Health Care Systems for Tracking Colorectal Cancer Screening Tests
2. Description of the Intervention
In this section, we first present a rationale and supporting evidence for the SATIS-PHI/CRC intervention. We then describe the role of a central entity in implementing it. We next provide a detailed description of each of the six component steps of the intervention (recruit practices, conduct academic detailing, identify eligible patients, mail screening information and materials, track patient screening and results, and provide feedback to practices) and our experience implementing them. We include the lessons we learned from implementing the intervention in the setting of the Lehigh Valley Physician-Hospital Organization and the ambulatory practices of its affiliated primary care providers. In the accompanying toolkit, we turn our lessons learned into "tips" to end users who may be interested in adopting the intervention.
Rationale and Supporting Evidence
SATIS-PHI/CRC is based on the premise that busy primary care practices could benefit from assistance in carrying out population-based screening programs (Zapka, 2008). We provide this assistance by having a central entity identify patients who are eligible for but not up to date in their CRC screening, send invitations to be screened and screening information and material to those patients, track whether patients get screened, send reminders to those who do not screen, and then issue reminders to the practices to follow up with screened patients. We modeled SATIS-PHI/CRC largely after the interventions developed by TJU researchers (Myers, et al., 2007; Myers, et al., 2004; Myers, et al., 2001).
The major components of SATIS-PHI/CRC are supported by recent literature. We incorporated a central entity to identify and communicate with eligible patients. Michael Pignone, M.D., a clinician and researcher with the University of North Carolina, Chapel Hill, medical school observed that most health care systems do "not have the ability to identify and then mass communicate with people who are not up to date with screening. More systems need to develop that kind of capability" (quoted in Pinkowish, 2009). We conducted academic detailing about CRC screening at participating practices and provided information about CRC and screening for it to eligible patients of these practices.
Educating both the provider and patient about the importance of CRC screening has been shown to be effective in increasing screening in patients (Levy, et al., 2007; Geller, 2008; Zapka, 2008). Receiving information and recommendations for CRC screening from their health care provider, particularly their primary care provider, has been found to be a predictor of patients becoming screened (Carcaise-Edinboro and Bradley, 2008; Griffith, et al., 2008; Sarfaty and Wender, 2007; Zajac, et al., 2010). Therefore, all the mailed communications to patients were sent by the central entity on behalf of the providers in each patient's practice. For example, all letters were signed by all providers in the practice rather than coming from the central entity itself. Mailed communications have been shown to be effective (Vernon, 1997; Snell and Buck, 1996). In one study comparing mailed reminders to patients and electronic reminders to physicians, the mailed reminders were more effective in increasing population-based screening rates (Sequist, et al., 2009).
Patients have varying preferences for CRC screening modalities (DeBourcy, et al., 2008; Hawley, et al., 2008), and opportunities for screening can be lost if patients are not given a choice of modality, especially when it comes to offering colonoscopy only. Therefore, we provided patients with a choice between colonoscopy and a less invasive test they could use themselves at home (stool test kit). The fact that colonoscopies may be a barrier to screening can be seen in a story that appeared in the Wall Street Journal in July 2009 (Mathews). A company that mandated that all employees would have to get certain exams and tests within a year or lose their insurance coverage excluded colonoscopy from the list of requirements because, according to the company's vice president for human resources, colonoscopies were "too intrusive" and mandating them might "create a lot of resistance and resentment."
Finally, since "failures to inform patients or to document informing patients of abnormal outpatient test results are common" (Casalino, et al., 2009), we built in both a feedback mechanism to remind providers to properly follow up with patients regarding test results and tools to assist practices and clinicians to track and document patient notification and followup.
The Central Entity
The SATIS-PHI/CRC system intervention is intended to be conducted by a central entity that has a relationship or affiliation with a formal or informal network of primary care practices. Examples of such a relationship or affiliation with practices include:
- Practices owned or operated by an integrated delivery network.
- Practices affiliated with or members of a PHO or independent practice association (IPA).
- Practices that provide care to the defined population for which an accountable care organization (ACO) is responsible.
- Practices that have a contractual relationship with a health insurance plan.
- Practices that are part of a centrally owned multilocation group practice.
- Practices affiliated with or members of a communitywide regional health information organization (RHIO).
