Once, medical care was relatively inexpensive because interventions were rare and personnel costs were low.
The Senate and House of Representatives of the United States of America designated the decade beginning 1 January 1990 as the ‘Decade of the Brain'. The goal of this proclamation was to stimulate multidisciplinary efforts of scientists from diverse areas to move toward the common aim of better understanding ‘…the structure of the brain and how it affects our development, health and behaviour' . During that decade, research in neurosciences received a lot of media coverage and funding. An immediate consequence of this was a significant increase in the number of new tools available for the monitoring of the central nervous system (CNS), both in the operating theatre and in the ICU . However, despite the many tools available for monitoring CNS, there are no clinical trials which prove that continuous monitoring of any single variable in the ICU has had a significant impact on the outcome of patients. Those with a sceptical view maintain that although neuromonitoring has significantly progressed in the last decade, patients with a traumatic brain injury (TBI), the paradigm of neurocritical care, have not benefited from this sophisticated technology.
Evidence-based medicine (EBM) has not assisted clinicians in solving these issues. Most of the time EBM has caused confusion and added fuel to the fire in an endless circular debate. An example of a clear contradiction is the case of intracranial pressure (ICP) monitoring in patients with severe TBI. While the evidence-based guidelines for the management of severe TBI in adults published by the Brain Trauma Foundation states that ICP monitoring should be performed on all patients with a severe TBI and abnormal CT scan , a recent systematic review conducted by the Cochrane collaboration concluded that there is no evidence to prove that ICP monitoring improves the outcome of comatose patients . Consequently, the debate among clinicians of whether or not neuromonitoring tools make a real difference in neurological outcome is still ongoing. In some ways, the issue has moved from the arena of scientific debate to something resembling a religious dispute in which there is a confrontation between the faith in the usefulness of these tools and the scepticism of non-believers . Even in the absence of robust evidence proving the efficacy of many neuromonitoring tools, we believe it is time to re-examine the basic objectives of neuromonitoring and leave the muddy waters of scientific fundamentalism.
Why do we monitor in the NeuroICU?
To avoid losing the focus of this debate, it is relevant to remember the main reasons as to why neurocritical patients are monitored. The justifications could be summarized as follows: (1) to detect early neurological worsening before irreversible brain damage occurs; (2) to individualize patient care decisions; (3) to guide patient management; (4) to monitor therapeutic response of some interventions and to avoid any adverse effects; (5) to allow clinicians to be able to understand the pathophysiology of complex disorders; (6) to design and implement management protocols; and (7) to improve neurological outcome and quality of life in survivors of severe brain injuries.
Sceptics maintain that to justify the use of monitoring tools, we should have evidence that monitoring devices are able to alter therapy, and that these modifications in management improve not only survival but also quality of life. This is obviously the aim of doctors who utilize neuromonitoring. To reach these goals, there is the need to overcome such unavoidable obstacles as the learning curve of any monitor and establishing consensus among experts on how to interpret monitor readings. Only when these obstacles have been overcome can these tools enter a stage where randomized clinical trials (RCTs) could establish their true repercussion on outcome. Furthermore, clinicians should abandon the illusion that every problem in medicine can be solved by RCTs. It is quite clear that RCTs designed to test the efficacy of some widely used monitoring devices, such as pulsioximetry, ICP, routine blood gases, etc., will probably never be conducted because of the lack of equipoise and the ethical issues that such trials can bring about. McCulloch and colleagues  published an excellent review of the problems that RCTs have in the field of surgery, and we believe their thoughts and conclusions can be easily extrapolated to the field of neurocritical care.
Obstacles in obtaining robust evidence in the field of neuromonitoring
Besides the concern of escalating costs associated with the use of new monitors and the subject of cost-benefit analysis in healthcare interventions, it is clear that most of the neuromonitoring tools we use in the NeuroICU, such as ICP, PtiO2, SjO2 or microdialysis, are acceptably safe for patients, even if eventually an RCT could demonstrate that they were useless. If we add this evidence to the reasons already mentioned, it is obvious that conducting an RCT to compare monitored with non-monitored patients will be complicated because of ethical reasons.
In the extremely unlikely scenario wherein a group of clinical investigators decided to conduct an RCT, and received the necessary funding, to prove the efficacy of ICP monitoring in improving outcome, it would be extremely difficult to convince most centres to enrol patients in such a trial. The main argument potential investigators would raise against such a trial would be that it has already been demonstrated that high ICP is the first cause of death and disability after TBI. In addition, it could be argued that no other neuroimaging or clinical exam is able to detect an increase in ICP early enough to be treated. Consequently, there is a lack of equipoise regarding ICP monitoring for most of the doctors who manage these patients.
