Development and validation of a Breast Cancer Genetic Counseling Knowledge Questionnaire

Abstract

Women who undergo genetic counseling concerning their increased risk of developing breast cancer confront large quantities of complex information in a short period of time. Clinical reports have suggested that many women may not retain what they learned during counseling. A validated questionnaire to measure their knowledge, however, is lacking. In this study, we describe the development and validation of a questionnaire to assess knowledge of information typically included in genetic counseling for breast cancer. Items were empirically derived from detailed content analyses of actual genetic counseling sessions. The instrument's content validity was high, as evidenced by high levels of independent interrater agreement (0.93) on items. Subsequent data reduction and confirmatory factor analytic techniques yielded a highly reliable (alpha = 0.92) 27-item Breast Cancer Genetic Counseling Knowledge Questionnaire (BGKQ). Direct comparison of this questionnaire to a scale previously developed in the literature (BCHK; [Breast Cancer Res. Treat. 53 (1999) 69]) supported the utility of the new questionnaire for evaluation of knowledge after counseling. Compared to non-counseled groups (n = 45), women who had undergone genetic counseling (n = 28) scored significantly higher (P < 0.0001) on the BGKQ, but not on the other questionnaire, establishing the BGKQ's criterion validity. The BGKQ may, thus, provide a useful clinical and research tool for assessing knowledge of information provided during genetic counseling and exploring the potential impact of distress on knowledge, as well as the impact of knowledge on screening behaviors.

Full text

Patient Education and Counseling 56 (2005) 182–191
Development and validation of a breast cancer genetic
counseling knowledge questionnaire
Joel Erblicha,∗, Karen Brownb, Youngmee Kimc, Heiddis B. Valdimarsdottira,
Barbara E. Livingstond, Dana H. Bovbjerga
aDerald H. Ruttenberg Cancer Center, Mount Sinai School of Medicine, One Gustave L. Levy Place, Box 1130, New York, NY 10029-6574, USA
bDepartment of Human Genetics, Mount Sinai School of Medicine, New York, NY 10029-6574, USA
cBehavioral Research Center, American Cancer Society, New York, NY 10029-6574, USA
dOncology Care Center, Mount Sinai School of Medicine, New York, NY 10029-6574, USA
Received 10 October 2003; received in revised form 23 January 2004; accepted 20 February 2004
Abstract
Women who undergo genetic counseling concerning their increased risk of developing breast cancer confront large quantities of complex
information in a short period of time. Clinical reports have suggested that many women may not retain what they learned during counseling.
A validated questionnaire to measure their knowledge, however, is lacking. In this study, we describe the development and validation of a
questionnaire to assess knowledge of information typically included in genetic counseling for breast cancer. Items were empirically derived
from detailed content analyses of actual genetic counseling sessions. The instrument’s content validity was high, as evidenced by high
levels of independent interrater agreement (0.93) on items. Subsequent data reduction and confirmatory factor analytic techniques yielded
a highly reliable (alpha =0.92) 27-item Breast Cancer Genetic Counseling Knowledge Questionnaire (BGKQ). Direct comparison of this
questionnaire to a scale previously developed in the literature (BCHK; [Breast Cancer Res. Treat. 53 (1999) 69]) supported the utility of the
new questionnaire for evaluation of knowledge after counseling. Compared to non-counseled groups (n=45), women who had undergone
genetic counseling (n=28) scored significantly higher (P<0.0001) on the BGKQ, but not on the other questionnaire, establishing
the BGKQ’s criterion validity. The BGKQ may, thus, provide a useful clinical and research tool for assessing knowledge of information
provided during genetic counseling and exploring the potential impact of distress on knowledge, as well as the impact of knowledge on
screening behaviors.
© 2004 Elsevier Ireland Ltd. All rights reserved.
Keywords: Breast cancer; Genetic counseling; Knowledge; Decision making; Test development
1. Introduction
Breast cancer is the most frequently diagnosed cancer
in the world; it is estimated that over 1 million people are
newly diagnosed with the disease in annually [1]. Current
estimates from the US indicate that women’s cumulative
lifetime risk of developing breast cancer is about 12% [2].In
addition, women with the disease in one or more first-degree
relatives are at even higher risk for the disease [3], esca-
lating further with the presence of other risk factors (e.g.
early menarche, nulliparity, etc.) [4]. The past decade has
seen major advances in our understanding of this increased
risk for breast cancer. A study by Claus et al. [5] estimated
∗Corresponding author. Tel.: +1-212-659-5516; fax: +1-212-659-5507.
