Does azelastine nasal spray prevent COVID-19?

Unfortunately, probably not.

If you follow me on instagram you may have already seen my post about this, but I wanted to also adapt it into a text post!

As a tangent, I am long-winded and a chronic nuance enjoyer (a term my friend introduced me to that I love haha), so I’m actually wanting to make text posts over IG posts more and more. I think IG posts are more shareable and a lot of misinfo is spread on there, so maybe I’ll try doing both. The thing is–science and studies are complicated and should be discussed precisely with a lot of nuance. And it’s hard to fit that in an instagram post. As a final aside, I have a text post in the works about red flags for misinformation that touches on this and many other points that I’m excited about :).

Anyway, back to the topic at hand!

Lots of people have been sharing (mis)information about this study(1):

Screenshot of study details

So let’s run through some issues with it!

Table of contents:

1. Results from the primary endpoint of the study

2. Other statistical issues

3. Lack of interference testing

4. Tipping point analysis results

5. Reporting on the wrong population

6. Weird graph issue

7. Sparse info on compliance

8. More examples of the study contradicting itself

9. Limitations they identified

10. Longest conflicts of interest section I’ve ever seen!

11. Other concerns

12. Previous studies

13. Summary/TLDR

1. Results from the primary endpoint of the study

The authors state:

“In the ITT population, which constituted the primary analysis set, the incidence of PCR-confirmed SARS-CoV-2 infection rate (primary end point) was significantly lower in the azelastine group (5 of 227 participants [2.2%]) compared with the placebo group (15 of 223 participants [6.7%]) (risk difference [RD], −4.5 percentage points; 95% CI, −8.3 to −0.7; P = .02), translating to an OR of 0.31 (95% CI, 0.11-0.87) (Table 2). These findings were supported by the PP analysis (5 of 179 infections [2.8%] in the azelastine, 13 of 174 [7.5%] in the placebo group; RD, −4.7 percentage points; 95% CI, −9.3 to −0.1 percentage points; P = .046; OR, 0.36; 95% CI, 0.12 to 1.02, respectively).”

Ignoring the unbolded part (see section 5 for why), this basically means that they’re pretty sure that if you use the azelastine nasal spray, your risk of testing positive for COVID-19 is between 12-102 % of the risk of testing positive for COVID-19 on the placebo nasal spray. This is a huge range which includes numbers >100 % (which would indicate that the placebo actually lowers your risk more than the azelastine spray).

2. Other statistical issues

If I plug their results into a slightly different odds ratio calculator, I get P values >0.05, and ≥0.05 is typically the cutoff for people to say that there is no significant difference between two things. This would mean there is no significant difference in the risk of testing positive for COVID-19 whether participants are on the azelastine spray or the placebo spray.

The results from the secondary endpoints lack P values which is suspicious, and the reason for that is not explained (see quote from the study below).

“Secondary outcomes related to SARS-CoV-2 infections are presented with between-group differences and 95% CIs only; P values are omitted. Risk and mean differences are calculated as azelastine minus placebo. Other secondary end points and frequency of infections with other respiratory pathogens are reported descriptively without inferential statistics.”

Since P values can tell us whether or not two treatments are significantly different, why would they exclude them for these results?

3. Lack of interference testing

Some nasal sprays and nasal spray ingredients have been shown to interfere with tests for COVID-19(2,3) and other viruses(4).

Like all other COVID-19 nasal spray studies to date, this study did not check whether or not the test spray nor the placebo spray interferes with tests for the pathogens they tested for.

Since nasal sprays are sprayed up the nose and COVID-19 and other respiratory pathogen tests rely on nasal or nasopharyngeal swabs, it is extremely important to check (and an inexcusable oversight to not check) whether or not the sprays interfere with the test results.

4. Tipping point analysis results

In this study, some participants dropped out (11 in the azelastine group and 13 in the placebo group). The researchers assumed they were uninfected when they analyzed the data and reported on the results.

Since only 5 people on azelastine and 13 people on the placebo tested positive for COVID-19 in the PP population, if some of these dropouts actually did get infected, this could change the results a lot.

To look at this, they conducted a tipping point analysis, which showed that in about half of all the possible combinations of dropouts being infected or uninfected, there was no longer a significant difference between the azelastine and placebo groups when it came to the risk of testing positive for COVID-19.

