The doctors to be included and patients to be surveyed were selected from lists provided by health plans, which included, depending on the metro area, Aetna, UnitedHealthcare, CIGNA HealthCare, BlueCross and BlueShield of Kansas City, and BlueCross BlueShield of Tennessee. The metro areas where each plan participated are as follows–
- Denver: Aetna, UnitedHealthcare
- Kansas City: Aetna, UnitedHealthcare, and BlueCross and BlueShield of Kansas City
- Memphis: Aetna, UnitedHealthcare, CIGNA HealthCare, and BlueCross BlueShield of Tennessee
- New York: Aetna, UnitedHealthcare (including Oxford), and CIGNA HealthCare.
The plans supplied lists of all of their physicians contracting with the plan with at least one practice location in the counties included in the survey.
For a one-year period, the plans listed all patient visits to these doctors involving certain evaluation and management CPT codes. (CPT codes describe the service/procedure rendered; to see a list of CPT codes, click here.) For each visit, the visit record included the date, the CPT code, the plan's identification code for the physician, and the plan's identification code for the plan member/patient.
The plans also provided CHECKBOOK/CSS lists of all members who had had a visit included on the visit list.
The patients on the lists the plans provided were predominantly members of commercial health plans (although a small number of Medicaid or Medicare members were included on some areas).
CHECKBOOK/CSS matched the physician lists across plans, using various physician identifiers supplied by the plans, including, where available, NPI number, UPIN number, date of birth, name, and various other identifiers.
Then, using the list of visits from each plan, CHECKBOOK/CSS identified patients who plan records showed had had visits to these physicians in a targeted time period (the period from November 1, 2007, through October 15, 2008 for the Denver, Kansas City, and Memphis areas, and the period from April 15, 2009, through April 14, 2010, for the New York area). From each doctor's list of patients with visits, CHECKBOOK/CSS selected a random sample of patients to survey. These patients were selected so that not more than one person in a household would be surveyed about the same doctor and not more than two persons in the same household would be surveyed at all.