Effect of Real-Time Computer-Aided Polyp Detection System (ENDO-AID) on Adenoma Detection in Endoscopists-in-Training: A Randomized Trial.

BACKGROUND: The effect of computer-aided polyp detection (CADe) on adenoma detection rate (ADR) among endoscopists-in-training remains unknown. METHODS: We performed a single-blind, parallel-group, randomized controlled trial in Hong Kong between April 2021 and July 2022 (NCT04838951). Eligible subjects undergoing screening/surveillance/diagnostic colonoscopies were randomized 1:1 to receive colonoscopies with CADe (ENDO-AID[OIP-1]) or not (control) during withdrawal. Procedures were performed by endoscopists-in-training with <500 procedures and <3 years' experience. Randomization was stratified by patient age, sex, and endoscopist experience (beginner vs intermediate level, <200 vs 200-500 procedures). Image enhancement and distal attachment devices were disallowed. Subjects with incomplete colonoscopies or inadequate bowel preparation were excluded. Treatment allocation was blinded to outcome assessors. The primary outcome was ADR. Secondary outcomes were ADR for different adenoma sizes and locations, mean number of adenomas, and non-neoplastic resection rate. RESULTS: A total of 386 and 380 subjects were randomized to CADe and control groups, respectively. The overall ADR was significantly higher in the CADe group than in the control group (57.5% vs 44.5%; adjusted relative risk, 1.41; 95% CI, 1.17-1.72; P < .001). The ADRs for <5 mm (40.4% vs 25.0%) and 5- to 10-mm adenomas (36.8% vs 29.2%) were higher in the CADe group. The ADRs were higher in the CADe group in both the right colon (42.0% vs 30.8%) and left colon (34.5% vs 27.6%), but there was no significant difference in advanced ADR. The ADRs were higher in the CADe group among beginner (60.0% vs 41.9%) and intermediate-level (56.5% vs 45.5%) endoscopists. Mean number of adenomas (1.48 vs 0.86) and non-neoplastic resection rate (52.1% vs 35.0%) were higher in the CADe group. CONCLUSIONS: Among endoscopists-in-training, the use of CADe during colonoscopies was associated with increased overall ADR. (ClinicalTrials.gov, Number: NCT04838951).

Incidence, epidemiology, and outcomes of acute allograft rejection following liver transplantation in Australia.

Acute allograft rejection is a well-known complication of liver transplantation (LT). The incidence, epidemiology, and outcomes of acute rejection have not been well described in Australia. We retrospectively studied consecutive adults who underwent deceased donor LT at a single center between 2010 and 2020. Donor and recipient data at the time of LT and recipient outcomes were collected from a prospective LT database. Liver biopsy reports were reviewed, and only a graft's first instance of biopsy-proven acute rejection was analyzed. During the study period, 796 liver transplants were performed in 770 patients. Biopsy-proven rejection occurred in 34.9% of transplants. There were no significant changes in the incidence of rejection over time (linear trend p =0.11). The median time to the first episode of rejection was 71 days after LT: 2.2% hyperacute, 50.4% early (≤90 d), and 47.5% late rejection (>90 d). Independent risk factors for rejection were younger recipient age at transplant (aHR 0.98 per year increase, 95% CI: 0.97-1.00, p =0.01), and ABO-incompatible grafts (aHR 2.55 vs. ABO-compatible, 95% CI: 1.27-5.09, p <0.01) while simultaneous multiorgan transplants were protective (aHR 0.21 vs. LT only, 95% CI: 0.08-0.58, p <0.01). Development of acute rejection (both early and late) was independently associated with significantly reduced graft (aHR 3.13, 95% CI: 2.21-4.42, p <0.001) and patient survival (aHR 3.42, 95% CI: 2.35-4.98, p <0.001). In this 11-year Australian study, acute LT rejection occurred in 35%, with independent risk factors of younger recipient age and ABO-incompatible transplant, while having a simultaneous multiorgan transplant was protective. Acute rejection was independently associated with reduced graft and patient survival after adjustment for other factors.

