Surgisphere, the Indian-American founded tiny data firm at the centre of HCQ-Lancet storm
Science

Surgisphere, the Indian-American founded tiny data firm at the centre of HCQ-Lancet storm

Sapan Desai, a 41-year-old surgeon, is the founder of the analytics company that prompted the retraction of two scientific articles, including one that spurred WHO to pause its drug trial.

   
Representational Image | Source : John Phillps | Getty | Bloomberg

Representational Image | Source : John Phillps | Getty | Bloomberg

Paris/New York: The founder of Surgisphere Corp., a data analytics firm based outside Chicago, quoted Sun Tzu’s Art of War in remarks to Utah’s Western Governors University.

“Opportunities multiply as they are seized,” Sapan Desai told fellow graduates in 2012. Desai, who obtained an MBA and also holds a medical degree and a Ph.D., sought to apply the Chinese strategist’s advice to his own career.

Now the 41-year-old surgeon and entrepreneur is at the center of a controversy with global health implications. On Thursday, doubts about data from Surgisphere prompted the retraction of two scientific articles, including an influential study in The Lancet that had shown antimalarial drugs promoted by President Donald Trump could be harmful in the treatment of Covid-19.

The World Health Organization had suspended clinical trial programs using hydroxychloroquine following the initial report, as did the U.K. and France. The Lancet’s peer-reviewed study, published on May 22, claimed to have analyzed Surgisphere data collected from almost 96,000 Covid-19 patients across the globe.

The Letter

But in the days after publication, concerns over the underlying data bubbled up. Questions arose over how Surgisphere, a little-known company that claims to have 11 employees, could have reached agreements on sharing sensitive patient information with some 1,200 hospitals around the world, much less received and processed the data so quickly.

Last week, more than 200 scientists signed a letter to The Lancet asking for greater transparency regarding the hospitals where patients’ medical records came from and the method of analysis, citing a list of inconsistencies and anomalies in the paper.

Late Thursday, co-authors Mandeep Mehra, Frank Ruschitzka, and Amit Patel requested the paper be retracted after Surgisphere declined to submit its full dataset and other information to independent peer reviewers because of client and confidentiality agreements.

“Based on this development, we can no longer vouch for the veracity of the primary data sources,” the authors said.


Also read: WHO to resume HCQ trial for Covid it had paused after controversial Lancet article


Rush for Data

In a statement, The Lancet said many outstanding questions surround the company and the data used in the study. “Institutional reviews of Surgisphere’s research collaborations are urgently needed,” it said.

Shortly after, the New England Journal of Medicine retracted a separate study that relied on the company’s data.

The scrutiny has placed Surgisphere and Desai, who was listed as a co-author on both of the articles, under a microscope, and also raised questions about research standards during the coronavirus pandemic.

“There is an almighty rush to understand this new disease — everybody is trying to get data quickly,” said Nicholas Day, a professor of tropical medicine at the University of Oxford’s branch in Bangkok. “All the journals are desperate to publish because there is a thirst to know about this disease. Therefore mistakes are made, stuff is rushed through.”

Desai and Surgisphere declined to comment on the retractions through a public relations firm.

Machine Learning

In emailed remarks before the articles were retracted, however, Desai defended his 12-year-old company, which consolidates medical records from around the world. He told Bloomberg the firm relied on artificial intelligence and machine learning to automate its processes, “which is the only way a task like this is even possible.”

“It is important to understand the nature of this database,” he added. “We are not responsible for the source data, thus the labor intensive task required for exporting the data from an EHR (electronic health record), converting it into the format required by our data dictionary, and fully de-identifying the data is done by the healthcare partner. Surgisphere does not reconcile languages or coding systems.”

No artificial intelligence or machine learning experts were listed as authors of the now-retracted Lancet paper. The article also indicated that of the four authors named on the study, Desai was the person who acquired and analyzed the data.

Headquartered in a residential house overlooking a small lake in an upscale area west of Chicago, Surgisphere has two branches that it says collect data and run machine learning software: QuartzClinical and Surgical Outcomes Collaborative.

Desai held positions in at least four medical centers following his Duke University surgery residency, according to his LinkedIn profile, and left his most recent position as a doctor at Northwest Community Hospital in Illinois in February.

Documents obtained by Bloomberg and filed by Desai in Illinois show him to be the sole owner of Surgisphere as of February. He didn’t respond to a query as to whether the company has a board or a scientific committee.

On Thursday, three employees were registered on LinkedIn as actively working at Surgisphere: a sales manager, a franchise representative and Desai himself. By Friday, following the retractions, only Desai remained.

A Sofia, Bulgaria-based computer scientist named Ivo Gelov, who told Bloomberg he had done part-time freelance programming for Surgisphere two years ago, said he had “no clue about their access to medical data.”


Also read: Modi govt to study WHO decision to halt HCQ trial, expert panel to review dosage


Data Questions

Scientists have highlighted potential inconsistencies in the data used in The Lancet study, from an unrealistically high number of electronic patient records in Africa to the doubtful origin of European figures given the continent’s strict rules around health privacy. Moreover, the dataset had more patients than would appear likely given the dates and progress of the virus, especially in the U.K. The Lancet last week published a brief correction of data from Australia.

Desai said the official figures “could have been under-reported early on during the pandemic, thus leading to the appearance that we are over-reporting numbers when in actuality we are capturing the true total number of Covid-19 infections at the hospital level, which is the true source for this data.”

Still, the Surgisphere studies were highly unusual in that they claimed to quickly assemble data from hundreds of anonymous hospitals, using numerous electronic medical records systems, under different privacy laws across many countries on multiple continents. And even more strangely, for studies that claimed a massive feat of data integration in record-setting time, they had no biostatisticians listed as authors that might have helped pull all this data together.

Health Records

More typically, when medical scientists do such studies they rely on clearly named and reputable government databases in one country or state that researchers are able to access.

Surgisphere said its information comes from “a registry, with data obtained from electronic health records” of a “very specific group of hospitalized patients with Covid-19.” The company “directly integrates with the EHRs of our hospital customers,” and “has permission to include these hospitals’ EHR data in its query-able registry/database of real-world, real-time patient encounters.”

Surgisphere didn’t provide the names of companies or institutions from which it obtained the data.

The retracted study published May 1 in the New England Journal of Medicine claimed to have records of 107 patients from five hospitals in France, including ethnic information such as skin color. However, it’s unlawful to collect such data in France. What’s more, the transmission or sale of hospital data and patient records are strictly limited, and often require approval by the CNIL privacy watchdog. The CNIL told Bloomberg it had not received requests from Surgisphere. The French health ministry didn’t immediately return an email seeking comment.

Scotland’s NHS

Surgisphere says on its website that it’s worked with Scotland’s National Health Service to find data-driven solutions to high rates of post-surgical complications and infections. No such relationship exists, the health agency said. “At no point have Surgisphere had any access to NHS Scotland data,” it said in an email.

As for Desai, in his remarks to Western Governors University, he paraphrased Britain’s wartime prime minister, Winston Churchill, on the importance of perseverance: “It is the challenges that teach us to never, never, never give in, to keep pressing forward no matter how hard it gets.”- Bloomberg


Also read: Every scientific success is built on a history of failures, Covid-19’s cure could be too