CHOICE is a collaboration between two university research groups (at Emory University and the University of North Carolina) and Aetna's Southeast Region Quality Management Department.
Identification of Potential Participants
This study received IRB approval from Emory University and the University of North Carolina. A partial HIPAA waiver was granted to allow access to information in Aetna's claims data repository for the purposes of identifying eligible health plan members. In addition, a full HIPAA waiver was obtained to send the decision aid intervention to non-responders who were potentially eligible in the intervention group and to permit access to follow-up claims data for non-responders in both the usual care and intervention groups.
Practice Recruitment and Randomization
To identify potential physician practices for participation, the study team obtained from Aetna a list of primary care physicians (family practice, general practice, and internal medicine) in the Atlanta, Tampa, and Orlando areas who participated in the Aetna HMO product. Medical practices recruited to the study had a minimum of 50 Aetna members between 52 and 75 years. Practices were recruited with a mailed introductory packet and telephone follow-up, supplemented with in-person visits, if requested by the practice.
As practices were enrolled into the study, they were grouped into three waves of 10 practices each. The first two waves were randomized into intervention and usual care groups. Randomization was done using matched pairs and a blocking procedure. Practices were matched on the number of study-eligible members per practice and rural vs. urban practices. These variables were chosen, because they may influence the effect of the intervention. When the third wave was recruited, intervention practice groups were unbalanced in practice size. Thus, purposive assignment to treatment group was used, resulting in a hybrid randomization procedure. Two additional practices were not randomized but included. These two practices initially were used as pilot practices (one intervention and one usual care). Because the intervention was not modified, and there were no differences in member characteristics for these two practices, the two practices also were included in the main trial.
Study Population and Recruitment
Potentially eligible participants were identified using claims data from Aetna for members between the ages of 52 and 80 whose primary care physicians had agreed to participate in CHOICE. Members who were not current with CRC screening were eligible for the study. Potentially eligible members were identified by claims data in three waves across 2005 and 2006 as practices were recruited into the study. A final claims extract was obtained in March 2007 to identify newly-eligible members in participating practices and to determine whether any of the eligibility survey non-respondents were no longer eligible for the study due to new medical exclusions. Each extract included data for the current year and the three previous years, which is the time frame contained in the member claims database.
Members were considered up-to-date with screening, and therefore ineligible, if claims data indicated that they had received fecal occult blood testing (FOBT) within the last year or a sigmoidoscopy, colonoscopy or barium enema during the period covered by the claims data. Other exclusions included individuals at increased risk for CRC (because the intervention was designed for average-risk patients) or with medical conditions that would limit their ability to participate in the study or might cause them tobe considered inappropriate candidates for screening and those unable to communicate effectively in English. Above average-risk persons were defined as adults with a personal history of CRC or polyps, a known history of CRC or polyps in a first-degree relative, or a known history of inflammatory bowel disease.
Potentially eligible members identified by claims data were sent a letter and a brief eligibility survey. The eligibility survey was used to obtain information about receipt of CRC screening that might not be in the claims data and to identify people at increased risk for CRC for whom the decision aid would not be appropriate (e.g. persons with a first degree relative with a history of CRC). Members who responded to the eligibility survey and who were eligible were sent a baseline survey . Those who responded to the baseline survey were enrolled.
The primary method of survey administration for health plan members was by mail. However, after two mailings (approximately one month apart), survey non-responders were given the opportunity to complete the surveys by telephone with trained research staff. An internet survey was available for those who preferred to respond online.
Theoretical Foundations of the Interventions
The intervention consisted of a combination of two evidence-based interventions, one targeted at the practice level (academic detailing) and the other at the patient level (decision aid). These two components of the intervention draw upon constructs from Social Cognitive Theory and The Transtheoretical Model or Stages of Change Model.10-12
Academic detailing sessions targeted constructs of collective self-efficacy and strategies of goal-setting and behavioral monitoring.10, 11
The decision aid intervention was based on constructs from the Transtheoretical Model (TTM)12
focusing on the pre-contemplation (not considering being screened), contemplation (considering being screened), and preparation (getting ready to be screened) stages.
