Some bad apples have spoilt the meta-studies which first promoted ivermectin (common deworming agent) as a promising treatment to COVID-19.
Some of these clinical trials overviews were made online within weeks and found to have impossible numbers, unexplainable cohort mismatches or inconsistent timelines. They also had significant methodological weaknesses.
Since then, one of the preprint analyses was withdrawn. A second has been revised after being found to contain fraudulent data.
Despite all the serious errors, millions of doses have been administered to COVID-19 victims around the globe. Others who have not contracted the virus are using it as a preventative and potentially putting their health at risk.
Scientists are calling for immediate remediation of the meta analysis process in order to prevent it from happening again.
The Nature authors published a letter arguing that we shouldn't include any studies in a metaanalysis unless we have access the individual patient data (IPD).
The clinical trial should be canceled if the original authors of the study are not able or willing to provide this information. Researchers claim that such simple standards would have prevented the publication of meta-studies on Ivermectin.
The authors acknowledge that IPD review is something they recommend, but the potential for harm to patients on a global level demands more.
Gideon MeyerowitzKatz, an epidemiologist, said that the meta-analysis process is almost entirely based on trust. Since fraud is almost impossible to prove, no checks are made.
This means that some meta-studies rely on experimental data that may not have been collected.
Meyerowitz-Katz stated that there was evidence that a number of studies from the literature were fraudulently included in meta-analyses of ivermectin. These have been included in dozens of meta analyses without any qualms for months," ScienceAlert.
"It's only by reviewing the actual line data that fraud can be detected of this nature, so that should become a standard practice."
This is what happened with ivermectin this summer. A few meta-analyses in July showed that the antiparasite medicine was effective for controlling COVID-19 infection. However, evidence from the drug quickly disappeared when a closer examination began.
Currently, there's no evidence that ivermectin is safe for COVID-19 treatment. The wrong dose can also be dangerous as the Centers for Disease Control and Prevention have warned repeatedly since August.
Five people were hospitalized just this weekend after they took the drug COVID-19 in Oregon.
Ivermectin can cause overdose if it is taken incorrectly. It can also interact with blood thinners and cause side effects such as nausea, vomiting and low blood pressure.
The Nature letter recently stated that the current pandemic "provides fertile grounds for even poorly supported claims of efficacy being amplified, both within the scientific literature as well as on social media."
The authors add that "this context" can lead to rapid translation of nearly any seemingly favorable conclusion from a relatively small trial or set of trials into widespread clinical practices and public policy.
Scientists have pointed this out for years, and many have called for an update to the long-accepted meta analysis practice.
Ofttimes, a meta analysis is deemed more reliable than a single well-designed clinical trial. However, this is not always true. The validity of a meta-analysis is ultimately dependent on the quality of the studies that were included. However, not all scientific journals have the same quality control.
A meta-analysis may not be able to select the most relevant trials for inclusion but it can include other questionable data. This can make all of the difference.
For example, ivermectin's meta-analyses were biased by a small number of studies that had falsified or possibly falsified data.
It's difficult to correct or clarify incorrect information once it is out there. It is difficult to change people's minds even if certain conclusions are proven wrong. This is evident from our experience with vaccines.
It is vital to stop false information from being leaked in the first instance. Some scientists believe that cracks in meta-analysis processes need to be fixed. Some scientists suggest that we eliminate meta-studies completely, as they may not contribute to scientific progress at all and can actually muddy the waters.
(Meta-analyses don't require any original laboratory work. It is possible that authors only desire a publication record.
The authors of the letter suggest that we double-check all raw data used in meta-analyses before making any bold claims. It is not clear if this will actually happen.
Meyerowitz-Katz stated that "our recommendations are simple, easy to adopt," ScienceAlert.
"I don't think many people will accept them, but they could."
Nature published the letter.
Disclosure statement: Gideon Meyerowitz–Katz previously wrote articles for ScienceAlert.