National Ethics Advisory Committee
Kāhui Matatika o te Motu
The type of study researchers choose for their research should be the one best suited to answering the study question while meeting ethical standards. These Standards broadly categorise research as either observational or interventional, while noting that many studies contain elements of both.
Studies can also be either invasive or non-invasive, either low risk or high risk, either therapeutic or non-therapeutic, and either comparative or non-comparative. A study’s features and design, and the context in which it is carried out, all factor into the ethical considerations that researchers must make.
In an observational study, in contrast to an interventional (or experimental) study, the researcher does not influence the assignment of any variable. Instead, the researcher observes and analyses natural relationships between variables and outcomes, and records them.
The prospective collection of data – such as from blood samples, imaging or questionnaires – does not change the status of a study from observational to interventional. Observational studies are not automatically of minimal risk; indeed, they may involve an invasive or high-risk means of collecting data from participants, and therefore pose a risk of privacy harm. Researchers must rigorously identify, gauge, minimise and manage such risks.
10.1 An invasive means of collecting data in an observational study is justified only when the importance of the objective outweighs the inherent risks and burdens to the participant.
Observational studies include case control studies, cohort studies, cross-sectional studies, case reports, case series and descriptive studies.
Qualitative research is a type of observational research in which researchers collect text-based rather than numeric information by methods such as interviews, case studies, focus groups, ethnography or direct observation. This strategy is suited to studies that seek to understand the health or treatment experience of individuals or communities.
Mixed-methods research incorporates both qualitative and quantitative methods. An example is research involving a debriefing interview that takes place after an intervention, to learn of barriers or facilitators to implementation of the intervention.
10.2 Researchers must consider whether the nature and duration of the research interaction may have a significant effect on a participant within a qualitative study.
10.3 Qualitative research methods may involve discussion of sensitive topics, or reliving upsetting experiences. In such cases, researchers should make participants aware of the nature of the methods beforehand, and develop a plan for managing any distress participants may experience.
10.4 A close relationship may develop between researcher and participant as a result of direct sharing of personal information over a prolonged period. This may result in researchers being more likely to step outside of the research role in their interactions with participants (e.g. to provide counselling support). Where this happens, researchers should make it clear to the participant involved that they are no longer acting in the capacity of a researcher.
10.5 Researchers should take care to only provide support if this is in line with their professional skills.
10.6 Researchers must take additional care to preserve confidentiality when they publish qualitative data, because of small sample sizes and potentially identifiable contextual information.
10.7 Where possible, researchers should provide transcripts of audio recordings to participants before they undertake analysis, so that participants may adjust potentially identifying or misleading content.
In an intervention study, the researcher controls and studies the intervention(s) that they provide to participants for the purpose of adding to knowledge of the health effects of the intervention(s). The term ‘intervention study’ is often used interchangeably with ‘experimental study’.
Intervention studies generally present more risk of harm than observational studies, however not all intervention studies pose significantly greater than minimal risk, and researchers should take care to avoid automatic classification.
An intervention study may evaluate:
A randomised controlled trial is often the best way of addressing questions about the effectiveness of treatments or preventions. Such a trial allocates participants to intervention arms in a way that minimises the influence of confounding factors (variables that are independently associated with both the exposure and the outcome of interest, are not on the causal pathway between the exposure and the outcome, and can distort a true relationship between the exposure and the outcome or create a spurious association).
10.8 Studies must be scientifically sound in order to be ethical. Researchers must design and conduct randomised controlled trials in a way that minimises systematic error (bias).
10.8.a Researchers should pay close attention to the means of randomisation (random sequence generation and allocation concealment), blinding of participants and personnel, blinding of outcome assessment, complete outcome data, and avoidance of selective reporting.
10.9 Researchers should ensure that participants enrolled in therapeutic intervention studies have post-study access to the best-proven intervention, where such an intervention is available. If the best-proven intervention is not going to be available to participants once the study has completed, researchers must clearly explain this to participants prior to their consent to participate.
10.10 Studies comparing two or more interventions should meet the standard of equipoise; that is, the expert medical community should be genuinely uncertain as to the overall balance of risks and benefits between the interventions offered in the study.
Potential harms to individual participants in intervention studies can include physical harms such as adverse events or lack of efficacy from the intervention, psychological harm and harm from receiving a placebo. At a community level, potential harms include an inequitable burden on a community without a corresponding benefit. Sometimes the potential benefits of an intervention study accrue to one group of individuals while its harms are experienced by a different group.
10.11 All trials of interventions for clinical conditions must include data collection and reporting of adverse events.
10.12 Researchers should conduct initial tests of a new intervention under lower-risk study conditions before escalating to higher-risk conditions, even if the new intervention is likely to be more therapeutically beneficial for a higher-risk population.
10.12.a This approach may not be clinically appropriate if a new intervention is not applicable to a lower-risk group.
