By Erica Berry
This article first ran on May 16, 2016 on the Columbia Journalism Review's website.
From the possibility of terrorists using encrypted apps to file-sharing software co-opted for pirated media, the gatekeepers of new technology are constantly confronted with what to do when users approach their tools in unexpected or suspect ways.
Last week, this issue landed closer to home in the journalism field, when ProPublica published “An Unintended Side Effect of Transparency,” an editorial about the site’s “Prescriber Checkup” database. Steve Engelberg, the site’s editor-in-chief, wrote that ProPublica had identified “clear signs” that a significant number of readers were using the database to search for doctors who freely prescribe opioids, raising questions about potential abuse.
Released in 2013, Prescriber Checkup relies on Medicare Part D data to show the prescribing habits of hundreds of thousands of doctors nationwide. The tool has had wide-ranging positive effects: ProPublica reporters have revealed how doctors often prescribe narcotics and antipsychotics in dangerously high doses, and patients, hospital administrators, and law enforcement officials have been able to compare, assess, and regulate doctor practices.
But according to the site’s Google Analytics data, up to 25 percent of database pageviews this year involved narcotic painkillers, anti-anxiety medications, and amphetamines, with many readers finding these pages after reading a “reporting recipe” framed to help local journalists pinpoint doctors who prescribed the highest number of narcotics. Others arrived after conducting web searches like “doctors who prescribe narcotics easily.”
The high readership and web traffic patterns led ProPublica to suspect the database was being used for, as Engelberg wrote, “not-so-beneficial” purposes. So the site, in addition to running the editorial, has added a warning to the database pages on narcotic drugs that links to the Centers for Disease Control and Prevention.
— ProPublica (@ProPublica) May 12, 2016
ProPublica also published a brief story last week about efforts to fight the opioid epidemic. What it didn’t do is pull anything down.
This is a fascinating ethical dilemma handled with great care. https://t.co/S1o9OqT9nG
— Lydia Polgreen (@lpolgreen) May 12, 2016
In an interview Friday, though, the healthcare data journalist Fred Trotter expressed concern that ProPublica, by calling attention to what it found, could trigger its own unintended consequences.
“The news here is that people are seeking opioids through ProPublica,” said Trotter, founder of CareSet Systems andDocGraph. But in his view, the data doesn’t support strong conclusions about motives: While some readers may be looking to support an illegitimate drug habit, others may simply be pain patients struggling to obtain legitimate opioids, given the concerns over abuse. In that climate, he worried that the steps taken by ProPublica might make things harder for people in need of pain treatment.
“If I was a person who was on opioids right now and I was reading this [editorial], I would have a knot in my stomach that you would not believe,” he said. “The number-one concern those people have…is that they will be lumped in with opioid abusers and their pain medication will be cut off.”
In his editorial, Engelberg did acknowledge that some of the people using the database were no doubt legitimately in need of treatment. Asked about Trotter’s concerns, he noted that the data in Prescriber Checkup continues to be fully available. “As far as I can see, we’ve done nothing that would affect the ability of patients hoping to find doctors who will treat their chronic pain,” he said.
Trotter said he believes he and ProPublica both subscribe to the same model of “radical transparency.” Rather than any big divergence in approach, then, the difference may just reflect different assessments about the risks in this context.
In an interview Friday, Engelberg added that he was skeptical of trying technical approaches to address ProPublica’s concerns, like obscuring the database from particular types of web searches. “I think that people who are determined to find things on the web go around you,” he said. “It’s kind of like stopping the ocean. You can build all the castles you want but the tides go where they go.”
He also discussed another possible consequence—that by publicizing its concerns, ProPublica could give a talking point to someone opposed to releasing other data sets.
“I wouldn’t be shocked if in our next argument about making public some certain data, someone in the government said, ‘Well this could happen,’ but I don’t think it’s a terribly valid argument,” said Engelberg. “Everything that you do in journalism when you bring something to light has potential benefits and potential downsides.”
Trotter agreed about that, noting that organizations on the fence about using open health data could “use this as an arrow” to articulate why they will avoid it.
He said he sees open data as a kind of “population health treatment.” In this sense, Trotter said, ProPublica’s recent findings of potential illegitimate use are a side effect, but one that should not distract from the larger good of the service. “If you have a drug that works, but five percent of the time it causes awful headaches, you don’t pull the drug.”