Only the first readmission or observation stay was included in the analysis. Patients dying during the hospital stay for a procedure were censored from the readmission and observation stay calculations.

Statistical Analysis

We defined the following 3 periods coinciding with the HRRP timeline for major joint surgery: prepolicy (before the inclusion of major joint surgery as an HRRP performance measure [January 1, 2006, to June 30, 2010]), performance (the 3-year measurement period used to calculate initial penalties [July 1, 2010, to June 30, 2013]), and penalty (the period in which initial penalties were levied and during which measurement occurred for subsequent years [July 1, 2013, to November 30, 2014]).6 Although the announcement to include total knee arthroplasty and total hip arthroplasty in the HRRP was not made until August 19, 2013, these procedures were included in the initial cohort of potential targets for the policy presented on September 21, 2010, to Congress.24 For this reason, we chose to define our periods of interest to match those of the HRRP policy for major joint surgery.

We calculated risk-adjusted quarterly hospital readmission and observation stay rates, consistent with the HRRP policy guidelines. First, we modeled rates for each procedure separately at the beneficiary level. To do this, we fit multivariable logistic regression models for each of our 5 procedures, adjusting for age and hierarchical condition category risk score.25 Generalized estimating equations with robust standard errors were used to account for clustering within hospitals, which was assumed to have independent correlation.26 For each hospital, the procedure-specific quarterly observed readmission rate was divided by the predicted rate and then multiplied by the average readmission rate for all patients undergoing that procedure during the study period. Risk-adjusted observation stay rates were calculated in a similar fashion. For each procedure, we calculated the changes in likelihood of readmission at the beneficiary level for the performance and penalty periods compared with the prepolicy period. We then combined procedure cohorts using a geometric mean for major joint and nontargeted procedures similar to the hospital-wide readmission rate method by the CMS27 to produce targeted and nontargeted procedure rates.

To examine trends in the quarterly hospital-level risk-adjusted rates of readmission and observation stays, we used interrupted time series analysis, which allows for evaluation of specific interventions in a long-term trend.28,29 To test for changes in readmission and observation stay rates over time, we used the calculated risk-adjusted hospital quarterly readmission and observation rates as the outcome variables in linear regression models. We fitted linear splines for each study period (prepolicy, performance, and penalty) and compared the slopes for each procedure over time. For the combined procedure cohorts, models included interaction terms for targeted vs nontargeted procedures, each of the seasonal indicators, linear time, and linear spline variables. In addition to comparing slopes between study periods within targeted and nontargeted procedures, we also compared slopes of targeted vs nontargeted procedures within each period.

All analyses were performed using statistical software (SAS, version 9.4; SAS Institute Inc). All statistical tests were 2 tailed, and the probability of type I error was set at .05.

Sensitivity Analyses

We performed several sensitivity analyses to test the stability of our results. First, we tested the association of our risk adjustment with our findings. We chose to use the hierarchical condition category risk score as our primary method of risk adjustment because this method allowed for a uniform measure across all procedures. For the sensitivity analysis, we re-ran our analysis using non–risk-adjusted readmission and observation rates, and we also performed comorbidity adjustment, similar to the CMS readmission measures. Because joint surgery is risk adjusted differently than other procedures, we used a procedure-specific hospital-level readmission measure30 for joints; for the remaining procedures, we used the hospital-wide readmission measure27 (eTables 2, 3, and 4 in the Supplement). We then removed the minimum hospital volume threshold for a given procedure and instead adjusted our models for quarterly hospital volume. Second, we combined readmission and observation stay rates and assessed for changes in the trends (eFigure 1 in the Supplement). Third, to better understand the trends in readmissions for the nontargeted conditions in the prepolicy period, we conducted our analyses separately based on surgical technique (open vs minimally invasive approaches) (eTable 5 and eFigure 2 in the Supplement). In all sensitivity analyses, the results did not differ from the findings of the primary analyses.

