Fax: (770) 488-4639
A vignette-based study
Article first published online: 25 JUL 2011
Copyright © 2011 American Cancer Society
Volume 117, Issue 23, pages 5334–5343, 1 December 2011
How to Cite
Trivers, K. F., Baldwin, L.-M., Miller, J. W., Matthews, B., Andrilla, C. H. A., Lishner, D. M. and Goff, B. A. (2011), Reported referral for genetic counseling or BRCA 1/2 testing among United States physicians. Cancer, 117: 5334–5343. doi: 10.1002/cncr.26166
We thank Blythe Ryerson from the Centers for Disease Control and Prevention for her early contributions to the development of the study's methods and Gilmore Research Group in Seattle, Washington, for conducting the survey.
The findings and conclusions of this study are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
- Issue published online: 18 NOV 2011
- Article first published online: 25 JUL 2011
- Manuscript Revised: 13 MAR 2011
- Manuscript Accepted: 4 MAR 2011
- Manuscript Received: 14 DEC 2010
- ovarian neoplasms;
- attitude of health personnel;
- health knowledge;
- genetic testing;
- primary health care
Genetic counseling and testing is recommended for women at high but not average risk of ovarian cancer. National estimates of physician adherence to genetic counseling and testing recommendations are lacking.
Using a vignette-based study, we surveyed 3200 United States family physicians, general internists, and obstetrician/gynecologists and received 1878 (62%) responses. The questionnaire included an annual examination vignette asking about genetic counseling and testing. The vignette varied patient age, race, insurance status, and ovarian cancer risk. Estimates of physician adherence to genetic counseling and testing recommendations were weighted to the United States primary care physician population. Multivariable logistic regression identified independent patient and physician predictors of adherence.
For average-risk women, 71% of physicians self-reported adhering to recommendations against genetic counseling or testing. In multivariable modeling, predictors of adherence against referral/testing included black versus white race (relative risk [RR], 1.16; 95% confidence interval [CI], 1.03-1.31), Medicaid versus private insurance (RR, 1.15; 95% CI, 1.02-1.29), and rural versus urban location. Among high-risk women, 41% of physicians self-reported adhering to recommendations to refer for genetic counseling or testing. Predictors of adherence for referral/testing were younger patient age [35 vs 51 years [RR, 1.78; 95% CI, 1.41-2.24]), physician sex (female vs male [RR, 1.30; 95% CI, 1.07-1.64]), and obstetrician/gynecologist versus family medicine specialty (RR, 1.64; 95% CI, 1.31-2.05). For both average-risk and high-risk women, physician-estimated ovarian cancer risk was the most powerful predictor of recommendation adherence.
Physicians reported that they would refer many average-risk women and would not refer many high-risk women for genetic counseling/testing. Intervention efforts, including promotion of accurate risk assessment, are needed. Cancer 2011;. © 2011 American Cancer Society.
Women with breast cancer susceptibility 1 and 2 (BRCA1/2) gene mutations have a substantially higher risk of breast and ovarian cancer than women without such mutations, with a cumulative risk of developing breast cancer by age 70 of 57% and 49% for BRCA 1 and 2 carriers, respectively.1 Cumulative risk for developing ovarian cancer incidence by age 70 is 40% for BRCA1 carriers and 18% for BRCA2 carriers.1 Women with a personal or family history indicative of a BRCA1/2 mutation may benefit from genetic counseling and testing. For women who test positive, interventions including chemoprevention, prophylactic mastectomy, and oophorectomy can decrease breast cancer incidence and ovarian cancer incidence by up to 95% and 80%, respectively.2-4
Several organizations have evidence-based recommendations for genetic testing and counseling for hereditary breast and ovarian cancer. In general, women at average risk for breast or ovarian cancer should not have genetic counseling or testing, whereas women at high risk should be offered these services. The United States Preventive Services Task Force (USPSTF) recommends that women be referred for genetic counseling and evaluation if their family history is associated with an increased risk for BRCA1/2 mutations (grade B recommendation).2 The USPSTF specifically recommends against routine referral for genetic counseling or testing for women without such family histories (grade D recommendation), because the potential harms outweigh the benefits.2 The American College of Obstetrics and Gynecology (ACOG) recommends a genetic risk assessment for patients with a >20%-25% chance of having an inherited predisposition to breast and ovarian cancer.5 The American College of Medical Genetics and the National Comprehensive Cancer Network both suggest risk assessment and referral to genetic counseling for breast cancer patients diagnosed before the age of 45.6, 7
Because of the potential benefits of genetic counseling and testing among women at high risk for BRCA1/2 mutations and the potential harms among women at average risk, it is important to learn more about the use of genetic tests by primary care physicians and what patient and physician characteristics predict use of genetic counseling and testing. Several studies have examined physician referral to genetic counseling and/or testing for hereditary breast and ovarian cancer in the United States as well as the predictors of referral and use.8-22 Although some studies have examined testing rates stratified by patient risk,9, 12, 17, 19-22 to our knowledge none have explored the appropriateness of referral to genetic counseling and testing by patient and physician characteristics using a nationally representative sample of physicians.
