That your “drunk” patient with presumed alcohol intoxication is not simply drunk but in fact has a different, potentially life-threatening cause for their reduced conscious level is a lesson best learned early in the Emergency Department and vicariously, if at all possible. Yet: it is a mistake we see made time and time again. Why is it so easy to think our patients are behaving in a particular way due to alcohol?
#EuSEM14 John Heyworth "we all make mistakes, unless we don't see any patients"
— Andywebster (@Andywebster) September 29, 2014
With 57% of the population of Great Britain reporting alcohol use as a part of their lifestyle, we are very used to seeing patients under the influence of alcohol – more so than other substances. A WHO report found that up to 45% of injured ED patients reported alcohol use prior to attendance – there’s really no doubt that alcohol use is common and it probably plays a significant role in the health of the patients we see, from contributing to illness and/or injury to affecting recovery and healing*. It is very easy, particularly in the evening and on night shifts when the proportion of attendees under the influence of alcohol may be as high as 78%, to wonder if everyone in the ED is drunk (apart from the staff).
But is that always the case? Of course not, as this study from the American Journal of Emergency Medicine shows.
What is this study about?
Recognising exactly the situation we described above, the authors of this paper set out to find out how often patients who were thought to be drunk actually weren’t, and to identify the aetiology or aetiologies of altered mental status for patients incorrectly presumed to be intoxicated with ethanol.
What did they do?
This was a retrospective, observational cohort over 4 years (Jan 2012 – Dec 2016), undertaken in a tertiary care county ED in the US with ED attendance figures of 110,000 patients per year. For comparison, here is the available data for four major trauma centres I have worked in at Consultant level:
- Central Manchester Foundation Trust >200,000 attendances across all sites In 2015/2016. These figures include the Manchester Royal Infirmary, Royal Manchester Children’s Hospital, the University Dental Hospital, St Mary’s Hospital (O&G), and Royal Eye Hospital Emergency Departments which are located on the same site but geographically separate. Figures from Trafford General Hospital are also included.
- Oxford University Hospitals >130,000 attendances across all sites in 2013/2014. These figures include Emergency Department attendances at the John Radcliffe Hospital and the Horton Hospital.
- Royal North Shore Hospital >83,000 in 2016/2017
- St George Hospital >77,000 in 2016/2017
The patients included were 18 years or older, with a chief complaint of “alcohol intoxication” or “altered mental status” – patients with initial breath alcohol of zero were included, while those with any detectable alcohol were excluded.
The way that they set about determining whether patients were intoxicated was interesting – they identified a cohort of patients “presumed to be intoxicated” if their electronic medical record had a chief complaint of “alcohol intoxication” or “altered mental status” and the patient was managed in a specific area of the emergency department designed and staffed specifically for the management of acute intoxication with alcohol or drugs (the “intoxication unit”). They have then included the patients with an initial breath alcohol concentration of zero, seeing this as evidence that alcohol was not responsible for the patient’s altered mental state.
This is an interesting practice point from a UK perspective – breath alcohol concentration measurement is not, in my experience, routinely performed in UK EDs and is often described as useless. Blood alcohol levels are sometimes measured in paediatric patients.
In Australia, blood alcohol levels are measured more routinely and used in a clinical context I haven’t quite managed to grasp as yet – seemingly the purpose is to determine whether or not a patient is fit for mental health assessment; I do not have a good feeling for how valid or reliable blood alcohol is for this purpose. I hadn’t really thought about breath or blood alcohol measurement as a rule out test and actually this may be where its usefulness lies; it is potentially important to think of it as rule out because even if alcohol is detectable, one or more of these other pathologies may be contributing but if negative, we can be pretty sure it’s not the booze. It’s hard to find data on the sensitivity of breath and/or blood alcohol measurement and on false negative rates, so there’s a reasonable assumption we are making here about how much we can rely on the results we are getting.
Aetiology data was obtained from electronic medical records; this was then categorised according to medical, trauma, psychiatric, or drug-related causes. They also recorded the hospital admission diagnoses presumably to capture circumstances where the cause of the altered conscious level and the ultimate diagnosis were not the same (eg hyponatraemia as cause of the altered mental state, lung cancer as the admission diagnosis causing the hyponatraemia through SIADH) and for drug-related causes they identified the substance where the drug was known or suspected substance if urinary drug analysis was not performed.
What did they find?
In the study, there were 29,322 patients with presumed alcohol intoxication – 1,875 had negative breath alcohol and were included in their analysis. That’s 6.39% of patients presumed to be intoxicated with alcohol who weren’t – around 1 in 16 patients.
The mean age of the “not drunk” patients was 41.7 years, with the range from 18-93 in the study. Take a moment to think about that – this could literally be any “intoxicated” ED patient.
The admission diagnoses were:
- drug-related 1,337 (71%)
- psychiatric in 354 (19%)
- medical in 166 (9%)
- traumatic in 18 patients (1%).
179 were admitted (10%), of whom 19 were admitted to ICU (1%).
Definition of “medical” diagnoses included sepsis/infection, hepatic encephalopathy, seizure, alcohol withdrawal, agitation/excited delirium, acute renal failure, intracranial haemorrhage (8 patients, 0.4%), hypoglycaemia (4 patients, 0.2%), hyponatraemia (2 patients, 0.1%), DKA (2 patients, 0.1%), other electrolyte disturbance (6 patients, 0.3%), CVA, TIA and GI bleeding.
Reading this paper reminded me strongly of another paper from earlier this year, this time from Annals of Emergency Medicine and addressing the other side of the coin – unexpected underlying pathology in patients who do have alcohol on board.
What is this paper about?
This paper addresses the opposite issue; here the authors are interested in those patients who do have alcohol on board but who go on to have coexisting significant diagnoses, particularly ones which necessitate critical care-level intervention.
