Factors Associated With Unplanned 30-Day Readmissions After Hematopoietic Cell Transplantation Among US Hospitals | Stem Cell Transplantation | JAMA Network Open | JAMA Network
[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.  Study Population and Selection Criteria for Autologous (Auto-) and Allogeneic (Allo-) Hematopoietic Cell Transplants (HCT)
Study Population and Selection Criteria for Autologous (Auto-) and Allogeneic (Allo-) Hematopoietic Cell Transplants (HCT)

Numbers in parentheses are weighted values. NRD indicates Nationwide Readmissions Database.

Figure 2.  Readmission Rates by Transplant Volume and Year
Readmission Rates by Transplant Volume and Year

A, Readmission rates in low-, medium-, and high-volume centers for autologous (auto-) and allogenic (allo-) hematopoietic cell transplant (HCT). B, Readmission rates in 2012 to 2014 for auto-HCT and allo-HCT. Error bars indicate standard errors.

Figure 3.  Reasons for 30-Day Readmission After Hematopoietic Cell Transplantation (HCT)
Reasons for 30-Day Readmission After Hematopoietic Cell Transplantation (HCT)

A, Readmission reasons after autologous HCT (auto-HCT) by hospital transplant volume. B, Readmission reasons after allogenic HCT (allo-HCT) by hospital transplant volume. Error bars indicate standard errors; GI, gastrointestinal; and GVHD, graft-vs-host disease.

