A Comprehensive Risk Score to Predict Prolonged Hospital Length of Stay After Heart Transplantation

Prediction of Outcomes After Heart Transplantation in Pediatric Patients Using National Registry Data: Evaluation of Machine Learning Approaches

BACKGROUND

The prediction of posttransplant health outcomes for pediatric heart transplantation is critical for risk stratification and high-quality posttransplant care.

OBJECTIVE

The purpose of this study was to examine the use of machine learning (ML) models to predict rejection and mortality for pediatric heart transplant recipients.

METHODS

Various ML models were used to predict rejection and mortality at 1, 3, and 5 years after transplantation in pediatric heart transplant recipients using United Network for Organ Sharing data from 1987 to 2019. The variables used for predicting posttransplant outcomes included donor and recipient as well as medical and social factors. We evaluated 7 ML models-extreme gradient boosting (XGBoost), logistic regression, support vector machine, random forest (RF), stochastic gradient descent, multilayer perceptron, and adaptive boosting (AdaBoost)-as well as a deep learning model with 2 hidden layers with 100 neurons and a rectified linear unit (ReLU) activation function followed by batch normalization for each and a classification head with a softmax activation function. We used 10-fold cross-validation to evaluate model performance. Shapley additive explanations (SHAP) values were calculated to estimate the importance of each variable for prediction.

RESULTS

RF and AdaBoost models were the best-performing algorithms for different prediction windows across outcomes. RF outperformed other ML algorithms in predicting 5 of the 6 outcomes (area under the receiver operating characteristic curve [AUROC] 0.664 and 0.706 for 1-year and 3-year rejection, respectively, and AUROC 0.697, 0.758, and 0.763 for 1-year, 3-year, and 5-year mortality, respectively). AdaBoost achieved the best performance for prediction of 5-year rejection (AUROC 0.705).

CONCLUSIONS

This study demonstrates the comparative utility of ML approaches for modeling posttransplant health outcomes using registry data. ML approaches can identify unique risk factors and their complex relationship with outcomes, thereby identifying patients considered to be at risk and informing the transplant community about the potential of these innovative approaches to improve pediatric care after heart transplantation. Future studies are required to translate the information derived from prediction models to optimize counseling, clinical care, and decision-making within pediatric organ transplant centers.

Use of Extracorporeal Membrane Oxygenation as Bridge to Replacement Therapies in Cardiogenic Shock: Insights From the Extracorporeal Life Support Organization

BACKGROUND

There has been increasing use of extracorporeal membrane oxygenation (ECMO) as bridge to heart transplant (orthotopic heart transplant [OHT]) or left ventricular assist device (LVAD) over the last decade. We aimed to provide insights on the population, outcomes, and predictors for the selection of each therapy.

METHODS

Using the Extracorporeal Life Support Organization Registry between 2010 and 2019, we compared in-hospital mortality and length of stay, predictors of OHT versus LVAD, and predictors of in-hospital mortality for patients with cardiogenic shock that were bridged with ECMO to OHT or LVAD. One hundred sixty-seven patients underwent LVAD versus 234 patients who underwent OHT.

RESULTS

The overall use of ECMO has increased from 1.7% in 2010 to 22.2% in 2019. Mortality was similar between groups (LVAD: 28.7% versus OHT: 29.1%) while length of stay was longer for OHT (LVAD: 49.6 versus OHT: 59.5 days, P=0.05). Factors associated with OHT included prior transplant (odds ratio [OR]=31.26 [CI, 3.84-780.5]), use of a temporary pacemaker (OR=6.5 [CI, 1.39-50.15]), and increased use of inotropes on ECMO (OR=3.77 [CI, 1.39-11.07]), whereas LVAD use was associated with weight (OR=0.98 [CI, 0.97-0.99]), cardiogenic shock presentation (OR=0.40 [CI, 0.21-0.78]), previous LVAD (OR=0.01 [CI, 0.0001-0.22]), respiratory failure (OR=0.28 [CI, 0.11-0.70]), and milrinone infusion (OR=0.32 [CI, 0.15-0.67]). Older age (OR=1.07 [CI, 1.02-1.12]), cannulation bleeding (OR=26.1 [CI, 4.32-221.3]), and surgical bleeding (OR=6.7 [CI, 1.26-39.9]) in patients receiving LVAD and respiratory failure (OR=5 [CI, 1.17-23.1]) and continuous renal replacement therapy (OR=3.82 [CI, 1.28-11.9]) in patients receiving OHT were associated with increased mortality.