E-mail address: joel.erblich@mssm.edu (J. Erblich).
that, while only a small minority of women (5–10%) carry
known mutations in breast cancer susceptibility genes,
those who are carriers are at extremely high-lifetime risk of
developing the disease, upwards of 92%. Subsequent stud-
ies identified BRCA1 and BRCA2 as strong risk factors for
the development of breast cancer, with estimates of up to
85% cumulative lifetime risk for carriers of these mutations
[5–7], and a lifetime risk of ovarian cancer of 40–60% for
carriers of BRCA1 [6]. Moreover, studies have indicated
that exogenous factors, such as low parity and young age at
last childbirth, can further increase risk for these diseases
in women who are carriers of susceptibility genes [4].
The rapid increase in risk information available has engen-
dered a need to better communicate to women the numerous
risk factors, their implications for women’s disease risks,
as well as the implications for their family members’ risks.
Indeed, many women with family histories of breast cancer
0738-3991/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.pec.2004.02.007

J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191 183
undergo genetic counseling, and often genetic testing for
BRCA1 and BRCA2/other susceptibility genes. During these
“pretest” genetic counseling sessions, women are faced with
complex information about basic genetic transmission, im-
plications of carrying mutations on their health, implications
for their family members’ health, more intensive screening
recommendations, and options for prophylaxis, including
chemoprevention (e.g. Tamoxifen) and mastectomy. These
considerations, while critical in and of themselves, may
also, to some extent, serve as a woman’s basis for making
the difficult decision of whether or not to undergo genetic
testing and/or to be notified of test results [8].
Previous work from our group [9] and others [10] has
demonstrated that women at elevated risk for breast can-
cer can experience substantial levels of psychological dis-
tress, and it comes as no surprise that women facing the
threat of learning of their elevated breast cancer risk also
exhibit high levels of distress, especially around the time of
genetic counseling [11–13]. Literature in the cognitive sci-
ences has long demonstrated that information processing at
times of stress can be severely impaired [14]. In a recent
report, we have demonstrated that such impairments exist
among women with family histories of breast cancer [15].
Consistent with this research, clinical reports have suggested
that while women undergoing genetic counseling do come
away more informed than they had been prior to counseling,
they do not retain much of what had been presented during
the session [16]. Other factors such as cultural background
and education level may have an impact on knowledge, as
well [17]. Unfortunately, few empirical studies have exam-
ined women’s processing of information presented in pretest
genetic counseling for breast cancer risk. One significant
barrier to this endeavor is the lack of an empirically de-
rived, psychometrically validated instrument to assess such
knowledge retention. In the first phase of the present study,
a novel breast cancer knowledge questionnaire designed to
assess women’s knowledge of information presented during
breast cancer genetic counseling is generated. In the second
phase, the initial validation and psychometric evaluation of
the instrument is described, and the utility of the novel in-
strument is compared to another instrument previously de-
veloped in the literature to assess knowledge among low- to
moderate-risk women [18].
2. Method—Phase 1
2.1. Overview
In the first phase of this study, we generated a question-
naire based on the content analyses of three genetic coun-
seling sessions held at the Mount Sinai School of Medicine.
As part of initial data reduction, questions were screened for
clarity and readability by expert genetic counselors. Content
validity was then assessed using interrater agreement (IR)
indices.
2.2. Item development
To generate items for the Breast Cancer Genetic Coun-
seling Knowledge Questionnaire (BGKQ), we performed a
detailed content analysis of three separate pretest genetic
counseling sessions of three different counselees, two who
had histories of breast cancer in first-degree relatives, and
one with a family history of breast and ovarian cancer, who
had already had breast cancer herself. The sessions were
conducted by a cancer genetic counselor (K.B.) at the Mount
Sinai Medical Center, consistent with consensus guidelines
for breast cancer genetic counseling [19]. The sessions lasted
about 1h each, and covered a broad range of topics, from
basic Mendelian transmission to information about prophy-
lactic measures. During two of the sessions, an investigator
took detailed notes in a corner of the room, with the patients’
consent. The key points of a third session were transcribed by
a counselee herself, after the session. The approach yielded
content from both the viewpoint of a counselee and an in-
dependent observer. A pool of potential questions were then
formed, based on the material presented in the sessions. Be-
cause session content followed consensus guidelines, there
was much overlap of information presented in these three
sessions. It was deemed, therefore, that three sessions were
sufficient for the development of appropriate questionnaire
items.
