Maybe This Has Something To Do With The Reproducibility Problem?

Recently, the Atlantic had an article about people who earn a living as clinical trial test subjects. Some ‘methodological notes of interest’ (boldface mine):

Intrigued by the promise of an easier way to make money, he enrolled as a guinea pig in a four-week study testing the effects of alcohol on a painkiller drug.

“It was pretty harsh,” he said. Many of the participants became violently ill; Stone vomited while having his blood drawn. The clinic staff told participants to use a bucket rather than the toilet, if possible, so that they could look through the vomit to see how much of the pill had been digested before it came back up. After the first round, Stone said, he began sneaking into the bathroom after each dose and forcing himself to throw up the pill, to stave off the side effects. The staff didn’t catch on, he told me, and he didn’t share his trick with any of the other participants. “I figured I could get away with it if I kept my mouth shut.”

…There are a few things most serious guinea pigs eventually learn. The first is: There’s such a thing as being too honest.

When Robert Helms, a former union organizer and now-retired guinea pig from Philadelphia, began doing studies in the early 1990s, other guinea pigs he knew from around the neighborhood “drilled me on what to say and what not to say,” he said. “Basically, it’s ‘Have you ever been sick a day in your life?’ ‘Has anyone you’ve related to ever been sick a day in their life?’ And the general answer is no.

…“A lot of places will ask if you’ve ever smoked a cigarette even once,” Stone said. “And if you say yes, boom, that’s the end of it. Sometimes you have to skirt around it.”

…Some studies, the well-paying ones, are competitive, and clinics will often admit more people than they need from the phone screen, expecting to cull the herd after the round of physicals. Pros know to avoid alcohol and drugs in the days leading up to the screening. Some of the more cautious ones will also abstain from exercise, out of worry that an increased creatinine level will make it appear as though they’ve been drinking.

Clinics will also ask about the so-called “washout period,” the 30-day window most clinics require in between when a person last participated in a study and when they can begin a new one. It’s a twofold purpose: protect the health of the subjects, and protect the integrity of the data.

There are some companies, like Verified Clinical Trials, that track participation across trial sites; researchers can subscribe to make sure their subjects aren’t enrolled in more than one study at the same time, or haven’t been in one too recently. Often, though, clinics rely on the participants themselves to enforce the washout….

Depending on when it happens, getting caught in a lie can either get a participant bounced from a study, or it can cause a company to scrap the whole thing. “It really jeopardizes the science and the integrity of the data you’re going to gather,” said John Lewis, the vice president of public affairs at the Association of Clinical Research Organizations. “If you go back and you find people who violated that or were not honest … then you’re potentially throwing out their data and maybe the whole study,” at a potential cost of millions of dollars.

As I’ve mentioned before, there are a lot of ways the science can go wrong long before you get to p-hacking.

This entry was posted in Drugs, Statistics. Bookmark the permalink.