The identification of strategies for the prevention, diagnosis, and treatment of cancer is a high priority for the nation. It is estimated that $10 billion is spent each year for hospital and physician services, $25 billion represents lost income, and over two million work-years are lost because of cancer.[1] With 1990 figures from the American Cancer Society (ACS) showing approximately 37,900 new cases in Michigan alone,[2] interventions aimed at early detection and treatment of cancer are actively being sought and implemented. The primary care physician, by virtue of practice location and accessibility to a large percentage of the population, has been identified as an important link in delivering the necessary education and early diagnostic procedures

Much of the literature on prevention of death and disability from cancer, including the ACS report on the cancer-related checkup,[4] assumes that the search for early cancer in asymptomatic individuals will afford the greatest medical benefit in a safe and practical way. Studies in various settings, however, have demonstrated that screening protocols are seldom implemented.[5-9] Patient, physician, test, and health care delivery system factors have all been cited as responsible for the failure to complete cancer screening.

Reports of cancer detection in practice settings are few. A retrospective study of cancer diagnoses from a single family practice over a 10-year period demonstrated that 69 cancers, or approximately one new cancer diagnosis every 2 months, were identified.[12] In the study, the majority of cancers were diagnosed among patients who were participating in cancer screening; however, only 2 of 11 patients with a diagnosis of colon cancer and 2 of 11 women with a diagnosis of breast cancer were asymptomatic. To improve the frequency and effectiveness of cancer screening by primary care physicians, more information is needed on how cancer is currently identified among their patients .

The purpose of the present study was to describe practitioner beliefs about cancer screening, early detection, and actual cancer detection rates (by both screening and case finding) in a population of patients cared for by a group of family practice physicians. It was hypothesized that the majority of cancer cases detected by these physicians would be among patients presenting with symptoms. In addition, it was hypothesized that those reporting a belief in more aggressive screening strategies would detect more asymptomatic cases and more cancers at an early stage of illness.

Methods

Subjects

Physicians who participated in this study were members of the Michigan Research Network (MIRNET), a voluntary network of Michigan practitioners interested in collaborating on primary care research projects. Twenty-nine family physicians and six physician assistants (PAs), representing 10 of the 18 MIRNET practices, participated in the study. These practices included 4 solo physician practices, 2 community-based family practice (FP) teaching faculty practices (7 physicians and 2 PAs), 1 academic FP faculty practice (four physicians), and 3 FP group practices (14 physicians and 4 PAs). Five practices were in rural locations (3 solo, 2 group practices). The 4 PAs in the group physician practices reported cases through the supervising physician rather than independently contributing information; therefore, the total number of practitioners participating in the study was 31.

Measures

A previously validated self-administered questionnaire on cancer screening practices was used with permission from Woo et al.[8] This questionnaire requested information on how often asymptomatic patients of varying ages should receive general (physical or pelvic examination) and specific (breast and/or rectal examination, fecal occult blood, sigmoidoscopy, Papanicolaou [Pap] smear, and mammography) cancer screening. Screening frequencies were reported as never, once in a lifetime, once every 10 years, every 4 to 5 years, every 2 to 3 years, or once annually. Subjects were also asked to rate themselves in general terms regarding the use of screening procedures (more, same, or less than recommended) and to list in rank order the reasons for following this approach. Information was requested on personal and family history of cancer.

A patient information card was used to identify each patient with a new diagnosis of cancer. This card was used to report the patient’s name and identification number, age, sex, diagnosis, screening and diagnostic tests used to identity the cancer, and whether the patient presented with symptoms attributable to the cancer. A test or examination w as considered part of screening if the patient was asymptomatic at the time of testing and the performance of the test was not prompted by the patient. A test or examination was considered diagnostic if the test was performed in response to either patient symptoms or a positive screening test. Cancers considered detectable by routine screening included breast cancer (clinical breast examination and mammography), cervical cancer (Pap smear), colorectal cancer (digital rectal examination, fecal occult blood, and sigmoidoscopy), prostate cancer (digital rectal examination with or without prostate specific antigen [PSA] test), and skin cancer (skin inspection during physical examination). The designation of these cancers as detectable was based on recommendations offered by the ACS.