Abstract
Background/Aims
We compared the 6 Point Scale, a screening tool to identify low income women for referral to genetic counseling, with genetic counselors’ (GC) recommendation and the Referral Screening Tool (RST).
Methods
RST and 6 Point Scale scores were computed for 2 samples: S1, public hospital mammography clinic patients in 2006–2010 (n=744), classified by GCs as high risk (meriting referral to counseling) or not high risk; and S2, primary care patients enrolled in an education intervention study in 2011–2012 (n=1425). Sensitivity, specificity, and area under the ROC curve (AUROC) were computed for the 6 Point Scale score versus GC and RST classification as high risk.
Results
The 6 Point Scale had low sensitivity (0.27, 95% confidence interval (CI) 0.21–0.34), but high specificity (0.97, 95% CI 0.95–0.99) and AUROC (0.85, 95% CI 0.81–0.90) versus GC classification, and high sensitivity (S1: 0.90, 95% CI 0.79–1.00; S2: 0.94, 95% CI 0.87–0.97), specificity (S1: 0.95, 95% CI 0.93`–0.97; S2: 0.94, 95% CI 0.93–0.96), and AUROC (S1:0.98, 95% CI 0.96–0.99; S2: 0.98, 95% CI 0.98–0.99) versus the RST.
Conclusion
The 6 Point Scale compared favorably with the RST, a validated instrument, and is potentially useful as a simple tool for administration in a safety net setting, requiring minimal time investment by primary care physicians and their staff and no financial investment in tablet computers or software.
Keywords: screening tool, validation, hereditary breast cancer, low income, genetic counseling referral
Introduction
While hereditary breast cancer is relatively rare, with 5–10% of breast cancers due to BRCA1 and BRCA2 mutations [1], carriers of deleterious mutations in these tumor suppressor genes are at much higher risk for breast and ovarian cancer compared with the general population. The risk of developing breast cancer by age 70 is 46–71% for women with a harmful BRCA1 or BRCA2 mutation [2]. The risk of developing ovarian cancer by age 70 is 41–46% for BRCA1 mutation carriers and 17–23% for BRCA2 mutation carriers [2]. In contrast, the lifetime risks of being diagnosed with breast cancer or ovarian cancer are 12% and 1.3%, respectively, in the general U.S. population [3]. Once identified, carriers have several options that have been shown to reduce incidence and/or mortality including prophylactic surgery, enhanced screening, and chemoprevention [4]. However, it is estimated that more than 90% of BRCA1/2 mutation carriers have not been identified [5], with the uninsured and people of color disproportionately represented in this undetected population [6–8]. Only 13% of non-Ashkenazi women tested for BRCA1/2 mutations in 2006–2008 were of non-European descent, with African American and Latina women 4–5 times less likely to receive testing than women of Western European ancestry [9].
Until recent years, the primary means through which women have accessed genetic counseling has been self-referral, typically by those who are well educated and have the means to pay or insurance coverage for the costly test, and there has been considerable variation in insurance coverage for counseling and testing [10]. In addition, primary care clinicians and many mammography clinics may collect family history information but referrals to genetic counseling have been inconsistent [11]. Access to counseling and testing for the uninsured and those of low income is a recent development. In 2002 foundation-funded risk programs offering free services were established in some public hospitals around the country [12], then Medicaid began to pay for counseling and testing in several states [10]; now the Affordable Care Act considers counseling and testing of high risk women without a personal history of BRCA-related cancer a covered preventive service [13].
In order to disseminate the benefits of genetic counseling and testing more widely, it is necessary to identify high-risk women in diverse and low-income communities and to provide them with appropriately tailored education and referral to care. Identifying women at risk of hereditary breast cancer involves an assessment of family history, including the number of family members affected with cancer, types of cancer, and ages of cancer onset. This can be a lengthy procedure, usually not compensated in primary care. Recent studies have called for an increased focus on efficient assessment tools based on recognized guidelines [14] to help primary care physicians make decisions more easily and reliably regarding when to refer patients for genetic counseling or discuss genetic risks with them [15]. Therefore, it is desirable to have an easy screening tool that can be administered with a minimum of time and effort in a low resource primary care setting. Toward that end, we developed a strategy to identify high-risk women among the callers to California’s statewide phone service called Every Woman Counts (EWC), which provides access to free breast and cervical cancer screening, and assist them in obtaining genetic counseling [16]. Implementing the strategy required a screening tool that could (1) determine quickly and efficiently whether a woman had a family history that warranted genetic counseling, and (2) be administered over the telephone by someone who is not trained in genetics. With future use in clinics and community settings in mind, we required a simple, non-tablet based tool that could be implemented easily as a self-administered scale.
