- Email: [email protected]
Impact of a cancer clinical trials web site on discussions about trial participation: a cluster randomized trial
original articles
Annals of Oncology Annals of Oncology 23: 1912–1918, 2012 doi:10.1093/annonc/mdr585 Published online 18 January 2012
Impact of a cancer clinical trials web site on discussions about trial participation: a cluster randomized trial R. F. Dear1,2*, A. L. Barratt1,3, L. M. Askie4, P. N. Butow1,5, K. McGeechan3, S. Crossing6, D. C. Currow7 & M. H. N. Tattersall1,2
Received 16 August 2011; revised 7 November 2011; accepted 28 November 2011
Background: Cancer patients want access to reliable information about currently recruiting clinical trials. Patients and methods: Oncologists and their patients were randomly assigned to access a consumer-friendly cancer clinical trials web site [Australian Cancer Trials (ACT), www.australiancancertrials.gov.au] or to usual care in a cluster randomized controlled trial. The primary outcome, measured from audio recordings of oncologist–patient consultations, was the proportion of patients with whom participation in any clinical trial was discussed. Analysis was by intention-to-treat accounting for clustering and stratification. Results: Thirty medical oncologists and 493 patients were recruited. Overall, 46% of consultations in the intervention group compared with 34% in the control group contained a discussion about clinical trials (P = 0.08). The mean consultation length in both groups was 29 min (P = 0.69). The proportion consenting to a trial was 10% in both groups (P = 0.65). Patients’ knowledge about randomized trials was lower in the intervention than the control group (mean score 3.0 versus 3.3, P = 0.03) but decisional conflict scores were similar (mean score 42 versus 43, P = 0.83). Conclusions: Good communication between patients and physicians is essential. Within this context, a web site such as Australian Cancer Trials may be an important tool to encourage discussion about clinical trial participation. Key words: clinical trials as topic, decision making, health knowledge, internet, neoplasms, patient selection
introduction Cancer patients and consumer advocacy groups want access to reliable information about recruiting clinical trials [1]. Studies consistently show that cancer patients want more information and to be involved in health care decisions [2–4]. Nevertheless, only ∼5% of cancer patients participate in clinical trials [5–8]. There are many barriers to enrollment including patients (and physicians) being unaware of trial opportunities [5]. Increasingly, patients are seeking information online [3, 9, 10], and their health care decisions can be affected by information on the internet [11, 12]. Online clinical trial registries collect information about trials but are not always consumer friendly [13]. In the first web site of its kind, we developed the Australian Cancer Trials (ACT), using data from the Australian New Zealand Clinical Trials Registry (ANZCTR) and ClinicalTrials.gov (CT.gov), to make it easy to
*Correspondence to: Dr R. F. Dear, Department of Medicine, Sydney Medical School, Room 391, Blackburn Building, D06, The University of Sydney, Sydney, NSW 2006, Australia. Tel: +612-9351-6171; Fax: +612-9351-4317; E-mail: [email protected]
search for consumer-friendly information about registered cancer trials [14]. A discussion about internet use may be an opportunity for oncologists to improve patient knowledge of and participation in cancer clinical trials [12,15–17]. Although cancer clinical trials web sites are available, no one has yet evaluated their impact using a rigorous research design. It has been acknowledged that this research is needed [12,15–17]. We report the results of the first cluster randomized controlled trial evaluating the impact of access to a consumerfriendly cancer clinical trials web site on discussions about trial participation, accrual rates, patient knowledge about clinical trials, decision making about joining a trial and consultation length. The cluster randomized controlled trial design was chosen so that we could deliver the intervention to both medical oncologists and their patients within a cluster and to minimize contamination of the control group. Subsequently, in 2011, the National Institute for Health Research developed the UK Clinical Trials Gateway that provides easy to understand information about clinical trials in all health conditions using CT.gov and the International
© The Author 2012. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: [email protected].
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
1 Centre for Medical Psychology and Evidence-based Decision-making, The University of Sydney, Sydney; 2Department of Medicine, Sydney Medical School, The University of Sydney, Sydney; 3Sydney School of Public Health, The University of Sydney, Sydney; 4National Health and Medical Research Council (NHMRC) Clinical Trials Centre, The University of Sydney, Sydney; 5School of Psychology, The University of Sydney, Sydney; 6Cancer Voices New South Wales, Sydney; 7Department of Palliative Care, Flinders University, Adelaide, Australia
original articles
Annals of Oncology
Standard Randomized Controlled Number Register [18]. Their web site has not been evaluated in a randomized trial.
patients and methods
Eligible patients who had cancer were scheduled to have an appointment with their oncologist in the near future and had email and internet access. Participating physicians were asked to recruit 20 patients with whom a discussion about treatment options was anticipated during the consultation. For example:
participants
• •
Approval was granted by the relevant ethics committees. Written informed consent was provided by all participants.
intervention We evaluated the ACT web site, which is a custom designed, consumerfriendly web site that directly interfaces two trial registries, ANZCTR and CT.gov [14]. Content is extracted directly from these registries via their data exporting services and downloaded to the ACT web site every
Figure 1. CONSORT flow diagram.
Volume 23 | No. 7 | July 2012
doi:10.1093/annonc/mdr585 |
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Eligible clinics consisted of at least one medical oncologist and a facility to administer chemotherapy. Eighty-four eligible clinics in New South Wales (NSW) and Victoria, Australia, were identified from cancer services directories. Clinics were randomly selected with probability proportionate to the number of oncologists and using a stratification procedure by state (NSW : Victoria, 1 : 1) and location (rural : urban, 1 : 2). One medical oncologist from each clinic was randomly selected and invited to participate. If this medical oncologist declined, a second medical oncologist from that clinic was invited. Once 30 medical oncologists at 30 clinics had agreed to participate they were randomized. Sixty-one medical oncologists at 33 clinics declined to participate (Figure 1).
Patients considering systemic treatment for the first time, such as new patients and Patients in whom a treatment change was expected.
original articles
procedures After randomization, all physicians completed a brief baseline demographic questionnaire. Medical oncologists in the intervention group were emailed a web link to access the ACT web site and encouraged to look at it. All physicians were asked to audio record the next scheduled consultation with each consenting patient. If the consultation was not recorded, the next scheduled appointment was recorded. Patients were invited to participate by a letter from their medical oncologist in a study called ‘Treatment options for cancer patients’. Some also received a phone call from their medical oncologist, oncologist’s secretary, clinical trials nurse or research staff (if requested) to check their eligibility. They were informed that the study’s aim was to characterize the manner in which medical oncologists discuss treatment options, including clinical trials, with their patients. All consenting patients initially received a ‘welcome’ email confirming their involvement in the trial. The welcome email to the intervention group also informed patients that ‘the real aim of this study is to test a new web site called ACT’. These patients were encouraged to look at the web site before their consultation. After receiving the welcome email, all patients received a link to a brief online questionnaire to be completed before their scheduled consultation. On completion of the questionnaire the intervention group were automatically taken to the ACT web site’s home page. Two weeks after their audiorecorded consultation, patients were asked to complete an online follow-up questionnaire.
objectives The objectives of the trial were to evaluate if access to the ACT web site increased the proportion of patients who discussed the possibility of joining a clinical trial with their oncologist and if access to the web site had an impact on the number and complexity of trial issues discussed, consultation length, accrual rates to recruiting trials, patients’ knowledge about randomized clinical trials and patients’ decisional conflict whether to join a trial. We hypothesized that oncologists, by searching the ACT web site, would know about all relevant recruiting trials and therefore discuss these with their patients. We hypothesized that patients, prompted by the information displayed on the ACT web site, would be aware that one of their treatment options may be a clinical trial and might discuss this with their oncologist.