2.3. Data analyses
Four different genetic counselors from Mount Sinai’s
Department of Human Genetics screened the questions for
appropriateness. Counselors were asked to agree or dis-
agree that the each of the questions was appropriate for the
questionnaire. To meet selection criteria, knowledge ques-
tionnaire items needed to have an interrater agreement of
at least 0.75 (3/4 agreed on appropriateness). Items that did
not meet criterion IR were either modified to increase clar-
ity or dropped. Overall IR was calculated as the mean of the
IRs for all retained items, and served as an index of content
validity.
3. Results—Phase 1
3.1. Item characteristics
Content analysis of the three genetic counseling sessions
yielded an initial pool of 50 items. To minimize the burden
of the questionnaire, items were generated with true/false
responses whenever possible. As a result, there were 36
true/false format items, and 14 multiple choice items. Con-
sistent with session content, the questionnaire included
items assessing basic Mendelian genetic information, trans-
mission of genetic risk for breast cancer, implications of
BRCA1/BRCA2 carrier status for risk of developing other
cancers, implications of carrier status for one’s own risk,

184 J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191
implications for family members’ risks, implications for
one’s screening behavior, and information regarding the
utility of prophylactic measures.
3.2. Content validity
Results of the four genetic counselors’ ratings of the ques-
tionnaire items indicated high levels of content validity. Of
the 50 items, 39 had IRs of 1.00 (4/4 endorsed the item), 8
had IRs of 0.75, 2 had IRs of 0.50, and 1 item received an
endorsement from only one of the raters. Overall IR for the
instrument was 0.92. After re-examination of questionable
items, five were dropped from the questionnaire. In addi-
tion, based on recommendations of the raters and the prin-
cipal genetic counselor, six items were modified to increase
clarity and re-rated. As a result of the modifications, the fi-
nal IR for the 45-item version of the questionnaire increased
slightly, to 0.93.
4. Method—Phase 2
4.1. Overview
In Phase 2 of the study, women (n=75) completed
the initial 45-item version of the Breast Cancer Genetic
Counseling Knowledge Questionnaire (BGKQ), along with
the Breast Cancer and Heredity Knowledge Questionnaire
(BCHK), an 11-item instrument previously developed to
assess knowledge among lower risk women [18]. Con-
firmatory factor analysis was employed to establish the
psychometric properties of each questionnaire. In addition,
to establish criterion validity, scores of women who under-
went genetic counseling were compared to those of women
who did not. Finally, the relative utility of the two instru-
ments for assessing knowledge of information presented in
genetic counseling of high-risk women was compared.
4.2. Participants
To validate the BGKQ, women who had undergone
breast cancer genetic counseling (n=28) at Mount Sinai’s
Department of Human genetics were recruited. Participants
were sequential counselees who visited the chief cancer ge-
netic counselor at Mount Sinai (K.B.). For purposes of com-
parison, two additional groups of women were recruited:
(1) nurses employed at Mount Sinai (n=26) thought to
be familiar with the type of complex health information
covered in genetic counseling, recruited by advertisement,
and (2) employees of Mount Sinai who were not health care
providers (e.g. administrative staff), who were “graduates”
of a recent study of healthy women with different family
histories of breast cancer at the medical center (n=21).
In the first comparison group, nurses with (n=11) and
without (n=15) oncology certification were recruited, as
they were thought to have differing levels of familiarity with
the information. Similarly, in the second comparison group,
employees with (n=11) and without (n=10) histories of
breast cancer (FH+,FH−) in one or more first-degree rela-
tives were recruited, as they were thought to have different
levels of knowledge as well. In sum, five groups of women
(n=75), including one Counselee group and four Compar-
ison groups, participated in Phase 2 of the study.
