Original Article
Attitudes Toward Molecular Testing for Personalized Cancer Therapy Rafeek A. Yusuf, MD1; Deevakar Rogith, MD1; Shelly R. A. Hovick, PhD2; Susan K. Peterson, PhD, MPH3; Allison M. Burton-Chase, PhD4; Bryan M. Fellman, MS5; Yisheng Li, PhD5; Carolyn McKinney, BSN6; Elmer V. Bernstam, MD1,7; and Funda Meric-Bernstam, MD6,8
BACKGROUND: This study assessed attitudes of breast cancer patients toward molecular testing for personalized therapy and research. METHODS: A questionnaire was given to female breast cancer patients presenting to a cancer center. Associations between demographic and clinical variables and attitudes toward molecular testing were evaluated. RESULTS: Three hundred eight patients were approached, and 100 completed the questionnaire (a 32% response rate). Most participants were willing to undergo molecular testing to assist in the selection of approved drugs (81%) and experimental therapy (59%) if testing was covered by insurance. Most participants were white (71%). Even if testing was financially covered, nonwhite participants were less willing to undergo molecular testing for the selection of approved drugs (54% of nonwhites vs 90% of whites, odds ratio [OR] 5 0.13, P 5.0004) or experimental drugs (35% vs 68%, OR 5 0.26, P 5.0072). Most participants (75%) were willing to undergo a biopsy to guide therapy, and 46% were willing to undergo research biopsies. Nonwhite participants were less willing to undergo research biopsies (17% vs 55%, OR 5 0.17, P 5.0033). Most participants wanted to be informed when research results had implications for treatment (91%), new cancer risk (90%), and other preventable/treatable diseases (87%). CONCLUSIONS: Most patients were willing to undergo molecular testing and minimally invasive procedures to guide approved or experimental therapy. There were significant differences in attitudes toward molecular testing between racial groups; nonwhites were less willing to undergo testing even if the results would guide their own therapy. Novel approaches are needed to prevent disparities in the delivery of genomically informed care and to increase minority C 2014 American Cancer Society. participation in biomarker-driven trials. Cancer 2015;121:243-50. V KEYWORDS: molecular testing, disparities, personalized cancer therapy, biomarkers, questionnaire, survey.
INTRODUCTION Although the concept of individualizing treatment is more than a century old,1,2 recent advances in molecular diagnostics have enabled personalized cancer therapy based on molecular characteristics of the patient or tumor and novel molecular marker–driven clinical trials. Although we know much about factors that influence general participation in biomedical research, little is known specifically about patients’ willingness to participate in testing for personalized cancer therapy and their perceptions of participation in biomarker-driven research and targeted therapy trials.3,4 Furthermore, molecular testing brings with it unique problems, such as the potential for the discovery of germline variants or mutations with health implications for the patient and the patient’s relatives. Little is known about patient wishes for the return of these incidental research results. In this study, we assess breast cancer patients’ attitudes toward molecular testing for personalized therapy and research as well as the return of incidental research results. MATERIALS AND METHODS Study Participants and Recruitment
This study was conducted at the University of Texas MD Anderson Cancer Center (MD Anderson), an academic cancer center located within the Texas Medical Center in Houston, Texas. The study was approved by the MD Anderson
Corresponding author: Funda Meric-Bernstam, MD, University of Texas MD Anderson Cancer Center, 1400 Holcombe Boulevard, FC8.3044, Houston, TX 77030; Fax: (713) 563-0566; [email protected] 1 School of Biomedical Informatics, University of Texas Health Science Center at Houston, Houston, Texas; 2School of Communication, Ohio State University, Columbus, Ohio; 3Department of Behavioral Sciences, University of Texas MD Anderson Cancer Center, Houston, Texas; 4Department of Basic and Social Sciences, Albany College of Pharmacy and Health Sciences, Albany, New York; 5Department of Biostatistics, University of Texas MD Anderson Cancer Center, Houston, Texas; 6Department of Surgical Oncology, University of Texas MD Anderson Cancer Center, Houston, Texas; 7Division of General Internal Medicine, Department of Internal Medicine, Medical School, University of Texas Health Science Center at Houston, Houston, Texas; 8Department of Investigational Cancer Therapeutics, University of Texas MD Anderson Cancer Center, Houston, Texas.