Genetic counseling is recommended in order to determine whether or not a patient’s personal and family history warrants genetic testing, and to provide guidance to the patient before and after the test [17]. Several screening tools have been developed to select patients for referral to genetic counseling, including the Pedigree Assessment Tool (PAT), a weighted scoring system that assigns points to family members in both maternal and paternal lineage extending to distant cousins [18] and the Family History Assessment Tool (FHAT) [19]. Both of these are relatively long and complex tools designed for staff administration in clinical settings. For best integration of our risk assessment and referral intervention into the infrastructure of EWC, we modified a simplified version of the PAT that had been developed in our public hospital risk clinic. Features of the modified tool, called the “6 Point Scale,” are that it consists of only 10 items, asks about fewer relatives, does not specify lineage, and assesses the most highly weighted indicators first so that the screening process can conclude if the threshold of 6 points, triggering referral to a genetic counselor, is reached (Figure 1). Details of the development and pretesting of the 6 Point Scale are presented elsewhere [16]. Another simplified tool, the Referral Screening Tool (RST) [20], was validated and published after we had begun using the 6 Point Scale; the 2009 RST defines a positive screening outcome as 2 or more checks on a 19-item checklist that requires a family history of breast and ovarian cancer by lineage.
Figure 1. The 6 Point Scale Items.
Originally published in Public Health Genomics 2015;18:65–66, erratum to the article Public Health Genomics 2012;15:172–180, reprinted with permission from S. Karger AG, Basel.
In order to validate the 6 Point Scale, we compared its selection of high risk family history profiles for referral to genetic counseling with classification as a positive screen according to the 2009 RST in two patient samples. We also compared the 6 Point Scale’s selection of family history screeners warranting referral to genetic counseling with genetic counselors’ retrospective assessment of the need for referral to formal genetic counseling in one of the samples. Here we present our validation of the family history screening tool we designed for brief and easy administration by phone and for maximum accuracy.
Materials and Methods
Patient Samples
We used two samples of patients to perform our comparisons. The first sample consisted of public hospital mammography clinic patients (S1). As part of their medical care, all women receiving screening mammography at San Francisco General Hospital (SFGH) are requested to complete a family history screener in their preferred language (English, Spanish, Chinese, or Russian) about their personal and family history of breast, ovarian, and other cancers (Figure 2). These screeners are read by a genetic counselor (GC) or genetic counseling assistant, who classifies screeners into three groups: a) family history of cancer highly suggestive of hereditary breast/ovarian cancer predisposition (Group A—high risk); b) family history with some cancer, but not suggestive of breast/ovarian hereditary cancer predisposition (Group B—not high risk); and c) no family history. In order to obtain approximately equal numbers of high risk and not high risk screeners, a stratified sampling scheme was used to select the validation sample from those completed in 2006–2010, including all from Group A and a simple random sample from Group B selected using a computer-generated sequence of random numbers. Our S1 sample size was designed to produce estimates of agreement and sensitivity with satisfactory precision. If kappa=0.70, a sample size of 700 screeners, with 50% high risk, produces a 95% confidence interval from 0.65–0.75. If the sensitivity is 0.90, a sample size of 350 high risk screeners produces a 95% confidence interval from 0.87–0.93. Because National Comprehensive Cancer Center (NCCN) guidelines for referral to counseling have changed over time [21,22], and to ensure that classification criteria were applied consistently, two licensed master’s level GCs independently reviewed the screeners and reclassified them as high risk or not high risk according to current (2011) guidelines; a third GC reviewed and adjudicated screeners with discrepant ratings.
Figure 2. Mammography Clinic Family History Screener.
Originally published in Genetic Testing, 2005, Volume 9, Issue 4, pp. 306–312, reprinted with permission from Mary Ann Liebert, Inc.
The second sample consisted of primary care patients participating in a randomized controlled trial (S2). Data were included from all participants who completed the baseline interview for the BreastCARE study, which tested a breast cancer risk reduction education intervention among women scheduled for an appointment with their primary care doctor at SFGH or UCSF’s general medicine practices [23]. Women completed the risk assessment questionnaire by telephone or tablet computer in their preferred language (English, Spanish, or Chinese) before their appointment in 2011–2012. Eligibility for the study included being age 40–75 and having no personal history of breast cancer. It is possible that some individuals in S2 were also included in S1.
Statistical Analysis
Patient Characteristics (Table 1): In S1, three family history screener risk groups were defined based on the review for this study in 2011 as 1) confirmed as high risk; 2) reclassified as high risk; and 3) confirmed as not high risk. In each risk group, frequencies and percentages were computed for demographic characteristics (age group, race/ethnicity, and language of screener) and personal/family history of cancer (breast/ovarian and other(s), breast/ovarian only, other(s) only, or none; any breast/ovarian); descriptive statistics (mean, standard deviation, minimum, and maximum) were computed for age. The age cut-points in Table 1 at 40 and 50 years correspond to age cut-points for mammography recommendations by the U.S. Preventive Services Task Force [24] ; at age 65 most U.S. residents become eligible for Medicare health coverage, which pays for screening mammograms and breast cancer treatment. The risk groups were compared with respect to these characteristics using chi-square tests for categorical variables and ANOVA for numeric age. Frequencies and descriptive statistics were also computed for S2.
