Hydroxychloroquine After Action Report
I was a vehement advocate of prescribing hydroxychloroquine (HCQ) off label while waiting for the results of clinical trials. I wasn’t all that much embarrassed to agree with Donald Trump for once. Now I feel obliged to note that my guess was totally wrong. I thought that the (uncertain) expected benefits were greater than the (relatively well known) costs.
The cost is that HCQ affects the heart beat prolonging the QT period (from when the atrium begins to contract to when the ventrical repolarizes and is read to go again). This can cause arrhythmia especially in people who already have heart problems. I understood that one might argue that all people with Covid 19 have heart problems but didn’t consider that argument decisive (I probably should have).
The positive expected value of the uncertain benefits was based on strong in vitro evidence that HCQ blocks SARS Cov2 infection of human cells in culture. (this is a publication in the world’s top general science journal).
Already in early May, there was evidence that any effect of HCQ on the rate of elimination of the virus must be small. In this controlled trial conducted in China, the null of no effect is not rejected. Much more importantly, the point estimates of the effects over time are all almost exactly zero. I considered the matter settled (although the painfully disappointed authors tried to argue for HCQ and that their study was not conclusive).
There are now four large retrospective studies all of which suggest no benefit from HCQ and two of which suggest it causes increased risk of death. I am going to discuss the two studies most recently reported.
One is a very large study (fairly big data goes to the hospital) published yesterday in The Lancet. In this study patients who received HCQ had a significantly higher death rate with a hazard of dying 1.335 times as high. The estimate comes from a proportional hazard model with a non parametric baseline probability and takes into account many risk factors including crucially initial disease severity. It is also important that only patients who were treated within 48 hours of diagnosis were considered.
I am, of course, dismayed by this result. I am also puzzled, because it is quite different from the result obtained in a smaller retrospective study published in JAMA
I think the practical lessons are that it seems unwise to give Covid 19 patients HCQ. Also maybe Robert Waldmann should be more humble. After the jump, I will discuss the two studies in some detail and propose an explanation of the difference in results.
To but my conclusion at the start, I think the difference in results occured because the JAMA paper used a more flexible specification to correct for differences in initial disease severity. The Lancet article reduced all information on severity to two indicator variables so patients were classified into one of 4 possible groups. This leaves out most of the data on disease severity. I think it likely that the gigantic data set could not have been assembled keeping track of more variables without the loss of many data points.
OK so Lancet massive study has an estimated increased hazard of dying a given day of 31% of the baseline hazard.
The New York specific JAMA study has an insignificant estimated change in the hazard of dying and point estimate close to one of the hazard ratio.
In the primary analysis, following adjustment for demographics, specific hospital, preexisting conditions, and illness severity, no significant differences in mortality were found between patients receiving hydroxychloroquine + azithromycin (adjusted HR, 1.35 [95% CI, 0.76-2.40]), hydroxychloroquine alone (adjusted HR, 1.08 [95% CI, 0.63-1.85]), or azithromycin alone (adjusted HR, 0.56 [95% CI, 0.26-1.21]), compared with neither drug
First I have to note that the Lancet point estimate is within the 95% interval estimated in the JAMA paper. The difference in results is striking, but not statistically significant at conventional levels. The difference between significantly greater than 1 and not significantly greater than 1 is not, itself, necessarily significant.
But I also note the coincidence that in the JAMA study the raw death rate was markedly and statistically significantly higher for patients treated with HCQ than for patients who weren’t.
“In unadjusted analyses, significant differences in in-hospital death were observed across the … hydroxychloroquine alone (n = 54, 19.9% [95% CI, 15.2%-24.7%]), … and neither-drug (n = 28, 12.7% [95% CI, 8.3%-17.1%]) groups ”
The combination of a significantly greater raw death rate and a near zero estimated effect on the risk of death from the model is due to the fact that patients treated with HCQ were sicker than patients not treated with HCQ. It isn’t surprising that an untested medicine was more likely to be used in more alarming cases. This makes the specification choices made in multiple regression to handle confounding variables critical. In particular, the choice, in both studies, to reduce continuous variables such as oxygen saturation to indicator variables for ranges must matter.
I was struck that, in the Lancet study, blood oxygen saturation was used in the regression as an indicator of saturation less than 94%. The modified variable clearly does not contain all the available information about oxygen saturation. Data on the mental status, respiratory rate, and systolic blood pressure (and other variables which were not described in the article) were summarized with a single index which was replaced with an indicator variable which took two possible values.
I think that Doctors and economists take statistics seriously in different ways. The data collecton ranges from impresseive (JAMA) to amazing (Lancet) with great care in sample selection. However, the specification is not well motivated and robustness to different specifications is not considered at all in the Lancet article (unless I missed it). Since there is a dramatic shift from the raw estimates to the estimates when confounding variables are considered, it is clearly necessary to be careful about how confounding variables were handled.
I salute you for openly admitting your mistake. I thought your advocacy was premature and poorly justified at the time, but since Angry Bear is not at the frontlines of COVID-19, I considered it an interesting exercise in prophecy. I hope you are chastened, but I look forward to your future posts.