This study fills this gap through a vignette-based survey of United States family physicians, general internists, and obstetrician/gynecologists that 1) examines adherence to genetic counseling and testing referral recommendations among women at average and high risk for ovarian and breast cancer and 2) identifies physician and patient predictors of adherence. Such national estimates are not currently available, but are critical for physician-based education and intervention efforts.
MATERIALS AND METHODS
United States physicians <65 years of age practicing in office or hospital settings comprised the eligible study population (n = 3200, equally divided among the primary specialties of family medicine, general internal medicine, and obstetrics/gynecology [OB/GYN]). We excluded physicians ≥65 years to ensure a sample of physicians with the highest likelihood of being in active clinical practice. Of these, 200 physicians pilot-tested the survey, and 3000 were administered the final survey. Physicians were randomly sampled from the August 2008 American Medical Association (AMA) Physician Masterfile (72,241 family physicians, 77,007 general internists, and 28,929 OB/GYNs) and equally drawn from the 3 specialties.
We developed a 12-page mailed survey booklet with 3 vignettes that asked about physicians' care for women's health and posed questions about physician demographics, practice characteristics, attitudes, beliefs, sources of information, and cancer experience. One of the vignettes asked about the services provided to a woman presenting for an annual examination, including frequency of referral to genetic counseling and offering or ordering BRCA1/2 testing (“almost never,” “sometimes,” or “almost always”). Patient characteristics (race [black, white], age [35 years, 51 years], insurance status [Medicaid, private insurance], request for ovarian cancer screening [yes, no]), and level of ovarian cancer risk [average, medium, high]) were varied across this clinical vignette, resulting in 48 vignette variations. Risk of ovarian cancer was defined according to family history: average risk, a woman whose mother developed breast cancer at age 70; medium risk, a woman whose mother developed ovarian cancer at age 62 and died at 65; and high risk, a woman with a personal history of breast cancer at age 30, a paternal grandmother with ovarian cancer, and a paternal first cousin with premenopausal breast cancer.
The 3000 physicians in the final survey were randomized to receive 1 of the 48 vignette versions. The survey was conducted using a modification of the Total Design Method,23 which entailed 2 priority mailings of the questionnaire, a midpoint reminder/thank you postcard, a $20 bill with the first mailing, and a handwritten note from the principal investigator with the second mailing. Return of the questionnaire was used as assurance of a physician's consent. The study was approved by the institutional review boards at the University of Washington and the Centers for Disease Control and Prevention.
The outcome variable was physician self-reported adherence to genetic counseling and testing guidelines. For women of average ovarian cancer risk, physician adherence was defined as “almost never” referring for genetic counseling or testing and “almost never” offering or ordering BRCA1/2 testing. For high-risk women, adherence was defined as “almost always” referring for genetic counseling or testing or “almost always” offering or ordering a BRCA1/2 test. All other combinations were considered “nonadherent.”