outcomes The primary outcome was the proportion of patients with whom the possibility of participation in any clinical trial was discussed. This was measured by coding transcripts of consultation audio recordings. All the transcripts were reviewed independently by two people (the primary researcher, RFD, and the transcriptionist, who was trained in this task). The agreement between RFD and the transcriptionist was very good for the primary outcome (kappa = 1). A random 10% of consultations in the control group were coded by a third person. Agreement for the primary outcome and secondary outcome (the proportion invited to participate in a trial) was also very good (kappa = 0.98 and 1, respectively). The secondary outcomes measured from the audio recordings were the proportion of patients invited to join a trial, the number of issues about trials discussed (18 in total were coded for), the overall complexity of the trial discussion and consultation length. The overall complexity of the trial discussion was coded as extended when the trial purpose and rationale were explained or as basic when only minimal information was given. The secondary outcomes obtained from patient questionnaires were the proportion of patients who consented to a trial, knowledge about randomized trials and decisional conflict about the possibility of joining a clinical trial. The proportion of patients who consented to a trial was verified from records provided by each medical oncologist. Knowledge about randomized trials was assessed using a validated seven-item knowledge scale [20]. Scores ranged from 0 ( poor knowledge) to 7 (very high knowledge). Among patients who discussed the possibility of trial participation, the quality of their decision making to join a trial was assessed by the 16-item Decisional Conflict Scale, which has demonstrated validity and reliability [21]. Scores ranged from 0 (no decisional conflict) to 100 (extremely high decisional conflict).
statistical analysis randomization and blinding Randomization was done by a central computer via a random allocation sequence method to conceal allocation. Block randomization, stratified by state (NSW or Victoria) and location (rural or urban), was used to ensure balanced numbers of clusters. Medical oncologists were recruited by either of two study investigators (RFD or MHNT) who were blind to their intervention group. As outlined above, all patients were initially blinded to the intervention and to allocation status but after consenting to the study, the intervention was revealed to patients in the intervention group only. All oncologists were aware of the trial intervention because they had to consent to participate before randomization but, after randomization, only oncologists in the intervention group received a web link to access the ACT web site. All participants were blinded to the primary outcome. Outcome assessors were blinded to the intervention status of the study transcripts.
| Dear et al.
The primary outcome sample size estimate assumed an increase from 10% to 22% in the proportion of consultations in which the possibility of participation in any trial was discussed. Assuming an intraclass coefficient of 0.05, the required sample size was 30 clusters and 20 patients at each site, with 80% power at a 5% level of significance. Participant characteristics were summarized by treatment group using, where appropriate, means, medians, standard deviations (SDs), minimums and maximums for continuously distributed variables and frequency counts and percents for categorical variables. The primary and secondary outcomes were analyzed on an intention-totreat basis. The primary outcome was binary and was compared with binomial regression and adjustment for clustering using generalized estimating equations and stratification by state and rural/urban location. Secondary outcomes included binary and continuous variables and were analyzed by comparison of proportions and comparison of means respectively, with adjustment for clustering and stratification by state and
Volume 23 | No. 7 | July 2012
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
5 minutes. ACT is hosted by Cancer Australia, an Australian federal government agency. The web site enables people to search for cancer trials by cancer type, stage, treatment, trial phase, recruitment status and age group. The information displayed about each trial includes routine data collected by trial registries as well as additional items requested by consumers during the web site’s design [19]. It also provides general information about clinical trials and two question prompt lists that suggest questions patients can ask their physician about clinical trial participation. The web site was publicly launched after recruitment and follow-up were completed and was thus only available to those participating in the trial. Medical oncologists in the intervention group received access to the ACT web site before recruiting patients. Consenting patients in the intervention group were given access to the web site before their scheduled appointment with their oncologist. Neither oncologists nor patients in the control group had access to the web site. The control group patients received the standard information provided by each medical oncologist before their scheduled appointment. See ‘Procedures’ section.
Annals of Oncology
original articles
Annals of Oncology location. Statistical analyses were carried out using SAS (version 9, SAS Institute, Cary, NC). This trial is registered with the ANZCTR, number ACTRN12608000273381 [22].
results
primary outcome The proportion of patients with whom the possibility of participation in a trial was discussed was 46% (84/183) in the intervention group and 34% (72/215) in the control group [adjusted difference 10%, 95% confidence interval (CI) 1% to 22%, P = 0.08] (Table 3).
discussion
Two percent (3/183) of the intervention group was invited to join a clinical trial compared with 5% (10/215) in the control group (adjusted odds ratio 2.86, 95% CI 0.58–13.99, P = 0.20).
This is the first evidence, albeit weak evidence, that a consumer-friendly cancer clinical trials web site can increase the proportion of patients with whom the possibility of participation in a clinical trial is discussed. We estimated the proportion of consultations in which the possibility of
Table 1. Physician characteristics at baseline for those in the main outcome analysis
Table 2. Patient characteristics at baseline for those in the main outcome analysis
secondary outcomes
Total number of physicians Mean age (years) Men Country of birth Australia Other Cancers treated Breast Lung Genitourinary State NSW Victoria Practice type Mostly salaried Mostly fee for service About equal Urban/rural Urban Rural
Intervention group
Control group
14 48 (7.3) 10 (71%)
14 47 (8.4) 10 (71%)
8 (57%) 6 (43%)
12 (86%) 2 (14%)
10 (71%) 7 (50%) 6 (43%)
11 (79%) 10 (71%) 9 (64%)
9 (64%) 5 (36%)
8 (57%) 6 (43%)
6 (42%) 4 (29%) 4 (29%)
4 (28%) 5 (36%) 5 (36%)
10 (71%) 4 (29%)
9 (64%) 5 (36%)
Data are N (N%) or mean (SD), unless otherwise stated. NSW, New South Wales.
Volume 23 | No. 7 | July 2012
Total number of patients Mean number of patients per cluster (range) Mean age (years) Men New patients Cancer site Breast Bowel Lung Other Cancer stage Early Advanced/recurrent Unknown Ability to search interneta Rarely or sometimes find Often or always find Do not search
Intervention group
Control group
183 13 (2–20)
215 15 (5–20)
60 (11.6) 84 (46%) 95 (52%)
59 (11.3) 91 (42%) 113 (53%)
55 (30%) 42 (23%) 24 (13%) 62 (34%)
77 (36%) 39 (18%) 21 (10%) 78 (36%)
89 (49%) 65 (36%) 29 (19%)
123 (57%) 67 (32%) 25 (12%)
38 (22%) 94 (53%) 44 (25%)
26 (13%) 135 (65%) 47 (23%)
Data are N (N%) or mean (SD), unless otherwise stated. Seven patients from the intervention group and seven patients from the control group did not answer this question. a
doi:10.1093/annonc/mdr585 |
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Thirty eligible medical oncologists working at 30 different medical oncology clinics were randomized. After randomization, two medical oncologists in each group withdrew and one replacement medical oncologist working at a different clinic was recruited to each group. As a result, there were 28 study clusters eligible for inclusion in the study with a total of 494 patients (Figure 1). Participant accrual commenced in June 2008 and was completed in July 2010. Follow-up was completed in October 2010. Physician characteristics assessed at baseline were similar between the intervention and control groups (Table 1), as were patient characteristics assessed at baseline (Table 2).