4.3. Procedure
Counselees were given the BGKQ-45 and the BCHK
[18] by their genetic counselor (K.B.) at the end of their
genetic counseling session. The BCHK was administered
in order to compare the utility of the BGKQ for assess-
ing genetic counseling knowledge with an instrument that
has already been validated for use with low-to-moderate
risk women, and which may not be as appropriate to as-
sess knowledge of information from genetic counseling of
higher risk women. They were instructed to complete the
questionnaire within two weeks of the session and anony-
mously return it in a prepaid mailer. A cover letter attached
to each questionnaire described the purpose of the study (to
help develop a valid questionnaire), ensured respondents that
their responses were voluntary and anonymous, labeled only
with a code indicating that they belonged to the “Counse-
lee Group”, and requested that they did not consult outside
references to answer the questions.
Nurses were recruited by advertisement to a “ques-
tionnaire lunch”, during which time they completed the
BGKQ-45 and the BCHK. Advertisements were placed
around the medical center with more concentrated place-
ment in and around the Oncology Care Center, to increase
the likelihood of recruiting a subset of oncology certified
nurses. As in the Counselee Group, nurses’ questionnaires
included a cover letter describing the study. Questionnaires
administered to the “Nurse Group” included an additional
question assessing whether or not they were certified in
oncology.
Employees not involved in provision of health care con-
sisted of graduates of a previous study at the cancer center.
Women were re-contacted and asked if they could be sent
a questionnaire. As above, questionnaires were anonymous,
respondents were asked to return them within two weeks (in
a prepaid mailer), and were asked not to consult any ref-
erences in completing the questionnaire. Study procedures
were approved by the Mount Sinai IRB.
4.4. Data reduction and analyses
The first step of the analyses was focused on item re-
duction. Items that were correctly answered or incorrectly
answered by at least 80% of the sample were dropped, as
they were deemed to be too simple or too difficult, respec-
tively. Next, the data were subjected to confirmatory factor
analysis using Normal Ogive Harmonic Analysis Robust
Method (NOHARM) analysis software [20,21], specially

J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191 185
designed to analyze instruments with binary response items
such as the BGKQ and the BCHK (i.e. correct/incorrect).
In this analysis, a parsimonious one factor solution was
compared to a priori solutions of two factors (with items
grouped into two categories: basic genetic information and
clinical implications/screening recommendations) and three
factors (basic genetic information, implications, screening
recommendations). Consistent with recommendations for
scales with binary items [20,21], those items with factor
loadings above 0.5 were retained and used in the next step.
In the second step of the analyses, we assessed criterion
validity of the reduced BGKQ. An ANOVA of the Groups’
responses to the instrument was performed, followed by
planned groupwise comparisons using two-tailed tests. A
higher mean score in the Counselee Group would suggest
good criterion validity. Finally, the BGKQ and the BCHK
were entered into a simultaneous logistic regression anal-
ysis as a “head-to-head” test of the relative contributions
of the two scales to predicting group membership. For the
purposes of this last analysis, Group was dichotomized as
“Counselee” versus “Other”.
5. Results—Phase 2
5.1. Response rate
A total of 105 questionnaires were administered, either
in person (genetic counselees, nurses) or via mail (employ-
ees not involved in health care provision). A total of 77
women returned the questionnaire, an overall response rate
of 73.3%. The response rates were similar across groups: 40
questionnaires were circulated to genetic counselees, yield-
ing 30 respondents (75% response rate), 35 questionnaires
were administered to the nurses, yielding 26 respondents
(74% response rate), and 30 questionnaires were given out
to the employees not involved in health care provision,
yielding 21 respondents (70% response rate). Two of these
respondents’ questionnaires were dropped from the analyses
because they each left an entire page of the questionnaire
blank, yielding a final sample size of 75.
Table 1
Demographic characteristics
Group Age (±S.D.) Education
(% W/graduate degree) Ethnicity
(% Caucasian) Income
(% >$ 60K per year) Marital status
(% currently married)
Counselees (n=28) 47.4 ±12.0a46.4 96.4c75.0e78.6
Nurses
Oncology (n=11) 40.6 ±9.4 63.6 54.5d90.9e54.5
General (n=15) 41.9 ±8.4 46.7 53.3d85.7e40.0
Other employees
FH+(n=11) 39.5 ±12.1 27.3 9.1d36.4f63.6
FH−(n=10) 37.4 ±8.1b30.0 40.0d40.0f50.0
Total (n=75) 42.8 ±11.0 44.0 61.3 68.9 61.3
Differing superscripts in each column are significant at P<0.05.