Table 1.  Baseline Patient and Hospital Characteristics for Index Transplantation Admission, 2012 to 2014
Baseline Patient and Hospital Characteristics for Index Transplantation Admission, 2012 to 2014
Table 2.  Multivariable Analysis of Variables Associated With 30-Day Unplanned Readmissions
Multivariable Analysis of Variables Associated With 30-Day Unplanned Readmissions
1.
Copelan  EA.  Hematopoietic stem-cell transplantation.  N Engl J Med. 2006;354(17):1813-1826. doi:10.1056/NEJMra052638PubMedGoogle ScholarCrossref
2.
Stranges  E, Russo  CA, Friedman  B.  Procedures With the Most Rapidly Increasing Hospital Costs, 2004-2007: Statistical Brief #82. Rockville, MD: Healthcare Cost and Utilization Project; 2006.
3.
Berenson  RA, Paulus  RA, Kalman  NS.  Medicare’s readmissions-reduction program—a positive alternative.  N Engl J Med. 2012;366(15):1364-1366. doi:10.1056/NEJMp1201268PubMedGoogle ScholarCrossref
4.
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. doi:10.1001/jama.2012.216476PubMedGoogle ScholarCrossref
5.
Fontanarosa  PB, McNutt  RA.  Revisiting hospital readmissions.  JAMA. 2013;309(4):398-400. doi:10.1001/jama.2013.42PubMedGoogle ScholarCrossref
6.
Joynt  KE, Orav  EJ, Jha  AK.  Thirty-day readmission rates for Medicare beneficiaries by race and site of care.  JAMA. 2011;305(7):675-681. doi:10.1001/jama.2011.123PubMedGoogle ScholarCrossref
7.
Thomas  CA, Ward  JC.  The oncology care model: a critique.  Am Soc Clin Oncol Educ Book. 2016;35:e109-e114. doi:10.1200/EDBK_156883PubMedGoogle ScholarCrossref
8.
Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project. Introduction to the HCUP Nationwide Readmissions Database (NRD) 2013. https://www.hcup-us.ahrq.gov/db/nation/nrd/NRD_Introduction_2013.jsp. Published November 2015. Accessed December 16, 2018.
9.
von Elm  E, Altman  DG, Egger  M, Pocock  SJ, Gøtzsche  PC, Vandenbroucke  JP; STROBE Initiative.  The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies.  Int J Surg. 2014;12(12):1495-1499. doi:10.1016/j.ijsu.2014.07.013PubMedGoogle ScholarCrossref
10.
van den Bosch  WF, Kelder  JC, Wagner  C.  Predicting hospital mortality among frequently readmitted patients: HSMR biased by readmission.  BMC Health Serv Res. 2011;11:57. doi:10.1186/1472-6963-11-57PubMedGoogle ScholarCrossref
11.
Mehta  KJB, Patel  K, Wang  H, Appleman  LJ, Parikh  RA.  30 day readmissions in cancer patients in United States.  J Clin Oncol. 2016;34(15_suppl):6577-6577. doi:10.1200/JCO.2016.34.15_suppl.6577Google ScholarCrossref
12.
Price  RA, Stranges  E, Elixhauser  A. Cancer Hospitalizations for Adults, 2009: Statistical Brief #125. Rockville, MD: Healthcare Cost and Utilization Project; 2012.
13.
Spring  L, Li  S, Soiffer  RJ, Antin  JH, Alyea  EP  III, Glotzbecker  B.  Risk factors for readmission after allogeneic hematopoietic stem cell transplantation and impact on overall survival.  Biol Blood Marrow Transplant. 2015;21(3):509-516. doi:10.1016/j.bbmt.2014.11.682PubMedGoogle ScholarCrossref
14.
Bashey  A, Zhang  X, Brown  S,  et al.  Myeloablative allogeneic hematopoietic cell transplantation performed without routine inpatient admission: a single center experience of 462 consecutive patients.  Blood. 2016;128(22):661. http://www.bloodjournal.org/content/128/22/661. Accessed May 22, 2019.Google Scholar
15.
Chhatrala  HS, Agrawal  Y, Ghetiya  S, Shah  C, Chandana  S.  Demographic variables and risk factors associated with 30-day readmission following bone marrow transplant: analysis of nationwide inpatient sample data.  J Clin Oncol. 2016;34(15_suppl):e13080. doi:10.1200/JCO.2016.34.15_suppl.e13080PubMedGoogle ScholarCrossref
16.
Jaglowski  SM, Ruppert  AS, Hofmeister  CC,  et al.  The hematopoietic stem cell transplant comorbidity index can predict for 30-day readmission following autologous stem cell transplant for lymphoma and multiple myeloma.  Bone Marrow Transplant. 2014;49(10):1323-1329. doi:10.1038/bmt.2014.155PubMedGoogle ScholarCrossref
17.
Horwitz  L, Partovian  C, Lin  Z,  et al. Hospital-wide (all-condition) 30-day risk standardization readmission measure. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/MMS/downloads/MMSHospital-WideAll-ConditionReadmissionRate.pdf. Updated August 10, 2011. Accessed December 16, 2018.
18.
Frassoni  F, Labopin  M, Powles  R,  et al; Acute Leukaemia Working Party of the European Group for Blood and Marrow Transplantation.  Effect of centre on outcome of bone-marrow transplantation for acute myeloid leukaemia.  Lancet. 2000;355(9213):1393-1398. doi:10.1016/S0140-6736(00)02137-1PubMedGoogle ScholarCrossref
19.
Horowitz  MM, Przepiorka  D, Champlin  RE,  et al.  Should HLA-identical sibling bone marrow transplants for leukemia be restricted to large centers?  Blood. 1992;79(10):2771-2774.PubMedGoogle Scholar
20.
Matsuo  K, Hamajima  N, Morishima  Y, Harada  M.  Hospital capacity and post-transplant survival after allogeneic bone marrow transplantation: analysis of data from the Japan Society for Hematopoietic Cell Transplantation.  Bone Marrow Transplant. 2000;26(10):1061-1067. doi:10.1038/sj.bmt.1702681PubMedGoogle ScholarCrossref
21.
Bejanyan  N, Bolwell  BJ, Lazaryan  A,  et al.  Risk factors for 30-day hospital readmission following myeloablative allogeneic hematopoietic cell transplantation (allo-HCT).  Biol Blood Marrow Transplant. 2012;18(6):874-880. doi:10.1016/j.bbmt.2011.10.032PubMedGoogle ScholarCrossref
22.
Ballen  KK, Joffe  S, Brazauskas  R,  et al.  Hospital length of stay in the first 100 days after allogeneic hematopoietic cell transplantation for acute leukemia in remission: comparison among alternative graft sources.  Biol Blood Marrow Transplant. 2014;20(11):1819-1827. doi:10.1016/j.bbmt.2014.07.021PubMedGoogle ScholarCrossref
23.
Rauenzahn  S, Truong  Q, Cumpston  A,  et al.  Predictors and impact of thirty-day readmission on patient outcomes and health care costs after reduced-toxicity conditioning allogeneic hematopoietic cell transplantation.  Biol Blood Marrow Transplant. 2014;20(3):415-420. doi:10.1016/j.bbmt.2013.12.559PubMedGoogle ScholarCrossref
24.
Bryant  AL, Deal  AM, Walton  A, Wood  WA, Muss  H, Mayer  DK.  Use of ED and hospital services for patients with acute leukemia after induction therapy: one year follow-up.  Leuk Res. 2015;39(4):406-410. doi:10.1016/j.leukres.2015.01.006PubMedGoogle ScholarCrossref
25.
Crombie  J, Spring  L, Li  S,  et al.  Readmissions after umbilical cord blood transplantation and impact on overall survival.  Biol Blood Marrow Transplant. 2017;23(1):113-118. doi:10.1016/j.bbmt.2016.10.012PubMedGoogle ScholarCrossref
26.
Sud  M, Yu  B, Wijeysundera  HC,  et al.  Associations between short or long length of stay and 30-day readmission and mortality in hospitalized patients with heart failure.  JACC Heart Fail. 2017;5(8):578-588. doi:10.1016/j.jchf.2017.03.012PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Hematology
    July 5, 2019

    Factors Associated With Unplanned 30-Day Readmissions After Hematopoietic Cell Transplantation Among US Hospitals

    Author Affiliations
    • 1Division of Hematology and Oncology, Medical College of Wisconsin, Milwaukee
    • 2Division of Hematology and Oncology, Yale University, New Haven, Connecticut
    • 3Division of Biostatistics, Medical College of Wisconsin, Milwaukee
    JAMA Netw Open. 2019;2(7):e196476. doi:10.1001/jamanetworkopen.2019.6476
    Key Points español 中文 (chinese)

    Question  What is the risk of readmission among adults receiving hematopoietic cell transplants in the United States?

    Findings  This cohort study of a US population-based database found that 30-day readmission rates for autologous and allogenic hematopoietic cell transplants were 11.6% and 24.4%, respectively. The most common reasons for readmission included infection and gastrointestinal symptoms, and hospital volume was inversely associated with readmission rate.

