Health Care Systems for Tracking Colorectal Cancer Screening Tests

Executive Summary

According to the Centers for Disease Control and Prevention (CDC) and United States Cancer Statistics (USCS) data, colorectal cancer (CRC) is the second leading cancer killer in the United States among cancers affecting both men and women. It is also one of the most commonly diagnosed cancers. In 2006, 139,127 people (70,270 men and 68,857 women) were diagnosed with CRC, and 53,196 people (26,801 men and 26,395 women) died from it (USCS, 2010). According to CDC, when CRC is found and treated early, survival is high (90 percent). However, many colorectal cancers are not found early due to low screening rates.

This report summarizes the experience of the CNA Health ACTION (Accelerating Change and Transformation in Organizations and Networks) Partnership in implementing and assessing a health care intervention to increase CRC screening and followup. The System Approach to Tracking and Increasing Screening for Public Health Improvement of Colorectal Cancer (SATIS-PHI/CRC) was a demonstration project conducted in primary care practices in the Lehigh Valley of Pennsylvania. The practices were affiliated with the Lehigh Valley Physician-Hospital Organization (LVPHO) and the Eastern Pennsylvania Inquiry Collaborative Network (EPICNet). This report also contains a description of our dissemination plans and efforts to date to spread the uptake of this intervention to other health care settings.

The project was funded by the Centers for Disease Control and Prevention (CDC). It was carried out as a task order (Contract No. HHSA290200600014, Task Order No. 290-06-0014-1) under the ACTION program of the Agency for Healthcare Research and Quality (AHRQ) between October 2007 and July 2010. We implemented the intervention in early 2009, and it ran through February 2010.

SATIS-PHI/CRC is a population-based system-redesign intervention designed to improve CRC screening rates and rates of diagnostic followup for positive screens. We based the major components of SATIS-PHI/CRC on prior studies developed in other settings. We used a case study approach, informed by the PRISM (Practical, Robust Implementation and Sustainability Model) framework, to determine whether we could:

  • Implement SATIS-PHI/CRC in the Lehigh Valley setting.
  • Increase screening and followup rates.
  • Achieve rate improvements similar to those previously reported.

SATIS-PHI/CRC is a six-step intervention that assists primary care practices in providing population-based CRC screening. Screening follows recommendations and guidelines jointly issued in 2008 by the American Cancer Society, the U.S. Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology and by the U.S. Preventive Services Task Force. This intervention seeks to influence the behavior of primary care providers and their patients regarding CRC screening and followup through targeted communications. It also is designed to facilitate the screening and followup process through improved eligibility identification and screening tracking systems.

The intervention is intended to be conducted by a central entity, such as a health care delivery system, accountable care organization, or insurer, affiliated with a network of primary care practices on behalf of and in conjunction with those practices. Intervention steps include:

  1. Recruiting primary care practices to participate.
  2. Conducting academic detailing to inform and influence the behavior of practice clinicians and other staff.
  3. Identifying patients of participating practices who are guideline eligible for, but not up to date in, CRC screening.
  4. Mailing information and reminders to patients regarding CRC screening and material facilitating screening.
  5. Tracking patient screening and followup of positive screens.
  6. Providing feedback to practices regarding patients who were screened and recommended for followup.

LVPHO, in conjunction with EPICNet, served as the central entity for our implementation study. We recruited 26 EPICNet practices to participate in the study: 20 practices to receive the intervention, 5 to act as controls not receiving the intervention, and 1 to be the site of a pilot of the intervention. We used purposive assignment to allocate practices to the intervention or control group to ensure a satisfactory mix of practice attributes in each group. We reviewed available electronic records (claims, billing, and electronic medical records) to identify eligible patients. We supplemented this record review with eligibility information provided by patients responding to a screening eligibility assessment (SEA) brief survey form that we mailed to them.

We identified 7,965 patients from intervention practices and 2,662 patients from control practices who met criteria for inclusion in our study. We further randomly assigned the 2,347 patients of two intervention practices to receive alternate versions of mailed screening materials (470 received a stool test kit and 1,877 received a mail-back card for requesting a kit) to estimate the effect of receiving a kit versus a card on screening rates. We reviewed electronic records and reports from a clinical laboratory that processed stool test kits, supplemented by audits of a sample of patients, to track screening and followup. We also conducted a short survey of participating practices, focus groups, and key informant interviews with each practice, as well as focus groups with a sample of patients of intervention practices to inform our assessment of our implementation effort and its outcome.

Overall, we were able to successfully implement the SATIS-PHI/CRC intervention in the LVPHO/EPICNet setting and generally achieve an effect on improving CRC screening rates that was comparable to previous intervention studies. However, we identified a number of factors that hindered implementation and that were a likely cause of lower than expected screening rates. The LVPHO/EPICNet central entity lacked some elements of the ideal implementation infrastructure. In particular, the central entity did not have electronic records systems set up for public health population-based patient outreach programs or experience implementing a program of this size based at the central entity rather than the practices.

