Careful How You Choose Your Doctors

A patient who seeks a recommendation from a friend, from a colleague at work, or from a doctor may not always be well-served. Consider first the friend or work colleague who recommends her own doctor. Such patients are seldom in a position to judge the quality of care delivered by their doctor. What can be said is that patients who are likely to recommend their doctor to others are those who have recovered and have not been harmed; they also like their doctor's professional manner. However, if a doctor harmed 10 percent of his patients, he could still have a large and growing practice. The patients he harmed would change doctors and would be recommending their new doctors. But the remaining 90 percent would continue to recommend the poorly performing doctor. For a person seeking a recommendation, those are not good odds.

Many people get their recommendations from another doctor, usually one in whom they have a lot of trust. And surveys show that just about everybody has a high degree of trust in his or her own doctor. But this selection method is hardly foolproof either. Health plan administrators, with insider knowledge of the better doctors, have informed us of irregularities in doctor referral patterns. Unfortunately, while the best doctors generally received more referrals than poorer doctors, their share of the total was not as large as one would have expected. Conversely, less-capable doctors received more referrals than warranted by their past results.

To understand why patients may not always be referred to the most qualified doctors, we need only consider referral incentives. First, there is friendship: a doctor may refer his patient to a close friend or colleague, even though that friend or colleague is not the best doctor he could have recommended. Fortunately, such a referral is more likely if the patient's condition is not serious, and the referring doctor is confident that his friend will perform adequately. However, if the patient's condition worsens, or if the patient subsequently develops a more serious ailment, then the referring doctor may have placed his patient at unnecessary risk.

Another influential factor is the desire to receive referrals from other doctors. Doctors welcome new referrals. They mean an increase in business and income, and they confer a degree of professional status. However, referrals often constitute a reciprocal relationship: "I'll send you referrals; but I expect you to send me referrals as well." From professional considerations alone, the best doctors will receive the most referrals. Yet, the more referrals they receive, the less able will they be to reciprocate; they will be especially disinclined to reciprocate to a mediocre doctor. Therefore, if a doctor wishes to use his own referrals as currency for receiving referrals, he may avoid referring his patients to the best doctors.

The Referral Satisfaction Index (RSI) is designed to overcome these perverse incentives by reporting the quality of each Diagnostician's referrals. The RSI is constructed from two components: 1) the treatment value that doctors recommended by the Diagnostician have provided to their patients, and 2) the satisfaction that the Diagnostician's patients have had with his previous recommendations. The first component motivates the Diagnostician to recommend a doctor who has given high value to his patients in the past. The second motivates the Diagnostician to select a doctor who is likely to give the patient a great deal of satisfaction.