5,976 Matching Annotations
  1. Last 7 days
    1. On 2021-08-21 14:43:23, user Paul-Olivier Dehaye wrote:

      This paper has now been published (and presumably peer-reviewed).

      New title: Adherence and Association of Digital Proximity Tracing App Notifications With Earlier Time to Quarantine: Results From the Zurich SARS-CoV-2 Cohort Study

      https://www.ssph-journal.or...

      It still includes in the introduction, despite showing the opposite:

      These findings provide evidence that DPT may reach exposed contacts faster than MCT, with earlier quarantine and potential interruption of SARS-CoV-2 transmission chains.

    1. On 2021-08-23 10:32:30, user ingokeck wrote:

      Dear authors, I have a problem following your comparison between vaccinated cases and non-vaccinated cases. I understand how you select the vaccinated cases with your flowchart (thank you for providing one, this really helps to understand!), but I don't understand how you create the non-vaccinated controls. If you simply add up all non-vaccinated cases, you will get a huge bias towards non-vaccinated cases, as the vaccination campaign was still ongoing during your analysis period. So you will need to account for the differences in exposure, i.e. for a vaccinated case in week 10 which got infected in week 8 you need to look at the dose2 percentage even 2 weeks earlier, i.e. week 6 to normalize the exposure risk between vaccinated and non-vaccinated persons. It would be interesting to see the result. If you can share the anonymized raw data, I may be able to help.

    1. On 2021-08-25 12:07:48, user Prof. W Meier-Augenstein, FRSC wrote:

      What other than the difference in antibody titer post-vaccination and post-infection is the take-home message of this study? Surely, the decline in antibody titer per se months after vaccination or primary infection is not a surprising finding but could be expected? Antibodies have a finite life-span given by their Ig specific half-life (for example 21 days for IgGs). In the absence of a subsequent challenge (e.g. by a secondary infection) antibodies formed in response to the challenge posed by vaccination or primary infection will have all but cleared from serum after 6+ months. Furthermore, the difference in antibody titer between mRNA vaccinated and SARS-CoV-2 infected could not have come as a big surprise either considering mRNA vaccination results in expression of spike-protein “only” which means in contrast to a viral infection host cells are actually not infected and do not reproduce copious amounts of the virus which will take longer to fight and clear from the body than the spike protein. For the same reason, macrophages (phagocytes) are unlikely to be involved in the mRNA vaccinated group to the same degree as they are in the group infected by the virus. The natural decline of IG antibodies produced in response to the mRNA vaccine does not offer an exclusive explanation for breakthrough infection. Instead, breakthrough infection occurring 146+ days post vaccination are most likely the result of a “perfect storm”, an unfortunate coincidence of the higher virulence of the Delta variant of <<7 days incubation time, the associated higher viral load produced, and the fact the production of neutralising antibodies by B-memory cells takes up to 4-5 days to reach its peak.

    1. On 2021-08-26 03:53:13, user IMO wrote:

      Interesting study. Funny how little discussion there is from the media or the public health machine about immunity conferred by infection. No mention from on high about extending "passport" privileges to those who have been infected and then chosen not to get vaccinated. What's up with that?

    2. On 2021-08-28 11:49:17, user Doug Truitt wrote:

      So if one lives in Kentucky previous infection is less protective than vaccination - https://www.cdc.gov/mmwr/vo... - and if one lives in Israel previous infection is more protective than vaccination (this study). I'd be interested in discussion as to why these two studies are at odds with one another.

    3. On 2021-08-29 05:35:10, user some_guy wrote:

      Is there any variable control?

      The first question that pops into my mind is: are we comparing the same populations?

      People who were first vaccinated are disproportionately old people with weaker immune systems, more likely to be infected anyways. I wonder how much does this impact the results.

    4. On 2021-08-30 07:37:30, user 4qmmt wrote:

      This paper has one major flaw that is not discussed: We know for certain who was vaccinated and of those, 199 had symptoms. Though 8 of the matched recovered had symptoms, they were assumed to have been infected because of a + PCR test in the past, which as we all know produces tons of false positives (Israel's Ct is 35-40). In fact, the paper's own data shows that of the 238 PCR+ in the vaccinated cohort, only199 were actually positive, i.e., symptomatic = ill, and in the recovered cohort, of 19 PCR+, only 8 were symptomatic, suggesting at least 60% false positives. Thus, while the number of vaccinated is certain, the true number of previously infected is at most 8, but could in fact be 0, as the Cleveland Clinic study found.

      In fact, last weeks' MoH data in Israel shows that 73% of PCR + tests are on people with no symptoms, i.e., not infected. So this paper is actually saying that the risk of infection for vaccinated vs. recovered is between 27X to Infinite X.

    5. On 2021-09-01 03:04:45, user bgoo2 wrote:

      To all the people commenting on this article who have no idea how to critically evaluate study methodology.... let's just skip over the lack of control variables and omission of addressing the Santa's naughty list of biases.

      Let's just get right to the bones of it... while you are searching for your "natural immunity" as opposed to vaccination (how did that go with Polio, by the way? And Tetanus? And Rubella? And Hepatitis? ... hint: those vaccines came out before the Twitter era) ... where the FAHQK do you intend to keep the millions of symptomatic infected people who require hospitalization while they recover?

      How did that work out in 2020? How is that working out now? Every whack job shouting about natural immunity doesn't seem to acknowledge that to get there... you have to sacrifice a whole lot of people who would've otherwise been protected by a vaccine.

      Alabama USA has less than 5 million people. 38% of the population is double vaccinated as of August 31. Nearly 2500 people are in hospital with mild to severe Covid symptoms.

      Ontario Canada has 15 million people. Triple the population. 67% double vaccinated. Less than 350 hospitalized.

      We can repeat these comparisons across the globe. The results of this years' vaccination program is completely undeniable.

      The results in an under-vaccinated population is undeniable.

      Thankfully, each day, the percentage of unvaccinated grows smaller and smaller.

      And it is very quickly isolating who the remaining lunatics are in your community. (hint: they're ironically usually the ones who clearly have trouble saying NO to refined carbohydrates)

      Literally, that's the end of the story.

    6. On 2021-09-04 14:10:39, user criticalscientist wrote:

      It is likely that emergence of the Delta variant, which has 10 mutations in the spike protein, is due to leakiness of the mRNA vaccines based on the single spike antigen. I more worry about the Mu variant that appears to be completely resistant to monoclonal antibodies. Time to develop new vaccines using multiple viral proteins or develop an "attenuated" virus mimicking natural immunity.

    7. On 2021-09-09 04:36:47, user Salvatore Chirumbolo wrote:

      Dear Colleagues, my congratulations for this valuable and excellent paper. With a Colleague of mine, Sergio Pandolfi (Rome) we too addressed the topic of natural immunized subjects against SARS-CoV2 (see Pandolfi S, Chirumbolo S. On reaching herd immunity during COVID-19 <br /> pandemic and further issues. J Med Virol. 2021 Sep 7. doi: <br /> 10.1002/jmv.27322. Epub ahead of print.) but we highlighted also the controversial issue of the immunization route, i.e. mucosal versus intramuscular, I mean sIgA-B cell mediated respect to a DC-mediated immunity towards the IgGs production, with obviously different B-cell memory. Do you think that this is one of the major differences between immunized vs vaccinated people? And if serum anti-RBD IgGs are the only clue for evaluating immunized people, why a "certain" discriminating attitude exists towards natural immunized subjects respect to vaccinated ones?

      Many thanks in advance<br /> Regards<br /> SC

    8. On 2021-09-10 07:18:00, user Jim Ayers wrote:

      I hate to ask a dumb question but did the study include people with long haul covid and people who died? Quoting the study '46,035 persons in each of the groups.' The fraction of a percent who died or the few percent who have long covid who may feel too sick to participate in a study could invert the study if not accounted for since they have been weeded out of the cohort and need to be adjusted for. This study could be 100 percent wrong if the up to 20% who have long haul covid aren't participating.

    9. On 2021-11-18 23:35:19, user Emily wrote:

      Does anyone know if this study has been published yet? Don't they usually prioritize data on current health threats to get them through peer review a little faster?

    1. On 2021-08-26 10:17:10, user Ollie wrote:

      This might be an interesting approach. However, something is worrying me:<br /> 1/ The first equation in this paper "r*dr/dt = ..." is not derived, just presented as a citation from a book consisting of 304 pages. A book that is not readily accessible lest one borrows or buys it. The reader thus cannot understand the validity of this equation. The number of open and close brackets is not equal, which implies that the citation is incorrect. Further, it is a pity that parameter e_s(T_a) in the equation is explained in a slightly sloppy manner by omitting the subscript a for T in the text.<br /> 2/ The second equation is stated by the authors, rather than derived from hypotheses. A derivation seems relevant here, as the intuition of the reader (at least mine) tells him or her that the relationship between evaporated volume and surface area reduction of spherical drops is only linear for evaporation that causes very little radius decrease, or in other words: only for evaporation (dV) where dR<<r, where="" dv="the" evaporated="" volume,="" r="the" initial="" radius="" of="" a="" droplet,="" and="" dr="" the="" change="" of="" r.="" if="" this="" intuition="" is="" correct,="" it="" should="" be="" evaluated="" why="" the="" indicator="" air="" drying="" capacity="" is="" indeed="" relevant,="" as="" it="" is="" likely="" that="" in="" a="" given="" timeframe="" for="" some="" drops="" who="" evaporate="" only="" slightly="" dr<<r="" indeed,="" but="" for="" other="" droplets="" dr="R" (complete="" evaporation).="">

    1. On 2021-08-26 18:46:13, user vicweast wrote:

      If a vaccinated person contracts covid-19 (breakthrough infection) and their symptoms do not include symptoms like coughing/sneezing... are they as contagious as a person who has coughing/sneezing symptoms? This is what I am not reading anywhere, and yet it seems to be exactly the point.

    2. On 2021-09-05 11:40:00, user OverSpun wrote:

      The term "infected" appears to translate in this article to the presence of virus (detectable SARS-Cov2 at Ct<25) rather than the presence of the disease (i.e clinical manisfestions of COVID-19). Chronic carriers of other pathogens such as Staphylococcus are sometimes referred to as "not infected".

      Beyond the cellular pathophysiology of bacteria vs virus, this is more than linguistic trivia because it challeges the assumption that SAR-Cov-2 asymptomatic carriers are in a transient subclinical state rather than chronic carriers. If such chronic carriers exist, are they more likely to have been vaccinated or obtained their partial immunity from a live virus infection, i.e. an acute case of COVID-19?

      The following quote from this article highlights the importance of determining if mRNA vaccines have the potential to create chronic carriers: <br /> "Notably, 68% of individuals infected despite vaccination tested positive with Ct <25, including at least 8 who were asymptomatic at the time of testing."

      Clinical management and public health policy require confirmation that all asymptomatic carriers are eventually clear of SARS-Cov-2 and any causative relationship between the vaccine and such carrier state is well undestood.

    1. On 2021-08-29 16:06:28, user Paul Wolf wrote:

      I wonder if the infectiousness of the delta variant could be a blessing in disguise, if it dominates over other, potentially more dangerous variants.

    1. On 2021-08-29 19:50:51, user Yvonne wrote:

      Based on Pfizer (6 month) study, the vulnerable started getting vaccinated on a great scale in February (USA) if you add 6 months that would put that vaccinated population at August, therefore the question would be asked, once the 6 month time frame of those vaccinated within a specific period, be deemed “unvaccinated” come August? With increasing spike cases/hospitalization in August (USA) and the term “unvaccinated” being used, who are within the description of “unvaccinated”, those never getting a vaccine? Or would the term include those who were vaccinated and now have passed the 6 months? I think August would be more of a complete study, if the term “unvaccinated” group is clearly defined. That would require maintaining that data, tracking the expiration of the 6 months when those vaccinated are spiking in cases. While this is helpful, the public should be shown how the spikes are increasing in August for full transparency and even comparison.

      The next spike of population vaccinated in April 2021 (USA) will hit the 6 month cycle in October. Therefore December will show if the spikes in November are from that vaccinated group.

    1. On 2021-09-01 16:46:01, user WGardner wrote:

      I would like to know more about how you controlled for fidelity of mask usage? This should consider whether or not people consistently wore masks, and wore them correctly. Having a mask mandate is a poor marker for whether or not masks work as it does not equate to people actually wearing masks.

    1. On 2021-09-01 20:00:22, user Peter Hanse wrote:

      This should be checked against "Experimental investigation of indoor aerosol dispersion and accumulation in the context of COVID-19: Effects of masks and ventilation" Physics of Fluids 33, 073315 (2021)

    1. On 2021-09-07 15:50:49, user Zach wrote:

      Im confused if it doesn't reduce death significantly statistically and increases serious adverse events by double. Why is this being pushed as effective or safe? This data proves both to be wrong. If your chance of death is unchanged and your hospitalization rate is nearly doubled it literally makes no sense to take this. What am I missing?

    1. On 2020-04-23 02:46:06, user Raspee wrote:

      (1) There appears to be a statistically significant imbalance in the arms with regard to disease severity.

      “However, hydroxychloroquine, with or without azithromycin, was more likely to be prescribed to patients with more severe disease, as assessed by baseline ventilatory status and metabolic and hematologic parameters.”

      The base line pulse oximetry data and baseline line absolute lymphocyte count (Table 2) - indicates a statistical difference at p = 0.024 - the subjects that received hydroxychloroquine had a worse baseline respiratory status - and a worse absolute lymphocyte count p = 0.021.

      This is an inherent bias in the design that has not been adequately addressed. The analysis should compare treatment in subjects with the same disease severity.

      (2) If we look at table 4 - (HC + AZ) - 82% were discharged without ventilation vs. 77% discharged without ventilation both in the HC and non- HC group - Apparently the HC + AZ group did better than the other two groups.

      This is supported by the observation that the adjusted HR for ventilation is 0.43 (0.16 - 1.12) - It was better than the control arm with regard to disease progression and no different than the control for death.

      So in patients that were sicker at baseline, HC + AZ appears to have had a better outcome - than the other two groups - with regard to being discharged without requiring an ICU admission.

      (3) Please provide a better justification to exclude the 17 women Please go back and perform the analysis including the 17 women.

      (4) What were the doses of azithromycin and hydroxychloroquine administered? How are the different doses and dose regimens adjusted in the analysis? Not everyone in the HQ and HQ + AZ groups were dosed in the same fashion. Is there a minimum number of doses that you used to include them in the treatment groups?

      (5) If the control group had less severe illness at presentation, it stands to reason that the mortality rate would be lowest in the control group.

      (6) Was there a sub analysis looking at impact of secondary bacterial pneumonia - which occurs in 5-15% of moderate to severe COVID-19 patients? Were the antibiotics utilized the same over the 3 cohorts or were they different?

      (7) How many patients were on ace inhibitors and/or angiotensin receptor blockers? Were these medications balanced in the 3 arms? What about corticosteroid use in the 3 cohorts? Was corticosteroid use balanced?

      (8) Please go back and re-run the analysis with an additional 14 days of COVID-19 data (using April 25th cut -off) as your sample size will undoubtedly be greater and we would expect that the HQ + AZ group will now have a p value < 0.05. for discharge without ventilation.

      (9) Please include length of stay in your analysis as well

      (10) Please include readmission rates to the hospital in your analysis

    2. On 2020-04-22 22:06:16, user Marie Benz wrote:

      Table 2 discrepancies which favored the non-treatment group, lack of randomization, lack of information on when treatment was begun as well as lack of number of doses completed make this paper unable to be interpreted since it is being heralded by news media as demonstrating that such treatment has now been proven ineffective. Clearly the jury is still out but the authors owe it to the country and the scientific community to point out in the media that this is not enough to conclude that HC or HC + AZ is ineffective or dangerous and that present therapies should not be altered one way or the other based on this report.

    3. On 2020-04-22 00:53:03, user Mike Cee wrote:

      This paper is flawed and should be withdrawn immediately.<br /> 1) This paper is flawed due to the limitation discussed on page 12 about the likelihood that the HCQ group, "However, hydroxychloroquine, with or without azithromycin, was more likely to be prescribed to patients with more severe disease, as assessed by baseline ventilatory status and metabolic and hematologic parameters."<br /> Doctors working the front lines have already noted that patients at SYMPTOMS ONSET + 14 days should not be prescribed HCQ<br /> 2) The important grouping by number of days since SYMPTOM ONSET was left out of this study. Previous studies, while anecdotal, suggest that patients should be prescribed HCQ early because it is believed to prevent the virus from infecting the type 2 lung cell through the ACE2 receptor and thus stops the progression of the disease.<br /> 3) This study did not document the dosage given to the patients. That would have been a helpful inclusion so we could understand if the patients who died were actually poisoned by excessive the treatment.<br /> 4) HCQ prescribed in patients in the first week after SYMPTOMS ONSET to include a zinc supplement which anecdotal evidence suggests a dual function of the combination: The HCQ provides an avenue for the zinc to enter the Type 2 lung cell where it interferes with the virus replication process.

