Wednesday, December 29, 2010
The media tends to fill the days before the new year with reviews of the year past and predictions for the new one coming up. The idea of looking back to predict the future is part of routine decision making, relying on experience to predict what might happen next. Sports tries to teach us that statistics make a difference. Baseball managers, fantasy sports fans and bookies in Las Vegas use what happened yesterday to predict tomorrow. But Wall Street advertisers use a disclaimer: past performance does not guarantee future results.
Decision making in medicine has tried to evolve from art to science with increasing amounts of available technology. Making the diagnosis of an illness or injury and considering the potential complications at the bedside still requires skill, experience and judgment but high tech has wedged its way into the equation. A generation ago, a baseball manager would use skill and experience to decide whether a new pitcher was needed or if a batter should bunt or swing away. Football plays would be called based on the ebb and flow of the game. Now, a statistician can produce numbers about left handed batters versus right handed pitchers with two out. Experience and intuition have been joined by a new partner
Statistics have similarly affected decision making in medicine. That decision making has at least two steps at the bedside. Step one is the diagnosis and step two is the treatment. Each has its own challenges to get it right.
Appendicitis is an interesting example. In the BCT era (before CT scans), appendicitis was a diagnosis made by history and physical examination. Blood tests might give some direction but the surgeon had to make a clinical decision at the bedside whether the patient needed an operation. The risk of major complications increased dramatically if an inflamed appendix was allowed to rupture so a 20 percent error rate was acceptable, meaning that 1 out 5 operations took out a normal appendix. The cause of the pain might have been diverticulitis, an inflamed ovary, swollen lymph glands or a mystery. But before CT, the only way to look into the belly was with a scalpel.
CT scans for appendicitis have their own challenges. It takes about 12 hours from onset of symptoms for inflammation to appear on a CT scan. There is significant radiation and there is a whole new skill set needed by a radiologist to interpret the images. The “error” rate of taking out normal appendixes has decreased and the individual benefits but a new generation of surgeons may feel obligated to use technology to confirm their bedside skill and decision before taking the non-reversible step of operating.
Statistics also matter for treatment recommendations and decisions. Statistics are used routinely by physicians in choosing the antibiotic most likely to combat an infection or the type of pain medication that might control symptoms. But statistics can cause decisions too hard to make. Patients in the midst of a stroke can have the clot that decreases blood flow to their brain dissolved with thrombolytic therapy, a clot busting drug . If TPA is injected within the appropriate time frame (the earlier the better but no later than 4 ½ hours for most strokes but up to 6 if injected directly into the clot) stroke symptoms can be reversed 1/3 of the time. The downside risk is that bleeding into the brain can occur in 6% of patients and the results are not reversible and can be life threatening and devastating. The sooner the drug is given the less risk of bleeding. Patients and family have to decide if it’s worth the risk. The clock is ticking.
Family and patient often turn to the doctor, looking not only for guidance but for the right answer when there isn’t one available. The risks and responsibility are high and the consequences can be catastrophic. It’s not like letting a pitcher face another batter; there may not be another inning.