Cherry-picking data was routine practice at Theranos, former lab worker says

At Theranos, lab techs would regularly omit two “outlier” data points from analysis in order to get the company's machines to pass quality checks, whistleblower Erika Cheung told jurors yesterday at the trial of Theranos founder Elizabeth Holmes. 

The process of removing outliers from an analysis isn’t necessarily unusual. But Theranos’ own lab manual did not say how outliers should be identified, so to get the company’s proprietary blood-testing devices to pass quality checks, employees could decide which results to keep, essentially “cherry-picking” data, Cheung said. 

“It was very concerning in a research context because once that translates to a patient setting, it’s giving you a good indication that the system isn’t working reliably enough to feel confident and comfortable in running patient samples,” she said.