Journal Logo

Original Clinical Science—General

Trends in Early Hospital Readmission After Kidney Transplantation, 2002 to 2014

A Population-Based Multicenter Cohort Study

Naylor, Kyla L. PhD1,2; Knoll, Gregory A. MD, MSc3; Allen, Britney MSc1; Li, Alvin H. PhD4; Garg, Amit X. MD, PhD1,5,6; Lam, Ngan N. MD, MSc7; McCallum, Megan K. MPH1; Kim, S. Joseph MD, PhD, MHS, FRCPC8

Author Information
doi: 10.1097/TP.0000000000002036

In the last decade, several outcomes in kidney transplant recipients have improved, with an increase in 1-year graft survival of 91.7% to 94.8% from 2001 to 2013.1,2 However, early posttransplant complications are common and often lead to hospital readmission. Of particular concern is an early hospital readmission (EHR), which is commonly defined as a hospital admission within 30 days of hospital discharge for kidney transplantation.3 EHRs are an important outcome in kidney transplant recipients because they are associated with morbidity and mortality.3-7 For example, McAdams-DeMarco et al4 reported that kidney transplant recipients with an EHR had a 50% increase in the relative hazard of long-term mortality compared to recipients with no EHR. Further, EHRs are costly with a mean cost per kidney transplant recipient of US $10 551.8

Current literature on EHR in kidney transplant recipients has limitations. First, there is limited information on the incidence of EHR in kidney transplant recipients residing outside of the United States.3 Estimates may vary across countries due to differences in recipient characteristics9 and healthcare systems.10 Second, many previous studies have been single center, limiting the external generalizability of findings.5,6,11,12 Lastly, in the past decade, there have been substantial changes in recipient characteristics,13 donor characteristics,14 and changes in clinical practice.15 It is unknown how these changes might have affected EHR rates in kidney transplant recipients.

To address this question, we conducted a multicenter cohort study to assess the incidence of EHR in kidney transplant recipients from Ontario, Canada, and to determine whether this incidence has changed from 2002 to 2014.


Design and Setting

We conducted a population-based cohort study using linked databases from Ontario, Canada, held at the Institute for Clinical Evaluative Sciences (ICES). Universal access to physician and hospital services are provided to all Ontario residents. These data sets were linked using unique encoded identifiers and analyzed at ICES. Study approval was obtained from the institutional review board at Sunnybrook Health Sciences Centre (Toronto, Canada). We followed the reporting guidelines for observational studies (Table S1, SDC,

Data Sources

We obtained information on our study population and outcomes using 5 linked databases. For vital status and demographics, we used Ontario’s Registered Person’s Database. To identify kidney transplant recipients, we used the Canadian Organ Replacement Register. Canadian Organ Replacement Register accurately identifies kidney transplant recipients when validated against chart review from transplant centers (sensitivity, >95%).17 We obtained information on acute care hospitalizations from the Canadian Institute for Health Information (CIHI) Discharge Abstract Database and emergency department visits from the CIHI National Ambulatory Care Reporting System database. The Ontario Health Insurance Plan database was used for data on Ontario physicians’ billing claims.

Cohort Creation

We included all kidney transplant recipients from Ontario, Canada, who survived to hospital discharge for kidney transplantation from April 1, 2002, to February 28, 2015. We excluded recipients of a simultaneous multiorgan transplant (eg, kidney-pancreas), recipients who died or experienced graft failure on or before the discharge date of the initial kidney transplant hospitalization, and recipients with a missing donor type. The cohort entry or index date was defined as the discharge date from hospital after kidney transplantation.


We defined EHR as an unplanned admission for any reason to an acute care hospital within 30 days of being discharged from the hospital after kidney transplantation. Admissions for elective procedures were excluded (Table S2, SDC, This definition is consistent with that used in prior studies.3 A transfer between hospitals was not considered an EHR but rather part of the same episode of care.

Statistical Analysis

We categorized recipients into four eras (2002 to 2004, 2005 to 2007, 2008 to 2010, and 2011 to 2014) based on their date of transplant. Baseline characteristics were described as medians (25th, 75th percentile) for continuous variables and as counts (percentages) for categorical variables. We compared recipient characteristics across eras using the Pearson χ2 test for binary or categorical variables and the Kruskal-Wallis test for continuous variables. We calculated the cumulative incidence of EHR using the exponential function which is based on the incidence rate.18 Recipients were censored at death and end of follow-up (30 days after discharge for kidney transplantation). We used the Cochran-Armitage test for trend to assess potential changes in the cumulative incidence of EHR across eras and year of transplant.