- Practices located in the jurisdiction of a county or municipal public health agency.
The central entity conducting the intervention in each of these examples would be the delivery network, the PHO or IPA, the ACO, the insurance plan, the group practice, the RHIO, or the public health agency, respectively.i In each instance, the entity conducting the intervention acts centrally on behalf of the practices to institute a population-based screening program for the practices' patients who are eligible, according to prevailing guidelines, to be screened based on their age, personal and family medical history, and previous screening history.ii
The central entity for this implementation of the SATIS-PHI/CRC intervention was the LVPHO in conjunction with Eastern Pennsylvania Inquiry Collaborative Network (EPICNet, a practice-based research network [PBRN] of primary care practices affiliated with LVHN).
To conduct this intervention, the central entity must be able to meet the following eight requirements:
- Be able to centrally and electronically determine likely eligibility based on the guideline recommendations.
- Be able to contact patients on behalf of the practices to confirm their eligibility, invite them to be screened, and remind them if they have not screened after a period of time.
- Be able to track patient response to the invitation and to screening results.
- Be able to feed back results to practice clinicians and remind them of appropriate recommended followup for positive screening findings.
- Be willing to fund or find funding for the intervention.iii
- Identify one or more suppliers of stool test kits and one or more clinical laboratories to process them (the processing lab may also be the supplier in many cases).
- Have or develop a business associate relationship with the clinical lab—or use a lab it operates—to allow the lab to report screenings and results to the central entity.
- Be able to notify colonoscopy providers identified by participating practices as those to whom they refer that this screening program will take place and when and that they should expect a potential increase in requests for screenings (this last condition is recommended rather than strictly required).
In addition, since we conducted the intervention as part of a study to assess its feasibility, transferability, and effectiveness, we performed research-related central entity activities, including obtaining Institutional Review Board (IRB) approval, collecting pre- and postintervention data needed to assess the intervention, expanding the demographic information collected from patients for use in evaluating the intervention, and conducting focus groups with practice staff and patients. Central entities adopting this intervention would generally not need to perform these additional activities unless they wanted to assess the intervention in their settings. Such central entities would need to consult with their IRBs to determine whether the assessment phase would require review and approval.
Step 1: Recruit Practices
The first step of the intervention is to recruit primary care practices to participate. This component consists of encouraging primary care practices affiliated with the central entity to participate by providing them information about the importance of CRC screening and the prevailing low rate of screening, the nature of the intervention and the evidence it is based on, the benefits to them and their patients of participating, and the requirements of them and their patients for participation.
In addition to recruiting practices, the central entity must also consider whether it wants to involve a stool test kit supplier. The central entity can choose to bear the cost or negotiate with a clinical laboratory or other test kit supplier to bear the cost of supplying kits for all eligible patients or only those patients who request them through the enclosed request card. As a third option, the central entity can send the kit to patient groups it especially wants to target and send request cards to all other eligible patients.
Our research design protocol specified that we purposefully recruit 25 primary care (family medicine and general internal medicineiv) practices from among the 111 such practices affiliated with the LVPHO and then purposively assign them to intervention and control arms (20 intervention and 5 controls). We classified practices based on five attributes:
- Size (smaller practices with one to three clinicians and larger practices with more than three clinicians).
- Affiliation or ownership (LVHN-operated hospital clinics, LVHN-owned practices of the Lehigh Valley Physician Group [LVPG], independent practices affiliated with Medical Associates of the Lehigh Valley [MATLV], and unaffiliated independent practices).
- Specialty (family medicine and general internal medicine).
- Location within the Lehigh Valley area (urban, rural, suburban).
- Presence or absence of an electronic medical record system.
Using a purposive recruitment process, we successfully recruited 25 practices that included adequate representation of each type of practice.
We then assigned the practices to the intervention and control arms based on target quotas for the five distinct attributes to ensure equitability between the two arms. The attribute that we gave the highest priority to ensure equal distribution between the intervention and control arms was practice size. We felt this was one of the most important practice characteristics that could influence screening.
Following assignment, but before the start of the intervention, 5 intervention practices dropped out of the study,v leaving us with 15 intervention and 5 control practices participating. We decided not to recruit replacement practices, as the randomization of the practices had already occurred and we were following an "intention to treat" research design. The net result of this decision, however, is that fewer patients were exposed to the intervention.