Table 1.
Characteristics of Two Samples: Mammography Clinic Patients, San Francisco General Hospital, 2006–2011, by Family History Screener Risk Group (n=744); Primary Care Patients, BreastCARE Study, San Francisco, 2011–2012 (n=1425)
| San Francisco General Hospital | BreastCARE Study |
|||
|---|---|---|---|---|
| Confirmed High Risk (n=351) n (%) |
Reclassified High Risk (n=59) n (%) |
Confirmed Not High Risk (n=334) n (%) |
Baseline (n=1425) n (%) |
|
| Age (years) | ||||
| Under 40 | 43 (12) | 1 (2) | 6 (2) | 0 (0) |
| 40–49 | 90 (26) | 9 (15) | 66 (20) | 379 (27) |
| 50–64 | 171 (49) | 29 (49) | 157 (47) | 747 (52) |
| 65 and over | 27 (8) | 9 (15) | 32 (10) | 299 (21) |
| Unknown | 20 (6) | 11 (19) | 73 (22) | 0 (0) |
| Mean ± SD | 51.6 ± 10.4 | 56.5 ± 10.1 | 54.6 ± 9.1 | 56.3 ± 8.9 |
| Min-Max | 21–87 | 38–83 | 27–84 | 40–75 |
| Race/Ethnicity | ||||
| Latino/Hispanic | 87 (25) | 14 (24) | 91 (27) | 326 (23) |
| NH Black | 55 (16) | 6 (10) | 26 (8) | 326 (23) |
| NH Asian/Pacific Islander | 90 (26) | 23 (39) | 142 (43) | 273 (19) |
| NH White | 100 (28) | 12 (20) | 47 (14) | 476 (33) |
| Other/Multiracial/Unknown | 19 (5) | 4 (7) | 28 (8) | 24 (2) |
| Languagea | ||||
| English | 237 (68) | 35 (59) | 162 (49) | 1146 (80) |
| Spanish | 50 (14) | 10 (17) | 72 (22) | 171 (12) |
| Russian | 9 (3) | 4 (7) | 6 (2) | 0 (0) |
| Chinese | 55 (16) | 10 (17) | 94 (28) | 108 (8) |
| Personal or Family History of Cancer | ||||
| Breast/ovarian and other(s) | 176 (50) | 11 (19) | 31 (9) | NA |
| Breast/ovarian only | 147 (42) | 40 (68) | 107 (32) | NA |
| Any breast/ovarianb | 323 (92) | 51 (86) | 138 (41) | 547 (39) |
| Other(s) only | 28 (8) | 8 (14) | 196 (59) | NA |
San Francisco General Hospital: language of family history screener; BreastCARE study: primary language
San Francisco General Hospital: subtotal of breast/ovarian and other(s) and breast/ovarian only;
BreastCARE study : eligibility criteria included having no personal history of breast cancer.
Risk Groups: Genetic counselors’ classifications of familial risk based on screener were confirmed or reclassified as high risk (warrants genetic counseling) or not high risk (does not warrant genetic counseling)
Note: NH=non-Hispanic; NA=not available
All variables differ significantly across risk groups by chi-square tests or ANOVA (p<.0001)
Distribution of 6 Point Scale and RST Scores (Table 2): The 6 Point Scale was computed for S1 and S2 using all 10 items, so that scores greater than 6 were possible. RST scores were computed for S1; the RST score had been computed previously for S2 [23]. When computing the 6 Point Scale and the RST from S1, we assumed that an entry in the column labeled “diagnosed before age 50?”, which is next to the column labeled “breast cancer” (Figure 2), referred to age at diagnosis of breast cancer for females whether or not there was an entry in the adjacent “breast cancer” box. Also, because the screener in S1 asked for diagnosis before age 50 and the survey in S2 asked for diagnosis before age 51, it was necessary to compute the 6 Point Scale and the RST using the same age cut-points (50 in S1, 51 in S2). Frequencies and percentages for the 6 Point Scale and RST, and their cross-classification, were computed for each risk group in S1 and separately in S2.
Table 2.