Results

A total of 672 135 Medicare beneficiaries treated at 2773 hospitals were included in our analysis. Patient and hospital characteristics for both targeted and nontargeted conditions are listed in Table 1. Over the study period, there were 358 655 total knee arthroplasties, 148 250 total hip arthroplasties, 25 020 abdominal aortic aneurysm repairs, 106 088 colectomies, and 23 320 lung resections performed. For major joint surgical procedures, 50.5% were performed in the prepolicy, 32.7% in the performance, and 16.8% in the penalty periods. Similarly, for the nontargeted procedures, 55.4% were performed in the prepolicy, 31.3% in the performance, and 13.3% in the penalty periods.

We found a significant decline in the beneficiary-level risk-adjusted readmission rate for all procedures in both the performance and penalty periods compared with the prepolicy period (Table 2). This decline ranged from 23% to 11%, with the largest decrease over the course of the study occurring after total hip arthroplasty (absolute reduction, 0.7%; adjusted odds ratio, 0.77; 95% CI, 0.72-0.82), while colectomy had the smallest decrease (absolute reduction, 1.6%; adjusted odds ratio, 0.89; 95% CI, 0.84-0.94). Trends in risk-adjusted hospital-level quarterly readmission rates for each procedure are shown in eFigure 3 in the Supplement.

The risk-adjusted hospital-level quarterly readmission rate declined significantly for both major joint surgery and nontargeted procedures from 2006 to 2014 (Figure 1). In the prepolicy period, the readmission rate for major joint procedures was flat (slope, −0.012; 95% CI, −0.027 to 0.034). This trend changed significantly during the performance period, with readmission rates decreasing at a quarterly rate of −0.060 (95% CI, −0.079 to −0.041), leading to a difference in slopes between the 2 periods of −0.049 (95% CI, −0.079 to −0.019) (P < .002). Conversely, we noted a rapid decrease in the readmission rate for the nontargeted procedures during the prepolicy period (slope, −0.200; 95% CI, −0.240 to −0.160) that was partly explained by greater use of minimally invasive surgery, which was associated with lower rates of readmission (eTable 4 in the Supplement). Nonetheless, readmission rates plateaued during the performance period (slope, 0.008; 95% CI, −0.049 to 0.066). Again, this result represented a significant difference between the 2 periods (slope difference, 0.210; 95% CI, 0.120-0.300) (P < .001). The readmission rate decreased for both groups during the penalty period; however, the rate of decrease for both major joint surgery and nontargeted procedures was not significant compared with the performance period (difference in slope, 0.029; 95% CI, −0.033 to 0.086 and −0.100; 95% CI, −0.300 to 0.093, respectively).

As shown in Figure 2, the use of observation stays increased over the study period for major joint surgery and nontargeted procedures. Rates by procedure are shown in eFigure 4 in the Supplement. After major joint surgery, we noted significant increases in the rate of observation stays during the prepolicy (slope, 0.005; 95% CI, 0.002-0.008) and performance (slope, 0.007; 95% CI, 0.002-0.012) periods. This trend tapered during the penalty period (slope, −0.014; 95% CI −0.029 to 0.000). The change in slope was statistically significant between the performance and penalty periods (difference in slope, 0.210; 95% CI, 0.039-0.003). However, the magnitude of the absolute increase in the rate of observation stays during the performance period was much smaller (0.08%) than the decrease in the readmission rate (0.66%) after major joint surgery, meaning that the use of observation stays accounts for 11% of the decrease in readmissions. The use of observation stays increased significantly for the nontargeted procedures only during the performance period (slope, 0.023; 95% CI, 0.091-0.036).

Discussion

Readmissions decreased significantly after major joint surgery and 3 other common procedures over the course of our study. Although readmission rates were decreasing well before the introduction of the HRRP, the change for the targeted major joint surgery was subtle and not statistically significant during this time. However, coinciding with the policy, readmissions decreased rapidly for major joint surgery compared with both prepolicy rates and with the observed change in nontargeted conditions. Interventions that resulted in this significant decrease for major joint surgery targeted by the policy do not appear to have spillover effects for other procedures examined. Readmission rates for these nontargeted procedures plateaued after the introduction of the HRRP. The increased use of observation stays, perhaps a means to circumvent penalties by the policy, contributed only minimally to the decrease in readmissions after major joint surgery.