For a practice to be eligible to participate, it had to be located in either Lehigh or Northampton County in Pennsylvania and a family medicine or general internal medicine practice. All participating practices agreed to participate in the study. The practices involved their patients in the study intervention as part of normal clinical care based on recommended screening guidelines.
We completed practice recruitment in November 2008. Intervention and control group assignment occurred once recruitment was complete.
We recruited a stool test supplier for the intervention to help facilitate screening. This supplier was the LVHN clinical laboratory (Health Network Laboratory, or HNL) that also would process and develop the stool tests. During the pilot, substantially fewer patients of the pilot practice chose to be screened by stool testing than initially expected. Based on the results of the pilot, HNL decided it would not be financially feasible to donate kits for all patients in the full intervention. HNL said it could not recoup the expense of the kits by processing and charging patients' insurance plans for a sufficient number to cover the costs of donating the kits.
In order to avoid charging patients for the cost of the kits, we negotiated an agreement with the lab to donate kits for patients who had a high likelihood of using them. We thus modified our full intervention protocol so that kits would be sent only to those patients who requested one using a request card enclosed with the invitation-to-be-screened mailing. We also arranged for the lab to donate some additional kits so that we could mail a subset of patients in two practices the kit directly, rather than the card. We felt the ease of screening might influence patients' decision to screen.
- Because five practices dropped out of the intervention before its start, the most important lesson learned is to stay engaged with the practices between the time you recruit them and the time you send the first mailing to their patients. While practice priorities will always be shifting and loss of practices from the intervention may be inevitable, maintaining engagement with the practices can help minimize this loss.
- While finding a stool test kit supplier was not the focus of this step, we learned some lessons during our pilot in terms of recruiting a supplier. When recruiting a stool test kit supplier, we recommend ensuring that the supplier is realistic in its understanding of the financial implications of supplying kits before agreeing to supply them. Since a clinical laboratory was the supplier for our implementation of SATIS-PHI/CRC, it planned to cover the cost of providing the kits by charging for developing those returned for testing. To break even, it required a certain percentage of kits to be returned. When the lab realized, based on its experience with the pilot, that fewer kits were likely to be returned than needed, it informed us that it could not continue to participate unless we could guarantee a higher rate of return. This condition resulted in our needing to revise the SATIS-PHI/CRC protocol by requiring most patients to request a kit before one would be sent to them. Even after insurance reforms associated with the recently passed Patient Protection and Affordable Care Act (P.L. 111-148) go into effect, this financial consideration will continue to apply.
- It is necessary to provide the practice with detailed written documentation of the practice's role in the intervention so that the practice office manager, staff, and clinicians understand who is supposed to do what and when. For the full intervention, we created a step-by-step instruction booklet to send to practices that outlined exactly what the practice and provider should do during each step of the process. This booklet also included a more detailed explanation of the lab's processing of stool test kits and a shortened form for practices to use when sending kits to the lab for processing. This approach proved very helpful for the full-intervention practices.
Step 2: Conduct Academic Detailing
The practice/clinician educational component of this intervention consists of an academic detailing session at each participating practice conducted by staff of the central entity. The detailers use informational material developed by this Task Order's staff (included as part of the dissemination kit produced for this Task Order). This material primarily consists of slides and notes for a detailing presentation to practice clinicians and staff. The presentation describes the intervention and provides information about currently recommended CRC screening and followup guidelines. Detailers also distribute and explain the use of a screening tracking spreadsheet developed for this intervention to be used by practices that do not already have an effective means of tracking screening tests. Detailers stress the importance of tracking screening, notifying patients of screening results, and following up on positive screens.
As needed, the central entity can also conduct followup academic detailing. This need may arise from information collected from an optional survey of clinicians and other practice staff regarding their perceptions of and behavior performing CRC screening and followup. It also may arise from record reviews (discussed below) or current events (e.g., changes in guideline recommendations). Detailers can develop an academic detailing "booster" containing material targeted at clarifying misperceptions or pointing out nonrecommended screening and followup behaviors that are revealed through the survey or focus groups.
We scheduled academic detailing sessions with each of the intervention practices at times that were most convenient for practice staff and clinicians. However, end-of-year holidays and busy schedules resulting from the winter cold and flu season resulted in scheduling delays and thus delays in completing these sessions. The academic detailing was not completed until the end of March 2009.