    Meaning  This study suggests that a substantial proportion of US adults receiving hematopoietic cell transplants are readmitted within 30 days of discharge, which may lead to significant health care expenditures.

    Abstract

    Importance  Hematopoietic cell transplantation (HCT) is a therapeutic strategy in the management of several hematological cancers. Limited data exist on the incidence and predictors of 30-day readmission after HCT.

    Objective  To measure the incidence of and risk factors associated with 30-day readmission following HCT in the United States.

    Design, Setting, and Participants  This cohort study examined patient data from the US population-based Nationwide Readmissions Database. All adults (age ≥18 years) who underwent autologous (auto-) or allogenic (allo-) HCT in US hospitals between January 1, 2012, and November 30, 2014, were included. The analysis was performed from June 2018 to February 2019.

    Main Outcomes and Measures  The main outcome was 30-day readmission rates for auto-HCT and allo-HCT. Factors associated with readmission, including baseline demographic characteristics and disease- and hospital-related characteristics (including annual case volume), were measured.

    Results  A total of 28 356 index admissions for auto-HCT in 244 centers (191 low-volume, 38 medium-volume, and 15 high-volume centers) and 17 217 index admissions for allo-HCT in 211 centers (161 low-volume, 37 medium-volume, and 13 high-volume centers) were identified during the study period. The overall 30-day readmission rates were 11.6% for auto-HCT and 24.4% for allo-HCT. The odds of readmission were significantly higher in low-volume hospitals compared with high-volume hospitals (adjusted odds ratio [aOR], 1.69; 95% CI, 1.08-2.64 for auto-HCT and aOR, 1.41; 95% CI, 1.09-1.82 for allo-HCT) but comparable to medium-volume hospitals (aOR, 1.06; 95% CI, 0.62-1.83 for auto-HCT and aOR, 1.19; 95% CI, 0.90-1.57 for allo-HCT). Other factors associated with readmission for auto-HCT included younger age (aOR for age ≥50 vs <49 years, 0.82; 95% CI, 0.68-0.98), female sex (aOR, 1.21; 95% CI, 1.06-1.36), disease type (aOR for other vs myeloma, 1.37; 95% CI, 1.06-1.77), and Elixhauser comorbidity index score (aOR for ≥20 vs 0, 1.5; 95% CI, 1.17-1.93). For allo-HCT, factors associated with readmission included disease type (aOR for acute lymphoblastic leukemia vs acute myelogenous leukemia, 1.30; 95% CI, 1.04-1.62), insurance (aOR for Medicare vs private, 1.18; 95% CI, 1.02-1.36), and Elixhauser comorbidity index score (aOR for 1-9 vs 0, 1.2; 95% CI, 1.04-1.39). Infections, neutropenic fever, and gastrointestinal symptoms were the most common reasons for readmission for both types of HCT.

    Conclusions and Relevance  This study found substantial rates of readmission for both types of HCT and an inverse association between hospital HCT volume and 30-day readmission. These results may provide guidance when developing quality indicators and policies penalizing hospitals for HCT readmission.

    Introduction

    Autologous (auto-) and allogeneic (allo-) hematopoietic cell transplantation (HCT) are widely used therapies in the management of several hematological disorders.1 Despite its curative potential, HCT is often associated with significant treatment-related morbidity, mortality, and health care utilization. A 2009 report from the Agency for the Healthcare Research and Quality (AHRQ) for the Healthcare Cost and Utilization Project (HCUP)2 noted that HCT was among the 10 procedures with the greatest increase in hospital costs from 2004 to 2007, with a growth rate of 84.9% from $694 million to $1.3 billion.

    Readmission after index hospitalization is a commonly used quality and cost-containment metric. Under the Hospital Readmissions Reduction Program, the Centers for Medicare & Medicaid Services uses 30-day unplanned readmission rate as a quality metric to penalize hospitals with high rates of risk-adjusted readmissions for medical conditions such as heart failure, pneumonia, myocardial infarction, and conditions requiring surgery.3-6 With increasing health care demands and cost of care for patients with cancer, the Centers for Medicare & Medicaid Services introduced the Oncology Care Model in 2015. This high-value care model aims to provide quality care at the same or lower price.7 To assess cancer care performance in the model, one of the quality metrics proposed is all-cause unplanned readmission. In this context, patients undergoing auto-HCT and allo-HCT are particularly vulnerable to unplanned readmissions due to immune suppression, myelosuppression, graft-vs-host disease (GVHD), and infections. To our knowledge, data on readmission rates and their associated factors in patients undergoing HCT are lacking. We designed this comprehensive study to assess the incidence and associated risk factors of unplanned 30-day readmission after auto-HCT and allo-HCT.