    4. On 2020-04-22 03:32:32, user Cheri Trigg wrote:

      According to MedCram Zinc is what actually helps. Chloroquine is how the zinc gets into the cells. MedCram is a medical journal on Youtube. Update #32 and 34 go into depth on this. It is tragic to use a medication meant as a delivery system and not use it to deliver. Not sure to laugh or bawl. I hope the word gets out to use zinc before the zithromyacin at least.

    5. On 2020-04-21 22:06:22, user Senad Hasanagic wrote:

      Totally flowed retrospective analysis because baseline Clinical characteristics are missing ( not reported) in significant number of patients. So adjustments not possible.

    1. On 2021-12-04 01:07:35, user Balazs wrote:

      What were the Ct values for the positive results? <br /> Are you sure you have not investigated how many people with questionable PCR <br /> positive results ended up with another questionable PCR positive result?<br /> I thought even the WHO early Jan 2021 declared that a "case" have to <br /> have clinical signs, and PCR reports should include Ct values...

    1. On 2021-12-06 16:42:17, user Kristi Leach wrote:

      Sociology student, here, currently writing a paper on issues with the online vax schedulers and the whole idea of using them. I would like to respectfully suggest that you consider focusing on other mechanisms in addition to the city's vax distribution strategy. On March 28, we were only 6 days into phase 1B+ and 50 days into Protect Chicago Plus. Meaning shots had been available to people in Plus neighborhoods much longer than it had been available to most other residents, unless we're suggesting it would have been appropriate to divert from nursing homes, jails, and healthcare workers. That's outside of my expertise, but as a lay person, it's unconvincing. I'm piggybacking on my advisor's findings that we are neglecting the social safety net as COVID mitigation https://www.newsweek.com/wh... For example Not to mention other efforts such as contact tracing and masking. Dr. Parker mentions the lack of hospitals in Black and Brown neighborhoods.

    1. On 2021-12-09 20:54:49, user El Fabsterino wrote:

      Interesting stuff. The sample sizes are very, very small, though. I'd refrain from using p-values here at all. I'm not sure the suggested statistical test (Student's t-test) to test for differences in the means is even applicable here. The original data is clearly not normally distributed and transformed into a 0/1 Bernoulli-variable by defining an arbitrary threshold.

    2. On 2021-12-25 13:45:26, user Blister wrote:

      Interesting that most evidence used to support boosting against omicron is in vitro. If anyone has a study that shows real clinical benefit to boosting with these legacy vaccines please share. This paper is being cited by authors as supporting the use of legacy virus boosters as opposed to a new generation variant booster.

    1. On 2021-12-13 18:28:05, user Kristen wrote:

      I just stumbled across this and I wonder what impact the Mullen's norms have to do with this drop. The Mullen norms are over 20 years old and many of the VR stimuli are very outdated and are not recognizable to children born in recent years. I always prefer to give the Bayley or WPPSI if I can given this issue. It looks like there has been an overall downward trend in Mullen scores in your sample. I know you wouldn't be able to go back and compare as easily, but I wonder how the COVID-19 babies would fare on a measure with more updated norms. Bayley-4 has been freshly updated and would capture those born during the pandemic.

    1. On 2021-12-15 06:28:56, user Robert Clark wrote:

      From the article:

      We used two-month periods as our basic time interval for defining the sub-cohorts, but combined months 12 to18 for the Recovered cohort and omitted months 8 to 10 for the Vaccinated and the hybrid cohorts due to the small number of individuals.

      And also:

      Typically, infection rates among recovered or vaccinated individuals are compared to the infection rate among unvaccinated-not-previously-infected persons. However, due to the high vaccination rate in Israel, the latter cohort is small and unrepresentative of the overall population; furthermore, the MoH database does not include complete information on such individuals. Therefore, we did not include unvaccinated-not-previously-infected individuals in the analysis.

      Frankly, I don’t think the researchers are being completely open about the real reason they aren’t including the unvaccinated/uninfected in their study. The vaccination rate in Israel is about 80%. At a population of 9 million, that would mean 1.8 million unvaccinated. Obviously they could get high statistical significance with that many people.

      I think the real reason is they would find the same as what was seen in the UK and in Sweden, post 6 months the vaccine effectiveness is worse than being unvaccinated to begin with.

      Stunning after this length of time so many countries are refusing to present this data. They’ll collect the data up to 6 months and find the vaccine has waned to having no effectiveness in comparison to the unvaccinated. But except for Sweden and the UK, they refuse to go beyond that point.

      Robert Clark

    1. On 2021-12-23 17:01:02, user Bonnie Taylor-Blake wrote:

      A quick comment, if you will. I'd urge the authors to double-check the contention that '[t]he US Surgeon General stated in 1969 that it was time to ...close the book on infectious diseases, declare the war on pestilence won.” In fact, Brad Spellberg and I looked high and low for corroboration that then-Surgeon General William H. Stewart had indeed made such a claim and we couldn't find it. Instead, we were able to discern how such a misattribution came to be. https://pubmed.ncbi.nlm.nih...

    1. On 2021-12-23 17:28:29, user Heather Madden wrote:

      Thank you for conducting this study, we were not included but as a family with a child with a ANKRD11 missense mutation that had never been seen before in a highly conserved region I was excited to see something written. I've actually had a Dr tell me missense mutations are harmless when they can pathogenic. My daughters half sister is still struggling to get a formal KBG dx, our mutation was proven pathogenic but the Dr used a different lab which did not have that data so its listed as a VUS. Frustrating for missense families, hopefully as more research is done other families won't need to go though long waits for answers.

    1. On 2021-12-28 17:10:57, user madmathemagician wrote:

      Small whole numbers, like "daily new cases and deaths", are not even expected to obey Benfords' distribution.

      Using that to "scientifically" cast doubt about the reliability of EU COVID-19 data, is just fraudulent science.

      The conclusion that higher vaccination rates correlates with larger deviation from Benfords' distribution is probably just because the "daily new cases and deaths" are smaller numbers.

    1. On 2021-12-29 01:27:15, user lowell2 wrote:

      conclusion? "in Omicron cases, these findings highlight the need for massive rollout of vaccinations and booster vaccinations." how so? if the vaccines DO NOT WORK, how is massive rollout going to help? vaccines every 3 months? every month? every day? The findings indicate the vaccines need to be adjusted to function better -- to actually provide IMMUNITY for a significant period of time. Not that one needs to continually use something that is ineffective.

    1. On 2021-12-29 09:29:29, user Željko Serdar wrote:

      Although some of us do not want to listen to good advice, I still give it to you, and it is up to you to decide whether to listen to it and apply it. These are the results of an analysis of more than 42 million people and can be considered reliable, which means that preliminary studies that said that young men have a higher risk of myocarditis after infection than after vaccination were wrong. Vaccination of the elderly and at-risk is justified because it significantly reduces the risk of severe disease. Vaccination of young and healthy people carries a significant risk of myocarditis and any form of forcing young people to get vaccinated is irresponsible.

    1. On 2022-01-02 12:06:34, user Marius wrote:

      Why did you use convalescent donors that had asymptomatic SARS-CoV-2 infection?<br /> I mean, it is great that even asymptomatic convalescents have robust T-cell response…but would it not be interesting to see the immune response of convalescents with moderate Covid? Possibly because this cohort does not even catch the omicron variant that often? At least I assume that unvaccinated people with moderate Covid Display an even better T-cell response compared to the vaccinated….

    1. On 2022-01-04 09:33:05, user Paolo Maccallini wrote:

      Dear all,

      I made an attempt at calculating the proportion of asymptomatic Omicron infections in function of the same parameter in the case of Delta infections, among unvaccinated individuals. I used the data presented in this paper, particularly the data relative to the subjects enrolled in the Ubunto trial (Omicron wave) and those of the population included in the Sisonke sub-study (Delta wave). The result is that, if we assume a prevalence of asymptomatic infections of 17% for the Delta variant, then we have that the prevalence of asymptomatic infections for the Omicron variant is about 60%, in unvaccinated subjects. The details of the calculation I performed can be found here: https://paolomaccallini.com...

    1. On 2022-01-04 13:28:50, user Richard in Bosstown wrote:

      One would like to know if there were people infected by prior versions were getting infected by omicron ("natural immunity").

      It is possible that the omicron is an escape variant selected for among vaccinated persons given that vaccination involves only the single, spike protein of the virus. In contrast, infection with virus exposes subjects to more than 20 viral proteins, any of which can provide immunodominant peptides for an individual's MHC composition to confer T cell immunity. T cell immunity is more critical than antibody for viral resistance, though rarely measured.

      With such a large number of viral protein targets for T cell recognition, the virus is much less likely to mutate during a virus's evolution to escape that "natural" T cell immunity. This is one reason why attenuated virus vaccines that include all viral proteins are generally more effective than protein vaccines. It has been suggested that natural immunity from multiprotein virus exposure is the reason the omicron surge was limited in South Africa where a much larger portion of the population was previously infected versus the US where higher proportions are vaccinated.

      Vaccination is important for individuals not previously infected, that is clear, if only to reduce the severity of infections, as we deal with annually with flu vaccines. However, the lack of presentation of these prior infection data, perhaps omitted from the study design, is a significant omission that could have added to our understanding of the biology and natural history of this virus.

      Richard P Junghans, PhD, MD<br /> IT Bio, LLC<br /> Boston University School of Medicine<br /> rpj@bu.edu

    1. On 2022-01-05 01:05:11, user Paul Wolf wrote:

      I would have liked to see a direct comparison to delta and omicron, which is what everyone has on their minds. This has two familiar mutations, N501Y and E484K, and no others? Where do most of them occur: on the RBD, N Terminal Domain, or furin cleavage site? Do the 46 mutations suggest a jump to another species? Omicron is acting like a completely different virus, and this one found in France is just as mutated.

    2. On 2022-01-11 17:07:10, user Bill wrote:

      Just saw a letter signed and sealed by Ministry of Health in Cameroon that expresses the displeasure of the Ministry for the preprint's methodology and apparent lack of follow up by the authors with Cameroon authorities.

    1. On 2022-01-05 21:36:37, user Bernie Fontaine, Jr., CIH, CSP wrote:

      The anedotal information is limited in population size and demographics. The study appeared to use only healthy individuals without any confounding factors. Secondly, the type of cloth mask was not identified and many people now use either N95 or KN95 respirators or double cloth masks for additional protection. The type of cloth mask was not identified. Many cloth masks may have multiple layers of material and the type of material does make a difference. In short, this study should be reviewed with caution since many variable were not considered.

    1. On 2022-01-08 00:16:51, user Claudia Lupu wrote:

      Excellent article, many of my friends Vax and no Vax got Covid, some fully vaxed had more severe symptoms that non vaxed because of their medical condition. This reinforce exactly what is said in the article.

    1. On 2022-01-12 18:43:00, user Thomas R. O'Brien wrote:

      This appears to be a very well-done study that provides important support for the hypothesis that Omicron is inherently less pathogenic than the Delta variant. I don’t understand the previous comments re: lack of information on age and immunization status. The paper clearly addresses those issues.

      Methods:<br /> · <br /> ‘Exposures of interest included demographic characteristics of patients (age…’<br /> · <br /> ‘We additionally recorded patients’ history of a positive SARS-CoV-2 test result of any type or COVID-19 diagnosis =90 days prior to their first positive RT-PCR test during the study period, as well as the dates of receipt of any COVID-19 vaccine doses (BNT162b2 [Pfizer/BioNTech], mRNA-1973 [Moderna/National Institutes of Health], or Ad.26.COV2.S [Janssen]).’<br /> · <br /> ‘We used Cox proportional hazards models to estimate the adjusted hazard ratio (aHR) for each endpoint associated with SGTF, adjusting for all demographic and clinical covariates listed above.’

      Results:

      ‘Adjusted hazard ratios for hospital admission and symptomatic hospital admission associated with Omicron variant infection, relative to Delta variant infection, were 0.48 (95% confidence interval: 0.36-0.64) and 0.47 (0.35-0.62), respectively (Table 1; Table S3).’

      The authors did not adjust the analyses of more severe outcomes (ICU admission, ventilation, death) for age and vaccination status, but that was because too few patients who were infected with Omicron had such outcomes (despite Omicron being ~3 times more common than Delta in the study population).

      To avoid this confusion, the authors might mention in the abstract that they adjusted the hospitalization analysis for age and vaccination status.

    2. On 2022-01-14 06:53:34, user Andy Bloch wrote:

      The hazard ratios for ICU admission and mortality were unadjusted. This is not clear from the abstract, but the full text explains: "The observed number of patients meeting each of these endpoints was inadequate for multivariate analyses due to the absence of counts within multiple covariate strata." Considering that non-SGTF (Delta) were more likely to be 60 or older, nearly twice as likely to be unvaccinated (49.7% v. 26.6%) and about 1/3 as likely to have had 3 shots boosted (4.6% v. 13.4%) there should have been some adjustment or stratification made, perhaps using rates from other studies. The CDC is citing this study as showing a "91% reduction in risk of mortality" and that is very misleading since the figure is unadjusted.

    1. On 2022-01-13 18:25:17, user Mackenzie Lee wrote:

      I think it's somewhat difficult to make solid claims re incident rates, etc, due to self-reported/-selected data collection via a Facebook site dedicated to survivors of COVID. The rapid data turnaround is nice of course, but a follow up with random sampling will be needed to substantiate claims.

    1. On 2020-05-25 19:37:34, user Laurent Roux wrote:

      Maybe focusing on Ab to the RDB epitope is too restricitve. A network of antibodies is likely to neutralize infectivity as well.

    1. On 2020-04-22 20:36:24, user Konstantin Momot wrote:

      The crucial issue here is sample selection. The participants essentially self-selected, but that is a potential source of huge bias. As a hypothetical scenario, if the people who chose to participate were predominantly people who’d had a cold and were curious to find out if it was COVID, then the cohort would a priori be hugely overweight with people who had a higher-than-average likelihood of COVID exposure. That would not be a good sample of the general population in that it would not represent the true percentage of CoV-exposed and CoV-naive people in the population as a whole. That would mean that the 2-4% figure is completely meaningless. Given how crucial this number is to any epidemiological modelling, I think it's important to remember that this is just one study with no guarantee of flawless methodology, and avoid making far-fetched conclusions based on limited evidence.

    2. On 2020-04-23 18:41:00, user Young-In Kim wrote:

      It’s not going to get peer reviewed well just like the studies on hydrochloroquine. That’s not a sample to apply to the whole population when we have data from all over the world. It’s picking and choosing data. I might well take my sample from worst hit parts of NYC n Italy to prove a point.

    3. On 2020-04-23 18:52:50, user Deplorable Kev wrote:

      So they tested for IgG antibodies in the population and found a high number of persons already had been exposed and likely immune. Interesting, but that depends on if their assay cross reacts with US Coronavirus and if so, then this is not correct. If their test is specific for COVID-19 antibodies then he is correct. PS: I believe it was here much earlier and we saw cases as early as mid November. Most of this early antibody testing has not been evaluated for cross reactivity yet.

    4. On 2020-05-16 23:36:43, user PANAGIOTIS AMPATZIS wrote:

      There is a hit piece from Buzz Feed news (I know lol) that says the Jet Blue founder (and anti lockdown proponent) gave 5000$ to the study to influence it. They make a huge deal out of it, so please release the full amount of the funding as an answer to show them that 5000 dollars is a small amount to make the study unreliable.

      Keep up the good work

    5. On 2020-04-20 20:43:45, user Kenneth Melendez wrote:

      So what these people are saying is that if you did the study LAST year, that you would get .5% false positives, about 10K people if you tested all of Santa Clara. You should declare them to be asymptomatic cases, and say that there are 9.5 times the number of asymptomatic cases as confirmed cases in Santa Clara. Time to call the media.

      And if you want to believe their results, then you surely want to believe the that the false positive rate (1-specificity) was accurately determined by 2 false positives out of 371 tests of old (assured non-infected) samples.

