Current recommendations for offering patients access to individual research results highlight participant choice. Recently, Bacon et al. (2015) evaluated how participants set preferences for genomic results beyond a simple opt-in/opt-out model.1 Using semi-structured interviews, they examined responses from 25 parents whose minor children received inpatient treatment in either the Medical Intensive Care Unit or Intermediate Care Program at Boston Children’s Hospital.
The team used participant feedback to create four adapted preference-setting models, which they piloted until reaching saturation:
Branching diagram model (n=3 parents)
Example-based model (n=4 parents)
Grid model with checklist (n=4 parents)
Step-wise grid model (n=14 parents)
Branching model: The first model comprised a series of branches sorting hypothetical conditions on the basis of age of onset, preventability, treatability and severity. Overall, the participants reported that the model was confusing and that the hypothetical results they received did not resonate with their actual preferences. However, this model did highlight two preference-setting dimensions: preventability and severity.
Example-based model: Using the information from the branching model, the team constructed an open-ended model where the parents could “talk through” 32 hypothetical conditions. Parents discriminated between conditions that can be prevented pre-symptomatically and those that can be treated post-symptomatically, but varied on whether they preferred to be notified of non-preventable conditions. The parents also varied on notification preference for non-severe conditions.
Grid model with checklist: Next, the researchers formulated a 2 x 2 decision grid that incorporated preventability and severity axes ( “non-severe and preventable,” “non-severe and non-preventable,” “severe and preventable,” “severe and non-preventable”) with an opt-out checklist for specific disease categories (e.g., allergies and sensitivities, cancer, heart disease, muscular conditions, neurologic conditions, mental health conditions, and developmental delays). The team established standard definitions for the axes:
“Preventable conditions are those for which, prior to onset of symptoms, there is some treatment, therapy or lifestyle change that can be made to help prevent or avoid symptoms, or reduce risk.”
“Severe conditions are those which may be fatal or which can cause significant pain, discomfort, or negative impact on quality of life.”
The addition of concrete definitions increased understanding compared to the previous models. However, many participants indicated that the model was cluttered and unsuitable for online navigation. No participant opted out of an entire category, but parents showed strong reactions to the following categories: mental illnesses, developmental conditions and adult-onset conditions. Almost every parent declined results from at least one of the following, citing complexity and stigmatization: attention deficit hyperactivity disorder, dyslexia, alcoholism and amyotrophic lateral sclerosis.
Step-wise grid model: Finally, the team revamped the 2 x 2 decision grid using a step-wise format with opt-out for disease categories that yielded strong reactions:
mental illness and psychological conditions
developmental disorders and learning disabilities
childhood-onset degenerative neurological conditions
Overall, parents reported the most accurate concordance between stated preferences and hypothetical results with this model. One participant’s responses inspired the team to add a fourth category to the opt-out conditions: adult-onset conditions with no interventions during childhood.
Applying the concepts of preventability and severity to the data derived from all four models, the research team observed that all participants preferred notification of preventable conditions, and over half (64%) preferred notification of both preventable and non-preventable conditions. These parents cited empowerment and the opportunity to take action via preventive therapies, lifestyle changes and behavioral interventions, as well as the ability to prepare themselves and their children for potential future diagnoses. The minority of parents (36%) who refused notification of non-preventable conditions did so to avoid unproductive anxiety.
For the severity axis, most parents (76%) preferred notification of both severe and non-severe conditions, with smaller groups preferring notification of only severe or non-severe conditions (16% and 18%, respectively). Parents who declined to be told about severe conditions cited psychological burden and anxiety as well concern that the results could change their expectations and relationships with the child. Parents with children previously admitted for chronic or genetic conditions demonstrated a higher tolerance for the concept of “severity” compared to those with limited experience.
Bacon et al. also made the following observations:
Familiarity with disease: Parents reported strong feelings about receiving or declining results based on family history. These reactions were divided between preferring results on familiar conditions or refusing results on the basis of superfluity or potential harm.
Autonomy: Parents reported mixed responses about whether receiving a child’s genetic results was a parental obligation or a violation of the child’s autonomy (and right not to know).
Previous experience with genetic testing: Parents with previous exposure to genetic testing better identified their own tolerance of uncertainty and made faster decisions regarding preference-setting. Parents with less experience most benefited from the grid structure.
Research characteristics: Most parents rated disease characteristics (preventability and severity) as more influential than research characteristics (degree of risk and clinical validity).
The team distilled their findings into three basic parental subgroups: information seekers (who prefer all results), action takers (who prefer preventable results) and worry avoiders (who prefer not to receive results regarding severe conditions). They note that the way that conditions overlap within categories may have influenced reported participant preferences.
1. Bacon, P.L., et al. (2015) “The development of a preference-setting model for the return of individual genomic research results,” Journal of Empirical Research on Human Research Ethics, 10(2) (pp. 107–120), doi: 10.1177/1556264615572092