Six Point Scale Scores and RST Scores in Two Samples: Mammography Clinic Patients, San Francisco General Hospital, 2006–2011 by Family History Screener Risk Group (n=744); Primary Care Patients, BreastCARE Study, San Francisco, 2011–2012 (n=1425)
| San Francisco General Hospital | BreastCARE Study |
|||
|---|---|---|---|---|
| Confirmed High Risk (n=351) n (%) |
Reclassified High Risk (n=59) n (%) |
Confirmed Not High Risk (n=334) n (%) |
Baseline (n=1425) n (%) |
|
| Six Point Scale score | ||||
| 0 | 27 (8) | 7 (12) | 195 (58) | 860 (60) |
| 2 | 22 (6) | 5 (8) | 93 (28) | 177 (12) |
| 4 | 120 (34) | 38 (64) | 35 (10) | 199 (14) |
| 6 or more | 182 (52) | 9 (15) | 11 (3) | 189 (13) |
| RST score | ||||
| 0 | 51 (15) | 12 (20) | 289 (87) | 1019 (72) |
| 1 | 178 (51) | 43 (73) | 42 (13) | 281 (20) |
| 2 or more | 122 (35) | 4 (7) | 3 (1) | 125 (9) |
| Six Point Scale × RST | ||||
| <6, <2 | 160 (46) | 50 (85) | 322 (96) | 1228 (86) |
| ≥6, <2 | 69 (20) | 5 (8) | 9 (3) | 72 (5) |
| <6, ≥2 | 9 (3) | 0 (0) | 1 (0) | 8 (1) |
| ≥6, ≥2 | 113 (32) | 4 (7) | 2 (1) | 117 (8) |
Risk Groups: Genetic counselors’ classifications of familial risk based on screener were confirmed or reclassified as high risk (warrants genetic counseling) or not high risk (does not warrant genetic counseling)
Note: RST=Referral Screening Tool (Bellcross et al., 2009): ≥2 warrants referral to genetic counseling
Six Point Scale: ≥6 warrants referral to genetic counseling
Comparison of the 6 Point Scale with GC Classification and RST (Table 3): Agreement between the 6 Point Scale score (≥6 vs. <6) and each of the other methods—final GC risk classification (high vs. not high) in S1, and RST score (≥2 vs. <2) in S1 and S2—was assessed using the kappa statistic, along with its 95% confidence interval, with strength of agreement interpreted as recommended by Landis & Koch [25]. Then, with the final GC risk classification and RST score as “gold standards”, the sensitivity, specificity, predictive value positive, and predictive value negative were computed for the 6 Point Scale score, as well as the proportion classified as high risk by each method; in S1, observations were weighted by the inverse probability of selection, and 95% confidence intervals were computed that accounted for the stratified sampling design. Treating the 6 Point Scale as numeric, the area under the ROC curve (AUROC) was computed with respect to each of the other methods using logistic regression; in S1, observations were weighted by the inverse probability of selection, and the bootstrap method [26] was used to compute 95% confidence intervals, as is recommended for the area under a weighted ROC curve [27]. Analyses for S1 were conducted using SAS, version 9.3; analyses for S2 were conducted using Stata, version 11.2.
Table 3.
Comparison of Six Point Scale with Genetic Counselors’ Classification and/or RST as Warranting Referral to Genetic Counseling in Two Samples: Mammography Clinic Patients, San Francisco General Hospital, 2006–2011 (n=744); Primary Care Patients, BreastCARE Study, San Francisco, 2011–2012 (n=1425)
| San Francisco General Hospital Clinic (n=744) |
BreastCARE Study (n=1425) |
||
|---|---|---|---|
| GC Classified High Risk Estimate (95% CI) |
RST Score Positive Screen Estimate (95% CI) |
RST Score Positive Screen Estimate (95% CI) |
|
| Sensitivity | 0.27 (0.21, 0.34) | 0.90 (0.79, 1.00) | 0.94 (0.87, 0.97) |
| Specificity | 0.97 (0.95, 0.99) | 0.95 (0.93, 0.97) | 0.94 (0.93, 0.96) |
| Predictive Value Positive | 0.68 (0.55, 0.82) | 0.44 (0.31, 0.56) | 0.62 (0.55, 0.69) |
| Predictive Value Negative | 0.83 (0.80, 0.87) | 0.99 (0.99, 1.00) | 0.99 (0.99, 1.00) |
| Kappa | 0.41 (0.36, 0.46) | 0.64 (0.58, 0.71) | 0.72 (0.66, 0.77) |
| Area Under ROC Curve | 0.85 (0.81, 0.90) | 0.98 (0.96, 0.99) | 0.98 (0.98, 0.99) |
| Proportion High Risk | |||
| GC or RST | 0.21 (0.18, 0.24) | 0.04 (0.03, 0.05) | 0.09 (0.07, 0.10) |
| Six Point Scale | 0.08 (0.06, 0.10) | 0.08 (0.06, 0.10) | 0.13 (0.12, 0.15) |
Note: RST=Referral Screening Tool (Bellcross et al., 2009); GC=Genetic Counselors; CI=Confidence Interval
All study activities were approved by the UCSF institutional review board.
