T-13: Exploring Subjects’ Opinions About Option to Skip Items in ePRO: An Online Survey Study
Jenny J. Ly
Clinical Science Advisor
ERT United States
Missing PRO data can reduce data quality and influence the validity of results. A way to prevent missing data is to design ePRO systems to require a response for each item. This study explored subjects’ opinion about required responses and sensitive topics subjects may wish to not answer.
Participants were recruited via ClinicalConnection and completed an online survey (May-August 2016). Open- and close-ended questions assessed the importance of having the option to skip items in electronic systems, rationale for their choice, and topics that they may be uncomfortable answering.
A total of 216 participants responded to our survey and 155 participants (67.5% female) completed our survey questions. Participants were diverse in age (40.6% age 18-45, 34.8% age 46-60, 24.5% over age 60). 34.0% of participants reported being diagnosed with at least one psychiatric disorder (e.g., depression, bipolar, generalized anxiety, schizophrenia), 24.7% reported at least one non-psychiatric medical disorder (e.g., cancer, COPD, arthritis, or multiple sclerosis), 14.0% reported having both a psychiatric and a non-psychiatric disorder, and 27.3% reported not having a diagnosis.
The majority of participants stated that subjects should be given the option to skip items (63.2% vs 36.8%). Rationale for skipping items (n=81) were recorded in an open ended format. Their responses were then coded into 7 categories and ranked as follows: 1. Not relevant to experience (36.0%), 2. Not comfortable/Do not want to answer (34.7%), 3. Privacy (9.3%), 4. Choices do not reflect answer (6.7%), 5. Do not know the answer (6.7%), 6. Do not understand the question (4.0%), and 7. Questionnaire too time consuming (2.7%). Six responses could not be understood and therefore, not categorized. Rationale differed with age ?2 (6)=15.3, p=.02, where 51.9% of participants age 18-45 endorsed “Not comfortable/Do not want to answer” and 43.8% of those age 46 and over endorsed “Not relevant”.
Most participants (69.6%) indicated they would answer questions about all topics, regardless of whether they endorsed skipping option. Topics participants cited as most desirable to skip were ranked as follows: 1. Sexual issues/history (19.9%), 2. Income (15.9%), 3. Upsetting past experiences (13.9%), 4. Intimate relationships (13.3%), 5. Employment status (6.6%), 6. Family relationships (5.9%), 7. Sexual orientation (5.3%), 8. Ethnicity (4.6%), and Other (i.e., drug or alcohol problems and suicidal ideation or behaviors; 2.6%).
PROs are often primary or secondary endpoints in clinical trials and missing PRO data can introduce systematic bias, which can reduce power to detect treatment efficacy. The FDA PRO Guidance (2009) encourages implementing specific procedures to avoid missing data, including instructions and training to patients and research staff. Despite this recommendation, reviews of randomized controlled trials consistently showed that 90-95% of trials reported missing data (Fielding et al 2008; Bell et al, 2014). Kyte et al (2016) found less than 50% of research staff routinely check for missing data and follow-up with participants to complete missing items. Capturing PRO data electronically can ameliorate this problem by requiring a response for each item and not allow out-of-range, contradictory, or extraneous responses. However, patients’ opinion regarding required responses and topics they may have reservations answering are not known.
Findings from our study showed that while the majority of participants preferred to have the option to skip, the majority also indicated they would answer all items. One of the leading reasons to skip items included the participant being not comfortable responding. Primary topics for wanting to skip questions involved sexual issues/history, income, upsetting past experiences, intimate relationships, and employment status. Thus, when a PRO instrument includes items of this nature, special consideration should be given to the potential value of enabling a skip option for responding subjects. Other reasons for skipping included irrelevant questions, choices that did not reflect participants’ answers, not knowing the answer, and not understanding the question. These results suggest the importance of developing well validated PRO measures with cognitive debriefing, as well as subject training for completing PRO measures to reduce missing data.
Ly JJ, Durand E, Gary S, Gerzon M, Tuller JM, Chan EKH, Jamieson J, McQuarrie K, Dallabrida SM