5.2. Demographic characteristics
The mean age of the women in the sample was 42.8
years (S.D.=11.0). Not surprisingly, as many of the
women were nurses and other professionals, 44% of the
sample reported having some postgraduate education, and
69% reported household annual income levels of at least
$ 60,000.00. Sixty-one percent of the sample reported being
Caucasian, 15% reported being Black, 13% reported being
Hispanic, and 11% reported other racial/ethnic backgrounds.
In addition, 61% reported being currently married, 20%
reported being never married, 17% reported being divorced
or separated, and one respondent reported being widowed.
Comparisons of the five groups of women revealed few
systematic differences on demographic variables (Table 1).
Counselees were older than FH−non-provider employees,
and had the largest percentage of Caucasian women (96%).
Counselees and nurses reported higher incomes than the
FH+and FH−employees who were not directly involved
in health care provision. To take a conservative approach,
age, ethnicity, and income, were included as covariates in
subsequent analyses of group differences.
5.3. Data reduction and construct validation
In the first phase of data reduction, item analyses were
performed to identify questions that were either too easy
(>80% correct) or too difficult (<20% correct). Results of
these analyses revealed that, of the 45 items, four were
too simple and four were too difficult (items labeled in
Appendix A). Because it was possible that the difficult
items were good “discriminators” between the Groups, we
ran individual group analyses on these items and found
that Groups answered the four items correctly at compara-
ble rates. Based on these findings, all eight questions were
dropped from the questionnaire.
The remaining 37 items were the entered into a confirma-
tory factor analysis, as described above, in which a parsimo-
nious one-factor solution was tested against two- (“basic ge-
netic information”, “clinical implications”) and three-factor
solutions (“basic genetic information”, “risk information”,

186 J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191
Table 2
Fit data for BGKQ and BCHK
Solution Sum of squared
residuals (SSR) Root mean squared
residuals (RMSR) Tanaka Goodness-of-Fit
index (TGFI)
BGKQ-37, one factor 0.314 0.022 0.86
BGKQ-37, two factors 0.304 0.021 0.87
BGKQ-37 three factors 0.304 0.021 0.87
BGKQ-27 one factor (final version) 0.178 0.025 0.93
BCHK 0.023 0.021 0.93
“screening recommendations”) hypothesized a priori. Re-
sults indicated that all three models were an excellent
fit with the data (Table 2), with Tanaka Goodness-of-Fit
indices (TGFI) ranging from 0.86 to 0.87. Because the
multifactorial models did not significantly fit the data better
than the single-factor solution, the parsimonious one-factor
solution was accepted. Examination of factor loadings re-
vealed that 27 of the 37 items loaded on the single factor
at or above the 0.5 level (see Appendix A for final 27-item
version, items asterisked). These items were retained for
the final version of the questionnaire. The 27-item version
of the BGKQ displayed excellent internal consistency (re-
liability), with a Cronbach’s alpha of 0.92 and a final TGFI
of 0.93. Cronbach’s alpha for the 37-item version was sim-
ilarly high, at 0.91. In contrast, while a one-factor solution
to the BCHK demonstrated an excellent fit (TGFI =0.93),
Cronbach’s alpha for the BCHK was only 0.48, exhibiting
poor internal consistency. Consistent with these findings,
only two of the items on the BCHK had factor loadings of at
least 0.5.
5.4. Criterion validation
To determine whether or not the 27-item BGKQ demon-
strated criterion validity, we conducted an ANOVA, com-
paring the five groups’ scores, controlling for demograph-
ics, as above. A significant main effect of Group was
observed; F(4,67) =8.39, P<0.0001. Planned compar-
isons of covariate-adjusted means (least-squares) indicated
that counselees scored significantly higher than all other
groups on the BGKQ-27. Also as expected, oncology
nurses scored higher than general practice nurses and other
FH+employees, and marginally higher than FH−employ-
ees (P<0.07). In contrast, groups did not differ signifi-
cantly in their performance on the BCHK; F(4,67) =0.96,
P<0.45 (see Table 3).