    Methods
    Data Source

    The study used data from the Nationwide Readmissions Database (NRD), part of a family of databases developed by AHRQ, from January 1, 2012, to November 30, 2014. The analysis was performed from June 2018 to February 2019. The NRD is a unique and powerful database of hospital inpatient stays for patients of all ages and payers. It is drawn from AHRQ’s state inpatient databases, which have a variety of sizes and densities.8 The database contains approximately 17 million discharges every year, weighted to roughly 36 million discharges nationwide. The NRD contains clinical and nonclinical variables that support readmission analyses within the same state and calendar year. The study was deemed exempt by the Medical College of Wisconsin institutional review board because the NRD is publicly available and contains deidentified data. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.9

    Study Population

    We identified the index admission for HCT using International Classification of Diseases, Ninth Revision (ICD-9) procedure codes (auto-HCT: 41.00, 41.01, 41.04, 41.07, and 41.09; allo-HCT: 41.02, 41.03, 41.05, 41.06, and 41.08) (eTable 1 in the Supplement). Survey-weighted domain analysis was conducted to study incidence of 30-day readmission by center volume. Annual hospital case volume was calculated as the sum of all discharges with HCT within the calendar year; groups with low, medium, and high annual case volume were created based on (survey-weighted) tertiles of patients in the analytic data domain (eFigure in the Supplement). Accordingly, hospitals were classified as low volume (<58 per year), medium volume (58-157 per year), and high volume (≥158 per year) for allo-HCT and low volume (<78 per year), medium volume (78-186 per year), and high volume (≥187 per year) for auto-HCT. The Centers for Medicare & Medicaid Services does not provide hospital classification as low-, medium-, and high-volume hospitals for HCT. Patients discharged in the month of December were excluded from the study (to allow 30-day readmission window in the full calendar year of data), as the primary outcome of the study was 30-day readmission. The analysis was limited to urban teaching hospitals because there were few eligible hospitalizations in rural and nonteaching hospitals. Index admissions for patients who died or transferred to another acute care hospital were excluded.

    Main Outcome Measure

    The main outcome measure was 30-day, unplanned, all-cause hospital readmission following an index admission for HCT. Readmissions were identified according to the methods outlined by the Centers for Medicare & Medicaid Services.10 For patients who had multiple readmissions within the first 30 days, only the first readmission was included. To determine the cause of readmission, the primary diagnosis of each readmission was independently reviewed by 2 of us (B.D. and S.G.) and grouped into clinically relevant categories. Secondary outcomes examined were length of stay (LOS) and hospital costs of readmission. Cost data are not directly available in the NRD; all costs were estimated from total charges using annual hospital-specific cost-to-charge ratios provided by HCUP. The estimated costs were then adjusted to the national mean hospital wage using the area wage index provided by HCUP.

    Index Admission Demographic and Clinical Characteristics

    We included patient demographic characteristics (age, sex, race, expected primary payer, and median household income) and relevant comorbidities. We used ICD-9 codes to determine the underlying disease for which HCT was performed. Hospital characteristics, such as volume and bed size, were also examined. The HCUP Clinical Classification Software and ICD-9 Clinical Modification codes used to define these variables are listed in eTable 1 in the Supplement.

    Statistical Analysis

    We used weights developed by AHRQ to estimate national index admissions for HCT throughout the United States.8 After the weighting, we summarized the patient demographic and clinical characteristics. We presented categorical variables with frequencies and percentages and continuous variables with means (standard deviations) and medians (interquartile ranges [IQRs]), as appropriate. Patient demographic characteristics, comorbidities, and hospital characteristics between low-, medium-, and high-volume hospitals were compared using the χ2 test for categorical variables and 1-way analysis of variance for continuous variables; nonnormally distributed continuous variables were log-transformed prior to analysis of variance. To examine differences in 30-day readmissions among low–, medium–, and high–HCT volume hospitals, survey-weighted multivariable logistic regression models were constructed. The model was used to compare odds of readmission between center volumes after controlling for potential confounders, including age, sex, insurance, median income (by zip code), hospital bed size, Elixhauser comorbidity index (ECI) for readmission (eTable 2 in the Supplement), disease type, presence of acute GVHD (aGVHD) for allo-HCT only, use of total body irradiation, stem cell source, discharge disposition, presence of infection, and LOS during index admission. We used adjusted odds ratios (aORs) and 95% confidence intervals to report the results of regression analysis. To address the issues of patients readmitted to a center in a different state than the one in which HCT was performed, we performed analysis for state residents only using the HCUP data element RESIDENT.

    All analyses were completed using SAS statistical software version 9.4 (SAS Institute Inc). All statistical tests were performed using SAS survey procedures to account for the NRD sampling design per HCUP guidelines. All P values were 2-sided and P < .05 was considered statistically significant.

    Results

    Between January 1, 2012, and November 30, 2014, there were 28 356 index admissions for auto-HCT in 244 centers and 17 217 index admissions for allo-HCT in 211 centers (Figure 1). For auto-HCT, 191 hospital-years (78%) were categorized as low–, 38 (16%) as medium–, and 15 (6%) as high–transplant volume centers, while for allo-HCT, 161 (76%) were classified as low–, 37 (18%) as medium–, and 13 (6%) as high–transplant volume centers.

    Baseline Characteristics

    Table 1 summarizes the patient and hospital characteristics for auto- and allo-HCT between low–, medium–, and high–HCT volume hospitals for patients who survived to discharge. Acute myelogenous leukemia was the most common indication for allo-HCT, while multiple myeloma was the most common indication for auto-HCT. These characteristics were well matched among the 3 transplant volume categories for both modalities. There was a significant difference in the LOS at index admission for auto-HCT, with longer LOS in low– and medium–transplant volume hospitals compared with high–transplant volume hospitals. The distribution of comorbidities was well matched between the groups, as shown by the ECI score in Table 1.

