The Effect of HEDIS Measurement of Colorectal Cancer Screening on Insurance Plans in Pennsylvania

Published Online: May 15, 2008
Mona Sarfaty, MD; and Ronald E. Myers, PhD

Objective: To determine the effect of Healthcare Effectiveness Data and Information Set (HEDIS) measurement of colorectal cancer (CRC) screening on insurance plans in Pennsylvania.

Study Design: Natural experiment tracking changes in CRC screening policies.

Methods: Survey data were collected in 2006 on screening policies of 13 Pennsylvania commercial insurers offering 37 plans. All companies that met the inclusion criteria were surveyed. Medical directors answered questions about how HEDIS measures affected plan benefit designs. Responses were analyzed using descriptive statistics.

Results: All companies responded and focused their responses on a particular plan as requested, including 2 health maintenance organizations, 3 point-of-service plans, 2 fee-for-service plans, and 6 preferred provider organizations. The survey results indicated that 39% of plans revised their screening guidelines, 46% established new or updated reminder systems, and 46% established new systems for tracking screening rates. Although only the health maintenance organization plans were linked to HEDIS with formal reporting, all types of plans reported changes that they attributed to the HEDIS measure.

Conclusion: The establishment of the new HEDIS measure on screening for CRC has been associated with changes in health plan policies and practices in Pennsylvania.

(Am J Manag Care. 2008;14(5):277-282)

The addition of screening for colorectal cancer to the publicly reported Healthcare Effectiveness Data and Information Set (HEDIS) measures in 2006 is associated with changes implemented by Pennsylvania health plans that have been shown to increase screening rates, such as the following:

  • Revised screening guidelines that shape physician recommendations for screening
  • Implementation of guidelines for those plans that did not previously have them
  • Measurement of screening rates
  • New reminder systems for patients
  • New tracking systems

These changes were implemented even by plans that did not have a formal reporting relationship with the National Committee for Quality Assurance, which collects and reports the HEDIS data.

Screening for colorectal cancer (CRC) is a beneficial and costeffective way to advance the public’s health by reducing the incidence and mortality of CRC, the nation’s second leading cause of cancer deaths.1 Colorectal cancer screening was designated as one of the highest priority preventive services by the National Commission on Preventive Priorities after 2 years of deliberation and analysis; the commission’s choice rested on 2 criteria, the burden of disease that could be prevented and the cost-effectiveness.2

Yet, the national screening rate of approximately 57% remains less than optimal.3 A nationwide survey of health plans conducted in 1999-2000 showed deficiencies in insurance coverage for recommended CRC screening tests and in the organizational systems that encourage enrollees to make use of such coverage where it exists.4 Only 57% offered coverage for colonoscopy, 41% had any system for monitoring delivery or outcome of screening, fewer than 25% had patient reminder systems, 16% had provider reminder systems, and 11% had tracking systems to determine whether invited enrollees completed screening. Only 5% tracked to determine whether individuals with positive screens received proper follow-up: 5% tracked the results of the follow-up tests, and 10% tracked adverse events related to the follow-up tests.

Evidence is strong that lack of insurance coverage is an impediment to the use of preventive services, including CRC screening tests.1,5 There is also considerable evidence supporting the usefulness of patient reminder systems and of tracking and feedback as effective approaches to raising screening rates.6-10

In 2003, in an effort to improve the country’s screening record, the National Committee for Quality Assurance11 (NCQA) added the CRC screening rate to the measures it requests from its health maintenance organization (HMO) member plans and announced that it would begin reporting these rates to the public in 2006. Given the earlier national survey, 4 this initiative on the part of the NCQA created the basis for a natural experiment in which it became possible to document the changes associated with this new public policy.

The NCQA is a voluntary organization with the mission of improving the quality of healthcare by creating accountability through application of measurement tools and transparency.11 The organization publishes measures and report cards that reflect the quality of care offered by managed care plans.12-15 Until recently, only HMOs were part of the NCQA reporting system. Preferred provider organizations (PPOs) subsequently became eligible in 2007. The Healthcare Effectiveness Data and Information Set (HEDIS) is a group of measures that has been published annually by the NCQA to provide information to employers and to other purchasers of health services about the performance of managed care (HMO) plans. The approach of the NCQA and other organizations that promote measurement, transparency, and accountability has been endorsed by the Institute of Medicine in its reports on quality and the healthcare system.16,17 The transparency achieved through public reporting of CRC screening rates was predicted to stimulate accountability that would raise screening rates among NCQA members. Whether there would also be an effect on non-HMO health plans that do not participate in the NCQA reporting was another question.