Is equipoise a necessary requirement for conducting RCT?
As recently remarked by Chiong , clinical investigators should reconcile clinical research with the therapeutic obligation of fidelity to their patients by applying the equipoise requirement. Traditionally, it is considered mandatory that clinicians, including clinicians involved in research, have to offer their patients the best available treatment and cannot compromise their patients' interests, even for the sake of great potential benefits to others . The term clinical equipoise, coined by Freedman in 1987, is defined as ‘…honest, professional disagreement among expert clinicians about the preferred treatment' . If two interventions are in equipoise, it means that there is no good reason for believing that one is superior to the other . Clinical equipoise has to be distinguished from individual equipoise, a term introduced years before Friedman's definition to describe the individual state of uncertainty within an individual physician . If we extrapolate this to the use of monitoring tools, it is difficult to believe that communities of experts in neurotrauma would feel uncertain about ICP monitoring being better than no monitoring at all in the management of patients with severe TBI. An additional reason to complicate this dilemma is the fact that most accepted treatments and decision-making in TBI are driven by protocols guided by ICP monitoring.
Sceptics could introduce the argument that recent evaluations of pulmonary artery catheters (PACs), considered for many years a standard of care, have shown that PACs do not improve outcome and are, at best, a waste of money . In our opinion, the PAC case merits further analysis outside the scope of this paper.
Facing this scenario, a possible alternative for testing new monitoring tools could be conducting this type of trial in resource-poor countries where, for instance, ICP monitoring is not used at all. However, this situation would clearly be against the Declaration of Helsinki, in which it is stated that local circumstances cannot justify the use of control interventions that are not considered the best treatment for the condition in question . However, debate about equipoise and international research is still ongoing .
The reasons for clinical neutrality of monitoring devices
The PAC story is appealing and comparable to that of any complex neuromonitoring tools, which retrieve information about the pathophysiology of the injured brain, such as ICP, PtiO2 or microdialysis. Essentially, the PAC, like the ICP catheter, has been considered a cornerstone in the management of critical patients in the ICU and, when used, many clinical decisions are taken from their readings. A recent meta-analysis conducted by Shah and colleagues  showed that the use of a PAC in critically ill patients neither increased overall mortality or days in hospital nor improved survival or increased their length of stay. We believe that if similar studies were conducted with certain neuromonitoring tools, the result would be similar.
What are the possible reasons for the neutrality of the PAC, or any potential neuromonitoring tools? As remarked by Shah and colleagues , the first reason is that these tools are diagnostic and cannot be considered interventions. With the exception of the obvious differences in cost, these tools are comparable to an echocardiogram or a chest radiograph . To expect that any diagnostic tool improves outcome is unrealistic, unless we have a treatment that can modify the diagnosed abnormality and consequently modify outcome. Except for high ICP, we do not as yet have any accepted standard for treating either brain tissue hypoxia or an increase in anaerobic metabolism, the abnormalities that PtiO2 and microdialysis probes monitor.
Another important issue is that monitors may increase the accuracy of any diagnosis, but we need good quality data and staff skilled in interpreting them. Without these conditions, data can be useless and even misleading. A good example of this was shown in the study published by Cunningham and colleagues , in which the authors assessed whether the provision of computerized physiologic trend data in a neonatal ICU improved outcome in infants requiring intensive care . In all, 600 consecutive neonates admitted to ICU were randomly allocated to either trend data monitoring at the cot side or no trend monitoring. Conventional monitoring was conducted in both arms. The study did not demonstrate whether computerized physiologic trend monitoring had any impact on outcome . Therefore, at first glance it would appear this study gives ammunition to the sceptics. However, a completely different interpretation of this study could be proposed based on the evidence gathered by the same group in a second study conducted to clarify the reasons for such negative results.
McIntosh and colleagues  decided to evaluate the cognitive processes of doctors and nurses in interpreting monitoring data, testing how junior doctors, senior doctors and nurses used, viewed and understood data. In their study, each staff member of a neonatal ICU was shown 14 trend events and asked to interpret them. This study showed unexpected findings, mainly that senior doctors were able to make a correct diagnosis in only 68% of cases, junior doctors in 58% and nurses in only 25% . We do not know of any similar study conducted in NeuroICUs, but similar findings could be expected. In the PAC meta-analysis, Shah and colleagues emphasized that ‘Without standard protocols for the PAC, there may have been errors in gathering hemodynamic data, which may have ultimately affected clinical outcomes' .