The safest possible manner of refining and testing techniques in an intervention study is to escalate doses incrementally, throughout testing. This approach also helps researchers to minimise the prospect of catastrophic events that might harm participants and undermine confidence in the development of interventions.
10.13 Researchers should justify dose level, dose escalation and cohort size in relation to international best practice.
10.14 Researchers should use methods such as sentinel dosing[1] along with careful safety monitoring, to protect participants from unnecessary risks.
10.15 Participants who benefit from a study intervention during a clinical trial should have ongoing access to the study intervention for as long as it is clinically beneficial.
10.15.a If continued access is not available, researchers must inform participants of this prior to seeking their consent to participate.
10.16 Researchers must clearly explain to all participants the arrangements for access to interventions after the study, including any uncertainties about that access.
10.17 Sponsors and researchers should seek access to effective interventions for study and target populations after the study, in discussion with relevant authorities.
An intervention study meets the equipoise standard if the evidence is ‘equally poised’ as to the overall balance of risks and benefits of each of the interventions offered in the study. As a result, in a study that meets the standard, no one can establish in advance which of the groups in a proposed study will be better off through participating in the research.
10.18 For any study comparing two or more interventions, researchers should design the study to meet the equipoise standard. They should not randomise or assign study participants to different interventions when available evidence demonstrates that one intervention has a better expected overall balance of benefits over risks than the other(s).
10.18.a However genuinely felt, an individual feeling of certainty or uncertainty is not enough to demonstrate the presence or absence of equipoise.
10.19 It may be justifiable to randomise participants to study arms that are not in equipoise if the better arm is not available as part of standard care and can only be offered to participants who are randomised to the treatment arm.
Using controls in clinical trials may create the potential for conflict between the demands of sound science and the obligation to safeguard the health and welfare of study participants. Controls in clinical trials can include a placebo (an inert substance or sham procedure having the goal of isolating the clinical effects of an investigational intervention) or an active control (where the investigational intervention is compared with an established effective intervention).
10.20 Participants in the control group of an interventional trial should receive an established effective intervention if one exists, unless researchers can ethically justify a different approach. The choice of control must be appropriate for the participants and the study design.
10.21 In general, researchers should design studies to generate accurate scientific information without delaying established effective interventions for, or withholding them from, participants. Established effective interventions may include interventions that, while not necessarily the best proven intervention, are professionally recognised as reasonable options.
10.22 Researchers who propose to delay or withhold established effective interventions must provide compelling justification for doing so. They must fully inform participants of treatments available to them outside the study and explain how these differ from study participation.
The risks of a placebo control are typically very low (e.g. ingesting an inert substance), but occasionally can be considerable (e.g. undergoing a sham procedure such as surgical incision under general anaesthesia).
10.23 Researchers must consider and minimise risks associated with placebos. They may use a placebo as a control when:
10.24 Compelling scientific reasons for placebo controls may exist when a trial cannot distinguish effective from ineffective interventions without a placebo control (Millum and Grady 2013). Examples of ‘compelling scientific reasons’ include the following.
10.25 Researchers must decrease the period of placebo use to the shortest possible time that is consistent with achieving the scientific aims of the study. They may reduce the risks by permitting the placebo arm to change to active treatment (‘escape treatment’), either during or after the study. In this case researchers should actively monitor participants, and should establish a threshold beyond which the participant should be offered the active treatment.
A cross-over study is a specialised type of randomised controlled study in which the order that treatments are given is randomised. There may be more than one active treatment period as well as a placebo period. All participants are exposed to each active or placebo intervention and thus act as their own control. This reduces interparticipant variability, so fewer participants are required, which brings cost and safety advantages. This design is often used in uncontrollably heterogeneous circumstances (e.g. drug-drug interaction studies, where baseline interpatient variability makes it difficult to see an effect if done in different patients). Treatment periods are typically separated by a ‘wash-out’ interval to prevent a carryover effect from the previous intervention.
10.26 Researchers must apply cross-over designs judiciously, having regard to any enduring effects of the intervention(s) beyond each treatment period and the stability of the background disease.
10.26.a A fundamental assumption inherent within the cross-over design is that participant measures return to baseline before the start of each new treatment period. Cross-over design is therefore unsuited to the study of interventions that researchers expect to produce an enduring response. A wash-out interval between periods of treatment should be sufficient to allow the condition to return to baseline.
10.26.b A cross-over design is unsuitable where the activity of the background condition is unstable.
10.26.c A cross-over design is unsuited to studies where there is a reasonable expectation of a high dropout rate.
Not all clinical trials are designed to test whether a new treatment is superior to existing therapies. An equivalence trial aims to demonstrate that the efficacy of a new treatment lies within predetermined upper and lower boundaries of a standard treatment, being neither better nor worse than the standard. A non-inferiority trial aims to demonstrate only that a new treatment is no worse than a standard treatment.
