In this issue of the NZMJ,1 Hadorn et al argue that recruitment to clinical trials in New Zealand is hampered by the willingness of patients to be allocated into the control arm of a study. They argue that we could improve the efficiency and effectiveness of research by a method of pre-randomisation. This would involve allocating half of the patients to a "standard" research arm (without consent), and only approaching the patients allocated to the interventional arm for consent. Recruitment rates are bolstered because control group patients are entered automatically and only patients in the interventional arm need be approached.The question that Hadorn et al pose is: why should we bother with individual patient consent, particularly if it is an obstacle to research? We consider that there are practical, scientific and ethical problems with this 'pre-randomisation' model that mean it should not be adopted in the setting of cancer drug trials.There are no New Zealand data available regarding the proportion of patients eligible for cancer trials that accept or decline participation. Hadorn et al cite US figures where only 5% of people with cancer participate in trials. However patients eligible for drug trials constitute only a fraction of cancer patients, yet represent the thrust of the Hadorn article.The majority of cancer patients do not have advanced disease, and do not require systemic therapy. Phase one studies and single arm phase 2 studies do not involve randomisation so the pre-randomisation model is not relevant to these trials. Therefore the pre-randomisation model is only potentially applicable to randomised phase 2 and 3 studies involving novel drugs or novel applications of existing treatments - a very narrow band of cancer patients—so is unlikely to produce a significant increase in trial participation.In our experience as clinical trialists we do not see the unwillingness of patients to be randomised to a control arm as the major issue. The main barriers are narrow eligibility criteria, administrative burden,2 lack of centrally funded infrastructure, too few and usually unfunded research support staff, and marginalisation of research from core DHB business. These features result in a small range of trials available, slow start up, and high burden on researchers.3 Where patients are eligible for studies there is frequently high uptake and a general willingness to accept randomisation.Pre-randomisation would result in studies of poor scientific quality. Standard clinical management is not analogous to the procedures for the control arm of a clinical trial. Firstly, a trial must offer the best existing therapy to the control arm, but in day to day practice there may be more than one standard option. For example there are at least 6 reasonable first line chemotherapy options that could be chosen for a patient with metastatic gastric cancer based on individual patient characteristics.4Pre-randomisation would require us to narrow the selection of therapies for "control" patients without their knowledge - failure to restrict would lead to problems with generalisability. Second, participation in a cancer clinical trial involves much more than "standard care" regardless of whether the patient is allocated to the control arm or the experimental treatment. Trial protocols almost invariably include additional CT scans, additional tests, additional tissue collection, and more rigorous recording and analysis of treatment and outcome related data from all participants.All contemporary phase 3 cancer studies require this type of detail for comparison of safety when using highly toxic therapies, for biomarker development and for identification of potential sub-groups who may derive particular benefit (or harm). This is in comparison to "pragmatic" phase 3 studies in other areas where perhaps only survival may be measured. Participation in these cancer trials, even in the control arm, always goes above and beyond the standard of care for non-trial patients. We believe it is unethical not to inform potential participants of the additional burdens of trial requirements and obtain their consent to participate.Thirdly, whilst routine data sources are valuable they are insufficient to provide the detailed, accurate or timely data that are required for cancer trials. Current routine data collection systems are poor at defining even stage; they collect negligible comorbidity or adverse event data, no information on radiographic response or when the tumour progresses, nothing on treatment intensity, dose reductions or interruptions, and certainly no information on quality of life. While these could be improved, standard cancer trial methods for data capture and management demonstrate the difficulty of collecting routine data with the rigour required for meeting Good Clinical Practice Guidelines for clinical trials.5Timely and accurate data are also required for preparing reports for independent Data Monitoring Committees, and this is unable to be provided by routine data sets. These committees are charged with safeguarding the rights of patients by reviewing current, accurate comparative data on the emerging benefit to risk profiles of the treatments, and, where warranted, making recommendations regarding early termination of the trial. Current routine data sets are therefore vastly inferior to clinical trial grade data.Changing our model of randomisation would set us apart from the international research community and likely result in marginalisation which would lessen the opportunity for patient participation in trials rather than enhance it. The FDA and EMEA require that data submitted for registration of products is collected from participants who have given their consent freely.6 If they do not accept data collected from pre-randomisation studies (in the context of cancer trials) then there would be no value at all to international companies conducting research in NZ as they would not be able to use this data for registration in major markets.It is exceedingly rare for Phase 3 trials to recruit solely from New Zealand. While improvements in recruitment may help to address this, in reality the evaluation of modern targeted therapies will involve much larger patient pools, so still require international collaborative trials.Therefore