We used the Cox proportional hazards model to examine the relationship between the era of transplant (2002 to 2004 being the referent) and EHR. The proportional hazards assumption was violated, therefore, hazard ratios were also estimated treating year as a continuous variable.19 When examining year as a continuous variable in the Cox proportional hazards model, there was evidence against the linear form. As a result, restricted cubic splines with 3 knots (placed at the 25th percentile, median and 75th percentile of the distribution of noncensored times) were used to examine the potentially nonlinear relationship between year of transplant and EHR. Based on a literature review and clinical expertise, we adjusted for the following covariates: recipient age (years), sex, race, residence (rural vs urban), income quintile, primary cause of end-stage renal disease (ESRD), dialysis vintage (years), Charlson Comorbidity Index,20 previous transplant, delayed graft function (ie, evidence of dialysis within the first 7 days of transplantation), donor age, type (living vs deceased), length of kidney transplant hospitalization, and intensive care unit stay during kidney transplant admission.

We also performed several subgroup analyses stratifying the cumulative incidence of EHR by age (<60 vs ≥ 60 years) and sex, length of initial hospitalization for kidney transplantation (≤5, 6-7, 8-9, ≥ 10 days), donor type (living vs deceased donor), and the 5 most common diagnoses for EHR.

In an additional analysis, we used the Pearson chi-square test to examine variation in EHR across the 6 Ontario adult transplant centers over the entire study period (2002-2014). To account for center- and patient-level factors, we fitted a random effects logistic regression model, allowing for transplant center to be modeled at the appropriate level (ie, level 1, patient characteristics and level 2, center characteristics). Briefly, each center had its own intercept (random effect) with the resulting odds ratio interpreted as the center-specific deviation from the grand mean (ie, difference between the model intercept and center-specific intercept). We adjusted for the following covariates: center volume (defined as the number of transplants in our cohort at each center over the entire study period), recipient age (years), recipient sex, Charlson Comorbidity Index, era of transplant (2002-2004, 2005-2007, 2008-2010, and 2011-2014), dialysis vintage (years), residence (rural vs urban), income quintile, race, previous transplant, and donor type (living vs deceased). We assigned each transplant center a unique random letter to maintain anonymity. For all analyses, a 2-sided P value less than 0.05 was considered statistically significant. All analyses were performed using Statistical Analysis Software, version 9.4.


Baseline Characteristics

We included 5437 kidney transplant recipients (Figure 1). The median age was 52 years (25th, 75th percentile: 42, 62) and 36.6% were female (Table 1). The primary cause of ESRD was glomerulonephritis (30.9%), and the median time on dialysis before transplant was 3 years (25th, 75th percentile: 1, 6). More recently transplanted recipients (2011 to 2014 vs 2002 to 2004) were older (54 years vs 50 years), were more likely to have diabetes (35.3% vs 25.8%), and had a shorter median length of stay for the kidney transplant admission (8 days vs 9 days). The minimum length of stay for the kidney transplant admission was 2 days, and the maximum was 251 days, with a median stay of 8 days (25th, 75th percentile: 7,11 days). A total of 5.9% of patients were managed in the intensive care unit at some point during their initial hospitalization, and a total of 25.6% developed delayed graft function within the 7 days after surgery.

Cohort selection for kidney transplant recipients.
Baseline characteristics of kidney transplant recipients in Ontario by era of kidney transplant, 2002-2014a


A total of 1128 of 5437 kidney transplant recipients experienced an EHR. The 30-day cumulative incidence of EHR was 20.8% (95% confidence interval [CI], 19.7-21.9%). The median length of stay during the EHR was 4 days (25th, 75th percentile: 2, 8 days). Of recipients with an EHR, 11 (0.98%) died during their admission, 27 (2.4%) were admitted to the intensive care unit, and 1028 (91.1%) were readmitted to the transplant hospital. The period of highest risk for EHRs was the 2 to 4 days after discharge from the kidney transplant admission, with 25% (n = 278) of all EHRs occurring during this time (Figure 2).

