Paediatrics is a specialty where lying about a diagnosis is normal practice. It's not because we're bad people. When you think about the challenges of diagnosis in children combined with the expectation of a diagnosis, it is completely unsurprising. The adult accompanying the child would like a diagnosis (please and thank you) and the clinician would very much like to give one (you're welcome).
While that all seems very reasonable, in child health it often isn't entirely truthful. It is one of the mantras of medicine that the diagnosis is going to come from history and examination in most cases. Hurrah for clinical diagnoses. In paediatrics, the history is often from a third party and will have an inevitable element of bias. The examination will also contain more uncertainties more of the time. You have to accept a significant lack of information when interpreting examination finding in children.
The result of this is that clinical diagnosis is more challenging in paediatrics. Here's the paradox: clinical diagnosis is the default position in child health. Why? Because we don't want to do tests on children or give them treatments "in case" unless these investigations or therapies are very likely to benefit the child.
This week, something big happened and it didn't even hit the news. The General Medical Council released some new and updated guidance: "Guidance on professional standards and ethics for doctors
Decision making
and consent." While much of the content is old news, there is a new emphasis on honesty when there is diagnostic uncertainty that is hugely relevant to paediatric practice, thanks to the fact that uncertainty is where we work.
So, when are we lying to our patients or the adults that accompany them? The truth is that there is a spectrum of how far what we tell people lies from the truth. What we should probably do in the light of the new GMC guidance is to re-evaluate our approach to a variety of clinical presentations and ask, "Should I change what I say about this?"
You could argue that nothing is certain in medicine, so what are the thresholds of uncertainty that decide when we should be honest in this way? That's a fair comment. We need to apply some measure here - enter the certometer.
The certometer takes the things that we are already using in our diagnostic approach and gives us an idea of how truthful it is to give that diagnosis. Last week, I asked the medical Twitter world for a few suggestions of diagnoses that we could feed into the Certometer and this seems like a good time to give this contraption a go.
First up is an intriguing suggestion: Diagnosis - Viral illness.
In my experience this diagnosis is usually given to children with a fever and signs or symptoms of upper respiratory tract infection without signs or symptoms of a more specific diagnosis.
Let's imagine a common scenario then: a 2yr old previously healthy child with a fever for 2 days. They have a runny nose but no cough. They have no respiratory abnormality. Pharynx and both tympanic membranes are inflamed.
The pre-test probability of this being a viral illness is high. It's a child with a fever so the probability that the illness is viral is around 90%.
Positive predictors of a viral cause do exist and include wheeze and urticarial rash in children. This child has none of these things.
Good negative predictors of a diagnosis of viral illness in this sort of case would be some signs of suppurative complications such as mastoiditis. We haven't seen any signs to suggest this.
So having looked for something specific that truly discriminates and found none, what you are left with is your pre-test probability, dialled down slightly by virtue of the absence of signs of another diagnosis. In other words, all we have truly achieved is to rule out complications. Since complications are rare, we're essentially no more certain this is a virus than before we started.
Calling it a viral illness implies that we've added some certainty to the underlying cause that in reality, we haven't. In fact, by calling it "a virus" we have admitted that there are no specific finding identifying a particular viral illness.
What we have done is far more important. We have looked for signs of complications and more serious infection (sepsis, meningitis etc) and found none. What we can say with honesty and certainty is that this is an uncomplicated upper respiratory tract infection.
To emphasise the point about how often the lack of specific signs and symptoms is the norm in paediatrics, I'll give a couple of examples of common, clinical diagnoses that are usually made with enough certainty to be considered completely honest.
- Croup
- Chickenpox
- Febrile convulsion
- Vasovagal syncope
Yep, that's pretty much it. Most other common problems are really labels given with real uncertainty due to the lack of specific signs or symptoms with good positive or negative predictive value.
Here are a few other examples of diagnoses that are commonly given in what is in reality a great deal of uncertainty that this problem is causing the symptoms or signs.
- Infant reflux disease
- Cow's milk protein allergy (non-IgE)
- Asthma
- in the under 5 yr old child
- where the diagnosis is based on chronic cough without wheeze
- Mesenteric Adenitis
- Hypermobility
Then there's a whole new level of diagnostic uncertainty. At the beginning I used colic as an example. Let's try feeding a classic colic presentation into the Certometer. You see a three week old baby whose only symptom is "crying all the time". The pregnancy and birth were uncomplicated. The baby examines normally and is thriving. They are feeding well and passing urine and stools normally.
What is the pre-test probability that this is colic? Unfortunately there's no good answer to that because it's not an actual disease. There is no pathology or treatment. Colic is simply a label to be given to crying infants that have no pathology. If you try to put this through the Certometer, you will break it because you can't have any certainty of something that doesn't exist.
It is often argued with colic that the label is therapeutic. The new GMC guidance should give us an opportunity to re-evaluate that approach. What would be wrong with telling the parent of the infant described above that their baby is normal and healthy? That would be honest and potentially just as therapeutic. We could then use the time that we might have spent (explaining a condition that doesn't exist) on being supportive and encouraging to the parent. The crying excessively phase does settle and in the meantime, it's all about making sure that it doesn't break the parent.
Here are a couple of other examples of diagnostic labels in children that are without evidence for any disease process. Neither of these has ever had any pathology associated or been shown to respond to any treatment:
- Growing pains
- Non-specific abdominal pain
Is it time to embrace the idea of greater honesty when we diagnose and explain symptoms in children? I certainly find that an explanation without a diagnosis is entirely acceptable to families when it comes to a situation where in the past I might have given a non-diagnosis. Changing that practice is relatively straightforward. You simply stop saying the thing.
For the situations where we are dealing with an actual diagnosis but there is significant uncertainty, we've got a few options. The infant with crying and regurgitation of feeds is a good example. Perhaps we should be stricter about starting off with a label of "possible GORD"? Perhaps we should go further and start with "Feeding symptoms under observation and follow-up." Increasingly, I don't give a diagnosis. Instead I tell the parents that (in the absence of red flags such as fatering growth) "crying and regurgitation can be normal, it can be early symptoms of reflux disease and it can be rarer problems such as allergy. We don't want to give unnecessary treatments to babies but we also want to treat problems when it's going to help. This is how we're going to try to get the right balance between those two things..."
It's a whole new world being honest about our uncertainty but it does work and it works like this:
Edward Snelson
@sailordoctor