10.27 Equivalence and non-inferiority trials must use an active control that is a current proven effective treatment for the indication being studied.
10.27.a To fairly test the equivalence or non-inferiority of a new intervention, the active control to which it is being compared must be used in a dose, formulation and population matching the trials in which it was proven to be effective. The selected response measure should be sufficiently sensitive to detect the expected effects of the intervention and control.
10.28 When a researcher proposes an equivalence or non-inferiority design, there must be a potential non-efficacy advantage associated with the trial intervention.
10.28.a Non-efficacy advantages include a more favourable side-effect profile, a dosing regimen that enhances compliance, lower cost and the potential to broaden treatment options to people with an idiosyncratic reaction to the standard treatment.
An adaptive design trial includes an opportunity planned in advance to modify one or more specified aspects of the study design and hypotheses based on analysis of data (usually interim data) from participants in a study. Researchers analyse the accumulating study data during the study, with or without formal statistical hypothesis testing. The adaptation process generally continues throughout the trial, following the trial protocol. Modifications may be to dosage, randomisation proportions, sample size, the intervention(s) undergoing trial and the patient selection criteria. Importantly, adaptation within the trial protocol is set before the trial begins, specifying the adaptation processes. In some cases, such as Bayesian adaptive trials, the adaption is driven by the early analysis of interim results of the participants previously recruited. The algorithm determining allocation may be specified, but the actual adaptive process is not determined until trial begins
Platform trial is a broad term for a type of clinical trial with a single master protocol in which multiple treatments and/or disease types are evaluated simultaneously. Adaptive platform trials may simultaneously investigate multiple categories of treatment for a single complex condition. By adding or dropping options within a category depending on analysis of interim results, researchers can investigate the possibility of synergy between treatments in a timely manner that is not possible if each combination is the subject of a single trial.
In oncology, umbrella trials allocate treatment for a single tumour type from a pool of treatment possibilities, according to participant biomarkers. Basket trials allocate differing tumour types with shared biomarkers to a common treatment. Both designs can include rules for adding or dropping treatment arms. Similar approaches can be used in other disease groups.
Adaptive trials using either Bayesian or standard frequentist statistical analysis have the potential to reduce participants’ exposure to ineffective treatments, hasten treatment development, conserve research resources and increase the likelihood that the trial will deliver a clinically useful result.
There are questions about how the complex designs of certain adaptive trials meet the substantial evidence standard required for new drug approvals, for example. Adaptive trials can create ethical challenges involving equipoise (given that randomisation rates may change throughout the study as one arm is shown to be more beneficial) and informed consent (because this kind of study is difficult to explain to participants). Safety monitoring and statistical analyses are especially important in this kind of study design, and researchers need specific expertise to perform them well.
10.29 In advance of initiating an adaptive clinical study, researchers should:
Cluster randomised trials (CRTs) involve randomly allocating groups of individuals or clusters such as communities, hospitals or medical practices to different interventions, either using randomisation or step-wedge design[2]. They pose distinct ethical challenges for several reasons, including the following[3].
Under New Zealand law, a CRT is considered research which a patient under the Code has a right to be fully informed about and for which consent must be in writing. Every consumer has the right to the information that a reasonable person in the circumstances would expect to receive, including notification of any proposed participation in teaching or research and including whether the research requires and has received ethical approval.[4]
NEAC recognises that there is a tension between ethics and the legal framework for consent, as cluster randomised trials generally are not designed to seek consent. This tension creates a legal barrier to some research that may otherwise meet ethical standards. NEAC is aware of the tension and support a review of the law in this area.
Generally, wherever the treatment (medical or otherwise, including gathering additional information from tests) under the study protocol is different, or for a CRT where it could be different to what the person would have received if not participating in the study, then consent for the research is required[5].
10.30 In the case of cluster randomised studies that do not involve health research, (i.e. are not under the Code of Rights) individual consent to participate in the trial should not be required if gaining that consent is impracticable, and the potential benefits from the study outweigh the harms.
10.30.a An example of such a study might be one examining the effects of a media campaign to reduce adolescent tobacco use.
10.31 When a CRT involves a group or community whose interests are substantially affected by the CRT, researchers should consult with representatives of the group to inform the study design, conduct and reporting, and to obtain their agreement to the study.
10.32 As far as possible, and whenever appropriate, researchers should involve community representatives in the planning and conduct of studies, and give community members the opportunity to contribute to them (e.g. through submissions or public meetings).
10.32.a lsquo;Community representatives’ include all the intended recipients of experimental (or control) interventions (including environmental manipulations), and those from whom the researcher intends to collect or access personal health information.
10.33 Participants may be patients or health care workers, or both. For example, in CRTs that target interventions to health care workers (e.g. an alternative hand-washing protocol), researchers may use aggregate data from patients’ records to judge the effectiveness of the intervention.
