Friday, November 06, 2009

The Needs Of The Few vs The Cost To The Many

One of the problems with trying to adopt an evidence based approach, to anything, but in particular, to medical testing and treatment, is that population studies do not tell individual stories.

The problem with using individual stories is that the plural of anecdote is not data, and association does not prove causation.


A young middle-aged woman with chest pains present to the ED. They had woken her from sleep, but been transient, and she had gone back to sleep. In the morning she had something of a dull ache behind her breastbone, and felt a little short of breath. Simple remedies had not helped. Her pain had settled at time of exam, and she examined normally. An ECG was normal.

Past history, family history, all negative.

How to proceed?

She was admitted for observation and serial cardiac enzymes. Should she have had something else? A CT scan? CT coronary angiography? CT triple testing, looking at aortic root, pulmonary vessels and coronaries?

While waiting the result of her serial enzymes, she collapsed on the way to the toilet. Immediately after the collapse, she was alert, if slightly clammy, and her pain had returned. ECG now showed some T wave inversion, laterally, in keeping with an ischaemic picture.

Before she could be further assessed, her pain became excruciating, and she developed marked cyanosis, centrally. ECG continued to show ischaemic changes, but had not changed from the immediate post-collapse trace.

Within about 15 minutes, she became profoundly bradycardic, and then arrested. Prolonged resuscitation was, ultimately, unsuccessful.

I don't know what the PM showed.


If she had been investigated more invasively, might the outcome have been different? Maybe... maybe not. But most folks who present as she did, DON'T keel over. So if we scan everyone, won't most of them just have a normal scan, and increased radiation load?

So not scanning makes sense for the population in general... just not for her, I guess.

No comments: