Saturday, May 23, 2020

A Lancet study with over-the-top data tries to sound a death knell for chloroquine, fails

A big news story came out today regarding the results of a Lancet study of chloroquine, hydroxychloroquine and azithromycin in hospitalized Covid-19 patients.  The first author is, naturally, from Harvard.

What the authors found were considerably higher rates of arrhythmia and death in the patients who received a chloroquine drug, with even worse outcomes if patients received azithromycin (Z-pak) too.

Important, smart doctors were interviewed, and they said things like, "Now we know these drugs kill."  "Stop using them, except possibly in a clinical trial setting."

Maybe the Lancet study is giving us the last word on the chloroquine drugs and Covid-19. 

But let me tell you a few things about this study that give me pause.

The retrospective study included 96,000 people, of whom nearly 15,000 received a chloroquine drug.  Now those are really large numbers, so this should be a well powered study.  How did the authors get so much data?  The second author, Sapan S Desai (SSD), founder of Surgisphere Corporation, appears to have provided it.  Which makes me wonder how the 671 hospitals or the 96,000 patients felt about their medical and financial data being used, with no ethical review...
"Acquisition of data and statistical analysis of the data were supervised and performed by SSD...
SSD is the founder of Surgisphere Corporation.
The Surgical Outcomes Collaborative (Surgisphere Corporation, Chicago, IL, USA) consists of de-identified data obtained by automated data extraction from inpatient and outpatient electronic health records, supply chain databases, and financial records. The registry uses a cloud-based health-care data analytics platform that includes specific modules for data acquisition, data warehousing, data analytics, and data reporting." 
The data came from 671 hospitals on 6 continents.  Wow.  And here is just a bit of what the authors tell us about data collection:
"The standardised Health Level Seven-compliant data dictionary used by the Collaborative serves as the focal point for all data acquisition and warehousing. Once this data dictionary is harmonised with electronic health record data, data acquisition is completed using automated interfaces to expedite data transfer and improve data integrity. Collection of a 100% sample from each health- care entity is validated against financial records and external databases to minimise selection bias. To reduce the risk of inadvertent protected health information disclosures, all such information is stripped before storage in the cloud-based data warehouse...The data collection and analyses are deemed exempt from ethics review." 
You harmonise the data, then you improve its integrity.  Wait, what?  The only way to improve the integrity of electronic data is to compare it to hard copies of the data.  I am guessing that what the authors mean is that data points that an algorithm determined were incorrect got changed or dropped.  And financial records were available, too.  How the heck did that happen?  Then Surgisphere stored all of this in the cloud, after de-identifying it. 

Besides the privacy issues (having a private US company get hold of blended medical and financial records from 671 hospitals on 6 continents) is the issue of the accuracy of this data and analysis.  Could the data have been manipulated?  I doubt data accuracy was checked with 671 individual hospitals...who might not have been happy their data were being used... How do you verify the validity of data from so many different sites?

In the UK, 33% of 17,000 hospitalized patients died from Covid-19.  A Chinese study found 28% of those hospitalized died. A US study revealed a 21% mortality rate in those hospitalized for Covid-19. Another US study had a 20.3% mortality in hospitalized patients, but found that those who received chloroquine drugs were sicker than those who did not.

Yet in the Surgisphere Corporation dataset of 96,000 hospitalized patients, only 11.1% of hospitalized patients died.  And in the control group, who did not get any chloroquine drugs, in-hospital mortality was only 9.3%. Mortality in the chloroquine groups ranged from 18% to 23.8%.

I find the data presented in this Lancet article hard to believe.  The mortality rates in the non-chloroquine patients are simply too good to be true.  It is also possible that the chloroquine patients were a sicker cohort. In any event, their mortality rates are in keeping with overall rates in the US, and are better than published rates in the UK and China.  

How did this group of hospitals do twice as well as the US, and 3 times as well as the UK in preventing Covid deaths???

I don't think the debate on use of these drugs is over.

Update May 31:  The Scientist has uncovered a long litany of lies and exaggerations from Dr. Sapan Desai in a detailed investigative piece here.  And I uncovered the odd fact that Dr. Desai wrote a 2013 article titled "Combating fraud in medical research."

No comments: