Gary Tse
Profile Url: gary-tse382
Researcher at Tianjin Medical University
BackgroundRecent studies have reported numerous significant predictors for adverse outcomes in COVID-19 disease. However, there have been few simple clinical risk score for prompt risk stratification. The objective is to develop a simple risk score for severe COVID-19 disease using territory-wide healthcare data based on simple clinical and laboratory variables. MethodsConsecutive patients admitted to Hong Kongs public hospitals between 1st January and 22nd August 2020 diagnosed with COVID-19, as confirmed by RT-PCR, were included. The primary outcome was composite intensive care unit admission, need for intubation or death with follow-up until 8th September 2020. ResultsCOVID-19 testing was performed in 237493 patients and 4445 patients (median age 44.8 years old, 95% CI: [28.9, 60.8]); 50% male) were tested positive. Of these, 212 patients (4.8%) met the primary outcome. A risk score including the following components was derived from Cox regression: gender, age, hypertension, stroke, diabetes mellitus, ischemic heart disease/heart failure, respiratory disease, renal disease, increases in neutrophil count, monocyte count, sodium, potassium, urea, alanine transaminase, alkaline phosphatase, high sensitive troponin-I, prothrombin time, activated partial thromboplastin time, D-dimer and C-reactive protein, as well as decreases in lymphocyte count, base excess and bicarbonate levels. The model based on test results taken on the day of admission demonstrated an excellent predictive value. Incorporation of test results on successive time points did not further improve risk prediction. ConclusionsA simple clinical score accurately predicted severe COVID-19 disease, even without including symptoms, blood pressure or oxygen status on presentation, or chest radiograph results.
BackgroundIn this territory-wide, observational, propensity score-matched cohort study, we evaluate the development of transient ischaemic attack and ischaemic stroke (TIA/Ischaemic stroke) in patients with AF treated with edoxaban or warfarin. MethodsThis was an observational, territory-wide cohort study of patients between January 1st, 2016 and December 31st, 2019, in Hong Kong. The inclusion were patients with i) atrial fibrillation, and ii) edoxaban or warfarin prescription. 1:2 propensity score matching was performed between edoxaban and warfarin users. Univariate Cox regression identifies significant risk predictors of the primary, secondary and safety outcomes. Hazard ratios (HRs) with corresponding 95% confidence interval [CI] and p values were reported. ResultsThis cohort included 3464 patients (54.18% males, median baseline age: 72 years old, IQR: 63-80, max: 100 years old), 664 (19.17%) with edoxaban use and 2800 (80.83%) with warfarin use. After a median follow-up of 606 days (IQR: 306-1044, max: 1520 days), 91(incidence rate: 2.62%) developed TIA/ischaemic stroke: 1.51% (10/664) in the edoxaban group and 2.89% (81/2800) in the warfarin group. Edoxaban was associated with a lower risk of TIA or ischemic stroke when compared to warfarin. ConclusionsEdoxaban use was associated with a lower risk of TIA or ischemic stroke after propensity score matching for demographics, comorbidities and medication use.
BackgroundThere is a bidirectional relationship between blood pressure variability (BPV) and generalized anxiety disorder (GAD), but few studies have examined the gender- and age-specific effects of visit-to-visit BPV on GAD incidence. We examined the predictive value of BPV for the incidence of GAD in a family clinic cohort. MethodsConsecutive patients with a first attendance to family medicine clinics in Hong Kong between January 1st, 2000, and December 31st, 2002, with at least three blood pressure measurements available thereafter were included. The primary endpoint was incident GAD as identified by ICD-9 coding from the local Clinical Data Analysis and Reporting System. ResultsThis study included 48023 (50% males) patients with a median follow-up of 224 (IQR: 217-229) months. Females were more likely to develop GAD compared to males (incidence rate: 7% vs. 2%), as were patients of older age. Significant univariate predictors were female gender, older age, pre-existing cardiovascular diseases, respiratory diseases, diabetes mellitus, hypertension, and gastrointestinal diseases, various laboratory examinations and the number of blood pressure measurements. Higher baseline, maximum, minimum, SD, CV, and variability score of diastolic blood pressure significantly predicted GAD, as did all systolic blood pressure measures (baseline, latest, maximum, minimum, mean, median, variance, SD, RMS, CV, variability score). ConclusionsThe relationships between longer term visit-to-visit BPV and incident GAD were identified. Female and older patients with higher blood pressure and higher BPV were at the highest risks of GAD.
