Machine learning algorithms for predicting determinants of COVID-19 mortality in South Africa.

Background: COVID-19 has strained healthcare resources, necessitating efficient prognostication to triage patients effectively. This study quantified COVID-19 risk factors and predicted COVID-19 intensive care unit (ICU) mortality in South Africa based on machine learning algorithms. Methods: Data for this study were obtained from 392 COVID-19 ICU patients enrolled between 26 March 2020 and 10 February 2021. We used an artificial neural network (ANN) and random forest (RF) to predict mortality among ICU patients and a semi-parametric logistic regression with nine covariates, including a grouping variable based on K-means clustering. Further evaluation of the algorithms was performed using sensitivity, accuracy, specificity, and Cohen's K statistics. Results: From the semi-parametric logistic regression and ANN variable importance, age, gender, cluster, presence of severe symptoms, being on the ventilator, and comorbidities of asthma significantly contributed to ICU death. In particular, the odds of mortality were six times higher among asthmatic patients than non-asthmatic patients. In univariable and multivariate regression, advanced age, PF1 and 2, FiO2, severe symptoms, asthma, oxygen saturation, and cluster 4 were strongly predictive of mortality. The RF model revealed that intubation status, age, cluster, diabetes, and hypertension were the top five significant predictors of mortality. The ANN performed well with an accuracy of 71%, a precision of 83%, an F1 score of 100%, Matthew's correlation coefficient (MCC) score of 100%, and a recall of 88%. In addition, Cohen's k-value of 0.75 verified the most extreme discriminative power of the ANN. In comparison, the RF model provided a 76% recall, an 87% precision, and a 65% MCC. Conclusion: Based on the findings, we can conclude that both ANN and RF can predict COVID-19 mortality in the ICU with accuracy. The proposed models accurately predict the prognosis of COVID-19 patients after diagnosis. The models can be used to prioritize COVID-19 patients with a high mortality risk in resource-constrained ICUs.

Evaluating determinants of treatment outcomes among tuberculosis patients in the mining district of Butha Buthe, Lesotho.

Background: Before the COVID-19 pandemic, tuberculosis (TB) was the leading infectious cause of death globally. In low- and middle-income countries (LMIC) including Lesotho, treatment outcome is lower than the recommended rate and poor TB treatment outcomes remain a programmatic challenge. The aim of this study was to determine unfavourable treatment outcomes and associated risk factors among TB patients in Butha Buthe district. Methods: This was a retrospective record review of TB patients registered between January 2015 and December 2020. Data were collected from TB registers and patients' files and entered Microsoft Excel 2012. Analysis was conducted using R and INLA statistical software. Descriptive statistics were presented as frequencies and percentages. The differences between groups were compared using Pearson's X 2 test in bivariate analysis. Frailty Cox proportional hazards model was used to determine the risk of unfavourable outcomes among the variables. Results: A total of 1792 TB patients were enrolled in the study with about 70% males (1,257). Majority (71.7%) of the patients were between 20 and 59 years old, with 48% of the patients being unemployed. Almost a quarter of the patients (23.1%) had unfavourable outcomes with death (342 patients) being the most common unfavourable outcome. Our study has shown that patients older than 59 years, and unemployment increased the risk of having unfavourable treatment outcomes. Death was the most common unfavourable outcome followed by lost-to-follow up. We also observed that the patients in the initiation phase of treatment died at a faster rate compared to those in the continuation phase (p=0.02). Conclusion: TB treatment programs should have efficient follow-up methods geared more toward elderly patients. Active case finding to identify population at risk should be part of a TB program which would improve early diagnosis and treatment initiation. Patients in the intensive phase of the treatment program should be monitored more closely to determine adverse drug effects and nutritional requirement to prevent death during this phase of treatment.

The association between acid-base status and clinical outcome in critically ill COVID-19 patients admitted to intensive care unit with an emphasis on high anion gap metabolic acidosis.