    30-Day Readmissions After HCT

    The overall incidence of 30-day readmission was 11.6% for auto-HCT and 24.4% for allo-HCT (eTable 3 in the Supplement). High-volume hospitals had significantly lower rates of readmission compared with low-volume hospitals (high vs medium vs low, 9.6% vs 9.5% vs 15.8% for auto-HCT and 21.5% vs 24.7% vs 27.2% for allo-HCT) (Figure 2A; eTable 3 in the Supplement). When including state residents only, readmission rates for low vs medium vs high volume were 16.4% vs 10.5% vs 10.4% for auto-HCT and 26.8% vs 25.1% vs 21.3% for allo-HCT. There was no difference in the readmission rates by year of transplant for both HCT modalities (Figure 2B). eTable 3 in the Supplement summarizes the univariate analysis of patients readmitted within 30 days. The rates of infection and mean LOS at index admission were significantly lower for readmitted patients for auto-HCT, while there was no difference in rates of infection, LOS, and aGVHD at index admission for patients undergoing allo-HCT.

    The readmission characteristics are summarized in eTable 4 in the Supplement. Most readmissions were to the hospital in which the transplant was performed. The median (range) time to readmission was 5.5 (2.9-11.3) days for auto-HCT and 9.3 (4.5-16.5) days for allo-HCT.

    On adjusted analysis, center volume remained independently associated with 30-day readmission after HCT (Table 2). Odds of readmission were significantly higher in low-volume hospitals compared with high-volume hospitals (aOR, 1.69; 95% CI, 1.08-2.64 for auto-HCT and aOR, 1.41; 95% CI, 1.09-1.82 for allo-HCT) but comparable to medium-volume hospitals (aOR, 1.06; 95% CI, 0.62-1.83 for auto-HCT and aOR, 1.19; 95% CI, 0.90-1.57 for allo-HCT). As shown in Table 2, other factors associated with higher 30-day readmission for auto-HCT included age (aOR for age ≥50 vs <49 years, 0.82; 95% CI, 0.68-0.98), female sex (aOR, 1.21; 95% CI, 1.06-1.36), higher ECI (aOR for ≥20 vs 0, 1.5; 95% CI, 1.17-1.93), disease type (aOR for other vs myeloma, 1.37; 95% CI, 1.06-1.77), and nonroutine discharge (discharge to skilled nursing facility, short-term hospital, or home health care). For allo-HCT, on the other hand, the associated factors were hospital bed size (aOR, 0.75; 95% CI, 0.57-0.95 for small-size hospitals), higher ECI (aOR for 1-9 vs 0, 1.2; 95% CI, 1.04-1.39), diagnosis with acute lymphoblastic leukemia (aOR for acute lymphoblastic leukemia vs acute myelogenous leukemia, 1.30; 95% CI, 1.04-1.62), insurance (aOR for Medicare vs private, 1.18; 95% CI, 1.02-1.36), and use of cord blood transplants (aOR, 2.40; 95% CI, 1.83-3.16). Length of stay at index admission was inversely associated with readmissions only for auto-HCT (aOR, 0.94; 95% CI, 0.90-0.95).

    LOS and Cost of 30-Day Readmissions

    There was a significant difference in readmission LOS among readmitted patients from low-, medium-, and high-volume index facilities (median [IQR], 6.9 [3.2-15.2] days, 5.4 [2.7-12.6] days, and 5.3 [2.7-11.7] days; P = .03) for allo-HCT but not for auto-HCT (median [IQR], 5.1 [2.7-8.5] days, 4.7 [2.4-10.5] days, and 4.2 [2.3-7.6] days; P = .24) (eTable 4 in the Supplement). The readmission cost was significantly higher in low-volume hospitals compared with medium- and high-volume hospitals (mean [SE], $19 794 [$1221], $20 968 [$1811], and $17 294 [$2019]; P < .001 for auto-HCT and $47 823 [$3215], $43 731 [$4028], and $35 889 [$3624]; P < .001 for allo-HCT). However, there was no significant difference in readmission costs by index facility volume for auto-HCT or allo-HCT (eTable 4 in the Supplement).

    Reasons for 30-Day Readmissions

    For auto-HCT, the most common reasons for readmission included infections (30%), neutropenic fever (24%), gastrointestinal (GI) symptoms (12%), and other causes (14%). For allo-HCT, infections remained the most common reason (43%), followed by GI symptoms (12%) and neutropenic fever (10%). Graft-vs-host disease constituted 5% of readmission after allo-HCT. Readmission reasons by transplant volume for both modalities are shown in Figure 3. In addition, our study found several other important reasons for readmission. Cardiac reasons were present in 8.3% for allo-HCT compared with 4.7% for auto-HCT. Acute renal failure was present in 7% of allo-HCT vs approximately 3% of auto-HCT. Dehydration leading to hypovolemia as the reason for readmission was comparable (7% in auto- vs 5% in allo-HCT). Fever without any specific diagnosis was also comparable, occurring in 1.7% and 2.1% of patients who underwent allo-HCT and auto-HCT, respectively.

    Discussion

    Using a large national readmission database, we report the incidence of and factors associated with unplanned 30-day readmission after HCT in the United States. The overall incidence of readmission was 11.6% for auto-HCT and 24.4% for allo-HCT. We report an inverse association between hospital transplant volume and 30-day readmissions in addition to several novel associations. Infections and neutropenic fever were the most common reasons for readmission.