    6. On 2020-04-20 22:26:21, user Andy wrote:

      Preliminary results of USC-LA County are out. Unfortunately, that study also suffers from self-selection bias. We need to know how many people were contacted initially; should we count those who declined to participate as presumed negatives?

      In the Santa Clara study, people were contacted initially through Facebook. We need to know how many people saw the ad, and should we also count those who declined to participate as presumed negatives? (Surely the authors know how many people saw the Facebook ad.)

      Edit to add: Wait a minute, I just saw comment by Anon who participated with a link but without Facebook. So the study is even more flawed.

    7. On 2020-04-21 20:29:57, user rerutled wrote:

      The Santa Clara results calibrated the test using known gold-standard "negative for covid19" blood. They had 401 "pre-covid19" samples; of those, 2 resulted in "positives". Thus, the test produces a "false positive" rate of 2 out of 401.

      They gave the test to 3330 people, and 50 of those tests returned positive. How many of them can be produced by the observed "false positives"?

      On average, for a sample of size 3300, one expects approximately 3330*(2/401) = 16.6 false positive results. Is that statistically significantly different from the 50 detected? It depends on what the uncertainty in that "16.6" is. Playing fast and dirty, take the Gaussian uncertainty in that false positive rate to be fractionally, 1/sqrt(2) = 0.70. So the false positive rate is 16.6+/-11.7. Which means, the 50 detected is only (50-16.6)/11.7 = 2.9 sigma above the false positive rate.

      Nobody in science should claim a detection which is only 2.9 sigma different from the false positive number. This result is not significantly different from NO DETECTION of positive anti-bodies.

      It's even worse than that, though, because I used the Gaussian uncertainty; and as you know - since you have taken an undergraduate class in statistics and understand the binomial, poisson, and Gaussian distributions - the Gaussian approximation isn't accurate for this situation, and one should use the Poisson distribution, or better yet the binomial distribution; but, again, because of that undergraduate class, you remember that both the Poisson and binomial distributions have a broad wingy excess in the higher direction above the mode; which means, the significance of even the detection of ANY positive covid19 antibodies in the Santa Clara County results is even LESS significant than 2.9 sigma.

      The authors of this result should answer this criticism, stating what they believe to the the probability of detecting 50 positive results among 3330 tested, given their measured false positive rate.

    8. On 2020-04-22 16:02:39, user Texas Longhorns wrote:

      The research paper does not indicate how many of those that participated had already been tested for Covid and what those test results were.

      If they over sampled people that had already tested positive and recovered of course you will get a higher rate of positive antibodies. That would not be indicative of the general population.

      There is also the problem of false positives because the test can trigger for the common cold that is also a coronavirus.

      I don't think this research passes muster as any reliable indication of antibodies in the general population and should absolutely not be used as a basis to reopen businesses and large public gatherings.

      Having antibodies to one strain of the virus may not give you any immunity to the more than 8 strains of Covid we know are out there.

      Even if the test results are accurate at 2% that is nothing and you need at least 60% solid immunity to consider any large population to have herd immunity protection.

    9. On 2020-04-17 22:20:15, user Teddy Weverka wrote:

      To take this test, people had to violate the ORDER OF THE HEALTH OFFICER OF THE COUNTY OF SANTA CLARA DIRECTING ALL INDIVIDUALS LIVING IN THE COUNTY TO CONTINUE SHELTERING AT THEIR PLACE OF RESIDENCE EXCEPT FOR ESSENTIAL NEEDS. What do you bet that people willing to violate that order have a higher prevalence of covid-19 antibodies? The paper ought to mention this source of bias.

    10. On 2020-04-18 14:47:42, user Chris S wrote:

      Were participants asked if they had been tested? If so, could the rate of actual COVID-19 testing among participants relative to the general population (at the time of survey) be used to assess ascertainment bias?

    11. On 2020-04-19 03:40:52, user disqus_B1vk25qxNZ wrote:

      Possibly other coronaviruses that the population has been exposed to confer cross immunity to SARS-CoV-2. If so, the benign virus could be a vaccine analogous to cowpox and smallpox.

    12. On 2020-05-24 07:09:29, user Animesh Ray wrote:

      Once again, this study uses previous studies' data (by the kit manufacturers) to estimate CL of their specificity estimate. This is flawed--a classic Type I error. In other words, the authors use "meta-analysis" of other studies data to establish the bounds of their own data interpretation. Meta-analysis requires very careful calibration of admissible data using several well-known metrics. None of that has been done here. These results will remain flawed until the authors use their kits under their own experimental conditions to determine the true negative frequencies using sufficient (at least 200) pre-COVID19 samples. Even then I will be worried because these samplings will be conducted non-contemporaneously with their main study. In other words, these studies have little hope of being salvaged because of their fundamentally flawed study design.

    1. On 2020-04-23 08:03:58, user excivilservant wrote:

      I'm amazed nobody has commented on this, given the mass of media coverage about tracing and testing. Isn't a fairly obvious point that the model on social interaction is based on a pre-virus, no restrictions world. That's not the world we will be in for the coming months at least. Many/most/all hospitality businesses will be closed and these are the places where one would imagine a substantial number of social interactions occur, especially ones that widen the social circle beyond family and close friends. The survey data underlying the model will have the details. Even leaving aside formal restrictions, a lot of people will be acting in a much more conservative way about going out and about. Numbers travelling by train, tube and bus are likely to be greatly reduced. Again, the model could be adjusted, or a sensitivity test done. So the efficacy of tracing should be much greater than the article implies - or, put another way, the effort required to trace people should be a lot less, as there will be a (much) lower average number of contacts. No doubt the authors are actually on to this, but I thought I should point it out anyway.

    1. On 2020-04-24 01:46:41, user Mike wrote:

      Note: the original version has been previously discussed at length. medrxiv is redirecting that paper to v2 (this paper), making those comments are no longer available. This is a link to those comments: https://disqus.com/home/dis...

      Below is my original comment with updates for this version of the paper


      This was certainly an interesting paper. They’ve done a lot of work and the findings are notable. IMHO it warrants as much attention as the original pro-HCQ study via Dr. Raoult (~3/15). While it is entertaining, I will add that it is not conclusive, nor without fault. A double-blind study is still required.

      It is important to note that this is version 2 of a previously released paper and it is much the same, with no major differences in the conclusions reached compared to v1. Therefore, my previous comment still holds true. Below I’ve included them followed by a new list of observations.

      Observations/Questions (updated)

      1. "hydroxychloroquine, with or without azithromycin, was more likely to be prescribed to patients with more severe disease” (p.12)<br /> 2. “as expected, increased mortality was observed in patients treated with hydroxychloroquine, both with and without azithromycin” (p.12) — I assume it’s expected because the patients given drugs were in a more severe state (and more likely die regardless of treatment)<br /> 3. "we cannot rule out the possibility of selection bias or residual confounding” (p.13)<br /> 4. demographic: 100% male, 66% black, median age ~70 (59 youngest); (Table 2, p.17)<br /> 5. uses PSM, which despite a common practice, could be considered controversial (https://gking.harvard.edu/f... "https://gking.harvard.edu/files/gking/files/psnot.pdf)")<br /> 6. Unless I missed it, I didn't see any specifics about how the treatments were administered.<br /> - How long before death were patients treated?<br /> - What was the quantity/frequency/duration of the treatments?<br /> - Were the treatments consistent between hospitals?<br /> 7. The rate of ventilation was less in HC+AZ (half of the HC and no-HC rates). Why was that? What does that suggest?<br /> 8. Although they were statistically insignificant, what was the result of the 17 women not included in the study?<br /> 9. Why does the paper seem to address political points? It seems like the Abstract is editorialized, which I'm not used to. The result seems to address topical issues of the times, having awareness of other similar studies being conducted, rather than a standalone independent study of its own. I interpret this as potential for some analysis/deciphering bias. I don't mind in the Discussion sentence as it's normal, I'm just not as accustomed to seeing it in the Abstract.

      New List of Observations

      1. There seems to be some bias in the number of healthy people with no-HC treatment, but left in the study. Those people are going to be unlikely to die to begin with. This is not a comparison of apples to apples.??

      To clarify:<br /> ?- Dramatic difference in percentage of people people that had fever temperatures (38.1-39.0ºC / 100.58-102.2ºF); HC:11.3%, HC+AZ:11.5%, no-HC:7.6%. There’s ~4% difference between treated and untreated fever temps (more likely to die) in favor of untreated cohort. ?<br /> - Compare that with the percentage of people that had normal temperatures (35-37ºC/95-98.6ºF);HC:56.7%, HC+AZ:52.2%, no-HC:61.4%. There’s a 5-9% difference between treated and untreated normal temps (likely to not die) in favor of untreated cohort.

      ??So in this study, there was a larger proportion of people that did not have fevers, suggesting the data may be padded. In absolute numbers it's approx. a 40 person swing, which is a fairly large percentage in such a small study/survey. Similar observations are for systolic blood pressure and breaths per minute. There appears to be more healthy people? again.

      2. Creatinine is created when muscles breakdown creatine. It’s a waste product removed by the kidneys. Levels are elevated when the kidneys begin to fail. Notice, there’s a much larger presence in the HC-groups, which suggests there was a larger percentage of these patients experiencing kidney failure.?

      3. There are an awful lot of missing data in solely the no-HC group for statistically significant criteria. For instance Erythrocytes (red blood cells that transport oxygen), there 11.4% (!) missing in no-HC patients, yet that category has a P-value of 0.001 (<0.05 is statistically significant).??

      Hermatocrit is the same way (missing 11.4% for no-HC). It’s also related to red blood cells, it is the ratio of red blood cells to the blood volume. Same missing amount for Leukocytes (white blood cells) test. And also Lymphocytes —white blood cells in lymphatic system, which transports fatty acids from the digestive system and white blood cells from the lymph nodes into the bones— not only has a lot of missing data, but the disproportional low count (<800 per mm^3) may warrant further investigation.

      4. Even looking at the statistically significant Cerebrovascular disease, there are a much larger percentage of HC-only patients per its cohort.?

      5. Table 4 describes a greater percentage of people using HC+AZ being discharged (recovering) w/o ventilation; 5% more than no-HC patients. Keep in mind that 30% of the no-HC patients were given AZ.

    2. On 2020-04-28 00:30:59, user Mark Reeder wrote:

      I am advocating that the authors, in the interest of public health, fill in the blanks of the following statement:<br /> "It was found that ___ of the 7 patients reclassified from the 'No HC' to the HC group (after ventilation began) died. Likewise, ___ of the 12 patients reclassified form the 'No HC group' into the 'HC+AZ' group died."<br /> Based on a comparison of Tables 4 and 3, the first blank must be either 6 or 7 whereas the second blank must be between 7 and 12, inclusive.

      If the groups are compared based on whether they were given the drug(s) PRIOR to ventilation or prior to discharge, the HC+AZ is better by either a 20% margin over the 'No HC' group or by a FACTOR of 6. <br /> To wit, let's assume that all 12 of reclassified as HC+AZ died. That would mean that only 2 in the original HC+AZ group died. Since we have no idea when the HC+AZ drug was administered to those who died without ventilation, a fair comparison would show that HC+AZ, one might justifiably count only the 2/90 (2.2%) in that group (excluding deaths w/o vent.) as having had HC+AZ early enough in treatment. It would also mean that 3+(6 or 7) + 12 out of 162 (13.6%, also excluding deaths prior to ventilation) eventually died. <br /> This would be a huge difference with HC+AZ coming out as a terrific alternative (factor of 6 better) if given early enough. By the same logic (pre-vent treatment only, excluding non-vent deaths), the worst case for HC+AZ would still mean a 20% IMPROVEMENT over the control group! <br /> But we cannot know unless the authors (or others) are ethical and transparent enough complete the sentence above. Even if they disagree with the foregoing analysis, what is the downside in providing those numbers?<br /> I understand the difficulty of dealing with imperfect data. But for that very reason, good science demands that all information be placed on the table.

    1. On 2020-04-25 13:53:09, user Rosemary TATE wrote:

      Since this is an observational study, and you answered yes to "(I have) uploaded the relevant EQUATOR Network research reporting checklist(s), could you please explain why you have not uploaded the STROBE guidelines? So many people are doing the same and I'm wondering why. I'm trying to think of ways of improving the process of sorting the wheat from the chaff in this explosion of preprints. While this looks like wheat, it would be really helpful if you could<br /> 1. State the type of study in the title (some journals require this and it is very helpful)<br /> 2. Upload the checklist.<br /> Many thanks.

    1. On 2020-04-26 15:17:51, user peopletrees wrote:

      Can you post the r script that you wrote in order to process and analyze this data? I'm particularly interested in the rstanarm model specification.

    1. On 2020-04-27 03:28:31, user Sam Leumas wrote:

      How was the data even collected? Veracity of this information?How can you tell if the contracted virus is from within house, transport area,outdoors,malls etc? Common sense tells stagnation of air is a probability but still,seems like there is a major flaw in the data collection process.

    2. On 2020-04-10 16:34:41, user Brian wrote:

      this looks at "3 or more case outbreaks" if outdoor spread is happening it's probably on a smaller scale, as people are out less.

    1. On 2020-04-28 13:51:34, user alasdair hay wrote:

      I am not convinced this paper provides any evidence on the safety of HFNC in respiratory infections. My understanding is the study shows similar number of respiratory particles were measured in all conditions from breathing air to any 02 device or coughing.

      I would be more reassured if the articlce demonstrated an increased number of respiratory aerosol particles with some airway devices or proccedures but not HFNC. In fact there is no control demonstrating the SMPS aerosol measurement probe detects respiratory droplets. The only particles the probe appeared to detect were produced by a candle. If the probe is picking up a background number of particles in the air and those produced by a candle it is not picking up anything of medical interest

    1. On 2020-05-01 09:48:59, user Kasper Kepp wrote:

      This paper on the state-of-the-art Danish blood-donor data finds a IFR = 0.08% for people between 0-69 years of age. The study is very important because the sampling bias from case fatality ratios (the iceberg effect of knowing almost all deaths but only the most symptomatic cases, i.e. missing the dark number) is largely removed.

      By interpolation, the Danish population now has approximately 1.6% infection, corresponding to 100,000 people out of 6 million. The dark number stands at 12-fold the known cases (7-18).

      Some minor sampling biases remain (people who are blood donors need to be healthy and may be socioeconomically skewed) but considering the wide blood donor representativeness in Denmark, I think all Danish researchers will agree that sampling bias must be small.

      The IFR is also fully in line with the most representative data we have from Iceland (14% of population tested, 48000 tests), where the sampling bias is essentially eliminated, which stands at approximately 0.56% (10 deaths / 1799 cases as of May 1) and includes all the high-risk individuals >70 years. https://www.worldometers.in...

      Compared to the Santa Clara study, which caried potential major sampling bias, this issue seems to be now largely removed. Consensus in Denmark is now emerging that the overall whole-population crude mortality of covid-19 is of the order of 0.25-0.6%, in excellent agreement with the Iceland data.

      These two countries have not have their health care systems strained, making them the relevant data also for this reason for pinpointing the "real" mortality of covid-19 absent overmortality by capacity exhaustion as seen in some other countries.

      Obviously, the fact that the IFR is 0.08% for the 0-69 year old has enormous implications for political decision making in Scandinavia, as it evidences that most of the population can build immunity at much reduced mortality than previously assumed.

    1. On 2020-05-02 04:43:05, user Ilya Zakharevich wrote:

      Thanks! This is the first potentially meaningful text on such stats I have seen so far. However, there are some (possibly significant) apparent problems too… (Below, I only address points about false-positivity rates.)

      The presentation: for your Table 2 (and other related presentations), would not it be more clear if you add a row “Pre-Covid” before the row “1–5 day”?

      “Samples from UCSF and ZSFG were assigned a random well position in one of four 96-well plates.”

      I suspect this is not clear enough. The crucial question for estimating the seropositive population is the “degree of double-blindness” of mixing the 108 pre-CoVid samples among the other samples. Can you be more explicit about this? (Separately for ELISA and the rest, if possible.)

      “Binomial exact 95% confidence intervals were calculated for all estimates.”

      With 14 different schemes tested, this is a very questionable choice. Basically, for laymen in statistics (which, in my experience, probably covers >99.9% of potential readers), only the 99.74%-confidence intervals can be useful. (Here 5%/14?0.357%.)

      The people who understand the pitfalls of using 95%-confidence with 14 schemes could also be interested in 95%-confidence numbers — but these numbers would just create an unneeded confusion among the overwhelming majority. (As this XKCD shows.)