There was no significant difference in mean consultation length (P = 0.69), the mean number of issues about clinical trials discussed (P = 0.96), whether the patient raised the possibility of participation in a clinical trial (P = 0.81), and the percentage of consultations that included an extended clinical trial discussion (P = 0.23) (Table 4). The proportion self-reporting to have consented to a clinical trial was 10% (14/146) in the intervention group and 10% (20/ 194) in the control group (adjusted odds ratio 1.20, 95% CI 0.54–2.69, P = 0.65) (Table 4). When verified by each participating medical oncologist, similar results were obtained: 10% (18/180) in the intervention group and 13% (28/215) in the control group (adjusted odds ratio 0.64, 95% CI 0.24–1.70, P = 0.37). The mean seven-item knowledge score was 3.0 (SD 1.8) in the intervention group compared with 3.3 (SD 1.9) in the control group (P = 0.03). Among those that discussed the possibility of participating in a clinical trial, the mean decisional conflict score was 42 in the intervention group and 43 in the control group (P = 0.83). In the intervention group, 146/183 (80%) patients who had an audio recording completed the follow-up questionnaire. In total, 59% (86/146) stated that they used the web site. The reactions of these patients to the web site are shown in Table 5.
original articles
Annals of Oncology
Table 3. Primary outcome from audio recordings: proportion of patients with whom the possibility of participation in a clinical trial was discussed
Number of patients analyzed Number of patients with whom the possibility of participation in a clinical trial was discussed
Intervention group
Control group
183/251 (73%) 84/183 (46%)
215/242 (89%) 72/215 (34%)
Adjustedb Risk difference (95% CI)
P value
10% (−1% to 22%)
0.08
Data are N (N%) or mean (SD), unless otherwise stated. Adjusted for state and urban/rural location. CI, confidence interval. a
Table 4. Secondary outcomes from audio recordings and patient follow-up questionnaires
Secondary outcomes from audio recordings Patients analyzed Patients invited to join a clinical trial Length of consultation, minutes (mean) Consultations where a clinical trial was discussed Number of issues about clinical trials discussed (mean score out of 18) Patient (rather than physician) raised the issue of a clinical trial Consultations that included extended (more complex) discussions Secondary outcomes from patient follow-up questionnaires Patients with follow-up questionnaires and audio recordings Patients who consented to a clinical trial Number of patients with valid seven-item knowledge scale Seven-item knowledge scale (mean score out of 7) Number of patients with a valid Decisional Conflict Scalec Decisional conflict score (mean score out of 100)
Control group
183/251 (73%) 3/183 (2%) 29 84/183 (46%) 5.3 (4.1) 11/84 (13%) 13/84 (16%)
215/242 (89%) 10/215 (5%) 29 72/215 (34%) 4.9 (3.8) 10/72 (14%) 9/72 (13%)
146/183 (80%) 14/146 (10%) 146/146 (100%) 3.0 (1.8) 69/84 (82%) 42 (12.4)
194/215 (90%) 20/194 (10%) 194/194 (100%) 3.3 (1.9) 71/72 (99%) 43 (13.0)
Adjusteda Difference (95% CI)
P value
2.86 (0.58 to 13.99)b −1.8 (−11 to 7)
0.20b 0.69
0.05 (−1.7 to 1.8) 1% (−5% to 7%) 5% (−3% to 13%)
0.96 0.81 0.23
1.20 (0.54 to 2.69)b
0.65
−0.4 (−0.03 to −0.75)
0.03
−0.6 (−4.7 to 5.9)
0.83
Data are N (N%) or mean (SD), unless otherwise stated. Adjusted for state and urban/rural location. b Odds ratio reported as logistic model fitted. Binomial regression model failed to converge. c Only patients who reported discussing a trial with their medical oncologist were asked to complete the Decisional Conflict Scale which began with the statement: “Thinking about the possibility of joining a clinical trial, please indicate your agreement or disagreement with the statements below.” CI, confidence interval. a
participation in any trial was discussed might be 10% in the control group. But it was 34%, possibly because physicians who agreed to participate in the study were more interested in clinical trials than those in everyday oncology practice. Furthermore, physicians in both arms of the study were unavoidably aware of the general study purpose so may have been primed to discuss clinical trials more than usual. The high rate of discussion about trial participation in the control group reduced our capacity to detect a statistically significant effect. The effect of the web site may also have been diluted because the web site was not actively endorsed by physicians (we did not ask them to do this). We chose discussion about the possibility of participation in a clinical trial as our primary end point rather than actual recruitment to a trial for several reasons. Firstly, it was directly detectable from our audio recordings whereas a decision to join a trial may be taken later and may depend on many factors (discussed below). In addition, it is important that patients and physicians discuss all treatment options because | Dear et al.
this respects their autonomy and assists them to make an informed decision. Barriers to patient participation in cancer clinical trials include lack of knowledge about trial opportunities [5]. Furthermore, to find suitable clinical trials, patients need to understand medical terminology related to their cancer diagnosis and treatment and must understand the clinical trial research process in general [5]. Physicians’ lack of awareness of recruiting clinical trials and their difficulty in adequately informing patients about trials are also barriers [23, 24]. Cancer clinical trials web sites may overcome these barriers [12, 15, 16]. In our study, there was no significant difference in the proportions invited to join a trial and the proportions that consented to a trial, reflecting that there are many factors that influence a physician’s decision to invite a patient to join a trial and the patient’ decision to consent. These factors may include other barriers to trial participation, e.g. patients’ dislike of randomization, discomfort with the research process, complexity of the trial protocol, potential side-effects and the Volume 23 | No. 7 | July 2012
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Intervention group
original articles
Annals of Oncology Table 5. Intervention group patient reactions to the Australian Cancer Trials (ACT) web site Patients N (N%)
physician’s attitudes toward trials [5]. This is consistent with other research showing that it is difficult to identify interventions that increase participation in randomized trials [25, 26]. Oncologists have expressed concern that patients’ internet investigation increases consultation times but this is not supported by our results [23]. The control group scored higher on the randomized clinical trial knowledge scale, but the absolute difference was small (0.3). The scores are comparable with other published studies and are consistent with poor knowledge about randomized clinical trials [20, 24]. In 1993, Baum called for a way to better inform people about randomized trials and it is clear that > 15 years later public and patient education is urgently required [27]. The mean decision conflict score was similar in both groups. Notably, scores over 37.5 are associated with decision delay or feeling unsure about implementing a decision. So, regardless of access to the ACT web site, patients considering clinical trial participation faced uncertainty about their decision. Effective interventions to support decision making about clinical trial participation are required. The main limitation to our study was the higher dropout rate in the intervention group compared with the control group. 27% (68/251) of consultations were not recorded in the intervention group compared with 11% (27/243) in the control group. Of the audio recordings missed in the intervention group, 41% (28/68) came from two medical oncologists who, despite reminders, repeatedly forgot to audio record consenting patients’ consultations. This may reflect the previously reported reluctance of medical oncologists to audiorecord consultations and also be the reason that oncologists declined to participate in our trial [28]. Only 59% of intervention group patients (86/146) stated that they used the ACT web site despite all being taken to the ACT web site after completing the baseline questionnaire. It has been reported patients do not recall looking at information despite evidence they have seen it [29]. Also, patients may be more likely to use a web site endorsed by their physician [4, 12, 15, 16, 30]. To check whether the patients in our trial Volume 23 | No. 7 | July 2012
acknowledgements We would like to acknowledge the medical oncologists who participated in the trial: Ehtesham Abdi, Stephen Begbie, Karen Briscoe, David Bell, Robert Blum, Frances Boyle, Walter Cosolo, Stephen Della-Fiorentina, Paul De Souza, Anthony Dowling, Vinod Ganju, Geraldine Goss, John Grygiel, Anne Hamilton, Andrew Haydon, Rina Hui, Michael Jefford, Richard Kefford, Georgina Long, Gavin Marx, Sue-Anne McLachlan, Roger McLennan, Paul Mitchell, Philip Parente, Nicholas Pavlakis, John Scarlett, Eva Segelov, Anne Sullivan, Janette Vardy and Desmond Yip. We would also like to acknowledge their patients who kindly participated.
