The dynamics of patient visits to a public hospital ED: a statistical model

Using a public hospital's computerized database, we formulated a statistical model to explain emergency department (ED) patient volume for better staffing and resource allocation. All patients visiting the ED over a 3-year period were included in this retrospective study. Each observation described the total daily number of referrals and was defined by the following variables: day of the week, month of the year, holiday/ weekday, relative order in a 3-year sequence, and number of visits to the ED on that day. The statistical method used to build the model was analysis of covariance. Periodicity in average number of daily visits existed for each of the seasonal factors that were examined, repeating every year during the study period. Based on a graphic analysis, the model was defined and explained 65% of the variance during the 3-year study, with a relatively low standard deviation of error. A statistically significant correlation existed between time-related factors and the number of visits to the ED. This statistical model may prove to be of value for planning emergency services, which operate under stressful, unpredictable situations.