Experimenting in an emergency: are clinical trials always justified?
Posted on 24th June 2013 by Jamie Loan
RCTs (randomized controlled trials or randomized clinical trials) are widely accepted as the gold standard for assessing the efficacy of a therapeutic. This is because they allow for comparison of treatments in a manner designed to minimise bias (the risk that any observed effect or lack of effect occurs as a result of a factor other than the difference in treatment)(Hill. 1952). However, they present many ethical challenges. Importantly, one needs to ask if it is ethical to treat one person with a more modern treatment – that the investigator believes to be more effective – and the person in the bed next to them with a sugar pill, or older treatment, just to prove a point?
In general, it has been accepted that, providing the RCT is well-designed, necessary and has sufficient safeguards to minimise the harmful effects of being in either the treatment or control arm, it’s conductance is ethically justified (Edwards, et al 1998a; Edwards et al 1998b). However, there is an important caveat: patients should be informed of the potential risks and benefits of entering into the trial and it should be clear that they understand the process of randomization prior to giving consent: experimenting on a patient who has not given consent may seem unthinkable (Edwards, et al 1998a).
However, there are circumstances in healthcare where it is not possible to get informed consent. These occur where, usually by nature of the illness that we intend to treat, the patient is unable to comprehend information, retain it, arrive at a decision or articulate the decision to provide or decline informed consent. These patients are said to lack “capacity”. What to do then? Clearly patients with these conditions are no less deserving of the benefits of the assessment of which treatment is most effective, yet also their right to personal autonomy should be preserved as much as possible.
Here I intend to work through this dilemma using the specific example of a trial conducted in the field of traumatic brain injury (TBI) to see how investigators deal with these issues and how this may generate controversy that is still unresolved.
The use of intracranial pressure (ICP) monitors to guide treatment in severe TBI is used by all neuro-intensive care units (NICU) in the UK (Panchatsharam et al. 2013). However, like many interventions for the management of TBI, there is a lack of consensus regarding the strength of evidence of therapeutic efficacy (Roberts, et al. 1998; Chesnut, et al. 2013; Hutchinson and Kirkpatrick. 2009). Because of this, Chesnut and colleagues (Chesnut, et al. 2013), decided to conduct an RCT which compared a treatment protocols for severe TBI (Glasgow Coma Scale Score <9; Teasdale and Jennet. 1974) based on either invasive ICP monitoring or clinical estimation of ICP and computed tomography (CT. The primary outcome of the study was a composite measure composed of 21 different scores (more on this later!) and the authors found that this measure did not significantly differ between the ICP and non-ICP monitor groups.
This trial has been widely criticised for a number of reasons that I will briefly detail before getting on to the meat of this blog entry: is it ethically justified to conduct trials such as this in emergency settings? So a brief crit: Firstly, the composite measure used by Chesnut is, to my mind, unhelpful – of the 21 different measures included, 15 measured psychological functions such as visuospatial memory, ability to name animals and verbal learning. Whilst measurement of these functions is laudable, in the context of severe TBI where 1/3rd – 40% in Chesnut et al – of patients die before discharge (Martins et al. 2009), subtle differences in complex cognitive functions are perhaps less relevant and more difficult to measure than alternative scales such as the Glasgow Outcome Scale (GOS) score (Jennett and Bond. 1975), or even Mortality. These are easily measured and the GOS is well validated and and long recognised as a useful measure of outcome following severe brain injury (Teasdale et al. 1998; Wilson et al. 2009). However, the trial had only 40% power to detect a statistically significant difference in the extended GOS (GOS-E) score alone and therefore not finding this is not surprising. Indeed, there was a non-significant trend to reduced mortality (especially at 14 days) and increased “Favourable outcome” as measured by GOS-E in the ICP-monitoring group. This study therefore used an odd, perhaps irrelevant, outcome measure and was underpowered to detect differences in an outcome that really matters.
The trial was conducted in Bolivia and Ecuador. The authors state that these sites were selected because they did not routinely use ICP monitoring. The reasons for this are not stated but this may be due to a lack of resources (Rubiano and Puyana. 2013; Ropper, 2012; Finkielman, 2012; Camputaro. 2012). As a result, it is possible that ICP monitoring, which is highly technical and perhaps new to some of the investigators, was ineffectively utilised. It is also possible that this represents a more widespread lack of infrastructure within these centres and this is reflected in the finding most patients were not admitted less than 1h post-injury (Rubiano and Puyana. 2013). It is oft acknowledged (although unproven) that this “Golden Hour” is the time during which outcome can be most effectively altered (Newgard, et al. 2010). Therefore, introduction of ICP monitoring at later time points may be less able to strongly influence outcome.
Ok, so this pretty much sums up my thoughts (largely stolen from commentators that I hope I have adequately referenced here!) regarding the primary outcome of this study. Because of these flaws, the efficacy (or non-efficacy) of ICP monitoring in TBI remains unproven and guidelines for its’ optimal use remain non-evidence-based! Why have so few trials been conducted in this field then?
