Article Text

Impact of the demand for ‘proxy assent’ on recruitment to a randomised controlled trial of vaccination testing in care homes
  1. Paul James Whelan1,
  2. Rebecca Walwyn2,
  3. Fiona Gaughran3,
  4. Alastair Macdonald4
  1. 1Older People and Healthy Ageing, Central and North West London NHS FT, London, UK
  2. 2Clinical Trials Unit, Institute of Psychiatry, London, UK
  3. 3National Psychosis Service, South London and Maudsley NHS Foundation Trust, London, UK
  4. 4Institute of Psychiatry, King's College London, London, UK
  1. Correspondence to Professor Alastair Macdonald, Institute of Psychiatry, King's College London, London, UK; alastair.macdonald{at}


Legal frameworks are in place to protect those who lack the capacity to consent to research, such as the Mental Capacity Act in the UK. Assent is sought instead from a proxy, usually a relative. However, the same legislation may, perversely, affect the welfare of those who lack capacity and of others by hindering the process of recruitment into otherwise potentially beneficial research. In addition, the onus of responsibility is moved from those who know most about the study (ie, the scientific community) to those who know less (the proxies). In this paper, we describe the characteristics of a sample at different stages of the recruitment process of an influenza vaccine-based randomised control trial in elderly care home residents (the FEVER study). 62% (602/968) of potential subjects lacked capacity but only 29% (80/277) of those actually randomised. Older age, being female and living in an Elderly Mentally Ill care home were the only variables associated with lacking capacity. Considering this was a study based in a care home setting where the prevalence of dementia approximates 80%, the trial, like many others, was thus significantly biased. We believe that difficulties seeking proxy assent contributed significantly to this problem. Further thought should be given to how assent to enter research for those who lack capacity should be provided, and we suggest avenues for further discussion such as independent risk/benefit expert panels.

  • Capacity
  • Autonomy
  • Informed Consent
  • Scientific Research
  • Mentally Ill and Disabled Persons

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Research on conditions that directly affect capacity to consent, as well as those on conditions often correlated with lack of capacity (eg, pressure sores), requires some method of entering the subjects to whom the research question is most relevant. In European and North American jurisdictions Research Ethics Committees or Institutional Review Boards generally demand that, in research projects on subjects who lack capacity to consent, assent be obtained from a proxy—usually a relative—before proceeding. In England and Wales, the relevant legislation is Section 32 of the Mental Capacity Act (MCA)1 and the Medicines for Human Use Act.2 These legal provisions establish a process of assent by a properly identified proxy or ‘legal representative’ in the case of trials or a ‘consultee’ (personal or nominated) in other research. This approach does not seem to sufficiently rectify the balance towards including people with dementia or delirium; in a trial of pressure-relieving mattresses, almost 40% of potential subjects lacked capacity, whereas only 4% of those who entered the trial were in this group.3

Of the distinct interest groups involved in clinical research (researchers, funders, patient groups, carers and specialists in ethics), specialists in ethics appear more likely to favour autonomy over other ethical principles such as avoidance of harm.4 It is therefore in keeping for them to demand that another person assumes this responsibility for an incapacitated subject (rather than, say, the researchers). However, there is concern that the hegemony of this principle encourages an inappropriate displacement of responsibility for the assessment of the risk–benefits of a study from those most knowledgeable on the subject itself (ie, the scientific community) to those who are less knowledgeable (ie, the subject or proxy),5 even though the subject or proxy may have greater knowledge of individual concerns. This transfer of the burden of responsibility to subjects with capacity or their proxies is accompanied by forms and information sheets to be signed and countersigned, and, in the case of industry funding, long and complex disclaimers, all contained within a burgeoning bureaucracy—albeit intended to protect the participant—and jargon-laden terminology. In addition to potentially disadvantaging certain populations, the legal status of much of this activity is unclear, as are its practical results.

