The problem is how to come up with a complete and accurate diagnosis in a very limited amount of time. Early in training, this is less of an issue, when you are encouraged to spend what you will later consider to be inordinate amounts of time interviewing your patients. But after training, you will quickly realize that there is a correlation between the number of patients that you see per day and your ability to afford a mortgage on that new home. You will be torn between the need to do things quickly and the need to do things right.

The way things are done in most busy community clinics is probably not so “right.” One recent study compared “routine diagnoses” as found in the medical chart with a “gold standard diagnosis” generated using the SCID [Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition (DSM-III-R)] plus chart review as well as an additional interview with a highly qualified psychiatrist or psychologist. There was only about a 50% rate of agreement between routine and gold standard diagnosis, and in one-half of all cases of disagreement, feedback to the original clinicians resulted in significant changes in patient care.

Does this mean that you should give the SCID to all of your patients before the interview? Thankfully not, because the techniques discussed in this section, involving screening and probing questions, mirror the SCID gold standard, adapting it to the realities of clinical practice.

One might assume that the best way to reach a diagnosis is to follow a two-step process:

1. Obtain all potentially relevant data about the patient.

2. Examine the data to determine which diagnosis fits best.

This strategy would work well if time were limitless. Because it isn’t, clinicians have developed ways of determining in advance what is likely to be relevant data for a particular patient, thereby vastly increasing the efficiency of the diagnostic interview.

How do expert clinicians make diagnoses? A number of researchers have done observational studies to answer this question (Elstein et al. 1978; Kaplan 1995). They have found that experienced clinicians begin by carefully listening to the patient’s initial complaint and asking open-ended questions. Based on this preliminary information, they generate a limited number of diagnostic hypotheses (the average being four) early in the interview, usually within the first 5 minutes. They then ask a number of closed-ended questions to test whether each hypothesis is true. This process is known as pattern-matching, in which the patient’s pattern of symptoms is compared with the symptom pattern required for a diagnosis.

Another way to view this approach is to think of a “closed cone” of questions (Lipkin 1987). The initial questions are open ended and exploratory; they become more closed ended to pursue a specific diagnosis to an endpoint of verification or exclusion.