As patients get older, acquire health problems, and undergo negative colorectal cancer (CRC) screening tests, the benefit of screening decreases and the potential harm of screening increases. Yet, existing population-centered efforts fail to adequately inform older adults about this changing balance in benefit and harm, often yielding screening utilization that is discordant with benefit.
Aim 1: Evaluate the impact of a 3-part Veteran-centered intervention on the frequency of CRC screening orders in a cluster-randomized controlled trial
Aim 2: Perform a process evaluation to identify barriers and facilitators to implementation of Veteran-centered CRC screening
Aim 3: Develop an implementable, Veteran-centered performance measure of CRC screening
We recruited average-risk screen-due Veterans aged 70-75 attending a primary care visit at 2 VA facilities (Ann Arbor VA Medical Center and Toledo Community-Based Outpatient Clinic) from fiscal years 2016-2018. The intervention group was cluster randomized (cluster = primary care provider) to receive either a simple screening informational booklet (control) or a decision aid booklet (DA) (intervention). The DA contained detailed information on screening benefits/harms that were personalized for each subject based on age, gender, prior screening, and comorbidity burden (using data from a validated microsimulation model developed by our team). Both the intervention and control groups received primary care provider (PCP) education and performance measurement and clinical reminder modification to allow Veterans and PCPs to make informed decisions about screening (including the decision to not be screened).
Aim 1: The primary outcome was whether screening was ordered within 2 weeks of the clinic visit. Data were analyzed using a multi-level clustered (site and PCP) logistic regression model, with a pre-specified interaction analysis examining whether the DA had a differential effect by magnitude of screening benefit. The secondary outcome was completion of screening at 6 months.
Aim 2: We conducted semi-structured interviews of Veterans and PCPs as well as audio-recordings of clinic visits. For interview analysis, we used a combination of deductive and inductive coding and, using matrices, identified patterns of similarities and differences across cases. For visit audio analysis, we quantified time spent discussing CRC screening and coded the content of the discussion for elements of informed decision making (IDM).
Aim 3: We presented measure concepts to an advisory panel of Veterans and to an expert audience both locally (at the VA Center for Clinical Management Research and University of Michigan) and nationally (through the National Quality Forum). We also developed several candidate measures using data from the intervention arm of the study (i.e., from well-informed patients), with the analysis following the approach in Aim 1.
Aim 1: 1,562 Veterans were eligible in the 30-month RCT period. 434 completed the study visit across 66 PCPs. 261 (60%) were randomized to intervention and 173 (40%) to control. 81% had undergone prior screening. The mean benefit of screening (CRCs prevented per 1,000 individuals screened) was 11.2, standard deviation 8.2. Screening orders were placed for 164/261 (63%) intervention versus 114/173 (66%) control patients (p=0.57). In our pre-specified interaction analysis, low-benefit intervention patients were less likely to receive screening orders than low-benefit controls; high-benefit intervention patients were more likely to receive orders than high-benefit controls (p=0.047). At 6 months after the study visit, 201 (46%) had completed a CRC screening test - 41% in the intervention group and 55% in the control group (p=0.03) - with the reduction in screening use largely confined to the lowest benefit patients.
Aim 2: We interviewed 31 patients from the intervention arm of the RCT. We categorized participants according to whether the DA: (1) influenced them to stop screening; (2) changed how they thought about screening but not their screening preference; or, (3) had no effect on how they thought about screening nor their screening preference. Veterans influenced by the DA to stop screening learned that CRC was slow-growing and that shortened life expectancy decreases screening benefits. Veterans in the second category understood that risk increases with age and, for some, that CRC is slow-growing. However, this information was mitigated by their perception of themselves as not old enough to stop screening and DA reinforcement of their preference for fecal immunochemical test (FIT) due to its low likelihood of harm compared to colonoscopy. Veterans for whom the DA had no influence had other health problems that they prioritized over CRC decision-making and/or had strong existing screening preferences. 53 PCP visits were recorded, 35 (66%) in the DA arm and 18 (34%) in the control arm. Discussions in the DA arm scored higher than control in 7 of 8 Informed Decision Making (IDM) elements and had more total IDM elements on average: 4.18 DA vs 3.32 control (p=0.07). Average time discussing CRC screening was ~ 2 minutes longer in the DA arm (p<0.0001). We interviewed 7 PCPs, 5 of whom were assigned to the intervention arm. Most PCPs valued that patients were more informed and prepared for a discussion regarding the potential harms and benefits of screening. Some voiced that they would have liked a copy of patient's personalized graph in advance of the visit.
Aim 3: Most participants of the Veteran advisory panel did not know that their PCPs are "graded" through performance measures or that some of these measures can affect performance pay. Many agreed that Veterans should be consulted on how their PCPs are "graded" and that process measures such as IDM should be considered rather than simply measuring screening completion. Experts agreed that individualizing the target rate offered a novel and implementable strategy for more nuanced measurement. They also suggested that patients and experts be consulted in a single forum rather than in separate forums.
Reducing use of low-value care and increasing use of high-value care is a high-priority topic for VHA. The results of this study suggest that Veterans presented with detailed, individualized information about screening benefits/harms (available at screeningdecision.com) are more likely to make a screening decision that is concordant with benefit than those who are not provided with such information. Thus, engaging older Veterans in decisions about cancer prevention can result in more appropriate and efficient use of resources in a way that is aligned with preferences and net benefit. Moreover, this can be accomplished in a way that is acceptable to PCPs with minimal provider burden. Lessons learned from this work may ultimately provide a model for delivering preventive care in a more Veteran-centered way.
- Veenstra CM, Abrahamse P, Wagner TH, Hawley ST, Banerjee M, Morris AM. Employment benefits and job retention: evidence among patients with colorectal cancer. Cancer medicine. 2018 Mar 1; 7(3):736-745.
- Caverly TJ, Kerr EA, Saini SD. Delivering Patient-Centered Cancer Screening: Easier Said Than Done. American journal of preventive medicine. 2016 Jan 1; 50(1):118-21.
- van Hees F, Saini SD, Lansdorp-Vogelaar I, Vijan S, Meester RG, de Koning HJ, Zauber AG, van Ballegooijen M. Personalizing colonoscopy screening for elderly individuals based on screening history, cancer risk, and comorbidity status could increase cost effectiveness. Gastroenterology. 2015 Nov 1; 149(6):1425-37.
- Saini SD, van Hees F, Vijan S. Smarter screening for cancer: possibilities and challenges of personalization. JAMA. 2014 Dec 3; 312(21):2211-2.
- Saini SD. Should my patient get a screening colonoscopy? 2016 Jun 30. Available from: http://screeningdecision.com/.