funding National Health and Medical Research Council (512380).
disclosure LA is the director of the Australian New Zealand Clinical Trials Registry and this nonfinancial interest may be relevant to the submitted work. All remaining authors declare no conflicts of interest.
references 1. Senate Community Affairs Committee. The Cancer Journey: Informing Choice, Report of the Inquiry into Services and Treatment Options for Persons with Cancer. Canberra, Australia: Australian Government 2005; 1–169. 2. Degner LF, Sloan JA. Decision making during serious illness: what role do patients really want to play? J Clin Epidemiol 1992; 45(9): 941–950.
doi:10.1093/annonc/mdr585 |
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Intervention group patients with a follow-up 146/183 (80%) questionnaire Patients who reported they used the ACT web site 86/146 (59%) The ACT web site was useful 76/86 (88%) The ACT web site was helpful for learning more about 73/86 (85%) cancer clinical trials There was enough information on the ACT web site 77/86 (90%) Would use the ACT web site in the future 57/86 (66%) Would recommend the ACT web site to other patients 68/86 (84%) Felt it was important for patients with cancer to have access 78/86 (91%) to the ACT web site
may have been different from other oncology patients in some way related to their likelihood of using the web site, we conducted a substudy of 100 consecutive unselected patients (including both new and follow-up patients) seeing a medical oncologist who had not participated in our trial. The same proportion (59%, 47/80) of patients recalled looking at the web site as in our main trial suggesting the patients in our trial were broadly representative of those in everyday oncology practice. Among those who used the web site, reactions to the web site were positive (Table 5). The cancer journey is often long and considering clinical trial participation is not a one-off event. Access to a web site like ACT, even briefly before one consultation, may positively inform patients for future clinical trial discussions and decision making. This is the first reported randomized controlled trial assessing the effect of a consumer-friendly clinical trials web site on physician–patient discussions about participation in clinical trials. The findings of our study are the first, albeit weak evidence, that the ACT web site increased discussion rates about participation in clinical trials. However, it did not increase accrual rates. Further research is required to assess the impact of such a web site when it is endorsed by physicians and following wider dissemination by cancer consumer organizations to people affected by cancer.
original articles
| Dear et al.
17. Simon C, Schramm S, Hillis S. Patient internet use surrounding cancer clinical trials: clinician perceptions and responses. Contemp Clin Trials 2010; 31(3): 229–234. 18. UK Clinical Trials Gateway. http://www.ukctg.nihr.ac.uk/default.aspx. (10 December 2011, date last accessed). 19. Dear RF, Barratt AL, Crossing S et al. Consumer input into research: the Australian Cancer Trials website. Health Res Policy Syst 2011; 9(1): 30. 20. Ellis PM, Butow PN, Tattersall MHN et al. Randomized clinical trials in oncology: understanding and attitudes predict willingness to participate. J Clin Oncol 2001; 19(15): 3554–3561. 21. O’Connor AM. User Manual–Decisional Conflict Scale. http://decisionaid.ohri.ca/ docs/develop/User_Manuals/UM_Decisional_Conflict.pdf (10 December 2011, date last accessed). 22. Australian New Zealand Clinical Trials Registry, ACTRN1260800027338. http ://www.anzctr.org.au/trial_view.aspx?id=82593 (10 December 2011, date last accessed). 23. Helft PR, Hlubocky F, Daugherty CK. American oncologists’ views of internet use by cancer patients: a mail survey of American Society of Clinical Oncology members. J Clin Oncol 2003; 21(5): 942–947. 24. Sabesan S, Burgher B, Buettner P et al. Attitudes, knowledge and barriers to participation in cancer clinical trials among rural and remote patients. Asia Pac J Clin Oncol 2011; 7(1): 27–33. 25. Mc Daid C, Hodges Z, Fayter D et al. Increasing participation of cancer patients in randomised controlled trials: a systematic review. Trials 2006; 7: 16. 26. Treweek S, Mitchell E, Pitkethly M et al. Strategies to improve recruitment to randomised controlled trials. Cochrane Database Syst Rev 2010; (4): MR000013. . 27. Baum M. New approach for recruitment into randomised controlled trials. Lancet 1993; 341(8848): 812–813. 28. Stockler M, Butow PN, Tattersall MH. The take-home message: doctors’ views on letters and tapes after a cancer consultation. Ann Oncol 1993; 4(7): 549–552. 29. Nisbett RE, Wilson TD. Telling more than we can know: verbal reports on mental processes. Psychol Rev 1977; 84(3): 231–259. 30. Atkinson NL, Massett HA, Mylks C et al. User-centered research on breast cancer patient needs and preferences of an Internet-based clinical trial matching system. J Med Internet Res 2007; 9(2): e13.
Volume 23 | No. 7 | July 2012
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
3. Chen X, Siu LL. Impact of the media and the internet on oncology: survey of cancer patients and oncologists in Canada. J Clin Oncol 2001; 19(23): 4291–4297. 4. Butow P, Devine R, Boyer M et al. Cancer consultation preparation package: changing patients but not physicians is not enough. J Clin Oncol 2004; 22(21): 4401–4409. 5. Mills EJ, Seely D, Rachlis B et al. Barriers to participation in clinical trials of cancer: a meta-analysis and systematic review of patient-reported factors. Lancet Oncol 2006; 7(2): 141–148. 6. Demmy TL, Yasko JM, Collyar DE et al. Managing accrual in cooperative group clinical trials. J Clin Oncol 2004; 22(15): 2997–3002. 7. Linda SE, Catherine C, Bekele BN et al. Generalizability of cancer clinical trial results. Cancer 2006; 106(11): 2452–2458. 8. Allison M. Can web 2.0 reboot clinical trials? Nat Biotechnol 2009; 27(10): 895–902. 9. Lopez-Gomez M, Feliu J, Sereno M et al. Internet use for medical research among cancer patients and their relatives in Spain. Ann Oncol 2008; 19(11): 1976–1977. 10. Helft PR. Breast cancer in the information age: a review of recent developments. Breast Dis 2004; 21: 41–46. 11. Dickerson SS, Boehmke M, Ogle C, Brown JK. Out of necessity: oncology nurses’ experiences integrating the internet into practice. Oncol Nurs Forum 2005; 32(2): 355–362. 12. Castleton K, Fong T, Wang-Gillam A et al. A survey of Internet utilization among patients with cancer. Support Care Cancer 2011; 19(8): 1183–1190. 13. Dear R, Barratt A, Askie L et al. Adding value to clinical trial registries: insights from Australian Cancer Trials Online, a website for consumers. Clin Trials 2011; 8(1): 70–76. 14. Australian Cancer Trials. http://www.australiancancertrials.gov.au. (10 December 2011, date last accessed). 15. Wallwiener M, Wallwiener CW, Brucker SY et al. The Brustkrebs-Studien.de website for breast cancer patients: user acceptance of a German internet portal offering information on the disease and treatment options, and a clinical trials matching service. BMC Cancer 2010; 10: 663. 16. Rodrigues MJ, Haberer S, Dionysopoulos D et al. Internet to boost patient accrual in oncology trials? A multiinstitutional AERIO study. Ann Oncol 2011; 22(2): 490–491.