Diagnostic and prognostic performance of the SAFE score in non-alcoholic fatty liver disease.

BACKGROUND & AIMS: The steatosis-associated fibrosis estimator (SAFE) score was developed to detect clinically significant liver fibrosis in patients with NAFLD in the United States. We compare the performance of the SAFE score and other non-invasive tests to diagnose liver fibrosis and to correlate the scores with liver-related outcomes in patients with NAFLD in Hong Kong. METHODS: This was a retrospective cohort study involving two data sets. The first cohort was a biopsy cohort of NAFLD patients (n = 279), and the second was a territory-wide cohort of NAFLD patients (n = 4603) retrieved from a territory-wide electronic healthcare database in Hong Kong. RESULTS: In detecting significant fibrosis, liver stiffness measured by transient elastography had the highest area under the receiver operating characteristic curve (AUROC) (.844), followed by SAFE score (.773). SAFE score had the highest AUROC among blood-based algorithms (.773 vs. .746 for FIB-4, .697 for APRI). Based on cut-off values of SAFE score (0 and 100 points), 854 (18.6%), 1596 (34.6%) and 2153 (46.8%) were in the low-, intermediate- and high-risk groups, respectively, in the territory-wide cohort. Six (.7%), 15 (.9%) and 59 (2.7%) developed liver-related events in those three groups respectively. Among patients who had liver-related events at 5 years, using the high cut-off, SAFE score could predict 84.9% of patients accurately, compared to 40.9% for FIB-4 and 27.2% for APRI. CONCLUSION: The SAFE score performed well and better than other blood-based markers in diagnosing significant fibrosis and predicting liver-related events in Asian patients with NAFLD.

Performance of continuous controlled attenuation parameter and liver stiffness measurement by the novel SmartExam in metabolic dysfunction-associated steatotic liver disease.

BACKGROUND & AIMS: FibroScan® Expert 630 and FibroScan® Mini+430 are novel vibration-controlled transient elastography devices equipped with the same SmartExam software, which allows continuous measurement of controlled attenuation parameter (CAP) during the entire examination. This study aims to compare the CAP variabilities and the quantification for liver fibrosis and steatosis between the conventional FibroScan and the SmartExam-equipped machines in patients with metabolic dysfunction-associated steatotic liver disease (MASLD). METHODS: This retrospective study included 118 patients with biopsy-proven MASLD who underwent liver biopsy at two tertiary centres between 2021 and 2023. Liver stiffness and steatosis measurements were performed using both FibroScan machines and M and XL probes for each individual. Liver histology was used as the reference standard for liver fibrosis and steatosis staging. RESULTS: Standard deviations of continuous CAP (cCAP) were significantly lower than those of CAP for all probes (p < .0001). CAP variability was significantly associated with body mass index (p < .01), probe selection (p < .001) as well as the random effect of centre. Only the effect of probe selection (p < .001) was significantly associated with cCAP variability. No significant difference was found in the performance of staging liver fibrosis and steatosis between two types of machines at the same cut-offs. CONCLUSIONS: The SmartExam-based VCTE reduces the variability of CAP measurement and achieves a similar accuracy as the FibroScan 502 device for the estimation of both hepatic steatosis and fibrosis. Future studies should determine if cCAP is a better tool to monitor changes in steatosis than the original CAP.

A pilot integrated model nurse clinic increases the uptake of antiviral treatment for the prevention of mother-to-child transmission of HBV.