10.33.a Research involving health care workers may involve a power imbalance (that between employer and employee or entailed by a hierarchy of employees). When their research involves this or a similar power imbalance, researchers must consider procedures that particularly safeguard all participants’ privacy and freedom to consent.
10.34 For cluster randomised trials that involve health or disability interventions (service provision), researchers must obtain written informed consent from all participants, or someone legally entitled to consent on their behalf.
10.34.a In such cases, integrated consent may facilitate recruitment for CRTs.
10.34.b In cases where participants are adults who are unable to provide informed consent, researchers should review the Standards outlined in research with adults who cannot provide consent.
Epidemiological and public health research studies often involve the use of different study methods and tools on a large number of research participants, in single or multiple settings. Many include features of observational studies (such as cross-sectional studies), case control studies, cohort studies, case reports, case series and other descriptive studies, as well as features of intervention studies (such as field trials and CRTs, stepped-wedge and quasi-experimental[6] study designs involving groups, geographic areas, institutions or systems collectively rather than individually).
10.35 For interventional research conducted in the context of health care or public health, researchers should additionally determine[7]:
10.35.a whether the project involves the systematic investigation of the safety, efficacy and/or effectiveness of an intervention;
10.35.b if the research involves exposure to an intervention for which the safety or efficacy, or both, is not well understood:
10.35.c whether it is likely or possible that the intervention will be of therapeutic benefit and
10.35.d whether there is a realistic possibility that the intervention being studied will be at least as beneficial overall as standard treatment, taking into account effectiveness, burden, costs and risks;
10.35.e where patient care is combined with intent to contribute to knowledge, that any risks of participation should be justified by potential benefits to which the participants attach significance. The prospect of benefit from research participation should not be exaggerated, either to justify to the reviewing body a higher risk than that involved in the participant’s current treatment or to persuade a participant to accept that higher risk;
10.35.f whether the intervention or other research procedures are without likely benefit to participants. For such research to be ethically acceptable, any known or emerging risks to the participants must not be greater than the risks that would be associated with the health condition and its usual care.
The health sector has a critical role in conducting health services research and in translating research findings into policy and practice[8]. It can encourage practitioners to take up new ideas by involving health professionals in research, evaluation, quality improvement and improved service delivery.
Research questions focus on what needs to be done to improve health system performance and how to influence policy to strengthen health systems, and frequently focus on ‘hardware aspects’ of healthcare, such as financing, information technology, service delivery, human resources and governance or ‘software aspects’ (norms, values and power relations) of health systems (Pratt et al. 2017). Importantly, they typically do not recruit patients or consumers directly, rather they are targeted at the population level, health system process level or health workforce level. Health systems research encompasses and overlaps with other types of research including comparative effectiveness research (CER), implementation research and activities (that may or may not be research) like quality improvement (QI).
The ethics of these areas of research are a new field and will require further work and frameworks to be developed.
Comparative effectiveness research (CER) compares established treatments (standards of care) or services when existing evidence is insufficient to determine which has the superior balance in terms of efficacy or safety. Such research aims to address variation in use of standards of care, or service delivery. It is often lower risk, and is integrated into routine delivery of care or support services and may require consideration of the dual role of service provider and researcher.
10.36 Researchers considering a CER design must first thoroughly assess the range and quality of published evidence to identify existing knowledge, along with points of uncertainty and disagreement. The trial design should respond to these disagreements by gathering additional evidence that may permit researchers to differentiate between treatments and identify specific groups to whom the treatments should be applied.
10.37 Researchers must clearly distinguish any specific risks associated with randomising individuals to one or other trial arm from the risks reasonably expected from assigned clinical treatment. Researchers must plan to minimise these risks by strategies such as:
10.38 CER designs must uphold the other ethical standards for conducting clinical research described in this publication. These standards include that participants:
Health systems research has been defined as ‘a search for knowledge which contributes to health systems strengthening and our understanding of health systems (Pratt et al. 2017).’. Health services research examines healthcare at the organizational level (e.g. clinics and hospitals).
The standards below provide high level principles that should guide health services research.
10.39 Researchers must consider health services research impact on reducing health inequity, and avoid widening of health disparities (i.e. where incentives[10] are taken up differently by different groups)
10.40 Researchers must ensure health services research includes people experiencing vulnerability.
10.41 Researchers must carefully consider what a ‘standard of care’ is when conducting health services research that involves changing or implementing a new standard of care.
10.42 Researchers should conduct community engagement when designing health services research.
10.43 Researchers must consider and balance social and cultural risks, as well as medical risks of harm.
10.44 Researchers must consider collective risks when conducting population level health services research.
10.44.a Examples are costs to healthcare systems, reputational damage to groups, lack of intervention sustainability.