OBJECTIVE: The aim of this study was to identify arterial blood gas (ABG) abnormalities, with a focus on a high anion gap (AG) metabolic acidosis and evaluate outcomes in coronavirus disease 2019 (COVID-19) patients admitted to the ICU. METHODS: A retrospective, observational study was conducted in a tertiary hospital in Cape Town during the first and second COVID-19 waves. Age, gender, sodium (Na), potassium (K), chloride (Cl), bicarbonate (HCO3std), pH, partial pressure of carbon dioxide (pCO2), creatinine, estimated glomerular filtration rate (eGFR), lactate levels and ABG results were obtained. The Pearson χ2 test or Fisher exact test and the Wilcoxon rank-sum test were used to compare mortality and survival. To identify factors associated with non-survival, a multivariable model was developed. RESULTS: This study included 465 patients, 226 (48%) of whom were female. The sample population's median (IQR) age was 54.2 (46.1-61.3) years, and 63% of the patients died. ABG analyses found that 283 (61%) of the 465 patients had alkalosis (pH ≥ 7.45), 65 (14%) had acidosis (pH ≤ 7.35) and 117 (25%) had normal pH (7.35-7.45). In the group with alkalosis, 199 (70.3%) had a metabolic alkalosis and in the group with acidosis, 42 (64%) had a metabolic acidosis with an increased AG of more than 17. Non-survivors were older than survivors (56.4 years versus 50.3 years, p < .001). CONCLUSION: Most of the COVID-19 patients admitted to the ICU had an alkalosis, and those with acidosis had a much worse prognosis. Higher AG metabolic acidosis was not associated with patients' characteristics.

Clinical characteristics associated with mortality of COVID-19 patients admitted to an intensive care unit of a tertiary hospital in South Africa.

BACKGROUND: Over 130 million people have been diagnosed with Coronavirus disease 2019 (COVID-19), and more than one million fatalities have been reported worldwide. South Africa is unique in having a quadruple disease burden of type 2 diabetes, hypertension, human immunodeficiency virus (HIV) and tuberculosis, making COVID-19-related mortality of particular interest in the country. The aim of this study was to investigate the clinical characteristics and associated mortality of COVID-19 patients admitted to an intensive care unit (ICU) in a South African setting. METHODS AND FINDINGS: We performed a prospective observational study of patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection admitted to the ICU of a South African tertiary hospital in Cape Town. The mortality and discharge rates were the primary outcomes. Demographic, clinical and laboratory data were analysed, and multivariable robust Poisson regression model was used to identify risk factors for mortality. Furthermore, Cox proportional hazards regression model was performed to assess the association between time to death and the predictor variables. Factors associated with death (time to death) at p-value < 0.05 were considered statistically significant. Of the 402 patients admitted to the ICU, 250 (62%) died, and another 12 (3%) died in the hospital after being discharged from the ICU. The median age of the study population was 54.1 years (IQR: 46.0-61.6). The mortality rate among those who were intubated was significantly higher at 201/221 (91%). After adjusting for confounding, multivariable robust Poisson regression analysis revealed that age more than 48 years, requiring invasive mechanical ventilation, HIV status, procalcitonin (PCT), Troponin T, Aspartate Aminotransferase (AST), and a low pH on admission all significantly predicted mortality. Three main risk factors predictive of mortality were identified in the analysis using Cox regression Cox proportional hazards regression model. HIV positive status, myalgia, and intubated in the ICU were identified as independent prognostic factors. CONCLUSIONS: In this study, the mortality rate in COVID-19 patients admitted to the ICU was high. Older age, the need for invasive mechanical ventilation, HIV status, and metabolic acidosis were found to be significant predictors of mortality in patients admitted to the ICU.

Haematological predictors of poor outcome among COVID-19 patients admitted to an intensive care unit of a tertiary hospital in South Africa.

BACKGROUND: Studies from Asia, Europe and the USA indicate that widely available haematological parameters could be used to determine the clinical severity of Coronavirus disease 2019 (COVID-19) and predict management outcome. There is limited data from Africa on their usefulness in patients admitted to Intensive Care Units (ICUs). We performed an evaluation of baseline haematological parameters as prognostic biomarkers in ICU COVID-19 patients. METHODS: Demographic, clinical and laboratory data were collected prospectively on patients with confirmed COVID-19, admitted to the adult ICU in a tertiary hospital in Cape Town, South Africa, between March 2020 and February 2021. Robust Poisson regression methods and receiver operating characteristic (ROC) curves were used to explore the association of haematological parameters with COVID-19 severity and mortality. RESULTS: A total of 490 patients (median age 54.1 years) were included, of whom 237 (48%) were female. The median duration of ICU stay was 6 days and 309/490 (63%) patients died. Raised neutrophil count and neutrophil/lymphocyte ratio (NLR) were associated with worse outcome. Independent risk factors associated with mortality were age (ARR 1.01, 95%CI 1.0-1.02; p = 0.002); female sex (ARR 1.23, 95%CI 1.05-1.42; p = 0.008) and D-dimer levels (ARR 1.01, 95%CI 1.002-1.03; p = 0.016). CONCLUSIONS: Our study showed that raised neutrophil count, NLR and D-dimer at the time of ICU admission were associated with higher mortality. Contrary to what has previously been reported, our study revealed females admitted to the ICU had a higher risk of mortality.