Additional supporting information may be found in the online version of this article. DOI: 10.1002/cncr.28966, Received: February 19, 2014; Revised: June 13, 2014; Accepted: June 25, 2014, Published online September 10, 2014 in Wiley Online Library (wileyonlinelibrary.com)
Cancer
January 15, 2015
243
Original Article
institutional review board. Patients registering at the breast center between October and December 2012 were invited to participate in a questionnaire study. Eligibility criteria included (1) being 18 years of age or older, (2) being female, (3) being fluent in English, and (4) having been previously diagnosed with breast cancer. Three hundred eight consecutive patients were approached, and 100 completed the questionnaire (a 32% response rate). Study Instrument
Because of the paucity of existing measures from which questions specific to this study could be adapted, new measures were developed that were informed by the literature and previously validated measures4-9 to assess attitudes of breast cancer patients toward molecular testing for personalized therapy and research. The questionnaire was developed and reviewed for content validity by a team of investigators with expertise in genetics, survey research, behavioral science, health communication, and bioinformatics. The questionnaire was pilot-tested on 8 subjects before the main study. Pilot testing stopped once 2 consecutive subjects were able to complete the questionnaire without encountering correctable problems. The final questionnaire contained 2 sections (demographic questions and an assessment of research perceptions) with a total of 20 questions. Items measured a patient’s socioeconomic status, willingness to be tested for approved or experimental therapies, willingness to provide blood and tissue samples for research, preference on being informed of test results, and willingness to permit sample DNA banking for future research. With the exception of demographic information, items were measured on a scale of 1 (strongly disagree) to 5 (strongly agree). The questionnaire, as administered to the subjects, is available on request. Sample Size
This was a descriptive study. The original sample size of 100 was chosen for acceptable precision of the estimates of means (for continuous variables) and proportions (for categorical variables) related to patients’ attitudes. Specifically, for a continuous variable, when the sample size is 100, a 2-sided 95% confidence interval (CI) for a single mean will extend 0.196 from the observed mean under the assumption that the standard deviation is known to be 1 and the CI is based on the large-sample z statistic. For a binary variable, when the sample size is 100, a 2-sided 95% CI for a single proportion using the large-sample normal approximation will extend 0.098 from the observed proportion for an expected proportion of 0.5. Furthermore, a logistic regression of a binary response 244
variable (Y) on a binary independent variable (X) with a sample size of 100 observations will achieve 80% power at a 0.05 significance level to detect an odds ratio (OR) of 3.2 to 4.2, which depends on the baseline probability of Y 5 1 and the distribution of X. Data Collection
After informed consent, each patient completed the questionnaire (see online supporting information). Research staff were available to answer questions and provide clarification on questionnaire items. A standard list of definition of terms was provided to participants. The same list was also used by researchers as a means of standardizing responses to questions about specific items in the questionnaire. Participants’ electronic health records also were reviewed to obtain clinical information relevant to the study. Data collected from the electronic health record included the following: participant age, self-reported race, tumor type and stage, and previous treatment history and participation in the institutional biobanking protocol at MD Anderson. Data Analysis