10.45 Researchers should consider whether informed consent from individuals is required. This determination will be based on who the target of the intervention or object of study is.
10.45.a For example, whether it is a system-level intervention, a healthcare worker intervention or an individual patient intervention. Other considerations are the level of risk for individuals.
Implementation research (IR) is growing in recognition as an important generator of practical knowledge that can be translated into health policy. IR increases understanding of how to improve access to health products and strategies that are already available and demonstrated to work, but which remain beyond the reach of many people who could benefit from them. It identifies practical problems faced by, for instance, disease control programmes, and seeks methods of improving access to health interventions which in turn lead to better health outcomes. It also addresses different aspects, such as social and contextual factors, implementation processes, and outcomes.
IR provides the link between theory and practice. For example, in case of a new vaccine for prevention of measles, traditional clinical trials will address safety and efficacy while IR seeks to answer questions of accessibility, acceptability, appropriateness, and feasibility in the communities where the vaccine is needed. However, IR is not always related to disease treatment or prevention, but may also focus on routine healthcare delivery, cost-effectiveness, policy, health education, and so on. It therefore draws flexibly on a variety of research approaches to address its diverse research questions.
This distinction between IR – which focuses on the application and practicalities of interventions – and biomedical and clinical research – which establishes evidence for use of these interventions – calls for a different application of Te Ara Tika and bioethical principles. The table below provides information on the ethical impact of these differences.
| Domain | Clinical research | Implementation research |
|---|---|---|
| Research participants | Individuals | Institutions, communities, and individuals |
| Informed consent | Informed consent by competent individuals, assent and or consent by legally authorised representatives | Consent may be difficult to obtain in cluster randomised trial design. There may be a need for a two level consent—consent for randomisation from gatekeepers[12] and consent for participation at the individual level. Sometimes individual consent may not be feasible. However, gatekeeper consent does not replace the need for individual consent. |
| Equipoise | Clinical equipoise | Clinical as well as contextual equipoise (genuine uncertainty that the implementation will work in a new context as well as whether the implementation package will work at all) |
| Pre-requisites |
Understanding of disease pathophysiology Intervention aimed at disease-specific management |
Identification of population health needs Understanding relative priority of need for intervention within local context Community engagement to understand community needs, ensure scalability, and sustainability |
| Research conditions | Generally controlled research environment | Real-life or pragmatic research environment |
| Research designs | Cross-sectional, case-control studies, Cohort studies, randomised clinical trials | Cluster randomised trials Pragmatic, mixed methods, effectiveness implementation hybrid designs, participatory action research, quasiexperimental design, realist review |
| Integration within health system | Often, there is no a priori plan for health system integration. Findings of clinical research go through IR before integration into health system | IR has a strong health system strengthening focus. It creates horizontal integration into the health system. There is an ethical imperative for health system integration |
| Predominant research disciplines | Physiology, genetics, biochemistry, and other basic sciences, epidemiology, clinical medicine | Anthropology, Economics, Epidemiology, Political science, Public health, Sociology |
| Control groups | In most epidemiological designs, control groups are required. But some phase 1 clinical trials and observational studies may not require control groups | Having a no intervention control group may not be acceptable. Alternative designs of quasi-experimental studies do not require a control group |
| Boundary between research and clinical care | This boundary is usually clear, but may be unclear in case of therapeutic misconception especially in cancer trials | Is often unclear, because the intervention is of proven efficacy |
| Types of research question | Efficacy and safety of a therapeutic strategy in the individual |
Operationalization of an intervention in local context Policy analysis Health system functioning at multiple levels |
| Anticipated outcomes | Well-defined hypothesis at the beginning of the clinical research. Expected outcomes clearly stated |
Multifaceted holistic impact on health systems functioning with regard to intervention tested. Sometimes outcomes may be unexpected |
| Risks assumed by: | Mostly, the risks are for the study participants. However, families and communities may also be affected in specific contexts | Usually population level risks. Moreover, the people getting the benefits and people suffering the risks may be different. |
| Benefits accrued by: | Benefits accrue to the participants, the community. The research finding may be a common good | Individuals, communities, health system, institutions may benefit. The research findings may be common good. The people accruing benefits may be different from those who suffer risks |
| Generalisability | Generalisability is sometimes possible in multicentric and well sampled studies, however most studies are specific to the target populations. | Generalisability may be limited by contextual factors. However, findings may be generalisable to similar contexts |
| Social justice implications |
Social justice is usually not a primary consideration. However, justice considerations are required in selection of research participants. Research on vulnerable participants is often contentious because of compromised autonomy and other logistics |
Social justice considerations are primary. Working with vulnerable groups essential to understand implementation issues in these groups so that the intervention can reach them |
The ethical considerations required for IR can be further divided into the three phases of its research life-cycle: the planning phase, the implementation phase, and the post-research phase. While these phases bring their own ethical challenges, a core ethical imperative of IR as a whole remains the strengthening of the health system.