The focus on readmission reduction by professional organizations and policymakers is consistent with reductions in readmission rates noted before implementation of the HRRP policy.1,2,5,31,32 In 2011, the American College of Surgeons National Surgical Quality Improvement Program began tracking readmissions.33 Medicare had also placed an emphasis on readmission reduction well before the introduction of the HRRP policy through the Hospital Compare program, which privately reported readmission data for select medical conditions to hospitals in 2009 and in 2010 began publicly reporting this information.34 The broader dissemination of minimally invasive surgery resulted in significant decreases in readmission rates for the nontargeted procedures. Because these nontargeted conditions had high baseline readmission rates, they served as natural targets for readmission reduction efforts. These included both systems-based interventions and substitution of surgical technology associated with lower risk of readmissions.

We noted a sharp decline in the slope for readmission after major joint surgery coinciding with the start of the HRRP measurement period. This finding aligns well with studies9,10,35,36 on targeted medical conditions that found similar declines. These results are not surprising in light of survey data showing that 66% of hospital leaders thought that the HRRP had a major influence on systems efforts to reduce readmissions.37 In addition, our data are consistent with a previous study10 on targeted medical conditions showing that hospitals concentrated efforts to reduce readmissions for targeted conditions in response to the threat of penalties because hospitals penalized under the policy decreased readmissions significantly more for targeted conditions than for nontargeted ones. The difference in the readmission trends between major joint surgery and nontargeted procedures in our study supports the notion that hospitals responded to the threat of financial penalties and diverted resources to prevention of readmission after targeted major joint and medical conditions at the expense of efforts directed toward procedures traditionally with much higher rates (ie, nontargeted procedures). An alternative explanation would be that readmissions after the nontargeted procedures reached a “floor rate.” However, the abrupt, opposing trends for targeted and nontargeted surgery at the start of the HRRP measurement period is more consistent with a shift in priorities rather than a tapering effectiveness of previous efforts, which would be expected to be gradual.

The beneficial association of the HRRP with targeted procedures and the absence of spillover effects for the nontargeted procedures are contrary to the policy associations noted previously after medical discharges.9,10,36 Prior analyses of the HRRP relative to medical conditions found less pronounced but commensurate changes in readmission trends for the nontargeted conditions.9,10,36 This spillover benefit was absent for surgical conditions in which readmission rates for our nontargeted procedures largely plateaued. This finding points to the marked heterogeneity of the surgical population both in terms of baseline function (patients undergoing the nontargeted procedures were older, with more comorbidities) and postoperative needs. In further contrast to medical conditions, readmissions after surgery are likely to result from novel, acute complications as opposed to exacerbations of chronic conditions. These complications are specific to each surgery and differ widely between procedures.38-41 As such, broad systems-level interventions designed to reduce interventions in medical patients may be less effective in a surgical population, and the benefits of procedure-specific interventions may be less transferrable across surgical procedures.

The trends in readmissions were only minimally affected by the increased use of observation stays. In this study, we confirm that previously reported increases in the use of observation stays also extend to surgical patients.9,23,42 We noted a significant increase in the rate of observation stays both in the prepolicy and performance periods, but there were no differences in this rate coinciding with the HRRP policy time frame. This finding likely represents a continuation of the secular trend to capitalize on better value (similar care at lower cost) of observation stays for some complications rather than a response by hospitals to divert potential readmissions and is in keeping with earlier studies.9,42

