The media celebrates another anti-Hydroxychloroquine paper Observational Study of Hydroxychloroquine in Hospitalized Patients with Covid-19 (Geleris et al.), published yesterday in the prestigious New England Journal of Medicine. The paper “finds no evidence of benefit” of HCQ for COVID-19, AP claims.

This was an observational non-randomized study repeats the errors of previous studies of this kind:

• The patients that received HCQ were in much worse condition than those who did not.
• HCQ treatment was started too late, when the patients were already in the emergency room.

Doctors could decide to whom to administer HCQ. They selected the most severe patients:

• Practically ALL patients in the HCQ group had acute respiratory distress syndrome (ARDS), as defined by PaO2/FiO2 ratio (160-303; P/F<300 indicates ARDS). The non-HCQ group had P/F in the 248-431 range. Normal P/F is about 500. The average P/F ratio for the HCQ group was 223 compared with 360 for the non-HCQ. ADRS is the last stage in most COVID-19 patients.
• 49% of the HCQ group had hypertension vs. <7% in the non-HCQ group

Statistical methods cannot eliminate such disparities. Even after torturing the data by Cox proportional-hazards regression models, propensity-score matching , multivariable logistic-regression model etc., a large difference of P/F ratios remained.

Weirdly, the study treated death and intubation as the same outcome. I wonder why.

via Science Defies Politics

https://ift.tt/3cfL9vt

May 8, 2020 at 10:14AM