A simple program to create exact person-time data in cohort analyses.
Wood-J; Richardson-D; Wing-S
Int J Epidemiol 1997 Apr; 26(2):395-399
A simple computer program for creating exact person/time data for use in cohort studies was described. The program was developed to create stratified person/time tables by counting each day from the date of first observation to the date of last observation for each person in the cohort. After the last day of observation, the vital status of each subject was determined. If the subject had died, a death was recorded and classified by the same levels of the variables recorded at the last day of observation. By this procedure, a temporary multidimensional array in which followup of each subject was counted could be created. Since this was actually a temporary array, a large number of cells could be included in a quickly constructed table. Within this array, an ongoing count of the number of person/days of observation could be kept, each being classified for followup according to the variables of interest. The program could categorize followup time and events by any desired unit of person/time. The program requires only a simple data file such as date of first observation, date of last observation, and an indicator of outcome status for each subject to create a simple person/time table. Sample SAS programs for calculating person/time data for crude outcome rates and for creating person/year tables stratified by age, race, gender, birth cohort, cumulative dose, and calendar year were presented. Sample SAS and BASIC programs for calculating data needed computing crude outcome rates were also presented.
NIOSH-Publication; NIOSH-Grant; Pulmonary-system-disorders; Epidemiology; Computer-models; Statistical-analysis; Mathematical-models; Occupational-medicine
Epidemiology University of North Carolina Cb#7400 Mcgavran-Greenberg HAl Chapel Hill, NC 27599-7400
International Journal of Epidemiology
University of North Carolina Chapel Hill, Chapel Hill, North Carolina