Annals of Oncology
Annals of Oncology Annals of Oncology 23: 1912–1918, 2012 doi:10.1093/annonc/mdr585 Published online 18 January 2012
Impact of a cancer clinical trials web site on discussions about trial participation: a cluster randomized trial R. F. Dear1,2*, A. L. Barratt1,3, L. M. Askie4, P. N. Butow1,5, K. McGeechan3, S. Crossing6, D. C. Currow7 & M. H. N. Tattersall1,2
Received 16 August 2011; revised 7 November 2011; accepted 28 November 2011
Background: Cancer patients want access to reliable information about currently recruiting clinical trials. Patients and methods: Oncologists and their patients were randomly assigned to access a consumer-friendly cancer clinical trials web site [Australian Cancer Trials (ACT), www.australiancancertrials.gov.au] or to usual care in a cluster randomized controlled trial. The primary outcome, measured from audio recordings of oncologist–patient consultations, was the proportion of patients with whom participation in any clinical trial was discussed. Analysis was by intention-to-treat accounting for clustering and stratification. Results: Thirty medical oncologists and 493 patients were recruited. Overall, 46% of consultations in the intervention group compared with 34% in the control group contained a discussion about clinical trials (P = 0.08). The mean consultation length in both groups was 29 min (P = 0.69). The proportion consenting to a trial was 10% in both groups (P = 0.65). Patients’ knowledge about randomized trials was lower in the intervention than the control group (mean score 3.0 versus 3.3, P = 0.03) but decisional conflict scores were similar (mean score 42 versus 43, P = 0.83). Conclusions: Good communication between patients and physicians is essential. Within this context, a web site such as Australian Cancer Trials may be an important tool to encourage discussion about clinical trial participation. Key words: clinical trials as topic, decision making, health knowledge, internet, neoplasms, patient selection
introduction Cancer patients and consumer advocacy groups want access to reliable information about recruiting clinical trials [1]. Studies consistently show that cancer patients want more information and to be involved in health care decisions [2–4]. Nevertheless, only ∼5% of cancer patients participate in clinical trials [5–8]. There are many barriers to enrollment including patients (and physicians) being unaware of trial opportunities [5]. Increasingly, patients are seeking information online [3, 9, 10], and their health care decisions can be affected by information on the internet [11, 12]. Online clinical trial registries collect information about trials but are not always consumer friendly [13]. In the first web site of its kind, we developed the Australian Cancer Trials (ACT), using data from the Australian New Zealand Clinical Trials Registry (ANZCTR) and ClinicalTrials.gov (CT.gov), to make it easy to
*Correspondence to: Dr R. F. Dear, Department of Medicine, Sydney Medical School, Room 391, Blackburn Building, D06, The University of Sydney, Sydney, NSW 2006, Australia. Tel: +612-9351-6171; Fax: +612-9351-4317; E-mail: [email protected]
search for consumer-friendly information about registered cancer trials [14]. A discussion about internet use may be an opportunity for oncologists to improve patient knowledge of and participation in cancer clinical trials [12,15–17]. Although cancer clinical trials web sites are available, no one has yet evaluated their impact using a rigorous research design. It has been acknowledged that this research is needed [12,15–17]. We report the results of the first cluster randomized controlled trial evaluating the impact of access to a consumerfriendly cancer clinical trials web site on discussions about trial participation, accrual rates, patient knowledge about clinical trials, decision making about joining a trial and consultation length. The cluster randomized controlled trial design was chosen so that we could deliver the intervention to both medical oncologists and their patients within a cluster and to minimize contamination of the control group. Subsequently, in 2011, the National Institute for Health Research developed the UK Clinical Trials Gateway that provides easy to understand information about clinical trials in all health conditions using CT.gov and the International
© The Author 2012. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: [email protected].
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
1 Centre for Medical Psychology and Evidence-based Decision-making, The University of Sydney, Sydney; 2Department of Medicine, Sydney Medical School, The University of Sydney, Sydney; 3Sydney School of Public Health, The University of Sydney, Sydney; 4National Health and Medical Research Council (NHMRC) Clinical Trials Centre, The University of Sydney, Sydney; 5School of Psychology, The University of Sydney, Sydney; 6Cancer Voices New South Wales, Sydney; 7Department of Palliative Care, Flinders University, Adelaide, Australia
original articles
Annals of Oncology
Standard Randomized Controlled Number Register [18]. Their web site has not been evaluated in a randomized trial.
patients and methods
Eligible patients who had cancer were scheduled to have an appointment with their oncologist in the near future and had email and internet access. Participating physicians were asked to recruit 20 patients with whom a discussion about treatment options was anticipated during the consultation. For example:
participants
• •
Approval was granted by the relevant ethics committees. Written informed consent was provided by all participants.
intervention We evaluated the ACT web site, which is a custom designed, consumerfriendly web site that directly interfaces two trial registries, ANZCTR and CT.gov [14]. Content is extracted directly from these registries via their data exporting services and downloaded to the ACT web site every
Figure 1. CONSORT flow diagram.
Volume 23 | No. 7 | July 2012
doi:10.1093/annonc/mdr585 |
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Eligible clinics consisted of at least one medical oncologist and a facility to administer chemotherapy. Eighty-four eligible clinics in New South Wales (NSW) and Victoria, Australia, were identified from cancer services directories. Clinics were randomly selected with probability proportionate to the number of oncologists and using a stratification procedure by state (NSW : Victoria, 1 : 1) and location (rural : urban, 1 : 2). One medical oncologist from each clinic was randomly selected and invited to participate. If this medical oncologist declined, a second medical oncologist from that clinic was invited. Once 30 medical oncologists at 30 clinics had agreed to participate they were randomized. Sixty-one medical oncologists at 33 clinics declined to participate (Figure 1).
Patients considering systemic treatment for the first time, such as new patients and Patients in whom a treatment change was expected.
original articles
procedures After randomization, all physicians completed a brief baseline demographic questionnaire. Medical oncologists in the intervention group were emailed a web link to access the ACT web site and encouraged to look at it. All physicians were asked to audio record the next scheduled consultation with each consenting patient. If the consultation was not recorded, the next scheduled appointment was recorded. Patients were invited to participate by a letter from their medical oncologist in a study called ‘Treatment options for cancer patients’. Some also received a phone call from their medical oncologist, oncologist’s secretary, clinical trials nurse or research staff (if requested) to check their eligibility. They were informed that the study’s aim was to characterize the manner in which medical oncologists discuss treatment options, including clinical trials, with their patients. All consenting patients initially received a ‘welcome’ email confirming their involvement in the trial. The welcome email to the intervention group also informed patients that ‘the real aim of this study is to test a new web site called ACT’. These patients were encouraged to look at the web site before their consultation. After receiving the welcome email, all patients received a link to a brief online questionnaire to be completed before their scheduled consultation. On completion of the questionnaire the intervention group were automatically taken to the ACT web site’s home page. Two weeks after their audiorecorded consultation, patients were asked to complete an online follow-up questionnaire.
objectives The objectives of the trial were to evaluate if access to the ACT web site increased the proportion of patients who discussed the possibility of joining a clinical trial with their oncologist and if access to the web site had an impact on the number and complexity of trial issues discussed, consultation length, accrual rates to recruiting trials, patients’ knowledge about randomized clinical trials and patients’ decisional conflict whether to join a trial. We hypothesized that oncologists, by searching the ACT web site, would know about all relevant recruiting trials and therefore discuss these with their patients. We hypothesized that patients, prompted by the information displayed on the ACT web site, would be aware that one of their treatment options may be a clinical trial and might discuss this with their oncologist.