CONCLUSIONS

ECMO use as a bridge to advanced therapies has increased over time, with more patients undergoing LVAD than OHT. Mortality was equal between the 2 groups while length of stay was longer for OHT.

A Risk Assessment Model for Stroke in the Early Post-Transplant Period in Adult Cardiac Allograft Recipients: A UNOS Database Analysis

BACKGROUND

Stroke risk in adult cardiac allograft recipients in the early postoperative period remains less defined. This study attempts to develop a risk assessment model in this population.

METHODS

Patients (≥18 years of age) between 2005 and 2015 who underwent cardiac transplantation were selected from the United Network for Organ Sharing database. The final study cohort had 20,915 patients. Risk factors were identified and multivariate logistic regression was used to test associations. SAS software was used for analyses.

RESULTS

Male recipients had a lower risk (odds ratio [OR] 0.7 [0.6-0.92]; P < .05) for stroke. Prolonged ischemic time (OR 1.2 [1.1-1.3), mechanical ventilation (OR 1.6 [1.2-1.9]), left ventricular assist device support (OR 1.8 [1.4-2.3]), black or Hispanic ethnicity (OR 1.33 [1.04-1.7]), days in status 1A (OR 1.05 [1.01-1.1]), recipient creatinine (OR 1.2 [1.02-1.4]), and type 2 diabetes (OR 1.4 [1.1-1.7]) were significant risk factors. A risk score was generated. Patients with a score of 8 had a 5-fold increase in event rate as compared with those with a score of 0. The c-statistic for this model was 0.65.

CONCLUSIONS

For the first time, a weighted risk score GIMVECH (female gender, ischemic time, mechanical ventilation, left ventricular assist device support, ethnicity, clinical history) was generated to assess stroke in the early post-transplant period.

Life impact of VA-ECMO due to primary graft dysfunction in patients after orthotopic heart transplantation

Aims

Primary graft dysfunction (PGD) is a feared complication after heart transplantation (HTX). HTX patients frequently receive veno‐arterial extracorporeal membrane oxygenation (VA‐ECMO) until graft recovery. Long‐term mortality of patients weaned from VA‐ECMO after HTX is comparable with non‐ECMO patients. However, impact on quality of life is unknown. This study investigated days alive and out of hospital (DAOH) as patient‐centred outcome in HTX patients at 1 year after surgery.

Methods and results

This retrospective single‐centre cohort study included patients who underwent HTX at the University Hospital Düsseldorf, Germany, from 2010 to 2020. Main exposure was VA‐ECMO due to PGD. VA‐ECMO and non‐VA‐ECMO patients were compared regarding the primary endpoint DAOH at 1 year after HTX. Subgroup analysis for patients weaned from VA‐ECMO was performed. In total, 144 patients were included into analysis; 1 year mortality was significantly lower in non‐ECMO patients [non‐ECMO 14.3% (14/98) vs. VA‐ECMO 34.8% (16/46), adjusted hazard ratio: 0.32, 95% confidence interval: 0.15–0.74; P = 0.002]. Mortality did not differ significantly between patients weaned from VA‐ECMO and non‐ECMO patients [non‐ECMO 14.3% (14/98) vs. VA‐ECMO (weaned) 18.9% (7/37), adjusted hazard ratio: 0.72, 95% confidence interval: 0.27–1.90; P = 0.48]. DAOH were significantly higher in non‐ECMO patients compared with VA‐ECMO patients and patients weaned from VA‐ECMO [non‐ECMO vs. VA‐ECMO: median 310 (inter‐quartile range 277–327) days vs. 243 (0–288) days; P < 0.0001; non‐ECMO vs. VA‐ECMO (weaned): 310 (277–327) days vs. 253 (208–299) days; P < 0.0001]. These results were still significant after multivariable adjustment with forced entry of predefined covariables.