To rule out independent effects of family history of
breast cancer on knowledge, we compared women with and
without such family histories, independent of the Group. In
each of the nurse subgroups, one woman reported having
a family history of breast cancer in a first-degree rela-
tive. Not surprisingly, 60% of the counselees were FH+,
as well (the remainder having family histories of ovarian
cancer or multiple second-degree relatives with breast can-
cer). We found, however, that FH+women (n=30–11
FH+employees, two nurses, and 17 counselees) and FH−
Table 3
ANOVA of performance on the BGKQ-27 and the BCHK by Group
Group BGKQ-27
score (±S.E.) BCHK score
(±S.E.)
Counselees (n=28) 20.7 ±1.2a6.9 ±0.4
Nurses
Oncology (n=11) 15.7 ±1.7b,c,e7.1 ±0.6
General (n=15) 11.4 ±1.4b,d6.0 ±0.5
Other employees
FH+(n=11) 9.5 ±1.9b,d5.9 ±0.6
FH−(n=10) 11.2 ±1.8b,f6.2 ±0.6
Groups (a–b, c–d) differ significantly at P<0.05 and Groups (e–f) differ
marginally at P<0.07.
Table 4
Logistic regression analysis of BGKQ-27 and BCHK
Instrument Beta S.E.
beta Wald χ2P-value Odds
ratio 95% CI
BGKQ-27 0.36 0.11 10.44 0.002 1.43 1.15, 1.78
BCHK −0.14 0.27 0.29 0.59 0.87 0.51, 1.46
women (n=45) did not differ on their BGKQ-27 scores,
nor did they differ in their performance on the BCHK, sug-
gesting that effects were not due to family history status
per se.
Finally, to compare the differential utility of the BGKQ-27
and the BCHK head to head, we performed a simultane-
ous entry logistic regression analysis, using the two instru-
ments as predictors of group membership, and age, income,
and ethnicity as covariates. To simplify analyses, Group was
dichotomized as Counselee versus Other. Results indicated
that the BGKQ-27 accurately predicted group membership;
χ2(1) =10.44, P<0.002, but the BCHK did not; χ2(1)
=0.29, P<0.59 (see Table 4).
6. General discussion
In this report, we describe the development of a question-
naire assessing knowledge of information generally provided
during breast cancer genetic counseling and the empirical
evaluation of its validity. Based on content analysis of actual
breast cancer genetic counseling sessions, the instrument
evidenced excellent initial psychometric properties, includ-
ing high levels of content validity. Evaluation with selected

J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191 187
samples of test-takers demonstrated that the instrument has
excellent reliability and criterion validity, and evidences a
confirmed single-factor structure. Reliability and validity of
this new instrument exceeded that of an existing measure
(BCHK) previously employed in the literature. Unlike the
BCHK, which was developed to assess knowledge in the
general population [18], items on the BGKQ-27 were specif-
ically generated to assess information from genetic coun-
seling sessions. Thus, while the BGKQ-27 is longer than
this other instrument, it may prove more useful for assess-
ment of knowledge specifically acquired during counseling.
As breast cancer genetic counseling becomes increasingly
standardized to cover key areas of information, the utility of
a standardized validated questionnaire to assess for knowl-
edge will become even greater. Thus, it is anticipated that
the BGKQ-27 may become increasingly useful to clinicians
interested in assessing how much information their patient
obtain/retain during counseling.
As indicated above, studies have suggested that distress
and worry associated with risk for breast cancer [9,10] and
attending genetic counseling [11–13] may have an impact on
processing of information [14–16]. The BGKQ-27 may pro-
vide one way to more accurately assess the impact of stress
and other factors (e.g. educational background) on knowl-
edge. In addition, initial reports have raised the possibility
that knowledge may be an important predictor of decisions
to undergo testing [8], engaging in health behaviors, and ad-
herence to screening guidelines [22,23]. Indeed, a major goal
of counseling is to allow women to make informed decisions
about both their own health and that of their family members
[23,24]. While studies have examined the impact of coun-
seling on behaviors such as undergoing genetic testing [8]
and screening/prophylactic treatments [22–24], researchers
have not had a validated questionnaire to examine whether
or not women who come away from counseling better able
to make informed decisions about their health care (e.g.
