Multimorbidity is strongly associated with long-term but not short-term mortality after cardiac valve replacement

Machine learning-based risk profile classification of patients undergoing elective heart valve surgery

OBJECTIVES

Machine learning methods potentially provide a highly accurate and detailed assessment of expected individual patient risk before elective cardiac surgery. Correct anticipation of this risk allows for the improved counselling of patients and avoidance of possible complications. We therefore investigated the benefit of modern machine learning methods in personalized risk prediction for patients undergoing elective heart valve surgery.

METHODS

We performed a monocentric retrospective study in patients who underwent elective heart valve surgery between 1 January 2008 and 31 December 2014 at our centre. We used random forests, artificial neural networks and support vector machines to predict the 30-day mortality from a subset of 129 available demographic and preoperative parameters. Exclusion criteria were reoperation of the same patient, patients who needed anterograde cerebral perfusion due to aortic arch surgery and patients with grown-up congenital heart disease. Finally, the cohort consisted of 2229 patients with a 30-day mortality of 3.86% (86 of 2229 cases). This trial has been registered at clinicaltrials.gov (NCT03724123).

RESULTS

The final random forest model trained on the entire data set provided an out-of-bag area under the receiver operator characteristics curve (AUC) of 0.839, which significantly outperformed the European System for Cardiac Operative Risk Evaluation (EuroSCORE) (AUC = 0.704) and a model trained only on the subset of features EuroSCORE uses (AUC = 0.745).

CONCLUSIONS

Advanced machine learning methods can predict outcomes of valve surgery procedures with higher accuracy than established risk scores based on logistic regression on pre-selected parameters. This approach is generalizable to other elective high-risk interventions and allows for training models to the cohorts of specific institutions.

Early mortality following percutaneous coronary intervention and cardiac surgery: Correlations within providers and operators

BACKGROUND

It is not clear whether correlations exist within hospitals or operators among risk-adjusted mortality rates (RAMRs) for the most common cardiac interventions and how much of variations in outcomes are residually explained by providers and physicians. We examined these aspects by using recent national data on percutaneous coronary intervention (PCI) and cardiac surgery.

METHODS

Publically available data from New York State aggregated at hospital and operator level were downloaded by Department of Health website for in-hospital/30-day mortality after PCI, coronary artery bypass graft (CABG) and valve surgery. Correlations between RAMRs were evaluated by using Spearman's coefficient (rho). The proportion of mortality variation attributed to hospitals and operators was estimated.

RESULTS

During the period 2008-2013, 390 cardiologists from 63 hospitals and 163 surgeons from 41 centres were evaluated. The RAMRs during 2008-2010 correlated with the RAMRs during 2011-2013 for valve surgery within providers (rho=0.55;p<0.001) and within interventionists for PCI (rho=0.21;p<0.001), isolated CABG (rho=0.25;p=0.009), and any valve surgery or CABG procedure (rho=0.49;p<0.001). The most recent hospital's RAMRs (year 2012 and 2013) significantly correlated in PCI (rho=0.40;p=0.002) but not in CABG (rho=0.13;p=0.413). <2% of mortality variations was attributed to providers and 2-3% to difference between operators.

CONCLUSIONS

A correlation exists at provider and operator level in RAMRs for PCI and cardiac surgery procedures performed in New York State. Beyond patient's risk profile, that is the strongest predictor of early mortality after a cardiac procedure, hospitals and operators have a small but statistically significant contribution to variation in post-operative outcome.

Impact of marital status and comorbid disorders on health-related quality of life after cardiac surgery

PURPOSE

To explore associations between HRQL, marital status, and comorbid disorders in men and women following cardiac surgery.

METHOD

A secondary analysis was completed using data from a randomized controlled trial in which 416 individuals (23% women) scheduled for elective coronary artery bypass graft and/or valve surgery were recruited between March 2012 and September 2013. HRQL was assessed using the Health State Descriptive System (15D) preoperatively, then at 2 weeks, and at 3, 6, and 12 months following cardiac surgery. Linear mixed model analyses were performed to explore associations between HRQL, social support, and comorbid disorders.

RESULTS

The overall 15D scores for the total sample improved significantly from 2 weeks to 3 months post surgery, with only a gradual change observed from 3 to 12 months. Thirty percent (n = 92) of the total sample reported a lower 15D total score at 12 months compared to preoperative status, of whom 78% (n = 71) had a negative minimum important differences (MID), indicating a worse HRQL status. When adjusted for age and marital status, women had statistically significant lower 15D total scores compared to men at 3, 6, and 12 months post surgery. Compared to pre-surgery, improvement was demonstrated in 4 out of 15 dimensions of HRQL for women, and in 6 out of 15 dimensions for men at 12 months post surgery. Both men and women associated back/neck problems, depression, and persistent pain intensity with lower HRQL; for women, not living with a partner/spouse was associated with lower HRQL up to 12 months.

CONCLUSION

Women experienced decreased HRQL and a slower first-year recovery following cardiac surgery compared to men. This study demonstrates a need for follow-up and support to help women manage their symptoms and improve their function within the first year after cardiac surgery. This was particularly pronounced for those women living alone.