Our goal in this descriptive study was to preliminarily assess the association of variables with attitudes toward molecular testing. We performed univariate logistic regression analyses, and we believe that findings based on these analyses remain informative toward this goal. Descriptive statistics were reported as proportions, means, and standard deviations. Univariate tests of association were conducted with Fisher’s exact test or the t test according to the variable’s distribution. Univariate logistic regression using the Wald chi-square test was performed to obtain ORs for answering yes on the questionnaire. Yes was defined as answering either “strongly agree” or “agree.” No was defined as answering “neutral,” “disagree,” or “strongly disagree.” A Bonferroni correction for multiple testing was used; statistical significance was defined as P < .01. McNemar’s test was used to compare categorical responses within patients/groups. Because race was found to be a primary variable associated with our outcomes of interest, we also conducted further analyses on this relationship while controlling for relevant covariants. Data analysis was performed with STATA/SE v12 (StataCorp, College Station, TX). RESULTS Patient Characteristics
Demographic, clinical, and pathological characteristics of the study participants are reported in Table 1. Most were Cancer
January 15, 2015
Attitudes Toward Molecular Testing/Yusuf et al
white (71.4%), were married (75%), had at least a college education (55.6%), had children (80%), and were covered by health insurance (98%). Only 37.2% earned less than $50,000 per year (the US median household income in 2011).10 Most had early-stage (stage 0-II) breast cancer (60.6%), had a positive family history of cancer (58.8%), and had some form of therapy (55.0%) before registration at MD Anderson. Patient Attitudes Toward Molecular Testing and Research Biopsies
Patients were generally willing to undergo molecular testing, biopsies, and blood draws to guide therapy and for research (Fig. 1A). Patients were more willing to undergo molecular testing for the selection of approved therapy versus experimental therapy both if the molecular testing was covered by insurance (81% vs 59%, P < .0001) and if the testing cost was not covered (64% vs 37%, P < .0001). Patients were more willing to undergo molecular testing if the cost of the testing was covered by insurance both to guide approved therapy (81% vs 64%, P 5 .0004) and to guide experimental therapy (59% vs 37%, P < .0001). Most participants were willing to undergo blood draws to guide treatment (88%) and for research (64%, P < .0001; Fig. 1B). Many also were willing to undergo biopsies for molecular testing for research (46%) but more so to guide treatment (75%, P < .0001). Interestingly, 84 participants (88% of those responding to the question) stated on the questionnaire that they were willing to have residual tumor or blood samples stored for future research, whereas 91 participants (93%) actually consented to the institutional biospecimen banking protocol under a separate consenting process. Given the debate over the return of incidental research results,11-18 we also asked about patient preferences regarding this issue. Most patients stated that they would like to be informed about their research results if these results had implications for therapy (91%), new cancer risk (90%), and other preventable/treatable diseases (87%). Association of Demographic and Clinicopathological Variables With Patient Attitudes
Next, we assessed whether specific demographic or clinicopathological characteristics were associated with patient attitudes toward molecular testing (Table 2 and Supporting Table 1 [see online supporting information]) and patient willingness to have blood draws and research Cancer
January 15, 2015
TABLE 1. Demographic, Clinical, and Pathological Characteristics of Patientsa Age at survey (y), mean (SD) Age at diagnosis (y), mean (SD) Race, n (%) White, non-Hispanic Hispanic or Latino Black or African American, non-Hispanic Asian Education, n (%) High school or less College Graduate or higher Marital status, n (%) Married/partnered Single Widowed Children, n (%) Yes No Income, n (%) $0-49,999 $50,000-99,999 $100,000 and above Insurance, n (%) Self-pay Medicare Medicaid Other insurance Stage, n (%) 0 I II III IV Duration of illness, n (%) 0-1 y 2-5 y 6-10 y >10 y Therapy to date, n (%) No therapy Chemotherapy only Surgery only Chemotherapy 1 surgery Surgery 1 radiotherapy Chemotherapy 1 radiotherapy 1 surgery Family history of cancer, n (%) Yes No
55.3 (12.9) 52.1 (13.0) 70 16 10 2
(71.4) (16.3) (10.2) (2.0)
21 (21.2) 55 (55.6) 23 (23.2) 75 (75.0) 18 (18.0) 7 (7.0) 80 (80.0) 20 (20.0) 32 (37.2) 26 (30.2) 28 (32.6) 2 24 8 66
(2.0) (24.0) (8.0) (66.0)
7 26 27 15 24
(7.1) (26.3) (27.2) (15.2) (24.2)
71 13 10 6
(71.0) (13.0) (10.0) (6.0)
45 5 16 14 1 19
(45.0) (5.0) (16.0) (14.0) (1.0) (19.0)
57 (58.8) 40 (41.2)
Abbreviation: SD, standard deviation. a Not all patients answered all questions; the analysis was based on questions answered.