    Patients with cancer generally have higher rates of 30-day unplanned readmissions (approximately 25%) compared with patients with other conditions.11,12 There is a paucity of data on readmissions in hematological malignant neoplasms, particularly for patients undergoing HCT. The reported rates of unplanned readmission in published studies, mainly comprising patients undergoing allo-HCT, vary from 20% to 58%.13-16 This wide range is partly due to variance in the time frame used for reporting readmission rates in HCT, depending on the study. We used a 30-day readmission rate, a time frame widely accepted as reasonable for assessing quality of care received during the preceding hospitalization.17 Our results showed that the readmission rates in allo-HCT were comparable to readmission among other patients with cancer, while rates were much lower after auto-HCT.

    The association between high volume and superior clinical outcomes has been replicated in several studies in the setting of HCT (mainly allo-HCT), although a threshold for what is considered high HCT volume has not been consistently defined.18-20 The clinical outcomes studied with HCT volume include transplant-related mortality, survival, and treatment failure. To our knowledge, no studies have investigated relapse and readmission rates, and this study is the first to report an association between HCT volume and readmission rates. In the absence of a universally accepted definition of high HCT volume, our annual case volume groups were created based on (survey-weighted) tertiles of patients in the analytic data domain. For both auto-HCT and allo-HCT, our results showed that readmission rates were significantly lower in high-volume and medium-volume hospitals compared with low-volume hospitals. For other characteristics of hospitals potentially associated with rates of readmission, we found no association between hospital bed size and auto-HCT readmission, but significantly lower readmission for larger hospitals for allo-HCT. Possible drivers of the observed inverse association between hospital transplant volume and 30-day readmissions are better patient selection, proficiency in HCT procedures by more experienced clinicians and ancillary staff, and better care coordination after HCT at high-volume centers. Several of these factors may affect LOS during readmission and, consequently, the lower cost observed in high-volume centers.

    Published studies have identified several risk factors associated with 30-day unplanned readmission after HCT. These include use of myeloablative or total body irradiation regimens, infections or GVHD during the index HCT admission, higher HCT comorbidity index, Medicaid coverage, and low median income for patients undergoing allo-HCT.13,15,21-23 In addition, patients undergoing cord blood transplants and patients with acute lymphoblastic leukemia were also at high risk for readmission.24,25 Of these, our study identified 3 factors in addition to center volume that were associated with readmission after allo-HCT: higher ECI, cord blood transplantation, and diagnosis of acute lymphoblastic leukemia. In addition, our study did not find an association between income and readmission risk. Several associations were identified for auto-HCT, but 2 worth highlighting are LOS at index admission and nonroutine discharge. In contrast to a previous study,16 we found an inverse association between the LOS at index admission and readmission rates. Of note, a recent study in patients with heart failure also showed an inverse association of LOS with readmission rates, likely due to hospitals’ early discharge policies.26 Discharge before the resolution of complications following transplant could result in readmission in patients who have undergone HCT, but the divergent results observed between auto-HCT and allo-HCT preclude more definite conclusions. The mean LOS for auto-HCT was approximately 17 days compared with 25 days for allo-HCT. Patients discharged to short-term hospitals, skilled nursing facilities, and home health care were more likely to be readmitted, indicating that this population may not be fully recovered following transplant. The association, however, was not observed in patients who underwent allo-HCT.

    Previous studies13-16,21,22 have reported fevers with and without infections, GI toxic effects, and graft complications (mainly GVHD) as some of the common causes of readmission. These results were corroborated by our study as the most common cause of readmission for both types of HCT was infection, followed by neutropenic fever and GI toxic effects. After infections, GI symptoms were the most common reason for readmission for allo-HCT, while neutropenic fever was the second most common for auto-HCT.

    Limitations

    This study has several limitations. All findings are national estimates generated from the NRD sample; the accuracy of the estimates depends on the weighting method generated by AHRQ. The NRD is an administrative database and information on conditioning chemotherapy, other treatments received during the transplant, and disease burden and risk information are not available. The database is not positioned to assess social determinants of health, outpatient care, or other factors that influence patients’ health during transitions from hospital to home. The NRD administrative data used in this study are not designed to distinguish preventable readmissions. Readmissions across states are not captured by NRD; however, this may be inconsequential because very few patients will be readmitted in a state other than the one in which the procedure was performed after HCT, and the results did not differ when nonresidents were excluded. The data do not take into account the outpatient transplants being performed in some centers. Likewise, the data cannot be used to assess the effect of readmission on the outcomes of these patients. Data on race are not available in the NRD, so racial disparity on readmission cannot be estimated. Given the variability in the number of HCT performed per year globally due to differences in population size and resource constraints, the results should be interpreted with caution.

    Conclusions

    Using a large national readmission database, we report the 30-day unplanned readmission rates after HCT in the United States and highlight that certain factors present during index admissions for HCT are associated with high adjusted readmission rates. This study found an inverse association between hospital HCT volume and readmission and distinguished several other factors associated with hospital readmission. Infections with and without neutropenic fever remain the most common reason for readmission with both HCT modalities. Lower readmission rates at high-volume hospitals may substantially reduce health care expenditure. These data will help identify various factors that may contribute to readmission after HCT and guide policy makers to identify and optimize HCT outcomes across hospitals.