Number of EHRs presented by days after initial discharge for kidney transplantation.

Trends in EHR

There was no trend in EHR across eras, with a 30-day cumulative incidence of 23.0% (95% CI, 20.4-25.9%), 21.4% (95% CI, 19.2-23.8%), 18.4% (95% CI, 16.4-20.5%), and 21.1% (95% CI, 19.3-22.9%) (P for trend = 0.197) for the eras 2002 to 2004, 2005 to 2007, 2008 to 2010, and 2011 to 2014, respectively (Table 2). Similarly, no trend was found when examining the 30-day cumulative incidence of EHR across years (P for trend = 0.946) (Figure S1, SDC, After adjusting for covariates in the Cox proportional hazards model, we found no association between era of transplant and EHR (Table 2). For example, when compared with recipients who received a transplant from 2002 to 2004, recipients who received a kidney in 2011 to 2014 had an adjusted hazard ratio of 1.0 (95% CI, 0.8-1.2). When using restricted cubic splines to examine year as a continuous variable, a significant nonlinear relationship was observed in both the unadjusted and adjusted analyses (P < 0.05).

Cumulative incidence and hazard ratio of EHR presented by the total cohort and by era of kidney transplant

Subgroup Analyses

Recipients 60 years or older had a higher cumulative incidence of EHR compared with recipients < 60 years in both men and women. For example, women 60 years or older had an EHR cumulative incidence of 23.2% (95% CI, 19.9-27.0%) versus 18.9% (95% CI, 17.0-21.0%) in women younger than 60 years. The cumulative incidence of EHR in men and women were similar (data not shown).

When examining EHR by sex and age, we found no significant trends across eras (Figure 3). Similarly, no significant trend was observed by donor type (living vs deceased) (Figure 3). When examining EHR by length of stay during the kidney transplant admission, we found that, generally, a longer initial hospital stay resulted in a higher incidence of EHR, and this was consistent across eras (Figure 3). For example, the cumulative incidence of EHR in recipients with a kidney transplant admission of 5 days or less was 13.6% (95% CI, 9.8-18.7%), whereas recipients with an admission of 10 days or longer had a cumulative incidence of 27.1% (95% CI, 25.2-29.2%). There were no statistically significant trends in EHR by length of kidney transplant admission across eras.

Cumulative incidence of EHR by era of transplant for men, women, length of initial hospitalization for kidney transplantation, and donor type.

Most Common Diagnoses for EHR

Using the first 3 digits of the most responsible International Classification of Diseases (10th Revision) code, the 5 most common diagnoses for EHR included failure and rejection of transplanted organs and tissues (18.7%); complications of procedures, not elsewhere classified (13.6%); acute renal failure (5.7%); other disorders of urinary system (4.3%); and postprocedural disorders of genitourinary system, not elsewhere classified (2.6%). The 5 most common diagnoses for EHR across eras are presented in Table 3. Of the diagnoses that were common, failure and rejection of transplanted organs and tissues significantly decreased across eras (P for trend < 0.001), complications of procedures, not elsewhere classified. significantly increased (P for trend = 0.003) and there was no significant change in acute renal failure (P for trend = 0.586).

Five most common diagnoses for EHR a

Variation in EHRs Across Transplant Centers

When examining variation in EHR across the 6 transplant centers, we found the 30-day cumulative incidence varied from 15.5% to 27.1% (P < 0.001). After adjusting for center- and patient-level factors, we found the odds of EHR at transplant center C was significantly higher, whereas the odds of EHR at transplant center D was significantly lower (Table 4).

Odds ratio of EHR using a random effects logistic model, presented by transplant center


In this study, we found that 1 in 5 kidney transplant recipients were readmitted to the hospital within 30 days of hospital discharge for kidney transplantation. Despite an increase in recipient age and comorbidities across eras, we found EHR had not increased in the last decade. This study highlights the high burden of EHR in kidney transplant recipients, the need to understand which EHRs might be preventable, and the importance of evaluating interventions to decrease these admissions.