10.46 Strengthening the capacity to translate research findings into health policy is ethically important in implementation research and must be a component of all phases of the IR process.
10.47 Researchers should engage with local health experts, communities, and stakeholders during the planning of IR.
10.47.a Situational analysis is crucial in implementation research, and engagement and consultation will assist in meeting the ethical requirements of IR.
10.48 Implementation research must reflect high local need.
10.48.a Engagement and analysis is essential in the planning stages of IR to determine if a health problem is indeed perceived to be a local priority.
10.49 The clinical or public health problem to be addressed by implementation research must be adequately identified. The epidemiology of the disease or health status must be understood, and the local situation analysed to identify accessibility of care.
10.49.a In the case of new interventions or those adopted from other countries, local situational analysis is important to determine the differences and similarities between communities where the intervention has been successfully implemented and communities in which the intervention will be tested.
10.50 Implementation research must meet the situational or contextual equipoise standard; that is, to ethically justify implementation research there must be reasonable doubt as to whether a new and untested intervention(s) will work in a specific context.
10.50.a This is related to the equipoise standard for clinical research, which requires it be in doubt whether an intervention is superior or inferior to what it is being compared with. See equipoise.
10.51 Researchers must balance the risks and benefits of implementation research to both individuals and communities.
10.52 Data ownership should be fairly negotiated during the planning of implementation research. Transparent stakeholder engagement should establish appropriate ethical oversight of data, and plan for future access to research findings.
10.53 Implementation research should ensure commitment upfront to the sustainability of any interventions found to be effective.
10.53.a If access to a proposed public health intervention cannot be ensured for a community after the IR, it may not be ethical to carry out such a research activity.
10.53.b Provision of interventions without a plan for sustainability could lead to harmful effects to the community, increased inequity, and loss of trust in the health system.
10.54 Researchers should consider the specific ethical issues raised by methods employed in undertaking IR.
10.54.a See table 10.2 for ethical issues relating to variable implementation research designs.[13]
| IR design | Features | Example | Ethical concerns |
|---|---|---|---|
| Cluster randomised trials (group randomised, place-based, community wide intervention trials) | Random allocation of groups or “clusters” to study arms and outcomes are measured in individual subjects and at community level | Randomization of clusters of obstetrics unit staff to education on hand washing or usual practice, measurement of rates of puerperal sepsis in women delivering at study clinics | Different units of intervention and outcomes measurement Consent before and after randomization, whom to consent? Choice of gatekeepers No opt-out option within cluster Risk: benefit balance Ethics of randomization to known intervention, equipoise, Identification of vulnerable groups |
| Effectiveness-implementation hybrid trials | Assess both effectiveness and implementation strategy simultaneously Identify intervention—implementation interactions |
Evaluate impact of ITN on reduction of malaria and assess robustness of availability and uptake of ITNs in the community | The trade-off between the scientific rigor required for effectiveness assessment and the realistic contextual considerations required for implementation is an important ethical consideration |
| Mixed-methods research | Use of both qualitative and quantitative methods Understands various perspectives Rationales: “participant enrichment”, “instrument validity”, implementation validity”, “meaning enhancement” |
Integration of HIV and TB management in single clinics—patient experience (qualitative) and adherence (quantitative) | The trade-off between the scientific rigor required for quantitative methods and the realistic contextual considerations required for the qualitative component |
| Participatory action research | Research question, design, and data collection in a participative manner by the research participants “Bottom-up” approach |
Peer support groups to improve adherence to ARV in HIV + subjects | There is a need for community engagement to ensure responsiveness, sustainability, and scalability |
| Pragmatic trials | Effects of intervention in routine practice Maximize variability of settings, practitioners, patients |
Introduction of community health workers for home management of malaria | There may be concerns of standards of care and ancillary care, which in pragmatic conditions may be ethically debatable. |
| Quasi-experimental study | Real-life conditions With or without control group No randomization |
Open label demonstration project of effectiveness of selfreported use of pre-exposure prophylaxis for HIV | There is a concern regarding scientific rigor of the research |
| Realist view | Analysis of how and why an intervention works in a context combining theory and empirical evidence. | Integration of traditional healers into home management of malaria strategies | Community engagement is of utmost importance to retain cultural and contextual sensitivity |
10.55 Implementation research must balance potential risks and benefits, and conduct diligent situational analysis to determine contextual risks and benefits.
10.55.a For example, in IR the community at large may benefit from a treatment which resulted in a side-effect for an individual who did not require that treatment. Alternatively, an intervention may be implemented in one group while the benefits accrue to a different group.
10.55.b Contextual risks may result from cultural or financial factors which bring risks of stigmatisation and/or discrimination in one setting where not present in another.