even if we changed our model, it is most unlikely that offshore pharmaceutical sponsors and collaborative trials groups would accept one model of randomisation and consent here whilst the rest of the world operates under a different model. We would risk becoming a research pariah rather than leader.There are clear circumstances where individual patient consent is unable to be obtained before randomisation. For example emergency situations, or where a patient is unable to provide consent (for example if they are unconscious in the Intensive Care Unit). In these situations consent is almost always sought at the first opportunity, or from the relatives. Cluster randomisation occurs where institutions or communities are randomised rather than individuals. These exemptions are not applicable or analogous to cancer trials.Over and above these scientific problems, the ethical problems are perhaps more troubling as pre-randomisation subordinates the right to consent to treatment to the desire to achieve recruitment targets. This threatens public trust in cancer trials generally, and would undermine the central values of trust and honesty in the doctor-patient relationship. Where an individual is competent and there is opportunity to gain consent, the argument to override consent is weak, particularly when the scientific benefits are not clear.For the doctor sitting with a patient the principles of honesty and respect for autonomy are paramount (save for real and imminent threats to public safety). Patients with cancer are particularly vulnerable. They are frightened for themselves and their families.Introducing a research proposition into this context can be challenging but this is not a reason to resile from the responsibility for honest and direct communication. We would feel deeply concerned about enrolling patients onto a control arm of a study without their explicit knowledge and consent, particularly where it may restrict the treatment options we discuss, and necessitate additional tests, data and often tissue collection. We do not assume the right to make these decisions on behalf of competent individuals simply for the perceived benefits of clinical research.We agree with the authors of the viewpoint that it is desirable to increase participation in clinical trials. We differ in how to achieve this. We agree that there is a need to build national capacity and the range of trials available. We need to encourage students, registrars, specialists, nurses and allied health staff to practice medicine with critical enquiry.We need to generate hypotheses from our observations, and then test them (with participants who agree to partake in that enquiry). We should be proposing questions where there is genuine equipoise, and have confidence in our ability to convey this. We should be reducing the burden on research administration through complex and often duplicative DHB contracting processes.We need clinicians to have enough time to explain complex ideas, and enough research staff to support the endeavours. If we are failing in delivering a high quality research environment, we do not need to blame the patients.\r\n

In this issue of the NZMJ,1 Hadorn et al argue that recruitment to clinical trials in New Zealand is hampered by the willingness of patients to be allocated into the control arm of a study. They argue that we could improve the efficiency and effectiveness of research by a method of pre-randomisation. This would involve allocating half of the patients to a "standard" research arm (without consent), and only approaching the patients allocated to the interventional arm for consent. Recruitment rates are bolstered because control group patients are entered automatically and only patients in the interventional arm need be approached.The question that Hadorn et al pose is: why should we bother with individual patient consent, particularly if it is an obstacle to research? We consider that there are practical, scientific and ethical problems with this 'pre-randomisation' model that mean it should not be adopted in the setting of cancer drug trials.There are no New Zealand data available regarding the proportion of patients eligible for cancer trials that accept or decline participation. Hadorn et al cite US figures where only 5% of people with cancer participate in trials. However patients eligible for drug trials constitute only a fraction of cancer patients, yet represent the thrust of the Hadorn article.The majority of cancer patients do not have advanced disease, and do not require systemic therapy. Phase one studies and single arm phase 2 studies do not involve randomisation so the pre-randomisation model is not relevant to these trials. Therefore the pre-randomisation model is only potentially applicable to randomised phase 2 and 3 studies involving novel drugs or novel applications of existing treatments - a very narrow band of cancer patients—so is unlikely to produce a significant increase in trial participation.In our experience as clinical trialists we do not see the unwillingness of patients to be randomised to a control arm as the major issue. The main barriers are narrow eligibility criteria, administrative burden,2 lack of centrally funded infrastructure, too few and usually unfunded research support staff, and marginalisation of research from core DHB business. These features result in a small range of trials available, slow start up, and high burden on researchers.3 Where patients are eligible for studies there is frequently high uptake and a general willingness to accept randomisation.Pre-randomisation would result in studies of poor scientific quality. Standard clinical management is not analogous to the procedures for the control arm of a clinical trial. Firstly, a trial must offer the best existing therapy to the control arm, but in day to day practice there may be more than one standard option. For example there are at least 6 reasonable first line chemotherapy options that could be chosen for a patient with metastatic gastric cancer based on individual patient characteristics.4Pre-randomisation would require us to narrow the selection of therapies for "control" patients without their knowledge - failure to restrict would lead to problems with generalisability. Second, participation in a cancer