Conclusions

Despite similar survival rates, VA‐ECMO due to PGD has a relevant life impact as defined by DAOH in the first year after HTX. As a more patient‐centred endpoint, DAOH may contribute to a more comprehensive assessment of outcome in HTX patients.

Post-Heart Transplant Care Pathway's Impact on Reducing Length of Stay

BACKGROUND

Prolonged length of stay (LOS) has undesirable consequences including increased cost, resource consumption, morbidity, and disruptions in hospital flow.

LOCAL PROBLEM

A high-volume heart transplant center in the Pacific Northwest had a mean index hospital LOS of 23 days, with a goal of 10 days according to the institutional heart transplant care pathway.

METHODS

A retrospective, regression analysis was used to identify the factors contributing to LOS of 41 post-heart transplant patients.

INTERVENTIONS

The post-heart transplant care pathway and order set were modified accordingly and reintroduced to the health care team.

RESULTS

Factors contributing to LOS included number of days (1) until the first therapeutic calcineurin inhibitor level, (2) until intravenous diuretics were no longer required, and (3) outside of a therapeutic calcineurin inhibitor range. The interventions reduced the mean LOS by 8 days.

CONCLUSIONS

Increased awareness of LOS, education, and consistent use of care pathways can significantly reduce length of stay.

A Nationwide Analysis of Cleft Palate Repair: Impact of Local Anesthesia on Operative Outcomes and Hospital Cost

BACKGROUND

This study investigates the associations between local anesthesia practice and perioperative complication, length of stay, and hospital cost for palatoplasty in the United States.

METHODS

Patients undergoing cleft palate repair between 2004 and 2015 were abstracted from the Pediatric Health Information System database. Perioperative complication, length of stay, and hospital cost were compared by local anesthesia status. Multiple logistic regressions controlled for patient demographics, comorbidities, and hospital characteristics.

RESULTS

Of 17,888 patients from 49 institutions who met selection criteria, 8631 (48 percent), 4447 (25 percent), and 2149 (12 percent) received epinephrine-containing lidocaine, epinephrine-containing bupivacaine, and bupivacaine or ropivacaine alone, respectively. The use of epinephrine-containing bupivacaine or bupivacaine or ropivacaine alone was associated with decreased perioperative complication [adjusted OR, 0.75 (95 percent CI, 0.61 to 0.91) and 0.63 (95 percent CI, 0.47 to 0.83); p = 0.004 and p = 0.001, respectively]. Only bupivacaine- or ropivacaine-alone recipients experienced a significantly reduced risk of prolonged length of stay on adjusted analysis [adjusted OR, 0.71 (95 percent CI, 0.55 to 0.90); p = 0.005]. Risk of increased cost was reduced in users of any local anesthetic (p < 0.001 for all).

CONCLUSIONS

Epinephrine-containing bupivacaine or bupivacaine or ropivacaine alone was associated with reduced perioperative complication following palatoplasty, while only the latter predicted a decreased postoperative length of stay. Uses of epinephrine-containing lidocaine, epinephrine-containing bupivacaine, and bupivacaine or ropivacaine alone were all associated with decreased hospital costs. Future prospective studies are warranted to further delineate the role of local anesthesia in palatal surgery.

CLINICAL QUESTION/LEVEL OF EVIDENCE

Therapeutic, III.

Percutaneous coronary intervention in patients with cardiac allograft vasculopathy: a Nationwide Inpatient Sample (NIS) database analysis