BACKGROUND AND AIMS: Mother-to-child-transmission (MTCT) of hepatitis B virus (HBV) may still occur despite birth-dose HBV vaccinations when pregnant women are positive for hepatitis B surface antigen (HBsAg) with high viral loads (HBV DNA ≥ 200 000 IU/mL). A pilot integrated model nurse clinic (IMNC) was started in 2020 to implement the pre-emptive antiviral therapy with tenofovir disoproxil fumarate (TDF). We aimed to evaluate the performance of IMNC on uptake of TDF. METHODS: This was a territory-wide retrospective cohort of all consecutive HBsAg-positive women of child-bearing age with pregnancy records in public hospitals 2019-2022. Demographic characteristics, liver biochemistries and virologic parameters, and TDF use were collected. Concurrently, data from a prospective audit in Union Hospital, the private hospital with the highest number of deliveries in Hong Kong, from June 2022 to May 2023 were compared. RESULTS: The prevalence rate of HBV DNA ≥ 200 000 IU/mL in pregnant women with available HBV DNA records was 29.2% (66/226) in 2019, 27.3% (99/363) in 2020, 15.9% (125/784) in 2021 and 17.2% (117/679) in 2022 (p < .001), out of 2052 pregnant women who had their HBV DNA checked within 1 year prior to delivery. An increasing uptake rate of TDF by highly viraemic pregnant women (i.e. ≥ 200 000 IU/mL) was noted after the commencement of IMNC in public hospitals, with 67% (45/67) in 2019, 83% (88/106) in 2020, 91% (117/128) in 2021 and 89% (149/167) in 2022. Moreover, all highly viraemic pregnant women from Union Hospital received TDF. Continuous use of TDF was associated with a reduced risk of postpartum biochemical flare. CONCLUSIONS: IMNC increases the uptake of antiviral treatment in pregnant women at risk of MTCT of HBV. IMNC contributes to hepatitis elimination through a structured care plan to prevent MTCT of HBV.

Risk of alanine aminotransferase flare in patients with previous hepatitis B virus exposure on biological modifier therapies-A population-based study.

BACKGROUND: It is uncertain whether biological therapies would increase the risk of hepatitis among patients with past hepatitis B virus (HBV) infection. This study aimed to evaluate the risk of alanine aminotransferase (ALT) flare in patients with past HBV infection while using biological therapies. METHODOLOGY: Patients who received biological therapies for ≥3 months from 2000 to 2019 were identified from a population-based database in Hong Kong. Patients with past HBV infection were compared with a control group without prior HBV exposure. The primary endpoint was development of ALT flare within 5 years of starting biological therapies, defined as ALT >80 IU/L. RESULTS: There were 2471 and 2394 patients with and without past HBV infection respectively. There was a non-significant increase in risk of ALT flare among the HBV-exposed group (27.6% vs. 23.7%, p = .055). In multivariable analysis, using prednisolone-equivalent dose of >20 mg daily, male sex and concomitant immunosuppressants were risk factors for ALT flare. The risk of ALT flare was significantly higher with anti-CD20 when compared to other biological agents (36.1% vs. 14.5%, p < .01), but was not significantly different among anti-tumour necrosis factor, anti-cytokine, Janus kinase inhibitors and T cell/B cell inhibitors or anti-integrin (15.2% vs. 14.6% vs. 11.7% vs. 11.1%, p = .82). Among patients with documented hepatitis B surface antigen seroreversion, 96% were on anti-CD20. CONCLUSIONS: Our study further supports the current suggestion of prophylactic anti-viral before starting anti-CD20 in HBV-exposed patients. While other biological therapies appear to have a lower risk for ALT flare, this result needs further confirmation.

Timing of endoscopy for acute upper gastrointestinal bleeding: a territory-wide cohort study.

OBJECTIVE: While it is recommended that patients presenting with acute upper gastrointestinal bleeding (AUGIB) should receive endoscopic intervention within 24 hours, the optimal timing is still uncertain. We aimed to assess whether endoscopy timing postadmission would affect outcomes. DESIGN: We conducted a retrospective, territory-wide, cohort study with healthcare data from all public hospitals in Hong Kong. Adult patients (age ≥18) that presented with AUGIB between 2013 and 2019 and received therapeutic endoscopy within 48 hours (n=6474) were recruited. Patients were classified based on endoscopic timing postadmission: urgent (t≤6), early (6

Effectiveness of prophylactic clipping in preventing postpolypectomy bleeding in oral anticoagulant users: a propensity-score analysis.

