Monday, 23 March 2015

How to diagnose overdiagnosis

In a BMJ editorial, within the framework of the campaign "Too much medicine," Professor Paul Glasziou and editor Fiona Godlee, among others, spoke of the phenomenon of overdiagnosis and analysing their characteristics, and since this is a trending concept, I think it's a good idea to pay attention to it.

In a medicine previous to cancer screening programs as we know them today, people were diagnosed according to the appearance of the first symptoms. It was the visible tip of the iceberg. But as it was suspected that when the symptoms appeared, the illness simmered below the clinical perception for long time, it was believed it was necessary to develop diagnostic techniques to allow seeing the iceberg below the water, with the assumption that the sooner the pathological process in motion is detected the sooner is possible to act, therefore obtaining a better prognosis.

But now we can see that things did not fully go as intended. In general, we can affirm that thanks to screening programs forecasts have improved, but unfortunately the mortality rates are the same, and voices criticizing this rummaging through the invisible parts of the iceberg, have enough arguments to defend that while early diagnosis may detect many cases that perhaps wouldn’t have emerged at all, this is done at the expense of improving survival rates in an artificial way.

According to the published reviews, there are studies that show that in Australia, thyroid cancer has tripled, while mortality remains unchanged and similar data is seen with pulmonary embolism which, after improving diagnostic technique -the CT angiography- it has been seen in the US that the number of cases has increased by 80%, this improvement in diagnostic accuracy is failing to cause a decrease in mortality. In other pathologies, diagnostic techniques have not been innovated, but instead the diagnostic threshold has been reduced, as it happens in hypertension, diabetes, osteoporosis, obesity and cognitive disabilities. These small changes in diagnostic criteria have caused an increase in the number of people affected, and therefore therapeutic consumers, but in exchange this has not presumed to obtain the desired improvements in clinical outcomes.

I think that from this BMJ editorial, physicians with clinical practice will find most useful the following tables showing how you can diagnose an "outbreak" of overdiagnosis:

Signs of over diagnosis (alarms):
  • A technique that increases the number of cases, but no reductions in mortality is observed.
  • A biomarker or a risk factor is labelled to sound like a disease.
  • There are changes in the definition of a disease, or reduced levels of evidence but no evidence that the benefits of new developments are larger than the negative effects.
Given the suspicions of overdiagnosis, what questions should be posed?
  • Are we measuring a risk factor or a symptom?
  • Are the labels making the distinction clear?
  • Who has defined the thresholds?
  • On what evidence have they relied?
  • Will the emergence of a new test detect more cases?
  • Do we know the natural course that the new cases would follow, if they wouldn’t have been detected?

I think it's time that when we speak of overdiagnosis we do so properly and for this reason, BMJ has opened a series within the campaign "Too much medicine" to review the main problems that this epidemic is bringing to everyday medicine.

Jordi Varela
Editor 

No comments:

Post a Comment