张丹薇
中国医学科学院阜外医院 心血管内科
AIMS:Preoperative risk assessment is crucial for cardiac surgery. Although previous studies suggested machine learning (ML) may improve in-hospital mortality predictions after cardiac surgery compared to traditional modeling approaches, the validity is doubted due to lacking external validation, limited sample sizes, and inadequate modeling considerations. We aimed to assess predictive performance between ML and traditional modelling approaches, while addressing these major limitations.METHODS AND RESULTS:Adult cardiac surgery cases (n = 168 565) between 2013 and 2018 in the Chinese Cardiac Surgery Registry were used to develop, validate, and compare various ML vs. logistic regression (LR) models. The dataset was split for temporal (2013-2017 for training, 2018 for testing) and spatial (geographically-stratified random selection of 83 centers for training, 22 for testing) experiments, respectively. Model performances were evaluated in testing sets for discrimination and calibration. The overall in-hospital mortality was 1.9%. In the temporal testing set (n = 32 184), the best-performing ML model demonstrated a similar area under the receiver operating characteristic curve (AUC) of 0.797 (95% CI 0.779-0.815) to the LR model (AUC 0.791 [95% CI 0.775-0.808]; P = 0.12). In the spatial experiment (n = 28 323), the best ML model showed a statistically better but modest performance improvement (AUC 0.732 [95% CI 0.710-0.754]) than LR (AUC 0.713 [95% CI 0.691-0.737]; P = 0.002). Varying feature selection methods had relatively smaller effects on ML models. Most ML and LR models were significantly miscalibrated.CONCLUSION:ML provided only marginal improvements over traditional modelling approaches in predicting cardiac surgery mortality with routine preoperative variables, which calls for more judicious use of ML in practice.
European heart journal. Quality of care & clinical outcomes 2024
BACKGROUND:We aimed to develop an administrative model to profile the performance on the outcomes of coronary artery bypass grafting across hospitals in China.METHODS AND RESULTS:This retrospective study was based on the Chinese Hospital Quality Monitoring System (HQMS) from 2016 to 2020. The coronary artery bypass grafting cases were identified by procedure code, and those of 2016 to 2017 were randomly divided into modeling and validation cohorts, while those in other years were used to ensure the model stability across years. The outcome was discharge status as "death or withdrawal," and that withdrawal referred to discharge without medical advice when patients were in the terminal stage but reluctant to die in the hospital. Candidate covariates were mainly identified by diagnoses or procedures codes. Patient-level logistic models and hospital-level hierarchical models were established. A total of 203 010 coronary artery bypass grafts in 699 hospitals were included, with 60 704 and 20 233 cases in the modeling and validation cohorts and 40 423, 42 698, and 38 952 in the years 2018, 2019, and 2020, respectively. The death or withdrawal rate was 3.4%. The areas under the curve were 0.746 and 0.729 in the patient-level models of modeling and validation cohorts, respectively, with good calibration and stability across years. Hospital-specific risk-standardized death or withdrawal rates were 2.61% (interquartile range, 1.87%-3.99%) and 2.63% (interquartile range, 1.97%-3.44%) in the modeling and validation cohorts, which were highly correlated (correlation coefficient, 0.96; P<0.001). Between-hospital variations were distinguished among hospitals of different volumes and across years.CONCLUSIONS:The administrative model based on Hospital Quality Monitoring System could profile hospital performance on coronary artery bypass grafting in China.
Journal of the American Heart Association 2024
BACKGROUND:With increasing surgical workload, it is common for cardiac surgeons to perform coronary artery bypass grafting (CABG) after other procedures in a workday. To investigate whether prior procedures performed by the surgeon impact the outcomes, we compared the outcomes between CABGs performed first versus those performed after prior procedures, separately for on-pump and off-pump CABGs as they differed in technical complexity.METHODS:We conducted a retrospective cohort study of patients undergoing isolated CABG in China from January 2013 to December 2018. Patients were categorised as undergoing on-pump and off-pump CABGs. Outcomes of the procedures performed first in primary surgeons' daily schedule (first procedure) were compared with subsequent ones (non-first procedure). The primary outcome was an adverse events composite (AEC) defined as the number of adverse events, including in-hospital mortality, myocardial infarction, stroke, acute kidney injury and reoperation. Secondary outcomes were the individual components of the primary outcome, presented as binary variables. Mixed-effects models were used, adjusting for patient and surgeon-level characteristics and year of surgery.RESULTS:Among 21 866 patients, 10 109 (16.1% as non-first) underwent on-pump and 11 757 (29.6% as non-first) off-pump CABG. In the on-pump cohort, there was no significant association between procedure order and the outcomes (all p>0.05). In the off-pump cohort, non-first procedures were associated with an increased number of AEC (adjusted rate ratio 1.29, 95% CI 1.13 to 1.47, p<0.001), myocardial infarction (adjusted OR (ORadj) 1.43, 95% CI 1.13 to 1.81, p=0.003) and stroke (ORadj 1.73, 95% CI 1.18 to 2.53, p=0.005) compared with first procedures. These increases were only found to be statistically significant when the procedure was performed by surgeons with <20 years' practice or surgeons with a preindex volume <700 cases.CONCLUSIONS:For a technically challenging surgical procedure like off-pump CABG, prior workload adversely affected patient outcomes.
