It was estimated that approximately 12,000 new cases of cervical cancer and 4,000 deaths attributed to cervical cancer would occur in the United States in 2003. Worldwide, cervical cancer is the leading cause of morbidity and mortality from all gynecologic cancers. Although screening programs can dramatically reduce mortality, they frequently do not reach the women most at risk because of cost, cultural problems, or problems in the organization of health services. Garcia and colleagues tested the feasibility of using patient-collected samples for cervical cytology and detection of oncogenic human papillomavirus (HPV) as a means of extending screening programs to women with limited resources.

The authors enrolled 334 women who attended colposcopy clinics in Arizona, Mexico, and Peru. Exclusions were pregnancy, hysterectomy, or history of vaginal trauma or laceration. Each woman was instructed in the use of the endocervical brush and correct technique for placement of the specimen in containers for transport to a laboratory. Each woman was then examined by a physician who collected samples for cervical cytology and HPV testing. Finally, all women underwent thorough colposcopic examination, including directed biopsies. All specimens were processed using standard techniques. All abnormal specimens and a randomly selected sample of 10 percent of normal specimens were reviewed by an experienced cytopathologist.

The mean age of the enrolled women was 36.9 years, and the mean parity was 2.3. One percent of patient-collected and 2 percent of physician-collected samples were assessed as unsatisfactory. Endocervical cells were absent in 66 percent of patient-collected and 14 percent of physician-collected samples. For diagnosis of atypical squamous cells, using colposcopy as the gold standard, the sensitivity of patient-collected specimens was 55 percent and that of physician-collected samples was 85 percent (see accompanying table). The corresponding specificities were 84 and 73 percent. For HPV, patient-collected samples had significantly lower sensitivity than physician-collected samples (49 percent compared with 82 percent) but comparable specificity (73 and 67 percent, respectively). When cytology (abnormal threshold of atypical squamous cells of undetermined significance [ASCUS] or greater) or HPV was positive, self-collected specimens had significantly lower sensitivity (75 percent compared with 95.1 percent) and negative predictive values (86.1 percent compared with 96.4 percent) than physician-collected samples. Conversely, specificities and positive predictive values were not significantly different between the two techniques of collection.

The authors conclude that their results, plus those of other studies in developing countries of self-sampling for HPV and cytologic evidence of cervical abnormalities, indicate that self-sampling techniques may help overcome barriers to screening for cervical cancer. Despite its limitations, self-sampling may be especially helpful in women at increased risk because of limited access to health services.