This week a study was reported about the change in belief about the best way to treat many breast cancer cases, a discovery that the long-time standard practice of surgical removal of the lymph nodes does not actually improve outcomes for such cases. There were discussions about whether the new recommendation implicit in that discovery should be adopted, whether the change would actually occur, and whether we should have actually changed our beliefs a few months ago when another study showed elements of the same result. There is also a bit of a mess about exactly which cases of breast cancer it applies to. Since any given study is limited in exactly what it can look at, there is a question of how far to extrapolate its results to similar cases that were not quite the kind studied.
What I found most interesting in all of this, which picks up on all the points in the previous paragraph, was how dumbfounded some commentators and consumers seemed about the fact that a common breast cancer surgery protocol was apparently wrong and how many unknowns remain. Most observers seem to think knowledge about optimal medical care is far better than it really is. Many readers have completely backlashed against news about nutrition and other public health information, reaching the frustrated conclusion that no one has any idea what is right. (This is not true, but I can understand the feeling – thus this series.) But they still think that medical decision making is far more sophisticated than it is.
I thought of a useful comparison. Consider something you are very familiar with and can judge, like the controls and options of your word processor, operating system, or some other piece of widely-used software. Every year there are major updates and you find yourself saying “why did they not figure out to do that before; it is not like it it required the latest graphics chip”. Yet we trust that there are not such “duh!” moments in medical decision making. But the person time devoted to improving the quality of one of those software tools dwarfs the amount devoted to improving our knowledge of the best way to do a particular medical treatment.
Yes there are thousands of highly-educated very well-paid people doing breast cancer surgery every day, and presumably they are trying to incrementally improve what they do, but they are like the users of a piece of software rather being researchers who are actively exploring possible major changes. Perhaps some of them are collecting data to assist non-incremental comparisons like the one that was just reported. But when you hear about these multi-million dollar medical studies, the impressive costs are just because everyone involved is highly paid and the medical activity is expensive. Ultimately the total amount of information generated by one of these impressively expensive studies is about the same as that from a twenty thousand dollar market research study that helps improve our software. There are a lot more untested standards about how to treat a particular disease – and a lot more market research studies – than there are medical studies
Every time we learn from one of these studies, it is great (so long as we learn the right lessons). But though cutting-edge medical machines put your personal computer to shame, the decision-making technology, e.g., about which surgical procedure to choose, is like running MS-DOS 3.1 and WordPerfect 4.2 – in other words something that is an unimaginable improvement over past decades, but not the marvel that what you might expect today. I am not holding my breath, however, for health news reports to bluntly say “yet another new study reveals that a million people have gotten the wrong treatment over the last ten years, even though they were told by guys in white coats that there was absolutely no doubt about what the right treatment was” and the in-depth analysis points out, “and ten years from now what we think is right now will seem just as quaint.”