Results
Patient Characteristics: S1: Of the screeners completed in 2006–2010, a total of 352 were initially classified high risk (Group A) and 5705 as not high risk (Group B), with a random sample of 467 selected from Group B. After excluding multiple screeners identified with the same medical record number and screeners with no personal or family history of cancer, the validation sample (n=744) consisted of 351 from Group A confirmed as high risk, 334 from Group B confirmed as not high risk, and 59 from Group B reclassified as high risk by GC review. As shown in Table 1, the 3 risk categories differed significantly with respect to demographic characteristics and, as expected, personal or family history of cancer (all p<.0001). Screeners confirmed high risk had higher proportions of younger women, non-Hispanic (NH) White and NH Black women, and English language screeners compared to those reclassified high risk or confirmed not high risk. In the confirmed high risk group, 92% had a personal or family history of breast or ovarian cancer, with half having a history of both breast/ovarian and other cancer(s). In contrast, the reclassified high risk group primarily consisted of women with a personal or family history of breast/ovarian but no other cancer (68%), and 59% of the confirmed not high risk group reported a history of other cancer(s) only. S2 overall had more NH Whites and English speakers than S1, and a lower proportion with personal/family history of breast/ovarian cancer (39%), reflecting the eligibility criterion excluding women with a personal history of breast cancer.
As shown in Table 2, these differences in history are reflected in the distribution of 6 Point Scale scores and RST scores in S1. In the confirmed high risk group, 52% had 6 Point Scale scores ≥6, compared to 15% of the reclassified high risk group and 3% of the confirmed not high risk group; similarly, 35% of the confirmed high risk group had RST scores ≥2 vs. 7% of the reclassified high risk group and 1% of the not high risk group. In S2, 13% had 6 Point Scale scores ≥6, and 9% had RST scores ≥2. The cross-classification of the 6 Point Scale and RST scores shows that most women who had RST scores ≥2 also had 6 Point Scale scores ≥6: 113/122 in the confirmed high risk group, 4/4 in the reclassified high risk group, 2/3 in the confirmed not high risk group, and 117/125 in S2.
As shown in Table 3, the 6 Point Scale classified a greater proportion of women in S2 as high risk (0.13, 95% confidence interval (CI) 0.12–0.15) than in S1 (0.08, 95% CI 0.06–0.10), and the GCs classified a much greater proportion of S1 women as high risk (0.21, 95% confidence interval (CI) 0.18–0.24). Consequently, the sensitivity of the 6 Point Scale with respect to the GC classification was low (0.27, 95% CI 0.21–0.34); however, the specificity was high (0.97, 95% CI 0.95–0.99) and the value of kappa indicated moderate agreement (0.41, 95% CI 0.36–0.46). In addition, the AUROC was high (0.85, 95% CI 0.81–0.90), suggesting that another cut-point corresponded more closely to GC classification.
In contrast, the RST classified a smaller proportion of women as high risk (S1: 0.04, 95% CI 0.03–0.05; S2: 0.09, 95% CI 0.07–0.10). The sensitivity of the 6 Point Scale with respect to the RST was high (S1: 0.90 95% CI 0.79–1.00; 0.94, 95% CI 0.87–0.97), as were the specificity (S1: 0.95, 95% CI 0.93–0.97; S2: 0.94, 95% CI 0.93–0.96) and AUROC (S1: 0.98, 95% CI 0.96–0.99; S2: 0.98, 95% CI 0.98–0.99); the value of kappa indicated substantial agreement (S1: 0.64, 95% CI 0.58–0.71; S2: 0.72, 95% CI 0.66–0.77).
Discussion
The 6 Point Scale was designed to capture systematically a genetic counselor’s selection of personal and family history of breast/ovarian cancer screeners for referral to counseling; however, counselors rated many more history screeners as warranting genetic counseling than did the 6 Point Scale. The proportion rated as high risk by genetic counselors (21%) was somewhat higher than the proportion high risk (17%) among those with a positive family history of cancer at the same mammography clinic reported in an earlier study [12], perhaps reflecting the change in guidelines. Among screeners confirmed high risk only about half (52%) had a 6 Point Scale score meeting the cut-off of 6 points; among those reclassified high risk, only 15% reached a score of 6. It seems that genetic counselors generally used the same facts as the 6 Point Scale, but set the bar lower for referral to formal genetic counseling when considering personal and family history of breast/ovarian cancer, with 86% of confirmed high risk and 80% of reclassified high risk having a 6 Point Scale score of 4 or higher. A score of 4 can be obtained by having (1) a diagnosis of breast cancer before age 50 in the patient or one relative (mother, sister, daughter, grandmother, or aunt); (2) a diagnosis of breast cancer after age 50 in the patient and one relative, in two sisters, or in relatives from two different categories (e.g., mother and aunt); (3) any blood relative with ovarian cancer; or (4) Jewish ancestors.
