Adults ages 48–75 not currently up-to-date with CRC screening (FOBT within the last year, sigmoidoscopy or barium enema in the last 5 years, colonoscopy within the last 10 years) were recruited to the UNC Decision Support Lab (DSL) from our DSL participant registry and through mass media recruiting (newspapers and email listservs). Individuals with a personal or family history of CRC, polyps, or inflammatory bowel disease were excluded from the study. This study was reviewed and approved by the Institutional Review Board at UNC-Chapel Hill.
Eligible patients completed informed consent at the beginning of their study session. They were randomized to view one of two decision aids in the lab: a 5-option version or a 2-option version. We used the random-number generator in STATA to create sequential study assignment envelopes. The envelopes were opened by the research assistant at the beginning of each participant session and determined which version of the decision aid participants viewed.
The development of the full, 5-option version of the decision aid has been described previously . The five screening options included are fecal occult blood test (FOBT), sigmoidoscopy, a combination of FOBT and sigmoidoscopy, colonoscopy, and barium enema. The decision aid contained introductory information about colon cancer and the screening decision, more detailed information about each of the tests, and comparative information for those who wished to decide between different tests. Viewing the entire decision aid, which was on DVD, required approximately 30 minutes. Participants were required to view the introductory portion of the decision aid, which lasted approximately 7 minutes, and were able to view additional information on each test by navigating the DVD's chapter menu.
The second, 2-option decision aid was a shortened (approximately 15 minutes total content) version of the full decision aid that included only the two options, FOBT and colonoscopy, most frequently chosen by patients in our previous study [3, 4]. The required introductory segment for this version was approximately 5 minutes. Participants could then view additional information on FOBT and colonoscopy.
The section comparing the different tests in terms of how often tests need to be completed, preparation for tests, time required for tests, ability of test to find polyps and cancer, discomfort during tests, and chance of complications was included in both versions. Both versions ended in a stage-based assessment of screening readiness .
The full questionnaires are included as Additional file 1. Our main outcomes were differences in screening interest and patient test preferences between versions of the decision aid. As a pilot study, our sample size was based on resources available, not a formal power calculation. This study was conducted as a lab-based study outside of a clinical setting and participants were told to respond to the best of their ability when indicating a test preference. We did not measure actual test completion, so test decisions should be interpreted as hypothetical decisions.
Participants completed questionnaires before and after viewing the decision aid. For screening interest, we used a single item with a 5-point Likert response scale that was assessed pre and post decision aid viewing.
Participant test preferences were also elicited after viewing the decision aid, first without any associated out-of-pocket costs, and then with the following costs: FOBT-$10, sigmoidoscopy-$50, barium enema-$50 and colonoscopy-$200. These out-of-pocket cost levels were based, partially, on estimates of co-payments required in the Medicare program .
We also measured several other secondary outcomes, including knowledge, decision satisfaction, and decisional conflict. We assessed knowledge using 3 questions that were administered pre and post decision aid viewing: 1 point was awarded for a correct response, resulting in scores ranging from 0 – 3. Decision Satisfaction was measured with a 6 item scale by Wills and Holmes-Rovner  after viewing the decision aid; scores ranged from 6 – 30; higher scores corresponded to higher satisfaction. O'Connor's 16-item Decisional Conflict Scale  was also assessed after viewing the decision aid. Scores calculated by adding the total for the responses and dividing by 16, scores ranged from 1 – 5; lower scores were associated with lower conflict.
We also assessed several subjective measures of decision aid content after decision aid viewing, including the amount of information presented, balance, and overall impression. These measures also employed Likert response scales and were drawn from previous studies by our group . Higher scores were associated with better subjective impressions of amount of information on advantages and disadvantages of screening, preparation for discussion with the doctor, and preparation for making a decision. For the subjective balance of the decision aid we asked: Do you think the video was: strongly in favor of screening, somewhat in favor of screening, neither in favor of nor against screening, somewhat against screening, strongly against screening.
We used means and proportions to report our descriptive statistics. To compare the different versions of the decision aid, we used t-tests and Wilcoxon rank sum for continuous measures and chi-square and Fisher's exact tests for proportions. When the results of the non-parametric tests did not differ from the parametric ones, we reported the parametric statistics. Statistical tests were considered significant with a p-value of 0.05 or less, but results that did not reach statistically significant were not necessarily considered clinically unimportant because of our small sample size. Because our small sample size did not permit equal distribution of potential confounders, we also performed multivariate analyses, using linear regression and logistic regression, to account for baseline differences in the intervention and control groups.
The study was approved by the University of North Carolina Biomedical Institutional Review Board.