Before an ethics committee will approve a trial, it must be convinced that the trial is necessary and that it will only be conducted for as long as is necessary for any observable difference between the two treatment arms to become manifest. As mentioned previously, ICP monitoring is a widespread practice in NICUs (Panchatsharam et al. 2013). That is, many clinicians are convinced that, based on personal experience, the benefits of ICP monitoring outweigh its potential risks (intracranial infection, bleeding or trauma to eloquent brain). Anecdotal evidence should not be discounted out of hand: if the treatment effect is obvious then a trial that denies a group of patients treatment that is clearly life-saving should not be conducted. The case of ICP monitoring does not appear to be so clear cut, however as many conflicting reports exist (Haddad and Arabi. 2012). This may be part of the problem: lots of clinicians feel strongly that there is little uncertainty about the value of ICP monitoring, despite the lack of convincing evidence one way or the other (Hutchinson et al. 2013; Haddad and Arabi. 2012; Shafi et al. 2008) and may therefore believe that a clinical trial is not ethically justified as the “uncertainly principle” – that a patient should only be enrolled in a trial if the clinican feels personally uncertain about the best treatment course – is not met (Weijer et al. 2000). In emergency situations, however, it is common for clinicians to be uncertain regarding the optimal treatment (Helmy et al. 2009), but this might reflect a lack of data for a particular patient, rather than uncertainty in published research. Conversely, clinicians trained to handle these situations may be required act decisively in the absence of a strong evidence base (or where the clinician does not have the time to look up the guideline/pubmed; Helmy et al, 2009). In this instance, the clinician may not feel that the uncertainty principle is met, even though the optimum treatment is unclear (Helmy et al, 2009).
Perhaps it is more appropriate therefore to use “clinical equipoise” as a standard for the enrollment of patients in clinical trials? This requires assimilation of the evidence prior enrollment of any patients to a trial (usually by systematic review and meta-analysis) to ascertain exactly where “controversy within the scientific community about whether the new intervention is better than standard therapy, including placebo” exists – either because evidence is lacking, or because researchers disagree about the standard of evidence (Djulbegovic et al. 2011). This therefore prevents circumstances unique to the urgent nature of emergency medicine or individuals views from preventing patient enrollment or introducing biases due to inappropriate patient selection. The uncertainty principle vs clinical equipoise dilemma is an important one encountered in RCTs (especially in the UK) and may explain why Chesnut et al used a developing healthcare service in their study (Weijer et al. 2000): By lieu of their expertise in TBI management without ICP monitoring, clinicians in Equador and Bolivia might feel uncertain that it provides optimum management whereas those in the USA believe that ICP monitoring is so integral to their practice that not using it in one arm of a trial would be ethically unacceptable (Finkielman. 2012).
ICP monitoring is only indicated when one suspects the patient to have, or be at risk of developing, raised ICP. In the context of TBI, most of these patients will either have a reduced GCS score (<9 for inclusion in Chesnut et al. 2013) because of their injury or because of sedation. As a result of this, patients may lack capacity to give informed consent for their entry into a clinical trial. In this circumstance, it is common practice to seek proxy consent from a relative of the patient (Kompanje et al. 2005). However, as 72% of patients admitted with major trauma are unaccompanied by relatives and contacting them takes, on average 60-80 minutes, it is rarely possible to allow immediate enrollment into a trial and commence therapy during the “Golden hour” if consent by proxy is required (Wright, et al. 2001). To expedite enrollment, differed consent may instead be used, whereby the patient is enrolled immediately and informed consent, or informed consent by proxy, obtained as soon as possible after this. If consent is not given the patient is then withdrawn from the trial (Kompanje et al. 2005). In some cases, consent may even be waived (Kompanje) and this is allowed by the WHO declaration of Helsinki, so long as the intention to use differed or waived consent is included in the protocol submitted for ethical approval and its’ use is essential. Chesnut et al state that “Informed consent was obtained for all participants”: because patients had to be GCS <9 this must have been either differed or by proxy and if it were the latter, this could explain delayed ICP bolt placement (even beyond the delay resulting from long transport to hospital times – median 3.5 hours; Chesnut, et al. 2013).
So what is the point of all this? Largely, I aimed to describe the flaws in Chesnut and colleagues’ landmark study that cause it to lack external validity: the lack of power to detect differences in meaningful outcomes and the use of healthcare centres in developing countries. However, these problems were surely known to Chesnut et al, and were likely even deliberate attempts to get around the intense and valid ethical issues that arise when questioning widely accepted, yet non-evidence based, practices. The ethics of conducting RCTs in emergency settings are vitally important when designing trials and it is important that ethicists and clinical researchers work together to develop protocols that allow trials to be conducted in a manner that protects participants, yet does not stifle research or deny future patients the benefits being treated in an evidence-rich environment.