      “Four assays … maintaining >95% specificity.”

      Sigh… Do you understand the expected number of outliers with 14 groups? Especially when you, essentially, say “<=5 false-positive samples per group”?!

      “We based minimum sample size calculations on expected binomial exact 95% confidence limits.”

      I think 108 samples is ruefully small for any reliable conclusion. (As your numbers, and 99.64%-confidence intervals show.) I do not see any way than to start with >=1,000 samples (with one method), or >=3,000 (with 14 methods, as you use).

    1. On 2020-04-06 17:15:59, user Matheus Macedo-Lima wrote:

      It seems to me some of the countries used in this analysis are still under-testing (Brazil, USA etc). Would a more suitable dependent variable be deaths/confirmed COVID cases?

    1. On 2025-08-21 08:10:06, user Taukim Xu wrote:

      Comments from the authors-01:

      Since the preprint was made available, we have been honored to receive numerous publication invitations from various journals. We would like to express our sincere gratitude for your interest and recognition of our work.<br /> However, we feel it is important to clarify that this manuscript is currently under consideration at another journal. Therefore, we are unable to submit it to your publication at this time.

      Dr. Taukim Xu

    1. On 2025-10-18 04:34:17, user CDSL JHSPH wrote:

      Hello!

      Thank you for sharing this preprint. I found it very interesting, and I believe this is an excellent contribution to improving statistical approaches in antibiotic trial designs. I believe that the comparisons between MCP-Mod and Fractional Polynomials (FP) models provided a very strong case for adopting model-based frameworks to more accurately identify duration-response relationships and estimate the Minimum Effective Duration (MED). I also believe that the demonstration of these methods outperforming traditional pairwise comparisons, particularly in smaller sample sizes, helps highlight the potential to make trials more efficient without compromising reliability and ultimately the patient.

      I also appreciate how you clearly discussed the limitations of traditional methods and how there is a need to move towards more model-driven designs. The emphasis on the simulation as a proof-of-concept framework was well-structured and persuasive. I think it would be very interesting to see how these approaches would perform under real clinical data. This is where patient adherence, comorbidities, and variable response rates could truly test the robustness of these models.

      I think that in future works, expanding on various factors like dose-spacing, variability in treatment adherence, or non-monotonic response patterns that affect model performance could help further strengthen the clinical applicability of your framework. Overall, I found that this paper sets a solid foundation for refining antibiotic trial methodologies, and also bridges the gap between statistical modeling and real-world trial optimization.

    1. On 2025-11-17 18:55:33, user David Stapells wrote:

      Interesting.

      Small comment:<br /> Rather than head size explaining larger amplitudes in females, Don and colleagues suggest cochlear travelling wave differences:

      Don, M., Ponton, C.W., Eggermont, J.J., Masuda, A. (1994). Auditory brainstem response (ABR) peak amplitude variability reflects individual differences in cochlear response times. J Acoust Soc Am, 96, 3476–3491.<br /> Don, M., Ponton, C.W., Eggermont, J.J., Masuda, A. (1993). Gender differences in cochlear response time: An explanation for gender amplitude differences in the unmasked auditory brain-stem response. J. Acoust. Soc. Am., 94, 2135–2148.

    1. On 2025-11-30 14:34:14, user Jeff Lubell wrote:

      This article has now been published in a peer-reviewed journal. Please use that version. Thank you. Here is the citation: Torok RA, Lubell J, Rudy RM, Eccles J, Quadt L. Variant connective tissue as a risk factor for long COVID: a case-control study of data from a retrospective online survey of adults in the USA and UK. BMJ Public Health. 2025 Sep 17;3(2):e002949. doi: 10.1136/bmjph-2025-002949. PMID: 41001233; PMCID: PMC12458677.

    1. On 2020-04-20 18:14:27, user Wladimir J. Alonso wrote:

      Not being temperature-dependent does not imply that it might not be seasonal. For instance, in equatorial regions seasonality of influenza is driven by the rainy and dry seasons, not temperature (see references below).

      Tamerius J, Nelson M, Zhou S, Viboud C, Miller M, Alonso WJ. Global Influenza Seasonality: Reconciling Patterns Across Temperate and Tropical Regions. Environ Health Perspect. 2010

      Alonso WJ, Viboud C, Simonsen L, Hirano EW, Daufenbach LZ, Miller MA. Seasonality of influenza in Brazil: a traveling wave from the Amazon to the subtropics. Am J Epidemiol. 2007;165(12):1434–42.

      Alonso WJ, Yu C, Viboud C, Richard SA, Schuck-Paim C, Simonsen L, et al. A global map of hemispheric influenza vaccine recommendations based on local patterns of viral circulation. Sci Rep. 2015;5:17214.

    1. On 2020-04-01 18:46:10, user Miklos Kertesz wrote:

      Does this modeling shed any light on the approximate number of those who have been infected, and are now presumably immune?<br /> Your curves fall off quite fast. Is this realistic?

    2. On 2020-04-04 21:18:14, user Lucy Branum wrote:

      Does this data reflect the lack of PPE for healthcare providers, and other essential workers? I appreciate that is a difficult factor to quantify, but if your healthcare providers are becoming infected and therefore spreading the virus to others, and becoming ill themselves, I would have to assume that would have a massive impact on the trajectory of the curve.

    3. On 2020-04-05 15:41:10, user Art Mills wrote:

      This was supposed to be updated yesterday and had a notification it was to be, then after the daily press conference, suddenly was not updated and the notification it was to be updated was removed. Why?

      The ICU and Vent numbers in the model to support the 93K deaths by Aug. 4 are off by a factor of 10 (meaning they have modeled far too high). The update yesterday should have brought those lines down. Why would we not update to reflect a line closer to the actual numbers?

    4. On 2020-04-06 07:06:24, user Tom T Walker wrote:

      Many of the State projections radically changed when updated 4/5. For example Colorado's peak date went from 4/17 to 4/4, hence an ICU deficit of 762 to a surplus of 421. Other states had similar swings in the opposite direction. Wondering if there might have been some data entry issue. Is Colorado really suddenly 1-day past it's peak date?

    5. On 2020-04-07 07:38:54, user dreich wrote:

      Every place in the US is significantly under-testing.<br /> This means that the infection rate is underestimated.<br /> And because there aren't enough tests almost no post mortem tests on performed for any deaths.<br /> This has the "advantage" of under-counting the number of fatalities which makes the organized incompetence (intentional or otherwise) seem less bad. The bigger the number the worse it looks for the guilty politicians.

    6. On 2020-04-10 13:51:26, user steve rubin wrote:

      Does anyone know peak hospitalization during the 2017-2018 flu season? From the cdc summary there were 808,000 total hospitalizations with 61,000 deaths and I remember the flu season was pretty long. I wonder how the curves for new cases, deaths, hospitalizations and icu use looked. I remember stories that the hospitals were crowded but I don't remember stories about people dying because there weren't beds or ventilators.

      I know that it's unpopular to compare coronavirus to the flu. Underestimating a threat is dangerous and could (and maybe did) lead to delay in ramping up testing and beds and ventilators and other necessary medical resources.

      When people were predicting 2,000,000 deaths in the US and then 200,000 deaths I could understand the fears. But now they're predicting 60,000 deaths and it may end up half of that, so I think it's reasonable to make the comparisons.

      Comparing situations to past situations is usually our best way to understand how to react. In terms of how contagious and how lethal this epidemic/pandemic is, it now seems that it and the flu are similarly contagious and that covid is much less lethal. The big difference is that we have a pretty big number of people with significant immunity to the flu while it's likely there was little immunity in our population to covid-19. If we end up with 200,000,000 people becoming infected but with only 60,000 deaths, then covid was a fifth as lethal as the flu for 2017-2018 with 3 in 10,000 infections dying vs 14 in 10,000 infections dying from the flu.

      However the comparison to the flu can lead to some counter-arguments. For example, the cdc uses a multiplier of ~80 for estimated current flu infections vs confirmed flu infections. Applying that to covid-19 means that we have ~500,000 confirmed cases so we would have had 40,000,000 total infections leaving another 160,000,000 to go assuming 60% of the population for herd immunity. Projecting deaths would mean 17,000 + 68,000 for a total of 85,000. We'll soon know what that multiplier is for covid-19 because there are a number of antibody surveys going on in the US and internationally. You can bet that the same thing will happen for the flu next year and we'll have a more accurate estimate of infections and lethality for the flu rather than the current guesstimates.

      A big question is how social distancing will have affected the final number of infections and deaths. It seems so logical that social distancing will curb infections and deaths, but many suggest that it may end up only prolonging the length of the pandemic while not making a significant difference in total final infections and total final deaths. The antibody tests may give us the answer to that as well.

    1. On 2020-04-22 08:47:13, user Katri Jalava wrote:

      This is a really interesting study from the environmental epidemiology perspective. However, the authors fail to present some of the key findings they surely have. The clinical presentation of their cases is "not interesting", as they are anyway, typical Covid-19 cases, that could be shortened. You could also try to get some of your food control experts as authors, they might be able to give you good clues about the potential spread modes with employees and hygienic situation in the supermarket. Also if one of the epidemiologists from China-CDC would help you to clear the epidemiological messages, that would be useful. Your current conclusions are "not really interesting".

      If possible, it would be very important to know how many customers were screened, at which time points? Also the time line for the employees when they were at work and when did their symptoms begin, ie. how long they were working while infectious? What type of supermarket this was, how much selling, customer profile? Also the description of the supermarket employees job duties, were they at the cashier, providing meat or other over-the-counter food etc.? How many children were exposed either as case household contacts or in the supermarket?

      You should try to follow up the asymptomatics. How many of them went on to develop symptoms, even mild? If not, please state the follow up and initial examination date. Comparison of clinical symptoms between groups A and B is not really interesting (you could just state in your text that the groups did not differ, data not shown), but comparison of symptomatic vs. truly asymptomatic might be. However, there is plenty of studies on that and unless you detect something extraordinary, that is of interest. Were your cases really 48 year old, plus minus 2 years, ie. 46-50 year old. All of them? What were the isolation (cases) and quarantine (close contacts, yet not symptoms) measures, please describe and how this succeeded. Did you consider (or would you do a next study) to do serological survey among all cases, their close contacts and supermarket customers to see if they had SARS-CoV-2 antibodies?

      Thanks for submitting this, this is a really nice and interesting study. Looking forward for the final paper.

    1. On 2020-04-07 16:22:04, user Kevin Niall wrote:

      "All received standard treatment (oxygen therapy, antiviral agents, antibacterial agents, and immunoglobulin, with or without corticosteroids)"

      This could play a huge part in the difference considering teh very small sample size and the difference of 8 patients.

    1. On 2020-04-07 19:20:32, user Andrei Drabovich wrote:

      There was no experimental proof presented that these variants impact ACE2 structure or function in any way.<br /> It is also not possible that ~3% genotype frequency for women and 1.5% for men (1.5% allele frequency for p.Asn720Asp) will result in additional >6% lethality. Were would additional 3-4.5% come from?

    1. On 2020-04-08 00:15:34, user Sinai Immunol Review Project wrote:

      Clinical features and the maternal and neonatal outcomes of pregnant women with coronavirus disease 2019

      Keywords

      Pregnancy, SARS-CoV2, neonatal and maternal Covid-19 outcome

      Key findings

      33 pregnant woman and 28 newborns were included in this retrospective multi-center study, conducted at 5 hospitals in Wuhan and Hubei province, China, between January 1 and February 20, 2020. All women were diagnosed with Covid-19 by qPCR or viral gene sequencing based on the Chinese New Corona Pneumonia Prevention and Control Program, 6th edition, and were further subdivided into four groups based on clinical severity: (1) mild, presence of mild clinical symptoms without radiological abnormalities; (2) moderate, fever or upper respiratory symptoms as well as radiological signs of pneumonia; (3) severe, at least one of the following: shortness of breath/respiratory rate >30/min, resting oxygen saturation SaO2<93%, Horowitz index paO2/FiO2 < 300 mmHg (indicating moderate pulmonary damage); and (4) severe-acute, acute respiratory distress with need for mechanical ventilation; systemic shock; multi-organ failure and transfer to ICU. Maternal admission to ICU, mechanical ventilation or death were defined as primary outcomes; secondary study outcomes comprised clinical Covid-19 severity in both mothers and newborns, including development of ARDS, neonatal ICU admission as well as mortality.

      Maternal characteristics and outcome: 3 out of 33 women were in their second trimester of pregnancy (17, 20 and 26 weeks), and 15/33 (45.5%) had a previous history of underlying chronic health disorders including cardiovascular, cerebrovascular or nervous system disease. Common Covid-19 symptoms at presentation were fever (63.6%), dry cough (39.4%), fatigue (21.2%), and shortness of breath (21.2%). Less common symptoms included diarrhea, post-partum fever, muscle ache, sore throat and chest pain. 4 (12.1%) pregnant women had no apparent symptoms. The majority of cases were classified as mild (39.4%) or moderate (57.6%); however, one woman developed severe Covid-19. 40.6% of women were diagnosed with bilateral pneumonia, 43.8% presented with unilateral pneumonia, and 15.6% showed radiological ground-glass opacity. 87.9% of women required oxygen administration, and one (3%) woman had to be put on non-invasive mechanical ventilation (primary outcome). 81.5% of women had a C-section and only 5% had vaginal deliveries. Obstetrical complications were seen in 22.2% of women, including three cases of preterm rupture of membranes, two cases of hypertensive disorders of pregnancy, and one case of spontaneous preterm labor. Five pregnancies were ongoing at the end of the observation point of this study; one woman decided to have her pregnancy terminated. Neonatal outcome: Out of 28 newborns included in this study, 35.7% were born preterm at <37 weeks of gestation with Apgar scores ranging from 8-10/10 at 1 min and from 9-10/10 after 5 min, indicating normal heart and respiratory rates. 17.9% of newborns were of low birth weight (not specified) and 14.3% showed signs of fetal distress (also not specified). According to the authors of this study, none of the newborns presented with clinical Covid-19 symptoms. However, one newborn, delivered at 34 weeks of gestation, was diagnosed with (apparently Covid-19 unrelated?) ARDS and transferred to NICU (secondary outcome). Of 26 newborns tested for SARS-CoV2, only one was found positive and showed radiological signs of pneumonia, but no clinical symptoms of Covid-19. It remains unclear whether this was the same case as the newborn diagnosed with ARDS. The affected newborn did not require any treatment and was discharged at 16 days post birth. In summary, the primary outcome “mechanical ventilation” in pregnant women was rare (3%), no other primary outcomes were reached. Most Covid-19 cases in pregnant women were described as mild to moderate. Only one of 28 (3.57%) newborns was diagnosed with ARDS (secondary outcome).

      Potential limitations

      Major limitations of this study are its small size and the rudimentary and at times inadequate description of patient specifics. For example, underlying health conditions that might be affecting Covid-19 outcome in pregnant women should have been clearly specified (other than being of be listed (not just <37 weeks). Given that maternal infection status seemed mostly unknown at the time of birth and, more importantly, that the majority of cases in this study were clinically asymptomatic or mild to moderate, it remains unclear whether the C-sections performed were a medical necessity or elective procedures. This is of importance and should have been discussed. With regard to neonatal outcome, it is also not apparent whether the newborn found to be infected with SARS-CoV2 and the case diagnosed with ARDS were the same individual. If this was the case, it would be incorrect to refer to all newborns as asymptomatic. Additionally, it seems somewhat unlikely that a newborn with a near-perfect Apgar score would present with ARDS immediately after birth. Likewise, any individual diagnosed with ARDS would certainly be expected to receive supportive treatment including (invasive) mechanical ventilation. While it is highly relevant that overall clinical outcome in pregnant women diagnosed with Covid-19 seems better than in SARS or MERS (as discussed by the authors), it nevertheless needs to be stressed that more than 37% of newborns in this study were delivered preterm and that the obstetric complication rate of 22% seems higher than non-Covid-19 average.

      Overall relevance for the field

      Observations in this study confirm some of the findings published in a case series by Yu N et al. (Lancet Infect Dis 2020; https://doi.org/10.1016/ S1473-3099(20)30176-6). However, due to the relatively small study size of 33 pregnant women and 28 newborns, this study lacks statistical power and final conclusions on Covid-19 outcomes in pregnant women and newborns cannot be drawn. Yet, the data collected here are important and should be incorporated into larger data sets for more insight. Understanding the clinical course and effects of Covid19 in both pregnant women and newborns is essential, and while there are some recent publications on vertical SARS-CoV2 transmission between mothers and newborns (Dong L et al, JAMA March 26, 2020, doi:10.1001/jama.2020.4621; Zeng H et al, JAMA March 26, 2020, doi:10.1001/jama.2020.4861) as well as on neonatal infection at birth (Zeng L et al, JAMA March 26, 2020, doi:10.1001/jamapediatrics.2020.0878), our knowledge of how these patient subsets are affected is still very limited.