To explore these issues further, and in particular the potential lack of applicability of research findings from trials affected by selective exclusion of those lacking capacity, we conducted additional analyses of a clinical trial in elderly care home residents. The FEVER trial was a randomised controlled trial (RCT) of the effectiveness of testing elderly care homes residents’ antibody response to the routine influenza vaccine and administering a booster dose where needed versus standard practice.6 We hypothesised that research in care homes—in which the presumption of capacity is tenuous because of the high prevalence of dementia7—would involve the exclusion of many subjects either because relatives (the highest in the hierarchy of acceptable proxies under UK guidance) could not be identified or because they or other possible proxies (eg, care home staff) would be unwilling to take on this responsibility. We also hypothesised that subjects entering the study would be unrepresentative of all who could have entered randomisation and that the exclusion of those whose relatives could not be contacted would significantly contribute to this recruitment bias.

A related question concerns the assessment of capacity. The need for proxy assent in a study is related to the numbers lacking capacity, and these are in turn related to the rigour of the capacity assessment and the expertise of the person conducting it. We have previously shown that an assessment of capacity strictly consonant with that of the UK Law Commission (and latterly the MCA) might subvert a research study's purpose to the extent of making the whole study unethical.8 In that study, investigating different methods of obtaining informed consent for entry into a study of the natural history of delirium, a formal assessment of capacity followed by obtaining assent from a proxy (vs an informal capacity/consent process) led to the recruitment of significantly fewer subjects (44% vs 74%) and a biased sample (with less cognitive impairment, lower severity of delirium, etc). We therefore set out in the present study to compare the characteristics of all subjects in care homes who were included in the RCT with all those who were originally approached and assessed for their capacity to consent to the trial. We also aimed to test whether the group excluded because they lacked capacity and a relative could not be contacted was responsible for any differences between those originally approached and assessed for capacity and those entering the study.


The original study was an RCT of the efficacy of assessing antibody response to influenza vaccine, and administering a booster vaccination if the antibody response was inadequate, in reducing hospitalisation and death rates compared with current standard practice.6 In brief, all care homes with nursing care in the boroughs of Lambeth, Lewisham or Southwark in London, UK, were approached just before and during the influenza season. All permanent residents aged 60 or above, including those in non-nursing care places within the same facilities, who would be routinely offered the influenza vaccine in 2004, were eligible for inclusion. Exclusion criteria were known contraindications to the influenza vaccine or current evidence of delirium. Residents in places designated ‘Elderly Mentally Infirm’ (EMI)—usually those with dementia and behavioural problems—were included.

Capacity assessment

Written informed consent was sought for all potentially eligible residents with capacity. Written informed assent was requested for all potentially eligible residents without capacity from the resident's nearest relative in the first instance or the assigned member of the care home staff in the second instance. In the study, which predated the MCA and current guidance, we deliberately used an informal method of capacity assessment to minimise the numbers deemed to lack capacity, and also those who had capacity but who were so irked by the assessment that they refused.

The trial power calculation required an overall sample size of 824 care home residents. Data available on all residents assessed for capacity were age, gender, ethnicity, whether they occupied a care home place with or without nursing care, or whether they were on an EMI unit specialising in mental illness (mainly dementia) with behavioural problems. No other comparisons were possible, as this would have involved using data without consent. Ethical approval for the study, and the analysis reported here, was obtained from the Local Research Ethics Committee of the South London and Maudsley Trust and the Institute of Psychiatry. Data were analysed using Statistical Package for Social Sciences V.9. Where necessary independent sample t tests were corrected for unequal variances according to Levene's Test. The purpose of the logistic regression analysis was to examine whether membership of any particular group of those not entering the study was associated with significant differences between the population eligible and those who ultimately entered trial randomisation.