On the other hand, a score of 6 or higher identified most patients classified as high risk by the RST in both S1 (90% sensitivity) and S2 (94% sensitivity). The positive predictive value of the 6 Point Scale with respect to the RST was substantially lower in S1 than in S2; this is mainly due to the lower prevalence of high risk women estimated from S1. The RST was somewhat more conservative than the 6 Point Scale in rating women as high risk, with 20% of the GC confirmed high risk women classified as high risk by the 6 Point Scale but not by the RST vs. 3% classified as high risk by the RST but not by the 6 Point Scale. The RST gives less weight to breast cancer diagnosed after age 50 than does the 6 Point Scale. For example, histories that are considered high risk by the 6 Point Scale but not by the RST include (1) one relative (e.g., sister) diagnosed with breast cancer before age 50 and another after age 50, (2) three relatives (e.g., sister, grandmother, aunt) diagnosed with breast cancer after age 50, (3) Jewish ancestry and one relative diagnosed with breast cancer after age 50, and (4) one relative with ovarian cancer and one relative diagnosed with breast cancer after age 50. The 6 Point Scale uses summary histories of ovarian cancer in relatives and breast cancer in relatives of the same type (except sisters); histories that are considered high risk by the RST but not by the 6 Point Scale include (1) ovarian cancer in two relatives and (2) breast cancer in two aunts before age 50. The comparison of the 6 Point Scale with the RST gave sensitivity and specificity results [20] similar to a comparison of the RST with the FHAT [19], another risk assessment screener designed to identify patients for referral to genetic counseling.
A fair number (15%) of screeners initially classified as not high risk were reclassified as high risk by GC review. The National Comprehensive Cancer Centers Network (NCCN) Guidelines for Genetic/Familial High-Risk Assessment changed over the time that screeners were collected and reviewed by GCs as part of clinic practice. The 2006 NCCN Guidelines [21] included individuals with breast cancer under age 50, personal and/or family history of male breast cancer, two primary breast cancers, and the combination of breast and ovarian cancer in either the same individual or on the same side of the family. Additional ‘red flags’ for cancer risk assessment included, combinations of cancers known to be associated with hereditary cancer syndrome, a known mutation in the family or a history of breast or ovarian cancer in a high risk population. The 2011 NCCN Guidelines [22] criteria were amended to include individuals with triple negative breast cancer, individuals with breast cancer at any age who had at least one close relative with breast cancer at age 50 or younger, or a relative with ovarian or fallopian tube cancer. For individuals without breast cancer, a family history of 2 primary breast cancers at any age on the same side of the family or family history of ovarian cancer, or the combination of breast and pancreatic cancer were sufficient to warrant a risk assessment. These guidelines cast a wider net than either the 6 Point Scale or the RST applied with their designated cut-points.
The demographic differences among the screener risk groups are not unexpected. Women with a family history of breast cancer are likely to undergo mammography screening at a younger age compared to other women [28,29]. Whites are more likely to report having a family history of cancer than non-Whites [30,31] and immigrants [32], which may be due in part to the reluctance of families to discuss cancer [33]. Therefore, women at high risk of hereditary breast/ovarian cancer who are unaware of their family history are likely to be classified inaccurately by any screener, unless their personal history of cancer is sufficient to warrant genetic counseling.
Our study has some limitations. First, the mammography clinic history screener was not designed to capture information with calculation of the 6 Point Scale and the RST in mind; therefore, it was necessary to make some assumptions, in particular, to assume that patients understood that the column labeled “diagnosed before age 50?” referred to a diagnosis of breast cancer. In addition, the screener does not have a question about bilateral breast cancer, so it was not possible to compute later versions of the RST [34]. Although the recommendations for referral to genetic counseling have changed over time [21,22], genetic counselors were able to apply the same standard to all screeners with a final review. In addition, the weighting of data from screeners that were reclassified as high risk resulted in larger confidence intervals for sensitivity than would have been the case with unweighted data. There may have been selection bias resulting in a high prevalence of family history of breast/ovarian cancer in S2; nevertheless, the comparison of the 6 Point Scale and the RST produced similar results in both samples.
Our study also has a number of strengths, including fairly large sample sizes, comparison with review by professional genetic counselors as well as a validated screening tool, and validation in both real world and research settings.
Conclusion
The 6 Point Scale is potentially useful as a simple tool that can be administered with paper and pencil over the telephone or at primary care clinics in safety net settings by staff not trained in genetics. In a low resource setting, this tool can be used to prioritize the referral of the highest risk women to genetic counseling easily and reliably, requiring minimal time investment by primary care physicians and their staff and no financial investment in tablet computers or software. It compares favorably with the RST, a validated instrument, in terms of sensitivity and specificity, and the larger proportion of patients classified as high risk compared to the RST is acceptable. Further research is needed to develop a tool suitable for self-administration by patients with low health literacy.
Acknowledgments
This study was funded by grants from Susan G. Komen for the Cure® (SAC110049, RJ Pasick; KG090504, CP Kaplan), the California Breast Cancer Research Program (15OB-0158, CP Kaplan), and the National Cancer Institute (R01 CA129096, RJ Pasick).
