Avoid 'deductive' disclosures by revamping informed consent
Qualitative research based on in-depth interviews has the potential to reveal rich details about the lives of participants. As a result, it also has the potential to identify those participants, even in the absence of protected health information such as name, address or birth date.
Social behavioral researchers struggle with how to prevent such "deductive disclosure" – the disclosure of subjects' identities through personal details that others know about them, says Karen Kaiser, PhD, a researcher at the Institute for Health Research and Policy at the University of Illinois at Chicago.
Kaiser says that while studies rarely raise issues of deductive disclosure, when it does happen, it often arises after some data already has been collected, making it difficult to determine how to proceed. She says there is little literature to guide researchers in how to handle the potential for deductive disclosure in qualitative research.
Common solutions tend to focus on manipulating the data afterward, by scrubbing details that might reveal too much or by providing pseudonyms or otherwise altering identifiers.
After dealing with concerns about deductive disclosure in one of her own studies, Kaiser began to examine the issue, and has arrived at an alternative solution that focuses on considering the audience for the research and altering the informed consent process to determine participants' wishes regarding disclosure.1
"By asking (participants) specifically how they want their data used, it gives them a greater opportunity to have a voice rather than leaving it to the researcher to decide behind closed doors," Kaiser says.
The case of 'Rachel'
Kaiser's first brush with deductive disclosure came in a study she did a few years ago on attitudes of women who had undergone treatment for breast cancer. She did extensive interviews with women whom she recruited through a small cancer center.
One of those women, "Rachel," revealed in the course of her interview that she did not participate in the center's support group because she was a lesbian, and didn't feel that she fit in with the other women. Her physicians knew of her orientation.
At that point, Kaiser says, she was unsure what to do with this information, particularly because she had told the center's director that she would present her findings to the doctors and nurses at the center to help them improve services.
On one hand, Rachel's experience could provide valuable insight for health professionals. However, because it was possible that Rachel was the only openly lesbian patient at the center, identifying the woman as a lesbian would likely lead to deductive disclosure by the staff there. The informed consent that Rachel signed specifically said that precautions would be taken to disguise her identity so that she could not be linked to the study.
In the end, Kaiser says, she removed Rachel's information because she couldn't think of a way to present it without identifying her.
She says the quandary she faced was typical of deductive disclosure situations: "It's those situations where someone shares something that a researcher really couldn't have anticipated and now they're left with a dilemma as to how to handle that information."
As a result of her experience, Kaiser began to look at possible ways to deal with such disclosures.
Consider the audience
She says the typical approach relies upon the assurances of confidentiality made during the informed consent process. If situations arise, they're typically dealt with by "cleaning" data (removing identifiers that others might use to link a person to a study) or changing data through the use of pseudonyms for names and places. For example, in writing about this process for a recent article in the journal Qualitative Health Research, Kaiser changed not just Rachel's name but the name of the cancer center where she was recruited.
While this approach generally works, Kaiser says that it may have drawbacks in certain cases.
"There are no studies out there that really have investigated how often researchers remove valuable information from a study to protect confidentiality," she says. "I think it's just in somewhat rare occurrences when really valuable information is lost in an attempt to maintain confidentiality."
To deal with those types of cases, she says, there may be another way to address the need for confidentiality while still including important details. It would require changes to the informed consent process, including a new form presented after an interview that would give a participant choices as to how his or her information could be used.
As part of this approach, the researcher would have to consider all the possible uses of the data – in a journal article, in a poster presentation, in media interviews, or in presentations to participants, physicians, etc. Kaiser says this should happen before the study is approved and patients are recruited.
Typically, researchers and IRBs don't talk about that question much, Kaiser says.
"There are questions about your intended use of data, but they're pretty general," she says. "Will it be published in a journal, shared at conferences, those kinds of things."
She says this is an area where IRBs can help guide researchers by asking questions about all the potential audiences for a study. For example, if a researcher is doing interviews with students, how much descriptive detail will school officials or teachers see?
"They need to be thinking about who the audience for the research is, and whether that audience is also closely connected to the study participants," she says. "Even if it's not highly sensitive – STDs or sexual practices or something like that – if the audience knows the study participants closely, it becomes an issue where they can start to identify study participants and there's consequences of that."
Kaiser says qualitative researchers, especially less experienced ones, could also benefit from being able to contact the IRB for guidance if they uncover potential deductive disclosure in the course of a study.
"If the researcher just can't figure out how to use a very sensitive piece of information, it would be nice to be able to roundtable or discuss that information with an IRB," she says.
Kaiser emphasizes that her alternative approach is not usually necessary and wouldn't be needed, for example, in a general survey, where personal details are not divulged.
"These types of situations are rare. They don't happen often. But they do leave a researcher a little confused as to how to proceed."
Reference
- Kaiser K. Protecting respondent confidentiality in qualitative research. Qual Health Res 2009 Nov;19(11):1632-41.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content