biopsies for testing (Supporting Tables 2 and 3 [see online supporting information]). Overall, race was the only consistent, statistically significant determinant (Table 3). Even if molecular testing was covered by insurance, more white respondents than nonwhite respondents were willing to undergo molecular testing for approved therapy (90% vs 54%, P 5 .0004; Table 3) and for experimental therapy (68% vs 35%, P 5 .0072; Tables 2 and 3). A larger majority of white respondents versus nonwhite respondents were willing to have blood drawn for 245
Original Article
P 5 .0116). Because race was found to be a primary variable associated with our outcomes of interest, we conducted further analyses on this relationship and controlled for education, stage, and history of chemotherapy.19 The corresponding results are summarized in Table 3. Race was an independent predictor of preferring not to undergo molecular testing (though covered by insurance) for the selection of approved and experimental therapy as well as tumor biopsies. In addition, patients with a history of chemotherapy were more willing to undergo molecular testing for the selection of experimental medication if it was covered by insurance; 76% of the patients who received chemotherapy were willing to undergo testing, whereas only 48% of the patients who did not receive chemotherapy were (OR 5 3.34, P 5 .0098; Table 2).
Figure 1. Willingness to participate in molecular testing, blood draws, and biopsies. (A) Willingness to participate in molecular testing: overall comparison of willingness to undergo molecular testing for approved therapy and experimental therapy if testing is covered by insurance (P 5.0001), for approved therapy and experimental therapy if testing is not covered by insurance (P
Attitudes Toward Molecular Testing for Personalized Cancer Therapy Rafeek A. Yusuf, MD1; Deevakar Rogith, MD1; Shelly R. A. Hovick, PhD2; Susan K. Peterson, PhD, MPH3; Allison M. Burton-Chase, PhD4; Bryan M. Fellman, MS5; Yisheng Li, PhD5; Carolyn McKinney, BSN6; Elmer V. Bernstam, MD1,7; and Funda Meric-Bernstam, MD6,8
BACKGROUND: This study assessed attitudes of breast cancer patients toward molecular testing for personalized therapy and research. METHODS: A questionnaire was given to female breast cancer patients presenting to a cancer center. Associations between demographic and clinical variables and attitudes toward molecular testing were evaluated. RESULTS: Three hundred eight patients were approached, and 100 completed the questionnaire (a 32% response rate). Most participants were willing to undergo molecular testing to assist in the selection of approved drugs (81%) and experimental therapy (59%) if testing was covered by insurance. Most participants were white (71%). Even if testing was financially covered, nonwhite participants were less willing to undergo molecular testing for the selection of approved drugs (54% of nonwhites vs 90% of whites, odds ratio [OR] 5 0.13, P 5.0004) or experimental drugs (35% vs 68%, OR 5 0.26, P 5.0072). Most participants (75%) were willing to undergo a biopsy to guide therapy, and 46% were willing to undergo research biopsies. Nonwhite participants were less willing to undergo research biopsies (17% vs 55%, OR 5 0.17, P 5.0033). Most participants wanted to be informed when research results had implications for treatment (91%), new cancer risk (90%), and other preventable/treatable diseases (87%). CONCLUSIONS: Most patients were willing to undergo molecular testing and minimally invasive procedures to guide approved or experimental therapy. There were significant differences in attitudes toward molecular testing between racial groups; nonwhites were less willing to undergo testing even if the results would guide their own therapy. Novel approaches are needed to prevent disparities in the delivery of genomically informed care and to increase minority C 2014 American Cancer Society. participation in biomarker-driven trials. Cancer 2015;121:243-50. V KEYWORDS: molecular testing, disparities, personalized cancer therapy, biomarkers, questionnaire, survey.