The incidence of EHR in our study was considerably lower than that found in a large population-based US study (21% vs 31%, n = 32 961).8 There are several potential explanations for this finding. First, there were several differences between studies in recipient and donor characteristics that reduced the risk of EHR in our study (eg, living donor kidney transplants 41% vs 25% and black recipient race 7% vs 31% in the current vs US study, respectively).8 However, recipients in our study also had risk factors that may have increased their risk of EHR (eg, older mean recipient and donor age).8 Second, there may be differences between the 2 studies in the EHR definition. We excluded elective procedures from our EHR definition; it is unclear whether these procedures were excluded in the US study. Of note, when we examined EHR, including elective procedures, the incidence changed minimally (less than 2%). Last, differences in healthcare systems and variation in practice patterns could account for some differences in the EHR incidence.9,10 Previous studies in the kidney transplant population, as well as in other populations, have found differences in outcomes between US and Canadian patients.9,21,22 However, it is important to note that wide variability in EHR has also been found across studies within the United States (11% to 32%)3,12,23 and between US transplant centers (18 to 47%).8 Similarly, we found significant variation across our transplant centers (16% to 27%), although to a lesser degree.

The incidence of EHR found in our study was higher when compared with the EHR in major surgeries (21% vs 13%), despite a significantly younger age in our study (52 years vs 78 years).24 Similarly, a Canadian study found that the cumulative incidence of unplanned EHR for surgical (7%) and medical patients (13%) was much lower compared to our study.25 Potential explanations for the higher risk of EHR in kidney transplant recipients versus other patient populations include the high burden of recipient comorbidities (eg, diabetes mellitus, hypertension), increased risk of infectious and noninfectious complications posttransplant, and the functional status of ESRD patients before kidney transplant.

We did not find an increase in EHR over time despite an increase in recipient and donor age and an increase in several comorbidities across eras. There are several possible reasons for this finding. Posttransplant care may have improved (eg, improvement in immunosuppression protocols and posttransplant rehabilitation programs). Although the selection of kidney transplant and deceased donor candidates has become less restrictive, advances in the medical management of dialysis patients may have improved recipients’ general fitness to undergo transplant surgery. Moreover, the significantly shorter length of stay for kidney transplantation across eras may have decreased the number of readmissions for hospital-acquired infections. Of note, we observed a significant nonlinear relationship between year of transplant and EHR. However, we are unclear about the reason for this observation and hypothesize this observation is likely due to chance.

Although no increase in EHR was observed across eras, the high incidence of EHR in kidney transplant recipients is concerning because previous studies have found that EHR is associated with graft failure, mortality, and increased economic costs.4-8 Moreover, EHRs are considered an indicator of healthcare quality.26 Additionally, although no increase in EHR was observed over time, we should not be satisfied since kidney transplant recipient outcomes have shown improvement in various domains over time (eg, reduction in graft failure, posttransplant diabetes).27 Therefore, these results highlight the need for a better understanding of EHR risk factors, which EHRs may be preventable, and the need to develop tools to accurately identify recipients who may benefit from interventions to reduce EHR risk.

Studies examining risk factors for EHR in kidney transplant recipients have found that many risk factors are not modifiable (eg, older recipient age, black race).3 Furthermore, only 1 study in the kidney transplant population has evaluated the predictive ability of a model for EHR (c-statistic = 0.73); however, this model has not been externally validated.11 Notably, there have been no randomized controlled trials to examine the impact of interventions to reduce EHR in kidney transplant recipients. In the general population, numerous randomized controlled trials have been conducted.28 A meta-analysis examining the efficacy of interventions to reduce EHR in nontransplant patients found several interventions to be effective in reducing EHR risk, with those aiming to increase self-care being the most efficacious.28 Recently published nonrandomized studies in kidney transplant recipients suggest that readmissions in kidney transplant recipients could be reduced through additional outpatient care, improving posttransplant medication knowledge, and reducing postdischarge anxiety.12,29,30 Cumulative incidence estimates from our study can be used to guide sample size calculations when designing future randomized controlled trials to reduce the impact of EHR in kidney transplant recipients.

Similar to previously conducted studies in kidney transplant recipients, we found that a longer initial hospital stay resulted in a higher number of EHR.5,6,8 Although not all studies in kidney transplant recipients have observed these results,11 most studies in kidney transplant and other patient populations have shown that a longer initial hospitalization is associated with an increased risk of EHR.31,32 Longer hospitalizations may be a marker of baseline health status, which may have implications for EHR risk and long-term outcome.

