Merck’s Molnupiravir paper in NEJM, imputed values study
From TrialSiteNews By Leo Goldstein December 20, 2021
This adds details to Merck’s Molnupiravir paper in NEJM contains fabricated data and corrects reasoning about the Annex, Table S6. Table S6 has the same baseline viral load value 6.81 (on log10 scale) in both Molnupiravir and Placebo arms. This is not a typing mistake. The respective values in the sub-groups also differ by only 0.05, and perfectly sum up weighted to 6.8069 and 6.8079, respectively. The difference is only 0.001, quite unlikely to happen by chance. And it is only the beginning.
The paper acknowledges in fine print that only nasopharyngeal viral load range of 500 to 500 million copies/mL (corresponding to the log10 range 2.7 – 8.7) were measured and recorded correctly. For samples with viral load outside of this range, imputed values were recorded:
“The quantitative assay to generate these data was the Q2 SARS-CoV-2 Viral Load Quantitation Assay, with lower limit of quantification of 500 copies/ml and upper limit of quantification of 500,000,000 copies/ml. Post-baseline results below or above these limits were included in the mean and the mean change from baseline, with imputed value 499 and 500,000,001, respectively.”(NEJM Annex, p.23)
This means that if the viral load grew to 5 billion after the start of treatment (which is within a usual range), it is recorded and used in calculations as 500M (plus 1).

The excuse Merck used for this was that these were the limits of the quantitation assay from Q2Solutions. This explanation does not work for the upper bound. The researchers could have diluted the sample as much as they needed. For the lower bound, they could have used a more sensitive quantitation assay.
As can easily be seen from the means and standard deviations, the value above the upper bound should have occurred quite frequently on days 3 and 5. Additionally, the lower bound became a limit for the mean, in all groups and subgroups, making accurate reporting of virus clearance by Day 29 impossible.
This suggests that Merck had something to hide, probably that less patients in the Molnupiravir treatment group cleared the virus by Day 29, compared to the placebo group. Given Molnupiravir’s mechanism of action, this raises the possibility that Molnupiravir breeds SARS-COV-2 variants that can survive for a long period, causing chronic or periodic disease.
Because of the mechanism of action, Day 3, rather than the “baseline”, should be used as the starting point. Molnupiravir driven mutations cannot do not take effect until the first generation of mutated virions became the majority. If NHC concentrations in the body remain constant, at ~3 mcg/mL, this would happen in approximately 36-48 hours. Even at this point, the effect may not be strong enough and may require a few more generations. So, at the earliest, Day 3 is when Molnupiravir starts to work.
Recalculated in reference to Day 3, viral load change values show that Molnupiravir is associated with slower viral clearance compared to placebo (the column Difference).