Practice-Level Intervention: Academic Detailing
The academic detailing intervention was modeled after previously successful detailing interventions to improve cancer screening in primary care practices.6-9
The goal of detailing was to prepare practices to facilitate CRC testing once patients were activated by the decision aid. Our previous work suggests that practice-level barriers are important barriers to test completion.13
Detailing sessions were led by two of the investigators from UNC (MP, CL) and two general internists at Emory. Academic detailers participated in a three-hour training session where materials and processes for the detailing were reviewed.
Physician academic detailers visited each intervention practice twice during the study, once soon after randomization and again after most patients from the practice had received their decision aids. At the first meeting, the two physician detailers and a research assistant met with participating physicians and support staff. The lead detailing physician gave a fifteen minute presentation to review the burden of colon cancer, current recommendations and screening tests, barriers to colon cancer screening, and previous interventions shown to increase screening test completion in primary care practices. Practice-specific claims-based screening rates were presented to the groups to provide a baseline measure of screening completion. The detailing physicians showed clips from the decision aid.
Detailing physicians then led a discussion about the practice's colon screening strategies, including preferred and available testing options, current office systems to facilitate screening such as reminders, electronic medical records, and whether the office performs in-office flex-sig or FOBT. Referral practices for colonoscopy procedures also were reviewed. Practices were asked to explicitly map current processes for colon cancer screening, including procedures for receiving patient phone calls and scheduling referrals.
During these meetings, detailers led discussions about plans to modify office practices to better accommodate patient requests for specific tests. Physicians and staff from the practice helped to develop a plan in consultation with the physician detailers. Detailers recorded information from the discussion and helped clarify the plan. These written plans were sent to the practices after the meeting for confirmation.
The second session was held after most patient interventions were mailed. The same detailing team returned to practices whenever possible. Detailers reviewed plans with the practices and addressed questions or concerns including any procedural barriers the practices had faced. For two practices, these sessions were conducted over the phone.
To standardize the academic detailing process and assure quality control, standardized notes were taken during detailing sessions, and detailers held a debriefing after each session.
Member-Level Intervention: Decision Aid
The decision aid used in the CHOICE trial is a modified version of a previously tested decision aid, which had been found to be effective at increasing CRC screening rates (Pignone 2000). We refined the content and layout of the decision aid to include web-based interactive delivery. Following these changes, usability testing was performed using cognitive interviewing techniques. The decision aid was produced in DVD and VHS videotape formats to maximize usability by participants.
The duration of the decision aid was about 22 minutes. It included: 1) information on colorectal cancer and screening and 2) information about the specific screening tests and comparative information between tests (e.g efficacy) In the final section of the decision aid, the patient is asked to choose stage-targeted information in the form of a color-coded brochure corresponding to their readiness to be screened.
Decision Aid Mailings and Follow-Up
Decision aids were sent to intervention group members in a package that contained a tailored letter; decision aid with instructions for viewing; stage-targeted brochures; a co-payment and referral information sheet specific to Aetna members; CRC screening options chart; and a decision aid survey. The survey assessed the use of and reactions to the materials, and evaluated perceived changes in knowledge as a result of viewing the decision aid.
Participants in the intervention group who had not been screened as determined by self-report on the one year survey were sent a “booster” intervention. The booster consisted of a letter encouraging screening, information on how to learn more about CRC and CRC screening tests, and an action plan with steps to complete recommended screening tests.
Usual Care Condition
Usual care practices received no academic detailing. All members affiliated with practices in both the intervention and usual care groups received Aetna annual reminders to obtain CRC screening.
Mailings to Non-Responders
To assess the effect of the intervention in a broader context, decision aid packages were sent to non-responding members in intervention practices. These were health plan members in participating practices who did not respond to either the eligibility survey or the baseline survey but had not refused participation in the study (). These members will be included in the intention to treat analyses for CRC screening test completion based on analysis of claims data.