Patient-level predictors of physician referral for genetic counseling and testing included race, age, family history, and insurance status. Physician and practice characteristics were elicited from the survey, and included practice type (group, solo), primary setting, weekly average number of patients, involvement in clinical teaching, and nonprofessional experience with cancer (personal, family, none). The survey asked physicians to choose the top 3 organizations that influenced their cancer screening recommendations, including those organizations with published recommendations about genetic counseling or testing (USPSTF and ACOG). Age and sex data were obtained from the AMA Masterfile. Years of practice were computed according to graduation year in the AMA Masterfile. Practice location was determined from the physician mailing address zip code and linked to rural urban commuting area codes (urban, rural).24, 25 Fear of malpractice and level of risk taking were calculated using previously described methods.26, 27 We used the primary physician specialty recorded on the survey, unless 2 specialties were reported, in which case we used the one that agreed with the specialty in the AMA Masterfile. Physicians estimated the patient's ovarian cancer risk in the scenario as “same as general population” (average risk), “somewhat higher than general population,” or “much higher than general population” (high risk).
Development and Weighting of Sample
Responses from both the pilot and main surveys were combined for a larger sample size. Of the 3200 physicians surveyed, 158 were excluded (33 duplicates; 95 undeliverable surveys; 19 retired, disabled, or deceased respondents; and 11 not practicing or on leave), resulting in 3042 physicians; of these, 1878 (61.7%) responded. We then excluded 200 physicians not providing outpatient care to women, 71 not working in outpatient/primary care settings, 10 reporting other specialties, and 23 in residency or fellowship training for a sample of 1574 respondents. Because our analysis demonstrated no notable specialty differences, we have chosen to report nationally representative data for these 3 specialties combined. Thus, the responses of the 591 family physicians, 414 general internists, and 569 OB/GYNs were weighted to their representative number in the practicing United States physician population using AMA Masterfile counts (63,418 family physicians, 62,573 general internists, and 26,677 OB/GYNs) for a final weighted sample size of 152,668.
Only physicians who received questionnaires with annual examination vignettes of women at average or high ovarian cancer risk were included (n = 513 and 466, respectively), because we wanted to focus on those vignettes for which there would be no question about whether genetic testing and counseling was indicated. The final weighted physician sample size was 93,771. We calculated weighted percentages of adherence by patient and physician characteristics, stratified by patients' ovarian cancer risk (average or high). We report unadjusted results on variables that were associated with adherence at the P ≤ .01 level due to multiple comparison testing. We chose the P ≤ .01 significance level because at this level, the rate of having an error due to chance would be <1 of the 38 bivariate comparisons in the analysis, whereas at the 0.05 level the error due to chance would be 2 comparisons. Stepwise multivariable models using logistic regression were developed for average- and high-risk women. Physician characteristics were entered in the stepwise model if they were associated with adherence at P≤.05 in unadjusted analyses. Patient age, race, and insurance status were included in the final regression model regardless of significance. All analyses were conducted with SUDAAN version 10.0 software (RTI International, Research Triangle Park, NC). Odds ratios from logistic regression models were converted to risk ratios using predictive marginals in SUDAAN.28
The physicians responding to the average- and high-risk vignettes in the survey were predominately 50-64 years of age (42%), white (71%), and men (59%). We weighted the 3 physician groups to their representative proportion in the physician population: 43% in family medicine, 39% in general internal medicine, and 18% in obstetrics/gynecology (Table 1).
|American Indian/Alaska Native, mixed race, other||8.1|
|General internal medicine||39.2|
|Years in practice|
|Office or freestanding clinic||80.2|
|Urgent care center||1.6|
|Hospital outpatient department||5.9|
|Health maintenance organization or other prepaid practice||2.1|
|Community health center, tribal health center/Indian Health Service, or nonfederal government clinic||3.0|
|Federal government clinic||2.5|
|Other, including institutional setting, family planning clinic, and missing primary setting||4.7|
|Weekly average number of patients|
|Involved in clinical teaching||40.1|
|Nonprofessional experience with cancer|
|Family (immediate and extended)||78.4|
|Urban location of practice||83.3|
|East north central||16.8|
|West north central||7.8|
|East south central||5.9|
|West south central||7.9|
|Level of risk taking|
|Fear of malpractice|
|Listed USPSTF among top 3 sources of cancer screening information||53.8|
|Listed ACOG among top 3 sources of cancer screening information||32.0|
Among average-risk women (Table 2), overall self-reported physician adherence to recommendations against BRCA1/2 counseling or testing was 71%. In bivariate analyses, physician adherence was significantly higher for women with Medicaid than those with private insurance (77% vs 66%).