As explained in relation to randomised cluster trials, non-consensual research raises complex legal issues in the New Zealand context, where ethics committee are unable to provide prospective waivers of consent.
10.56 Implementation research may involve modifications to the traditional informed consent process depending on the design. Researchers should consult the Code of Health and Disability Service Consumers’ Rights to consider whether the IR project meets informed consent requirements under New Zealand law.
10.56.a In IR the informed consent process can be difficult to operationalise at the individual level, and these activities often involve ‘relational’ rather than ‘individual’ autonomy, where participants are recruited into the research as groups or communities.
10.57 Where implementation research requires access to individual-level data, safeguards must be put in place to protect individuals’ privacy.
10.57.a This is especially important when data is accessed without explicit consent. Risks to the individual’s autonomy can be off-set by increased protection of data confidentiality.
10.57.b The most effective safeguard of privacy is the provision of data to IR projects in a de-identified form.
10.58 When deciding on the standard of care or prevention to be provided to a control group, researchers should consider its adequacy against the justice principle.
10.58.a For example, simply providing an existing standard of care which is obviously insufficient is unfair and ethically unacceptable. Alternatively, a standard of care which is not currently in use but agreed upon by public health experts and the community could be employed. Fairness should determine whether a local de facto or local de jure (respectively) standard acts as a control.
10.59 Implementation research should strengthen the health system in which it was conducted.
10.59.a For example, IR should identify any technologies or expertise required for post-intervention scale-up.
10.60 IR projects should focus on horizontal integration of public health interventions and avoid vertical program structures.[14]
10.60.a Vertical structures can be disempowering to the health system through inefficient resource utilisation.
10.61 Implementation research should lead to strengthened research capacity of the local institution and individuals’ capacity to conduct research in settings where such capacity is weak.
10.61.a This can range from creating a trained workforce of researchers to building capacity and infrastructure to allow independent conduct of IR in the future.
10.62 Given the important public health impact and objectives of implementation research, findings should be widely disseminated, including feedback to communities and stakeholders who participated in the research.
10.63 Communities and individuals who have contributed data should have access to implementation research findings.
10.64 Participants who acted as controls should gain access to interventions which were withheld from them during the study.
10.65 The results of implementation research should be disseminated irrespective of the results.
10.65.a Both negative and positive implementation research findings may be important for planning similar interventions elsewhere, and could enhance resource utilisation globally.
10.66 Knowledge generated in implementation research should be translated into public health action.
10.66.a Potential barriers to knowledge translation include lack of prior consultation with policy-makers, lack of funding, weak health systems, and absence of a culture of evidence-based decision-making.
10.67 Researchers should communicate their findings promptly to policy-makers and health system officials.
10.68 Researchers should propose actionable suggestions based on their research findings to facilitate uptake and scale-up of successful interventions.
10.68.a Barriers identified during IR may require further study to develop strategies to overcome them.
10.69 Knowledge generated in implementation research should be used for public education.
10.70 The interventions demonstrated to be successful in implementation research should be sustained post-research.
10.70.a Responsibility for sustainability lies with all stakeholders.
10.71 The benefits of implementation research should be shared regardless of context.
10.71.a Benefits may be direct as a result of the intervention being studied, or indirect and not related to the intervention. Benefits may also accrue to individual participants – for example therapeutic benefits – or to the community as a whole – such as knowledge about barriers affecting better provision of care. Direct benefits are shared by the sustainable translation of research into action, and indirect benefits by capacity-building and health system strengthening.
10.71.b Optimal benefit sharing is important in IR where the individuals who bear the risk do not receive direct benefit from the intervention.
The definition of translational research is less clear than the definitions of basic and clinical research (Weijer et al. 2012). Translational research fosters the multidirectional integration of basic research, patient-oriented research, and population-based research, with the long-term aim of improving the health of the public. One form of translational research expedites the movement between basic research and patient-oriented research that leads to new or improved scientific understanding or standards of care. The second form of translational research facilitates the movement between patient-oriented research and population-based research that leads to better patient outcomes, the implementation of best practices, and improved health status in communities. The third form of translational research promotes interaction between laboratory-based research and population-based research to stimulate a robust scientific understanding of human health and disease.
An important component of translational research concerns research aimed at enhancing the adoption of best practices in the community. Cost-effectiveness of prevention and treatment strategies is also an important part of translational science.
Most ethical considerations of a translational research programme are common to any research involving human subjects.
10.72 The interests of participants in translational research must always have priority over those of the wider community for whom the research benefit is envisaged. The objective of rapid knowledge transfer from basic research to clinical practice must not compromise participant safety.
10.73 Translational research should focus on diseases with high health impact and significant outcome inequity.
Innovation in practice in health and disability care is a change to practice aimed at providing improvements in the outcomes or experiences of people receiving health or disability services.