outcomes The primary outcome was the proportion of patients with whom the possibility of participation in any clinical trial was discussed. This was measured by coding transcripts of consultation audio recordings. All the transcripts were reviewed independently by two people (the primary researcher, RFD, and the transcriptionist, who was trained in this task). The agreement between RFD and the transcriptionist was very good for the primary outcome (kappa = 1). A random 10% of consultations in the control group were coded by a third person. Agreement for the primary outcome and secondary outcome (the proportion invited to participate in a trial) was also very good (kappa = 0.98 and 1, respectively). The secondary outcomes measured from the audio recordings were the proportion of patients invited to join a trial, the number of issues about trials discussed (18 in total were coded for), the overall complexity of the trial discussion and consultation length. The overall complexity of the trial discussion was coded as extended when the trial purpose and rationale were explained or as basic when only minimal information was given. The secondary outcomes obtained from patient questionnaires were the proportion of patients who consented to a trial, knowledge about randomized trials and decisional conflict about the possibility of joining a clinical trial. The proportion of patients who consented to a trial was verified from records provided by each medical oncologist. Knowledge about randomized trials was assessed using a validated seven-item knowledge scale [20]. Scores ranged from 0 ( poor knowledge) to 7 (very high knowledge). Among patients who discussed the possibility of trial participation, the quality of their decision making to join a trial was assessed by the 16-item Decisional Conflict Scale, which has demonstrated validity and reliability [21]. Scores ranged from 0 (no decisional conflict) to 100 (extremely high decisional conflict).
statistical analysis randomization and blinding Randomization was done by a central computer via a random allocation sequence method to conceal allocation. Block randomization, stratified by state (NSW or Victoria) and location (rural or urban), was used to ensure balanced numbers of clusters. Medical oncologists were recruited by either of two study investigators (RFD or MHNT) who were blind to their intervention group. As outlined above, all patients were initially blinded to the intervention and to allocation status but after consenting to the study, the intervention was revealed to patients in the intervention group only. All oncologists were aware of the trial intervention because they had to consent to participate before randomization but, after randomization, only oncologists in the intervention group received a web link to access the ACT web site. All participants were blinded to the primary outcome. Outcome assessors were blinded to the intervention status of the study transcripts.
| Dear et al.
The primary outcome sample size estimate assumed an increase from 10% to 22% in the proportion of consultations in which the possibility of participation in any trial was discussed. Assuming an intraclass coefficient of 0.05, the required sample size was 30 clusters and 20 patients at each site, with 80% power at a 5% level of significance. Participant characteristics were summarized by treatment group using, where appropriate, means, medians, standard deviations (SDs), minimums and maximums for continuously distributed variables and frequency counts and percents for categorical variables. The primary and secondary outcomes were analyzed on an intention-totreat basis. The primary outcome was binary and was compared with binomial regression and adjustment for clustering using generalized estimating equations and stratification by state and rural/urban location. Secondary outcomes included binary and continuous variables and were analyzed by comparison of proportions and comparison of means respectively, with adjustment for clustering and stratification by state and
Volume 23 | No. 7 | July 2012
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
5 minutes. ACT is hosted by Cancer Australia, an Australian federal government agency. The web site enables people to search for cancer trials by cancer type, stage, treatment, trial phase, recruitment status and age group. The information displayed about each trial includes routine data collected by trial registries as well as additional items requested by consumers during the web site’s design [19]. It also provides general information about clinical trials and two question prompt lists that suggest questions patients can ask their physician about clinical trial participation. The web site was publicly launched after recruitment and follow-up were completed and was thus only available to those participating in the trial. Medical oncologists in the intervention group received access to the ACT web site before recruiting patients. Consenting patients in the intervention group were given access to the web site before their scheduled appointment with their oncologist. Neither oncologists nor patients in the control group had access to the web site. The control group patients received the standard information provided by each medical oncologist before their scheduled appointment. See ‘Procedures’ section.
Annals of Oncology
original articles
Annals of Oncology location. Statistical analyses were carried out using SAS (version 9, SAS Institute, Cary, NC). This trial is registered with the ANZCTR, number ACTRN12608000273381 [22].
results
primary outcome The proportion of patients with whom the possibility of participation in a trial was discussed was 46% (84/183) in the intervention group and 34% (72/215) in the control group [adjusted difference 10%, 95% confidence interval (CI) 1% to 22%, P = 0.08] (Table 3).
discussion
Two percent (3/183) of the intervention group was invited to join a clinical trial compared with 5% (10/215) in the control group (adjusted odds ratio 2.86, 95% CI 0.58–13.99, P = 0.20).
This is the first evidence, albeit weak evidence, that a consumer-friendly cancer clinical trials web site can increase the proportion of patients with whom the possibility of participation in a clinical trial is discussed. We estimated the proportion of consultations in which the possibility of
Table 1. Physician characteristics at baseline for those in the main outcome analysis
Table 2. Patient characteristics at baseline for those in the main outcome analysis
secondary outcomes
Total number of physicians Mean age (years) Men Country of birth Australia Other Cancers treated Breast Lung Genitourinary State NSW Victoria Practice type Mostly salaried Mostly fee for service About equal Urban/rural Urban Rural
Intervention group
Control group
14 48 (7.3) 10 (71%)
14 47 (8.4) 10 (71%)
8 (57%) 6 (43%)
12 (86%) 2 (14%)
10 (71%) 7 (50%) 6 (43%)
11 (79%) 10 (71%) 9 (64%)
9 (64%) 5 (36%)
8 (57%) 6 (43%)
6 (42%) 4 (29%) 4 (29%)
4 (28%) 5 (36%) 5 (36%)
10 (71%) 4 (29%)
9 (64%) 5 (36%)
Data are N (N%) or mean (SD), unless otherwise stated. NSW, New South Wales.
Volume 23 | No. 7 | July 2012
Total number of patients Mean number of patients per cluster (range) Mean age (years) Men New patients Cancer site Breast Bowel Lung Other Cancer stage Early Advanced/recurrent Unknown Ability to search interneta Rarely or sometimes find Often or always find Do not search
Intervention group
Control group
183 13 (2–20)
215 15 (5–20)
60 (11.6) 84 (46%) 95 (52%)
59 (11.3) 91 (42%) 113 (53%)
55 (30%) 42 (23%) 24 (13%) 62 (34%)
77 (36%) 39 (18%) 21 (10%) 78 (36%)
89 (49%) 65 (36%) 29 (19%)
123 (57%) 67 (32%) 25 (12%)
38 (22%) 94 (53%) 44 (25%)
26 (13%) 135 (65%) 47 (23%)
Data are N (N%) or mean (SD), unless otherwise stated. Seven patients from the intervention group and seven patients from the control group did not answer this question. a
doi:10.1093/annonc/mdr585 |
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Thirty eligible medical oncologists working at 30 different medical oncology clinics were randomized. After randomization, two medical oncologists in each group withdrew and one replacement medical oncologist working at a different clinic was recruited to each group. As a result, there were 28 study clusters eligible for inclusion in the study with a total of 494 patients (Figure 1). Participant accrual commenced in June 2008 and was completed in July 2010. Follow-up was completed in October 2010. Physician characteristics assessed at baseline were similar between the intervention and control groups (Table 1), as were patient characteristics assessed at baseline (Table 2).