|Characteristics||Average-Risk Women (n=46,370)a||High-Risk Women (n=43,073)b|
|Adherencec (Not offering genetic counseling or testing)||P||Adherenced (Offering genetic counseling or testing)||P|
|General internal medicine||73.9||41.1|
|Years in practice||.95||.13|
|Weekly average number of patients||.68||.14|
|Involved in clinical teaching||.60||.63|
|Nonprofessional experience with cancer||.08||.09|
|East north central||68.8||33.6|
|West north central||79.8||38.3|
|East south central||77.8||24.3|
|West south central||62.9||43.4|
|Level of risk taking||.16||.06|
|Fear of malpractice||.85||.10|
|Listed USPSTF among top 3 sources of cancer screening information||.34|
|Listed ACOG among top 3 sources of cancer screening information||.95|
For high-risk women (Table 2), 41% of physicians would recommend referral for genetic counseling or testing, which is consistent with all guidelines. Age was the only patient characteristic associated with physician adherence; physicians were more likely to report adherence for a 35-year-old (57%) patient versus a 51-year-old (27%) patient. Female physicians reported higher adherence than male physicians (50% vs 35%). OB/GYNs had the highest adherence (57%), followed by general internists (41%) and family physicians (34%). Physicians with practices in urban areas were more likely to be adherent (44%) than those in rural areas (25%). Physicians who listed the USPSTF among their top 3 sources of information about cancer screening were less adherent (35%) than those who did not (48%).
For both average-risk and high-risk women, physician-estimated ovarian cancer risk was the strongest predictor of adherence to recommendations (Table 3). The more accurate the physician's estimated risk of the patient, the better the adherence. Among average-risk women, 78% of physicians who correctly estimated the patient as average risk would adhere to recommendations and not offer genetic testing or counseling versus only 26% of physicians who incorrectly labeled the patient as having high risk. Similarly, when physicians accurately estimated the risk of a patient at high risk of ovarian cancer, they were much more likely to report adherence to recommendations to offer genetic testing or counseling (65%) than if they misclassified a high-risk woman as being average risk (5%).
|MD Estimated Risk of Ovarian Cancer||Actual Risk: Average Risk (n=46,370)a||Actual Risk: High Risk (n=43,073)b|
|Adherence (No Counseling or Testing)||Nonadherence (Counseling or Testing)||Adherence (Counseling or Testing)||Nonadherence (No Counseling or Testing)|
|Somewhat higher risk||55.4||44.6||33.5||66.5|
|Much higher risk||26.0||74.0||64.6||35.4|
The results from adjusted models combining patient and physician characteristics were largely similar to bivariate results. Among average-risk women (Table 4), physician adherence to recommendations against genetic counseling and testing was significantly higher for patients who were black versus white and had Medicaid versus private insurance. Physicians who practiced in rural settings and accurately assessed the patient as being at average risk were more adherent compared with physicians who practiced in urban areas and assessed the patient at somewhat higher risk of ovarian cancer, respectively. For high-risk women (Table 5), young patient age, female physician, OB/GYN specialty, and accurate physician risk assessment were predictive of physician adherence. OB/GYNs were 64% more likely to report being adherent than family physicians (relative risk [RR], 1.64; 95% confidence interval [CI], 1.31-2.05). Even after adjustment, physicians who correctly identified women as high-risk were 8.46 (95% CI, 2.39-30.03) times more likely to refer them for counseling or testing than physicians who identified their high-risk patients as average risk.