      This review was undertaken as part of a project by students, postdocs and faculty at the Immunology Institute of the Icahn school of medicine, Mount Sinai.

    1. On 2020-04-09 07:07:30, user Daniel Corcos wrote:

      This calculation is made by dividing the number of deaths by the number of days since the FIRST Covid death, and by comparing to the number of deaths per day by driving during the same period. This means that the death rate by day is diluted with the period of low mortality per day from Covid.

      But the highest mortality per day is yet to come. And, already, social distancing has been implemented and has some effect.<br /> This means that there is no evidence to support the hypothesis that there is overreaction to the pandemic.

    2. On 2020-04-17 14:53:53, user Petard Stamo wrote:

      "Absolute risk" is a very good measure to quantify the probability of dying from cancer and heart attack but it doesn't capture the dynamics of an infectious disease. It doesn't capture transmissivity. I am sure other people will find other features of an infectious disease that are missed with the "absolute risk" analysis. It has very complex dynamics. Ample Testing and exact classification of death as pointed out by dr. Ioannidis is the only way to APPROXIMATE "infection fatality rate". <br /> His analysis of the diamond princess cruise ship is very interesting IMO.

    1. On 2020-04-11 12:00:55, user Sherlock Holmes wrote:

      From an article in The Hill: "France has recorded 100 health incidents and four fatalities linked to experimental drugs for those with the coronavirus since late March..

      France has has some 13.500 deaths of coronavirus. Is someone doing the maths?

    2. On 2020-04-14 10:26:28, user Franko Ku wrote:

      New trials must include use of zinc sulfate in <br /> combination with HCQ as well as as in latest French study with over 1000 patients in early stages added to HCQ and azithromycin, using same regimen to prove the great reported results.

    1. On 2020-04-11 21:57:12, user H van Woerden wrote:

      I think that this is a really important paper as it starts to explore the cost effectiveness of different approaches.

      I am concerned about the effect of the fall in average incomes over the next decade on life expectancy and analysis of that issue by this team would be helpful. Particularly the fall in income for lower socio-economic groups.

    1. On 2020-04-12 04:43:17, user Bárbara Souto wrote:

      It seems that the authors want to reinforce an "almost-effect" of chloroquine. They wrote that matched 25 pairs of patients from Changsa, the only place using chloroquine, to compare the 25 individuals who have taken chloroquine, as a kind of case-control nested study. They said they found a difference of 12% and none patient in chlorquine group got worse (with a almost significance. p = 0.074). Well, two groups of 25 with zero outcome in one of them and 12% difference means that only three people in the comparison group got worse. A very small number to consider. But what about the p-value? 0.074 is a very promising p-value, suggesting that the non-significance was just a type II error. We should believe that this p-value is a probability derived from a Fisher exact test. But not. The p-value of a Fisher test from these numbers is 0.234. The 0.074 p-value came from a chi-square calculation. The authors have forgotten that the chi-squared test must not be used when there is a zero in one of the cells. I hope that the reviewers realize this shameful mistake and ask for correction.<br /> Francisco Souto<br /> Brazil

    1. On 2020-04-12 10:21:12, user Lev Yampolsky wrote:

      wait a minute - there is one more confounding co-variable that radically affects death per million for which no control has been done as far as I can tell - timing of epidemics onset in each county. Clearly everything started early in large cities which are both travel hubs and tend to have higher PM2.5 than some nice place in Montana. This analysis will only be possible after the outbreak is over everywhere.

    1. On 2020-04-13 08:45:15, user Alberto Villena wrote:

      Please, have a look to this preprint: Lianne Abrahams, 2020: Covid-19: acquired acute porphyria hypothesis. https://osf.io/4wkfy/<br /> "Macaques infected with SARS-CoV-2 also have<br /> decreased red blood cell numbers (Munster 2020) and susceptibility to<br /> SARS-CoV-2 appears to be determined by blood group; blood group A is most<br /> affected whereas blood group O seems to be protected (Yang 2020). This finding<br /> is concordant with previous studies showing that susceptibility to the 2003<br /> strain of SARS-CoV was determined by blood group (Guillon 2008)."

    2. On 2020-05-30 02:40:13, user M Del wrote:

      If you look at the data table from the study of NY Columbia university the mortality rate is higher for blood type O+ median age 54.8 was 12.4% of the infected and for blood type A+ median age 61.8 it was 12% of the infected , type O has less positive cases compared to the representation of percentage of population but mortality was a little higher even when they were younger.

    1. On 2020-04-13 14:09:18, user Ian Sinclair wrote:

      This study seems to me potentially of enormous significance. I think it would gain greater acceptance if a) the authors explain why they chose to publish before they had reached the numbers specified in the protocol (100 for TAU and 100 for 4 mg group b) they say why they did not report the results for the 2 mg per day group c) they report the actual data on coughs temperature, numbers improved on radiology examination rather than just the significance levels d) they remedy a minor error in the summary (quotes 32 cases as against 31 e) they confirm that the measures were also made by staff who were blind to allocation f) they got themselves an editor who is a native English speaker. I absolutely do not think that the authors have anything to hide but they need to cope with a Western Audience that has been trained to be ultra critical, looking among other things for investigators who stop a trial the moment that it looks to be going their way. My guess is that this was not the case in this instance and that the study was running out of subjects or the authorities were asking for results or some other event that was out of the control of those running the trial. Given the potential world importance of this trial everyone should be trying to offer constructive suggestions for its greater acceptability rather than exercising their brains on ways in which mistakes might have been made.

    1. On 2020-04-13 17:55:08, user Izz Thatso wrote:

      The country benefiting the most from its BCG vaccination program, South Africa, probably has too few deaths in the defined 30 day period for it to be of statistical significance.

      Why do my comments need to be approved?

    1. On 2020-04-14 00:10:20, user Sinai Immunol Review Project wrote:

      Main findings<br /> This report is a retrospective analysis of clinical data collected from 47 patients with confirmed COVID-19 disease (median age 64.91 years; 26 males; 24 severe cases). Upon admission, patients were assessed with an APACHE II score (mortality risk) and SOFA score (organ system dysfunction). Among other clinical parameters, including lymphocyte count, CRP, and enzyme levels, serum lactate dehydrogenase (LDH) level was most positively correlated with APACHE II and SOFA. Additionally, strong positive correlations were found between LDH and observational assessment of severity of pneumonia and of underlying coronary disease. Notably, serum LDH was also positively correlated with other univariate parameters, including troponin and CRP. Meanwhile, LDH was negatively correlated with lymphocyte count across different subsets, including CD4+ and CD8+ T cells.

      Limitations<br /> Several studies have previously identified serum LDH as a predictive biomarker for COVID-19 progression (see references). Therefore, it is important to note that while LDH correlates with disease severity, it does not suffice as a reason for severe disease, so whether it serves as a robust predictive factor is not thoroughly supported; an explanation is still needed between the release of LDH from infected cells and disease severity.

      Significance<br /> The identification of serum LDH as a predictive biomarker does glean insight into possible links between its release into the circulation and immune response to SARS-CoV-2 pathogenesis. It is important to note that LDH release itself requires significant damage to and permeabilization of plasma membranes, leading to necrotic cell death. Therefore, this report, and others like it, should warrant investigation into the link between SARS-CoV-2 infection and potential necrosis or pyroptosis of infected host cells. It is important to note that previous studies concerning biomarkers for SARS-CoV-1 disease have also identified serum LDH as an important correlative factor with disease severity1. And subsequent investigations have studied the activation of inflammasome structures involved in pyroptosis in the presence of SARS-CoV2-3.

    1. On 2020-04-14 07:12:46, user Ole 500 wrote:

      So they dosed the subjects at x2 and x3 the normal dosage of 400 mg/day. If I'm reading this right, they overdosed the patients and rediscovered known overdose side effects?

    1. On 2020-04-14 08:38:01, user Quyen Vu wrote:

      I am not scientist or sort of , however, BCG vaccine has been already used in Vietnam in 70 or 80 years until now , now covid?-19 death rate in Vietnam today is zero (today April 14th) , yes 0? ?at current infected 265 and recovered 155 . BCG is total make sense to me .

    1. On 2020-04-15 13:58:04, user Buck Rogers wrote:

      I don't understand why there is nothing here about the fact that Hydroxychloroquine opens a Zync channel to the cell and prevents the virus from replicating. Hydroxychloroquine was given 200MG twice a day with Azithromycin 500MG once a day. And most important Zync sulfate 220MG once a day. If you don't test the correct protocol what is the value of the research?

    1. On 2020-04-15 14:38:14, user Lawrence Mayer wrote:

      I am very suspicious of articles that are data driven but do not allow us to see the data. Why aren't papers like this being submitted to peer-reviewed journals? They are guaranteeing two week turn around from submission to posted on line.

      I suggest people interested in learning about what has been reviewed by at least one expert join our discussion group on clinical epidemiology and science. We are over 2000 clinicians and epidemiologists discussing the latest empirical results on Covid19

      Lawrence S. Mayer, MD, MS, PhD<br /> Faculty Associate<br /> Harvard University<br /> Professor Emeritus of Public Health, Medicine and Biostatistics

      https/facebook.com/groups/covidnerds

    1. On 2020-04-16 06:58:38, user Kratoklastes wrote:

      It would have been useful to tabulate critical illness (and deaths) - both by age cohort - and to have given some indication of the statistical properties of the estimators (beyond p-values).

      The OR of 66× for the 75+ age cohort in the hospitalisation regression seems outlandish; the raw OR is 4× (i.e., the raw ratio of (Admitted|PosTest) for over-75s compared to the same quantity for 19-44 year olds).

      That looks (to me) like a collinearity issue in the regressor matrix - a really wide CI for one really-obviously-important variable is another clue. (Call me a Bayesian!).

      If your regressors were boolean (i.e., presence/absence) for comorbidities, VIF is not an appropriate test for collinearity: VIF performs poorly for categorical variables. Why not simply test the determinant of X´X, or its condition number, or its smallest eigenvalue?It's not a large matrix by modern standards - so it can't be a computational constraint. R's mctest package does a good job too (omcdiag includes the Farrar-Glauber test)

      I would suspect some rank deficiency caused by correlation between hypertension and variables that represent CVD-ish things; not necessarily pairwise - and this is the problem with booleans.

      Weak collinearity can happen because of weighted sums of columns - 3 boolean columns can give a run of '2' values, that correspond 'enough' with the '1's in the hypertension column. Add in other correlates with hypertension (age, obesity and maleness) and it would be suspicious if there wasn't collinearity.

      I don't think it would be viewed negatively if you dropped the 19 newborns (who are confounders in the 'hospitalisation' regression, since they are always hospitalised), so long as it was clearly disclosed: the presence of those 19 observations will also mess up the 'critical illness' regression as well (it would only require a handful of newborns to need critical care for things unrelated to COVID19, to bias up the OR for their age group, and their presence already biases up recovery rates).

      Lastly: it would be relatively straightforward to furnish the R script and the parameters (and residuals) without furnishing the data - that way interested people could generate their own pseudo-data and run some Monte Carlo experiments to get an idea of the asymptotic properties of the estimates. (To do it properly it would be good to have a covariance matrix so that the pseudodata could be generated by an appropriate cupola).

      .

      It's a pity that (as far as I am aware) Disqus does not permit any type of maths markup.

    1. On 2020-04-16 10:58:04, user Pete Quinn wrote:

      Hello, I expect that your results are far too conservative, by one or two orders of magnitude. I expect it's far more likely that by the end of this first wave (say mid-May, plus or minus), the total reported fatalities will lie in a range between about 1700 to 3500, given current restrictions (which are more lax than in neighboring Norway and Finland, which I estimate to report between about 200-450 and 100-250 respectively), and the new daily cases and fatalities will be small numbers by that point (handful or less fatalities daily). I hope I'm right and you are wrong, not for the sake of being right, but because, well, fatalities...

    1. On 2020-04-16 12:24:09, user Eric Zorrilla wrote:

      Hi,

      I tried to comment on the preprint, but I cant tell if my comment is going through. I noticed some sig/near-sig OR relations in your Suppl. Table 1 that I thought were potentially of interest to highlight in your paper in addition to those you highlighted?

      NON-HOSPITALIZATION also associated with<br /> Oseltamivir (of interest to field imo?)

      And sig or nearly sig with<br /> Atenolol<br /> Lisformin<br /> Metformin<br /> Rosuvastatin

      The joint presence of these anti-metabolic syndrome drugs (hypertension, diabetes, dyslipidemia) seems of interest as may support the emerging hypothesis/view that UNTREATED obesity/metabolic syndrome disorders may be worse than TREATED obesity/metabolic syndrome disorders as COVID-19 comorbidities.

      DECREASED VENTILATION was sig. associated with<br /> Cetirizine (Zyrtec)<br /> Diphenhydramine (Benadryl)<br /> Cholecalciferol (Vitamin D3)<br /> I’m not sure if the first two are an age confound or something interesting vis-à-vis immune reactivity/hypersensitivity, but may be of interest to deconvolute/deconfound?<br /> And the Vitamin D3 may be of interest given various other emerging preprints on it?

      Nice work and best wishes,<br /> Eric

    1. On 2020-04-16 17:51:35, user James Bell wrote:

      In the assumptions, was the ILI resulting from flu an average number or some lower number? This flu season peaked in November/December. A simple Google search reflects this through articles that were published about it at the time. If, in fact, an average ILI was used, doesn't that means the flu ILI are too high in this paper and that the overestimated value presumably belongs to COVID-19 instead? Also, since some in the media are arguing this paper "proves" we don't need the lockdown, is that really true (assuming all the assumptions in the paper are correct)? After all, we're talking about a virus for which none of us (except those who have recovered from it) have any immunity. If you have viruses with equal fatality rates, but we have herd immunity to one and not the other, doesn't common sense dictate there would be more fatalities with the latter (all else being equal)?

    1. On 2020-04-17 20:31:38, user Brian Byrd wrote:

      The authors have made a major error in asserting that

      In order to evaluate whether CCBs have therapeutic effect in COVID-19 patients, we<br /> 199 retrospectively analyzed the medical record of 487 adult COVID-19 patients with<br /> 200 hypertension...

      The effect of a drug is not discernible in this retrospective observational design.

      This major error is extended and compounded when the authors claim that have identified a mortality reduction effect:

      For the primary outcome of mortality, a beneficial effect in reducing the case fatality<br /> 234 rate (CFR) was observed in patients receiving amlodipine besylate, with the CFR<br /> 235 being significantly decreased from 26.1% (12/46) in non-amlodipine besylate treated<br /> 236 group to 6.8% (3/44) in amlodipine besylate treated group (P = 0.022).

      It is irresponsible of the authors to include these incorrect claims of public health significance in a manuscript that is likely to be widely read prior to peer review. It is not clear how the authors can fix any damage that may already have been done, but they should most certainly immediately remove these unsupported and inappropriate claims.

    1. On 2020-04-20 00:46:27, user deutsch wrote:

      A conclusion with only two days oh HCQ and such a smal sample is dishonest!<br /> The probability of missing real differences between the treatments is very high with such a small sample. For example with real death rate 2% vs 4% the probability of false conclusion is 90%....

    2. On 2020-04-15 12:57:31, user ZStat wrote:

      When will the studies be done for patients who get HCQ right after diagnosis of covid19 ? Anybody doing these studies with a control group ? We need to know if HCQ effective if given early.

    1. On 2020-04-21 08:34:15, user Maria wrote:

      Don't you think that lung capacity differences are instead the key to explain the lower incidence and severity of the disease in women and children? Thank you.

    1. On 2023-06-15 13:36:40, user Rachel Gibson wrote:

      This Scientific Correspondence has also been submitted to eLife.

