Presentation: Statistical evidence from case series data in support of early outpatient COVID-19 treatment protocols

by Eleftherios Gkioulekas, Ph.D.

No transcript...

On March 23, 2020, Dr. Vladimir Zelenko announced his treatment protocol for COVID-19, which consisted of risk stratifying patients as high-risk vs low-risk, and treating the high-risk patients with a multidrug therapy consisting of a 5 day treatment with hydroxychloroquine, azithromycin, and zinc sulfate. Dr. Zelenko defined three categories of high-risk patients: (a) all patients older than 60 years; (b) patients that are immunocompromised or have comorbidities or have BMI > 30kg/m2; (c) all patients not satisfying the previous two conditions that develop shortness of breath.

Dr. Zelenko's protocol was incrementally improved by including a quercetin protocol for low-risk patients to reduce severity of symptoms, the adoption of dexamethasone and nebulized budesonide at the beginning of May 2020, and the adoption of the blood thinner Eliquis at the end of May 2020 and beginning of June 2020. With these adaptations, Dr. Zelenko's protocol evolved into what is now the well-known McCullough protocol [1].

The original Derwand-Scholz-Zelenko paper (DSZ) [2] reported 335 patients seen by Dr. Zelenko with PCR confirmed COVID-19 infection, of which 141 patients were high-risk and treated with the triple-drug therapy, resulting in 4 hospitalizations and 1 death. However, using Zelenko's complete data set, which included all patients that he diagnosed clinically with COVID-19, we learn that by April 28, 2020, Dr. Zelenko had seen 1,450 patients of which 405 patients were high-risk and were treated with his Zelenko protocol, with 6 hospitalizations and 2 deaths. By June 14, 2020, Dr. Zelenko had seen a total of 2200 patients, with 800 high-risk patients treated, resulting in 12 hospitalizations and 2 deaths.

Likewise, by December 2020, Dr. Brian Procter in North Texas had treated 869 high-risk patients, using an early version of the McCullough protocol, with 20 hospitalizations and 2 deaths, with high-risk classification defined using age > 50 years as a threshold [3,4]. Dr. Didier Raoult, in Marseilles, France, also treated, by December 2020, 10,429 patients, of which we can identify 1,495 patients as high-risk patients (with age > 60) with 106 hospitalizations and 5 deaths [5]. It should be highlighted that unlike early treatment doctors in America, who tried to prevent hospitalizations, due to the unsatisfactory protocols used in American hospitals, Dr. Raoult hospitalized his patients very liberally for close monitoring and in order to provide them very high-quality care, which was to a large extent consistent with the principles of the McCullough protocol. 

These observational results are all examples of case series. A case series is one half of a case control study, where we have a treatment group but do not have a control group. In a case control study, especially if it is randomized, it becomes possible to provide an unbiased measurement of the efficacy of treatment. However, when the signal of efficacy in a case series is sufficiently strong, in comparison against historical or population level outcomes without treatment, it becomes possible and sufficient to simply answer the question of whether using the treatment is better than doing nothing. That is, instead of providing an unbiased measurement of treatment efficacy, we can simply address the weaker question of whether any efficacy exists.

In the paper that I co-authored with Dr. Peter McCullough and Dr. Vladimir Zelenko [6], we presented a statistical technique for analysis of case series that calculates efficacy thresholds and random selection bias thresholds for establishing the existence of a mortality reduction benefit and a hospitalization reduction benefit. These thresholds can then be compared against the expected mortality and hospitalization rates for high-risk patients at the population-level to determine whether a beneficial effect exists. An assumption that underlies the analysis is that the medications used have a known acceptable safety profile, which can only be satisfied if we are considering treatment using repurposed medications.

When the efficacy threshold, for either mortality or hospitalization rate reduction, is exceeded, it shows the existence of some treatment benefit by the preponderance of evidence, which means that the effect is statistically significant if not overturned by selection bias and furthermore that it is more likely than not that random selection bias has not overturned the result. When the random selection bias threshold is exceeded, it shows that the existence of benefit is clear and convincing, which means that the effect is statistically significant and furthermore that we have statistically significant confidence that the result is not overturned by random selection bias.

Given a result that it is clear and convincing that using the proposed treatment is better than doing nothing, we argue that it becomes unethical to conduct randomized control trials, where a control group is randomly denied treatment. Given a result that there is preponderance of evidence that the proposed treatment is better than doing nothing, we argue, is sufficient for emergency adoption by treating physicians, within the ethical guidelines of article 37 of the 2013 Helsinki declaration [7].

In the paper [6] we argued that without treatment, for pre-omicron variants, the probability of hospitalization for high-risk patients is at least 10%, the probability of death for high-risk patients with comorbidities or shortness of breath is at least 5%, and the probability of death for patients with age > 60 years is at least 3.5% and for age > 50 it is at least 2.3%. As an underestimate of the probability of death without treatment for high-risk patients, we can also use the case fatality rate (CFR) of the entire population of the United States and France, including both low-risk and high-risk patients, since the treatments proposed by Zelenko and McCullough were unfortunately used only on a very small proportion of the overall population in both countries. In the pre-omicron period, until November 2021, the overall average CFR for both countries converged to 1.7%.