INTRODUCTION Although the concept of individualizing treatment is more than a century old,1,2 recent advances in molecular diagnostics have enabled personalized cancer therapy based on molecular characteristics of the patient or tumor and novel molecular marker–driven clinical trials. Although we know much about factors that influence general participation in biomedical research, little is known specifically about patients’ willingness to participate in testing for personalized cancer therapy and their perceptions of participation in biomarker-driven research and targeted therapy trials.3,4 Furthermore, molecular testing brings with it unique problems, such as the potential for the discovery of germline variants or mutations with health implications for the patient and the patient’s relatives. Little is known about patient wishes for the return of these incidental research results. In this study, we assess breast cancer patients’ attitudes toward molecular testing for personalized therapy and research as well as the return of incidental research results. MATERIALS AND METHODS Study Participants and Recruitment
This study was conducted at the University of Texas MD Anderson Cancer Center (MD Anderson), an academic cancer center located within the Texas Medical Center in Houston, Texas. The study was approved by the MD Anderson
Corresponding author: Funda Meric-Bernstam, MD, University of Texas MD Anderson Cancer Center, 1400 Holcombe Boulevard, FC8.3044, Houston, TX 77030; Fax: (713) 563-0566; [email protected] 1 School of Biomedical Informatics, University of Texas Health Science Center at Houston, Houston, Texas; 2School of Communication, Ohio State University, Columbus, Ohio; 3Department of Behavioral Sciences, University of Texas MD Anderson Cancer Center, Houston, Texas; 4Department of Basic and Social Sciences, Albany College of Pharmacy and Health Sciences, Albany, New York; 5Department of Biostatistics, University of Texas MD Anderson Cancer Center, Houston, Texas; 6Department of Surgical Oncology, University of Texas MD Anderson Cancer Center, Houston, Texas; 7Division of General Internal Medicine, Department of Internal Medicine, Medical School, University of Texas Health Science Center at Houston, Houston, Texas; 8Department of Investigational Cancer Therapeutics, University of Texas MD Anderson Cancer Center, Houston, Texas.
Additional supporting information may be found in the online version of this article. DOI: 10.1002/cncr.28966, Received: February 19, 2014; Revised: June 13, 2014; Accepted: June 25, 2014, Published online September 10, 2014 in Wiley Online Library (wileyonlinelibrary.com)
Cancer
January 15, 2015
243
Original Article
institutional review board. Patients registering at the breast center between October and December 2012 were invited to participate in a questionnaire study. Eligibility criteria included (1) being 18 years of age or older, (2) being female, (3) being fluent in English, and (4) having been previously diagnosed with breast cancer. Three hundred eight consecutive patients were approached, and 100 completed the questionnaire (a 32% response rate). Study Instrument
Because of the paucity of existing measures from which questions specific to this study could be adapted, new measures were developed that were informed by the literature and previously validated measures4-9 to assess attitudes of breast cancer patients toward molecular testing for personalized therapy and research. The questionnaire was developed and reviewed for content validity by a team of investigators with expertise in genetics, survey research, behavioral science, health communication, and bioinformatics. The questionnaire was pilot-tested on 8 subjects before the main study. Pilot testing stopped once 2 consecutive subjects were able to complete the questionnaire without encountering correctable problems. The final questionnaire contained 2 sections (demographic questions and an assessment of research perceptions) with a total of 20 questions. Items measured a patient’s socioeconomic status, willingness to be tested for approved or experimental therapies, willingness to provide blood and tissue samples for research, preference on being informed of test results, and willingness to permit sample DNA banking for future research. With the exception of demographic information, items were measured on a scale of 1 (strongly disagree) to 5 (strongly agree). The questionnaire, as administered to the subjects, is available on request. Sample Size