      Comment on ‘The clinical pharmacology of tafenoquine in the radical cure of Plasmodium vivax malaria: an individual patient data meta-analysis’<br /> Authors: Raman Sharma1, Chao Chen2, Lionel Tan2, Katie Rolfe1, Ioana-Gabriela Fita2, <br /> Siôn Jones2, Anup Pingle3, Rachel Gibson1, Navin Goyal4*, Isabelle Borghini Fuhrer5, <br /> Stephan Duparc5, Hema Sharma2†, Panayota Bird2<br /> Affiliations: 1GSK, Stevenage, UK; 2GSK, Brentford, UK; 3GSK, Mumbai, India; 4GSK, Upper Providence, PA, USA; 5Medicines for Malaria Venture, Geneva, Switzerland<br /> *At the time of submission of this Letter, Navin Goyal is no longer an employee of GSK and is affiliated to Johnson and Johnson<br /> †At the time of submission of this Letter, Hema Sharma is no longer an employee of GSK and is affiliated to AstraZeneca

      Abstract<br /> A single 300 mg dose of tafenoquine, in combination with chloroquine, is currently approved in several countries for the radical cure (prevention of relapse) of Plasmodium vivax malaria in patients aged >=16 years. Watson et al.’s recent publication suggests, however, that the approved dose of tafenoquine is insufficient for radical cure and that a higher 450 mg dose should be recommended. In this response, the authors challenge Watson et al.’s assertion based on empirical evidence from dose-ranging and pivotal studies (published) as well as real-world evidence from post-approval studies (ongoing, therefore currently unpublished). The authors confidently assert that, collectively, these data confirm that the benefit–risk profile of a single 300 mg dose of tafenoquine, co-administered with chloroquine, for the radical cure of Plasmodium vivax malaria in patients who are not G6PD deficient, continues to be favourable.

      Introduction<br /> The Plasmodium vivax malarial parasite has a major economic and public health impact, especially in regions such as East Africa, Latin America and South and East Asia.1,2 When present in blood, P. vivax can cause acute malaria with episodes of chills, fever, muscle pains and vomiting. The parasite also has a dormant liver hypnozoite stage, which can reactivate after weeks, months or years, leading to relapses and, potentially, to severe anaemia, permanent brain damage and death.1,2 For effective treatment, eradication of both the blood and liver stages of P. vivax is required (radical cure).2<br /> Since 2018, regulators from the United States initially, and subsequently from Australia, Brazil, Colombia, Thailand, Peru and The Philippines, have approved tafenoquine (as a single oral dose of 300 mg in combination with standard doses of chloroquine) for the radical cure (prevention of relapse) of P. vivax malaria in patients aged >=16 years.1,3-5 A paediatric formulation that allows weight-band-based dosing of children (aged >=2 years) and adolescents is also approved in Australia (since 2022).5 Like primaquine, tafenoquine is an 8-aminoquinoline derivative effective against hypnozoites and all other stages of the P. vivax lifecycle; however, although the World Health Organization (WHO) recommends a 7- or 14-day treatment course for primaquine, tafenoquine is the first single-dose treatment for the radical cure of P. vivax malaria and therefore has patient adherence and convenience advantages.1,3,6 Nonetheless, as an 8 aminoquinoline, the safety profile of tafenoquine is similar to that of primaquine, and both agents can cause oxidant haemolysis in people with glucose-6-phosphate dehydrogenase (G6PD) deficiency.7,8 Acute haemolysis is usually short-lived and does not need specific treatment; however, in rare cases, severe haemolysis may lead to life-threatening anaemia (requiring red blood cell transfusions) or haemoglobinuric renal failure.9 In malaria-endemic regions it has been estimated that 8% of the population are G6PD deficient, although significant variation is reported across regions, with the highest country-specific prevalence estimated in Africa and Western Pacific countries.10,11 G6PD deficiency is an X-linked disorder; males are either G6PD deficient or have normal G6PD activity, whereas females exhibit a wide range of G6PD deficiency.2 Females may be symptomatic if they are homozygous, or if they are heterozygous and inactivation of their normal X chromosome (lyonisation) is skewed towards a deficient phenotype.2,12 Caution is needed because inter-individual variability in the pattern of lyonisation may cause heterozygous females with levels of enzyme activity between 30% and 70% of normal to test as normal for G6PD deficiency using qualitative, phenotypic, rapid diagnostic screening tests.13,14 To reduce the risk of haemolysis, the G6PD status of all potential tafenoquine patients must be determined with a quantitative test capable of accurately differentiating deficient, intermediate and normal G6PD activity levels, and tafenoquine should be withheld from patients with G6PD enzyme levels below 70% of normal.3<br /> Importantly, appropriate clinical practice for the use of 8-aminoquinolines in P. vivax malaria has always been precariously balanced between providing adequate activity against hypnozoites and the real risk of haemolytic harm to patients with G6PD deficiency.15 The cautious benefit–risk balance involved with the single 300 mg dose of tafenoquine has been questioned in a recently published paper in which Watson et al., hypothesise that the current recommended dose of tafenoquine 300 mg is insufficient and that a 450 mg dose of tafenoquine would reduce the risk of relapse.16 That dose is 50% greater than the 300 mg dose approved by the US Food and Drug Administration (FDA), Australian Therapeutic Goods Administration (TGA) and other international regulatory authorities.1,3-5 Herein, the authors discuss concerns regarding the conclusions of Watson et al.<br /> • The benefit–risk profile of tafenoquine 450 mg is not appropriately considered. For example, there is minimal discussion of tafenoquine safety data and key findings from a phase 1 study in healthy female volunteers heterozygous for the G6PD Mahidol variant. This important study demonstrated not only that the haemolytic potential of tafenoquine was dose dependent but also that a single 300 mg dose of tafenoquine had the same potential to cause haemolytic harm as the WHO-recommended dose of primaquine for uncomplicated P. vivax malaria (15 mg/day for 14 days).17,18<br /> • The authors acknowledge that data from the phase 2b, paediatric, pharmacokinetic (PK) bridging study TEACH19 were not available before submission of the Watson et al. manuscript. However, in the TEACH study, in which the tafenoquine dosage in paediatric patients was chosen to match blood exposure in adults receiving 300 mg, tafenoquine was efficacious and generally well tolerated: no patients withdrew from the study because of adverse events.19<br /> • The model used by Watson et al. to predict the recurrence-free rate at 4 months after a 450 mg dose is hypothetical and does not consider data regarding the tafenoquine exposure–response relationship. Importantly, tafenoquine exposure achieved with a single 300 mg dose approaches the plateau of the exposure–response curve; therefore, the incremental recurrence-free rate gained by the proposed 50% increase in dose is small and unlikely to be justified by overall benefit–risk considerations.3 In addition, as primaquine and tafenoquine have different PK and metabolic profiles, the authors consider the extrapolation of data from primaquine to tafenoquine to be problematic.2,9<br /> • The authors feel that, overall, some of the conclusions do not acknowledge evidence-based safety concerns for a >300 mg dose of tafenoquine and do not consider additional data from the INSPECTOR study that the recurrence rate of P. vivax infection within 6 months of tafenoquine treatment was not significantly affected by bodyweight.20<br /> Watson et al. mentioned the phase 2b dose-selection study (DETECTIVE) of tafenoquine,21 from which a single 300 mg dose was chosen for phase 3 evaluation in adults. However, the authors did not point out that, in this study, exposure was a significant predictor of efficacy and doubling the tafenoquine dose from 300 mg to 600 mg was associated with only a marginal increase (from 89.2% to 91.9%) in the primary efficacy endpoint, relapse-free efficacy at 6 months.21 Moreover, in addressing the INSPECTOR study of tafenoquine in Indonesian soldiers, the authors did not specify that this was a study of tafenoquine administered with an artemisinin-based combination therapy rather than chloroquine and, as such, is not directly comparable due to poorly understood but confirmed interactions impacting tafenoquine efficacy.20 Watson et al. also suggest that tafenoquine 300 mg is likely inferior to ‘optimal primaquine regimens’, but it is unclear whether such regimens are the WHO-recommended schedules of primaquine or regimens defined as optimal based on non-regulatory studies of primaquine. The authors provided no specific reference or dosage characterising optimised primaquine therapy, so this a priori inferiority cannot be evaluated.<br /> Methods<br /> The hypothetical causal model proposed by Watson et al. for the clinical pharmacology of tafenoquine for the radical treatment of P. vivax malaria is similarly problematic. Central to this model are methaemoglobin (MetHb) production and active metabolites. However, MetHb is not a validated biomarker of tafenoquine efficacy, and currently there is no evidence, from non-clinical or clinical studies, of circulating active metabolites of tafenoquine; if such metabolites were fleetingly present, they would require extraordinary potency to exert any significant pharmacodynamic effect.22<br /> Regarding radical curative efficacy, Watson et al. selected P. vivax recurrence within 4 months as their primary endpoint. However, the trial-defined primary endpoint at 6 months from the pivotal tafenoquine clinical trials8,21,23 was an FDA requirement and was mandated for analysis purposes. This was to maximise the probability of capturing relapses, including those from regions with longer latency periods. Watson et al. used the INSPECTOR study20 as one of two reasons to justify the selection of a 4-month endpoint. Relapse rates differ greatly from country to country, so the duration of the endpoint should not be based on rates observed in a single country. Moreover, the 6-month rate of loss to follow-up (only 9.1%) does not justify a change of treatment endpoint from 6 months to 4 months.<br /> In their efficacy models, Watson et al. explored the association between the odds of P. vivax recurrence and the following predictors: mg/kg dose of tafenoquine; AUC0–?; peak plasma tafenoquine concentration; terminal elimination half-life; and Day 7 MetHb level. However, details of how the best predictor was selected and how statistical significance was judged were not provided.<br /> Results<br /> Use of a 4-month versus 6-month follow-up period<br /> A key focus of the Watson et al. manuscript is that the authors describe a possible association between tafenoquine mg/kg dose and the odds of recurrence (using logistic regression), with a 4-month rather than the original 6-month follow-up. An odds ratio of 0.66 (95% confidence interval [CI]: 0.51, 0.85) is cited by Watson et al. in their analysis of the effect of tafenoquine mg/kg dose in patients who received tafenoquine 300 mg, but descriptive details for this result and the analysis are limited. Figure 2 in the Watson et al. manuscript shows Kaplan–Meier survival curves for time to first recurrence, based on tafenoquine mg/kg dosing category, but some areas require clarification, such as how the dosing bands were selected.<br /> Rationale for tafenoquine dose selection<br /> Importantly, the classification and regression tree analysis, in which a clinically relevant breakpoint tafenoquine AUC value of 56.4 ug·h/mL was identified, was not discussed.24 Population PK modelling revealed that tafenoquine 300 mg would provide systemic exposure greater than or equal to the AUC breakpoint in approximately 93% of individuals, who would have a high probability (85%; 95% CI: 80, 90) of remaining relapse-free at 6 months.24 Therefore, this ‘… model-based approach was critical in selecting an appropriate phase 3 dose’ for tafenoquine.24 Although data from the TEACH paediatric study19 were not available when Watson et al. conducted their analysis, had the data been available, they would have validated the AUC approach to tafenoquine dose selection, with an overall efficacy of approximately 95%.19 Individuals (aged 2–15 years) were given tafenoquine, based on bodyweight, to achieve the same median AUC as the 300 mg dose in adults (children weighing >10–20 kg received tafenoquine 100 or 150 mg; >20–35 kg received 200 mg; and >35 kg received 300 mg). The recurrence-free rate at 4 months was 94.7% (95% CI: 84.6, 98.3),19 and the TEACH study supported the successful approval of tafenoquine for children aged 2–16 years by the Australian TGA in March 2022.5<br /> Another important counter to the mg/kg-based dose selection is that, when bodyweight categories were fitted as a continuous variable in the INSPECTOR study (using data for the time to recurrence for all participants), neither bodyweight nor bodyweight-by-treatment interactions were statistically significant (p=0.831 and p=0.520, respectively).20<br /> Use of an unvalidated biomarker<br /> Although Watson et al. state that increases in blood MetHb concentrations after tafenoquine administration were highly correlated with mg/kg dose, no correlation coefficients were presented. It should also be re-emphasised that MetHb is not a validated, surrogate biomarker of antimalarial treatment efficacy as a radical cure for P. vivax malaria and was used as a safety measure in the INSPECTOR study.20<br /> Potential safety concerns<br /> In the Tolerability and safety section, Watson et al. state that severe haemolytic events were rare; however, this is because all the studies were randomised and controlled, which excluded patients with <70% G6PD activity. In addition, no mention was made that, in one of the constituent studies (which examined the dose–response for haemoglobin decline in participants with 40–60% G6PD enzyme activity),17 dose escalation of tafenoquine from 300 mg to 600 mg was not attempted due to safety concerns about potential haemolysis in patients with G6PD deficiency. In tafenoquine-treated patients in the real-world setting, some instances of severe haemolysis might be expected, and it is already known from the previously highlighted phase 1 study that the haemolytic potential of tafenoquine increases with increasing dose.17 Watson et al.’s Tolerability and safety section also mentions that one tafenoquine-treated patient had a >5 g/dL decrease in haemoglobin level, but the baseline haemoglobin level and tafenoquine dose are not mentioned. The section may have benefitted from a holistic discussion of safety parameters per tafenoquine dose group: for example, the occurrence of serious adverse events, gastrointestinal adverse events (beyond the selective discussion of vomiting within 1 hour post dose) and neuropsychiatric adverse events.<br /> Discussion<br /> Watson et al. conclude that ‘the currently recommended adult dose is insufficient … increasing the adult dose to 450 mg is predicted to reduce the risk of relapse’; however, the authors have raised several concerns relating to these conclusions. In particular, the authors feel that the safety concerns associated with a higher-than-approved tafenoquine dose have not been thoroughly considered: the safety analysis is limited, and the increased risk of haemolysis in patients with G6PD deficiency that a 450 mg tafenoquine dose (which is 50% greater than the approved 300 mg dose) would pose in vulnerable populations in limited-resource settings is not adequately discussed. In some malaria-endemic regions, 8% of the population may be G6PD deficient, although wide variability exists, and in sub Saharan Africa and the Arabian peninsula the prevalence of G6PD deficiency may exceed 30%.10,11 Therefore, in regions with fragile healthcare systems and limited availability of relevant testing for G6PD deficiency, potential exists for a significantly increased risk of haemolysis if tafenoquine is administered at an above recommended dose (450 mg). Importantly, off-label use of a dose not robustly evaluated in clinical trials would pose a considerable risk to patient safety.<br /> Regarding tafenoquine efficacy, the rationale for a dose increase to 450 mg has limitations. Watson et al. suggest that a 50% increase in the adult dose of tafenoquine (from 300 mg to 450 mg) would prevent one relapse of malaria for every 11 patients treated. However, this number-needed-to-treat estimate is not balanced by a number-needed-to-harm estimate for acute haemolytic anaemia. In addition, the phase 2b part of the DETECTIVE study21 showed that, in countries where the trial was carried out, single doses of tafenoquine 300 mg and 600 mg had similar relapse-free efficacy at 6 months (89.2% and 91.9%, respectively); therefore, the lack of additional benefit for tafenoquine 600 mg in DETECTIVE and the phase 1 study, which demonstrated dose-dependent haemolytic potential for tafenoquine, favour a 300 mg dose.<br /> In summary, based on currently available data, dosing tafenoquine at the approved 300 mg dose, in combination with chloroquine, carefully balances efficacy and safety in the radical cure of P. vivax malaria; indeed, tafenoquine 300 mg demonstrated a favourable benefit–risk profile in a comprehensive clinical development programme that included at-risk populations in regions with fragile or resource-restricted healthcare systems. The arguments raised by Watson et al. come with the concerns articulated here, and the authors confidently assert that a tafenoquine dose increase from 300 mg to 450 mg is not supported by available fact-based evidence for the radical cure of P. vivax malaria in adults aged >=16 years.

