Objective
In 2002, major US guideline groups recommended cervical cancer screening discontinuation in 2 groups of low-risk women: those >65 years old with previous normal Papanicolaou test results and those with no previous cervical neoplasia who have had a total hysterectomy. A survey administered to obstetrician-gynecologists around that time indicated that only 1 in 5 would not screen these women. To investigate whether screening practices have changed over time, we surveyed obstetrician-gynecologists about their practices and examined characteristics that are associated with adhering to screening guidelines.
Study Design
We conducted a national probability survey with a random sample from the American Medical Associations’ Physician Masterfile of 1020 obstetrician-gynecologists practicing in the United States. Obstetrician-gynecologists were asked to evaluate clinical vignettes ( Table ) and to indicate whether they would perform a Papanicolaou test, a human papillomavirus (HPV) test, or both and to rank the importance of each as very, moderately, a little, or not important. The vignettes were constructed such that neither a Papanicolaou nor an HPV test was indicated as per guidelines. Age, gender, race, and ethnicity characteristics were collected. Physicians were asked to indicate their practice setting (university-based, solo, group, hospital-based, or other), clinic volume, proportion of patients covered by Medicaid and Medicare, and geographic region. Data were collected between May 2010 and January 2011. Our response rate was 62%. Our methods have been previously reported. We performed multivariable logistic regression to identify provider characteristics that are associated with being guideline-adherent, defined as responding “no” to both Papanicolaou and HPV testing in both vignettes. Analyses were conducted with statistical software (STATA, version 13.1; StataCorp, College Station, TX). The Committee on Human Research at the University of California, San Francisco, approved the protocol.
| Vignette | n | Percentage that would not perform test | ||
|---|---|---|---|---|
| Papanicolaou | Human papillomavirus | Either | ||
| Woman (55 years old): her cervix and ovaries were removed last year at the time of hysterectomy for symptomatic fibroid tumors. She had no history of dysplasia, was not immunocompromised, and had no symptoms. | 514 | 76.3 | 89.9 | 75.1 |
| Woman (70 years old): she had had annual Papanicolaou tests with normal findings for the past 30 years. She had not been sexually active for the last 10 years. She had no history of dysplasia, was not immunocompromised, and had no symptoms. | 513 | 84.0 | 94.9 | 83.4 |
Study Design
We conducted a national probability survey with a random sample from the American Medical Associations’ Physician Masterfile of 1020 obstetrician-gynecologists practicing in the United States. Obstetrician-gynecologists were asked to evaluate clinical vignettes ( Table ) and to indicate whether they would perform a Papanicolaou test, a human papillomavirus (HPV) test, or both and to rank the importance of each as very, moderately, a little, or not important. The vignettes were constructed such that neither a Papanicolaou nor an HPV test was indicated as per guidelines. Age, gender, race, and ethnicity characteristics were collected. Physicians were asked to indicate their practice setting (university-based, solo, group, hospital-based, or other), clinic volume, proportion of patients covered by Medicaid and Medicare, and geographic region. Data were collected between May 2010 and January 2011. Our response rate was 62%. Our methods have been previously reported. We performed multivariable logistic regression to identify provider characteristics that are associated with being guideline-adherent, defined as responding “no” to both Papanicolaou and HPV testing in both vignettes. Analyses were conducted with statistical software (STATA, version 13.1; StataCorp, College Station, TX). The Committee on Human Research at the University of California, San Francisco, approved the protocol.