So, is the Orange Baboon taking hydroxychloroquine or not?
Thank you for being willing to change your mind in the light of new evidence. Not everyone does that, and even fewer are willing to admit it. I was one of the commenters who ragged you about the difference between promising drug leads and useful drugs. Drug development is a humbling and nightmarish business, and when there is a disease out of control any possible lead looks attractive.
Joel, I think that response was particularly ungracious. Do you always dig in the knife when someone admits they were wrong? And
does doing so encourage people to admit their errors? rather, no, it does not.
And for whether his advocacy was a mistake does not really depend on whether his guess was right, but upon how reasonable it was given the information available at the time and judgements of risk and necessity.
I believe Robert was being complimented by a peer of a different discipline of authority. Knowing them both by what I read and backgrounds, I find Joel’s comment complimentary of Robert. There are so many of us here who come from varied backgrounds discussing issues and scenarios that we may not have an expertise in but have questions about. I think it is important we place our beliefs in the open for review. We may not agree in the end; but, we are bound to review the basis for what we believe.
In this case, I do not agree with either for various reasons; but, I am willing to listen and wait and see how this evolves.
the broader lesson here is to always question yourself when you find you’re in agreement with Donald Trump..
I led with a compliment. I used no knives. We apparently have different ideas of what graciousness is. Perhaps instead of attacking me, you might direct your future comments to the subject of the post, like I did.
I consider Joel’s comment to be very civil. However, while he might wish I were chastened it is not wise to hope for that.
In fact, I continue to advocate expanded access to pharmaceuticals while phase 3 trials are ongoing. In particular, I advocate expanded access to favipiravir except in China and Japan where it is approved for influenza. I advocate off label prescription if favipiravir there.
One warning. If Remdesivir is available, it is necessary to check the safety of the combination before combining them.
I also look with hope at the phase II trial of merimepodib and remdesivir. If it suggests safety and efficacy (a phase II trial can’t approach proof of either) I will advocate expanded access to merimepodib too.
Finally I am very very actively advocating off label prescription of hydroxychloroquine in a specific case of a disease which is not Covid 19, malaria, lupus, or rheumatoid arthritis.
As neither a doctor nor an economist I still find this an academically interesting conversation about statistics.
“I understood that one might argue that all people with Covid 19 have heart problems”
As alluded to by the quoted passage, I wonder if the statistical problem here is that hospitalized Covid-19 patients are not representative of the population. Most people who get Covid-19 do not end up in the hospital. If there is a correlation between those who need hospitalization and those who have mal-effects from HCQ, then normal statistical usage is inadequate.
The hopeful in vitro impact of HCQ does not carry forward when the actual patient population is a factor. Conditional probability messes with human expectation in non-intuitive ways.
If one looks at the dosages being applied in these studies, they are well above what some countries have been using to treat Covid. Remember too, HCQ is the ionophore and not the drug/med which prevents replication. The same as Remdesivir, it must be used early on in the virus prediction.
Joel, led with a compliment, finished with a dagger, is how I saw it. But if you deny it, and Robert says you were civil then I rescind my comment, given the fundamental ambiguity a written communication, and apologize.
“The World Health Organization says it is temporarily halting its clinical trials that use hydroxychloroquine to treat COVID-19 patients over published concerns that the drug may do more harm than good.
The move comes after the medical journal The Lancet reported on Friday that patients getting hydroxychloroquine were dying at higher rates than other coronavirus patients.”
My Mrs. is very familiar with this drug, having administered it for Lupus and RA.
She would not take it for preventative purposes. Even if recommended.
Nothing like experience.
The initial data from China and France was statistically highly significant for a positive effect from hydrochloroquine. The large early 1990’s large British Leicester study (LIMIT 2)using magnesium sulfate (costing pennies) to reduce death in coronary artery disease in the 90’s had a p value of .001-.009. LIMIT 2 was later refuted by a larger US ISIS study. The utility of medical procedures and pharmaceutical efficacy, as determined by the inconsistency of prospective blinded scientific studies, makes Medicine at best a questionable science as compared to physics and chemistry, although we try. Big Pharma makes really big profits on these ‘scientific studies.’ Visit the blog site NNT and your eyes will be open to modern medicine as a science.
On the other hand, Asset Debt Growth and Decay Asset Debt Macroeconomics is a highly patterned self-assembly science…. easily observable to Economists, Physicists, Chemists, and nonscientists.
The Observational and Empirical Case for Observational Quantitative Asset Debt Growth and Decay Saturation Fractal Macroeconomics as a Science akin to physics, chemistry, and biology…
The data that follows offsets the immediately assumed possibility of quackery.
From the December 2018 Composite Equity nadir valuation low: x/2-2.5x/2-2.5x/1.5y :: 11/26/26/15 of 16 weeks : on a daily basis nonlinearity can be observed between the 22nd and 23rd week of the second 26 week fractal. (see main page regarding second fractal nonlinearity)
This correlates to a 3/7/7/4 of 5 month fractal series of similar x/2-2.5x/2-2.5x/1.5-1.6y proportionality. On a daily basis for the CRB, the fractal progression is 5/11/10/4 of 7 days. A 1987 like collapse is expected over the next three trading days.
