Predictive risk factors for death, hospitalization, and colchicine response in SARS-CoV-2 people

Summaries 2 min 233

Font Size

Created On: 25/02/2022

This study was carried out to investigate the risk factors for hospitalization or mortality in ambulatory coronavirus-infected people and subsequently recognize high-risk people who will significantly benefit from early treatment with colchicine (a potent anti-inflammatory drug).

See All

Key take away

A simple predictive model (based on 7 risk factors) for SARS-CoV-2 hospital admission or mortality permits the identification of high risk people who can significantly benefit from colchicine treatment.

Background

Method

In placebo group (2084 volunteers), a stepwise multivariable logistic regression model with risk factors was evaluated. Additional assessment of the model was done on clinical grounds. C-statistic was assessed for final predictive model along with optimal threshold value for predicted probability as per Youden's index.

For internally validating the final model, a 10-fold cross-validation was performed on volunteers in the placebo group. For 4159 volunteers, risk scores and their anticipated probability were estimated as per beta coefficients of final predictive model derived in placebo arm and their individual values of retained risk factors.

For classifying low-risk and high-risk subjects, the optimal threshold value recognized from the final predictive model in the placebo arm was utilized. Utilizing logistic regression model, the subgroup assessment was carried out. It included risk subgroup variable (low or high risk), therapeutic arm (placebo/colchicine), and therapy group by risk subgroup variable interaction.

Result

Body-mass index, age, usage of diabetes drugs, gender, respiratory disorder history, anticoagulants usage and oral steroids usage at the randomization time were the seven variables retained in predictive model. For classifying low-risk people (with an anticipated probability below ideal threshold) and elevated-risk people (with an anticipated probability above ideal threshold), an optimal threshold value identified from predictive model was utilized.

The number required to treat/prevent 1 hospitalization or mortality with colchicine therapy dropped from 71 in whole study population (4159 subjects) to 29 in high-risk subgroup (1692 subjects).

Conclusion

In non-hospitalized people with early SARS-CoV-2 infection, a simple predictive model on the basis of medications and clinical data can serve to identify high-risk people who will principally benefit from early colchicine treatment.