|Characteristics||RR (95% CI)a for Not Offering Testing or Counseling (Guideline- Adherent Care)|
|Race, black vs white||1.16 (1.03-1.31)|
|Age, y, 35 vs 51||0.95 (0.84-1.07)|
|Insurance, Medicaid vs private||1.15 (1.02-1.29)|
|Estimated risk of ovarian cancer|
|Somewhat higher risk||0.72 (0.60-0.87)|
|Much higher risk||0.35 (0.07-1.66)|
|Geographic location, urban vs rural||0.88 (0.77-0.99)|
|Characteristics||RR (95% CI)a for Offering Genetic Testing or Counseling (Guideline- Adherent Care)|
|Race, black vs white||1.01 (0.82-1.25)|
|Age, y, 35 vs 51||1.78 (1.41-2.24)|
|Insurance, Medicaid vs private||0.90 (0.73-1.11)|
|Estimated risk of ovarian cancer|
|Somewhat higher||4.39 (1.23-15.70)|
|Much higher||8.46 (2.39-30.03)|
|General internal medicine||1.15 (0.89-1.51)|
|Sex, women vs men||1.30 (1.07-1.64)|
Because physicians' estimated risk of ovarian cancer was such a strong predictor of adherence for the hypothetical patient, we re-ran the multivariable models adding the estimated risk variable at the last step. In this way, we identified whether physicians' estimated risk explained the association between other physician characteristics and adherence to genetic counseling and testing referral recommendations. The results were largely unchanged using this approach, with 1 exception. In the high-risk model, the significantly greater adherence to guidelines among urban compared with rural physicians (RR, 1.60; 95% CI, 1.05-2.45) was explained after entering physicians' estimated ovarian cancer risk (RR, 1.39; 95% CI, 0.98-1.98), confirming the strong association found in the unadjusted analysis (19% of rural physicians vs 38% of urban physicians accurately assessed high-risk patients as high risk).
Adherence to recommendations that referral for genetic testing and counseling be offered to women at high risk of breast and ovarian cancer is low (41%). This finding is concerning, given the availability of evidence-based interventions to decrease breast and ovarian cancer risk in these high-risk women.2-4 At the same time, many physicians self-report referring average-risk women to genetic counseling and testing services (29%). This is an inefficient use of resources, and is associated with, at most, a small clinical benefit.2 It is difficult to estimate the absolute numbers of average-risk women who may be tested unnecessarily, but considering that most women are average-risk, even a modest degree of testing among this group could translate into substantial numbers of women receiving unnecessary care. Other studies have also demonstrated a similar percentage (24%, 45%, and 92%) of referral to genetic services among women at low to moderate risk of carrying BRCA1/2 mutations.19, 20, 22 Among high-risk women, the published literature demonstrates that referral to genetic counseling or testing ranges from 9% to 60%,9, 12, 17, 20-22 which is comparable to our estimates. One study reported wide discrepancies in referral depending on whether the high-risk family history was related to the patient's mother (29%) or father (9%).20 Methodologies of previous studies have included physician self-report,12, 17, 19, 22 use of standardized patients,20 and record review.9, 21
In this study, self-reported adherence to recommended genetic counseling and testing practices was strongly related to the physicians' estimated risk of ovarian cancer among their patients. Physicians were more likely to report adherence when they accurately assessed their patients (65% among high-risk women and 78% among average-risk women); however, adherence is still not optimal, even with accurate risk assessment. In fact, a higher percentage of average-risk women who were incorrectly labeled as high risk would be sent for counseling or testing (74%) than high-risk women correctly labeled as high risk (65%). Other studies observed that a barrier to appropriate genetic services referral is lack of knowledge and confidence in genetic risk and risk reduction skills,16, 17 despite the widespread belief that assessing breast cancer risk is a primary care provider's responsibility.17 Primary care physicians also report knowledge deficits related to hereditary breast and ovarian cancer.8, 11, 20 Physician education ranging from basic genetic concepts to the presence of referral guidelines may remedy the mismatch between recommendations and practice.
Few physician practice characteristics, including fear of malpractice, influenced reported adherence with genetic counseling/testing recommendations. This suggests that concerns about legal ramifications are not influencing counseling or testing practices.