    Back to top
    Article Information

    Accepted for Publication: May 11, 2019.

    Published: July 5, 2019. doi:10.1001/jamanetworkopen.2019.6476

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Dhakal B et al. JAMA Network Open.

    Corresponding Author: Mehdi Hamadani, MD, Division of Hematology and Oncology, Medical College of Wisconsin, 9200 W Wisconsin Ave, Milwaukee, WI 53226 (mhamadani@mcw.edu).

    Author Contributions: Dr Dhakal 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.

    Concept and design: Dhakal, Giri, Szabo, Hari, Hamadani.

    Acquisition, analysis, or interpretation of data: Dhakal, Levin, Rein, Fenske, Chhabra, Shah, Szabo, D'Souza, Pasquini, Hari, Hamadani.

    Drafting of the manuscript: Dhakal, Rein, Hari, Hamadani.

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

    Statistical analysis: Dhakal, Rein, Szabo, Hari, Hamadani.

    Administrative, technical, or material support: Hari.

    Supervision: Dhakal, Fenske, D'Souza, Pasquini, Hari.

    Conflict of Interest Disclosures: Dr Dhakal reported serving on an advisory board for Takeda and Amgen and receiving honorarium from Celgene outside the submitted work. Dr Fenske reported personal fees from Genentech, Adaptive Biotechnologies, AbbVie, Sanfoi, Seattle Genetics, AstraZeneca, and Celgene outside the submitted work. Dr Shah reported research funding from Lentigen and serving on advisory boards for Kite, Juno, and Cellectar outside the submitted work. Dr D'Souza reported grants and personal fees from Prothena; grants from Merck, EDO Mundapharma, Sanofi, Celgene, and Janssen; and personal fees from Pfizer outside the submitted work. Dr Pasquini reported personal fees from MediGene, Pfizer, Celgene, and Amgen and grants from Novartis, Gilead/Kite, and Bristol-Myers Squibb outside the submitted work. Dr Hamadani reported research support and/or funding from Takeda, Otsuka Pharmaceutical, Spectrum Pharmaceuticals, and Astellas Pharma; serving as a consultant for MedImmune LLC, Janssen, Incyte Corporation, ADC Therapeutics, Cellerant Therapeutics, Celgene, Pharmacyclics, and DOVA Oncology; and participation in speaker’s bureaus for Celgene and Sanofi Genzyme outside the submitted work. No other disclosures were reported.