Cardiac allograft vasculopathy (CAV) is a major cause of heart transplant failure and mortality. The role of percutaneous coronary intervention (PCI) in these patients remains unknown.Methods: The National Inpatient Sample (NIS) (2015-2017) was queried to identify all cases of CAV. The merits of PCI were determined using a propensity-matched multivariate logistic regression model. Adjusted odds ratios (aOR) for in-hospital complications were calculated.Results: A total of 2,380 patients (PCI 185, no-PCI 21,95) with CAV were included in the analysis. There was no significant difference in the odds of major bleeding (OR 1.87, 95% CI 0.94-3.7, P = 0.11), post-procedure bleeding (P = 0.37), cardiogenic shock (OR 0.87, 95% CI 0.45-1.69, P = 0.80), acute kidney injury (uOR 0.92, 95% CI 0.68-1.24, P = 0.64), cardiopulmonary arrest (OR 0.84, 95% CI 0.34-2.11, P = 0.88), and in-hospital mortality (OR 1.59, 95% CI 0.91-2.79, P = 0.14) between patients undergoing PCI compared to those treated conservatively. A propensity-matched analysis closely followed the results of unadjusted crude analysis.Conclusion: PCI in CAV may be associated with increased in-hospital complications and higher resource utilization.

Complications of left ventricular assist devices causing high urgency status on waiting list: impact on outcome after heart transplantation

Aims

Heart transplantation (HTx) represents optimal care for advanced heart failure. Left ventricular assist devices (LVADs) are often needed as a bridge‐to‐transplant (BTT) therapy to support patients during the wait for a donor organ. Prolonged support increases the risk for LVAD complications that may affect the outcome after HTx.

Methods and results

A total of 342 patients undergoing HTx after LVAD as BTT in a 10‐year period in two German high‐volume HTx centres were retrospectively analysed. While 73 patients were transplanted without LVAD complications and with regular waiting list status (T, n = 73), the remaining 269 patients were transplanted with high urgency status (HU) and further divided with regard to the observed leading LVAD complications (infection: HU1, n = 91; thrombosis: HU2, n = 32; stroke: HU3, n = 38; right heart failure: HU4, n = 41; arrhythmia: HU5, n = 23; bleeding: HU6, n = 18; device malfunction: HU7, n = 26). Postoperative hospitalization was prolonged in patients with LVAD complications. Analyses of perioperative morbidity revealed no differences regarding primary graft dysfunction, renal failure, and neurological events except postoperative infections. Short‐term survival, as well as Kaplan–Meier survival analysis, indicated comparable results between the different study groups without disadvantages for patients with LVAD complications.

Conclusions

Left ventricular assist device therapy can impair the outcome after HTx. However, the occurrence of LVAD complications may not impact on outcome after HTx. Thus, we cannot support the prioritization or discrimination of HTx candidates according to distinct mechanical circulatory support‐associated complications. Future allocation strategies have to respect that device‐related complications may define urgency but do not impact on the outcome after HTx.

Development of a scoring tool for predicting prolonged length of hospital stay in peritoneal dialysis patients through data mining

Background

The hospital admission rate is high in patients treated with peritoneal dialysis (PD), and the length of stay (LOS) in the hospital is a key indicator of medical resource allocation. This study aimed to develop a scoring tool for predicting prolonged LOS (pLOS) in PD patients by combining machine learning and traditional logistic regression (LR).

Methods

This study was based on patient data collected using the Hospital Quality Monitoring System (HQMS) in China. Three machine learning methods, classification and regression tree (CART), random forest (RF), and gradient boosting decision tree (GBDT), were used to develop models to predict pLOS, which is longer than the average LOS, in PD patients. The model with the best prediction performance was used to identify predictive factors contributing to the outcome. A multivariate LR model based on the identified predictors was then built to derive the score assigned to each predictor. Finally, a scoring tool was developed, and it was tested by stratifying PD patients into different pLOS risk groups.

Results

A total of 22,859 PD patients were included in our study, with 25.2% having pLOS. Among the three machine learning models, the RF model achieved the best prediction performance and thus was used to identify the 10 most predictive variables for building the scoring system. The multivariate LR model based on the identified predictors showed good discrimination power with an AUROC of 0.721 in the test dataset, and its coefficients were used as a basis for scoring tool development. On the basis of the developed scoring tool, PD patients were divided into three groups: low risk (≤5), median risk [5–10], and high risk (>10). The observed pLOS proportions in the low-risk, median-risk, and high-risk groups in the test dataset were 11.4%, 29.5%, and 54.7%, respectively.