Figure 1 shows the derivation of the sample. There were 1295 residents living in care homes that were participating in the study at its inception, of whom the research team was able to assess the capacity of 968—the remainder could not be assessed before the seasonal vaccination was administered for practical reasons (ie, time pressure)—thus, no data are available for these residents. Of those assessed, 602 (62.2%) lacked the capacity to consent to the study. Lack of capacity was significantly associated with older age (mean 83.6 vs 80.8, t=4.72, d.f.=963, p<0.001), female gender (66.2% vs 50.2%, 2 χ=12.6, d.f.=2, p=0.001) and with residence in an EMI place (83.1% vs 54.3% 2 χ=68.1, d.f.=2 p<0.001). There was no association with ethnic group (white vs non-white) with whether in a care home with nursing versus without nursing care staff or with duration of residence.

Figure 1

Derivation of the sample. Letters in red are referred to in the on-line appendix.

Of the 968 residents whose capacity was assessed, 277 (28.6%) were randomised to the two treatment arms of the trial. In those in whom definite consent or assent was not established (D1 and D2 in figure 1) this was for reasons that applied sometimes before and sometimes after the test of capacity was carried out; for clarity, they are all shown as after. Eleven residents with capacity and seven without were not available when the time for their prevaccination blood test was due.

Residents with capacity

Of the 255 eligible residents with capacity, 19 were excluded because they asked that their closest relative also agreed, but in 10 of these cases this person could not be found, and in the remainder the relative disagreed. Of the 255 with capacity and available, 157 (61%) consented with or without the agreement of a relative and 79 (31%) declined.

Residents without capacity

Of the 557 eligible residents who lacked capacity and were not excluded for other reasons, 146 (26%) were considered by the research nurses to be likely to physically resist the research procedures and so participation was not feasible even if relatives assented; in 27 (5%), relatives declined to assent and in 82 (14%) relatives assented. A relative could not be contacted in 304 (55%) residents without capacity. Of these, 40 (13%) were entered into randomisation via staff assent. The remaining 264 patients—those whose relative could not be found and who were not assented by staff—constituted by far the largest single group of those not entering the study (only 13 fewer than all those entering the randomisation).

Table 1 shows the differences in overall resident characteristics at each stage of recruitment. Logistic regression analysis showed that bias in recruitment by care home type was associated with exclusion from the study for three reasons, significantly more than for any other reason: lacking capacity, no relative found and deemed likely to dissent.

Table 1

Characteristics of residents at different stages of recruitment

Characteristics of residents at different stages of recruitment and the logistic analysis are further described in online supplementary appendix 1.


There was significant recruitment bias in this study of influenza vaccination in care homes. Residents from non-nursing care homes and EMI care homes specialising in dementia were disproportionately excluded. The latter are important since there may be an immunological factor in some dementias, especially Alzheimer's disease,9 and the response of people with dementia to vaccination may be different.10 ,11 In this study, the biggest single reason for exclusion was because relatives of patients without capacity could not be contacted within the recruitment timeframe of the study. This concurs with findings of Mason et al3 although they focussed on the numerically much smaller issue of relative assent. In our study, although ethical approval was given for care home staff to assent on residents’ behalf, this took place in only 13% of these residents. This may have reflected a careful decision about the residents’ likely wishes if they had had capacity, but could equally have been because of reluctance to take responsibility for any adverse consequences, the risk of the displeasure of relatives who subsequently might become available or a simple practical desire to minimise extra work. Further research on these possibilities would be fruitful.

The next most important reason for exclusion—those who were thought likely to actively dissent or resist the procedures—was one which may not be avoidable. It is difficult to see how these residents could be included.

Bias against recruitment of residents in EMI compared with non-EMI homes was probably associated with higher prevalence of cognitive impairment and thus lack of capacity to consent. The reasons for the greater exclusion of residents in non-nursing versus nursing care are not obvious. It may be that people needing a care home but not nursing care are for some reason less likely to be in contact with relatives who could assent, but this is speculative.

