Colorectal cancer (CRC) is a significant and preventable disease, yet CRC screening rates are lower than those for many other conditions. Although the prevalence of CRC screening exceeds 80% across the VA health care system (VA), many Veterans remain non-adherent to CRC screening. Incorporating patient preferences into CRC screening decisions has been identified as one method for increasing adherence in these under screened groups. However the impact of such a preference-based strategy on important outcomes, including screening adherence and cost, is unknown.
The broad hypothesis of the study was that an educational intervention designed to help Veterans clarify their preferences for different CRC screening tests would lead to increased adherence. The framework conceptualized that preference clarification leads to a more activated patient, more informed decision-making and, eventually, to better adherence with screening. The specific aims were: 1) To test the effectiveness of a preference-tailored (PT) vs. standard information (SI) intervention for increasing Veterans' CRC screening adherence 6- and 12-months later; 2) To assess the impact of the PT intervention on informed decision making, knowledge and attitudes toward screening, decisional outcomes, and intention to get screened, relative to standard information; and 3) To conduct a cost effectiveness analysis of the PT intervention for increasing CRC screening across the VA.
A randomized controlled trial of the PT vs. SI decision tool was conducted in the Ann Arbor and Pittsburgh VA healthcare systems from 4/1/08 to 3/31/12. The intervention was developed by the PI working with the Center for Health Communication Research at the University of Michigan. Eligible patients were Veterans between 50 and 78 who were due for CRC screening during the screening period, and who had a non-urgent primary care visit scheduled within the upcoming 4-6 weeks. Those at high risk (e.g., family or personal history of CRC, or life threatening comorbid conditions) were excluded. Providers were informed of the study via regular mail and email, and given the chance to opt out their patients from the study. Potential participants were mailed an introductory letter, and those who did not opt-out were contacted by telephone by the local study research assistant.
The Pittsburgh VA site IRB initially rejected the "opt out" approach described in the grant application and approved by the Ann Arbor site IRB. Thus, for the first 6 months of recruitment at the Pittsburgh site, potential participants were required to opt-in rather than opt-out. This opt-in approach resulted in extremely low enrollment into the study; thus the Pittsburgh site PI (Dr. Ling) was able to work with their IRB to approve an opt-in approach similar to that being used in Ann Arbor toward the end of the first year of recruitment. Enrollment increased considerably at Pittsburgh after this, however we were never able to fully meet our targets due to the slow nature of initial enrollment. Dr. Ling presented a methodologic abstract comparing the impact of opt-out vs. opt-in approaches at the 2012 VA National Meeting (see citation below).
Participants who agreed to be in the study came to their primary care visit about 45 minutes early to meet with the local study RA, complete the informed consent process, and be logged onto the computer which randomized them to the PT or SI condition.
Study measures were consistent with the conceptual framework and based on the Preventive Health Model. The primary outcome variable was participant adherence with CRC screening (using any test modality) 6- and 12- months post enrollment. Due to difficulty meeting enrollment targets, we only collected 6-month adherence data as our primary outcome. The approval for this change was submitted through both local IRBs and discussed with the DSMB during the course of the study. It was deteremined that there were likely to be few differences between 6- and 12-month adherence, making the 6-month collection point reasonable and feasible.
Secondary outcomes and independent variables were collected via telephone survey 3-5 days following the intervention. Preferences for screening were collected following completion of the PT or SI program. Specific values for screening test attributes were collected via computer only for the PT group who received the interactive exercise.
Information in a sub-sample of Veterans not in either the PT or SI groups was also collected to measure "usual care."
To date, we have conducted analyses related to the primary hypothesis (i.e,. that screening adherence rates would be higher in the PT vs. SI group 6-months post enrollment) using chi-square tests and regression to control for other factors (patient age, comorbid conditions, study site). We have evaluated associations between computer generated preferences, participant stated preferences and test received in the PT group. The study team is in process conducting various other analyses, including looking at factors associated with screening uptake in this group of Veterans, factors associated with having preferences for different CRC screening tests, use of CRC screening in older Veterans with comorbid conditions, and the methodologic paper noted above.
