Back in 2013, I wrote a post here on petechiae in well kids; it received a phenomenal number of views for reasons I don’t quite understand (I think it ended up as a link somewhere significant) and for a while it was our most read post. What is clear though is that whatever the mechanism that drove up the views of that particular post, rashes in children remain a significant diagnostic challenge and a source of anxiety for Emergency clinicians across the globe.
Petechial rashes (and purpura, and other non-blanching rashes) are much feared because of their association with meningococcal sepsis, that nasty little disease that presents insidiously and often with benign non-specific symptoms (at least, in every patient I’ve ever seen), before progressing to a severe and sometimes fatal overwhelming sepsis. Of course, there are many causes of these rashes; from traumatic, to viral, to haematological, but we live on a heightened state of alert because the implications of missing a case of meningococcal sepsis are both potentially devastating for the patient and potentially opening the doctor up to criticism, both at work and in the media.
This paper, published online in the Lancet Infection, describes a study in which the authors attempted to validate clinical practice guidelines in identifying and appropriately treating paediatric patients with non-blanching rashes. The abstract is below, but as we always say, please read the full paper yourself.
What is this paper about?
This study was a collaboration by the Paediatric Emergency Research in the UK and Ireland (PERUKI) group. The authors wanted to compare how accurate UK clinical practice guidelines were for diagnosing meningococcal disease in febrile children with a non-blanching rash. They have called it the “Petechiae in Children (PiC)” study, which is a little misleading in itself as they actually recruited children with a fever (>38ºC) and a new-onset rash of petechiae or purpura, or that was non-blanching. It’s a little hard to elicit this from the manuscript itself as they describe the symptomatic inclusion criteria as:
“a new onset non-blanching rash or features suggestive of meningococcal infection.”
The line immediately made me stop and wonder what they meant, since I’ve seen more meningococcal infection without a rash than with it. However, when you dig down into the supplementary material you can see the case report form used for initial data collection (CRF 1), which specifies “new petechia/purpura or non-blanching rash.” So yes, they are talking about children with fever and petechiae/purpura/non-blanching rash.
What did they do?
The study was carried out collaboratively across 37 Emergency Departments in the UK. Sensibly, those patients known to have other reasons to have non-blanching rashes were excluded, and due to the nature of the illness in question, consent was deferred so that clinical care could be delivered. The idea was that these patients presenting with potential meningococcal disease were investigated and treated in accordance with local protocols (or NICE Guidelines, where local protocols were not in place), and the outcome measure was not to do with the patients themselves but the performance of the protocols or guidelines in appropriately recognising and treating meningococcal disease when it was present and in avoiding over-investigation and over-treatment where it was not.
Here the authors considered the protocols and guidelines as “tests” and used test characteristics – sensitivity, specificity, NPV and PPV – as their performance measures. They also did some work around the cost inherent in following each protocol or guideline, though this was not the primary outcome.
The presence or absence of disease was determined by identification of meningococcal disease or other invasive bacterial infection, defined as positive PCR for N.meningiditis or other bacterial pathogen from a “sterile body site” (here they have clarified blood and CSF). There were various attempts to ensure that patients who did not have bloods at this first presentation didn’t bounce back with significant illness later.
There were 1329 children (<18yrs) included in the analysis, which is pretty impressive and a clear benefit of this networked style of research.
What did they find?
There are a few things we can learn from this paper.
Firstly, the rate of invasive bacterial infection was 2% (26 of 1329 patients), of whom 19 had confirmed meningococcal disease (1% of the total). The authors felt this was low but it is certainly comparable to previous studies, such as this one from 1997 which found a rate of 1.9% bacteraemia or clinical sepsis in a cohort of 411 children with fever and petechiae.
The most widely used guideline was NICE CG102, used in 29 of the 37 sites. Interestingly, adherence to the departmental guidance sat at around 46%, irrespective of whether it was a local or national protocol. Where clinicians did not follow the guideline, it usually meant fewer incidences of antibiotic administration. In this cohort, that also meant that two of the 1329 patients were “incorrectly discharged from the hospital with early meningococcal disease,” both of whom subsequently reattended and were treated without a need for critical care and with resultant survival.
Looking at the individual protocols, all of the local and national variations would have advocated treatment for the 19 patients subsequently found to have meningococcal disease, giving a broad sensitivity of 100% (95% confidence interval 82-100%). Because it was deviation from protocol that sent 2 of the 19 patients home without treatment inappropriately, this was been grouped as “clinician practice” and demonstrated to have a lower sensitivity than following protocol (89%: 95% confidence interval 67-99%).
Specificity was low across the board and lowest with the NICE guidelines (CG102 and CG51). The highest specificity was clinician practice (56%:95% confidence interval 53-59%) and the highest specificity for a clinical practice guideline was the London guideline at 36%(95% confidence interval 34-39%).
There’s another nugget hidden away in the manuscript too – only 5% of the patients without meningococcal disease had a purpuric rash, versus 68% of those with meningococcal disease. Compare this to petechial rashes, present in 95% of those without meningococcal disease and 32% of those with meningococcal disease.
What does it mean?
