The Gist: The Glasgow Coma Scale (GCS) is widely used, yet complicated by clunkiness and poor inter-rater reliability (explanation of kappa). The Simplified Motor Score (SMS) is easier to use and equivalent, although this is prone to similar limitations. Until a better means of communicating mental status comes, it may be best to communicate what the patient is doing (opening eyes to voice, moaning incomprehensibly, localizing pain). See this ScanCrit post.
The Case: A 29 year old male involved in a MVC with multiple traumatic injuries resulting in a prolonged ICU course at Janus General had a tracheostomy placed for respiratory failure. The patient was responding appropriately to questions, following commands, opening his eyes spontaneous and lacked any signs of confusion or delirium, mouthing words, but was awaiting tracheostomy exchange for a fenestrated trach with a Passy-Muir valve. What's the patient's GCS? Does this patient's GCS reflect their mental status?
- Documented at as 10NT, 11T, and 15 by various providers. The arguments behind each: 10NT - cannot test verbal, 11T -one point for showing up, 15 -patient oriented, and saying appropriate words, just without phonation.
Other critical components to the physical exam and evaluation are less easily quantified. The mental status, for example, is a key component to evaluating a patient. The GCS was developed to communicate the mental status of a head injured individual among providers during continuing care in a neurosurgical unit [1]. It is often used to track neurologic status when transferring care or over time. A particular GCS in the prehospital setting may also qualify a patient for a trauma activation in some settings.
Limitations: Unfortunately, unlike other vital signs, scores don't have explicit meaning. The total GCS is often reported, yet this 13 point scale (3-15) actually has 120 different possible combinations. A patient with a GCS of 10 may be completely oriented but totally paralyzed or be moaning incomprehensibly with their eyes own and a withdrawal reflex present. Further, the sum of the GCS does not equal the parts, with regard to mortality. Healey et al used the National Trauma Database to model mortality predictions based on GCS and found that the same total sum score could be associated with double the mortality (ex: from 27% to 52%) depending on the individual components. Further, the mortality associated with scores is not linear [3]. So a GCS of 11, for instance may mean very different things for two different patients.
Limitations: Unfortunately, unlike other vital signs, scores don't have explicit meaning. The total GCS is often reported, yet this 13 point scale (3-15) actually has 120 different possible combinations. A patient with a GCS of 10 may be completely oriented but totally paralyzed or be moaning incomprehensibly with their eyes own and a withdrawal reflex present. Further, the sum of the GCS does not equal the parts, with regard to mortality. Healey et al used the National Trauma Database to model mortality predictions based on GCS and found that the same total sum score could be associated with double the mortality (ex: from 27% to 52%) depending on the individual components. Further, the mortality associated with scores is not linear [3]. So a GCS of 11, for instance may mean very different things for two different patients.
Yet, even if the numbers did mean something, the GCS has been found to have abysmal inter-rater reliability. In one study, 19 emergency physicians rated 131 patients within five minutes of each other found a concordant GCS 32% of the time (Spearson's rho 75, weighted kappa 0.40) [4]. Even in the rather protected setting of case based written scenarios, emergency providers the overall GCS accuracy was 33.1% (95% CI, 30.2-36.0) [5]. In a written mock scenario, EMS personnel (n=178) generated an accurate GCS one-quarter of time without a scoring aid and a shocking 57% with the use of a scoring aid [7].
Alternatives: Given the inaccuracy of the GCS, Thompson et al set out to determine whether the performance of the SMS, a truncated version of the GCS was equivalent to the GCS in a retrospective cohort of out of hospital head injured patients. In the SMS, points are awarded for obeying commands (2), localizing pain (1), and withdrawing to pain or worse (0). They found that the predictive nature of the SMS paralleled that of the GCS, although the GCS seemed to predict mortality slightly better, 0.90 using GCS (0.88-0.01) vs 0.87 using SMS (0.86-0.88) [7].
Alternatives: Given the inaccuracy of the GCS, Thompson et al set out to determine whether the performance of the SMS, a truncated version of the GCS was equivalent to the GCS in a retrospective cohort of out of hospital head injured patients. In the SMS, points are awarded for obeying commands (2), localizing pain (1), and withdrawing to pain or worse (0). They found that the predictive nature of the SMS paralleled that of the GCS, although the GCS seemed to predict mortality slightly better, 0.90 using GCS (0.88-0.01) vs 0.87 using SMS (0.86-0.88) [7].
So, what do we do?
