Longitudinal serum chloride as a marker for poor prognosis in severely ill COVID-19 patients: A joint model approach

Coronavirus Disease 2019 (COVID-19) has been responsible for an international health crisis that demanded incredible efforts from the scientific community to understand the disease, its risk factors, and consequences. In this work, we seek to investigate the role of serum chloride and main Strong Ion Difference (mSID) on the survival of severely ill COVID-19 patients treated in the Intensive Care Unit (ICU). Electrolyte measurements were taken daily from each patient from their admission to death or discharge. ICU survival time was measured in days from admission, and discharged patients were considered censored. The longitudinal trajectories of chloride and mSID were associated with patients’ survival times using joint models for longitudinal and time-to-event data. A total of 58 COVID-19 patients were enrolled, and 21 died during hospitalization. Older patients had lower concentrations of chloride overall, and both chloride and mSID increased over time for patients on average. Age was a significant risk factor for ICU mortality, along with the slope of estimated longitudinal trajectory of chloride. Patients with decreasing chloride levels during ICU stay had increased hazard of death. This results of this study suggest that acquired hypochloremia may be an important marker of disease progression and risk of death in patients with severe COVID-19. As such, chloride should be further validated as longitudinal marker for monitoring prognosis during the course of ICU stay. Neither the current values or the slope of the trajectory of mSID were associated with mortality in this sample. The association between the longitudinal trajectory’s slope and ICU mortality is important to understand the dynamics of the disease on a patient level, and could only be quantified by using the joint model framework.