Prior to academic detailing at a practice, we distributed CRC screening surveys to the practice and requested that all staff complete them and have them available for collection prior to or at the beginning of the detailing session. We adopted this approach to avoid confounding preintervention baseline survey results by allowing staff to use information presented in the detailing session to respond to survey questions. For the same reason, we conducted a short focus group to learn about the practice's prevailing screening procedures before beginning the detailing portion of the session. We audio recorded the focus group portions of the sessions and later transcribed them for analysis.
It is important to note that the Multi-Society Task Force and the USPSTF revised their CRC guidelines late in 2008 and the revised guidelines were relatively new to clinicians. We intended the detailing to serve as a way to ensure that all clinicians were aware of the updates. We did not provide academic detailing to control practices as detailing is a component of the intervention and controls were not to be exposed to intervention components. Nothing, however, prevented clinical staff at control practices from reading the revised guidelines on their own, especially as these guidelines received coverage in publications aimed at clinicians.
Early results from analyzing the preintervention practice survey revealed that a not-insignificant percentage of clinicians at intervention practices held screening beliefs and engaged in screening behaviors that were not in accordance with the prevailing guidelines. To address this finding, we developed an academic detailing "booster" that was E-mailed and mailed to clinicians at the 15 intervention practices. This booster was designed to reemphasize which screenings were recommended and to differentiate between an in-office stool test done during a digital rectal exam (which is not recommended) and the use of an annual multisample at-home stool test (which is recommended). The booster also emphasized that positive tests should be followed up by colonoscopy (and not by a repeat stool test or flexible sigmoidoscopy).
During our pilot, we learned four key lessons regarding the academic detailing that affected how we implemented this step in the full intervention. Our lessons learned included the following:
- Some physicians may not agree with the new recommended CRC screening guidelines. During the pilot, we encountered a physician who believed that colonoscopy is the only acceptable screening modality and that other modalities recommended by the current guidelines are not correct, especially stool testing. We were concerned that his belief, if he communicated it to his patients, could negatively affect the intervention's ability to get his patients screened, especially if they were uncomfortable with colonoscopy and would prefer stool testing. We talked with him about his concerns and about the efficacy of stool testing as a screening modality. For the full intervention, we stressed the acceptability of all recommended screening modalities and noted that including both stool testing and colonoscopy to accommodate patient preference could increase screening rates.
- It was not always clear to practice staff, especially physicians, that participating in the intervention included agreeing to attend the academic detailing sessions (with the embedded focus group component) and to complete the survey of CRC beliefs and behaviors. For the full intervention, we reminded the practices and their physicians of this agreement before distributing the surveys and conducting the academic detailing sessions. This reminder helped alleviate the issue, but we still were not able to have all practice staff attend the academic detailing sessions.
- It was not clear to all practice staff that they needed to complete the survey before the academic detailing session. (We wanted to avoid having them answer questions about CRC screening after they had been detailed about prevailing recommended guidelines, thus confounding the results of the baseline [preintervention] survey). For the full intervention, we emphasized that surveys had to be completed before the detailing session and that completed surveys would be collected at the beginning of the session. We also allowed a short time at the beginning of the session prior to providing information regarding current guidelines for any practice staff without a completed survey to complete one.
i In appropriate circumstances, a single large primary care practice with a sizable patient population could institute this intervention on its own by acting as the central entity.
ii Although this intervention specifically targets CRC screening, the same general approach could be used for screening for other conditions or for other appropriate preventive services such as immunizations as long as there is a central entity able to meet the requirements stated below and there exists evidence-based and generally accepted guidelines recommending who should and should not receive the service. Thus, SATIS-PHI/CRC could become, for example, SATIS-PHI/DM for diabetes mellitus screening or SATIS-PHI/HTN for hypertension screening.
iii This funding is minimal considering the potential public health benefit to be derived. It is primarily required for conducting electronic reviews of records, mailing material to patients, and providing stool test kits for those patients preferring to be screened by stool testing.
iv We excluded pediatric practices since their patient populations do not meet the age eligibility requirements (50-79) for average-risk CRC screening.
v We list the reasons these practices cited for dropping out in our discussion of adoption of the intervention.