ACTIV-6-Ivermectin Found No COVID-19 Effect but Fraud and a Suspect Statistic Limit Impact

The ACTIV-6 trial reported that a “higher” dose of ivermectin failed to shorten symptom duration in a group of patients with mild to moderate severity COVID-19. The trial considered the effect of a maximum targeted dose of 600 μg/kg daily for 6 days. An earlier arm of the trial found no effect of 400 μg/kg, daily for 3 days. The investigators concluded:

“These findings do not support the use of ivermectin in patients with mild to moderate COVID-19.”

The impact of the trial is limited by an obvious public fraud and by a suspect statistic.

Fraud

The actual endpoint of the trial was:

The primary outcome was time to sustained recovery, defined as at least 3 consecutive days without symptoms.

However, in Orwellian defiance of reality, the trial endpoints are declared in ClinicalTrials.gov as:

Primary Outcome Measures :

1. Number of hospitalizations as measured by patient reports. [ Time Frame: Up to 28 days ]

2. Number of deaths as measured by patient reports [ Time Frame: Up to 28 days ]

3. Number of symptoms as measured by patient reports [ Time Frame: Up to 28 days ]

Suspect Statistic

The primary endpoint, time to sustained recovery, is estimated using a Bayesian logistic regression model. In addition to the randomization assignment (ivermectin/placebo) 8 additional covariates are included. Presumably, the purpose for inclusion of the additional covariates is to compensate for randomization failure, although that is not stated. The primary endpoint covariates are:

1. Randomization assignment

2. Age as a restricted cubic spline with 3 knots

3. Gender

4. Duration of symptoms prior to treatment

5. Calendar time as restricted cubic spline with 4 knots

6. Vaccination status

7. Geographic region (Northeast, Midwest, South, West)

8. Call center indicator

9. Additional appendix-specific covariates relevant to baseline disease severity and patient risk.

Additionally, an analysis was performed for “Heterogeneity of treatment effect”; the interaction between the treatment and various clinical, demographic and administrative variables. These are similar but not identical to the variables used in the regression model:

1. Vaccination status

2. Calendar time (pandemic epoch) as a restricted cubic spline with 3 knots

3. Duration of symptoms (time from symptom onset) as a restricted cubic spline with 3 knots

4. Age as a restricted cubic spline with 3 knots

5. BMI as a restricted cubic spline with 3 knots

6. Patient reported symptom severity (none, mild, moderate, severe)

Investigators claim that all of the variables from the primary endpoint regression model are analyzed in the heterogeneity of treatment effect:

“For each characteristic, a proportional hazards regression model was constructed using the same covariates as the primary endpoint model plus additional interaction terms between treatment assignment and the characteristic of interest.”

However, that is obviously false. For example, “Geographic Region” is obviously excluded from the heterogeneity analysis. That variable might show variations in outcomes between trial sites. More to the point, if there were any irregularities in randomization etc.

Conclusions

The ACTIV-6 trial is a disgrace to the medical establishment and to the nation. If the data relating to the original endpoints is subsequently found to support the use of ivermectin in COVID-19, the suppression of that data is treason, pure and simple.