Potential and limitations of machine meta-learning (ensemble) methods for predicting COVID-19 mortality in a large inhospital Brazilian dataset

The majority of early prediction scores and methods to predict COVID-19 mortality are bound by methodological flaws and technological limitations (e.g., the use of a single prediction model). Our aim is to provide a thorough comparative study that tackles those methodological issues, considering multiple techniques to build mortality prediction models, including modern machine learning (neural) algorithms and traditional statistical techniques, as well as meta-learning (ensemble) approaches. This study used a dataset from a multicenter cohort of 10,897 adult Brazilian COVID-19 patients, admitted from March/2020 to November/2021, including patients [median age 60 (interquartile range 48-71), 46% women]. We also proposed new original population-based meta-features that have not been devised in the literature. Stacking has shown to achieve the best results reported in the literature for the death prediction task, improving over previous state-of-the-art by more than 46% in Recall for predicting death, with AUROC 0.826 and MacroF1 of 65.4%. The newly proposed meta-features were highly discriminative of death, but fell short in producing large improvements in final prediction performance, demonstrating that we are possibly on the limits of the prediction capabilities that can be achieved with the current set of ML techniques and (meta-)features. Finally, we investigated how the trained models perform on different hospitals, showing that there are indeed large differences in classifier performance between different hospitals, further making the case that errors are produced by factors that cannot be modeled with the current predictors.

Hospital characteristics associated with COVID-19 mortality: data from the multicenter cohort Brazilian Registry

The COVID-19 pandemic caused unprecedented pressure over health care systems worldwide. Hospital-level data that may influence the prognosis in COVID-19 patients still needs to be better investigated. Therefore, this study analyzed regional socioeconomic, hospital, and intensive care units (ICU) characteristics associated with in-hospital mortality in COVID-19 patients admitted to Brazilian institutions. This multicenter retrospective cohort study is part of the Brazilian COVID-19 Registry. We enrolled patients ≥ 18 years old with laboratory-confirmed COVID-19 admitted to the participating hospitals from March to September 2020. Patients' data were obtained through hospital records. Hospitals' data were collected through forms filled in loco and through open national databases. Generalized linear mixed models with logit link function were used for pooling mortality and to assess the association between hospital characteristics and mortality estimates. We built two models, one tested general hospital characteristics while the other tested ICU characteristics. All analyses were adjusted for the proportion of high-risk patients at admission. Thirty-one hospitals were included. The mean number of beds was 320.4 ± 186.6. These hospitals had eligible 6556 COVID-19 admissions during the study period. Estimated in-hospital mortality ranged from 9.0 to 48.0%. The first model included all 31 hospitals and showed that a private source of funding (β = - 0.37; 95% CI - 0.71 to - 0.04; p = 0.029) and location in areas with a high gross domestic product (GDP) per capita (β = - 0.40; 95% CI - 0.72 to - 0.08; p = 0.014) were independently associated with a lower mortality. The second model included 23 hospitals and showed that hospitals with an ICU work shift composed of more than 50% of intensivists (β = - 0.59; 95% CI - 0.98 to - 0.20; p = 0.003) had lower mortality while hospitals with a higher proportion of less experienced medical professionals had higher mortality (β = 0.40; 95% CI 0.11-0.68; p = 0.006). The impact of those association increased according to the proportion of high-risk patients at admission. In-hospital mortality varied significantly among Brazilian hospitals. Private-funded hospitals and those located in municipalities with a high GDP had a lower mortality. When analyzing ICU-specific characteristics, hospitals with more experienced ICU teams had a reduced mortality.

Predictors of venous thromboembolism in COVID-19 patients: results of the COVID-19 Brazilian Registry

Previous studies that assessed risk factors for venous thromboembolism (VTE) in COVID-19 patients have shown inconsistent results. Our aim was to investigate VTE predictors by both logistic regression (LR) and machine learning (ML) approaches, due to their potential complementarity. This cohort study of a large Brazilian COVID-19 Registry included 4120 COVID-19 adult patients from 16 hospitals. Symptomatic VTE was confirmed by objective imaging. LR analysis, tree-based boosting, and bagging were used to investigate the association of variables upon hospital presentation with VTE. Among 4,120 patients (55.5% men, 39.3% critical patients), VTE was confirmed in 6.7%. In multivariate LR analysis, obesity (OR 1.50, 95% CI 1.11-2.02); being an ex-smoker (OR 1.44, 95% CI 1.03-2.01); surgery ≤ 90 days (OR 2.20, 95% CI 1.14-4.23); axillary temperature (OR 1.41, 95% CI 1.22-1.63); D-dimer ≥ 4 times above the upper limit of reference value (OR 2.16, 95% CI 1.26-3.67), lactate (OR 1.10, 95% CI 1.02-1.19), C-reactive protein levels (CRP, OR 1.09, 95% CI 1.01-1.18); and neutrophil count (OR 1.04, 95% CI 1.005-1.075) were independent predictors of VTE. Atrial fibrillation, peripheral oxygen saturation/inspired oxygen fraction (SF) ratio and prophylactic use of anticoagulants were protective. Temperature at admission, SF ratio, neutrophil count, D-dimer, CRP and lactate levels were also identified as predictors by ML methods. By using ML and LR analyses, we showed that D-dimer, axillary temperature, neutrophil count, CRP and lactate levels are risk factors for VTE in COVID-19 patients.

Systematic review and meta-analysis of ivermectin for treatment of COVID-19: evidence beyond the hype

BACKGROUND

The role of ivermectin in the treatment of COVID-19 is still under debate, yet the drug has been widely used in some parts of the world, as shown by impressive market data. The available body of evidence may have changed over the last months, as studies have been retracted and "standards of care" (SOC) used in control groups have changed with rapidly evolving knowledge on COVID-19. This review aims to summarize and critically appraise the evidence of randomized controlled trials (RCTs) of ivermectin, assessing clinical outcomes in COVID-19 patients.

METHODS

RCTs evaluating the effects of ivermectin in adult patients with COVID-19 were searched through June 22, 2022, in four databases, L.OVE platform, clinical trial registries and pre-prints platforms. Primary endpoints included all-cause mortality and invasive ventilation requirement. Secondary endpoint was the occurrence of adverse events. Risk of bias was evaluated using the Cochrane Risk of Bias 2.0 tool. Meta-analysis included only studies which compared ivermectin to placebo or SOC. Random-effects were used to pool the risk ratios (RRs) of individual trials. The quality of evidence was evaluated using GRADE. The protocol was register in PROSPERO (CRD42021257471).

RESULTS

Twenty-five RCTs fulfilled inclusion criteria (n = 6310). Of those, 14 compared ivermectin with placebo, in night ivermectin associated with SOC was compared to SOC and two studies compared ivermectin to an active comparator. Most RCTs had some concerns or high risk of bias, mostly due to lack of concealment of the randomization sequence and allocation, lack of blinding and high number of missing cases. Ivermectin did not show an effect in reducing mortality (RR = 0.76; 95%CI: 0.52-1.11) or mechanical ventilation (RR = 0.74; 95%CI: 0.48-1.16). This effect was consistent when comparing ivermectin vs. placebo, and ivermectin associated with SOC vs. SOC, as well as in sensitivity analysis. Additionally, there was very low quality of evidence regarding adverse effects (RR = 1.07; 95%CI: 0.84-1.35).

CONCLUSIONS

The evidence suggests that ivermectin does not reduce mortality risk and the risk of mechanical ventilation requirement. Although we did not observe an increase in the risk of adverse effects, the evidence is very uncertain regarding this endpoint.