Innovative practice sits on a continuum with normal or everyday practice. Health and disability services must always be tailored to the individual needs, circumstances and conditions of each consumer. The care of individual patients may therefore vary around a core of standard accepted practice.
Health professionals must be allowed to make minor deviations from accepted practice to adjust health care to suit the individual needs of each consumer, and sometimes new techniques or procedures may result from unplanned responses to medical complications arising from the treatment of an individual consumer.
However, innovative practice proper arises where there is a planned deviation from the currently accepted practice involving an untested or unproven intervention or change that is intended to be used on an ongoing basis (Ministry of Health 2008). An innovative practice may arise from a series of small incremental changes to accepted practice; practice becomes properly innovative where it reaches a point of significant shift, and the service provider plans to continue it.
The term ‘innovative practice’ extends to the application of known procedures in new or novel circumstances in which those procedures have not previously been tested. It may involve new delivery practices by health practitioners, or by those working in the disability field; new devices; new investigative procedures; new management options; or systems changes.
Where innovative practice does not require formal research, ethical obligations remain. For example, service providers should generally seek fully informed consent to the use of innovative surgical techniques or devices. A recognised danger in this area is that practitioners – and patients – may tend to believe that if something is new it is necessarily better. This belief may not be well founded; service providers need to make this clear to patients. To reduce the risk of false belief, the innovator may not always be the best person to seek consent.
10.74 To protect clinicians and patients, service providers should seek peer appraisal of innovative approaches.
10.75 Service providers should keep good records of the nature of innovations, the outcomes of their use for the patients, and developments and refinements.
10.76 Providers should clearly describe innovations and investigate ways to inform the wider community of health providers about the innovations being practiced. The recognised danger of runaway diffusion[15] (Jake Earl 2019) needs also to be guarded against.
10.77 Only those with appropriate qualifications and expertise must undertake innovative practice, and only for the purpose of improving outcomes or experience for an individual consumer or consumer group.
10.78 Appropriate safeguards should be in place to ensure that independent assessment occurs through an innovative process so that should it become apparent the innovative practice is not achieving positive results or is exposing consumers to unnecessary harm, service providers can resume providing standard practices.
10.79 Service providers must not prematurely adopt innovative practice into standards of care. They should put appropriate evaluative mechanisms in place to assess the effectiveness of any innovative practice; these may include formal research.
10.80 Where innovative practice in health or disability services requires research, the practitioners involved must ensure this happens at the appropriate time and in the appropriate way.
10.80.a Research into innovative practice must meet these Standards, including the requirement for ethical review.
10.81 It can be challenging to identify when a difference from existing practice requires formal research, and what the nature of the research should be. As a guide, service providers should consider formally researching an innovative practice when:
10.82 Service providers may not need to conduct full research on minor innovations.
10.82.a While it is beyond the scope of this discussion to fully specify such guidelines, some reasons that could justify engaging in innovative practice rather than clinical research include:
[1] Where one person in a first cohort of participants receives a single dose of investigational product in advance of the full study cohort
[2] In the context of a cluster randomised trial, the stepped-wedge design involves the collection of observations at a baseline period in which no clusters are exposed to the intervention. Following this, at regular intervals, or steps, a cluster (or group of clusters) is randomised to receive the intervention.
[3] For more information see The Ottawa Statement on the Ethical Design and Conduct of Cluster Randomised Trials(external link).
[4] Code of Health and Disability Consumer Rights right 6(1)(d).
[5] However as the terms “health research” and “disability research” are open to broad interpretation and are not defined in the law, it is unclear whether the requirements of the Code may also apply to enrolment in research that involves the collection and analysis of data even where the treatment is the same for all participants (as opposed to using data retrospectively where it was originally obtained during the usual course of health care).
[6] A quasi-experiment is an empirical interventional study used to estimate the causal impact of an intervention on target population without random assignment
[7] Adapted from the Australian National Statement on Ethical Conduct in Human Research (2007)
[9] This section has been adapted from Ethics of health policy and systems research: a scoping review of the literature 2017.
[10] Incentives may be financial or reputational. See How Financial and Reputational Incentives Can Be Used to Improve Medical Care(external link).
[11] This section has been developed from Gopichandran et al., Developing the ethics of implementation research in health(external link). Implementation Science (accessed 07/11/19).
[12] For instance, a GP may be a gatekeeper in the health context.
[13] Gopichandran et al., 5.
[14] Here, the terms ‘horizontal’ and ‘vertical’ refer to two levels or modes of integration, which are general and specialised respectively. An intervention integrated in the healthcare system vertically may, for example, target only a select group of patients. However, a public health intervention integrated horizontally will be applied broadly and be accessible to as much of the population as possible.
[15] Runaway diffusion refers to innovations which, due to non-therapeutic factors such as enthusiasm and profit, are adopted without adequate ethical oversight or scientific due process.