It is beyond the scope of this paper to analyse in depth the different reasons as to why some monitoring events are not recognized, but the following factors may contribute to the high rate of error: (1) the problem of data overload in the ICUs; (2) the ICU is a very distracting scenario and facilitates human error; (3) the fact that multiple alarms are often disregarded for lack of reliability in some of them (‘crying wolf' situation) ; (4) the lack of adequate training of staff; (5) the wide variability of expertize in detecting clinically relevant data and trends (human error); and (6) the lack of scientific consensus on the interpretation of data, known as ‘consolidated' knowledge.
It is obvious that we need more translational research to get more evidence assisting in the interpretation of metabolic alterations found in brain-injured patients. Consequently, strong collaboration between basic and clinical research is needed to achieve this goal. The problem of cognitive overload in doctors and nurses needs to be addressed and better graphic display of adverse events should be implemented at the bedside. Also, we need to use new computer-based methodologies and algorithms to analyse time-series data and extract useful information, independent of the skill of the end-user. It is well known that in medicine, pattern-recognition is the easiest way to diagnose problems, and this ability is the domain where new statistical tools do extremely well . Continuous education of ICU staff in interpretation of monitoring data is mandatory to avoid suboptimal detection or even the failure to recognize adverse events that can influence neurological outcome. Once the improvement in the detection of adverse events is achieved, research in new interventions effective in treating them could increase.
To conclude, we have to remember that to achieve success in the field of monitoring, the human factor is most important. To achieve optimal results, we need units specialized in neurocritical care in which a multidisciplinary approach is used. Continuous education and training of staff in these units is probably the most important factor in changing the future of monitoring. A trained staff in an adequate environment would provide the opportunity to address the issue of efficacy in some monitors that sceptics demand. Even if these trials were definitely negative, we remain convinced, as Roizen and Toledano wrote many years ago, that ‘even the patients who may not benefit directly from a new technology may benefit indirectly from what physicians learn by using it' .
Competing interests: None.
We thank Sabrina Voss for correcting the manuscript. This study is supported in part by grants PI030153 and PI051092 from Fondo de Investigaciones Sanitarias (FIS) received by Dr J Sahuquillo.
1. Fleisher LA, Mantha S, Roizen MF. Medical technology assessment: an overview. Anesth Analg
2. Bush G. Project on the decade of the brain. Presidential Proclamation 6158. 17-7-1990.
3. Andrews RJ. Monitoring for neuroprotection. New technologies for the new millennium. Ann NY Acad Sci
4. Bullock RM, Chesnut RM, Clifton GL et al.
. Management and prognosis of severe traumatic brain injury. Intracranial pressure treatment threshold. J Neurotrauma
5. Forsyth R, Baxter P, Elliott T. Routine intracranial pressure monitoring in acute coma. Cochrane Database of Systematic Reviews
, 2001, Issue 3. Art. No.: CD002043. DOI: 10.1002/14651858.CD002043.
6. Bakker J. Lactate: may I have your votes please? Intensive Care Med
7. McCulloch P, Taylor I, Sasako M, Lovett B, Griffin D. Randomised trials in surgery: problems and possible solutions. BMJ
8. Chiong W. The real problem with equipoise. Am J Bioeth
9. Freedman B. Equipoise and the ethics of clinical research. N Engl J Med
10. Halpern SD. Evidence-based equipoise and research responsiveness. Am J Bioeth
11. Shah MR, Hasselblad V, Stevenson LW et al.
. Impact of the pulmonary artery catheter in critically ill patients: meta-analysis of randomized clinical trials. JAMA
12. Cunningham S, Deere S, Symon A, Elton RA, Mcintosh N. A randomized, controlled trial of computerized physiologic trend monitoring in an intensive care unit. Crit Care Med
13. Mcintosh N, Lyon AJ, Reiss J et al.
. The cognitive processes of doctors and nurses in the interpretation of physiological monitoring data in the neonate. Early Hum Dev
14. Tsien CL, Fackler JC. Poor prognosis for existing monitors in the intensive care unit. Crit Care Med
15. Mcintosh N. Intensive care monitoring: past, present and future. Clin Med
16. Roizen MF, Toledano A. Technology assessment and the “learning contamination” bias. Anesth Analg