Limitations

Our findings should be considered in the context of several limitations. First, we were unable to separate the associations of the HRRP from other readmission reduction efforts (continued efforts by clinical leaders5 and accountable care organizations43) implemented over a similar time frame. However, we noted significant changes in trends coinciding with specific dates in the implementation of the HRRP for both targeted and nontargeted procedures, making other explanations for our findings less likely. Second, it is possible that readmission reduction efforts were already under way before the start of the performance period for surgical conditions, thus tempering the observed association; however, this outcome is unlikely. The measurement period for targeted surgical conditions began only 4 months after initial announcement of the HRRP. The implementation of any substantial readmission reduction efforts in this short time frame is doubtful. In addition, there was no significant difference in readmissions for targeted procedures before the introduction of the HRRP. Third, because the addition of major joint surgery was not announced until 2013 (2 years after the announcement of the 3 initial medical conditions), it is possible that spillover benefits from efforts targeting medical conditions accounted for some of the observed association with surgical procedures. However, in the absence of targeted efforts by hospitals to reduce readmissions after major joint surgery, we would expect a similar magnitude in spillover effects for our targeted and nontargeted procedures. The contrast in trends between these 2 groups suggests that hospitals were anticipating the inclusion of major joint surgery to the policy-directed readmission reduction efforts accordingly. Fourth, because our models parallel the CMS readmission models, it is possible that the omission of some clinical and demographic information may have led to omitted variable bias: that is, patients undergoing elective surgery during the performance period may have had lower risk of readmission, which is not entirely accounted for by the risk adjustment (“cherry picking”). However, this is unlikely because the number of total knee arthroplasties and total hip arthroplasties increased during each year of the policy. Fifth, our analysis represents the early results of the HRRP on surgical procedures. Because the first set of penalties for major joint surgery was levied near the end of our study period, it is possible that the trends in readmissions may have been altered as hospitals actually experienced the financial ramifications of these penalties. Despite this early time frame and similar to the evaluation of medical conditions,9,10 we found that the greatest changes in the readmission rates occurred during the performance period, with diminished association afterward, suggesting possible limitations in the capacity of hospitals to further decrease readmissions with current approaches.

Despite these limitations, our findings have significant implications. To our knowledge, this study is the first analysis to demonstrate the ability of the HRRP to reduce readmissions after targeted surgical procedures. This study confirms the ability of policies leveraging financial penalties to drive clinical changes in a surgical population. Similar to previous work,9,10 we noted muted associations with the policy over time, raising questions of whether basal rates of readmissions have been reached and if continued financial pressures can remain effective drivers of change. Contrary to what was noted for medical conditions,9,10 we found no spillover benefit in readmission reduction for the nontargeted procedures. To maximize the benefit of this policy, it is necessary to cultivate greater understanding of factors that drove down readmission in targeted procedures and to develop strategies to scale these interventions across all surgical patients. Finally, the number of truly preventable readmissions after surgical procedures may differ significantly from those of medical conditions. Early recognition of complications and readmission, thus limiting further clinical deterioration, may lead to both improved outcomes and decreased costs.

Conclusions

The readmission rate for major joint surgery targeted by the HRRP decreased significantly over the study period, with the largest changes coinciding with policy landmarks. While both targeted and nontargeted procedures experienced a significant reduction in readmissions, there was a lack of spillover benefit of the policy to the nontargeted procedures. The increased use of observation stays had a minimal association with the readmission trends. A better understanding of differences in the association of the HRRP with medical and surgical discharges will be necessary to further enhance its associations and generalizability.

Back to top
Article Information

Accepted for Publication: July 23, 2017.

Corresponding Author: Tudor Borza, MD, MS, Dow Health Services Research Division, Department of Urology, University of Michigan, 2800 Plymouth Rd, Bldg 16, Ann Arbor, MI 48109 (tborza@med.umich.edu).

Published Online: November 22, 2017. doi:10.1001/jamasurg.2017.4585

Author Contributions: Dr Borza had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Borza, Norton, Ellimoottil, Dimick, Shahinian, Hollenbeck.

Acquisition, analysis, or interpretation of data: Borza, Oerline, Skolarus, Norton, Ryan, Hollenbeck.

Drafting of the manuscript: Borza, Skolarus, Hollenbeck.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Borza, Oerline, Norton, Ryan.

Obtained funding: Shahinian, Hollenbeck.

Study supervision: Borza, Skolarus, Ellimoottil, Shahinian, Hollenbeck.

Conflict of Interest Disclosures: Dr Dimick reported having a financial interest in ArborMetrix. No other disclosures were reported.