5. Reporting on the wrong population

The researchers state repeatedly that the ITT population contains participants that made major protocol violations, and that therefore the ITT population won’t be used when looking at the primary endpoint of the study. See a few examples:

“The intention-to-treat (ITT) and safety populations consisted of all randomized participants (n = 450), while the per-protocol (PP) population included participants without major protocol deviations (n = 353, Figure 1; eTable 6 in Supplement 3 for details on protocol deviations).”

”The per protocol (PP) analysis set will include all evaluable probands who comply with the protocol in all points defined in the blind review and delivering a complete data set of measurements for the evaluation of the primary efficacy variable. Probands presenting major deviations will be excluded from the Per Protocol efficacy population and therefore will not be considered in the primary efficacy analysis.”

However, they also report on the ITT population repeatedly and make other conflicting statements such as:

“In the ITT population, which constituted the primary analysis set,“

The difference between the azelastine and placebo groups is larger in the ITT population than the PP one, which may be why they report on it. they also state:

“The PP analysis supported the ITT findings, though its statistical significance should be interpreted with caution given the low number of events.“

6. Weird graph issue

Upon close inspection of Figure 2 on the ITT population and supplemental eFigure 1 on the PP population, the number of azelastine infections looks consistent with what the authors reported but the number of placebo infections looks like one less than they report.

See Figure 2 below as an example. They report 5 azelastine infections and 15 placebo infections by day 56, but on the figure it looks like 5 azelastine infections and 14 placebo infections.

Figure 2 from the study

7. Sparse info on compliance

In the main text and supplemental, the authors make these statements:

“Participants daily documented the use of the investigational product and, if applicable, the occurrence of respiratory symptoms or adverse effects.”

”Significant deviations from the dosing regimen (e.g., missing multiple doses).”

Does this mean compliance is based on self reporting and not weighing of the spray bottle at the end of the study or something more objective?

Does it mean that participants used 3-5 sprays per nostril per day for 56+ days and were only allowed to miss one dose before they were removed from the study?

They report that only 6 of the 227 azelastine group and 5 of the 223 placebo group had insufficient adherence to the study medication. That seems low when we’re talking a minimum of 168 doses.

8. More examples of the study contradicting itself

In the results section of the study, they state:

“The mean (SD) duration of SARS-CoV-2 positivity, as measured by participant-reported RAT, was shorter in the azelastine group (3.40 [1.34] vs 5.14 [2.98] days with an MD of −1.74 [95% CI, −2.17 to −1.31] days).“

Whereas, in the discussion section of the study, they state:

“A reduction in duration of SARS-CoV-2 positivity, as measured by participant-reported RAT, could not be shown in this study.“

These statements directly contradict each other and one must be false.

9. The limitations they identified

“This randomized clinical trial has some limitations. The modest sample size and low incidence of infections for certain pathogens limited the statistical power for subgroup analyses. The PP analysis supported the ITT findings, though its statistical significance should be interpreted with caution given the low number of events. Limitations in the sensitivity of the RAT could have led to underreporting of asymptomatic SARS-CoV-2 infections despite close-meshed testing, in particular for the azelastine group. Such an effect with significant reduction in viral load and lack of symptoms would nevertheless be considered a benefit for the individual and likely result in decreased disease propagation due to the reduction in viral shedding. Symptom-triggered testing for non–SARS-CoV-2 pathogens likely has resulted in underreporting of non–SARS-CoV-2 infections. In addition, the bitter taste of azelastine nasal spray may have unblinded participants, potentially introducing a bias. Conversely, it cannot be ruled out that the placebo had an effect on the probability of infection because rinsing and diluting effects as well as the barrier-stabilizing properties of hypromellose could have contributed to infection prophylaxis.24 Because this was a single-center trial in a mostly healthy, vaccinated population, the generalizability of the findings to other settings may be limited.“

(includes using RATs as first tests before PCRS, low number of infections, azelastine tastes bitter and placebo doesn’t so people on azelastine may have guessed they were on the test spray)