Camputaro L. Effectiveness in the use of new technology. N Engl J Med Comments; Dec 19 2012. Available from URL: http://www.nejm.org/doi/full/10.1056/NEJMoa1207363#t=comments
Chesnut RM, Temkin N, Carney N, Dikmen S, Rondina C, Videtta W, Petroni G, Lujan S, Pridgeon J, Barber J, Machamer J, Chaddock K, Celix JM, Cherner M, Hendrix T. A trial of intracranial-pressure monitoring in traumatic brain injury. N Eng J Med 2013; 368(18):1751-2
Djulbegovic, B. Uncertainty and equipoise: At interplay between Epistemology, Decision-Making and Ethics. Am J Med Sci, 2011. 342(2): p. 282-289.
Edwards SJL, Lilford RJ, Hewison J. The ethics of randomised controlled tirals from the perspectives of patients, the public, and healthcare professionals. BMJ 1998a; 317(7167):1209-12
Edwards SJL, Lilford RJ, Braunholtz DA, Jackson JC, Hewison J, Thornton J, Ethical issues in the design and conduct of randomised controlled trials. Health Technol Assessment 1998b; 2(15)
Finkielman J. Ethical issues. N Engl J Med Comments; Dec 16 2012. Available from URL: http://www.nejm.org/doi/full/10.1056/NEJMoa1207363#t=comments
Haddad SH, Arabi YM. Critical care management of severe traumatic brain injury in adults. Scandinavian Journal of Trauma, Resucitation and Emergency Medicine 2012; 20:12
Helmy A, Timofeev I, Santarius T, Hutchinson P. What constitutes clinical equipoise. Br J Neurosurg, 2009. 23(5): p. 564-565.
Hill AB. The clinical trial. N Engl J Med 1952; 247(4):113-9
Hutchinson P, Kirkpatrick P. Randomised Evaluation of Surgery with Craniectomy for Uncontrollable Elevation of Intra-Cranial Pressure (RESCUE ICP): Study Protocol version 4 (2009).
Hutchinson PJ, Kolias AG, Czosnyka M, Kirkpatrick PJ, Pickard JD, Menon DK. Intracranial pressure monitoring in severe traumatic brain injury. BMJ 2013; 346:f1000
Jennett B, Bond M. Assessment of outcome after severe brain damage. Lancet 1975; 1(7905):480-4
Kompanje EJO, Maas AIR, Hilhorst MT, Slieker FJA, Teasdale GM. Ethical considerations on consent procedures for emergency research in severe and moderate traumatic brain injury. Acta Neurochir (Wein) 2005; 147:633-40
Martins ET, Linhares MN, Sousa DS, Schroeder HK, Meinerz J, Rigo LA, Bertotti MM, Gullo J, Hohl A, Dal-Pizzol F, Walz R. Mortality in severe traumatic brain injury. A multivariated analysis of 748 Brazilian patients from Florianopolis City. J Trauma 2009; 67(1):85-90
Newgard CD, Schmicker RH, Hedges JR, Trickett JP, Davis DP, Bulger EM, Aufderheide TP, Minei JP, Hata JS, Gubler KD, Brown TB, Yelle JD, Bardarson B, Nichol G; Resuscitation Outcomes Consortium Investigators. Emergency medical services intervals and survival in trauma: assessment of the “golden hour” in a North Americal prospective cohort. Ann Emerg Med 2010; 55(3):235-46
Panchatsharam S, Lewinsohn B, De La Cerda G, Wijayatilake D. Monitoring of severe traumatic brain injury patient in UK ICUs: a national survey. Crit Care 2013; 17(Suppl 2):335.
Ropper AH. Brain in a Box. N Engl J Med 2012; 367:2539-41
Rubiano AM, Puyana JC. Intracranial-Pressure Monitoring in Traumatic Brain Injury. N Engl J Med 2013; 268:1748-52
Shafi S, Diaz-Arrastia R, Madden C, Gentilello L. Intracranial pressure monitoring in brain-injured patients is associated with worsening of survival. J Trauma 2008, 64(2):335-340.
Teasdale GM, Jennett B. Assessment of coma and impaired consciousness. A practical scale. Lancet 1974; 2(7872):81-4
Teasdale GM, Analyzing outcome of treatment of severe head injury: A review and update on advancing the use of the Glasgow Outcome Scale. J Neurotraum 1998; 15(8):587-597
Weijer C, Shapiro SH, Cranley K. Clinical equipoise and not the uncertainty principle is the moral underpining of the randomized trial. For and against. BMJ. 2000;321:756–758
Wilson L, Pettigrew LEL, Teasdale GM. Structured interviews for the Glasgow Outcome Scale and Extended Glasgow Outcome Scale: Guidelines for their use. J Neurotraum 1998; 15(8):573-85
Wright DW, Lancaster RT, Lowery DW. Necessary time to achieve next of kin proxy consent for acutely injured altered status patients. Academic Emerg Med 2001; 8:419-20