When examining the most common diagnoses of EHR across eras, we found that the proportion of recipients with failure and rejection of transplanted organs and tissues as the reason for readmission significantly decreased across eras. This suggests that improvements in posttransplant care (eg, counseling on medication adherence, quality of immunosuppression protocols) may have reduced early graft failure events. However, although declines in graft failure as a cause of EHR were observed, there were increases in other causes resulting in no overall decrease in EHR across eras. Caution should be taken when interpreting these results. To define diagnoses for EHR, we only included the most responsible diagnosis; recipients could still have failure and rejection as a secondary diagnosis on readmission. Therefore, to more accurately understand trends in EHR diagnoses a medical chart abstraction would need to be performed.

This is the first study to examine trends in EHR events in kidney transplant recipients over time. We clearly defined EHR and ensured that elective readmissions were excluded from the definition.3 We were able to include all Ontario kidney transplant recipients (total of 6 transplant centers), due to universal healthcare coverage, thus minimizing selection bias. Despite these strengths, several limitations of our study merit discussion. First, we were not able to adjust for all relevant confounders (eg, smoking, body mass index, rejection risk factors, prescription drugs) due to limitations of our administrative healthcare databases. Second, we were not able to determine EHRs that were preventable; this would require medical chart abstraction. Previous studies conducted in the U.S. have found that preventable admissions after kidney transplantation range from 10 to 21%.3 Third, we could only hypothesize reasons for finding no increase in EHR over time. Finally, approximately 65% of recipients were of white race, potentially limiting generalizability to predominantly non-white kidney transplant populations. Further, our results may not be generalizable to other Canadian provinces and countries.

In conclusion, the burden of EHR in kidney transplant recipients is high but an increase in the incidence of EHR has not been observed over time. These results serve as a call to action to better understand factors that predict EHR, to understand the long-term outcomes in patients with EHR, and to develop/test interventions to prevent EHR in this unique patient population.


This research was made possible by infrastructure support from the Lilibeth Caberto Kidney Clinical Research Unit. This study was supported by the ICES Western site. ICES is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). Core funding for ICES Western is provided by the Academic Medical Organization of Southwestern Ontario (AMOSO), the Schulich School of Medicine and Dentistry (SSMD), Western University, and the Lawson Health Research Institute (LHRI). The research was conducted by members of the ICES Kidney, Dialysis and Transplantation team, at the ICES Western facility, who are supported by a grant from the Canadian Institutes of Health Research (CIHR). The opinions, results and conclusions are those of the authors and are independent from the funding sources. No endorsement by ICES, AMOSO, SSMD, LHRI, CIHR, or the MOHLTC is intended or should be inferred. Parts of this material are based on data and information compiled and provided by CIHI. However, the analyses, conclusions, opinions and statements expressed herein are those of the author, and not necessarily those of CIHI. The authors thank Eric McArthur for his help with data analysis and interpretation.