Latent class analysis: an innovative approach for identification of clinical and laboratory markers of disease severity among COVID-19 patients admitted to the intensive care unit.

Objective: The aim of this study was to identify clinical and laboratory phenotype distribution patterns and their usefulness as prognostic markers in COVID-19 patients admitted to the intensive care unit (ICU) at Tygerberg Hospital, Cape Town. Methods and results: A latent class analysis (LCA) model was applied in a prospective, observational cohort study. Data from 343 COVID-19 patients were analysed. Two distinct phenotypes (1 and 2) were identified, comprising 68.46% and 31.54% of patients, respectively. The phenotype 2 patients were characterized by increased coagulopathy markers (D-dimer, median value 1.73 ng/L vs 0.94 ng/L; p < 0.001), end-organ dysfunction (creatinine, median value 79 µmol/L vs 69.5 µmol/L; p < 0.003), under-perfusion markers (lactate, median value 1.60 mmol/L vs 1.20 mmol/L; p < 0.001), abnormal cardiac function markers (median N-terminal pro-brain natriuretic peptide (NT-proBNP) 314 pg/ml vs 63.5 pg/ml; p < 0.001 and median high-sensitivity cardiac troponin (Hs-TropT) 39 ng/L vs 12 ng/L; p < 0.001), and acute inflammatory syndrome (median neutrophil-to-lymphocyte ratio 15.08 vs 8.68; p < 0.001 and median monocyte value 0.68 × 109/L vs 0.45 × 109/L; p < 0.001). Conclusion: The identification of COVID-19 phenotypes and sub-phenotypes in ICU patients could help as a prognostic marker in the day-to-day management of COVID-19 patients admitted to the ICU.

Predicting COVID-19 outcomes from clinical and laboratory parameters in an intensive care facility during the second wave of the pandemic in South Africa.

Background: The second wave of coronavirus disease 2019 (COVID-19) in South Africa was caused by the Beta variant of severe acute respiratory syndrome coronavirurus-2. This study aimed to explore clinical and biochemical parameters that could predict outcome in patients with COVID-19. Methods: A prospective study was conducted between 5 November 2020 and 30 April 2021 among patients with confirmed COVID-19 admitted to the intensive care unit (ICU) of a tertiary hospital. The Cox proportional hazards model in Stata 16 was used to assess risk factors associated with survival or death. Factors with P<0.05 were considered significant. Results: Patients who died were found to have significantly lower median pH (P<0.001), higher median arterial partial pressure of carbon dioxide (P<0.001), higher D-dimer levels (P=0.001), higher troponin T levels (P=0.001), higher N-terminal-prohormone B-type natriuretic peptide levels (P=0.007) and higher C-reactive protein levels (P=0.010) compared with patients who survived. Increased standard bicarbonate (HCO3std) was associated with lower risk of death (hazard ratio 0.96, 95% confidence interval 0.93-0.99). Conclusions: The mortality of patients with COVID-19 admitted to the ICU was associated with elevated D-dimer and a low HCO3std level. Large studies are warranted to increase the identification of patients at risk of poor prognosis, and to improve the clinical approach.

A Review of Adherence and Predictors of Adherence to the CONSORT Statement in the Reporting of Tuberculosis Vaccine Trials.

The statement on Consolidated Standards of Reporting Trials (CONSORT) ensures transparency in the reporting of randomized trials. However, it is unclear if the statement has led to improvement in the quality of reporting of tuberculosis (TB) vaccine trials. We explored the quality of reporting of TB vaccine trials according to the latest version of the CONSORT statement, released in 2010. We searched PubMed and the Cochrane Central Register of Controlled Trials in August 2019. We conducted screening, study selection, and data extraction in duplicate; and resolved differences through discussion. We assessed reporting to be adequate if trials reported at least 75% of the CONSORT 2010 items. We conducted a trend analysis to assess if there was improvement in reporting over time. We also used logistic regression to assess factors associated with adequate reporting. We included 124 trials in the analyses. The mean proportion of adherence was 67.3% (95% confidence interval 64.4% to 70.1%), with only 46 (37%) trials having adequate reporting. There was a significant improvement in the quality of reporting over time (p < 0.0001). Trials published in journals with impact factors between 10 and 20 were more likely to have adequate reporting (odds ratio 9.4; 95% confidence interval 1.30 to 67.8), compared to lower-impact-factor journals. Despite advances over time, the reporting of TB vaccine trials is still inadequate and requires improvement.