BMJ quality & safety 2023
OBJECTIVE:The optimal preoperative antiplatelet strategy for patients with acute coronary syndrome (ACS) requiring surgical revascularization remains unclear because of competing risks of bleeding and ischemic events. We evaluated the effect of clopidogrel within 5 days before coronary artery bypass grafting (CABG) on outcomes in patients with ACS.METHODS:Consecutive patients with ACS who underwent isolated CABG at a single center were included in this retrospective study. The primary outcome was a composite of death, myocardial infarction, and stroke within 30 days after surgery. Secondary outcomes were CABG-related major bleeding and perioperative transfusion. Inverse probability weighting using propensity score was performed to evaluate the risk-adjusted effect of preoperative clopidogrel on outcomes.RESULTS:Of 5543 patients with ACS, 820 (14.8%) patients continued clopidogrel within 5 days before CABG. After adjustment for differences in baseline factors, clopidogrel use ≤5 days before CABG was associated with significantly increased odds of the primary composite outcome (odds ratio [OR], 1.63; 95% confidence interval [CI], 1.16-2.29; P = .005), stroke (OR, 3.13; 95% CI, 1.82-5.39; P < .001), major bleeding (OR, 2.01; 95% CI, 1.56-2.58; P < .001), and transfusion (OR, 2.05; 95% CI, 1.82-2.30; P < .001). The effects of preoperative clopidogrel use ≤5 days on primary outcome and major bleeding were greater in patients older than 65 years.CONCLUSIONS:Among patients with ACS undergoing CABG, clopidogrel therapy within 5 days before surgery was associated with increased odds of major cardiac and cerebrovascular events and bleeding complications than discontinuing clopidogrel for >5 days.
The Journal of thoracic and cardiovascular surgery 2022
OBJECTIVES:To develop a model of in-hospital mortality using medical record front page (MRFP) data and assess its validity in case-mix standardisation by comparison with a model developed using the complete medical record data.DESIGN:A nationally representative retrospective study.SETTING:Representative hospitals in China, covering 161 hospitals in modelling cohort and 156 hospitals in validation cohort.PARTICIPANTS:Representative patients admitted for acute myocardial infarction. 8370 patients in modelling cohort and 9704 patients in validation cohort.PRIMARY OUTCOME MEASURES:In-hospital mortality, which was defined explicitly as death that occurred during hospitalisation, and the hospital-level risk standardised mortality rate (RSMR).RESULTS:A total of 14 variables were included in the model predicting in-hospital mortality based on MRFP data, with the area under receiver operating characteristic curve of 0.78 among modelling cohort and 0.79 among validation cohort. The median of absolute difference between the hospital RSMR predicted by hierarchical generalised linear models established based on MRFP data and complete medical record data, which was built as 'reference model', was 0.08% (10th and 90th percentiles: -1.8% and 1.6%). In the regression model comparing the RSMR between two models, the slope and intercept of the regression equation is 0.90 and 0.007 in modelling cohort, while 0.85 and 0.010 in validation cohort, which indicated that the evaluation capability from two models were very similar.CONCLUSIONS:The models based on MRFP data showed good discrimination and calibration capability, as well as similar risk prediction effect in comparison with the model based on complete medical record data, which proved that MRFP data could be suitable for risk adjustment in hospital performance measurement.
BMJ open 2021
Background Characterizing and assessing the prevalence, awareness, and treatment patterns of patients with isolated diastolic hypertension (IDH) can generate new knowledge and highlight opportunities to improve their care. Methods and Results We used data from the China PEACE (Patient-centered Evaluative Assessment of Cardiac Events) Million Persons Project, which screened 2 351 035 participants aged 35 to 75 years between 2014 and 2018. IDH was defined as systolic and diastolic blood pressure of <140 and ≥90 mm Hg; awareness as self-reported diagnosis of hypertension; and treatment as current use of antihypertensive medications. Of the 2 310 184 participants included (mean age 55.7 years; 59.5% women); 73 279 (3.2%) had IDH, of whom 63 112 (86.1%) were untreated, and only 6512 (10.3%) of the untreated were aware of having hypertension. When compared with normotensives, participants who were <60 years, men, at least college educated, had body mass index of >28 kg/m2, consumed alcohol, had diabetes mellitus, and prior cardiovascular events were more likely to have IDH (all P<0.01). Among those with IDH, higher likelihood of awareness was associated with increased age, women, college education, body mass index of >28 kg/m2, higher income, diabetes mellitus, prior cardiovascular events, and Central or Eastern region (all P<0.05). Most treated participants with IDH reported taking only 1 class of antihypertensive medication. Conclusions IDH affects a substantial number of people in China, however, few are aware of having hypertension and most treated participants are poorly managed, which suggests the need to improve the diagnosis and treatment of people with IDH.
Journal of the American Heart Association 2019