I was interested in this, from the context of the Bangladesh ivermectin + doxycycline trials. Oh, can we subtract out the doxy effect?

Unfortunately, this trial has the same problem as every other no-control study: what would have happened without the doxy? Are we seeing a dramatic improvement or is this just the normal course of disease?

That said, IMO this is an impressive result.

Results

Eighty-nine (89) high-risk patients, who developed a sudden onset of fever, cough, SOB, and hypoxia and were diagnosed with COVID-19, were treated with DOXY (100 mg PO or intravenous (IV) for seven days) and regular standard of care. Eighty-five percent (85%) of patients (n=76) demonstrated clinical recovery that is defined as resolution of fever (average 3.7 days, Coeff = -0.96, p = 0.0001), resolution of SOB (average 4.2 days), and improvement of POX: average 84% before treatment and average 95% after treatment (84.7 ± 7% vs. 95 ± 2.6%, p = 0.0001). Higher pre- and post-treatment POX is associated with lower mortality (oxygen saturation (Spo2) vs. Death, Coeff = -0.01, p = 0.023; post-Spo2 vs. Death, Coeff = -0.05, p = 0.0002). Within 10 days of symptom onset, 3% of patients (n=3) were transferred to hospital due to clinical deterioration and 11% of patients (n=10) died. The result was followed for 30 days from the onset of symptoms in each patient.

Thanks for the excellent title! :)

Edit: I'm confused.

Eighty-five percent (85%) of patients (n=76) demonstrated clinical recovery that is defined as resolution of fever (average 3.7 days, Coeff = -0.96, p = 0.0001)

I expect a p-value to represent the probability that something is not due to chance. Typically a study will have two arms and show the p-value of the difference in some outcome measure.

What's the p-value 0.0001 comparing? I can't figure it out.

permalink