This was a descriptive study. The original sample size of 100 was chosen for acceptable precision of the estimates of means (for continuous variables) and proportions (for categorical variables) related to patients’ attitudes. Specifically, for a continuous variable, when the sample size is 100, a 2-sided 95% confidence interval (CI) for a single mean will extend 0.196 from the observed mean under the assumption that the standard deviation is known to be 1 and the CI is based on the large-sample z statistic. For a binary variable, when the sample size is 100, a 2-sided 95% CI for a single proportion using the large-sample normal approximation will extend 0.098 from the observed proportion for an expected proportion of 0.5. Furthermore, a logistic regression of a binary response 244
variable (Y) on a binary independent variable (X) with a sample size of 100 observations will achieve 80% power at a 0.05 significance level to detect an odds ratio (OR) of 3.2 to 4.2, which depends on the baseline probability of Y 5 1 and the distribution of X. Data Collection
After informed consent, each patient completed the questionnaire (see online supporting information). Research staff were available to answer questions and provide clarification on questionnaire items. A standard list of definition of terms was provided to participants. The same list was also used by researchers as a means of standardizing responses to questions about specific items in the questionnaire. Participants’ electronic health records also were reviewed to obtain clinical information relevant to the study. Data collected from the electronic health record included the following: participant age, self-reported race, tumor type and stage, and previous treatment history and participation in the institutional biobanking protocol at MD Anderson. Data Analysis
Our goal in this descriptive study was to preliminarily assess the association of variables with attitudes toward molecular testing. We performed univariate logistic regression analyses, and we believe that findings based on these analyses remain informative toward this goal. Descriptive statistics were reported as proportions, means, and standard deviations. Univariate tests of association were conducted with Fisher’s exact test or the t test according to the variable’s distribution. Univariate logistic regression using the Wald chi-square test was performed to obtain ORs for answering yes on the questionnaire. Yes was defined as answering either “strongly agree” or “agree.” No was defined as answering “neutral,” “disagree,” or “strongly disagree.” A Bonferroni correction for multiple testing was used; statistical significance was defined as P < .01. McNemar’s test was used to compare categorical responses within patients/groups. Because race was found to be a primary variable associated with our outcomes of interest, we also conducted further analyses on this relationship while controlling for relevant covariants. Data analysis was performed with STATA/SE v12 (StataCorp, College Station, TX). RESULTS Patient Characteristics
Demographic, clinical, and pathological characteristics of the study participants are reported in Table 1. Most were Cancer
January 15, 2015
Attitudes Toward Molecular Testing/Yusuf et al
white (71.4%), were married (75%), had at least a college education (55.6%), had children (80%), and were covered by health insurance (98%). Only 37.2% earned less than $50,000 per year (the US median household income in 2011).10 Most had early-stage (stage 0-II) breast cancer (60.6%), had a positive family history of cancer (58.8%), and had some form of therapy (55.0%) before registration at MD Anderson. Patient Attitudes Toward Molecular Testing and Research Biopsies