Trashing the GCS is simply not an option for most of us; yet, score cards don't seem to do us any favors. For example, during a trauma activation, the expectation (at least at Janus General) is to communicate to the room the patient’s GCS. This may seem to convey more neurologic information than the actual exam discriminates. Recognizing the limitations to the GCS is important in discerning both what we do with this information and how we communicate what we mean, whether it's in documentation, to other providers, or family members.
- Describe the exam. With the knowledge of the subjectivity and poor reliability of the GCS, one may give the breakdown of points rather than a simple total GCS and describe the neurologic examination. Documenting descriptors in medical records this may aid other teams in tracking the patient's exam.
- We may also engage in interdisciplinary discussions about use of simplified scoring systems such as the SMS or about the ways we communicate and document neurologic exams.
References:
1. Green SM. Cheerio, laddie! Bidding farewell to the Glasgow Coma Scale. Ann Emerg Med. 2011;58(5):427–30. doi:10.1016/j.annemergmed.2011.06.009.
2. Singh B, Murad MH, Prokop LJ, et al. Meta-analysis of Glasgow Coma Scale and Simplified Motor Score in predicting traumatic brain injury outcomes. 2013;27(March):293–300.
3. Healey C, Osler TM, Rogers FB, et al. Improving the Glasgow Coma Scale score: motor score alone is a better predictor. J Trauma. 2003;54(4):671–678; discussion 678–680.
4. Beveridge R, Ducharme J, Janes L, Beaulieu S, Walter S. Interrater reliability of Glasgow Coma Scale scores in the emergency department. Ann Emerg Med. 2004;43(February):215–223.
5. Bledsoe BE, Casey MJ, Feldman J, et al. Glasgow Coma Scale Scoring is Often Inaccurate. Prehosp Disaster Med. 2014. doi:10.1017/S1049023X14001289.
5. Bledsoe BE, Casey MJ, Feldman J, et al. Glasgow Coma Scale Scoring is Often Inaccurate. Prehosp Disaster Med. 2014. doi:10.1017/S1049023X14001289.
6. Feldman A, Hart KW, Lindsell CJ et al. Randomized controlled trial of a scoring aid to improve glascow coma scale scoring by emergency medical services providers. Ann Emerg Med. 2015 Mar;65(3):325-329.e2.
7. Thompson DO, Hurtado TR, Liao MM, Byyny RL, Gravitz C, Haukoos JS. Validation of the Simplified Motor Score in the out-of-hospital setting for the prediction of outcomes after traumatic brain injury. Ann Emerg Med. 2011;58(5):417–25.
Interesting site and concept. I am a fellowship trained subspecialist in GU function, and old enough to doubt everything I'm told by textbooks, patients, and fellow clinician/scientist, but also have had time to research and publish about my questions (another discussion: what is the value of "academic success"?). You are at the other end of your career. Your posts reflect questioning of practice, the metaphysics of paradigm. I hope this leads to an academic and dedicated level of questioning the paradigm however. GCS, normative lab values, the importance of patient texting - these are the first or second level of descriptive language.
ReplyDeleteThe real question is, do we know what we're doing? In my world, "overactive bladder," "stress vs. urge incontinence," and more recently "urge incontinence" are regarded as descriptors of disease entities or at least the patient experience which then has something to say regarding diagnosis. But is that true? Have these boxes produced promising and effective results? I argue that by and large they have not produced positive results out of proportion to their expense and risks. And the problem is not the application of intelligence to the paradigm....the problem is the paradigm itself.
Humans have an evolutionary need to classify the world. A low kappa of the GCS is more about a failure to deeply understand the pathophysiology than it is a failure of the scale's modelling. Antimuscarinics don't work because it's not about an "overactive" bladder. Vaginal mesh is the result of financial drive (directly and/or indirectly) moreso than any basis in biomechanical engineering. Such great things we bring to our patients, the very people who give us license to poison and injure them in the hopes of improving their health.
Do we really do anything more than facilitate maladaptive lifestyles?
Argument/Challenge: other than helping the potentially moribund avoid premature death from acute trauma, what is it that patients tell us by their symptoms? Little more than they perceive themselves to not be euboxic. Our job as physicians should be to understand that single fact and then be the detective that sorts out the brain/body interaction leading to that perception of dis-ease. Is it real? What possibly can be the texter's perception of reality, to not give the physician one's full attention?........that this kind of thing happens should be a wake-up call to the medical profession that we are marginalizing our usefulness.