Funding/Support: This work was supported by grant T32 CA180984 (Drs Borza and Hollenbeck) from the National Cancer Institute and by grants R01 AG039434 (Dr Dimick) and R01 AG048071 (Dr Hollenbeck) from the National Institute on Aging.

Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The contents do not represent the views of the US government. Dr Dimick is the Surgical Innovation Editor for JAMA Surgery but was not involved in the editorial review or decision process.

References
1.
Gonzalez  AA, Abdelsattar  ZM, Dimick  JB, Dev  S, Birkmeyer  JD, Ghaferi  AA.  Time-to-readmission and mortality after high-risk surgery.  Ann Surg. 2015;262(1):53-59.PubMedGoogle ScholarCrossref
2.
Varacallo  MA, Herzog  L, Toossi  N, Johanson  NA.  Ten-year trends and independent risk factors for unplanned readmission following elective total joint arthroplasty at a large urban academic hospital.  J Arthroplasty. 2016.PubMedGoogle Scholar
3.
Dharmarajan  K, Hsieh  AF, Lin  Z,  et al.  Diagnoses and timing of 30-day readmissions after hospitalization for heart failure, acute myocardial infarction, or pneumonia.  JAMA. 2013;309(4):355-363.PubMedGoogle ScholarCrossref
4.
Jencks  SF, Williams  MV, Coleman  EA.  Rehospitalizations among patients in the Medicare fee-for-service program [published correction appears in N Engl J Med. 2011;364(16):1582].  N Engl J Med. 2009;360(14):1418-1428.PubMedGoogle ScholarCrossref
5.
Dimick  JB, Ghaferi  AA.  Hospital readmission as a quality measure in surgery.  JAMA. 2015;313(5):512-513.PubMedGoogle ScholarCrossref
6.
Centers for Medicare & Medicaid Services. Readmissions Reduction Program (HRRP). https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Readmissions-Reduction-Program.html. Published 2016. Accessed March 11, 2017.
7.
Rau  J. Half of nation’s hospitals fail again to escape Medicare’s readmission penalties. http://khn.org/news/half-of-nations-hospitals-fail-again-to-escape-medicares-readmission-penalties/. Published August 3, 2015. Accessed March 15, 2017.
8.
Boccuti  C. Aiming for fewer hospital U-turns: the Medicare Hospital Readmission Reduction Program. http://kff.org/medicare/issue-brief/aiming-for-fewer-hospital-u-turns-the-medicare-hospital-readmission-reduction-program/. Published March 10, 2017. Accessed March 15, 2017.
9.
Zuckerman  RB, Sheingold  SH, Orav  EJ, Ruhter  J, Epstein  AM.  Readmissions, observation, and the Hospital Readmissions Reduction Program.  N Engl J Med. 2016;374(16):1543-1551.PubMedGoogle ScholarCrossref
10.
Desai  NR, Ross  JS, Kwon  JY,  et al.  Association between hospital penalty status under the Hospital Readmission Reduction Program and readmission rates for target and nontarget conditions.  JAMA. 2016;316(24):2647-2656.PubMedGoogle ScholarCrossref
11.
Centers for Medicare & Medicaid Services. Comprehensive Care for Joint Replacement (CJR) model. https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2015-Fact-sheets-items/2015-11-16.html. Published November 16, 2015. Accessed March 15, 2017.
12.
Hicks  CW, Bronsert  M, Hammermeister  KE,  et al.  Operative variables are better predictors of postdischarge infections and unplanned readmissions in vascular surgery patients than patient characteristics.  J Vasc Surg. 2017;65(4):1130-1141.PubMedGoogle ScholarCrossref
13.
Gani  F, Lucas  DJ, Kim  Y, Schneider  EB, Pawlik  TM.  Understanding variation in 30-day surgical readmission in the era of accountable care: effect of the patient, surgeon, and surgical subspecialties.  JAMA Surg. 2015;150(11):1042-1049.PubMedGoogle ScholarCrossref