So following the documented protocol for patients with fever and rash (petechiae, purpura or non-blanching rash) mandates treatment and thus ensures these patients won’t get “missed”. That’s pretty straightforward and it shouldn’t be a surprise to us. These guidelines are set up to be a catch-all, blanket approach, aiming to ensure we don’t let any super sick kids slip through the net.
But what of all the other patients, who didn’t have meningococcal disease? The most interesting thing for me was that NICE CG51 advocated treatment for all 1329 patients. That’s a pretty blunt tool! And CG102, the most widely used guideline, wasn’t much better, with only 15 patients not being treated.
When we get to the local guidelines, the numbers of patients not being treated are a bit bigger: the London guideline avoided treatment for 475 of the patients without meningococcal disease, which was the highest number. The guideline taken from the DFTB site is shown below. I’ve you’ve not read their blog on this study then you should as soon as you’ve finished this one. You can find it here.
The paper really demonstrates to us that we should be following guidelines as they are better than we are (if we consider the “clinician practice” row of the diagnostic performance table, which represents deviation from protocol to be a form of Gestalt), but that in turn following the NICE guidelines lacks the nuanced approach that some local protocols can offer.
I’m avoiding getting dragged into the cost part of the paper as I am not a health economist, nor am I prepared to put a price on avoidance of missing a case of meningococcal disease. I will say, though, that a purely financial analysis doesn’t capture entirely the cost of treating patients who do not need it (consider harms that might be caused by inappropriate antibiotics, blood tests, complications from IV cannulae, the time and psychological costs of having a child admitted to hospital…). There’s a bigger picture here that we aren’t quite seeing.
Still, I think this paper does reinforce something important:
- PURPURA are BAD (I do a rashes teaching session for ED registrars and this is always one of my take-home messages, it’s nice to clearly be able to reference this now).
And some things to think about
- We should follow our local guidelines, or national guidelines if we don’t have local ones, because in this case we really aren’t as smart as we like to think we are
- NICE guidelines are a blunt tool; in this case, local guidelines seem to perform better across the board. Maybe we should all be using the London guidelines, but it’s certainly hard to justify the extra testing and treatment mandated by NICE CG51 and CG102 without demonstrable benefit.
- And, reassuringly, meningococcal disease remains rare.
Natalie May @_nmay
- Don’t Be Rash – Petechiae in Well Kids at St Emlyn’s https://www.stemlynsblog.org/dont-be-rash-petechiae-in-well-kids-at-st-emlyns/
- Validating clinical practice guidelines for the management of children with non-blanching rashes in the UK (PiC): a prospective, multicentre cohort study https://www.thelancet.com/journals/laninf/article/PIIS1473-3099(20)30474-6/fulltext
- PERUKI https://www.peruki.org/
- NICE guidelines on assessing petechial rashes https://cks.nice.org.uk/topics/meningitis-bacterial-meningitis-meningococcal-disease/diagnosis/assessing-the-rash/
- St Emlyn’s lesson plan on diagnostics https://www.stemlynsblog.org/lesson-plan-introduction-to-diagnostic-testing/
- Incidence of bacteremia in infants and children with fever and petechiae https://pubmed.ncbi.nlm.nih.gov/9329416/
- Meningitis (bacterial) and meningococcal septicaemia in under 16s: recognition, diagnosis and management https://www.nice.org.uk/guidance/cg102
- Sepsis: recognition, diagnosis and early management https://www.nice.org.uk/guidance/NG51
- What is Gestalt https://www.stemlynsblog.org/gestalt-st-emlyns/
- Tessa Davis. Petechiae in Children – the PiC Study, Don’t Forget the Bubbles, 2020. Available at: https://doi.org/10.31440/DFTB.30782
1 thought on “JC: Spot the Difference, what can we do about petechiae? St Emlyn’s”
Great review! Thank you so much!
Good to see guidelines being validated .
To small points
– 1: I agree that purpura is bad and should always demand our attention and considered worked up aggressively . However in this study they seem (relevantly) to have excluded the most common benign(ish) causes of purpura (I.e HSP, ITP etc) so the “risk of meningitis in the child with fever and purpura”-number is maybe a bit inflated?
– 2: even though the conclusion might be that guidelines are the way to go with fever and petechiae in children , I just wanted to highlight, Trisha Greenhalgs point, that it requires clinical judgement (incl gestalt) to know whether at all to use a guideline in the particular patient (as Simon Carley stated in my favourite blog on St Emlyns “Making good decisions in the ED”: first and foremost you need to be looking for petechiae as part of your work up. Then evaluate: is that a petechiae or not? And then: is this child in a bad way or not?).
This might not be that hard in this particular area (IRR on petechiae I suspect is probably quiet good) . But in most areas guidelines should be validated tools (Streetlights) to be specifically used once we have found our way through “the shadows” of patient interaction (ref: Trisha greenhalg: Of lamp posts, keys, and fabled drunkards: A perspectival tale of 4 guidelines + Gary Klein: streetlights and shadows)
Thanks again for all your great work!
All the best
Peter, EM resident in Denmark/ Sweden, Akutmedicineren.dk