Surgisphere's phantom registry — two top journals retracted within days

On 22 May 2020, in the middle of the first COVID-19 wave, The Lancet published a blockbuster analysis: hydroxychloroquine, with or without a macrolide, was linked to a higher risk of death in COVID-19 patients. The paper was billed as a multinational registry analysis of 96,000 patients from 671 hospitals — and all of that data came from one source: Surgisphere, a US company run by CEO Sapan Desai.

The paper changed policy within days. The WHO halted the hydroxychloroquine arm of its Solidarity trial, and several countries suspended their own randomised controlled trials of the drug, which had become a political flashpoint after being promoted by world leaders.

Then the data fell apart. A Guardian investigation found inconsistencies in the figures, initially explained as patients wrongly allocated to Australia instead of Asia — more anomalies followed, raising serious questions about the company itself. The authors launched an independent third-party review of Surgisphere with Desai's consent, but the reviewers withdrew when Surgisphere refused to transfer the full dataset, client contracts and ISO audit report to their servers, citing client confidentiality.

The Lancet's editor, Richard Horton, called it "a shocking example of research misconduct in the middle of a global health emergency." The WHO and several countries restarted the trials they had suspended. The Lancet called for institutional reviews of Surgisphere's research collaborations, noting there were "many outstanding questions about Surgisphere" — questions the company never answered.

The Lancet and NEJM retracted within days when Surgisphere refused to hand over the dataset, after WHO and countries had already halted trials on those numbers.

The paper's evidence existed only inside one company: 96,000 patients in 671 hospitals, with no way for editors, reviewers or readers to see the underlying data

The lead authors themselves never held the raw data — they analysed what Surgisphere gave them, so the whole chain of trust ended in a company that controlled every number

Policy moved faster than verification: WHO and countries halted trials within days of publication, before anyone outside the company had seen the data

When the audit finally came, Surgisphere's refusal was decisive — no dataset, no client contracts, no ISO report — and the papers fell within days

When a paper rests on data only one company holds, it rests on trust. Surgisphere refused the audit; days later both papers were gone — and policy had already changed on numbers nobody had seen.

Both journals retracted on 4 June 2020, and the WHO and several countries restarted the trials they had suspended. The Lancet called for institutional reviews of Surgisphere's research collaborations. The episode forced both journals to revisit their review processes and became a defining example of what crisis-speed publishing can cost. The retracted papers continue to be cited years later.