14.
Kim  Y, Gani  F, Lucas  DJ,  et al.  Early versus late readmission after surgery among patients with employer-provided health insurance.  Ann Surg. 2015;262(3):502-511.PubMedGoogle ScholarCrossref
15.
Jenq  GY, Doyle  MM, Belton  BM, Herrin  J, Horwitz  LI.  Quasi-experimental evaluation of the effectiveness of a large-scale readmission reduction program.  JAMA Intern Med. 2016;176(5):681-690.PubMedGoogle ScholarCrossref
16.
Jack  BW, Chetty  VK, Anthony  D,  et al.  A reengineered hospital discharge program to decrease rehospitalization: a randomized trial.  Ann Intern Med. 2009;150(3):178-187.PubMedGoogle ScholarCrossref
17.
Hamar  B, Rula  EY, Wells  AR, Coberley  C, Pope  JE, Varga  D.  Impact of a scalable care transitions program for readmission avoidance.  Am J Manag Care. 2016;22(1):28-34.PubMedGoogle Scholar
18.
Feigenbaum  P, Neuwirth  E, Trowbridge  L,  et al.  Factors contributing to all-cause 30-day readmissions: a structured case series across 18 hospitals.  Med Care. 2012;50(7):599-605.PubMedGoogle ScholarCrossref
19.
Tsai  TC, Joynt  KE, Orav  EJ, Gawande  AA, Jha  AK.  Variation in surgical-readmission rates and quality of hospital care.  N Engl J Med. 2013;369(12):1134-1142.PubMedGoogle ScholarCrossref
20.
Brooke  BS, Goodney  PP, Kraiss  LW, Gottlieb  DJ, Samore  MH, Finlayson  SR.  Readmission destination and risk of mortality after major surgery: an observational cohort study.  Lancet. 2015;386(9996):884-895.PubMedGoogle ScholarCrossref
21.
Tsai  TC, Orav  EJ, Jha  AK.  Care fragmentation in the postdischarge period: surgical readmissions, distance of travel, and postoperative mortality.  JAMA Surg. 2015;150(1):59-64.PubMedGoogle ScholarCrossref
22.
Barrett  ML, Fingar  KR, Owens  PL, Stocks  C, Steiner  C, Sheng  M. Identifying observation services in the Healthcare Cost and Utilization Project (HCUP) state databases. HCUP Methods Series Report 2015-05. US Agency for Healthcare Research and Quality. http://www.hcup-us.ahrq.gov/reports/methods/methods.jsp. Published September 1, 2015. Accessed March 28, 2017.
23.
Feng  Z, Wright  B, Mor  V.  Sharp rise in Medicare enrollees being held in hospitals for observation raises concerns about causes and consequences.  Health Aff (Millwood). 2012;31(6):1251-1259.PubMedGoogle ScholarCrossref
24.
Stone  J, Hoffman  GJ; Congressional Research Service. Medicare hospital readmissions: issues, policy options and PPACA. http://www.hhnmag.com/ext/resources/inc-hhn/pdfs/resources/opencrs-R40972_20100921.pdf. Published September 21, 2010. Accessed March 28, 2017.
25.
Pope  GC, Kautter  J, Ellis  RP,  et al.  Risk adjustment of Medicare capitation payments using the CMS-HCC model.  Health Care Financ Rev. 2004;25(4):119-141.PubMedGoogle Scholar
26.
Norton  EC, Bieler  GS, Ennett  ST, Zarkin  GA.  Analysis of prevention program effectiveness with clustered data using generalized estimating equations.  J Consult Clin Psychol. 1996;64(5):919-926.PubMedGoogle ScholarCrossref
27.
Dorsey  K, Grady JN, Horwitz  LI, et al. 2016 All-Cause Hospital-Wide Measure Updates and Specifications Report: Hospital-Level 30-Day Risk-Standardized Readmission Measure. Version 5.0. New Haven, CT: Yale New Haven Health Services Corp/Center for Outcomes Research & Evaluation; March 2016. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Measure-Methodology.html. Accessed October 11, 2017.
