Ivermectin and COVID-19: How a Flawed Database Shaped the Pandemic Response of Several Latin-American Countries

Ivermectin and COVID-19: How a Flawed Database Shaped the Pandemic Response of Several Latin-American Countries

ivermectina preprint post


[Authors: Carlos Chaccour, Assistant Research Professor at ISGlobal and BOHEMIA Chief Scientific Officer; Joe Brew, data scientist; and Alberto García-Basteiro, Assistant Research Professor at ISGlobal and medical doctor at the International Health Service of Hospital Clínic]


Over the last few weeks we have seen the inclusion of ivermectin in the national therapeutic guidelines for COVID-19 of Peru, mass drug administration of ivermectin to 350,000 people for treatment or prevention of COVID-19 in Bolivia, Paraguay restricting the ivermectin market and advocacy groups in Colombia aiming for national ivermectin policy. Why is this happening?

Ivermectin is an antiparasitic drug used for river blindness, lymphatic filariasis and other Neglected Tropical Diseases. It also has some antiviral effect against single-strain RNA viruses like Dengue and Yellow fever. Early in April, researchers from Australia reported that Ivermectin inhibits the replication of SARS-CoV-2 in vitro. They used concentrations that are not readily achieved in the human body but the biological plausibility opened the doors for clinical trials given the drugs excellent safety profile and lack of effective treatment for COVID-19. In this guest editorial at the American Journal of Tropical Medicine and Hygiene we call for scientific rigor and provide rationale for conducting trials. Our own trial on this subject SAINT, was launched on May 13.

However, the policy decisions in Latin-American have been largely based on the analysis presented in a pre-print posted in the SSRN repository in early April by Patel et al. Though only a pre-print, this manuscript has been very impactful: it has been downloaded 15,655 times, its abstract has been viewed 89,895 times (as of May 28, 2020).

The policy decisions of ivermectin in Latin-American have been largely based on the analysis presented in a pre-print posted in the SSRN repository in early April

The authors claim to have used data from the Surgical Outcomes Collaborative (Surgisphere Corporation, Chicago, IL, USA). According to a recent publication in The Lancet (discussed below), the data included in this collaborative platform is “de-identified data obtained by automated data extraction from inpatient and outpatient electronic health records, supply chain databases, and financial records. In other words, there is some form of collaboration agreement with hundreds of hospitals using electronic records from around the world that allow this private corporation to automatically retrieve patient’s data periodically. At least in the EU, this seems to go directly against several points of the EU General Data Protection Regulation (GDPR) and the Privacy Shield EU-US collaboration scheme for the transfer of personal data.

Several concerns about this database have been raised based on the recent Hydroxychloroquine analysis published in The Lancet by the same authors of the ivermectin pre-print. These concerns are largely addressed elsewhere.

The first version of the ivermectin pre-print was posted on April 6*. This version evaluated data from 1,970 critically ill hospitalized patients diagnosed with COVID-19 with lung injury requiring mechanical ventilation from 169 hospitals across Asia, Europe, Africa, North and South America between January 1st 2020 and March 1st 2020. This included 52 patients treated with ivermectin, three of these patients (critically ill, requiring ventilation) came from African hospitals, but by March 1st, only two COVID-19 cases had been confirmed in the whole African continent.

After finding out this discrepancy, we contacted the authors via email and the answers received left our concerns unchanged. Additionally, the manuscript presented a survival analysis with serious methodological flaws (Figure 1):

Figure 1


The first version was removed and substituted by a second version of this by April 19*. This new version included data from 1,408 PCR-confirmed, hospitalized patients diagnosed with COVID-19 between January 1, 2020 and March 31, 2020. The data came from 169 hospitals in three continents. Half of these patients (704) had received a single dose of ivermectin (150 mcg/kg) and were matched “exactly on age, sex, race, underlying co-morbidity including chronic obstructive pulmonary disease (COPD), history of smoking, history of hypertension, diabetes mellitus, coronary artery disease, other cardiac disease, an index of illness severity (qSOFA) as well as medication use including hydroxychloroquine, azithromycin and corticosteroids”.

The outcomes from this analysis are a 65% reduction in the need for mechanical ventilation (7.3 vs 21.3%) and an 83% reduction in the overall death rate (1.4% vs 8.5%) in patients treated with ivermectin (figure 2). There are however two problems** with these data:

  • The need for mechanical ventilation in untreated patients seems quite high. As an example, in Spain, as of May 20, 2020, only 9% of all hospitalized patients (11,454 out of 124,521) have required ICU admission, a prerequisite for mechanical ventilation [Source: official statistics from ISCiii).
  • The data described in the manuscript do not match the figure.


In spite of these flaws this analysis has been cited in a white paper advocating for ivermectin to be included in the national COVID-19 treatment guidelines of Peru. This was followed shortly by a ministerial level communication recommending the use of ivermectin for COVID-19, albeit recognizing the lack of evidence and requesting informed consent. This however led to a black market and alleged distribution of veterinary formulations. All in spite of some strong voices from local scientific leaders including the ex-minister of Health Patricia García.

In spite of these flaws this analysis has been cited in a white paper advocating for ivermectin to be included in the national COVID-19 treatment guidelines of Peru

Health authorities from Bolivia has followed closely and even went a step further into distributing 350,000 ivermectin doses in the city of Trinidad. In Paraguay the authorities had to restrict ivermectin sales after a surge in demand.

This off-label use of ivermectin entails several risks:

  • Diversion of drug supply, causing shortages for its use in proven indications.
  • The use of veterinary formulations or non-supervised doses could lead to unforeseen side effects that can harm ongoing mass treatment schemes at community. level such as the Mectizan Donation Program which managed to eradicate river blindness in Colombia just a few years ago.
  • Rural regions of Latin America have a high prevalence of intestinal helminths. These parasite are known to modulate one type of immune response that favors viral clearance. Mass deworming due to ivermectin could have repercussions on the severity of COVID-19.
  • Moral hazard, due to a false feeling of protection or treatment with the drug.
  • Impossibility to conduct clinical trials should ivermectin become the new standard of care.

Once again, scientific rigor is needed, even in pandemic times.

Related resources

ASTMH editorial

One-hour conversation with clubes de ciencia Bolivia on this subject

Our own clinical trial, SAINT


[*06/02/2020: The first version of this text indicated that the preprint on ivermectin appeared on April 16, while it was actually April 6. The release date of the second version of the preprint has also been updated. Initially it was written that it had been published on April 21, while the correct date is April 19.]

[**06/04/2020: Initially it was stated that the second version of the preprint had three problems with the data. The second of the problems said that the mortality rate in hospitalized COVID-19 patients receiving mechanical ventilation was too low, based on a case series of 5,700 patients hospitalized with COVID-19 in New York which showed a mortality among those receiving mechanical ventilation was 88%. These data came from a paper by Richardson et al. which has been corrected. For more details, see https://jamanetwork.com/journals/jama/fullarticle/2765367]