We enrolled 468 Veterans from two VAMCs (234 per group), who completed the intervention and for whom we have adherence data from the medical record. Of these, 85% (N=395) completed the follow up survey where secondary outcomes were obtained. The average age was 60, and 95% were male. CRC screening adherence 6-months post-enrollment was 38.6% overall, and was not significantly different between groups (37.4% intervention, 39.7% control). However, rates of CRCS in both intervention groups was higher than in the "usual care" cohort (26.5% of those identified and followed via electronic medical records were screened 6 months after being identified). Adherence was not effected by the intervention (Figure 1). Of enrolled subjects, most did not get tested (51%); more were adherent with fecal occult blood test (FOBT) (22.4%), followed by colonoscopy (16.7%).
The most commonly recommended test by physicians was COL (59.2% of tests recommended), though Veterans most commonly stated a preference for FOBT (60.1%).
The features of tests most important to Veterans were nature of the test (32.8%), effectiveness (27.1%), and risk of complications (17%). Adherence was significantly (P<0.001) higher with the test that the patient selected immediately post-intervention (20.2% vs. 6.9% % for FOBT and 32.6% vs. 7.8% for COL) than with a different test. (Figure 2).
Among intervention/PT subjects (N=234), there was a significant (P<0.05) association between the test the computer suggested from the interactive preference elicitation exercise and the test that subjects chose. However, more subjects chose FOBT when the computer suggested COL suggesting that there may be a preference for a less invasive test among some Veterans. (Figure 3).
In the follow up survey (N=395), intervention subjects more often reported they were clear about what test was best for them (64% vs. 57%, P<.10), and were more confident in making a screening decision (69% vs. 63%, NS). Intervention subjects were also significantly more likely than control subjects to report wanting to make a patient-based decision (51% vs. 37%, P<0.05).
Our results thus far suggest that CRC screening adherence in this population of Veterans was low and the preference-tailored intervention did not significantly improve adherence relative to standard information. However, both web-based programs appeared to improve adherence relative to usual care (no decision tool). These results were similar to those from a non VA RCT of a similar version of the Decider Guider decision tool. These data are important because they suggest that while it is important to incorporate preferences into screening decisions, in reality most patients in both studies chose not to be screened at all. In the context of preventive health and cancer screening decision making, when offered the option to "do nothing" it may be preferable to many patients. Our study suggests that further work to improve adherence with CRC screening in target populations may require more than matching patients to the test that they say they would prefer to receive.
We also found that the preference tailored intervention positively impacted elements of the decision making process compared to standard information. Veterans continue to prefer a non-invasive test (FOBT) despite higher rates of recommendation for colonoscopy by physicians. These results suggest that a standard information decision tool delivered over the web may be as effective as a preference-tailored decision tool on CRC screening adherence in this group of Veterans, yet preference-tailoring may prepare patients for discussions with their providers. Improving communication between primary care clinicians and their patients around CRC screening could result in greater concordance between physician recommendations and patients' preference. This may be one method for improving CRC screening adherence in under screened groups, such as Veterans who have a history of non-compliance with screening. Further work is needed to understand barriers to incorporating preferences into screening discussions during clinical encounters.
External Links for this Project
- Hawley ST. Evaluating a preference-based intervention for increasing CRC screening in the Veterans Health Administration. Paper presented at: VA HSR&D / QUERI National Meeting; 2012 Jul 17; National Harbor, MD. [view]
- Ling BS, Dobrosky SM, Larkin A, Hawley S. Is an opt-in recruitment strategy associated with a higher rate of study enrollment. Poster session presented at: VA HSR&D / QUERI National Meeting; 2012 Jul 18; National Harbor, MD. [view]
- Martinez K, Ling B, Vernon S, Partin M, Hawley ST. The association between patient preferences for colorectal cancer screening and screening receipt in the VA. Poster session presented at: AcademyHealth Annual Research Meeting; 2013 Jun 24; Baltimore, MD. [view]
- Hawley ST, Larkin A, Dobrosky SM, Vernon SW, Partin M, Ling BS. The effect of a preference-tailored decision tool on CRC screening adherence in the VA. Poster session presented at: VA HSR&D / QUERI National Meeting; 2012 Jul 18; National Harbor, MD. [view]