Conclusions

This study developed a scoring tool to predict pLOS in PD patients. The scoring tool can effectively discriminate patients with different pLOS risks and be easily implemented in clinical practice. The pLOS scoring tool has a great potential to help physicians allocate medical resources optimally and achieve improved clinical outcomes.

Impact of implantation time on early function of cardiac transplant

BACKGROUND

Data on out-of-ice implantation ischemia in heart transplant are scarce. We examined implantation time's impact on allograft dysfunction.

METHODS

We conducted a single-site retrospective review of all primary adult heart transplants from June 2012 to August 2019 for implantation warm ischemic time (WIT), defined as first atrial stitch to aortic crossclamp removal. Univariate regression was used to assess the relationship of perioperative variables to primary graft dysfunction (PGD) and to pulmonary artery pulsatility index (PAPi) at postoperative hour 24. A threshold of p < .10 was set for the inclusion of covariates in multivariate regression. Secondary analyses evaluated for consistency among alternative criteria for allograft dysfunction. A post hoc subgroup analysis examined WIT effect in prolonged total ischemia of 240 min or longer.

RESULTS

Complete data were available for 201 patients. Baseline characteristics were similar between patients who did and did not have WIT documented. In univariate analysis, female gender, longer total ischemic time (TIT), longer bypass time, greater blood transfusions, and pretransplant intensive care unit (ICU) care were associated with PGD, whereas longer bypass time was associated with worse PAPi and pretransplant ICU care was associated with better PAPi. In multivariate analysis, longer bypass time predicted PGD, and worse PAPi and preoperative ICU admission predicted PGD and better PAPi. Results did not differ in secondary or subgroup analyses.

CONCLUSIONS

This study is one of few examining the functional impact of cardiac implantation ischemia. Results suggest allograft implantation time alone may not impact postoperative graft function, which was driven by intraoperative bypass duration and by preoperative ICU care, instead.

The Frequency of, and Factors Associated with Prolonged Hospitalization: A Multicentre Study in Victoria, Australia

Background: Limited available evidence suggests that a small proportion of inpatients undergo prolonged hospitalization and use a disproportionate number of bed days. Understanding the factors contributing to prolonged hospitalization may improve patient care and reduce the length of stay in such patients. Methods: We undertook a retrospective cohort study of adult (≥20 years) patients admitted for at least 24 h between 14 November 2016 and 14 November 2018 to hospitals in Victoria, Australia. Data including baseline demographics, admitting specialty, survival status and discharge disposition were obtained from the Victorian Admission Episode Dataset. Multivariable logistic regression analysis was used to identify factors independently associated with prolonged hospitalization (≥14 days). Cox proportional hazard regression model was used to examine the association between various factors and in-hospital mortality. Results: There were almost 5 million hospital admissions over two years. After exclusions, 1,696,112 admissions lasting at least 24 h were included. Admissions with prolonged hospitalization comprised only 9.7% of admissions but utilized 44.2% of all hospital bed days. Factors independently associated with prolonged hospitalization included age, female gender, not being in a relationship, being a current smoker, level of co-morbidity, admission from another hospital, admission on the weekend, and the number of admissions in the prior 12 months. The in-hospital mortality rate was 5.0% for those with prolonged hospitalization compared with 1.8% in those without (p < 0.001). Prolonged hospitalization was also independently associated with a decreased likelihood of being discharged to home (OR 0.53, 95% CI 0.52–0.54). Conclusions: Patients experiencing prolonged hospitalization utilize a disproportionate proportion of bed days and are at higher risk of in-hospital death and discharge to destinations other than home. Further studies are required to identify modifiable factors contributing to prolonged hospitalization as well as the quality of end-of-life care in such admissions.

Prognosis of heart transplant patients in Mashhad University of Medical Sciences

Introduction

Heart transplant is the ultimate treatment for patients with end-stage heart failure.

Aim

To assess 50 heart transplant patients for underlying diseases, transplantation outcome and mortality rate during a 5-year follow-up program.

Material and methods

Fifty heart transplant patients who underwent heart transplantation from 2012 to 2017 were assessed for underlying diseases, organ rejection, duration of hospitalization, extubation time, cardiac output and survival. Biopsy samples were obtained after surgery for evaluation of rejection.

