[Estimating the denominator of a sentinel practice using accounting data of the federal physicians association]

In order to interpret the regional and temporal variation in the number of cases reported for sentinel topics, rates have to be calculated which relate these figures to a common denominator. For this denominator different definitions are possible, where this paper focuses on the contact denominator, counting all contacts between the patient and the doctor in a specified period of time and a specified population (i.e. children between 2 and 3 years of age by sex for the first MORBUS topic). One patient may produce different contacts. Due to work load considerations contact denominators could not be directly recorded in sentinel practices in the MORBUS-project. A method was developed which combined minimal denominator information obtained from sentinel practices with routine (accounting) data from the doctor's associations, in order to estimate denominators with a high degree of precision. Data elements in the accounting data base include number of personal patient-doctor-contacts by type of treatment (screening/cure/emergency/referral) and insurance status (ordinary member, family member, retired) itemized list of actions that were reimbursed through the insurance company structural data about the doctor/practice (specialty, accreditation date) etc. For a first assessment of the degree of accuracy with which denominators can be estimated by such data, a pilot test was scheduled for April 1991. Doctors recorded the number of contacts with children between two and three years of age (by sex) for one week on a simple checklist. A linear model which included specialty, region, number of vouchers per quarter, number of infant health assessments and date of accreditation as predictors, was fitted to the data.(ABSTRACT TRUNCATED AT 250 WORDS)