Objective: Our goal was to develop reliable data on the probability of specific diagnoses among patients of family physicians (FPs) presenting with common symptoms.
Study design: A group of 54 Dutch FPs recorded the reasons for encounter, diagnoses, and interventions for all episodes of care between 1985 and 1995. Diagnoses could be modified during the episode of care, and a modified diagnosis was applied to all episode data.
Population: We used the listed patient populations of the 54 Dutch FPs, representing 93,297 patient years, 236,027 episodes of care, and 267,897 patient encounters.
Outcomes measured: The top 20 diagnoses related to 4 selected presenting symptoms (cough, shortness of breath, general weakness/tiredness, and low back symptom/complaints without radiation), per 100 patients, with 95% confidence intervals, stratified by age. In the standard tables, age-specific cells with fewer than 10 observations were excluded.
Results: The availability of an accurate estimate of prior (pretest) probabilities for common symptoms/complaints has great potential value for family practice as an academic discipline and for family physicians in that it can support their medical decision making. Stratifying data by age groups increases the clinical relevance of the prior probabilities.
Conclusions: Though collected by Dutch FPs, the data in our study have a high face validity for other clinicians. Still, FPs in other countries should give priority to collecting their own probability databases.