      References<br /> 1. GSK. US FDA approves Krintafel (tafenoquine) for the radical cure of P. vivax malaria [press release]. July 20, 2018. https://www.gsk.com/en-gb/media/press-releases/us-fda-approves-krintafel-tafenoquine-for-the-radical-cure-of-p-vivax-malaria/ (accessed 26 April 2023).<br /> 2. Hounkpatin AB et al. Clinical utility of tafenoquine in the prevention of relapse of Plasmodium vivax malaria: a review on the mode of action and emerging trial data. Infect Drug Resist 2019;12:553–570.<br /> 3. GSK. Krintafel. Highlights of prescribing information. https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/210795s000lbl.pdf (accessed 26 April 2023).<br /> 4. GSK, Medicines for Malaria Venture. Perú becomes second malaria-endemic country in Latin America to approve single-dose tafenoquine for radical cure of P. vivax malaria [press release]. https://www.vivaxmalaria.org/sites/pvivax/files/content/attachments/2021-01-25/GSK%20-%20MMV%20PRESS%20RELEASE%20TAFENOQUINE%20APPROVED%20IN%20PERU.pdf (accessed 26 April 2023).<br /> 5. Medicines for Malaria Venture. Single-dose Kozenis (tafenoquine) approved for children with Plasmodium vivax malaria by Australian Therapeutic Goods Administration. https://www.mmv.org/newsroom/press-releases/single-dose-kozenis-tafenoquine-approved-children-plasmodium-vivax-malaria (accessed 26 April 2023).<br /> 6. World Health Organization. WHO guidelines for malaria, 14 March 2023. https://www.who.int/teams/global-malaria-programme (accessed 26 April 2023).<br /> 7. Milligan R et al. Primaquine at alternative dosing schedules for preventing relapse in people with Plasmodium vivax malaria. Cochrane Database Syst Rev 2019;7:CD012656.<br /> 8. Llanos-Cuentas A et al. Tafenoquine versus primaquine to prevent relapse of Plasmodium vivax malaria. N Engl J Med 2019;380:229–241.<br /> 9. Baird JK. 8-Aminoquinoline therapy for latent malaria. Clin Microbiol Rev 2019;32.<br /> 10. Howes RE et al. G6PD deficiency prevalence and estimates of affected populations in malaria endemic countries: a geostatistical model-based map. PLoS Med 2012;9:e1001339.<br /> 11. P. vivax information hub. G6PD global prevalence. https://www.vivaxmalaria.org/diagnosis-treatment/g6pd-deficiency/g6pd-global-prevalence#:~:text=G6PD%20Global%20Prevalence,-Photo%3A%20Jaya%20Banerji&text=G6PD%20deficiency%20affects%20around%20400%20million%20people%20globally (accessed 26 April 2023).<br /> 12. Domingo GJ et al. Addressing the gender-knowledge gap in glucose-6-phosphate dehydrogenase deficiency: challenges and opportunities. Int Health 2019;11:7–14.<br /> 13. Chu CS et al. Haemolysis in G6PD heterozygous females treated with primaquine for Plasmodium vivax malaria: a nested cohort in a trial of radical curative regimens. PLoS Med 2017;14:e1002224.<br /> 14. Baird JK et al. Noninferiority of glucose-6-phosphate dehydrogenase deficiency diagnosis by a point-of-care rapid test vs the laboratory fluorescent spot test demonstrated by copper inhibition in normal human red blood cells. Transl Res 2015;165:677–688.<br /> 15. Shanks GD. Historical 8-aminoquinoline combinations: not all antimalarial drugs work well together. Am J Trop Med Hyg 2022;107:964–967.<br /> 16. Watson JA et al. The clinical pharmacology of tafenoquine in the radical cure of Plasmodium vivax malaria: An individual patient data meta-analysis. Elife 2022;11:e83433.<br /> 17. Rueangweerayut R et al. Hemolytic potential of tafenoquine in female volunteers heterozygous for glucose-6-phosphate dehydrogenase (G6PD) deficiency (G6PD Mahidol variant) versus G6PD-normal volunteers. Am J Trop Med Hyg 2017;97:702–711.<br /> 18. World Health Organization. Guidelines for the treatment of malaria, 3rd ed. https://apps.who.int/iris/handle/10665/162441 (accessed 26 April 2023).<br /> 19. Velez ID et al. Tafenoquine exposure assessment, safety, and relapse prevention efficacy in children with Plasmodium vivax malaria: open-label, single-arm, non-comparative, multicentre, pharmacokinetic bridging, phase 2 trial. Lancet Child Adolesc Health 2022;6:86–95.<br /> 20. Sutanto I et al. Randomised, placebo-controlled, efficacy and safety study of tafenoquine co-administered with dihydroartemisinin-piperaquine for the radical cure of Plasmodium vivax malaria (INSPECTOR). Lancet Infect Dis [2023 May 23:S1473-3099(23)00213-X doi: 101016/S1473-3099(23)00213-X Epub ahead of print PMID: 37236221].<br /> 21. Llanos-Cuentas A et al. Tafenoquine plus chloroquine for the treatment and relapse prevention of Plasmodium vivax malaria (DETECTIVE): a multicentre, double-blind, randomised, phase 2b dose-selection study. Lancet 2014;383:1049–1058.<br /> 22. GSK. Investigator brochure. Data on file.<br /> 23. Lacerda MVG et al. Single-dose tafenoquine to prevent relapse of Plasmodium vivax malaria. N Engl J Med 2019;380:215–228.<br /> 24. Tenero D et al. Exposure-response analyses for tafenoquine after administration to patients with Plasmodium vivax malaria. Antimicrob Agents Chemother 2015;59:6188–6194.

      Authors’ contributions<br /> Hema Sharma, Lionel Tan, Katie Rolfe, and Navin Goyal contributed to the conception or design of the studies the paper contains data from. All authors contributed to data analysis or interpretation. All authors contributed to the development and writing of this correspondence and approved the final submitted version.

      Conflicts of interest statements <br /> Raman Sharma, Siôn Jones, Rachel Gibson, Katie Rolfe, Lionel Tan, Ioana-Gabriela Fita, Chao Chen, Panayota Bird, and Anup Pingle are employees of, and shareholders in GSK.<br /> Hema Sharma is a former employee of GSK, a shareholder in GSK and a current employee of AstraZeneca. Navin Goyal is a former employee and shareholder in GSK and a current employee of Johnson and Johnson. Isabelle Borghini Fuhrer and Stephan Duparc have no conflict of interest to report. <br /> Acknowledgements <br /> Medical writing support was provided by David Murdoch, a contract writer working on behalf of Apollo, and Alex Coulthard of Apollo, OPEN Health Communications, funded by GSK Biologicals SA, in accordance with Good Publication Practice 3 (GPP) guidelines (www.ismpp.org/gpp-2022) "www.ismpp.org/gpp-2022)").

      Funding<br /> Funding for this article was provided by GSK Biologicals SA.

      Data availability<br /> Data sharing is not applicable to this article as no datasets were generated or analysed.

    1. On 2020-06-03 14:31:19, user Nick Bauer wrote:

      Your paper specifically picks out "A-positive" individuals. I may be missing something, but I don't see any discussion of Rh factor in the text or presence in the statistics?

    2. On 2020-06-08 21:16:33, user Christian Lehmann wrote:

      nice article - the problem is the communication in by the public media - The authors show a ASSOCIATION - this is something completely different from a causality - so please be careful, having a certain blood group does not tell you something about your possible disease outcome.

    3. On 2020-06-09 20:00:05, user Anne Smith wrote:

      I'm blood group O. That should require two of the same allele. The article reports an association between ARDS and a specific SNP, but all of the discussion in the paper is about the relationship between major blood group and chance of ARDS. At 23andme, for rs657152, on a ABO gene, my genotype is A/C. My first question is how is that even possible, since if this SNP determines blood group and I'm type O I should have either two A's or two C's. Second, I and many others would really think you to present the odds of specific variants on rs657142 and ARDS! For two A's, two C's, and one of each.

    4. On 2020-06-10 00:22:53, user Michael Jolley wrote:

      From Fig. 3A, it would seem that rs11385942 sits in the LZTFL1 gene, which the authors did not mention. This gene is highly expressed in lung epithelial cells (and is mutated in many lung cancers). Since LZTFL1 protein is involved in regulating ciliary trafficking and controlling ?-catenin nuclear signaling, might not this be an important clue as to how the virus operates (which the authors missed)? https://www.nature.com/arti...

    1. On 2020-04-11 00:50:14, user Kirsten McEwen wrote:

      There's clearly a need to monitor COVID-19 sequence mutations over primer sites. Is there an initiative to share this info among testing facilities?

    1. On 2020-06-10 22:02:36, user mark wrote:

      There seems to be an error that needs to be corrected in this line: "Results: Univariate analysis showed lower mortality in the ivermectin group (25.2% versus 15.0%......"

    1. On 2021-01-12 19:27:05, user Antonio Bernabe-Ortiz wrote:

      A little favor to the authors... please adjust the estimates by ischemic heart disease as there is a potential disbalance between arms as seen in Table 2... thanks...

    1. On 2020-06-12 15:57:05, user Rosemary TATE wrote:

      What is the correlation hypothesis? With all due respect I think that a course in statistics could be very beneficial for the author.

    1. On 2020-05-07 03:52:42, user Alisha Geldert wrote:

      We thank the authors for their detailed analysis of a suite of N95 decontamination approaches, with specific appreciation for the direct applicability to medical center needs. We see the manuscript – once published in a peer-reviewed journal – as being an excellent resource for medical center decision makers, as well as those working to implement the decontamination methods. With a spirit of attention to the existing peer-reviewed literature and rigor needed in this crisis, we offer a review of areas where improvements would benefit the study as well as (and more importantly) any readers who may adopt the approaches. The authors are aware of the following major comments summarized below, and are working diligently to provide necessary clarifications and revisions.

      1. The UV-PX experimental design and choice of combination approach does not appear to be consistent with evidence on effective approaches for UVGI/UV-C ultraviolet decontamination of N95s, presenting a major concern. To address this, consider providing a reader with clearer justification for the ‘unconventional’ approach by perhaps answering the following questions:<br /> ---Were longer duration UV-PX treatments investigated? The fluence delivered during the 5-minute treatment time is unsupported by the evidence for UV-C decontamination of N95s [Lore et al., 2012; Mills et al, 2018; Heimbuch & Harnish, 2019].<br /> ---It is not clear why the authors suggest coupling of UV-PX with moderate RH heat before testing UV-PX alone, when the benefit of adding UV-PX is not described (perhaps stemming from the very low pathogen inactivation observed with UV-PX alone, as would be expected from the ~50X too low delivered germicidal fluence using this protocol). As the protocol deviates from CDC guidance [CDC, 2020], a rationale and supporting peer-reviewed references would be essential.

      2. Important details are missing in the methods section. Please provide key details about the UV-PX setup to ensure replicable research reporting, specifically:<br /> ---Measures taken, if any, to ensure respirators are directly illuminated on both sides. <br /> N95 respirator placement relative to and distance from the light source. As irradiance, and therefore fluence, depends on distance between source and target, this is a critical parameter.<br /> ---Please specify the reflective material used in the UV room, the make and model number of the flame irradiance spectrometer, and whether the irradiance measurements reported in Supplementary Table 3 were measured within the UV room with reflective walls or within an alternative setting. Do the irradiance measurements represent the irradiance at the side of the N95 facing the Xenex UV-PX source or irradiance at areas indirectly exposed to UV light? <br /> ---Please clarify whether the measured irradiance represents the irradiance of one pulse or the average irradiance over multiple cycles.

      3. There appears to be a potential issue with the conclusions reported in the abstract: the specific experimental parameters shown to yield high levels of pathogen inactivation (moderate RH heat) were not tested for N95 function, so the following statement might be confusing or misleading:<br /> “High levels of biological indicator inactivation were achieved following treatment with either moist heat or VHP. These same treatments did not significantly impact mask filtration or fit.”<br /> The limitations of the proposed approaches and the need for additional testing should be clarified.

      References cited: <br /> 1. Lore et al., 2012: https://academic.oup.com/an...<br /> 2. Mills et al., 2018: https://www.ncbi.nlm.nih.go...<br /> 3. Heimbuch & Harnish, 2019: https://www.ara.com/sites/d...<br /> 4. CDC guidance on N95 decontamination: https://www.cdc.gov/coronav...

    1. On 2020-02-10 20:39:56, user Hoku Toki wrote:

      Would be interesting to supplement positions median with mean to evaluate the degree of symmetry of the distribution, and dispersion stats as SD. For ex on age.

    1. On 2020-09-14 12:51:47, user Iván Williams wrote:

      Simple and useful methodology, thanks. I think there is a lag of two weeks between onset and death, so maybe the estimated stock could be lagged too.

    1. On 2020-05-12 04:19:20, user Gordon Lehman wrote:

      Testing so far has not been rigorous or systematic. Solely based so far on clinical and primary care presentations, and who shows up at burger king drive byes.

    1. On 2020-05-12 13:44:41, user Abdul Mannan Baig wrote:

      What a brilliant contribution to ongoing COVID-19 research.<br /> The olfactory bulb findings took the science way forward to understand the loss of smell and taste.<br /> Wow<br /> Abdul Mannan Baig

    1. On 2020-03-23 20:09:08, user halrhp wrote:

      wha is the mechnism that changes at 2 m? Gravity or particle mass or both. my guess is the combination. ahow large are COV19 virus droplets. Y. Li is very smart.

    1. On 2020-05-22 14:10:19, user C'est la même wrote:

      The very high female proportion and narrow recruitment strategy suggest these results might not represent the community as a whole, due to participation and response biases.

    1. On 2021-11-20 21:08:59, user MICHAEL A WALLACH wrote:

      Does the data include unvaccinated workers in the study who never tested positive for covid? I assume it does but was not addressed. How did they compare with the two groups compared in the study?

    1. On 2020-05-27 17:48:21, user John G. wrote:

      I'm a non health or research person. But I thank you and others who are quickly researching and reporting on observations. We don't know exactly how it interacts but these results could be very promising. Thank you for sharing this.

    1. On 2020-05-30 01:10:44, user ajaxthegreat wrote:

      This should be the final nail in the coffin for lockdowns. In a nutshell, lockdowns don't work, and especially belated ones are in fact worse than useless compared to more moderate NPIs applied earlier in the curve. Thus, ending stay-home orders and reopening most non-essential businesses can and should be done *yesterday* with practically no risk of resurgence of the disease. Higher-risk businesses like restaurants, bars, and nightclubs, as well as other amusements, should be reopened a bit more gradually of course, but once a country is at least two weeks post-peak (almost every country) they can begin reopening those as well, albeit with restrictions.

      For masks, they probably do work well (just ask Japan and Taiwan) but confounding may have hampered the finding of any good signal in the noise for these policies beyond the first two weeks, especially since these policies were applied belatedly.

      As for schools, closing them (even very briefly and locally like Taiwan did) likely worked in the short term, but more recent data show that reopening them does not seem to carry much if any risk of resurgence. For this particular virus, unlike with influenza, children are not superspreaders, in fact they are far less susceptible and infectious compared with adults, but school closures likely prevented at least some adults from infecting each other, both at school and by effectively forcing parents to work from home temporarily at a critical point in the epidemic curve.

    1. On 2020-05-31 07:59:11, user ashkan homayuni wrote:

      I have noticed a preprint entitled "covid-19 in iran, a comprehensive investigation from exposure to treatment outcomes" <br /> I have read the article and noticed authors mentioned that during 22 feb to 5 march, 100 patients with covid-19 were included; however I working as an internist physician in YAS hospital (where study has noted to be consucted in) have to report you that we had only 46 cases during above mentioned period of time. I am so concerned about reliability and honesty of the data presented by authors and I am afraid that it may subject to data falsification or data duplication.

    1. On 2020-06-01 15:30:23, user Nathan Goodman wrote:

      Regarding RJones1’s question about the data: WA Dept of Health provides downloads updated weekly https://www.doh.wa.gov/Emer.... The download is near the bottom of the page.

      Timothy Siegel’s comment about the validity of using positive test results as a proxy for prevalence is absolutely correct. Sadly it’s the only data available. Various models use the data to estimate actual prevalence, eg, https://covid19-projections..., and more specifically. the WA estimate at https://covid19-projections...

      For what it’s worth, weekly case counts (meaning the number of positive tests) has declined steadily in WA from 2,555 week of Mar 22 to 1,256 week of May 17 (the last week for which the DOH download is reasonably complete). Over the same period, case counts for the youngest group (age 0-19) have increased from 65 to 180.

      The pressing question is whether this increase is real or an artifact of testing strategy. Are more kids getting sick, or is it simply that more kids are getting tested.

      Dr. Malmgren’s preprint is silent on this question.

    1. On 2020-06-02 19:09:27, user Irene Petersen wrote:

      There are two key issues with this study.

      1) The majority of people in this study (about 16 million) are NOT at risk of the outcome (dying from COVID19) as they will not have been infected.

      2) Dying from COVID19 is a two-stage process - A) Risk of getting infected B) Risk of dying once infected. This study conflates the two. Thus you can't tell whether an elevated risk is due to A or B.

    1. On 2020-06-03 04:00:14, user ??????? ??????? wrote:

      You can not trust the testimony of patients who are admitted to hospitals with a serious illness. Currently, the entire public health system is anti-smoking. For example, insurance companies may apply certain restrictions for smokers. Therefore, patients hide the fact of smoking. In connection with the above, the fact of smoking should be documented by objective methods.

