Computerized approach to verifying study population data in occupational epidemiology.

Abstract

IN a previous paper a method was presented for verifying the completeness of occupational study populations independently of company-held records. The basis for the verification scheme was the Employers Quarterly Report on Earnings (Internal Revenue Service Form 941) that is submitted to the IRS each quarter of every year for all employees on a company payroll. In employing the verification scheme two approaches can be taken in determining how many and which quarterly reports are selected: (1) If the work force is large and/or many years are surveyed, a probability sample of years and quarters within years can be selected, allowing inferences to be made with respect to the entire study population. (2) If the number of survey years and/or the size of the work force is sufficiently small, copies of all quarterly returns for each survey year can be requested thus allowing a comprehensive check for completeness to be conducted. Te verification method described in the previous paper concentrated primarily on the first-mentioned approach. The second approach is herein being dealt with as well as a computerized method for verifying the completeness of study populations. In addition a new computerized method is described that utilizes the quarterly reports as an unbiased data base for checking the accuracy of certain work history data as developed from company records. An application of the computerized methods is given using data from a recent historical-prospective study of chemical production workers.