The last time I posted I wrote about the frequency and danger of diagnostic errors. How do we make them less frequent? The information on how to do that is sparse and definitive solutions are lacking. However, several groups of doctors are involved in active quests for answers. To see what they are up against, let us look at how doctors first learn to make diagnoses.
After gorging their brains with anatomy, physiology, pathology and biochemistry for months on end, medical students are finally allowed onto a hospital ward to examine their very first patient. This can be a scary event for many of them and for which they have taken many days of preparation. The first and diagnostically most important part of the exam is the history -- the story the patient tells. This must be obtained with as little prodding as necessary so as to not "lead the witness". "I noticed this lump in my breast", "I began to vomit up blood" ,"I noticed that my eyes and skin had turned yellow" and "I had the worst headache of my life all of a sudden" are typical worrying chief complaints.
Beginning with the patient's chief complaint, the doctor must listen creatively, make up hypotheses about what is going on and think of multiple diagnoses based on what the patient is saying and ask questions which separate one possible diagnosis from another. A doctor must be ready to ask the patient those questions that can either seal a diagnosis on the one hand or force him or her to consider a whole spectrum of disorders that can produce identical symptoms on the other. Making up a rolling list of possibilities as one draws out the patient's story is a purely mental process. This takes time, patience, skill, knowledge and experience to perfect. The physical exam that follows usually confirms what the doctor suspected from taking the history. At other times, however, it can yield a key finding not suggested by interviewing the patient.
With the history and physical completed, the doctor, neophyte or veteran, must put everything found at the bedside together and state formally in writing a differential diagnosis, a reasonable list of possible ailments to explain the findings. This always seemed obligatory to me until I read an excellent paper from the University of Iowa about missed diagnoses by veteran primary physicians. With the benefit of hindsight, the majority of the doctors who missed diagnoses admitted that they had failed to make up an adequate differential diagnosis. As many clinicians do in their busy practices, they had used a common method in which they had stopped thinking after arriving at the diagnosis that came most forcefully to mind. Old hands in medicine are accurate with this approach quite often because they have seen so much. The name for it is "pattern recognition". Pattern recognition takes less time than formulating and writing out a differential diagnosis. It is not an entirely bad way to diagnose disease and it must suffice in dire emergencies where there is no time to lose.
Scientists who study how best to reach the correct diagnosis have found that when the two processes, formulating a differential diagnosis and pattern recognition are combined, the rate of getting the correct diagnosis is better than when either is used alone. Another way of avoiding missing a diagnosis is repeating the examination at a later date if time permits and getting a colleague to do a consultation to check if something was missed or not thought of. There is no shame in either approach and I have seen the best doctors of my acquaintance get help with puzzling cases.
Researchers in the field of diagnostic errors do not put much faith in check lists or computer programs designed to make diagnoses. They feel that the field is wide open and must be rethought. We are dealing with the intellectual functioning of experts on the highest plane of their proficiency so that the going may well be arduous and slow in finding the solutions.