Regarding research site adoption, we approached 13 San Francisco and Sacramento-area health care organizations. At each research site, we invited a clinical trials manager or physician in charge to coordinate submission of active breast cancer study protocols and institutional review board approval documents to our research team. Of the 13 research sites, 12 completed the requirements for participation, by providing sufficient information for our research team to code their trials using our structured data entry forms. Of the 12 research sites, 2 were academic medical centers, 6 were community hospitals, 1 was a participant in the National Cancer Institute-funded Community Clinical Oncology Program, 1 was a health maintenance organization, and 2 were private oncology practices (). These 12 research sites contributed all of their breast cancer protocols, representing 55 studies during the study period. The BreastCancerTrials.org team coded the trials in the BreastCancerTrials.org database and sent them to the research sites for review and approval. The research sites reported no disagreements with the BreastCancerTrials.org codes, so the team activated the trials on the production website. A total of 11 research sites remained in the study through its completion; 1 withdrew after the first year because the clinical trial manager submitting active study documents felt this required too much time.
Research sites providing trials at the launch of BreastCancerTrials.org.
With regard to patients using the site and matching to clinical trials, the population for this study question consisted of 733 patients registering on BreastCancerTrials.org between June 2005 and September 2006. Registration involved providing a name and email address, at which point patients could navigate the website and consent (or not) to use the matching service. Of 733 registrants, 614 (83.8%) consented to use the matching service, of whom 491 (80.0%) also reported demographics ().
Demographics characteristics of respondents to acceptability and accuracy study questions.
We tracked website usage as an indication of BreastCancerTrials.org’s acceptability to patients. Of the 614 patients who registered and consented, 428 (69.7%) completed the minimum health history elements required to match to a trial. Of these 428, 407 (95.1%) matched to at least one trial. Of the 407 matched visitors, 70 (17%) connected with a research site through BreastCancerTrials.org’s Message Center.
Between July 1, 2005 and June 30, 2006, we sent email surveys to 375 website users who had completed the health history questionnaire during the first 12 months after launch, and invited them to respond anonymously. We found that 75 of 375 (20%) responded, and 23 of the respondents (31% ) reported contacting a research site. Among the 23 who contacted a research site, 12 (52%) reported being told they were eligible for a trial, and 5 of these 12 (42%) reported enrolling in a trial.
We present demographics reported by 75 survey respondents in (middle columns). Respondents rated satisfaction with BreastCancerTrials.org at a mean level of 7 out of a maximum of 10. Respondents rated willingness to recommend BreastCancerTrials.org at a mean level of 7 out of 10. Finally, respondents rated the ease with which they completed the health history questionnaire at a mean level of 8 out of 10.
With regard to study questions about patients’ accuracy, 26 patients consented out of 57 approached (46%), 20 of whom completed the study requirements, for a response rate of 20 out of 57 (35%) among all approached, or a completion rate of 20 out of 26 (77%) among consenting patients. We invited them to fill out the health history questionnaire so we could compare their responses with the information in their medical record. summarizes the demographic profiles of the final sample (last columns). These patients provided a total of 1456 items, matching the chart for 1324. Therefore, the overall accuracy rate was 90.93%. For the standard items in the My Health and the My Cancer sections, the accuracy rate was 469 out of 520 (90.2%). For the variable items in the My Treatment section, the accuracy rate was 855 out of 936 (91.4%).
On an item-by-item level, accuracy ranged from 65% to 100% (see ). On the low end of the range, 13 of the 20 (65%) respondents to the standard items correctly reported their progesterone status. At the high end of the range, respondents were 100% accurate in reporting their status with respect to pregnancy, current well-being, hypertension, cardiac arrhythmia, disease in other sites, local recurrence, type of bisphosphonate therapy taken, and type of biologic therapy taken.
Accuracy of patients’ responses to health history questionnaire items, compared with a study coordinator’s abstraction of the data from the patients’ medical charts.
While patients were accurate overall, we call attention to items for which patients had difficulty matching the chart. A total of 5 patients did not match the chart with respect to the stage of current cancer; 5 patients were not accurate in responding to response to last treatment; and 5 patients were inaccurate in responding to items about disease in lymph nodes, 4 of whom were wrong about axillary and 1 about nonaxillary lymph nodes.