Consort Diagram of Member Eligibility, Participation and Group Assignment
Data Collection and Measures
Practice level measures
Intervention and usual care practices completed brief surveys to assess their CRC screening practices, referrals, and quality improvement initiatives prior to the intervention and again at the end of the study.
CRC Outcome Measures
We measured completion of CRC screening for health plan members through Aetna claims data and participants self-report. Claims data were assessed for CRC screening test completion for members in both the intervention and usual care groups who had completed the eligibility and baseline surveys, as well as those who did not complete the eligibility and/or baseline surveys but did not refuse participation (). We measured CRC screening test completion by participant self-report through the baseline survey and follow-up surveys. Follow-up surveys were sent at 12 months, and at 18 months for those who were unscreened at one year. Interest in CRC screening, intent to ask providers about screening, and readiness to be screened also were assessed on the baseline and follow-up surveys.
Social Cognitive Construct Measures
We measured social cognitive constructs in baseline and follow-up surveys as potential mediating variables. We performed factor analyses on survey items measuring attitudes, beliefs14
, and perceptions about colorectal cancer and cancer risk15
in order to group the individual items into discrete dimensions. Factor analyses revealed four clearly delineated composite variables: 1) Perceived Risk of colon cancer (4 items: e.g., what is your chance of getting colon cancer in your lifetime; response was 5 point scale from very low to very high); 2) General Beliefs about colon cancer (4 items: e.g., a person with colon cancer would always have symptoms; response options were 5 point Likert scale from strongly agree to strongly disagree); 3) Fears about colorectal cancer screening (4 items: e.g., I am afraid of having an abnormal screening test result for colon cancer; response options were 5 point Likert scale from strongly agree to strongly disagree); and 4) Intrusive Thoughts about colon cancer (3 items: e.g., during the past month, how often have you thought about your chances of getting colon cancer; response options on 4 point scale of not at all to almost all of the time).
Other social cognitive items measured participants’ knowledge of the age at which CRC screening should begin; perceptions about whether participants had enough information on CRC screening in general, as well as whether they had enough information about specific screening tests, intentions to talk to their doctor about CRC screening (two items), willingness to pay for screening tests16
, and desire to participate in medical decisions.17
12 and 18 Month Follow up Surveys
Approximately one year from receipt of baseline surveys, we mailed participants 12-month surveys. Participants who reported that they were not screened at 12 months were sent the 18-Month Survey. Outcome measures and social cognitive construct measures were identical to the baseline survey.
Power Calculations and Statistical Analyses
Our power calculations were based on our expectation that the proportion screened in the intervention group would be 15 percentage points higher than the proportion screened in the usual care group (.40 Intervention Group and .25 Usual Care). This estimate was based on the efficacy trial of the CRC decision aid5
and related research on academic detailing.7, 9
To account for the design, a cluster randomized trial, we assumed a conservative estimate of the intraclass correlation, 0.0225 or less. Based on alpha=.05, a two-tailed test, and power set at 0.80, 10 practices in each arm would have sufficient power to detect a 15 percentage point difference in the proportion screened if at least 25 participants from each practice, or 500 total participants, were enrolled. To ensure a sample of this size, allowing for some practice drop-out and member non-participation, we aimed to recruit at least 30 practices (15 in each group), with each practice having a minimum of 50 Aetna members between the ages of 52 and 75 years.
We will employ an intent-to-treat approach for analyses, using outcome data on recent CRC screening. We will perform two separate analyses. First, we will analyze claims data for all health plan members who were determined to be eligible at the start of the study, and who did not opt-out (and thus received the intervention if their practice was randomized to that treatment group), even if they did not complete survey questionnaires. These analyses will include both active participants and non-respondents in practices participating in the CHOICE trial. Second, we will analyze CRC screening test completion for those who actively enrolled in the trial and completed surveys.
For baseline results in this paper, we report practice and member participation rates. We also report descriptive statistics from members’ eligibility and baseline surveys and the participating practice surveys. We compared demographic characteristics of those who responded to the eligibility survey to those who did not respond or refused, using chi-square tests. We also compared demographic and social cognitive characteristics for intervention and usual care groups using chi-square tests.