Results

Dilated cardiomyopathy (DCM) and ischemic cardiomyopathy (ICM) were the most common underlying diseases with prevalence of 56% and 12%, respectively. Significant improvement in ejection fraction was observed following heart transplant. Minimum and maximum extubation and hospitalization times were 3–408 hours and 1–51 days, respectively. Organ rejection evaluation 10 days after heart transplantation revealed that 50% of patients did not show any rejection while 10% had severe rejection. At 30 days post-operatively the number of patients with grade III rejection decreased to 2% while 56% of patients had no sign of rejection. The 5-year survival rate was 66% while infection and arrhythmia were the most common causes of death.

Conclusions

DCM and ICM are considered the most prevalent underlying diseases in heart transplant candidates. Ejection fraction reached normal ranges following transplant, which provides good quality of life. Low incidence of severe acute rejection demonstrates the effectiveness of our immunosuppressive therapy. In the cases of increased rejection, the patient’s immunosuppressive regimen was re-assessed accordingly.

Review of interactions between high-risk pediatric heart transplant recipients and marginal donors including utilization of risk score models

BACKGROUND

Donor organ acceptance practices vary among pediatric heart transplant professionals. We sought to understand what is known about the interactions between the "high-risk" recipient and the "marginal donor," and how donor risk scores can impact this discussion.

METHODS

A systematic review of published literature on pediatric HTx was undertaken with the assistance of a medical librarian. Two authors independently assessed search results, and papers were reviewed for inclusion.

RESULTS

We found that there are a large number of individual factors, and clusters of factors, that have been used to label individual recipients "high-risk" and individual donors "marginal." The terms "high-risk recipient" and "marginal donor" have been used broadly in the literature making it virtually impossible to make comparisons between publications. In general, the data support that patients who could be easily agreed to be "sicker recipients" are at more risk compared to those who are clearly "healthier," albeit still "sick enough" to need transplantation. Given this variability in the literature, we were unable to define how being a "high-risk" recipient interplays with accepting a "marginal donor." Existing risk scores are described, but none were felt to adequately predict outcomes from factors available at the time of offer acceptance.

CONCLUSIONS

We could not determine what makes a donor "marginal," a recipient "high-risk," or how these factors interplay within the specific recipient-donor pair to determine outcomes. Until there are better risk scores predicting outcomes at the time of organ acceptance, programs should continue to evaluate each organ and recipient individually.

Hospitalisation and life support in the year before and during heart transplantation: a French national study

Objective

The objective of this study was to define the characteristics of hospital care use during the year prior to heart transplantation.

Methods

A retrospective cohort of heart transplant recipients registered on the national hospital discharge database between 2010 and 2015 was analysed.

Results

In this cohort of 2379 heart transplant recipients (mean age: 48 years, 74% men), 91% had been admitted at least once, for at least 1 day, to a short-stay hospital during the year before transplantation (mean: 4.3 days), and 84% had at least one cardiovascular diagnosis (heart failure and cardiogenic shock: 60%). At least one intensive care stay was reported for 61% of patients, a rehabilitation stay was reported for 30%, mechanical circulatory support was reported for 12%, and ventilatory support was reported for 15%. The median interval between admission and transplantation was 5 days (interquartile QI-Q3, 0-16). The hospital mortality was higher for patients not hospitalised before transplantation (18% vs 15%). Their transplantation hospitalisation diagnosis, compared with patients hospitalised before transplantation, was more frequently cardiomyopathy (31% vs 27%) or heart failure (28% vs 18%), and less frequently myocardial infarction (1% vs 12%).

Conclusion

This study demonstrates a high hospitalisation rate before heart transplantation and identifies three groups of patients: 1-patients with a high hospitalisation rate and terminal heart failure requiring circulatory support who experienced at least one intensive care unit stay; 2-patients with a history of hospitalisation, mainly for heart disease, without circulatory support requirement; and 3-patients with no pretransplantation overnight stay. These findings provide useful information to evaluate the medical benefits and needs for transplantation, and identify areas for improvement in heart transplantation listing criteria.