clinical trial involves much more than "standard care" regardless of whether the patient is allocated to the control arm or the experimental treatment. Trial protocols almost invariably include additional CT scans, additional tests, additional tissue collection, and more rigorous recording and analysis of treatment and outcome related data from all participants.All contemporary phase 3 cancer studies require this type of detail for comparison of safety when using highly toxic therapies, for biomarker development and for identification of potential sub-groups who may derive particular benefit (or harm). This is in comparison to "pragmatic" phase 3 studies in other areas where perhaps only survival may be measured. Participation in these cancer trials, even in the control arm, always goes above and beyond the standard of care for non-trial patients. We believe it is unethical not to inform potential participants of the additional burdens of trial requirements and obtain their consent to participate.Thirdly, whilst routine data sources are valuable they are insufficient to provide the detailed, accurate or timely data that are required for cancer trials. Current routine data collection systems are poor at defining even stage; they collect negligible comorbidity or adverse event data, no information on radiographic response or when the tumour progresses, nothing on treatment intensity, dose reductions or interruptions, and certainly no information on quality of life. While these could be improved, standard cancer trial methods for data capture and management demonstrate the difficulty of collecting routine data with the rigour required for meeting Good Clinical Practice Guidelines for clinical trials.5Timely and accurate data are also required for preparing reports for independent Data Monitoring Committees, and this is unable to be provided by routine data sets. These committees are charged with safeguarding the rights of patients by reviewing current, accurate comparative data on the emerging benefit to risk profiles of the treatments, and, where warranted, making recommendations regarding early termination of the trial. Current routine data sets are therefore vastly inferior to clinical trial grade data.Changing our model of randomisation would set us apart from the international research community and likely result in marginalisation which would lessen the opportunity for patient participation in trials rather than enhance it. The FDA and EMEA require that data submitted for registration of products is collected from participants who have given their consent freely.6 If they do not accept data collected from pre-randomisation studies (in the context of cancer trials) then there would be no value at all to international companies conducting research in NZ as they would not be able to use this data for registration in major markets.It is exceedingly rare for Phase 3 trials to recruit solely from New Zealand. While improvements in recruitment may help to address this, in reality the evaluation of modern targeted therapies will involve much larger patient pools, so still require international collaborative trials.Therefore even if we changed our model, it is most unlikely that offshore pharmaceutical sponsors and collaborative trials groups would accept one model of randomisation and consent here whilst the rest of the world operates under a different model. We would risk becoming a research pariah rather than leader.There are clear circumstances where individual patient consent is unable to be obtained before randomisation. For example emergency situations, or where a patient is unable to provide consent (for example if they are unconscious in the Intensive Care Unit). In these situations consent is almost always sought at the first opportunity, or from the relatives. Cluster randomisation occurs where institutions or communities are randomised rather than individuals. These exemptions are not applicable or analogous to cancer trials.Over and above these scientific problems, the ethical problems are perhaps more troubling as pre-randomisation subordinates the right to consent to treatment to the desire to achieve recruitment targets. This threatens public trust in cancer trials generally, and would undermine the central values of trust and honesty in the doctor-patient relationship. Where an individual is competent and there is opportunity to gain consent, the argument to override consent is weak, particularly when the scientific benefits are not clear.For the doctor sitting with a patient the principles of honesty and respect for autonomy are paramount (save for real and imminent threats to public safety). Patients with cancer are particularly vulnerable. They are frightened for themselves and their families.Introducing a research proposition into this context can be challenging but this is not a reason to resile from the responsibility for honest and direct communication. We would feel deeply concerned about enrolling patients onto a control arm of a study without their explicit knowledge and consent, particularly where it may restrict the treatment options we discuss, and necessitate additional tests, data and often tissue collection. We do not assume the right to make these decisions on behalf of competent individuals simply for the perceived benefits of clinical research.We agree with the authors of the viewpoint that it is desirable to increase participation in clinical trials. We differ in how to achieve this. We agree that there is a need to build national capacity and the range of trials available. We need to encourage students, registrars, specialists, nurses and allied health staff to practice medicine with critical enquiry.We need to generate hypotheses from our observations, and then test them (with participants who agree to partake in that enquiry). We should be proposing questions where there is genuine equipoise, and have confidence in our ability to convey this. We should be reducing the burden on research administration through complex and often duplicative DHB contracting processes.We need clinicians to have enough time to explain complex ideas, and enough research staff to support the endeavours. If we are failing in delivering a high quality research environment, we do not need to blame the patients.\r\n