BACKGROUND AND AIMS: Evidence of prophylactic clipping is inconsistent except for proximal and large colonic lesions in the general population. Although warfarin and direct oral anticoagulants (DOACs) are significant risk factors of postpolypectomy bleeding (PPB), dedicated studies to examine the benefit of prophylactic clipping in these high-risk patients remain limited. METHODS: We performed a propensity score-weighted retrospective cohort study from 2012 to 2020. Patients who received an oral anticoagulant and underwent colonoscopic polypectomy were included. Data were collected on baseline demographics, medications (anticoagulant, antiplatelet, and heparin bridging), and endoscopies (polyp number, location, size, morphology, histopathology, resection method and prophylactic clipping). Propensity-score models with inverse probability of treatment weighting were developed between prophylactic clipping and no clipping groups. Unbalanced variables were included in a doubly robust model with multivariate analysis. The primary outcome was clinically significant delayed PPB, defined as a composite endpoint of hemoglobin drop ≥2 g/dL, blood transfusion, or repeat colonoscopy for hemostasis within 30 days. RESULTS: Five hundred forty-seven patients with 1485 polyps were included. Prophylactic clipping was not associated with a reduced risk of PPB (odds ratio [OR], 1.19; 95% confidence interval [CI], .73-1.95; P = .497). The hot resection method was associated with a significantly higher risk of PPB (OR, 9.76; 95% CI, 3.94-32.60; P < .001) compared with cold biopsy or snare polypectomy. In a subgroup analysis, prophylactic clipping was associated with a lower PPB risk in patients on DOACs (OR, .36; 95% CI, .16-.82; P = .015). CONCLUSIONS: Prophylactic clipping was not associated with an overall reduced risk of PPB in patients on oral anticoagulants. The use of cold snare polypectomy should be maximized in anticoagulated patients.

Association of genetic variations with NAFLD in lean individuals.

BACKGROUND & AIMS: How adiposity influences the effect of genetic variants on non-alcoholic fatty liver disease (NAFLD) in the Asian population remains unclear. We aimed to study the association between genetic risk variants and susceptibility/severity of NAFLD in the lean, overweight and obese individuals. METHODS: Nine hundred and four community subjects underwent proton-magnetic resonance spectroscopy and transient elastography examination. Lean (<23 kg/m2 ), overweight (23-24.9 kg/m2 ) and obesity (≥25 kg/m2 ) were defined according to the body mass index cut-offs for Asians. NAFLD was defined as intrahepatic triglycerides ≥5%. PNPLA3, TM6SF2, MBOAT7 and 9 other gene polymorphisms were analysed by rhAMPTM SNP assays. RESULTS: Five hundred and twenty-nine (58.5%), 162 (17.9%) and 213 (23.6%) subjects were lean, overweight and obese, respectively. The prevalence of NAFLD was 12.4%, 41.4% and 59.1% in the three groups (P < .001). Amongst those with NAFLD, lean subjects (30.3%) were more likely to carry the PNPLA3 rs738409 GG genotype than overweight (17.9%) and obese subjects (17.4%) (P = .003). Compared with the CC genotype, the GG genotype was associated with the greatest increase in the risk of NAFLD in lean subjects (odds ratio [OR] 6.04), compared with overweight (OR 3.43, 95% CI [1.06, 11.14]) and obese subjects (OR 2.51, 95% CI [0.93, 6.78]). Additionally, the TM6SF2 rs58542926 TT genotype was associated with reduced serum triglycerides only in lean subjects. A gene-BMI effect was not observed for the other gene polymorphisms. CONCLUSIONS: The PNPLA3 rs738409 gene polymorphism has a greater effect on liver fat in Asian lean individuals than in overweight or obese ones.