Pennsylvania is the sixth largest state, with a population of 12.4 million people (or 4% of the US population according to US census data). In 2006, the Pennsylvania General Assembly sought to gather information that would inform policy making in the area of CRC screening. The assembly was considering the establishment of a state mandate on insurers for the provision of CRC screening coverage that included the full range of options recommended by national consensus panels. An earlier effort to pass such legislation in 2002 was opposed by the insurance industry based on the claim that such screening coverage was already universal and that a mandate was unnecessary. In 2006, an assembly standing committee funded a survey of health insurance companies to provide unbiased information regarding CRC screening coverage. The survey included questions on the effect of HEDIS measurement of screening rates on plan policies. The findings of the survey are reported herein.

Study Design
Insurance companies that market health plans in Pennsylvania were identified from a listing in the Directory of Health Plans published by Atlantic Information Systems, Inc.18 The listing included all health plans offered by each company with business in the state. There were 37 plans in all. The listing included the size of the enrollment in each plan type and the name of the medical director and his or her contact information.

The following 2 inclusion criteria applied to insurance companies: (1) the insurance company had to offer at least 1 commercial health plan in the state (non-Medicare and non-Medicaid) and (2) at least 1 of its plans had to have a significant enrollment in Pennsylvania, defined as more than 25,000 enrollees. A minimum of 25,000 enrollees was chosen because only 5 plans had fewer than 25,000 enrollees, which added together produced a combined enrollment of less than 1% of the total enrollment, and the next smallest enrollment was 4 times that large. Plans designated for Medicare or Medicaid enrollees were excluded because their policies are determined at the state or federal level and not by the insurance company and because they would not have been affected by a proposed state insurance mandate. The study was given an expedited review and was approved by the Thomas Jefferson University Institutional Review Board.

Sample and Population
The medical directors were the principal recipients of the survey. They were asked to complete the survey based on the policies of a specific plan offered by their company. Their response about a single plan protected them from responding to more than 1 survey and assured that an assortment of plan types would be included in the survey. The plans included HMOs, PPOs, point-of-service (POS) plans, and fee-for-service (FFS) plans. To ensure adequate reporting about each type of plan, an effort was made to distribute plan types so that the sample would have the largest number of patients possible in each plan type. The company with the largest enrollment in a specific type of plan across all companies was asked to complete the survey about that plan. When a company had the largest plan in 2 classes of plans, the company was designated to represent the plan that was the largest product line for that company, and another company was selected to represent the next plan type.19 Enrollment in the designated plans of each plan type accounted for more than 25% of the total enrollment in that plan type. Health maintenance organizations offered by several companies were excluded because they were designed for Medicare or Medicaid enrollment.

The survey questionnaire was adapted from one used by the National Cancer Institute20 for an earlier national survey that the institute updated with  questions pertaining to HEDIS and posted on its Web site. The survey may be found at the Web site. For the purpose of this research, the investigators created hard copy and Web-based surveys. The survey domains were the following: (1) plan characteristics and benefits, (2) guidelines and clinical policies, (3) systems for cancer screening such as patient recruitment and reporting of results, and (4) plan experience with screening. Every survey page had a unique identification number that corresponded to the specific respondent. The Web-based survey was posted on the Internet (

An advance telephone call was made to identify the correct survey respondent, up-to-date contact information, and availability in the next 2 weeks. His or her availability in the next 2 weeks was documented. The surveys were mailed in the summer of 2006 with a stamped return envelope and a cover letter that explained the purpose of the survey and options for responding (hard copy vs Internet). Data from the responses to the hard copy were input into the Web-based survey during the next several months by research staff. The follow-up protocol involved reminder telephone calls, e-mails, and repeat mailings.

Descriptive statistics were used to report frequency and percentage of responses to each question. Commercially available statistical software was used in the data analysis (SAS version 9.1; SAS Institute, Cary, North Carolina). All percentages that describe the proportion of plans that responded are calculated based on the total number of 13 (the number of plans that received surveys). To maintain confidentiality, results were not analyzed by type of plan.

Thirteen companies met the inclusion criteria. The companies insured 8.3 million people. Eleven of 13 companies offered several types of plans, for a total of 37 plans. Two companies offered only HMOs; the rest offered several types of plans. Every company offered at least 1 HMO, 11 companies offered PPOs, 9 companies offered POS plans (an HMO or a PPO plus an indemnity plan), and 8 companies offered FFS plans. Although HMO was the most common plan type in the state, the enrollment in PPOs exceeded that of HMOs.

PDF is available on the last page.


Recommended Reading

No Result Found