Author: Andy Neill / Codes: IP1, IP2, SLO1 / Published: 06/02/2017
It’s probably fair to say that few of us give any great thought on how a temperature is measured. I admit it’s always amusing when the new start student nurse puts the temperature in the respiratory rate box but rarely do we consider the accuracy of the number presented to us.
A fever catches our eye in Emergency Medicine, it often radically changes the diagnostic pathway, treatment and disposition of our patients. An 80 year old with a temp of 39 is a different kettle of fish from an 80 year old with a temp of 37.
A reasonable number of us could cite a sensitivity and specificity of a CT at 6 hrs for ruling out SAH or the sensitivity of the newest high sensitivity troponin T but I doubt any of us could cite a sensitivity or specificity for one of the most commonly used devices in our EDs.
So what is the sensitivity and specificity then?
Let’s just say at this stage of the blog that the diagnostic characteristics of the tympanic thermometer used in your ED is somewhat poor.
Ah.. such impatience..First a little, needless, background.
To many of us in the sepsis-sphere, the English word ‘fever’ probably stems from the latin root ‘hypoTazocinaemia’ and its use in the modern ED has likely triggered more sepsis alerts for 19 year olds with tonsillitis than any other vital sign in the known world.
So how can we measure people’s temperature?
Probably the best way to do so is via some kind of intravascular catheter or some other catheter lying deep within some body cavity unspeakable in polite company.
In trials looking at this subject, reference standards include pulmonary artery catheters, oesophageal catheters and urinary catheters. The most commonly used central temperature I’ve seen used in the ED is the rectal thermometer.
And think back to the last time you saw a rectal temp recorded in your ED – was it that 80 year old on New Year’s day who’d been lying on the kitchen floor for 9 hours with a broken hip and an ECG like this?
[LITFL ECG library]
This seems to be a fairly accepted and commonly practised indication for abandoning the humble tympanic thermometer and searching the cupboards for the rectal thermometer that gets somewhat awkwardly placed in a bit of frozen poo in the rectum.
How do thermometers work?
Honestly I would have said ‘pixie dust’ if asked this in an exam but it turns out it’s more scientific than that.
Tympanic thermometers work by infrared.
They measure the infra-red spectrum of light emitted by the tympanic membrane. As the tympanic membrane gets warmer it emits a different frequency of infra-red light. The machine correlates the infra red light with a temperature and Robert’s your Mother’s Brother you have a temperature reading.
Rectal thermometers, and I think most central thermometers, work by a ‘thermocouple’ (Not two hot people on a date…) and utilise the Seebeck effect.
The problem isn’t actually the tympanic method, it’s probably because the ear is too peripheral to accurately reflect the central temperature. In the same way that many patients with severe sepsis and high central temperatures have cool peripheries so many patients with high central temperatures have relatively cold tympanic membranes.
Think the opposite applies for those with otitis media getting their temp measured tympanically? Think again, apparently, as studies have shown that having a red, bulging tympanic membrane does not give you a different tympanic temp than if it was taken via the axilliary route.
However, be cautious with tympanic thermometers in the very young. The rule is simple: if it doesn’t fit in their ear canal, then it’s not likely to be accurate.
And if parents tell you they own a disposable forehead thermometer – the kind which have colour codes for different temperature ranges- please do encourage them to invest in a more accurate thermometer while they have a child (unless they want to risk a PUO – pyrexia of unknown origin- screen).
So what are the diagnostic characteristics of the non-invasive thermometers then?
Well, according to one of the most recent systematic reviews and meta analyses on the subject the sensitivity is 64% and the specificity is 96%. The reported range of inaccuracy was plus or minus 1.5 degrees.
This was actually a fairly nicely done study, with a well defined search strategy and appropriate inclusion and exclusion criteria (prospective studies only, all had to have a clear ‘gold standard’ comparator, ie rectal, oesophageal, endovascular temp etc…) and it’s by no means the first paper to question the use of non invasive techniques for measuring temperatures.
What does that mean for my practice then?
If your tympanic thermometer reads 39 degrees then it’s probably true. It’s a very specific test.
However if your tympanic thermometer reads 37.5 degrees then it doesn’t mean a great deal. The patient may well have a temperature and you’ve just missed it with your infra-red duck/goose detecting machine.
So in short, a peripheral temperature is good as a rule-in test for fever/hypothermia, but not so great as a rule-out.
Perhaps the best comparison would be the FAST examination in trauma. All of us who have done our level 1 ultrasound competencies will be able to recognise that a positive FAST examination is probably reflective of reality – the patient probably does have free fluid in the abdomen – it’s a fairly specific test. However we’re all savvy enough to know that the FAST exam in trauma will often show no free fluid even when in reality there is free fluid there- in other words it’s not a very sensitive test.
In the situation of trauma, if we have a reasonably high pre-test probability that the patient has a serious injury we don’t just trust the poorly sensitive FAST exam, we go on and arrange a more definitive test like a CT.
So maybe, when we have an 80 year old who we’re sending to the medical team with the oh so unsatisfactory label of ‘collapse query cause’ or ‘acute confusion’ when we were suspicious of an infectious cause we should really consider getting the definitive test and getting a rectal temperature.
P.S. A non-scientific word of warning here for those planning on changing practice in their work up of patients: If you suddenly decide to request rectal temperatures on every punter who rolls through the door with a ‘normal’ temperature on your tympanic be prepared for some very poor multi source feedback for your eportfolio and perhaps you better invest in a new Nespresso machine for the tea room and learn from Cliff Reid’s tips on how to make things happen.
- Accuracy of peripheral thermometers for estimating temperature: a systematic review and meta-analysis.
- International Journal of Paediatric Otorhinolaryngology: Infrared transtympanic temperature measurement and otitis media with effusion
- LITFL Hypothermia ECG
- Cliff Reid, Making things Happen, SMACC 2013
- Don’t Forget the Bubbles
I just found this paper published in print in the EMJ just now. It’s an observational trial looking at temp measurement and it found just what we all expected it to find. ‘None of the non-invasive methods met benchmarks for diagnostic accuracy using the criterion of 38°C to detect rectal temperature of 38°C’