    References
    1.
    Copelan  EA.  Hematopoietic stem-cell transplantation.  N Engl J Med. 2006;354(17):1813-1826. doi:10.1056/NEJMra052638PubMedGoogle ScholarCrossref
    2.
    Stranges  E, Russo  CA, Friedman  B.  Procedures With the Most Rapidly Increasing Hospital Costs, 2004-2007: Statistical Brief #82. Rockville, MD: Healthcare Cost and Utilization Project; 2006.
    3.
    Berenson  RA, Paulus  RA, Kalman  NS.  Medicare’s readmissions-reduction program—a positive alternative.  N Engl J Med. 2012;366(15):1364-1366. doi:10.1056/NEJMp1201268PubMedGoogle ScholarCrossref
    4.
    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. doi:10.1001/jama.2012.216476PubMedGoogle ScholarCrossref
    5.
    Fontanarosa  PB, McNutt  RA.  Revisiting hospital readmissions.  JAMA. 2013;309(4):398-400. doi:10.1001/jama.2013.42PubMedGoogle ScholarCrossref
    6.
    Joynt  KE, Orav  EJ, Jha  AK.  Thirty-day readmission rates for Medicare beneficiaries by race and site of care.  JAMA. 2011;305(7):675-681. doi:10.1001/jama.2011.123PubMedGoogle ScholarCrossref
    7.
    Thomas  CA, Ward  JC.  The oncology care model: a critique.  Am Soc Clin Oncol Educ Book. 2016;35:e109-e114. doi:10.1200/EDBK_156883PubMedGoogle ScholarCrossref
    8.
    Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project. Introduction to the HCUP Nationwide Readmissions Database (NRD) 2013. https://www.hcup-us.ahrq.gov/db/nation/nrd/NRD_Introduction_2013.jsp. Published November 2015. Accessed December 16, 2018.
    9.
    von Elm  E, Altman  DG, Egger  M, Pocock  SJ, Gøtzsche  PC, Vandenbroucke  JP; STROBE Initiative.  The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies.  Int J Surg. 2014;12(12):1495-1499. doi:10.1016/j.ijsu.2014.07.013PubMedGoogle ScholarCrossref
    10.
    van den Bosch  WF, Kelder  JC, Wagner  C.  Predicting hospital mortality among frequently readmitted patients: HSMR biased by readmission.  BMC Health Serv Res. 2011;11:57. doi:10.1186/1472-6963-11-57PubMedGoogle ScholarCrossref
    11.
    Mehta  KJB, Patel  K, Wang  H, Appleman  LJ, Parikh  RA.  30 day readmissions in cancer patients in United States.  J Clin Oncol. 2016;34(15_suppl):6577-6577. doi:10.1200/JCO.2016.34.15_suppl.6577Google ScholarCrossref
    12.
    Price  RA, Stranges  E, Elixhauser  A. Cancer Hospitalizations for Adults, 2009: Statistical Brief #125. Rockville, MD: Healthcare Cost and Utilization Project; 2012.
    13.
    Spring  L, Li  S, Soiffer  RJ, Antin  JH, Alyea  EP  III, Glotzbecker  B.  Risk factors for readmission after allogeneic hematopoietic stem cell transplantation and impact on overall survival.  Biol Blood Marrow Transplant. 2015;21(3):509-516. doi:10.1016/j.bbmt.2014.11.682PubMedGoogle ScholarCrossref
    14.
    Bashey  A, Zhang  X, Brown  S,  et al.  Myeloablative allogeneic hematopoietic cell transplantation performed without routine inpatient admission: a single center experience of 462 consecutive patients.  Blood. 2016;128(22):661. http://www.bloodjournal.org/content/128/22/661. Accessed May 22, 2019.Google Scholar
    15.
    Chhatrala  HS, Agrawal  Y, Ghetiya  S, Shah  C, Chandana  S.  Demographic variables and risk factors associated with 30-day readmission following bone marrow transplant: analysis of nationwide inpatient sample data.  J Clin Oncol. 2016;34(15_suppl):e13080. doi:10.1200/JCO.2016.34.15_suppl.e13080PubMedGoogle ScholarCrossref
    16.
    Jaglowski  SM, Ruppert  AS, Hofmeister  CC,  et al.  The hematopoietic stem cell transplant comorbidity index can predict for 30-day readmission following autologous stem cell transplant for lymphoma and multiple myeloma.  Bone Marrow Transplant. 2014;49(10):1323-1329. doi:10.1038/bmt.2014.155PubMedGoogle ScholarCrossref
    17.
    Horwitz  L, Partovian  C, Lin  Z,  et al. Hospital-wide (all-condition) 30-day risk standardization readmission measure. https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/MMS/downloads/MMSHospital-WideAll-ConditionReadmissionRate.pdf. Updated August 10, 2011. Accessed December 16, 2018.
    18.
    Frassoni  F, Labopin  M, Powles  R,  et al; Acute Leukaemia Working Party of the European Group for Blood and Marrow Transplantation.  Effect of centre on outcome of bone-marrow transplantation for acute myeloid leukaemia.  Lancet. 2000;355(9213):1393-1398. doi:10.1016/S0140-6736(00)02137-1PubMedGoogle ScholarCrossref
    19.
    Horowitz  MM, Przepiorka  D, Champlin  RE,  et al.  Should HLA-identical sibling bone marrow transplants for leukemia be restricted to large centers?  Blood. 1992;79(10):2771-2774.PubMedGoogle Scholar
    20.
    Matsuo  K, Hamajima  N, Morishima  Y, Harada  M.  Hospital capacity and post-transplant survival after allogeneic bone marrow transplantation: analysis of data from the Japan Society for Hematopoietic Cell Transplantation.  Bone Marrow Transplant. 2000;26(10):1061-1067. doi:10.1038/sj.bmt.1702681PubMedGoogle ScholarCrossref
    21.
    Bejanyan  N, Bolwell  BJ, Lazaryan  A,  et al.  Risk factors for 30-day hospital readmission following myeloablative allogeneic hematopoietic cell transplantation (allo-HCT).  Biol Blood Marrow Transplant. 2012;18(6):874-880. doi:10.1016/j.bbmt.2011.10.032PubMedGoogle ScholarCrossref
    22.
    Ballen  KK, Joffe  S, Brazauskas  R,  et al.  Hospital length of stay in the first 100 days after allogeneic hematopoietic cell transplantation for acute leukemia in remission: comparison among alternative graft sources.  Biol Blood Marrow Transplant. 2014;20(11):1819-1827. doi:10.1016/j.bbmt.2014.07.021PubMedGoogle ScholarCrossref
    23.
    Rauenzahn  S, Truong  Q, Cumpston  A,  et al.  Predictors and impact of thirty-day readmission on patient outcomes and health care costs after reduced-toxicity conditioning allogeneic hematopoietic cell transplantation.  Biol Blood Marrow Transplant. 2014;20(3):415-420. doi:10.1016/j.bbmt.2013.12.559PubMedGoogle ScholarCrossref
    24.
    Bryant  AL, Deal  AM, Walton  A, Wood  WA, Muss  H, Mayer  DK.  Use of ED and hospital services for patients with acute leukemia after induction therapy: one year follow-up.  Leuk Res. 2015;39(4):406-410. doi:10.1016/j.leukres.2015.01.006PubMedGoogle ScholarCrossref
    25.
    Crombie  J, Spring  L, Li  S,  et al.  Readmissions after umbilical cord blood transplantation and impact on overall survival.  Biol Blood Marrow Transplant. 2017;23(1):113-118. doi:10.1016/j.bbmt.2016.10.012PubMedGoogle ScholarCrossref
    26.
    Sud  M, Yu  B, Wijeysundera  HC,  et al.  Associations between short or long length of stay and 30-day readmission and mortality in hospitalized patients with heart failure.  JACC Heart Fail. 2017;5(8):578-588. doi:10.1016/j.jchf.2017.03.012PubMedGoogle ScholarCrossref
    ×