No increased risk of flare in ulcerative colitis patients in corticosteroid-free remission after stopping 5-aminosalicylic acid: A territory-wide population-based study.

BACKGROUND AND AIM: Whether 5-aminosalicylic acid (ASA) can be stopped in patients with stable ulcerative colitis (UC) remains unclear. We aimed to examine whether 5-ASA can be safely withdrawn in UC patients who have been in corticosteroid-free clinical remission for ≥ 1 year. METHODS: This is a retrospective cohort study using territory-wide healthcare database in Hong Kong. Primary outcome was development of UC flare, defined as new corticosteroid use or UC-related hospitalizations within 5 years. UC patients on oral 5-ASA ≥ 2 g daily for ≥ 1 year with C-reactive protein (CRP) < 10 mg/dL and no 5-ASA dosage escalation, UC-related hospitalization or corticosteroid use in the past year were included. Patients on biological agents were excluded. Patients were classified as "stopping" if 5-ASA was withdrawn for ≥ 90 days within follow-up period. We performed multivariable Cox regression models adjusting for demographics, blood parameters and immunosuppressants used. Adjusted hazard ratio (aHR) with 95% confidence interval (CI) was reported comparing stopping and continuous-use groups. RESULTS: A total of 1408 patients were included with a median follow-up duration of 41.8 months (interquartile range [IQR]: 17.2-60.0 months). Stopping 5-ASA was not associated with an increased risk of UC flare (aHR 0.91; 95% CI 0.64-1.31; P = 0.620). A higher CRP levels at the time of stopping 5-ASA (aHR 1.15; 95% CI: 1.01-1.30; P = 0.037) were associated with increased risk of flare. CONCLUSION: Stopping 5-ASA in UC patients in corticosteroid-free remission for ≥ 1 year was not associated with increased risk of flare. Future prospective trials should evaluate the role of stopping 5-ASA in stable UC patients.

Secular trend of treatment uptake in patients with chronic hepatitis B: A territory-wide study of 135 395 patients from 2000 to 2017.

BACKGROUND AND AIMS: The uptake of antiviral treatment for patients with chronic hepatitis B (CHB) has been suboptimal. We aimed to determine the secular trend of treatment uptake in the territory-wide CHB cohort in Hong Kong from 2000 to 2017 and the factors for no treatment despite fulfilling treatment criteria. METHODS: Chronic hepatitis B patients under public clinics and hospitals were identified through electronic medical records. The treatment indications were defined according to the Asian-Pacific guidelines published at the time of patients' first appearance in four periods: 2000-2004, 2005-2009, 2010-2013, and 2014-2017. RESULTS: There were 135 395 CHB patients were included; 1493/12472 (12.0%), 7416/43426 (17.1%), 10 129/46559 (21.8%), 8051/32 938 (24.4%) patients fulfilled treatment criteria in the four periods, respectively. The treatment uptake rate increased with time: 35.1%, 43.4%, 60.2%, and 68.6% respectively. High fibrosis indices (APRI, FIB-4, and Forns indices) appeared to be the main factors for treatment indication in non-cirrhotic patients, with over 90% fulfilling treatment criteria due to high fibrosis indices alone. Of those fulfilling treatment criteria by high fibrosis indices, less than 60% of patients (25.2%, 36.1%, 46.0%, and 58.9%, respectively) had antiviral treatment initiated. Normal platelet count (odds ratio 0.42, P < 0.001) was the independent factor associated with not initiating antiviral treatment in patients fulfilling treatment criteria. CONCLUSIONS: Treatment uptake rates have been increasing over time. Normal platelet count, which reflects less advanced liver disease, precludes patients from receiving antiviral treatment even if treatment indication is fulfilled. Hence, the importance to identify non-cirrhotic patients with significant liver fibrosis should be emphasized.