We think that the present assent system leads to biased (and therefore unethical) studies of incapacitated adults in settings where relatives may not be available like hospitals and care homes. In our remaining discussion we assume that any bias due to active dissent or resistance cannot be avoided; it is our contention that bias caused by exclusion of patients who would readily participate but lack capacity is the central issue. In this study, we used an informal method of assessing capacity to consent to research. Had we used a more formal method, recruitment might well have been more biased still away from impaired patients.

One suggestion is to restrict research on incapacitated people altogether. Corrigan and Williams-Jones5 suggest that assent is ethically unsound, providing ‘only the illusion of protection’. They assert that clinical trials ‘can never be primarily in a patient's best interests’. They suggest that the responsibility of weighing up the risks and benefits to individual participants in a research study on incapacitated adults should not be delegated to proxies, however well informed, but rather accepted by Research Ethics Committees and the MCA system, including the suggestion that every time an incapacitated patient is admitted into a study the Research Ethics Committee is informed. The main thrust of their argument is that ‘Protection of participants must be the paramount concern, even if it restricts research’. We think that this approach would discriminate against those suffering from conditions that directly affect capacity. However, we sympathise with the view that the very necessary risk–benefit analysis for individual patients without capacity should not be expected of assentees.

The present system of using staff to assent as nominated consultees has advantages. If there is knowledge of how individual patients might have reacted to a research proposal if they had had capacity, this should be used in assent decisions. This would, at least in long-term settings, clearly be best discharged by people who know the patient well—the staff. It is also likely that some of the problems of trials in these patients—such as the assessment of side-effects—would be best moderated by the ability of staff to withdraw assent if necessary. However, the increasing heterogeneity of ethnicity and culture among staff and residents in care home settings (ref Badger)12 may interfere with this process, especially if there are significant language barriers, as will low levels of understanding and recognition of dementia (Macdonald and Carpenter,13 Macdonald and Woods).14 Qualified care staff may not know the patients as well as unqualified care workers, but the latter are among the lowest paid15 and poorly educated and trained16 employees in the UK.

Decisions about participation in research must be based on written information sheets, scrutinised by the Research Ethics Committees. The approved sheet for relatives and staff possibly assenting to the project described here contained, under the heading ‘What are the possible disadvantages and risks of taking part?’ just two statements, one relating to mild local pain at the blood test site and one about an unspecified ‘reaction’ to the booster vaccine, adding ‘the risk of this is similar to when they have their usual flu vaccine’. There was no requirement to quantify or elaborate the balance of these risks to the individual. The provision of lists of possible adverse consequences, unordered and unquantified has been criticised by Corrigan and Williams-Jones.5 Information sheets are anyway provided by the researchers carrying out the study, whose interests may not be served by lucidity, and Research Ethics Committees in the UK, at present proscribed from much scientific assessment of projects, may be inadequately qualified to comment on anything other than the language or means of communication.

How then could the scientific and research community take more of the risk–benefit analysis burden from the shoulders of the assentee?

One possible solution in the UK is to expand the role of Independent MCA advocates. These would have better support, training and understanding of research than staff in care homes, but are less likely than staff to be able to make a judgment about whether or not a person would have permitted research had they had capacity, and would not be easily in a position to review or withdraw assent. Also, the onus of assessing risk–benefits in general would still be removed from those best able to carry it.

Another possibility, not necessarily exclusive, might be that the risks and benefits of participation (to the subject and the community) is assessed by a scientific expert panel independent of the funders (and their peer reviewers), the Research Ethics Committees and the researchers. Their recommendations would make clear, in plain English, the risks and benefits to both the participants and the common weal, with due attention to scope and defining the groups likely to be affected. These recommendations would be made available to relatives or staff who may give assent. In the study reported here, the risks were minimal, and there was a strong possibility of benefit to all care home residents and even individual participants; we believe that such a judgement would have influenced the staff asked to assent on behalf of residents in their care, and bias would have been reduced.

