COVID Rx news

I'm not sure what happened to all those promised chloroquine double-blind trials.  Lately I mostly read arguments from doctors on the front lines that they're liking the results of using chloroquine rather early in a case, though they're unsure whether it's helpful when delayed until symptoms are severe.  Unfortunately, this may not mean much except that a patient sample consisting of people who aren't yet very sick are going to do pretty well for the most part no matter what treatment they get, because only a small minority of patients draw the black bean and get horrifically sick.  You really need careful studies to tease out slight gains in the rate of patients who avoid getting much more ill depending on whether they receive a particular early treatment.  This is how we get widespread stories that melatonin, vitamin C, zinc, etc., are working great, because lots of people took them and didn't die.  Nevertheless, I'd still ask my doctor for immediate chloroquine and a Z-pack if I tested positive or showed symptoms, and would cheerfully take the treatment if he prescribed it, because it shows promise of helping and seems to have an awfully small downside.

Meanwhile, there's perhaps a little good news for the much smaller subset of patients who did draw the black bean and are now seriously ill:  Remdesivir appears to help quite a bit, though it's definitely no magic cure.  Unfortunately the recent good-news study was conducted by Remdesivir's manufacturer, so we have to take it with a grain of salt.

9 comments:

RonF said...

The original protocol included a daily 200mg zinc sulfate supplement. I would include that as well.

David Foster said...

Double-blind studies are desirable, but they're not the only was of identifying the effect of a drug or other treatment. Multivariate statistical methods, such as multiple regression, can be used to analyze the effects of multiple factors...in this case, such things as age, pre-existing conditions, and disease progression...and isolate them from the effects of the treatment. It's not perfect...you can never be totally sure that you've identified *all* the things that need to be 'held constant'...but it's a lot better than nothing, and in some circumstances, double-blind tests just aren't feasible or take too long.

E Hines said...

Double-blind studies are desirable, but they're not the only was of identifying the effect of a drug or other treatment. Multivariate statistical methods....

Actually, in drug studies--and in quite a number of other types of studies--double-blind isn't just desirable, it's critical. The placebo effect is real, and patient expectations can heavily impact the outcome of a drug trial--both positively and negatively--so it's important he not know what he's getting. Similarly, experimenter bias, no matter how hard he tries to control it in himself, can heavily impact his interpretation of inherently fuzzy results. It's just as important that he not know what he's administering. Third parties handle the identifications and don't reveal anything to anyone until after the experiment is completed and the data collected and ready for analysis.

Multivariate statistics are highly useful, but the need for them (not the technique itself) has its own serious weakness: overspecification ("overtuning" in the investment world). If too many variables are brought in, the experimenter runs the serious risk of fixing his results to his specific experiment and not allowing their generalization to the wider environment he was trying to estimate.

double-blind tests just aren't feasible or take too long

It's not feasible at all to use the results of a poorly done test, which results are meaningless. Being double-blind adds very little to the time required to run a test. That's just an isolation of who knows what until when.

Eric Hines

David Foster said...

Suppose there is an outbreak of snakebites, most of which are fatal, and the existing remedies don't work. There is a possible new treatment, based on repurposing of a drug with minimal side effects. It is given to everyone who wants it.

If the outcome of snakebites falls from 90% fatal among the population as a whole to 10% fatal among those who have taken the drug...and this ratio is sustained when you correct for age, health condition, etc...wouldn't this be a meaningful result? Would it not justify ramped-up production and wide availability of the product in the area?

E Hines said...

There's a world of difference between repurposing a drug--that's already been through its double-blind trials--where the only question is whether it also works for on a new item and administering a drug that's never been tested before.

Even here, though, another effective treatment is to deal with the snakes directly and reduce the incidence of snake bites. Folks in Australia know about this alternative.

Eric Hines

Aggie said...

Yes the zinc is am important component - it inhibits the ability of the virus to attach and replicate in the cell. Without it, the beneficial effects of hydroxychloroquine are muted.

David Foster said...

E Hines...I"d be interested in your take on this study of hydroxychloroquine and azithromycin effectiveness:

https://archive.is/UIkDh#selection-5001.261-5001.295

E Hines said...

Mr Foster,

I'll leave aside the hype in the Highlights and Abstract, since it's impossible to tell when the thing was written. This claim, though Globally, the overall crude mortality from SARS-COV-2 is estimated to be approximately 6-7% suggests the paper was written some considerable time before it was published. That tends to mitigate the misstatement of the stage we're in, but it doesn't excuse the hype. The delay itself is no big deal; it just illustrates the long period of delays between writing and peer review-publishing.

Some questions immediately arise, though: Patients with a COVID-related admission: What does this mean? A (GP, perhaps) diagnosed SARS-CoV-2 infection drove the (suggested? needed?) hospitalization? Hospitalization with a concurrent, but not driving, SARS-CoV-2 infection? defined as hospitalization during which the patient had a positive SARS-CoV-2 test does not answer the question.

The first COVID-19 case confirmed at HFHS by RT-PCR was on March 10, 2020, any patients admitted before March 10th and subsequently tested positive were also included in the analyses. Did they come in with the infection and only lately tested, or did they pick up the infection in the hospital? Age of the infection matters as much as its severity. Picking up the infection post-admission would go some way to answering the question of why the hospitalization.

For Hydroxy + Azithro, were the doses simultaneously administered or staggered? If the latter, by how much? Drugs interact with each other as well as with the body and the targeted disease. See for instance how much more effective Azithro got in combination with Hydroxy compared with its use alone.

Plainly not a double-blind study; administering according the cardiac risk blows that up. No mention of patient-blind administration, but my experience with hospitals, as a user and as a relative and friend of users suggests that the patients know their meds unless they've agreed a priori to participate in a blind experiment. No mention of voluntary participation.

Indeed, Limitations to our analysis include the retrospective, non-randomized, non-blinded study design. The retrospective nature of the study is minor compared to the others. Nor are retrospectives themselves valueless; such studies can give strong indications of where to go look when other data come available or new ideas of the meaning of existing data arise. Experimenter blindness could have been approached by interposing a third party between the drug administer-ers and the analysts. Tough to backdate techniques in a retrospective, though.

Finally, I continue to be puzzled by the repeated inclusion of PRC results in serious papers. Also, the French study is no longer worth citing--it provided an initial, useful anecdote, but it was only anecdotal--aside from being non-random, it had a much too dinky n for that study to matter anymore. It's fortuitous that the study indicated favorable results, but non-favorable would have been meaningless--absence of evidence, and all that. Lethal results would have warranted further study just like the favorable results did. Now the aggregating data have moved on. So should paper writers.

The results can hardly be taken as dispositive, but they do support conducting proper--randomized, double-blind--studies on the efficacy of hydroxychloroquine and of hydroxychloroquine plus azithromycin. I'm not sure it adds to an EUA or reissuance of an EUA, though, this drug and the two together, already proven safe, have always been in the category of what have you got to lose. There shouldn't be a need for an EUA. Off-label uses have long and easily been authorized; these two drugs for this off-label purpose should be no different.

Eric Hines

Tom Bridgeland said...

Remdezivir does appear to help. At least, we are not seeing nearly as many covid patients do the sharp, fast slide into respiratory distress that we were seeing a while back. Might be the Remdesivir, the Dexamethasone. Heck, it might be the vitamin C we are giving them all now.

We are not currently using Hydroxychloroquine, though we did for a few weeks. No opinion on that one.