Human Subjects Case Unfolds

Institutional Review Boards are in the top news at outlets such as the New York Times, as a research debacle unfolds. I looked through the documents that are publicly available to figure out what happened and what to expect.

Researchers at 22 universities or hospitals in the US enrolled premature babies in a randomized controlled trial between 2004 and 2009. This was the second part of a broader study, but the first part of the study “raised no concerns” according to the US Office of Human Research Protections on page 2 of its determination letter to the lead institution, University of Alabama-Birmingham. OHRP is the federal agency in charge of enforcing human-subjects regulation.

For the second part of the study, though, OHRP found that all 23 IRBs that approved the study (at 22 research sites) violated federal regulations: IRBs should have made researchers tell the parents that they knew their babies would be at higher risk of death, neurological damage, or blindness if they enrolled in the study (pages 2 and 10 of UAB letter). OHRP has only posted a determination letter for UAB at this point, but it explains that at all of the sites, the agency found violations with consent documents “similar to those described” to UAB. The UAB IRB is in especially hot water because it seems first to have first approved the 2.5-page template consent form, which the other institutions used (page 5). If you read the last page of UAB’s letter, you can make a good guess at who may officially be getting bad news from OHRP soon.

In 2010, a year after all 22 sites stopped enrolling babies, the research team published its findings in JAMA. The article showed the seriousness of the gamble the parents had taken without knowing. More than 1000 babies were given either more oxygen or less oxygen than (what OHRP determined) is standard of care:

  • The babies that got low levels of oxygen were significantly more likely to die—but to be able to see if they did live to time of discharge.
  • The babies that got high levels of oxygen were significantly more likely to live—but to be blind.

OHRP almost breaks its bureaucratic cool in the determination letter when it explains why one researcher (who was not directly involved in the study) was especially misguided (page 10). Last year the commentator, Dr. Simon N Whitney of Baylor Medical College, had publicly taken to task the study’s IRBs for requiring consent at all. Whitney had used the study (called SUPPORT) as an example of the “greatest error” of the IRB system, which is that they cause “patients die” by needlessly delaying research, for example by requiring documented consent. Simon describes IRBs as “pythons” – metaphorically strangling science (or perhaps babies?) – because they are too restrictive. Whitney’s commentary ran with an article that was written by SUPPORT researchers and others that OHRP used to as evidence that the clinicians had an incorrect sense of the study’s risks and thus what parents might want–specifically, that the study but babies at “minimal additional risk compared to the alternative clinical treatment” and thus that parents would not “have a preference between the 2 treatments” (page 10, OHRP quoting the team’s article). According to OHRP, this was a wrong-headed “interpretation of facts” (page 10).

This case points to the bind that human-subjects regulations create for IRBs that review multi-site studies. Boards do not want to be criticized yet again for killing patients with red tape. Nor do they want to be called baby killers because they let researchers enroll preemies in research that is life-threatening without telling parents. This case seems to be a tragic and perfect example of how poorly IRBs deal with multi-site studies as they are currently designed.

In my own research from inside IRBs, I found that boards were reluctant to require major changes to boilerplate consent documents for multi-site studies, like the one AUB’s IRB approved and the other sites used, which OHRP said was incomplete in potentially life-altering ways and should not have been approved. The IRBs I observed knew that forcing changes to boilerplate language for a multi-site consent would require that all of the other IRBs involved approve the change, which would delay the study—and put them in the crosshairs of critics like Simon. At the same time, no IRB would want to be on the receiving end of the lawsuits that are no doubt forthcoming from this case. The parents enrolled in the study have yet to be told that researchers did not give them full information.

OHRP has settled many open questions in this particular — and particularly tragic — multi-site study. It remains to be seen, though, how OHRP will remedy the broader problem of multi-study review as it revamps human-subjects regulations this year.

3 thoughts to “Human Subjects Case Unfolds”

  1. Hi Laura,

    According to the NYT: “Babies assigned to a high-oxygen group were more likely to develop eye disease and blindness, the letter said, and babies assigned to a low-oxygen group were more likely to die.”


  2. Thanks, Holly! Indeed, I’d flipped the outcomes of the two arms. I’ve now corrected it. I should note that OHRP got it right (page 11). For the stats-lovers out there, the P value for low oxygen arm (often died, but if lived could see) was point 001. The P value for the high oxygen arm (often lived, but were blind) was point 04.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.