This web site makes the observation that the asset debt economic system is mechanistic and quantitative in its nature of its hourly, daily, weekly, monthly, and yearly composite asset class valuations following simple growth and decay fractal valuation patterns so precise that ‘the mathematical laws’ and ‘self assembly’ of asset valuation growth and decay are similar to physics and chemistry and biology.
Asset Debt Saturation Macroeconomics likewise has the quality and property of a science.
The simple ever recurring and easily observed quantitative fractal ‘mathematical laws’ determined by the nadir asset valuation are:(y connotes final valuation low for the individual fractal series pattern)
and x/2-2.5x/1.5 to 2.5y
(the second fractal length of 2-2.5x determines the ideal base first fractal length; the third fractal is a 1.5 multiple of this ideal base.)
Qualitatively, the facilitated creation of excessive debt leads to overvaluation, overproduction, and over-ownership of assets. The system is self correcting with liquidation of bad debt and a lower re-equilibrium of asset valuations with a lower total denominator of composite system wealth near the nadir of bad debt liquidation and lower asset composite valuation.
At any given time period, all individual asset valuations are denominated in first time derivative of the composite total worth of all other asset valuations.
The fractal mathematical laws of the composite asset valuations of the asset debt system are elegantly simple.
While global central banks’ interventions can cause observational rises of subfractal components, the fractal grouping patterns are still easily observable.
In fact these observational patterns show the direct effect of central bank intervention.
The US Hegemonic Asset Debt Macroeconomic grand Fractal series had an initiating fractal base of about 18 years near the initiation of its constitution in 1790.
The First Fractal started in 1807-8 and ended after the panic of 1837 in 1842-43 for a base fractal of 36 years. Its 90 year Second Fractal ended with nadir composite equity valuations in 1932. The US 89 year Third Fractal composed of two subfractal series of 51 years and 39 years and is expected to end very shortly (three trading days) in 2020. A fourth fractal is expected to end in 2074. (1.5y) The US 54 year fourth fractal will be supported with necessary debt creation.
A Look at the 1982 second subfractal series: 9/20/12 year :: x/2-2.5x/1.5y concluding US 1932 third fractal series:
The monthly fractal progression of US composite Equities from the low in 2003 was made of two fractal series: 6/13/15/10 months :: x/2-2.5x/2.5x/1.6y and a decay fractal of x/2-2.5x/1.5y : 9/20/12 months: The ideal base of a second 20 month fractal is 8 months with 1.5 times 8 months yielding a 12 month third fractal.
What was the composite equity and CRB valuation fractal effect of the global Central Bank intervention on the 2008-2009 collapse? The 2/5/5/3 month fractal series composing the 12 month third decay fractal begins a valuation climb in March 2009 at the beginning of its third 5 month fractal.
Note the x/2-2.5x/1.5y fractal similarity of the 1982 9/20/12 year fractal series (completing the 89 year US Third Fractal) to the 9/20/12 month fractal series completing the second 20 year subfractal series which started in 1990.
Sans global central Bank coordinated intervention, the expected unassisted starting point for the observed March 2009 composite nadir was at the end of the 2/5/5/3 month natural self assembly fractal series or September 2009
From the expected September 2009 low (unassisted by Central Bank assumption of toxic debt and collaborative interCentral bank money printing and interbank borrowing), the two monthly subfractal series – 2/5/4/3 and 3/7/8 months :: x/2.5x/2x/1.5y and x/2-2.5x/2-2.5y, respectively – make up a 26 month base first fractal sequence of the final 12 year third subfractal.
The final 12 year third fractal sequence of the 1982 9/20/12 year :: x/2-2.5x/1.5y decay fractal series (this second fractal subseries follow a 1932 10-11/21/21-22 51 year first fractal subseries ) is composed of 26/53/52 of 53 months. (x/2-2.5x/2-2.5y)
The second 53 month subfractal of the 26/53/52 of 53 series is composed of two fractal subseries 3/7/6 months and 8/17/17 months (x/2-2.5x/2-2.5y – both subseries)
The third 52 of 53 month series is composed of 10/26/18 of 19 months. The integrative final series is 10/25/20 months)
The first 10 month fractal is composed of a 2/4/4/3 month series; the second 26 month fractal is composed of a 5/11/11 month series, and the third 19 month series a 3/7/7/4 of 5 months series.
The patterned asset composite valuation activity of the Asset Debt macroeconomic system is directly observational and is indisputable. What causes the ideal self assembly of mathematically precise fractal asset valuation growth and decay patterns?
What causes the mathematical laws and derived numerical constants of physics and the naturally occurring self assembly of subatomic particles, atomic particles, molecules, plant and animal embryological development, stars, solar systems, galaxies and the universe?
The observational self assembly highly patterned fractals defining the counterbalancing growth and decay of valuations of composite assets composing the asset debt macroeconomic system confers upon that macroeconomic system the properties of a science.