Following the evidence base is excellent – or is it?

I have started my section on EBM, as it is commonly known, with an example of decision-making by management based on no evidence – or, to be fair, bits of evidence that do not add up (this is the bare below the elbows issue). However there are numerous examples of clinical trials which have produced good evidence that cannot be relied upon.[1]

Three examples in my own specialty come to mind. In one case a lack of anatomical knowledge prejudiced the outcome (also true of the second) and in the third a failure to understand exponential data meant that like was not compared with like.

Bear these examples in mind when you look at any clinical trial.

Do you know your anatomy (1)

Colleagues in a neighbouring Trust performed a sequential audit on the benefit of steroid injection for frozen shoulder – technically known as capsulitis, and basically a severe inflammation of the joint between long arm bone (humerus) and shoulder blade (scapula) – the glenohumeral joint. The injections were administered by a single practitioner, and the results suggested little or no benefit. The injections had been given by the lateral approach, which enters another part of the shoulder mechanism – the joint under the tip of the shoulder blade (subacromial joint). This and the glenohumeral joint are quite separate unless the main shoulder muscle, the rotator cuff, has torn – an anatomical fact not appreciated by the study sponsors, probably because rheumatologists don’t read orthopaedic textbooks and never sat in orthopaedic clinics. So, one can reasonably suppose that injection of one joint will only rarely have any effect on the other, and that as the wrong joint had been injected the study was useless.

This failure led to some debate. It became clear that there was disagreement on the management of shoulder problems that extended further to the actual accurate assessment of them. We set up a little study which showed that even experts did not agree on a diagnosis even when they saw the patients together[2]. There are a number of published trials which have fallen into the same trap – but are still adduced as evidence when people attempt to assess the efficacy of joint injections.  The first thing you must do is ensure that you are actually treating the part that need treating.

Do you know your anatomy (2)

People are still treating sacroiliac joint strain. This is diagnosed by finding – pain over the sacroiliac joints. Injection treatment relieves symptoms. QED.

I am not so sure. A registrar in our department, when I was a senior registrar, came to me asking my views on this syndrome. He could not see no logic in it. Inflammation and infection – yes; you can see changes on X-ray and scans and intrinsically you know that pain could result. But how do you strain a rigid joint – so stable that if you are in a dreadful car accident the pelvic bones are more likely to fracture than the SIJ disrupt? Also, despite the book descriptions of how to inject the joint, he was not clear, anatomically, how it was possible to get a needle into it. So if injections seemed to work, how could this be?

So he set up a little experiment. He decided to do a traceable injection. Taking a series of corpses (necessary permissions were obtained) he performed injections by marking out the surface anatomy as per the books, and then injected under X-ray control using Indian ink. Then he dissected down to see where the ink was.

He did not get into a single sacroiliac joint. The ink was everywhere but. Interestingly some got into the venous plexuses and spread up in the vertebral veins but most spread out over the surface of the pelvis under the attachments of the gluteal muscles.

Could it be, therefore, that sacroiliac region pain is actually a gluteal strain syndrome? This little piece of evidence suggests so. If pain is in a particular area that has more than one anatomical structure, you should beware of deciding it comes from one and not another.

You may not be clear if a drug is working…

 …but if you stop it, and everything goes haywire, then you can be sure that it was. This is sometimes the only way to persuade a patient to stay on their pills.

The converse applies. If a patient appears to have side-effects, and you stop it, and everything returns to normal, you can be fairly sure of cause and effect – and make certain by a re-challenge that produces the same problem.  I did this with statins.  More fool I.

You may not be clear if a drug is working (2)…

…and you rely on the evidence as is. However the evidence can change. My biochemistry professor used to say that medical knowledge had a half-life of seven years, so by that time half of what we had been taught might be wrong. Only we don’t know which half.

There are dozens of examples of this with drugs. Take bisphosphonates and osteoporosis. There is no doubt that you can prove that their administration may stop the progression of bone loss. However as time goes by other things surface; thus there has been a significant scare over osteonecrosis of the jaw, and undoubtedly there a many patients who have significant side-effects, either with dreadful indigestion or worse (hence the instruction to wash it down with lots of fluid and then stay upright for half an hour) or acute allergic reactions to the infusions designed to get round this. Just recently reports have appeared of an increased risk of unusual site hip fractures in people on bisphosphonates.[3] So the drug you give causes the very thing you are trying to stop. Not good.  By the same token the elderly patient (more prone to fracture) has been put on blood pressure pills.  So when he or she stands up suddenly they pass out and fall over.  So before starting, ask why (actually with the blood pressure thing you may be risking trouble, as narrower vessels need a higher driving pressure, so you might be at greater risk of a stroke, or coronary, if the pressure drops too far.).  The owner of our local dry cleaners worked that one out so it’s odd that doctors cannot.

[1] For more examples, see Malcolm Kendrick’s “Doctoring data”

[2] Bamji AN, Erhardt CC, Price RP & Williams P. The Painful Shoulder. Can Consultants agree? Brit J. Rheumatol 1996; 35: 1172-74

[3] Agarwal S, Agarwal S, Gupta P, Agarwal PK, Agarwal G, Bansal A. Risk of atypical femoral fracture with long-term use of alendronate (bisphosphonates) : a systemic review of literature. Acta Orthopaedica Belgica 2010; 76 (5): 567-71


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s