Of course, were such a system implemented, there is no logical reason why it should be confined to assent. At present, ‘informed consent’ by people with capacity places all the responsibility for risk–benefit analysis on themselves, provided it is executed according to the law. In the tragic TGN1412 trial it probably was,17 although the readability of the information sheet given to participants has been questioned.18 No one in the study team appeared to know of an unpublished previous incident in another centre,19 but it is just possible that an expert review panel charged with weighing up the risks and benefits might have heard about it informally, and, even if the trial had gone ahead, this would have been communicated to the participants. Where there is capacity we accept the hegemony of autonomy but see no reason why this should not be strengthened by considered expert opinion as well as, at present, ‘information’.

This approach has problems to be overcome and issues to be resolved. It will add to the expense and bureaucracy of research. When the risk–benefit issues are contentious, will the independent panel be culpable if harm comes to the subject? If a study is deemed to pose a very low risk to the patient but has some potential benefit for them (for instance, a sympathetic and skilled interview survey with amenable care home residents which would punctuate an otherwise eventless day), could the panel assent directly and dispense with the consultee? And could this also pertain in low-risk research of great potential benefit but to the community at large rather than the subject? In any higher-risk instance the panel is likely to be more cautious than either the researchers or the peer-review opinion given to the funders; some otherwise worthwhile but risky research may not take place. Otherwise, if more scientific and thus more ethical research on incapacitated subjects can take place, perhaps these are prices worth paying.


In this study, potential subjects who lacked capacity, primarily because of cognitive impairment, were far less likely to enter a trial that had, for them, significant potential health benefits and relatively low risk. The main and avoidable barrier was that relatives could not be found to assent on their behalf. People with cognitive impairment are already disadvantaged enough without being also excluded from research and also the benefits of research on their conditions. The ‘opt-in’ assent system presently accepted by most ethics of research bodies leads, as in this study, to the exclusion of a group of subjects sufficiently large to vitiate the research results. At worst, this makes such research unethical for all participants, but at best it introduces enough avoidable bias to significantly impair its validity. We suggest solutions to mitigate this blight on research with incapacitated adults and the conditions from which they suffer.


We would like to thank the residents who participated in the study and the other members of the FEVER Trial team: Elizabeth Atere-Roberts, Olubukola Awodiya, Orla Barnewell, Lynne Batty, Nas Belazka, Joanne Brooke, Kate Chatterton, Elizabeth Francis, Colin Gelder, Veronica Ibizugbe, Rob Lambkin-Williams, Niamh Keane, Pat Meeking, John Oxford and Jonathan Riley. We are also indebted to the members of the Trial Steering Committee, chaired by Professor Geoffrey Schild, and including Joanna Murray, Jo Leonardi-Bee, Mehool Patel, Noreen Jakeman and Gill Dale. We thank Benjamin Spencer for his help with some of the legal aspects of this paper.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Contributors PW assisted in recruitment, data collection, trial management, analysis and interpretation of data, and wrote various drafts of the paper. FG developed the concept and design, assisted in recruitment and data collection, assisted in analysis and interpretation of data, and assisted in preparing the final manuscript. RW helped develop the concept and design, assisted in acquisition of subjects and trial management, performed the analysis, and assisted in interpretation of data and preparing the final manuscript. AM developed the concept and design, trained researchers, assisted with the analysis and interpretation of data, and wrote the resubmitted manuscript.

  • Funding Charitable Trustees of Guys and St. Thoma’s Hospitals, London, UK.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Ethical approval Local Research Ethics Committee (LREC) of the South London and Maudsley NHS Trust and the Institute of Psychiatry (Reference number 04/045).

  • Data sharing statement We have not shared these data with other organisations other than those involved in the creation of the series of papers generated from the FEVER study. The final paper in the series has just been submitted and we are planning no further work from this dataset. However, we have no objection to the data being used by others as long as it does not conflict with the interests of the participants from the trial.

Other content recommended for you