Background: The coronavirus disease 2019 (COVID-19) has become a pandemic, placing significant burdens on the healthcare systems. In this study, we tested the hypothesis that a machine learning approach incorporating hidden nonlinear interactions can improve prediction for Intensive care unit (ICU) admission. Methods: Consecutive patients admitted to public hospitals between 1st January and 24th May 2020 in Hong Kong with COVID-19 diagnosed by RT-PCR were included. The primary endpoint was ICU admission. Results: This study included 1043 patients (median age 35 (IQR: 32-37; 54% male). Nineteen patients were admitted to ICU (median hospital length of stay (LOS): 30 days, median ICU LOS: 16 days). ICU patients were more likely to be prescribed angiotensin converting enzyme inhibitors/angiotensin receptor blockers, anti-retroviral drugs lopinavir/ritonavir and remdesivir, ribavirin, steroids, interferon-beta and hydroxychloroquine. Significant predictors of ICU admission were older age, male sex, prior coronary artery disease, respiratory diseases, diabetes, hypertension and chronic kidney disease, and activated partial thromboplastin time, red cell count, white cell count, albumin and serum sodium. A tree-based machine learning model identified most informative characteristics and hidden interactions that can predict ICU admission. These were: low red cells with 1) male, 2) older age, 3) low albumin, 4) low sodium or 5) prolonged APTT. A five-fold cross validation confirms superior performance of this model over baseline models including XGBoost, LightGBM, random forests, and multivariate logistic regression. Conclusions: A machine learning model including baseline risk factors and their hidden interactions can accurately predict ICU admission in COVID-19.
BackgroundProgrammed death 1 (PD-1) and programmed death 1 ligand (PD-L1) inhibitors, such as pembrolizumab, nivolumab and atezolizumab, are a major class of immune checkpoint inhibitors that are increasingly used for cancer treatment. However, their use is associated with adverse cardiovascular events. We examined the incidence of new-onset cardiac complications in patients receiving PD-1 or PD-L1 inhibitors. MethodsPatients receiving PD-1 or PD-L1 inhibitors since their launch up to 31st December 2019 at the Hospital Authority of Hong Kong, without pre-existing cardiac complications were included. The primary outcome was a composite of incident heart failure, acute myocardial infarction (AMI), atrial fibrillation (AF) or atrial flutter with the last follow-up date of 31st August 2020. ResultsA total of 1959 patients were included. Over a median follow-up of 136 days (IQR: 42-279), 320 (16.3%) patients met the primary outcome (heart failure: n=244, AMI: n=38, AF: n=54, atrial flutter: n=6) after PD-1/PD-L1 treatment. Univariate Cox regression showed that age, respiratory diseases, gastrointestinal diseases, a shorter readmission interval and total number of hospitalizations before PD-1/PD-L1 inhibitor prescription, PD-L1 inhibitor use, hyponatraemia, and reduced triglyceride levels were significant predictors of the primary outcome. On multivariate adjustment, older age, a shorter average readmission interval, and a higher number of hospital admissions remained significant predictors. Patients who developed cardiovascular complications had a shorter average readmission interval and a higher number of hospitalizations after PD-1/PD-L1 treatment. ConclusionsCardiovascular complications are found in 16% of patients receiving PD-1 or PD-L1 inhibitors and are associated with more healthcare utilization.
IntroductionLong QT syndrome (LQTS) and catecholaminergic ventricular tachycardia (CPVT) are less prevalent cardiac ion channelopathies than Brugada syndrome in Asia. The present study compared paediatric/young and adult patients with these conditions. MethodsThis was a territory-wide retrospective cohort study of consecutive patients diagnosed with LQTS and CPVT attending public hospitals in Hong Kong. The primary outcome was spontaneous ventricular tachycardia/ventricular fibrillation (VT/VF). ResultsA total of 142 LQTS (mean onset age= 27{+/-}23 years old) and 16 CPVT (mean presentation age=11{+/-}4 years old) patients were included. For LQTS, arrhythmias other than VT/VF (HR=4.67, 95% confidence interval=[1.53-14.3], p=0.007), initial VT/VF (HR=3.25 [1.29-8.16], p=0.012) and Schwartz score (HR=1.90 [1.11-3.26], p=0.020) were predictive of the primary outcome for the overall cohort, whilst arrhythmias other than VT/VF (HR=5.41 [1.36-21.4], p=0.016) and Schwartz score (HR=4.67 [1.48-14.7], p=0.009) were predictive for the adult subgroup (>25 years old; n=58). All CPVT patients presented before the age of 25 but no significant predictors of VT/VF were identified. A random survival forest model identified initial VT/VF, Schwartz score, initial QTc interval, family history of LQTS, initially asymptomatic, and arrhythmias other than VT/VF as the most important variables for risk prediction in LQTS, and initial VT/VF/sudden cardiac death, palpitations, QTc, initially symptomatic and heart rate in CPVT. ConclusionClinical and ECG presentation vary between the pediatric/young and adult LQTS population. All CPVT patients presented before the age of 25. Machine learning models achieved more accurate VT/VF prediction.