References
- 1.Campeau PM, Foulkes WD, Tischkowitz MD. Hereditary breast cancer: new genetic developments, new therapeutic avenues. Hum Genet. 2008;124(1):31–42. doi: 10.1007/s00439-008-0529-1. [DOI] [PubMed] [Google Scholar]
- 2.Nelson HD, Fu R, Goddard K, Mitchell JP, Okinaka-Hu L, Pappas M, Zakher B. Risk Assessment, Genetic Counseling, and Genetic Testing for BRCA-Related Cancer: Systematic Review to Update the U.S. Preventive Services Task Force Recommendation. Rockville, MD: Agency for Healthcare Research and Quality; 2013. Evidence Synthesis No. 101. AHRQ Publication No. 12-05164-EF-1. [PubMed] [Google Scholar]
- 3.Howlader N, Noone AM, Krapcho M, Garshell J, Miller D, Altekruse SF, Kosary CL, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, Chen HS, Feuer EJ, Cronin KA. SEER Cancer Statistics Review, 1975–2012, National Cancer Institute. Bethesda, MD: 2015. Apr, http://seer.cancer.gov/csr/1975_2012/, based on November 2014 SEER data submission, posted to the SEER web site. [Google Scholar]
- 4.Moyer VA. Risk assessment, genetic counseling, and genetic testing for BRCA-related cancer in women: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2014;160(4):271–281. doi: 10.7326/M13-2747. [DOI] [PubMed] [Google Scholar]
- 5.Narod SA, Foulkes WD. BRCA1 and BRCA2: 1994 and beyond. Nat Rev Cancer. 2004;4(9):665–676. doi: 10.1038/nrc1431. [DOI] [PubMed] [Google Scholar]
- 6.Forman AD, Hall MJ. Influence of race/ethnicity on genetic counseling and testing for hereditary breast and ovarian cancer. Breast J. 2009;15(Suppl 1):S56–S62. doi: 10.1111/j.1524-4741.2009.00798.x. [DOI] [PubMed] [Google Scholar]
- 7.Armstrong K, Micco E, Carney A, Stopfer J, Putt M. Racial differences in the use of BRCA1/2 testing among women with a family history of breast or ovarian cancer. JAMA. 2005;293(14):1729–1736. doi: 10.1001/jama.293.14.1729. [DOI] [PubMed] [Google Scholar]
- 8.Levy DE, Byfield SD, Comstock CB, Garber JE, Syngal S, Crown WH, Shields AE. Underutilization of BRCA1/2 testing to guide breast cancer treatment: black and Hispanic women particularly at risk. Genet Med. 2011;13(4):349–355. doi: 10.1097/GIM.0b013e3182091ba4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Dean M, Boland J, Yeager M, Im KM, Garland L, Rodriguez-Herrera M, Perez M, Mitchell J, Roberson D, Jones K, Lee HJ, Eggebeen R, Sawitzke J, Bass S, Zhang X, Robles V, Hollis C, Barajas C, Rath E, Arentz C, Figueroa JA, Nguyen DD, Nahleh Z. Addressing health disparities in Hispanic breast cancer: accurate and inexpensive sequencing of BRCA1 and BRCA2. Gigascience. 2015;4:50. doi: 10.1186/s13742-015-0088-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Wang G, Beattie MS, Ponce NA, Phillips KA. Eligibility criteria in private and public coverage policies for BRCA genetic testing and genetic counseling. Genet Med. 2011;13(12):1045–1050. doi: 10.1097/GIM.0b013e31822a8113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Bellcross CA, Peipins LA, McCarty FA, Rodriguez JL, Hawkins NA, Hensley Alford S, Leadbetter S. Characteristics associated with genetic counseling referral and BRCA1/2 testing among women in a large integrated health system. Genet Med. 2015;17(1):43–50. doi: 10.1038/gim.2014.68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Lee R, Beattie M, Crawford B, Mak J, Stewart N, Komaromy M, Esserman L, Shaw L, McLennan J, Strachowski L, Luce J, Ziegler J. Recruitment, genetic counseling, and BRCA testing for underserved women at a public hospital. Genet Test. 2005;9(4):306–312. doi: 10.1089/gte.2005.9.306. [DOI] [PubMed] [Google Scholar]
- 13.Walcott FL, Dunn BK. Legislation in the genomic era: the Affordable Care Act and genetic testing for cancer risk assessment. Genet Med. 2015;17(12):962–964. doi: 10.1038/gim.2015.18. [DOI] [PubMed] [Google Scholar]
- 14.Wood ME, Flynn BS, Stockdale A. Primary care physician management, referral, and relations with specialists concerning patients at risk for cancer due to family history. Public Health Genomics. 2013;16(3):75–82. doi: 10.1159/000343790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jbilou J, Halilem N, Blouin-Bougie J, Amara N, Landry R, Simard J. Medical genetic counseling for breast cancer in primary care: a synthesis of major determinants of physicians' practices in primary care settings. Public Health Genomics. 2014;17(4):190–208. doi: 10.1159/000362358. [DOI] [PubMed] [Google Scholar]