28.
Wagner  AK, Soumerai  SB, Zhang  F, Ross-Degnan  D.  Segmented regression analysis of interrupted time series studies in medication use research.  J Clin Pharm Ther. 2002;27(4):299-309.PubMedGoogle ScholarCrossref
29.
Penfold  RB, Zhang  F.  Use of interrupted time series analysis in evaluating health care quality improvements.  Acad Pediatr. 2013;13(6)(suppl):S38-S44.PubMedGoogle ScholarCrossref
30.
Dorsey  K, Grady JN, Desai  N,  et al. 2016 Procedure-Specific Measures Updates and Specifications Report: Hospital-Level 30-Day Risk-Standardized Readmission Measures. Version 5.0. New Haven, CT: Yale New Haven Health Services Corp/Center for Outcomes Research & Evaluation; March 2016. https://www.qualitynet.org/dcs/ContentServer?cid=1228774371008&pagename=QnetPublic%2FPage%2FQnetTier4&c=Page. Accessed October 11, 2017.
31.
Benbassat  J, Taragin  M.  Hospital readmissions as a measure of quality of health care: advantages and limitations.  Arch Intern Med. 2000;160(8):1074-1081.PubMedGoogle ScholarCrossref
32.
Greenblatt  DY, Weber  SM, O’Connor  ES, LoConte  NK, Liou  JI, Smith  MA.  Readmission after colectomy for cancer predicts one-year mortality.  Ann Surg. 2010;251(4):659-669.PubMedGoogle ScholarCrossref
33.
Lucas  DJ, Haider  A, Haut  E,  et al.  Assessing readmission after general, vascular, and thoracic surgery using ACS-NSQIP.  Ann Surg. 2013;258(3):430-439.PubMedGoogle ScholarCrossref
34.
Centers for Medicare & Medicaid Services. Hospital Compare datasets. https://data.medicare.gov/data/hospital-compare. Updated July 26, 2017. Accessed June 6, 2017.
35.
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(5):324-331.PubMedGoogle ScholarCrossref
36.
McGarry  BE, Blankley  AA, Li  Y.  The impact of the Medicare Hospital Readmission Reduction Program in New York State.  Med Care. 2016;54(2):162-171.PubMedGoogle ScholarCrossref
37.
Joynt  KE, Figueroa  JE, Oray  J, Jha  AK.  Opinions on the Hospital Readmission Reduction Program: results of a national survey of hospital leaders.  Am J Manag Care. 2016;22(8):e287-e294.PubMedGoogle Scholar
38.
Ellimoottil  C, Ryan  AM, Hou  H, Dupree  JM, Hallstrom  B, Miller  DC.  Implications of the definition of an episode of care used in the Comprehensive Care for Joint Replacement model.  JAMA Surg. 2017;152(1):49-54.PubMedGoogle ScholarCrossref
39.
Krell  RW, Girotti  ME, Fritze  D, Campbell  DA, Hendren  S.  Hospital readmissions after colectomy: a population-based study.  J Am Coll Surg. 2013;217(6):1070-1079.PubMedGoogle ScholarCrossref
40.
Orr  NT, El-Maraghi  S, Korosec  RL, Davenport  DL, Xenos  ES.  Cost analysis of vascular readmissions after common vascular procedures.  J Vasc Surg. 2015;62(5):1281-1287.e1.PubMedGoogle ScholarCrossref
41.
Ricciardi  BF, Oi  KK, Daines  SB, Lee  YY, Joseph  AD, Westrich  GH.  Patient and perioperative variables affecting 30-day readmission for surgical complications after hip and knee arthroplasties: a matched cohort study.  J Arthroplasty. 2017;32(4):1074-1079.PubMedGoogle Scholar
42.
Venkatesh  AK, Wang  C, Ross  JS,  et al.  Hospital use of observation stays: cross-sectional study of the impact on readmission rates.  Med Care. 2016;54(12):1070-1077.PubMedGoogle ScholarCrossref
43.
Centers for Medicare & Medicaid Services. Accountable care organizations (ACOs): general information. https://innovation.cms.gov/initiatives/aco/. Updated September 5, 2017. Accessed June 6, 2017.
×