changing dietary habits, screening). Similarly, research must
still determine whether or not levels of knowledge persist
over time, and whether they predict subsequent compliance
with recommended screening guidelines and/or other po-
tentially risk modifying behaviors (e.g. diet, exercise). One
important approach to test the possibility that knowledge is
gained through genetic counseling, short of a clinical trial,
would be to compare women’s performance on the BGKQ
before and after counseling. The possibility that pretesting
might influence the counselees’ behavior during the session
(i.e. requesting that the counselor address questions that the
counselee could not answer during the pretest), however,
would need to be addressed. From a clinical standpoint,
such “pretest sensitization” may actually prove useful—as
a method of guiding the counseling session toward those
points that require the most attention for that individual.
Larger scale randomized trials of this possibility may be
warranted.
It must be emphasized, though, that while the BGKQ-27
exhibited sound psychometric properties in this sample, it
would be important to validate the instrument on additional
diverse samples, varying in cultural and demographic char-
acteristics, to confirm generalizability and to address the
inevitable statistical phenomenon of “shrinkage”, the ap-
parent decrease in statistical effects upon cross-validation.
In addition, the possibility that a shorter version of this
27-item instrument could be developed to perform equally
well needs to be examined in a larger sample. A brief ver-
sion of the BGKQ would further enhance its utility in both
the research and clinical settings. In addition, as genetic
counseling is a dynamic process, with the content being
modified and updated as new research becomes available, a
test such as the BGKQ-27 would also need to be modified
periodically to keep up with such changes. The instrument
might also require the addition of site/session-specific items
if genetic counseling sessions do not follow consensus
protocols. At the same time, it must be emphasized that
the empirical approach to discarding items (e.g. those that
were too simple or too difficult) necessarily resulted in a
more streamlined questionnaire, representing only a sam-
pling of the total information covered. In spite of this fact,
though, the questionnaire accurately discriminated groups,
indicating that the 27-item version is a valid index of
knowledge.
Finally, the use of this approach to develop brief knowl-
edge questionnaires for other types of genetic counseling,
and more generally, for other types of critical medical
interactions (i.e. counseling about risk for cardiovascular
disease), might prove useful, as well. In sum, the reliabil-
ity and validity of the newly developed BGKQ-27 suggest
its potential utility. Research on the knowledge obtained
by women during genetic counseling may help clinicians
better understand women’s decisions about undergoing
genetic testing, and engaging in health and screening be-
haviors, ultimately enhancing informed decision-making
and compliance with recommended health and screening
guidelines.
Acknowledgements
This research was sponsored in part by grants from the Na-
tional Cancer Institute (#R01 CA72457—Bovbjerg) and the
Department of Defense (#DAMD 17-99-1-9305—Erblich;
DAMD 17-99-1-9303—Bovbjerg). We are required to in-
dicate that the content of the information contained in this
report does not necessarily reflect the position or policy of
the United States Government.

188 J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191
Appendix A

J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191 189

190 J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191

J. Erblich et al./Patient Education and Counseling 56 (2005) 182–191 191
References
[1] Ferlay J, Bray F, Pisani P, Parkin D. GLOBOCAN 2000: cancer
incidence, mortality and prevalence worldwide, version 1.0. IARC
CancerBase No. 5. Lyon: IARC Press; 2001.
[2] Ries L, Eisner M, Kosary C, Hankey B, Miller B, Clegg L, Ed-
wards B, editors. SEER cancer statistics review, 1973–1999. Bethesda
(MD): National Cancer Institute; 2002.
[3] Easton D. Familial risks of breast cancer. Breast Cancer Res
2002;4:179–81.
[4] Gail M, Brinton L, Byar D, Corle D, Green S, Schairer C, Mulvihill
J. Projecting individualized probabilities of developing breast cancer
for white females who are being examined annually. J Natl Cancer
Inst 1989;81:1879–86.
[5] Claus E, Risch N, Thompson W. Genetic analysis of breast can-
cer in the cancer and steroid hormone study. Am J Hum Genet
1991;48:232–42.