The data from the original Derwand-Scholz-Zelenko study [2], based on our threshold calculations, show the existence of some hospitalization rate reduction benefit by the preponderance of evidence. However, using Zelenko's complete data set shows that, by the end of April 2020, it was clear and convincing that the Zelenko protocol prevented hospitalizations and the preponderance of evidence was in favor of mortality rate reduction. By June 2020, the claim of mortality rate reduction crossed over to clear and convincing, based on Zelenko's results. By December 2020, we also had a clear and convincing claim of mortality rate reduction by the case series of Dr. Brian Procter and Dr. Didier Raoult, thus corroborating Zelenko's results.

Dr. Marc Rendell published a commentary discussing the implications of our work in further detail for a broader audience [8]. Suffice to say, millions of lives could have been saved, if we had adopted Zelenko's protocol by the end of April 2020. If we had done so then, by the end of summer 2020, it would have been evident that the epidemic curves are crashed, and we could have gone back to normal from September 2020 and onwards, at the latest.

Download slides

2022 Early Tx Talk
1.53MB ∙ PDF file

2022 Early Tx Talk Bw
827KB ∙ PDF file


[1] P.A. McCullough, P.E. Alexander, R. Armstrong, C. Arvinte, A.F. Bain, R.P. Bartlett, R.L. Berkowitz, A.C. Berry, T.J. Borody, J.H. Brewer, A.M. Brufsky, T. Clarke, R. Derwand, A. Eck, J. Eck, R.A. Eisner, G.C. Fareed, A. Farella, S.N.S. Fonseca, C.E. Geyer, Jr., R.S. Gonnering, K.E. Graves, K.B.V. Gross, S. Hazan, K.S. Held, H. Thomas Hight, S. Immanuel, M.M. Jacobs, J.A. Ladapo, L.H. Lee, J. Littell, I. Lozano, H.S. Mangat, B. Marble, J.E. McKinnon, L.D. Merritt, J.M. Orient, R. Oskoui, D.C. Pompan, B.C. Procter, C. Prodromos, J.C. Rajter, J-J. Rajter, C. V.S. Ram, S.S. Rios, H.A. Risch, M.J.A. Robb, M. Rutherford, M. Scholz, M.M. Singleton, J.A. Tumlin, B.M. Tyson, R.G. Urso, K. Victory, E.L. Vliet, C.M. Wax, A.G. Wolkoff, V. Wooll, V. Zelenko. "Multifaceted highly targeted sequential multidrug treatment of early ambulatory high-risk SARS-CoV-2 infection (COVID-19)", Reviews in Cardiovascular Medicine 21 (4) (2020), 517-530

[2] R. Derwand, M. Scholz, V. Zelenko. "COVID-19 outpatients - early risk-stratified treatment with zinc plus low dose hydroxychloroquine and azithromycin: a retrospective case series study", International Journal of Antimicrobial Agents 56 (2020), 106214

[3] B.C. Procter, C. Ross, V. Pickard, E. Smith, C. Hanson, P.A. McCullough. "Clinical outcomes after early ambulatory multidrug therapy for high-risk SARS-CoV-2 (COVID-19) infection", Reviews in Cardiovascular Medicine 21 (4) (2020), 611-614

[4] B.C. Procter, C. Ross, V. Pickard, E. Smith, C. Hanson, P.A. McCullough, "Early Ambulatory Multidrug Therapy Reduces Hospitalization and Death in High-Risk Patients with SARS-CoV-2 (COVID-19)", International Journal of Innovative Research in Medical Science 6 (2021), 219-221

[5] M. Million, J-C. Lagier, H. Tissot-DuPont, I. Ravaux, C. Dhiver, C. Tomei, N Cassir, L. DeLorme, S. Cortaredona, S. Gentile, E. Jouve, A. Giraud-Gatineau, H. Chaudet, L. Camoin-Jau, P. Colson, P. Gautret, P-E. Fournier, B. Maille, J-C. Deharo, P. Habert, J-Y. Gaubert, A. Jacquier, S. Honore, K. Guillon-Lorvellec, Y. Obadia, P. Parola, P. Brouqui, D. Raoult, IHU COVID-19 Task Force. "Early Treatment with Hydroxychloroquine and Azithromycin in 10,429 COVID-19 Outpatients: A Monocentric Retrospective Cohort Study", Reviews in Cardiovascular Medicine 22 (2021), 1063-1072

[6] E. Gkioulekas, P.A. McCullough, V. Zelenko: "Statistical analysis methods applied to early outpatient COVID-19 treatment case series data", COVID 2(8) (2022), 1139-1182

[7] World Medical Association. "World Medical Association Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Subjects", JAMA 310(20) (2013), 2191-2194.

[8] M. Rendell: "Commentary on 'Statistical Analysis Methods Applied to Early Outpatient COVID-19 Treatment Case Series Data' by Gkioulekas, Mccullough and Zelenko: A Return Back to the Future" Journal of Health Care Communications 7 (10) (2022), 130-135

Share Discover with Dr. Eleftherios Gkioulekas


Discover with Dr. Eleftherios Gkioulekas
Discover with Dr. Eleftherios Gkioulekas Podcast
Research, teaching, and commentary
Listen on
Substack App
RSS Feed
Email mobile setup link
Appears in episode
Eleftherios Gkioulekas