You don't need to be signed in to read BMJ Blogs, but you can register here to receive updates about other BMJ products and services via our site.


Basics: AVID

21 Aug, 15 | by Bob Phillips

060314_1404_Whenatestis1.jpg The shortcut world of acronyms for critical appraisal was lacking one for diagnostic test accuracy – we have RAMbo for RCTs, FAST for systematic reviews, but what of the poor reader of studies evaluating a new test?

We know the basic idea – patients who are considered to potentially have the diagnosis in question have both the test-under-evaluation and the as-good-as-we, these are assessed without looking  at the results from the other one, and that if there are cut-offs these are reproducible.

Wait! That’s it … more…

When a test isn’t a test

8 Jun, 14 | by Bob Phillips

There are many reasons why we request tests, in medicine. One imaginary patient’s journey picks up a number of them.

Take a patient who presents with a painless lump on their arm, who’s tired and a bit pale & washed out. You might send a series of blood tests, including a full blood count to diagnose anaemia. You may also request an ultrasound of the lump, which may show an ugly mass with features consistent with sarcoma. Your friendly local plastic sarcoma surgeon might do a biopsy for you after an MRI, and the histopathologists confirm it’s a rhabdomyosarcoma.

All these tests are aimed at making a diagnosis: to clarify if the patient in front of us has, or does not have, the condition.

The oncologist who then takes up the patient’s care will move to undertake a series of further investigations; more…

StatsMiniBlog: ROC plots

19 May, 14 | by Bob Phillips

A (while ago) we published an explanatory page about ROC plots in the Education and Practice journal. There are a few great reasons why we should replicate it here:

1. So people can read it more easily

2. Because it fits into the stuttering series on diagnostics

3. It saves me having to write the same thing in different words.


Springing into action

2 Apr, 14 | by Bob Phillips

If you could get a multiplex PCR result back to you within 2 hours that told you your hot, grumpy, 2 month old patient did not have bacteraemia, would you discontinue antibiotics?

How sure would you need to be of that result – 95% certain? 98% certain? 99.5% certain?

What – in diagnostic analysis speak – would be your ‘threshold’?

The deciding of a threshold can be emotional, or rational, and both. The rational part of the idea relies on an equation:

Probability of wrong decision * consequence vs. probability of right decision * advantages

Rationality – based on a good understanding of some of the attributes of the diagnostic test in question – provides the probabilities. The ’emotional’ part comes in when assessing what weight the consequences and the advantages both have. (Now there are rational approaches to coming up with these – utility values they get called – but they’re still emotional at heart.)

Sometimes the consequence is so emotionally overwhelming that there is no degree of ‘chance’ that is allowed to be acceptable.

But what about you? What value would you place? If the test was 99% correct – would that be enough? Comments welcome.

  • Archi

(You may recall this is the same maths that runs the decision
in treatment too.)


Gambling, alcohol and division.

23 Mar, 14 | by Bob Phillips

No, not an average afternoon at the Houses of Parliament, but another in our diagnostics series.

Moving yourself from looking at the predictive values of the tests as evaluated, to taking this information but using it in the situation you face, is a case of Bayesian mathematics.

Which sounds hard.

But its absolutely what you do. Take a serum potassium of 7.8 in a baby who you struggled +++ to get some blood out of, every drop taking minutes to leak from its squealing heel. You may re-test, but you’re unlikely to call ICU on hearing it.

What if that same value came back from the central line sample from a child on peritoneal dialysis who presented with vague abdominal pain?

There is a different ‘pre-test probability’ – expectation – associated with different clinical settings. The same result has a different ‘meaning’ – chance of being real – in different settings. However, sensitivity and specificity don’t change lots across different clinically relevant pre-test probability values. And this leads to the ability to use them to move from a known/estimated pre-test probability of disease, to a post-test value. This needs the use of the likelihood ratio (LR) and, most often, a nomogram.


Positive about predictions

19 Mar, 14 | by Bob Phillips

In a previous post I muttered about how unhelpful sensitivity and specificity are to practicing clinicians, and how what we really want to know are the predictive values of a test.

Remembering the Table




Really diseased Really not diseased
Test +ve A B 1.. A/(A+B)
Test -ve C D 2.. C/(C+D)

5.. D/(C+D)

3… A/(A+C) 4… D/(B+D)


Sensitivity and specificity

9 Mar, 14 | by Bob Phillips


Sensitivity and specificity are those sorts of things that can really get knickers twisted up something rotten. They sound like something you should be able to understand, they get used as if you understand them, and then you realise … it’s not quite as you thought …


Really diseased 

Really not diseased 


Test +ve 

1.. A/(A+B) 

Test -ve 

2.. C/(C+D) 


3… A/(A+C) 

4… D/(B+D) 



Looking at the Table above – which of 1.. to 4.. is sensitivity?

And what does sensitivity tell us?


Diagnostic test accuracy

19 Feb, 14 | by Bob Phillips


The main things we look for when examining a new diagnostic test are “Is it as good as, or better than our usual one”, “Is it quicker?”, “Is it cheaper?” and “It is easier for patients/less dangerous?”

While the latter three questions can be assessed by asking the folk who do the test, asking the managers who pay for the test, and undertaking an adverse effects systematic review, it’s the first of these that we tend to call “diagnostic test accuracy”, and as clinicians we want to look for “phase III” studies.

The premise of such studies is that we can evaluate how accurate a test is by comparing its results with that of a ‘reference standard’ – a thing by which we will judge if the patient really does, or really doesn’t, have the diagnosis in question* – in a group of patients in whom we want to know the answer.


Differential Diagnosis

16 Feb, 14 | by Bob Phillips

The essential elements of a differential diagnosis study are, like most of critical appraisal, really simple and straightforward.

You need to start with a bunch of children/young people who turn up with the symptom, or symptom-complex, you’re interested in. Ideally, you need these folk to not already be known to have something, to attend a similar type of facility (e.g. office for general practitioners, or ED for ED types), and to be a consecutive group or random sample.

Testing. Diagnostic tests – why?

5 Feb, 14 | by Bob Phillips


“But what’s wrong with him, doctor?”

The constant refrain from many a consulting room is not “How can you make her better?” but “Can you name the problem so I can own and understand it?”. When addressing this need we will each develop our  own approaches; some of us will explore differentials, others state the top of our list, others delve into the concerns sitting behind the inquiry. When addressing the content – what’s actually wrong – we’ll probably want to come up with a diagnosis.


ADC blog homeapage

ADC Online

Education, debate, and meandering thoughts on child health, using evidence and research.Visit site

Creative Comms logo

Latest from Archives of Disease in Childhood

Latest from Archives of Disease in Childhood