[6] Easton D, Ford D, Bishop D. Breast and ovarian cancer incidence in
BRCA1-mutation carriers. Breast Cancer Linkage Consortium. Am
J Hum Genet 1995;56:265–71.
[7] Wooster R, Bignell G, Lancaster J, Swift S, Seal S, Mangion J,
Collins N, Gregory S, Gumbs C, Micklem G. Identification of the
breast cancer susceptibility gene BRCA2. Nature 1995;378:789–92.
[8] Thompson H, Valdimarsdottir H, Duteau-Buck C, Guevarra J, Bovb-
jerg D, Richmond-Avellaneda C, Amarel D, Godfrey D, Brown K,
Offit K. Psychosocial predictors of BRCA counseling and testing de-
cisions among urban African-American women. Cancer Epidemiol
Biomarkers Prev 2002;11:1579–85.
[9] Erblich J, Bovbjerg D, Validimarsdottir H. Looking forward and
back: distress among women at familial risk for breast cancer. Ann
Behav Med 2000;22:53–9.
[10] Kash K, Holland J, Halper M, Miller D. Psychological distress and
surveillance behaviors of women with a family history of breast
cancer. J Natl Cancer Inst 1992;84:24–30.
[11] Bish A, Sutton S, Jacobs C, Levene S, Ramirez A, Hodgson S.
Changes in psychological distress after cancer genetic counselling:
a comparison of affected and unaffected women. Br J Cancer
2002;86:43–50.
[12] Cull A, Anderson E, Campbell S, Mackay J, Smyth E, Steel
M. The impact of genetic counseling about breast cancer risk on
women’s risk perceptions and levels of distress. Br J Cancer 1999;79:
501–8.
[13] Meiser B, Halliday J. What is the impact of genetic counselling in
women at increased risk of developing hereditary breast cancer? A
meta-analytic review. Soc Sci Med 2002;54:1463–70.
[14] McEwen B, Sapolsky R. Stress and cognitive function. Curr Opin
Neurobiol 1995;5:205–15.
[15] Erblich J, Montgomery G, Valdimarsdottir H, Cloitre M, Bovbjerg
D. Biased cognitive processing of cancer-related information among
women with family histories of breast cancer: evidence from a cancer
Stroop task. Health Psychol 2003;22:235–44.
[16] Hallowell N, Murton F, Statham H, Green J, Richards M. Women’s
need for information before attending genetic counselling for familial
breast or ovarian cancer: a questionnaire, interview, and observational
study. Br Med J 1997;314:281–3.
[17] Weil J. Multicultural education and genetic counseling. Clin Genet
2001;59:143–9.
[18] Ondrusek N, Warner E, Goel V. Development of a knowledge scale
about breast cancer and heredity (BCHK). Breast Cancer Res Treat
1999;53:69–75.
[19] American College of Medical Genetics Foundation. Genetic sus-
ceptibility to breast and ovarian cancer: assessment, counseling and
testing guidelines, executive summary. New York (NY): American
College of Medical Genetics; 1999.
[20] Fraser C, McDonald R. NOHARM: least squares item factor analysis.
Multivariate Behav Res 1988;23:267–9.
[21] McDonald R. Normal-ogive multidimensional model. In: van der
Linden W, Hambleton R, editors. Handbook of modern item response
theory. New York (NY): Springer-Verlag; 1997. p. 257–86.
[22] Butow P, Lobb E, Meiser B, Barratt A, Tucker K. Psychological
outcomes and risk perception after genetic testing and counselling
in breast cancer: a systematic review. Med J Aust 2003;178:77–81.
[23] Lynch H, Lemon S, Durham C, Tinley S, Connolly C, Lynch J,
Surdam J, Orinion E, Slominski-Caster S, Watson P, Lerman C,
Tonin P, Lenoir G, Serova O, Narod S. A descriptive study of
BRCA1 testing and reactions to disclosure of test results. Cancer
1997;79:2219–28.
[24] Morris K, Johnson N, Krasikov N, Allen M, Dorsey P. Genetic coun-
seling impacts decision for prophylactic surgery for patients perceived
to be at high risk for breast cancer. Am J Surg 2001;181:431–3.