Defining frequent attendance in general practice

Background: General practitioners (GPs) or researchers sometimes need to identify frequent attenders (FAs) in order to screen them for unidentified problems and to test specific interventions. We wanted to assess different methods for selecting FAs to identify the most feasible and effective one for use in a general (group) practice. Methods: In the second Dutch National Survey of General Practice, data were collected on 375 899 persons registered with 104 practices. Frequent attendance is defined as the top 3% and 10% of enlisted patients in each one-year age-sex group measured during the study year. We used these two selections as our reference standard. We also selected the top 3% and 10% FAs (90 and 97 percentile) based on four selection methods of diminishing preciseness. We compared the test characteristics of these four methods. Results: Of all enlisted patients, 24 % did not consult the practice during the study year. The mean number of contacts in the top 10% FAs increased in men from 5.8 (age 15-24 years) to 17.5 (age 64-75 years) and in women from 9.7 to 19.8. In the top 3% of FAs, contacts increased in men from 9.2 to 24.5 and in women from 14 to 27.8. The selection of FAs becomes more precise when smaller age classes are used. All selection methods show acceptable results (kappa 0.849 - 0.942) except the three group method. Conclusion: To correctly identify frequent attenders in general practice, we recommend dividing patients into at least three age groups per sex