10. Longest conflicts of interest section I’ve ever seen!

“Conflict of Interest Disclosures: Dr Lehr reported grants from Ursapharm as a study sponsor during the conduct of the study; personal fees from Saarmetrics GmbH as a founder and shareholder outside the submitted work. Dr Meiser reported personal fees from URSAPHARM Arzneimittel GmbH for employment outside the submitted work; and Dr Meiser is employed at URSAPHARM Arzneimittel GmbH, the sponsor of the CONTAIN trial. Dr Selzer reported grants from URSAPHARM Arzneimittel GmbH as a study sponsor during the conduct of the study; grants from the Scientific Consilience GmbH for constultant work for the company outside the submitted work. Dr Holzer reported personal fees from URSAPHARM Arzneimittel GmbH as CEO of the company outside the submitted work; and Frank Holzer is the CEO of URSAPHARM Arzneimittel GmbH, the sponsor of the CONTAIN trial. Dr Mösges reported personal fees from Ursapharm GmbH and grants from Ursapharm GmbH during the conduct of the study; personal fees from ALK, grants from ASITbiotech, personal fees from Allergopharma, personal fees from Bencard, grants from Leti, grants from Lofarma, nonfinancial support from Roxall, personal fees from Stallergenes, grants from Bencard, personal fees from Lofarma, nonfinancial support from Lofarma, grants from Stallergenes, personal fees from Optima, Friulchem, Hexai, Servier, and Klosterfrau, nonfinancial support from Atmos, personal fees from Bayer, nonfinancial support from Bionorica, personal fees from FAES, GSK, MSD, Johnson & Johnson, Meda, and Novartis, nonfinancial support from Novartis, nonfinancial support from Otonomy, personal fees from Stada and UCB, nonfinancial support from Ferrero, grants from Hulka, personal fees from Nuvo, Menarini, Mundipharma, and Pohl-Boskamp, grants from Inmunotek, personal fees from Cassella-med GmbH&Co KG, Laboratoire de la Mer, and Sidroga, grants from HAL BV, personal fees from HAL BV, Lek, PRO-AdWISE, Angelini Pharma, and JGL, nonfinancial support from JGL, grants from bitop, personal fees from bitop and Sanofi, grants from Probelte Pharma, personal fees from Probelte Pharma, Diater, and Worg Pharma, grants from Allergy Therapeutics, and personal fees from Allergy Therapeutics outside the submitted work. Dr Smola reported grants from URSAPHARM Arzneimittel GmbH as institutional funding to perform laboratory analysis for the present study during the conduct of the study. Dr Bals reported grants from Ursapharm Pharmaceuticals during the conduct of the study; grants from Deutsche Forschungsgemeinschaft, German Ministry for Research and Education, Schwiete-Foundation, and the State of Saarland, personal fees from CSL Behring, Grifols, AstraZeneca, GSK, and Regeneron outside the submitted work. No other disclosures were reported.

Funding/Support: Supported by URSAPHARM Arzneimittel GmbH.

Role of the Funder/Sponsor: URSAPHARM Arzneimittel GmbH (Saarbruecken, Germany) is the sponsor of the clinical trial and designed the trial in cooperation with academic partners. Data were collected by investigators in collaboration with a contract research organization (ClinCompetence Cologne GmbH, Cologne, Germany) and analyzed by the academic partners.“

11. Other concerns

Figure 2 (shown in section 6) and supplemental eFigure 1 could be misleading, as they show infections beyond day 56 that are not included in their analysis nor results.

This post is not an exhaustive list of the issues in this study, just some important ones.

12. Previous studies

Previous studies(5,6) looking at azelastine nasal spray and COVID-19 from the same group (which are also studies 11 and 12 in my post about how there’s no convincing evidence that nasal sprays prevent, nor treat, COVID-19), funded by the same company, have comments on PubPeer(7,8) pointing out many concerns about the studies. The authors have not responded to the concerns.

13. Summary/TLDR

This latest study on COVID-19 and azelastine nasal sprays is extremely flawed and extremely funded by the company that makes the spray.

COVID-19 influencers uncritically and unjustifiably hyping up this study is irresponsible and spreads misinformation. Those of us who still care about COVID-19 are supposed to be the science-backed side and should be spreading accurate information. Without taking the time to critically analyze these studies, we shouldn’t be posting about them.

Ultimately and unfortunately, pretending there’s convincing evidence that nasal sprays prevent or treat COVID-19 is not going to fight COVID-19.