There was no significant difference in mean consultation length (P = 0.69), the mean number of issues about clinical trials discussed (P = 0.96), whether the patient raised the possibility of participation in a clinical trial (P = 0.81), and the percentage of consultations that included an extended clinical trial discussion (P = 0.23) (Table 4). The proportion self-reporting to have consented to a clinical trial was 10% (14/146) in the intervention group and 10% (20/ 194) in the control group (adjusted odds ratio 1.20, 95% CI 0.54–2.69, P = 0.65) (Table 4). When verified by each participating medical oncologist, similar results were obtained: 10% (18/180) in the intervention group and 13% (28/215) in the control group (adjusted odds ratio 0.64, 95% CI 0.24–1.70, P = 0.37). The mean seven-item knowledge score was 3.0 (SD 1.8) in the intervention group compared with 3.3 (SD 1.9) in the control group (P = 0.03). Among those that discussed the possibility of participating in a clinical trial, the mean decisional conflict score was 42 in the intervention group and 43 in the control group (P = 0.83). In the intervention group, 146/183 (80%) patients who had an audio recording completed the follow-up questionnaire. In total, 59% (86/146) stated that they used the web site. The reactions of these patients to the web site are shown in Table 5.
original articles
Annals of Oncology
Table 3. Primary outcome from audio recordings: proportion of patients with whom the possibility of participation in a clinical trial was discussed
Number of patients analyzed Number of patients with whom the possibility of participation in a clinical trial was discussed
Intervention group
Control group
183/251 (73%) 84/183 (46%)
215/242 (89%) 72/215 (34%)
Adjustedb Risk difference (95% CI)
P value
10% (−1% to 22%)
0.08
Data are N (N%) or mean (SD), unless otherwise stated. Adjusted for state and urban/rural location. CI, confidence interval. a
Table 4. Secondary outcomes from audio recordings and patient follow-up questionnaires
Secondary outcomes from audio recordings Patients analyzed Patients invited to join a clinical trial Length of consultation, minutes (mean) Consultations where a clinical trial was discussed Number of issues about clinical trials discussed (mean score out of 18) Patient (rather than physician) raised the issue of a clinical trial Consultations that included extended (more complex) discussions Secondary outcomes from patient follow-up questionnaires Patients with follow-up questionnaires and audio recordings Patients who consented to a clinical trial Number of patients with valid seven-item knowledge scale Seven-item knowledge scale (mean score out of 7) Number of patients with a valid Decisional Conflict Scalec Decisional conflict score (mean score out of 100)
Control group
183/251 (73%) 3/183 (2%) 29 84/183 (46%) 5.3 (4.1) 11/84 (13%) 13/84 (16%)
215/242 (89%) 10/215 (5%) 29 72/215 (34%) 4.9 (3.8) 10/72 (14%) 9/72 (13%)
146/183 (80%) 14/146 (10%) 146/146 (100%) 3.0 (1.8) 69/84 (82%) 42 (12.4)
194/215 (90%) 20/194 (10%) 194/194 (100%) 3.3 (1.9) 71/72 (99%) 43 (13.0)
Adjusteda Difference (95% CI)
P value
2.86 (0.58 to 13.99)b −1.8 (−11 to 7)
0.20b 0.69
0.05 (−1.7 to 1.8) 1% (−5% to 7%) 5% (−3% to 13%)
0.96 0.81 0.23
1.20 (0.54 to 2.69)b
0.65
−0.4 (−0.03 to −0.75)
0.03
−0.6 (−4.7 to 5.9)
0.83
Data are N (N%) or mean (SD), unless otherwise stated. Adjusted for state and urban/rural location. b Odds ratio reported as logistic model fitted. Binomial regression model failed to converge. c Only patients who reported discussing a trial with their medical oncologist were asked to complete the Decisional Conflict Scale which began with the statement: “Thinking about the possibility of joining a clinical trial, please indicate your agreement or disagreement with the statements below.” CI, confidence interval. a
participation in any trial was discussed might be 10% in the control group. But it was 34%, possibly because physicians who agreed to participate in the study were more interested in clinical trials than those in everyday oncology practice. Furthermore, physicians in both arms of the study were unavoidably aware of the general study purpose so may have been primed to discuss clinical trials more than usual. The high rate of discussion about trial participation in the control group reduced our capacity to detect a statistically significant effect. The effect of the web site may also have been diluted because the web site was not actively endorsed by physicians (we did not ask them to do this). We chose discussion about the possibility of participation in a clinical trial as our primary end point rather than actual recruitment to a trial for several reasons. Firstly, it was directly detectable from our audio recordings whereas a decision to join a trial may be taken later and may depend on many factors (discussed below). In addition, it is important that patients and physicians discuss all treatment options because | Dear et al.
this respects their autonomy and assists them to make an informed decision. Barriers to patient participation in cancer clinical trials include lack of knowledge about trial opportunities [5]. Furthermore, to find suitable clinical trials, patients need to understand medical terminology related to their cancer diagnosis and treatment and must understand the clinical trial research process in general [5]. Physicians’ lack of awareness of recruiting clinical trials and their difficulty in adequately informing patients about trials are also barriers [23, 24]. Cancer clinical trials web sites may overcome these barriers [12, 15, 16]. In our study, there was no significant difference in the proportions invited to join a trial and the proportions that consented to a trial, reflecting that there are many factors that influence a physician’s decision to invite a patient to join a trial and the patient’ decision to consent. These factors may include other barriers to trial participation, e.g. patients’ dislike of randomization, discomfort with the research process, complexity of the trial protocol, potential side-effects and the Volume 23 | No. 7 | July 2012
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Intervention group
original articles
Annals of Oncology Table 5. Intervention group patient reactions to the Australian Cancer Trials (ACT) web site Patients N (N%)
physician’s attitudes toward trials [5]. This is consistent with other research showing that it is difficult to identify interventions that increase participation in randomized trials [25, 26]. Oncologists have expressed concern that patients’ internet investigation increases consultation times but this is not supported by our results [23]. The control group scored higher on the randomized clinical trial knowledge scale, but the absolute difference was small (0.3). The scores are comparable with other published studies and are consistent with poor knowledge about randomized clinical trials [20, 24]. In 1993, Baum called for a way to better inform people about randomized trials and it is clear that > 15 years later public and patient education is urgently required [27]. The mean decision conflict score was similar in both groups. Notably, scores over 37.5 are associated with decision delay or feeling unsure about implementing a decision. So, regardless of access to the ACT web site, patients considering clinical trial participation faced uncertainty about their decision. Effective interventions to support decision making about clinical trial participation are required. The main limitation to our study was the higher dropout rate in the intervention group compared with the control group. 27% (68/251) of consultations were not recorded in the intervention group compared with 11% (27/243) in the control group. Of the audio recordings missed in the intervention group, 41% (28/68) came from two medical oncologists who, despite reminders, repeatedly forgot to audio record consenting patients’ consultations. This may reflect the previously reported reluctance of medical oncologists to audiorecord consultations and also be the reason that oncologists declined to participate in our trial [28]. Only 59% of intervention group patients (86/146) stated that they used the ACT web site despite all being taken to the ACT web site after completing the baseline questionnaire. It has been reported patients do not recall looking at information despite evidence they have seen it [29]. Also, patients may be more likely to use a web site endorsed by their physician [4, 12, 15, 16, 30]. To check whether the patients in our trial Volume 23 | No. 7 | July 2012
acknowledgements We would like to acknowledge the medical oncologists who participated in the trial: Ehtesham Abdi, Stephen Begbie, Karen Briscoe, David Bell, Robert Blum, Frances Boyle, Walter Cosolo, Stephen Della-Fiorentina, Paul De Souza, Anthony Dowling, Vinod Ganju, Geraldine Goss, John Grygiel, Anne Hamilton, Andrew Haydon, Rina Hui, Michael Jefford, Richard Kefford, Georgina Long, Gavin Marx, Sue-Anne McLachlan, Roger McLennan, Paul Mitchell, Philip Parente, Nicholas Pavlakis, John Scarlett, Eva Segelov, Anne Sullivan, Janette Vardy and Desmond Yip. We would also like to acknowledge their patients who kindly participated.
