BackgroundThere is overwhelming volume of confirmed cases of COVID-19, despite this numerous knowledge gaps remain in the diagnosis, management, and prognostication of this novel coronavirus infection, making prevention and control a challenge.MethodsThis retrospective cohort study included patients with real-time reverse transcriptase polymerase chain reaction (rRT-PCR)-confirmed COVID-19. Binary logistic regression was used to determine the association between the cardiac biomarkers and in-hospital mortality. ROC, AUC, and cutoff analyses were used to determine optimal cutoff values for the cardiac biomarkers.ResultsA total of 90 subjects with a complete panel of cardiac biomarkers out of the 224 rRT-PCR confirmed cases were included. The median age was 57 years (IQR, 47–67 years), majority were males. Sixty-six (77.6%) subjects survived while 19 (22.4%) expired. The most common presenting symptom was fever (75.6%), and the most common comorbidity was hypertension (67.8%). Spearman rho correlation analysis showed moderate positive association of high sensitivity troponin I (hsTnI) with in-hospital mortality (R, 0.434, p = <0.001). Multivariate binary logistic regression analysis showed that creatine kinase and hsTnI were independently associated with in-hospital mortality (OR, 4.103 [95% CI, 1.241–13.563], p=0.021; and OR, 7.899 [95% CI, 2.430–25.675], p=0.001, respectively). ROC curve analysis showed that hsTnI was a good predictor for in-hospital mortality (AUC, 0.829 [95% CI, 0.735–0.923], p = <0.001) and that creatine kinase was a poor predictor (AUC, 0.677 [95% CI, 0.531–0.823], p=0.018). Optimal cutoff point derived from the ROC curve for hsTnI was 0.010 ng/ml (J, 0.574) with a sensitivity of 84% (TPR, 0.842 [95% CI, 0.604–0.966]), specificity of 73% (TNR, 0.732 [95% CI, 0.614–0.386]), and an adjusted negative predictive value of 99% (Known prevalence*adjusted NPV, 0.989), a positive likelihood ratio of 20% (LR+, 3.147 [95% CI, 2.044–4.844]) and a negative likelihood ratio of 30% (LR−, 0.216 [95% CI, 0.076–0.615]).ConclusionHigh sensitivity troponin I level was a good tool with a very high negative predictive value in significantly predicting in-hospital mortality among rRT-PCR positive COVID-19 patients.Funding AcknowledgementType of funding sources: None. ROC Curve
Read full abstract