- 16.Joseph G, Kaplan C, Luce J, Lee R, Stewart S, Guerra C, Pasick R. Efficient identification and referral of low-income women at high risk for hereditary breast cancer: a practice-based approach. Public Health Genomics. 2012;15(3–4):172–180. doi: 10.1159/000336419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hampel H, Bennett RL, Buchanan A, Pearlman R, Wiesner GL Guideline Development Group. American College of Medical Genetics and Genomics Professional Practice and Guidelines Committee and National Society of Genetic Counselors Practice Guidelines Committee: A practice guideline from the American College of Medical Genetics and Genomics and the National Society of Genetic Counselors: referral indications for cancer predisposition assessment. Genet Med. 2015;17(1):70–87. doi: 10.1038/gim.2014.147. [DOI] [PubMed] [Google Scholar]
- 18.Hoskins KF, Zwaagstra A, Ranz M. Validation of a tool for identifying women at high risk for hereditary breast cancer in population-based screening. Cancer. 2006;107(8):1769–1776. doi: 10.1002/cncr.22202. [DOI] [PubMed] [Google Scholar]
- 19.Gilpin CA, Carson N, Hunter AG. A preliminary validation of a family history assessment form to select women at risk for breast or ovarian cancer for referral to a genetics center. Clin Genet. 2000;58(4):299–308. doi: 10.1034/j.1399-0004.2000.580408.x. [DOI] [PubMed] [Google Scholar]
- 20.Bellcross CA, Lemke AA, Pape LS, Tess AL, Meisner LT. Evaluation of a breast/ovarian cancer genetics referral screening tool in a mammography population. Genet Med. 2009;11(11):783–789. doi: 10.1097/GIM.0b013e3181b9b04a. [DOI] [PubMed] [Google Scholar]
- 21.NCCN Clinical Practice Guidelines in Oncology™, Genetic/Familial High-Risk Assessment: Breast and Ovarian V.1.2006. 2006 [Google Scholar]
- 22.NCCN Clinical Practice Guidelines in Oncology™, Genetic/Familial High-Risk Assessment: Breast and Ovarian V.1.2011. 2011 [Google Scholar]
- 23.Kaplan CP, Livaudais-Toman J, Tice JA, Kerlikowske K, Gregorich SE, Pérez-Stable EJ, Pasick RJ, Chen A, Quinn J, Karliner LS. A randomized, controlled trial to increase discussion of breast cancer in primary care. Cancer Epidemiol Biomarkers Prev. 2014;23(7):1245–1253. doi: 10.1158/1055-9965.EPI-13-1380. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. http://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/breast-cancer-screening1. [Google Scholar]
- 25.Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159–174. [PubMed] [Google Scholar]
- 26.Efron B, Tibshirani RJ. An introduction to the bootstrap. New York: Chapman and Hall/CRC; 1998. [Google Scholar]
- 27.Izrael D, Battaglia AA, Hoaglin DC, Battaglia MP. SAS Institute Inc., Proceedings of the Twenty-Seventh SAS® Users Group International Conference. Cary, NC: SAS institute, Inc.; 2002. Use of the ROC curve and the bootstrap in comparing weighted logistic regression models. Paper 248-27. [Google Scholar]
- 28.Shah M, Zhu K, Palmer RC, Jatoi I, Shriver C, Wu H. Breast, colorectal, and skin cancer screening practices and family history of cancer in U.S. women. J Womens Health (Larchmt) 2007;16(4):526–534. doi: 10.1089/jwh.2006.0108. [DOI] [PubMed] [Google Scholar]
- 29.Gierisch JM, O'Neill SC, Rimer BK, DeFrank JT, Bowling JM, Skinner CS. Factors associated with annual-interval mammography for women in their 40s. Cancer Epidemiol. 2009;33(1):72–78. doi: 10.1016/j.cdp.2009.03.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Orom H, Kiviniemi MT, Underwood W, 3rd, Ross L, Shavers VL. Perceived cancer risk: why is it lower among nonwhites than whites? Cancer Epidemiol Biomarkers Prev. 2010;19(3):746–754. doi: 10.1158/1055-9965.EPI-09-1085. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Ramsey SD, Yoon P, Moonesinghe R, Khoury MJ. Population-based study of the prevalence of family history of cancer: implications for cancer screening and prevention. Genet Med. 2006;8(9):571–575. doi: 10.1097/01.gim.0000237867.34011.12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Orom H, Coté ML, Gonzalez HM, Underwood W, 3rd, Schwartz AG. Family history of cancer: is it an accurate indicator of cancer risk in the immigrant population? Cancer. 2008;112(2):399–406. doi: 10.1002/cncr.23173. [DOI] [PubMed] [Google Scholar]
- 33.Corona R, Rodríguez V, Quillin J, Gyure M, Bodurtha J. Talking (or not) about family health history in families of Latino young adults. Health Educ Behav. 2013;40(5):571–580. doi: 10.1177/1090198112464495. [DOI] [PubMed] [Google Scholar]
- 34.Bellcross C. Further development and evaluation of a breast/ovarian cancer genetics referral screening tool. Genet Med. 2010;12(4):240. doi: 10.1097/GIM.0b013e3181d4bc3a. [DOI] [PubMed] [Google Scholar]