We implemented the intervention during a period of economic uncertainty and limitations, resulting in fewer staff available for implementation at both the central entity and the practices. We also had to change the intervention protocol to accommodate the decision of the stool test kit supplier to restrict the number of kits it would make available free of charge. We thus could only send the kit directly to a small subsample of patients; the large majority of patients had to request a kit from us by mailing back a request card. The implementation timeframe was also a period of change and transformation among primary care practices in the Lehigh Valley that affected both their electronic medical record (EMR) systems and their ability to focus on the SATIS-PHI/CRC intervention.

These factors affected our ability to fully eliminate ineligible patients from our rate denominators and to fully identify completed screenings (especially colonoscopies) for our rate numerators, leading to low observed screening rates. Implementation delays shortened the period available for observing screening and followup. Shortages of colonoscopy providers in the Lehigh Valley to accommodate an increase in demand for screening likely depressed observed screening as well as followup rates during the shortened observation period.

Despite these implementation shortcomings, we found that the odds of being screened during the observation period were significantly greater among patients of intervention practices than control practices. This finding persisted even after controlling for age, gender, and various practice attributes, including the completeness of the tracking data available. Factors increasing the odds of being screened included receiving the stool test kit directly rather than having to request it and having commercial insurance.

We also found that our observed screening rate was substantially lower than that achieved by the previous study on which we modeled the patient outreach elements of SATIS-PHI/CRC. But when we more closely approximated the research conditions of that earlier study, our rate more closely approximated that study's rate. In particular, when we sent stool test kits directly to patients rather than request cards and included only patients responding to the baseline SEA survey in our analysis, we observed more comparable rates.

We found evidence of 786 patients being screened: 682 (8.6 percent) from intervention practices and 104 (3.9 percent or 4.7 percent with an adjusted denominator needed for comparison purposes) from control practices. Of those 786 screens, 363 were by stool test (almost all of the others were by colonoscopy), of which only 8 were positive (abnormal); we could not ascertain the results of an additional 18. We tracked the followup experience of these 26 patients for evidence of complete diagnostic examinations; however, their small number precluded any meaningful assessment of the intervention's effectiveness in improving followup rates. A comparison of the pre- and postintervention survey of intervention practices suggests that the academic detailing element of SATIS-PHI/CRC was somewhat effective in educating providers about current CRC screening guidelines.

Overall, our assessment of our implementation of the SATIS-PHI/CRC intervention in the LVPHO/EPICNet setting demonstrates that we were able to:

  • Successfully implement the SATIS-PHI/CRC intervention in a setting that differed from those that prevailed in studies on which we based our development of SATIS-PHI/CRC.
  • Achieve comparable effectiveness in improving the odds of becoming screened among guideline-eligible patients not up to date in their screening.

Further, we were able to extract a set of lessons learned from our implementation experience that could help others to successfully implement SATIS-PHI/CRC in their settings and to achieve comparable effectiveness outcomes. We describe these lessons learned in this report and introduce the implementation toolkit we developed to assist those who want to implement the intervention. This toolkit contains descriptions of how to implement each of SATIS-PHI/CRC's steps. It also has implementation tips based on our lessons learned and tools for each step (e.g., forms, materials to mail to patients, patient eligibility criteria, and practice recruitment and academic detailing material). We then describe our current efforts to disseminate the intervention through presentations at professional meetings, publication of papers, and involvement with health care delivery systems and with clinical and policy working groups.

Based on our assessment findings and our lessons learned, we believe that the SATIS-PHI/CRC can be a transferable intervention that can improve CRC screening and followup. It is most transferable to health care system settings with a central entity that:

  1. Is motivated to take the lead in organizing and implementing the effort.
  2. Has easy access to up-to-date and reasonably complete electronic records.
  3. Understands and accepts the time and resource commitment needed to undertake the intervention.
  4. Has experience with large, targeted, population-based mailings to patients (either by conducting such mailings themselves or outsourcing them to reliable contractors).
  5. Has strong relationships with its affiliated primary care practices.

For successful implementation, it is important to have a sufficient number of willing colonoscopy providers serving the medical service areas participating in the intervention to accommodate any increased demand for colonoscopies resulting from the intervention. In addition, it is best not to have other competing population-based initiatives in the service area or at the participating practices that could detract from the support and attention needed to implement SATIS-PHI/CRC.

Our experience with SATIS-PHI/CRC also demonstrates that this intervention can be successfully implemented in a wide range of practices. These include those that are more closely and less closely affiliated with the central entity and those that have and do not have fully functional EMR systems. However, the central entity would need access to sufficient other electronic records (in particular, claims or other evidence of medical services provided to patents) for practices without fully functional EMR systems. Successful implementation would also be enhanced if participating practices are dedicated to population-based preventive health in general and have strong leadership supportive of the SATIS-PHI/CRC intervention effort. In addition, it helps to have a clinical champion for the intervention.

Page last reviewed August 2015
Page originally created August 2015
Internet Citation: Executive Summary. Content last reviewed August 2015. Agency for Healthcare Research and Quality, Rockville, MD.