An interesting comment in The Lancet published online July 9th.
The health emergency triggered by the COVID-19 pandemic has led to the massive use of pharmacological treatments whose efficacy has not been sufficiently evidenced.
Gradually, thanks to the results of standardised clinical trials, the efficacy of the different treatments used have become clearer, of which only dexamethasone and tocilizumab have so far been shown to reduce mortality associated with COVID-19.
One of the most widely used drugs in the treatment of COVID-19 has been azithromycin, since it has known antiviral properties and immunomodulatory effects, from which benefits have been obtained in the treatment of other respiratory diseases.
However, clinical trials with results published so far have not been able to show its efficacy in the treatment of COVID-19.
Large clinical trials such as RECOVERY have reasonably ruled out any benefit of azithromycin as a standalone therapy in patients admitted to hospital, in terms of reducing mortality and duration of hospital stay.
However, the efficacy of antiviral drugs often varies depending on the stage of the disease course at the time they are administered because of factors related to the host’s immune reaction and the existing organic damage at the time of therapy.
The ATOMIC2 study by Timothy Hinks and colleagues recruited 295 patients whose symptoms brought them to the emergency room, where they were diagnosed with COVID-19, but whose condition was not considered serious enough to be admitted (participant age was 45.9; 152, 52% were men and 143, 49% were women).
Therefore, the study focuses on a specific stage in the evolution of COVID-19, of intermediate severity, which is not well represented in previous studies with patients admitted to hospital, not in the principal study, which recruited patients diagnosed in primary care.
The ATOMIC2 study did not show a reduction in the number of admissions or deaths in the first 28 days of follow up. Neither did principle study show any varying from these same result. Nor did the two previous trials that tested efficacy demonstrate clinical benefits of azithromycin, in combination with the hydroxychloroquine, in patients seen outside the hospital.
The most immediate conclusion is that azithromycin is not useful in any phase of COVID-19. However, many of these studies share a series of limitations, imposed by the difficult emergency situation in which they were conducted, requiring a cautious interpretation in the absence of positive results from azithromycin.
First, placebo control or the masking of patient’s investigators to treatment has often not been visible.
Furthermore, many of the patients enrolled in ATOMIC2 and in other studies did not have diagnostic confirmation by PCR.
Obtaining a PCR test result in patients not admitted to hospital is especially difficult, given that the test was unavailable at some points during the epidemic, given the waiting times that have often been necessary to obtain results of the tests outside the hospital.
On the one hand, limiting the analyses to patients with positive PCR results would have caused the exclusion of a large number of patients who actually had COVID-19 but did not have a PCR test, or had falsely negative results. On the basis of high clinical suspicion and the efficacy studies of azithromycin who do not actually have COVID-19.
In the case of the ATOMIC2 study, the authors controlled for this limitation, replicating analysis in the subgroup of patients with positive PCR results. Although these analyses were under powered, no substantially different results were obtained.
The adherence of patients to treatment represents another important challenge in studies with patients cared for in the community, because it is much more difficult to a guarantee than in patients admitted to hospital.
Low adherence, in combination with intention – treat statistical analyses, might cause a potential effect of azithromycin to be missed, because of its dilution in the total set of patients, of which some did not take the drug in the end, or did not take it in sufficient doses.
For this reason, it is advisable to size the studies sufficiently, as it is also possible to carry out analyses per protocol, with statistical power adding in equation on the potential effects that might have under initial adherence conditions.
In the ATOMIC2 studies, a long course of high dose azithromycin was prescribed 500 mg for 14 days, which turned out to be partly protective against low adherence, since even the patients in that group did not comply with the planned schedule took an average of six doses of azithromycin – demonstrating the strength of his clinical trial.
Additionally, it is worth in analysing the important difference between the expected and observed effect sizes in the ATOMIC2 study.
In the ATOMIC2 study, the results show a frequency of primary outcome events (hospital admission or death) in the 28 days from randomisation of 10.3% (15 of 145 patients) in the azithromycin group 11.6% (17 of 147 patients) in the control group.
This difference represents a small risk reduction of 1.3%, which was not statistically significant, corresponding to a number needed to treat of 77, whereas the study was sized to detect an expected risk reduction of 10%, corresponding to an NNT of 10.
The question is, was a 10% risk reduction in 28 days to be expected?
It has been argued that a smaller reduction will probably not change clinical practice, but against this argument is the indisputable fact that the RECOVERY study changed clinical practice worldwide when it showed that dexamethasone treatment reduced mortality by 2.9% at 28 days in patients who needed oxygen.
The RECOVERY study has also since shown that tocilizumab produces a 4.1% reduction in the risk of death in 28 days in patients admitted to hospital.
Notably, this result contradicts those obtained in most previous trials on these drugs since these trials did not have a sufficient sample size to show such modest efficacy.
Thus, the actual efficacy that some drugs have shown so far in COVID-19 is not as great as researchers who designed ATOMIC2 expected, nor the other studies with azithromycin in out-of-hospital patients.
Therefore, the results of these trials being very valuable, only established a limited efficacy, ruling out a prominent effect of azithromycin, but they do not have the power to detect more modest effects which could have clinical relevance.
The ATOMIC2 adds much to our knowledge, but the body of evidence is not yet complete.
There is still room for studies with power to demonstrate modest therapeutic effects, and studies focused on other outcomes that are of great interest. For example, symptomatic persistence in COVID-19, sequalae, and other results that need longer follow up periods, about which we still know little and for which we do not know therapies capable of their prevention.
The London General Practice, the leading London doctors’ clinic, undertakes all forms of COVID testing, including fit for fly, day 2, day 5 and day 8 testing, test for release testing.
It also undertakes a full diagnostic service for COVID-19 and operates a COVID assistance safety net service for those suffering with acute COVID as well as looking after those with long COVID.
Dr Paul Ettlinger
BM, DRCOG, FRCGP, FRIPH, DOccMed