funding National Health and Medical Research Council (512380).
disclosure LA is the director of the Australian New Zealand Clinical Trials Registry and this nonfinancial interest may be relevant to the submitted work. All remaining authors declare no conflicts of interest.
references 1. Senate Community Affairs Committee. The Cancer Journey: Informing Choice, Report of the Inquiry into Services and Treatment Options for Persons with Cancer. Canberra, Australia: Australian Government 2005; 1–169. 2. Degner LF, Sloan JA. Decision making during serious illness: what role do patients really want to play? J Clin Epidemiol 1992; 45(9): 941–950.
doi:10.1093/annonc/mdr585 |
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
Intervention group patients with a follow-up 146/183 (80%) questionnaire Patients who reported they used the ACT web site 86/146 (59%) The ACT web site was useful 76/86 (88%) The ACT web site was helpful for learning more about 73/86 (85%) cancer clinical trials There was enough information on the ACT web site 77/86 (90%) Would use the ACT web site in the future 57/86 (66%) Would recommend the ACT web site to other patients 68/86 (84%) Felt it was important for patients with cancer to have access 78/86 (91%) to the ACT web site
may have been different from other oncology patients in some way related to their likelihood of using the web site, we conducted a substudy of 100 consecutive unselected patients (including both new and follow-up patients) seeing a medical oncologist who had not participated in our trial. The same proportion (59%, 47/80) of patients recalled looking at the web site as in our main trial suggesting the patients in our trial were broadly representative of those in everyday oncology practice. Among those who used the web site, reactions to the web site were positive (Table 5). The cancer journey is often long and considering clinical trial participation is not a one-off event. Access to a web site like ACT, even briefly before one consultation, may positively inform patients for future clinical trial discussions and decision making. This is the first reported randomized controlled trial assessing the effect of a consumer-friendly clinical trials web site on physician–patient discussions about participation in clinical trials. The findings of our study are the first, albeit weak evidence, that the ACT web site increased discussion rates about participation in clinical trials. However, it did not increase accrual rates. Further research is required to assess the impact of such a web site when it is endorsed by physicians and following wider dissemination by cancer consumer organizations to people affected by cancer.
original articles
| Dear et al.
17. Simon C, Schramm S, Hillis S. Patient internet use surrounding cancer clinical trials: clinician perceptions and responses. Contemp Clin Trials 2010; 31(3): 229–234. 18. UK Clinical Trials Gateway. http://www.ukctg.nihr.ac.uk/default.aspx. (10 December 2011, date last accessed). 19. Dear RF, Barratt AL, Crossing S et al. Consumer input into research: the Australian Cancer Trials website. Health Res Policy Syst 2011; 9(1): 30. 20. Ellis PM, Butow PN, Tattersall MHN et al. Randomized clinical trials in oncology: understanding and attitudes predict willingness to participate. J Clin Oncol 2001; 19(15): 3554–3561. 21. O’Connor AM. User Manual–Decisional Conflict Scale. http://decisionaid.ohri.ca/ docs/develop/User_Manuals/UM_Decisional_Conflict.pdf (10 December 2011, date last accessed). 22. Australian New Zealand Clinical Trials Registry, ACTRN1260800027338. http ://www.anzctr.org.au/trial_view.aspx?id=82593 (10 December 2011, date last accessed). 23. Helft PR, Hlubocky F, Daugherty CK. American oncologists’ views of internet use by cancer patients: a mail survey of American Society of Clinical Oncology members. J Clin Oncol 2003; 21(5): 942–947. 24. Sabesan S, Burgher B, Buettner P et al. Attitudes, knowledge and barriers to participation in cancer clinical trials among rural and remote patients. Asia Pac J Clin Oncol 2011; 7(1): 27–33. 25. Mc Daid C, Hodges Z, Fayter D et al. Increasing participation of cancer patients in randomised controlled trials: a systematic review. Trials 2006; 7: 16. 26. Treweek S, Mitchell E, Pitkethly M et al. Strategies to improve recruitment to randomised controlled trials. Cochrane Database Syst Rev 2010; (4): MR000013. . 27. Baum M. New approach for recruitment into randomised controlled trials. Lancet 1993; 341(8848): 812–813. 28. Stockler M, Butow PN, Tattersall MH. The take-home message: doctors’ views on letters and tapes after a cancer consultation. Ann Oncol 1993; 4(7): 549–552. 29. Nisbett RE, Wilson TD. Telling more than we can know: verbal reports on mental processes. Psychol Rev 1977; 84(3): 231–259. 30. Atkinson NL, Massett HA, Mylks C et al. User-centered research on breast cancer patient needs and preferences of an Internet-based clinical trial matching system. J Med Internet Res 2007; 9(2): e13.
Volume 23 | No. 7 | July 2012
Downloaded from http://annonc.oxfordjournals.org/ at Serials Section, Dixson Library on July 10, 2015
3. Chen X, Siu LL. Impact of the media and the internet on oncology: survey of cancer patients and oncologists in Canada. J Clin Oncol 2001; 19(23): 4291–4297. 4. Butow P, Devine R, Boyer M et al. Cancer consultation preparation package: changing patients but not physicians is not enough. J Clin Oncol 2004; 22(21): 4401–4409. 5. Mills EJ, Seely D, Rachlis B et al. Barriers to participation in clinical trials of cancer: a meta-analysis and systematic review of patient-reported factors. Lancet Oncol 2006; 7(2): 141–148. 6. Demmy TL, Yasko JM, Collyar DE et al. Managing accrual in cooperative group clinical trials. J Clin Oncol 2004; 22(15): 2997–3002. 7. Linda SE, Catherine C, Bekele BN et al. Generalizability of cancer clinical trial results. Cancer 2006; 106(11): 2452–2458. 8. Allison M. Can web 2.0 reboot clinical trials? Nat Biotechnol 2009; 27(10): 895–902. 9. Lopez-Gomez M, Feliu J, Sereno M et al. Internet use for medical research among cancer patients and their relatives in Spain. Ann Oncol 2008; 19(11): 1976–1977. 10. Helft PR. Breast cancer in the information age: a review of recent developments. Breast Dis 2004; 21: 41–46. 11. Dickerson SS, Boehmke M, Ogle C, Brown JK. Out of necessity: oncology nurses’ experiences integrating the internet into practice. Oncol Nurs Forum 2005; 32(2): 355–362. 12. Castleton K, Fong T, Wang-Gillam A et al. A survey of Internet utilization among patients with cancer. Support Care Cancer 2011; 19(8): 1183–1190. 13. Dear R, Barratt A, Askie L et al. Adding value to clinical trial registries: insights from Australian Cancer Trials Online, a website for consumers. Clin Trials 2011; 8(1): 70–76. 14. Australian Cancer Trials. http://www.australiancancertrials.gov.au. (10 December 2011, date last accessed). 15. Wallwiener M, Wallwiener CW, Brucker SY et al. The Brustkrebs-Studien.de website for breast cancer patients: user acceptance of a German internet portal offering information on the disease and treatment options, and a clinical trials matching service. BMC Cancer 2010; 10: 663. 16. Rodrigues MJ, Haberer S, Dionysopoulos D et al. Internet to boost patient accrual in oncology trials? A multiinstitutional AERIO study. Ann Oncol 2011; 22(2): 490–491.
Annals of Oncology