There are a few common author names between the two papers and the description of the unit is familiar so I suspect this paper comes out of a research group at the same institution – good on them for tackling this important issue.
What did they do?
This was a retrospective, observational cohort study of patients over a longer time period (October 2011 to September 2016), presenting with the same complaint as above (“alcohol intoxication” or “altered mental status”) plus detectable breath alcohol and the patient had to be managed in the “intoxication unit” initially because this area was thought to represent low-risk ED encounters. They also stipulated exclusion of patients relocated out of the “intoxication unit” in the first ten minutes as this was felt to represent incorrect initial triage and thus patients more obviously unwell from the outset.
Again, outcome data was abstracted from the electronic medical record but where critical care interventions were necessary these data were extracted from the records by clinicians with standardised data forms.
Their outcome of interest, “critical care resource utilisation”, was defined as evaluation and management in a critical care area of ED (presumably ED resus) or ICU admission (including medical, surgical, burns or paediatric ICU).
What did they find?
There were 46,633 attendances with the complaint criteria they had pre-specified; 35,435 had a blood alcohol level > 0 and of these 31,788 were managed in the “intoxication unit” with a further 424 excluded because they were relocated within ED in the first 10 mins.
The remaining 31,364 attendances represented 11,175 unique patients with per patient attendances during the study period ranging from 1 to 227 – a point I’ll come back to later.
Of these patients, critical care resources were utilised during 325 attendances (1% of encounters), most often due to:
- respiratory compromise (96 patients, 30% of the 325)
- alcohol withdrawal (54 patients, 17% of the 325)
- sepsis or infection (34 patients, 10% of the 325)
205 patients were intubated (0.65% overall, 63% of the 325) and 256 required ICU admission (0.8% overall, 79% of the 325). There were two deaths, both due to CVA. Three patients had a cardiac arrest while in the intoxication unit; one with a subarachnoid haemorrhage, one with a tachydysrhythmia, one with hyperkalaemia – all were resuscitated and subsequently discharged “without neurological deficit”, although the narrative part of the paper describes one 29-year-old patient as being discharged to a long-term care facility .
The authors also looked at factors potentially predicting the need for critical care – as you might expect, markers of illness such as hypoglycaemia, fever, hypotension, hypoxia and hypothermia all featured but alcohol level was not associated with increasing odds ratio for critical care.
What are the problems with taking these papers at face value?
There are a couple of issues with the external validity of these papers. Both look at US populations and while there are undoubtedly similarities between these patients and the ones we see in the UK and Australia, it is always helpful to think critically about how patients, care and healthcare systems may differ in different nations.
I have already identified some alcohol intoxication-related practice differences with regard to testing. The “intoxication area” is also an unfamiliar concept for me; while it may be practical for the ED I can foresee significant issues with confirmation bias in as much as patients moved to this area will likely be difficult to relocate as you may know from your experiences trying to relocate unexpectedly unwell patients from the fast track/minors area. Diagnostic momentum in the ED is incredibly hard to overcome.
There’s also the issue of the preservation of the remit of the “intoxication area” in the era of ED overcrowding. When we are short of beds it’s tempting to use those resources for patients who wouldn’t normally be in that area and once the lines are blurred it may be very difficult to re-establish them.
But there are some very important messages evidenced in these papers, ones that we as ED clinicians should ponder carefully and probably read out loud at regular intervals.
What do these papers this tell us?
Firstly, we can’t assume that patients are drunk. In 6.39% of cases at the hospital in the first study, the cause of altered mental state wasn’t alcohol and it’s very easy for us to miss this due to errors such as anchoring bias, confirmation bias and fixation – particularly as 55% of these patients were delivered by EMS and 31% by police; these colleagues often helpfully provide us with their assessment along with the patient. It remains our duty to be open-minded.
Secondly, alcohol breath testing might be a useful test to prevent us from falling into this trap but we have to be wary of its reliability. It would only really be useful as a rule-out strategy – the second paper tells us, as we already know, that alcohol level tells us nothing about the likelihood of requiring critical care, but if there is no alcohol to detect then we certainly shouldn’t be attributing altered conscious level to alcohol and should actively seek out an alternative explanation.
Thirdly, the second paper highlights the issue of regular attenders in this particular context. Many Emergency Departments see a cohort of patients who attend frequently; in addition to having a number of social and medical issues confounding their assessment, their presence in this group reflects the increased prevalence of substance misuse in this challenging population but also the significantly higher chance we will give all of these confounders greater weight and potentially miss significant pathology. This is a real and regular challenge for Emergency Physicians. These patients are often “difficult” for more than one reason – and that should give us pause.
Risky ED pt groups
Christian Hohenstein #EuSEM14
— Natalie May (@_NMay) September 29, 2014
There is increased risk in patients who reattend.
Full assessment mandatory.
John Heyworth #EuSEM14
— Natalie May (@_NMay) September 29, 2014
Finally, even if alcohol testing is positive we still have to convince ourselves that there is no other pathology present. These patients also deserve a thorough examination and assessment. All patients with altered mental state, with or without alcohol in evidence, should have a blood sugar level checked at the very minimum, not least because we know alcohol can cause hypoglycaemia in people with type I diabetes and reduces hypoglycaemic awareness. There’s insufficient evidence here to recommend a further strategy but there is enough to remind us to take these patients seriously, to observe them appropriately and to dismiss them has having had “one too many” at our peril.
*Note – it’s reasonably tricky to find good quality evidence around proportion of ED patients who drink at all or who have ingested alcohol before attending; I’m sure this is multifactorial including the issues of relying on self-reporting, but I am still interested in comparing general proportions of alcohol use among ED patients with the general population. If you have some reliable data on this, please let me know.