Women of average risk who were white, were privately insured, and lived in an urban area were more likely to receive genetic counseling or testing than their black, Medicaid-insured, and rural counterparts, despite the fact that these services are not recommended for an average-risk population. This corresponds to previous literature demonstrating that women who are insured or live in urban areas were more likely to receive all types of health care, both recommended care (eg, breast, cervical, and colorectal cancer screening)29-35 and nonrecommended care (eg, too frequent cervical cancer screenings).36 Similar to our results, others have observed that doctors in urban areas were more likely to refer for genetic counseling or testing.13, 15
Physician specialty has been associated with referral to genetic counseling and testing in published studies, with OB/GYNs having higher referral for genetic evaluations than general internists10-12, 16 or family physicians.10, 16 Previous research indicates that OB/GYNs have more training in, and thus, greater knowledge about genetics.11 One study demonstrated that almost all OB/GYNs had a genetics professional for patient referral versus two-thirds of internists.11 The association between physician specialty and self-reported adherence persisted even after adjustment for physicians' estimated risk of ovarian cancer, suggesting both independently contribute to physician adherence.
Because this was a vignette-based study, we can only describe physicians' self-reported intentions. True practice may differ from intended practice. However, vignette-based surveys have been validated as a way to measure receipt and quality of health care.37 Another limitation is the variation in recommendations among professional medical associations and other organizations. It is difficult to know which recommendations physicians are following, if any. For example, USPSTF only mentions family history and not personal history of breast or ovarian cancer in its recommendation. In addition, the USPSTF is the only organization that specifically recommends against genetic counseling and testing among average-risk women. Although it can be reasonably assumed that an organization with only a recommendation for high-risk women implies that testing in average-risk women should not routinely occur, there is a void of clear recommendations for average-risk women. The definition of what constitutes a high-risk family history varies across the major medical organizations, potentially causing confusion for physicians. Finally, because of the complexity of the recommendations, substantial time is needed to elicit the necessary family history. Simple, reliable, and validated referral tools38 could help primary care physicians appropriately refer patients for a more detailed genetic risk assessment. Educational interventions to improve confidence and increase appropriate genetic services referrals have been successful.39
The strengths of this study include the large sample size and variation of ovarian cancer risk across the vignettes. This study was nationally representative and had a good response rate, whereas previous studies have either been limited to specific clinical sites such as academic medical institutions or to certain states and were not widely representative. To our knowledge, our study is the first to provide nationally representative estimates of self-reported adherence to genetic counseling and testing recommendations among United States primary care physicians by patient and physician characteristics.
Despite the existence of evidence-based guidelines related to referral for genetic counseling and testing for hereditary breast and ovarian cancer, many physicians report practices contrary to these recommendations, with too many average-risk women being referred for genetic counseling or testing, and too few high-risk women receiving these important services. Efforts are needed to encourage appropriate counseling and genetic testing for women at high risk of hereditary breast and ovarian cancer, particularly among male physicians, family physicians, and general internists. Simultaneously, efforts to discourage referral and testing among average-risk women should be implemented.
This work was supported by the Centers for Disease Control and Prevention through the University of Washington Health Promotion Research Centers Cooperative Agreement (U48DP001911), and through the Alliance for Reducing Cancer, Northwest, funded by both the Centers for Disease Control and Prevention (Grant U48DP001911, V. Taylor, PI) and the National Cancer Institute.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 6American College of Medical Genetics. Genetic susceptibility to breast and ovarian cancer: assessment, counseling and testing guidelines 1999. Available at: http://www.health.state.ny.us/nysdoh/cancer/obcancer/contents.htm Accessed March 16, 2010.
- 7National Comprehensive Cancer Network. The NCCN 1.2010 clinical practice guidelines in oncology genetic/familial high-risk assessment: breast and ovarian. Available at: http://www.nccn.org Accessed June 14, 2010.
- 23Mail and internet surveys: the tailored design method. 2nd ed. Hoboken, NJ: John Wiley & Sons; 2007..
- 25Economic Research Service. Measuring rurality: rural-urban commuting area codes. Available at: http://www.ers.usda.gov/briefing/Rurality/RuralUrbanCommutingAreas/ Accessed June 11, 2010.