1. Canadian Institute for Health Information. Canadian Organ Replacement Register Annual Report: Treatment of End-Stage Organ Failure in Canada, 2001 to 2010. Ottawa, ON: CIHI; 2011.
2. Canadian Institute for Health Information. Canadian Organ Replacement Register Annual Report: Treatment of End-Stage Organ Failure in Canada, 2004 to 2013. Ottawa, ON: CIHI; 2015.
3. Li AH, Lam NN, Naylor KL, et al. Early hospital readmissions after transplantation: burden, causes, and consequences. Transplantation. 2016;100:713–718.
4. McAdams-Demarco MA, Grams ME, King E, et al. Sequelae of early hospital readmission after kidney transplantation. Am J Transplant. 2014;14:397–403.
5. Luan FL, Barrantes F, Roth RS, et al. Early hospital readmissions post-kidney transplantation are associated with inferior clinical outcomes. Clin Transplant. 2014;28:487–493.
6. Harhay M, Lin E, Pai A, et al. Early rehospitalization after kidney transplantation: assessing preventability and prognosis. Am J Transplant. 2013;13:3164–3172.
7. King EA, Bowring MG, Massie AB, et al. Mortality and graft loss attributable to readmission after kidney transplantation: immediate and long-term risk. Transplantation. 2017;101:2520–2526.
8. McAdams-Demarco MA, Grams ME, Hall EC, et al. Early hospital readmission after kidney transplantation: patient and center-level associations. Am J Transplant. 2012;12:3283–3288.
9. Kim SJ, Schaubel DE, Fenton SS, et al. Mortality after kidney transplantation: a comparison between the United States and Canada. Am J Transplant. 2006;6:109–114.
10. Lasser KE, Himmelstein DU, Woolhandler S. Access to care, health status, and health disparities in the United States and Canada: results of a cross-national population-based survey. Am J Public Health. 2006;96:1300–1307.
11. Taber DJ, Palanisamy AP, Srinivas TR, et al. Inclusion of dynamic clinical data improves the predictive performance of a 30-day readmission risk model in kidney transplantation. Transplantation. 2015;99:324–330.
12. Lubetzky M, Yaffe H, Chen C, et al. Early readmission after kidney transplantation: examination of discharge-level factors. Transplantation. 2016;100:1079–1085.
13. Lam NN, Kim SJ, Knoll GA, et al. The risk of cardiovascular disease is not increasing over time despite aging and higher comorbidity burden of kidney transplant recipients. Transplantation. 2017;101:588–596.
14. Knoll G. Trends in kidney transplantation over the past decade. Drugs. 2008;68:3–10.
15. Ferguson RM, Henry ML, Elkhammas EA, et al. Twenty years of renal transplantation at Ohio State University: the results of five eras of immunosuppression. Am J Surg. 2003;186:306–311.
16. Benchimol EI, Smeeth L, Guttmann A, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015;12:e1001885.
17. Lam NN, McArthur E, Kim SJ, et al. Validation of kidney transplantation using administrative data. Can J Kidney Health Dis. 2015;2:20.
18. Walker AM. Estimation. In: Observations and Inference: An Introduction to the Methods of Epidemiology. Newton, Massachusetts: Epidemiology Resources Inc.; 1991:107–108.
19. Wilson MG. Assessing Model Adequacy in Proportional Hazards Regression [Internet]. 2013. Accessed January 3 2017.
20. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383.
21. Kaul P, Armstrong PW, Chang WC, et al. Long-term mortality of patients with acute myocardial infarction in the United States and Canada: comparison of patients enrolled in Global Utilization of Streptokinase and t-PA for Occluded Coronary Arteries (GUSTO)-I. Circulation. 2004;110:1754–1760.
22. Gorey KM, Kliewer E, Holowaty EJ, et al. An international comparison of breast cancer survival: Winnipeg, Manitoba and Des Moines, Iowa, metropolitan areas. Ann Epidemiol. 2003;13:32–41.
23. Harhay MN, Hill AS, Wang W, et al. Measures of Global Health Status on dialysis signal early rehospitalization risk after kidney transplantation. PLoS One. 2016;11:e0156532.
24. Tsai TC, Joynt KE, Orav EJ, et al. Variation in surgical-readmission rates and quality of hospital care. N Engl J Med. 2013;369:1134–1142.
25. Canadian Institute for Health Information. All-Cause Readmission to Acute Care and Return to the Emergency Department. Ottawa, ON: CIHI; 2012.
26. Centers for Medicare & Medicaid Services. Readmissions Reduction Program. [Internet]. 2012. Accessed March 4 2017.
27. U.S. Department of Health and Human Services. 2012 annual data report. Accessed September 2 2017.
28. Leppin AL, Gionfriddo MR, Kessler M, et al. Preventing 30-day hospital readmissions: a systematic review and meta-analysis of randomized trials. JAMA Intern Med. 2014;174:1095–1107.
29. Covert KL, Fleming JN, Staino C, et al. Predicting and preventing readmissions in kidney transplant recipients. Clin Transplant. 2016;30:779–786.
30. Chandrasekaran A, Anand G, Sharma L, et al. Role of in-hospital care quality in reducing anxiety and readmissions of kidney transplant recipients. J Surg Res. 2016;205:252–259.
31. van Walraven C, Dhalla IA, Bell C, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. CMAJ. 2010;182:551–557.
32. Westert GP, Lagoe RJ, Keskimaki I, et al. An international study of hospital readmissions and related utilization in Europe and the USA. Health Policy. 2002;61:269–278.

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.