Differential Impact of Extended Criteria Donors After Brain Death or Circulatory Death in Adult Liver Transplantation.

Using grafts from extended criteria donors (ECDs) and donation after circulatory death (DCD) donors is a strategy to address organ shortage in liver transplantation (LT). We studied the characteristics and outcomes of ECD and DCD grafts. We retrospectively studied consecutive adults who underwent deceased donor LT between 2006 and 2019. ECD was defined using modified Eurotransplant criteria. Our primary outcomes were graft and patient survival. A total of 798 grafts were used for LT, of which 93.1% were donation after brain death (DBD; 59.9% were also ECD) and 6.9% were DCD grafts (49.1% were also ECD). Among DBD graft recipients, donors having >33% liver steatosis or 3 ECD criteria resulted in poorer graft survival. Otherwise ECD graft recipients had similar graft and patient survival compared with non-ECD graft recipients. DCD graft recipients also had similar patient survival compared with DBD recipients. However, DCD grafts from an ECD appeared to have worse outcomes. DCD graft recipients experienced higher rates of early allograft dysfunction (50.9% versus 24.7%; P < 0.001) and ischemic biliopathy (16.4% versus 1.5%; P < 0.001) compared with DBD graft recipients. Use of DBD grafts from ECDs did not impact outcomes unless there was significant donor steatosis or 3 Eurotransplant criteria were met. DCD graft recipients have similar patient survival compared with DBD graft recipients as long as the donor was not an ECD. We recommend that DBD donors with 3 or more ECD features or >33% steatosis and DCD donors with any ECD features be used with caution in adult LT.

Development and validation of a model for hepatitis B e antigen seroconversion in entecavir-treated patients with chronic hepatitis B.

Achieving hepatitis B e antigen (HBeAg) seroconversion is a satisfactory endpoint during antiviral treatment for chronic hepatitis B (CHB). This study aimed to develop and validate a novel scoring system to predict HBeAg seroconversion during entecavir (ETV) treatment. A total of 526 patients with HBeAg-positive CHB treated with ETV for at least 1 year were randomly assigned to the training and validation cohorts. Baseline parameters including hepatitis B virus DNA, hepatitis B surface antigen (HBsAg), hepatitis B core antibody (HBcAb), and alanine aminotransferase level were quantified. Patients who achieved HBeAg seroconversion were compared with those without HBeAg seroconversion. A prediction model was established to predict HBeAg seroconversion during ETV treatment. After a median follow up of 2.67 years, 93 (36.0%) and 87 (32.5%) patients in the training and validation cohorts developed HBeAg seroconversion. A prediction score composed of age, HBsAg and HBcAb quantification was derived. Areas under receiver operating characteristic curve at 5 years of this prediction score were 0.70 and 0.72 in the training and validation cohorts. By using the dual cutoff values of 0.28 and 0.58, the model was endowed with high sensitivity and specificity to exclude or identify patients developing HBeAg seroconversion (90.3% sensitivity and 90.2% specificity in the training cohort as well as 92.8% sensitivity and 84.4% specificity in the validation cohort, respectively). A novel prediction score that uses baseline clinical variables was developed and validated. The score accurately estimates the probabilities of developing HBeAg seroconversion at 5-years ETV therapy in patients with CHB.

The safety of stopping nucleos(t)ide analogue treatment in patients with HBeAg-negative chronic hepatitis B.

