Correspondence AuSCR Office Public Health Florey Institute of Neuroscience and Mental Health 245 Burgundy Street Heidelberg Victoria, 3084 AUSTRALIA Free Call: 1800 673 053 Web: www.auscr.com.au Email: admin@auscr.com.au 12th July 2013 [ONHMRC has removed personal information] Professor Warwick Anderson National Health & Medical Research Council Dear Professor Anderson, Re: Response to Public Consultation on Review of Chapter 2.3 of the NHMRC National Statement: Qualifying or waiving conditions for consent submission We applaud the NHMRC for this initiative in providing greater clarity over situations of waiving conditions for consent in medical research. In particular, the use of ‘limited disclosure’, ‘waiver’ and ‘opt-out’ consent approaches in ‘low’ or ‘negligible risk’ situations or where explicit consent is not possible. Such information is important for researchers and participants in research, but particularly for governing bodies and Health Research Ethics Committees which vary in their understanding over such situations in the protection of the interests and wellbeing of potential participants in research. While the document is detailed and comprehensive, we wish to highlight some aspects in particular sections that may have been omitted or need emphasis. Section 2. Comparing different approaches Although this section serves the purpose of providing an overview comparison of the different approaches, further elaborated on in the next section, it would be important to emphasise 3 points: To qualify that ‘the provision of information to participants or ‘next of kin’ be included as it is not always practical to provide the information directly to the patient/subject, for example where subjects with critical illness are included in clinical registries during admission to hospital Table 1 would benefit from a section regarding the different level of participation of the clinical population, that is the success of registries in achieving very high levels of participation in order to avoid selection bias, whereas studies that use explicit consent generally include only 2-10% of potential patients due to various inclusion/exclusion criteria being applied to enhance internal validity. The section could be: Level of participation from source population - Explicit Consent 2-10%; Opt-out 95-100%; Waiver 80-100% As clinical quality registries suffer from being defined as ‘quality assurance’ rather than research, and therefore a priority for health funding agencies rather than research charities, we would recommend that the statement in section 2.3 be ‘recruitment of potential participants into research or clinical quality registries’. . Section 3. The rationale of an opt-out approach As emphasised in section 3.3, an opt-out approach is the preferred method for research that requires near 100% data collection to avoid unacceptable statistical bias for inclusion but also for outcome data during follow-up. For example, the recovery from disabling illnesses such as stroke, heart failure or hip fracture. We, therefore, recommend including a statement that ‘clinical quality registries, health services research or epidemiological studies which include primary data related to the onset of the illness or condition also require the ability to retain personal identifiable _____________________________________________________________________________________________________________________________________________________ Project Consortium: National Stroke Research Institute, The George Institute for Global Health, National Stroke Foundation and Stroke Society of Australasia. information to allow direct follow-up of people to assess outcomes some time from the initial hospitalisation or to probabilistically link with another dataset for analysis (e.g. cohort studies of the quality of care and outcomes)’. Section 4. Limiting the use of the opt-out approach Although we agree that the merit and integrity of any proposed research that requires use of optout consent should be subject to testing in regards to the National Statement on Ethical Conduct in Research Involving Humans, HRECs need education in appreciating that ‘opt out consent or waiver’ is the default gold standard for clinical quality registries. There should be an additional section, perhaps section 4.6, indicating that waiver of consent may be required for cluster controlled trials in health services research where subjects who are required to consent to a procedure that is perceived to be ‘out of the ordinary’, may be concerned or even alarmed, which would influence their participation (i.e. selection bias) or response to assessment of their health status or management and care during follow-up (i.e. drop-out and responder bias). Thus, in order to obtain the least biased, most reliable objective assessment of a low risk intervention within the range of routine clinical care, it is often better to use a guardian consent and avoid individual consent to the intervention. In such situations, individual consent may be required to non-routine interventions, access to medical records and the follow-up of subjects by agencies external to the health service. Flow Chart 1 – Opt-out Approach Flow Chart Under the flow diagram for 2a, ‘Data use for research anticipated’, it would be important to include ‘Waiver of consent’ as a relevant option as 3c, with 3d being ‘Exempt from ethical review’. For example, for a clinical quality registry that collects information about hospital care, a waiver of consent would apply to patients who die or are unconscious to avoid inconveniencing or distressing next-of-kin. In addition, it would mean that family members are not required to be contacted if a patient has died, again alleviating unnecessary potential distress and allowing near complete collection of data on all subjects. In addition, item 3b and elsewhere in the document, there is reference to the use of an opt-out approach only in low risk/negligible risk situations. However, this is potentially misleading and could confuse HRECs. Where a clinical quality registry obtained personal identifying information to be able to contact registrants for follow-up, we have found that the majority of HRECs consider this is not low risk and have generally approved the use of an opt-out rather than a waiver of consent. Indicating that opt-out is only applicable for 'low risk situations' may compromise a large number of existing initiatives and the wording around these statements should be reviewed. Furthermore, data that would qualify for box 2b, ‘Data use for research not anticipated’, would generally fall under the category of administrative data. It is not clear what the value of an opt-out approach would be over waiver of consent for these types of data, especially if the use of the data has already been identified as being of low or negligible risk and of significant value. We do not support the use of opt-out consent in situations where waiver of consent would otherwise be applicable; it seems meaningless. This is particularly relevant to health services research where it is impractical to contact large numbers of people to undertake low risk research. Section 5. Proposed amendments to the National Statement We suggest that the following be amended in 2.3.10a to cover the situation where a HREC application for a clinical quality registry or other research project may not be deemed ‘low risk’ since personal identifying information are collected, but use of an opt-out approach is still justified. 2 A statement on data linkage should be included. For example, that opt-out consent provides consent for data linkage activities in situation where participants have been made aware that their data may be linked with other health information data for quality assurance purposes and for HREC approved research purposes. The timeliness between when the data were collected and when opt-out consent is obtained needs to be taken into account. For secondary use of data that may have been collected a year or more ago, consideration needs to be given to the research population and the likelihood that participants may have changed address or died. In these cases it would not be possible to know if most people had been given the opportunity to opt-out, and waiver of consent may be a better option. General Comments: We believe the addition of information on the opt-out approach in Chapter 2.3 of the National Statement on Ethical Conduct in Research Involving Humans is important and completes the information in this section reflecting what currently happens in the field. It would be useful for the document to include specific definitions to differentiate between: research studies; clinical quality registries; quality of care assessment projects such as clinical audit; administrative data; and stored data. Yours sincerely Dr Craig Anderson Chair, AuSCR Management Committee Professor of Stroke Medicine, University of Sydney Head, Neurology Department, Royal Prince Alfred Hospital Senior Director, The George Institute for Global Health A/Prof Dominique Cadilhac Data Custodian, AuSCR Head, Public Health Stroke Division Florey Institute of Neuroscience and Mental Health 3