Patients were generally willing to undergo molecular testing, biopsies, and blood draws to guide therapy and for research (Fig. 1A). Patients were more willing to undergo molecular testing for the selection of approved therapy versus experimental therapy both if the molecular testing was covered by insurance (81% vs 59%, P < .0001) and if the testing cost was not covered (64% vs 37%, P < .0001). Patients were more willing to undergo molecular testing if the cost of the testing was covered by insurance both to guide approved therapy (81% vs 64%, P 5 .0004) and to guide experimental therapy (59% vs 37%, P < .0001). Most participants were willing to undergo blood draws to guide treatment (88%) and for research (64%, P < .0001; Fig. 1B). Many also were willing to undergo biopsies for molecular testing for research (46%) but more so to guide treatment (75%, P < .0001). Interestingly, 84 participants (88% of those responding to the question) stated on the questionnaire that they were willing to have residual tumor or blood samples stored for future research, whereas 91 participants (93%) actually consented to the institutional biospecimen banking protocol under a separate consenting process. Given the debate over the return of incidental research results,11-18 we also asked about patient preferences regarding this issue. Most patients stated that they would like to be informed about their research results if these results had implications for therapy (91%), new cancer risk (90%), and other preventable/treatable diseases (87%). Association of Demographic and Clinicopathological Variables With Patient Attitudes
Next, we assessed whether specific demographic or clinicopathological characteristics were associated with patient attitudes toward molecular testing (Table 2 and Supporting Table 1 [see online supporting information]) and patient willingness to have blood draws and research Cancer
January 15, 2015
TABLE 1. Demographic, Clinical, and Pathological Characteristics of Patientsa Age at survey (y), mean (SD) Age at diagnosis (y), mean (SD) Race, n (%) White, non-Hispanic Hispanic or Latino Black or African American, non-Hispanic Asian Education, n (%) High school or less College Graduate or higher Marital status, n (%) Married/partnered Single Widowed Children, n (%) Yes No Income, n (%) $0-49,999 $50,000-99,999 $100,000 and above Insurance, n (%) Self-pay Medicare Medicaid Other insurance Stage, n (%) 0 I II III IV Duration of illness, n (%) 0-1 y 2-5 y 6-10 y >10 y Therapy to date, n (%) No therapy Chemotherapy only Surgery only Chemotherapy 1 surgery Surgery 1 radiotherapy Chemotherapy 1 radiotherapy 1 surgery Family history of cancer, n (%) Yes No
55.3 (12.9) 52.1 (13.0) 70 16 10 2
(71.4) (16.3) (10.2) (2.0)
21 (21.2) 55 (55.6) 23 (23.2) 75 (75.0) 18 (18.0) 7 (7.0) 80 (80.0) 20 (20.0) 32 (37.2) 26 (30.2) 28 (32.6) 2 24 8 66
(2.0) (24.0) (8.0) (66.0)
7 26 27 15 24
(7.1) (26.3) (27.2) (15.2) (24.2)
71 13 10 6
(71.0) (13.0) (10.0) (6.0)
45 5 16 14 1 19
(45.0) (5.0) (16.0) (14.0) (1.0) (19.0)
57 (58.8) 40 (41.2)
Abbreviation: SD, standard deviation. a Not all patients answered all questions; the analysis was based on questions answered.
biopsies for testing (Supporting Tables 2 and 3 [see online supporting information]). Overall, race was the only consistent, statistically significant determinant (Table 3). Even if molecular testing was covered by insurance, more white respondents than nonwhite respondents were willing to undergo molecular testing for approved therapy (90% vs 54%, P 5 .0004; Table 3) and for experimental therapy (68% vs 35%, P 5 .0072; Tables 2 and 3). A larger majority of white respondents versus nonwhite respondents were willing to have blood drawn for 245
Original Article
P 5 .0116). Because race was found to be a primary variable associated with our outcomes of interest, we conducted further analyses on this relationship and controlled for education, stage, and history of chemotherapy.19 The corresponding results are summarized in Table 3. Race was an independent predictor of preferring not to undergo molecular testing (though covered by insurance) for the selection of approved and experimental therapy as well as tumor biopsies. In addition, patients with a history of chemotherapy were more willing to undergo molecular testing for the selection of experimental medication if it was covered by insurance; 76% of the patients who received chemotherapy were willing to undergo testing, whereas only 48% of the patients who did not receive chemotherapy were (OR 5 3.34, P 5 .0098; Table 2).
Figure 1. Willingness to participate in molecular testing, blood draws, and biopsies. (A) Willingness to participate in molecular testing: overall comparison of willingness to undergo molecular testing for approved therapy and experimental therapy if testing is covered by insurance (P 5.0001), for approved therapy and experimental therapy if testing is not covered by insurance (P