BACKGROUND: The rates of hepatitis B surface antigen (HBsAg) seroclearance after stopping nucleos(t)ide analogues (NA) in European (19% in 2 years) and Asian (13% in 6 years) patients with chronic hepatitis B (CHB) vary dramatically. We evaluated the incidence of hepatitis flare and HBsAg seroclearance in hepatitis B e antigen (HBeAg)-negative Chinese CHB patients who had stopped NA. METHODS: This was a territory-wide retrospective study in Hong Kong. We identified HBeAg-negative CHB patients from January 2000 to December 2017 who had stopped NA treatment for more than 3 months. Hepatitis flare was defined as ALT >2×ULN. RESULTS: The 1076 patients were predominantly middle-aged men (mean age 52 years, male 74.8%) when starting NA; they stopped NA after 82 ± 35 months of treatment. At 44.3 ± 24.6 months after stopping NA, 147 (13.6%) patients had hepatitis flare, which led to resumption of NA; whereas 77 (7.2%) patients had flare but did not resume NA. Decompensation occurred in 7/914 (0.8%) patients. A total of 695 (64.6%) patients remained on NA treatment at the last visit. Eleven patients had achieved HBsAg seroclearance (6 of them had hepatitis flare and 1 of these 6 patients achieved HBsAg seroclearance after NA was restarted). Hepatic events developed in 75/695 (10.8%) patients who had NA resumed vs 43/381 (11.3%) patients who did not resume NA (P = .677). CONCLUSIONS: Hepatitis flare and retreatment were common in HBeAg-negative CHB patients who stopped NA treatment; whereas HBsAg seroclearance rarely occurred. Stopping NA to achieve functional cure should not be recommended at this moment.

Glucose-lowering drugs and outcome from COVID-19 among patients with type 2 diabetes mellitus: a population-wide analysis in Hong Kong.

OBJECTIVES: To investigate the association between baseline use of glucose-lowering drugs and serious clinical outcome among patients with type 2 diabetes. DESIGN: Territory-wide retrospective cohort of confirmed cases of COVID-19 between January 2020 and February 2021. SETTING: All public health facilities in Hong Kong. PARTICIPANTS: 1220 patients with diabetes who were admitted for confirmed COVID-19. PRIMARY AND SECONDARY OUTCOME MEASURES: Composite clinical endpoint of intensive care unit admission, requirement of invasive mechanical ventilation and/or in-hospital death. RESULTS: In this cohort (median age 65.3 years, 54.3% men), 737 (60.4%) patients were treated with metformin, 385 (31.6%) with sulphonylureas, 199 (16.3%) with dipeptidyl peptidase-4 (DPP-4) inhibitors and 273 (22.4%) with insulin prior to admission. In multivariate Cox regression, use of metformin and DPP-4 inhibitors was associated with reduced incidence of the composite endpoint relative to non-use, with respective HRs of 0.51 (95% CI 0.34 to 0.77, p=0.001) and 0.46 (95% CI 0.29 to 0.71, p<0.001), adjusted for age, sex, diabetes duration, glycated haemoglobin (HbA1c), smoking, comorbidities and drugs. Use of sulphonylureas (HR 1.55, 95% CI 1.07 to 2.24, p=0.022) and insulin (HR 6.34, 95% CI 3.72 to 10.78, p<0.001) were both associated with increased hazards of the composite endpoint. CONCLUSIONS: Users of metformin and DPP-4 inhibitors had fewer adverse outcomes from COVID-19 compared with non-users, whereas insulin and sulphonylurea might predict a worse prognosis.

Temporal matrix completion with locally linear latent factors for medical applications.

Regular medical records are useful for medical practitioners to analyze and monitor patient's health status especially for those with chronic disease. However, such records are usually incomplete due to unpunctuality and absence of patients. In order to resolve the missing data problem over time, tensor-based models have been developed for missing data imputation in recent papers. This approach makes use of the low-rank tensor assumption for highly correlated data in a short-time interval. Nevertheless, when the time intervals are long, data correlation may not be high between consecutive time stamps so that such assumption is not valid. To address this problem, we propose to decompose matrices with missing data over time into their latent factors. Then, the locally linear constraint is imposed on the latent factors for temporal matrix completion. By using three publicly available medical datasets and two medical datasets collected from Prince of Wales Hospital in Hong Kong, experimental results show that the proposed algorithm achieves the best performance compared with state-of-the-art methods.