    1. On 2020-06-04 09:38:49, user Rosemary TATE wrote:

      This is very interesting. However, the most recent report by ICNARC, which is based on 9347 UK patients, suggests that ethnicity is associated with worse outcome in hospital and also with increased hospital admissions.<br /> https://www.icnarc.org/Our-...<br /> They are preparing a manuscript - hopefully it will be out soon.

    1. On 2020-06-06 13:59:32, user Nayo57 wrote:

      While the result of this interesting and meaningful analysis may be statistically correct: a reduction of R of 0.04 with 10% mobility reduction does not explain the vast reductions from R = 3-4 at outbreak to below 1. A rough analysis of WHO reported case data and Google mobility data gave a similar result e.g. for time to reach R<1 or cumulative cases per population, measures one would expect to be impacted by effective social distancing. The best conclusion may be that mobility index as provided is not a suitable measure to assess or guide policies to contain COVID-19: Fig1 (Germany: increase in mobiilty index while R stays <1), Fig. 2(USA: decrease in mobility by further increase in R) and the scatter in Fig 3 support this view.

      The interpretation would rather be that BEHAVIOUR during mobility activities matter much more than the QUANTITY of mobility. Alternatively, more focussed indexes (restaurants/bars; cinemas/theaters..) may be worth while to examine if they could be useful.

    1. On 2020-06-08 10:58:42, user ReviewNinja wrote:

      Nice and very useful work! Saliva could be a great diagnostic tool! It would fix some problems about NP swab shortages and the necessity of accurate sampling by skilled personnel.

      I was wondering about some details:<br /> 1.) The Ct values that you observe are lower for saliva compared to NP swabs. Could this (sometimes) be a matrix effect? Did you run standard curves of positive controls in a constant negative background of human RNA obtained as well from saliva as from NP swabs? A swab might sometimes just lead to more PCR inhibition? NP swabs and saliva might just have different Cqs for their LOD. Another way to assess this is with digital PCR.

      2.) Just out of interest: What if you normalize the viral values for RNAseP expression (as a surrogate for sampling efficiency)? Are the values comparable then? Or is there really just more virus present in saliva? (I know, for diagnostics, this does not matter.)

      3.)The RNAseP results indicate that sampling with NP swabs can indeed be an issue. <br /> It would have been nicer to also take standard curves here and plot RNAseP values, as Cq's on itself can be quite variable in between runs due to run-to-run variability.

      3.) Just a last question, maybe for another paper;): are you also working on extraction free detection in saliva?

    1. On 2020-06-09 18:29:42, user Chris Winchester wrote:

      Would it be possible to study in your data set the quality of RCTs from different funders (e.g. commercial funders vs governmental funders, charities and NGOs)?

    1. On 2020-06-10 08:42:20, user Rita Van Dingenen wrote:

      Although more confounding factors have been taken into account in the revised version, I am not convinced that they can capture the specific dynamics of the COVID-19 epidemic where the spreading seems to be strongly driven by sparse super-spreading events, taking place under specific circumstances (e.g. relatively crowded social/cultural/sports events involving shouting, singing, strong breathing). The number and frequency of such events (or infrastructure) could be an important determinant of which PM2.5 may just be a proxy. Further, in stead of mortality rate, case fatality rate would be the more appropriate indicator to establish a possible impact of air pollution, removing a lot of noise from confounding factors.

    1. On 2020-06-13 03:34:34, user kpfleger wrote:

      Why are the 25(OH)D levels reported here (43 or 44 nmol/L w/ IQRs of 32 or 31 respectively for the n=580 C19+ and n=723 C19- UK Biobank cases) so much higher than those reported in Hastie et al, "Vitamin D concentrations and COVID-19 infection..." Diabetes Metab Syndr., 2020: https://pubmed.ncbi.nlm.nih...<br /> which reported median 25(OH)D of 29nmol/L w/ IQR 10-44 for C19+ & 33 IQR 10-47.<br /> This is a huge difference for 2 papers with online publication dates 2 days apart both pulling data from the same source.

      See also D'Avolio et al, "25-Hydroxyvitamin D concentrations are lower in patients with positive PCR for SARS-CoV-2", Nutrients, 2020 and Meltzer et al, “Association of vitamin D deficiency and treatment with COVID-19 incidence”, medRxiv, 2020 for 2 different studies that found in contrast to the top level conclusion here, that low D was associated with higher C19 incidence. Discussion of all 4 paper of these papers in the "D deficiency might be associated with higher infection risk" of the review: "Low vitamin D worsens COVID-19": http://agingbiotech.info/vi...

    1. On 2020-06-14 10:02:05, user Wen Minneng wrote:

      Our paper have analyzed so many variables: new case, new death, latitude, temperature, humidity, rainfall, sunshine UV. The article, COVID-19 and climate: global evidence from 117 countries, only focus on two variables: cases and latitude. Their result is: "A one-degree increase in absolute latitude is associated with a 2.6% increase in cases per million inhabitants."

    1. On 2020-06-14 20:56:36, user Marm Kilpatrick wrote:

      Thank you for this very nice synthesis of the literature on this topic and for your careful thoughts on shortcomings of the available data. One small suggestion is that in Fig 2 you propose a few options for the "Hypothetical distributions of S ARS -CoV-2 viral load". One option that has been proposed and fit to data in one of your cited papers (He et al 2020 Nat Med) is a gamma distribution that starts before symptom onset. In that paper they suggest it starts 2.3 d before symptom onset. <br /> Also, there are now several papers linking viral loads to infectious virus. Several can be found in this thread:<br /> https://twitter.com/Disease...

    1. On 2020-06-15 22:51:11, user Marm Kilpatrick wrote:

      One more quick comment. The paper indicates that those tested were asymptomatic but in the Methods and other sections it's not clear how this was verified. Were those being tested asked to report on whether they currently had symptoms? If so, were they not tested that week? Do you have information on when those that tested positive developed symptoms? This data is rare and would be very valuable. Thank you!<br /> marm

    1. On 2020-06-17 11:54:35, user Dr. D. Miyazawa MD wrote:

      Possible reason for why in multivariable analysis with BMI >=30, diabetes, hypertension, was outside the limits of standard statistical significance in Black patients.<br /> Obesity correlates with hypertension, diabetes, and is more prevalent in Black people.<br /> https://doi.org/10.22541/au...

    1. On 2020-06-17 15:34:18, user Dr. Amy wrote:

      NSAID users were MUCH older and more likely to have comorbidities associated with adverse outcomes. Can't make too much of this without controlling for those big confounders.

    1. On 2020-06-18 07:29:29, user Hilda Bastian wrote:

      Thank you for this interesting preprint. I could not figure out how these particular decedents came to be selected for this study (apologies if I overlooked the explanation). That's a critical piece of information, and it would be helpful if it were detailed in the next version.

    1. On 2020-06-18 12:17:43, user Paul Warren wrote:

      the discussion states that: ‘There are three types of risk for medical staff. The first relates to their biology, the second their environment and the third to the exposure. This tool evaluates the former in order to advise mitigation of the latter...’ Leaving aside the clumsy english, as I understand it the tool is evaluating data on covid deaths from ICNARC. To die from covid you have to catch it and then fall ill, so it is a composite measure of exposure, biological factors, and other bits and pieces – health behaviors and so on. How this composite cashes out is generally thought to depend on the risk factor in question. For<br /> instance the risk of increasing age may be more to do with biology than exposure, whereas for ethnicity many commentators say that exposure probably is an important factor. So, to a variable extent, the tool uses a measure of exposure across society as a whole to advise on exposure mitigation in healthcare settings. Now in practice this may not be a great problem, but isn’t it at least worth a mention?<br /> Hardly a concern, maybe just something as a nonscientist I struggled with was relative<br /> and absolute risk. Does normalising the risks to 40-49yr old women have any effect on the relative scores of the different risk factors, when compared with other possible choices, say 20-29yr women or whatever? My logic isn’t up to it, but if it does make a difference the choice of 40-49 would need justification. Also, does the correlation with opensafely and PHE document absolute risk scores mean that this tool could be used to judge absolute risk? While for justice and equity relative risk may be the best measure, in other situations both employer and employee may want some idea of absolute risk.

    1. On 2020-06-18 22:51:21, user Sasha Bruno wrote:

      I think there’s some issues in the paper with some of the assumptions made— especially about virus exposure to healthcare workers without confirmed positive tests via RT-PCR. Assuming virus exposure to healthcare workers during the early months of the outbreak, assumes that all the healthcare workers tested were also working during the early months of the outbreak and worked in wards, units and areas of the hospital likely for virus exposure.

      Some units in those hospitals likely pose a higher risk of exposure than others. For example, a radiologist reviewing imaging results in an office at a hospital probably has a lower risk of exposure than a nurse in the ER or a pulmonologist in the ICU.

      While I agree that some significant proportion of healthcare workers were likely exposed— without confirmed RT-PCR tests it’s a tenuous leap to draw conclusions about the low proportion (4%) of IgG antibodies in the group because we don’t know how many of them were in fact infected by the virus. It’s a logical premise but, too many assumptions and extrapolations to get there. Further investigation is needed.

      But, I think the >10% of confirmed Covid-19 patients who no longer had detectable IgG antibodies post 21 days is a significant finding. 10% is too large of a percentage to be easily explained by potential errors in testing or sample collection.

    1. On 2020-06-18 23:37:52, user Florin wrote:

      There was no mention of Japan, Taiwan, or Hong Kong, countries and territories which have successfully dealt with the pandemic without doing lockdowns. Taiwan didn't close its schools. Japan didn't do much testing or tracing. Hong Kong was late in imposing a complete travel ban. The only common and early responses to the pandemic in Japan, Taiwan, Hong Kong, and South Korea was universal mask wearing and a decrease in large gatherings.

    1. On 2020-06-21 14:24:56, user Robin H wrote:

      Hi. I would like to ask for a clarification. The observations seem favorable to the treatments, notably before statistical adjustment.

      Based on Table 1, some parameters needed obvious adjustments, such as male sex, obesity (26% in HCQ+AZI vs 12% in CTRL), current smokers, hepatic Failure (11% HCQ+AZI vs 4% CTRL), diabetes, asthma or COPD (12% HCQ+AZI vs 7% CTRL), overrepresented in the treated groups.

      Figure S4 shows covariates before and after the IPTW.<br /> But PaCO2 and PaO2 are shown to be disbalanced before IPTW, and so were finally reweighted. However, in Table 1, there is no apparent disbalance for those parameters!

      Am I wrong? Did I miss something?<br /> Since the adjustment "erase" the decrease in mortality for HCQ treatment, we need to be sure that reweighting was correctly done. There are a lot of transformation of the raw datas in this article… And also by a pretty "interventional" causal adjustment. <br /> So it is strange to have this final result after causal adjusment+IPT, when we expect that adjusting for obesity, hepatic Failure, male sex, asthma, etc, overepresented in the HCQ(+/-AZI) patients, would have favored the results for the treatment groups… Right?

      EDIT: In complement, there are other mismatches between Table 1 and Figure S4, peculiarly for HCQ+AZI treatment.<br /> In Table 1, asthma, COPD and obesity are largely overrepresented in HCQ+AZI.<br /> But in Figure S4, those parameters are displayed as pretty balanced before adjustment...

      Asthma is depicted as the most balanced parameter before adjustment in Figure S4, despite an important difference in Table 1 (13.2% HCQ+AZI vs 7.4% CTRL)...<br /> Something seems wrong. Or please, do not hesitate to indicate what I am missing. In this state, I would only consider raw and not adjusted datas.

    1. On 2020-05-24 08:44:09, user Raúl H. Sánchez wrote:

      There is an erratum between lines 146-149 (HLHF instead of ~~HLLF~~ in the stratification).

      The correct statements for the stratification are:

      a. Audiometric group-a: HLHF < 50 dBHL, and HLLF < 30 dBHL.

      b. Audiometric group-b: HLHF > 50 dBHL, and HLLF < 30 dBHL.

      c. Audiometric group-c: HLHF > 50 dB HL, and HLLF > 30 dBHL.

      d. Audiometric group-d: HLHF < 50 dB HL, and HLLF > 30 dBHL.

    1. On 2021-01-23 14:32:57, user Michael J. McFadden wrote:

      You state, "there is reason to believe that there are unknown confounds that may be spuriously suggesting a protective effect of smoking."

      Can you expand a bit on what that reason is? I'm guessing you mean there is evidence pointing to such?

      Also: I have seen seemingly strong arguments made for Carbon Monoxide blood/cell levels as forming the base of this resistance. Do you have any thoughts on that?

      :?<br /> Michael McFadden

    1. On 2021-01-27 16:50:34, user Eric O'Sogood wrote:

      A couple things I noticed. Studies that have been peer reviewed and published with large statistically significant effect sizes are reported here as "no data" or, selectively negative individual outcomes from trials which did have positive effect sizes were chosen. I would be interested more in the source of these authors' methodologies. Standardized, widely validated methods were not used here. Considering Kory, Marik et al's meta-analysis has passed peer review and is accepted for publication, and Dr. Hill and Dr. Lawrie, both experienced systematic reviewers for WHO and Cochrane, came to opposite conclusions to these authors, I would say there is an extremely low likelihood this meta-analysis will pass peer review.

    2. On 2021-02-04 13:18:44, user Daniel Hervas Masip, MD, pHD wrote:

      It is shocking to observe such a big difference between this meta-analysis and others, For example A. Hills group (https://www.researchgate.ne... "https://www.researchgate.net/publication/348610643_Meta-analysis_of_randomized_trials_of_ivermectin_to_treat_SARS-CoV-2_infection/link/6007a57ea6fdccdcb868a4b3/download)"). It also goes against Tess Lawrie meta-analysis preliminary data. The FLCCC members are not exactly a gang of gangsters; they are serious colleagues. It is starting to be very confusing.

    1. On 2021-02-02 03:28:10, user Kenneth Sanders wrote:

      Given the prevalence of individuals with previous asymptomatic infection due to SARS-<br /> CoV-2, is there an implication that all individuals (not already confirmed to have had the disease) should be tested for existent antibodies to SARS-CoV-2 prior to first dose of vaccine? Subsequently, only those naive to SARS-CoV-2 before vaccination would receive two doses.

    1. On 2021-02-02 22:30:54, user Philippe Marchal wrote:

      The authors write "We believe that the large excess mortality seen around the world during the COVID-19 pandemic is robust to the exact model specification". This is clearly false.

      Consider for instance controlling for age structure. See

      https://www.math.univ-paris...

      which is taken from

      https://www.ons.gov.uk/peop...

      At the end of week 24, there is no excess mortality in France, while the graph on p.5 shows a substantial excess mortality. See also Bulgaria and Czechia, which have a substantial *negative* excess mortality at that time. This does not appear in the graphs.

      It should be clear that given the age structure of countries where a baby boom occured after WW2 and given the fact that the mortality rate grows superlinearly as a function of the age, the number of deaths will grow superlinearly in time. A linear regression as used by the authors will not<br /> capture this phenomenon. Thus the related baseline will be lower than the baseline computed by controlling for age structure.

      Another major concern is the way the authors modelize the noise $epsilon$ on p.3. I suppose $epsilon$ should be $epsilon_t$, i.e. the noise depends on time, otherwise this makes no sense. But the model seems to assume that the random variables $(epsilon_t)$ are independent, which is obviously not the case: otherwise, there would be no epidemics lasting more than a week! It is a bit ironic that, in a paper studying a pandemic, the authors use a model that cannot describe the annual flu epidemics.

    1. On 2021-02-04 15:57:58, user JP Monet wrote:

      I know that this is in pre-print, but did someone mention that the description of your Group 1, 2 and 3 are inconsistent in your "Methods" section with the description in the Results/Table? This needs to be clarified or it invalidates the conclusions. " Group 1= SARS-CoV-2 IgG negative healthcare worker (HCW). Group 2= asymptomatic SARS-CoV-2 IgG positive HCW. Group 3= symptomatic SARS-CoV-2 IgG positive HCW. Box plots represent 25% to 75% percentile, with individual dots representing outliers using Tukey’s method (1.5 x IQR)." But in Methods, "Group 1: IgG positive with history of symptomatic COVID-19; Group 2: IgG positive and with asymptomatic COVID-19; and Group 3: IgG antibody negative." In this day in age of misinformation, I would want to see your validated raw data to confirm you conclusions.