APPROACHES TO QUALITY-CONTROL WITH AN APPLICATION TO A NEW CANCER REGISTRY IN A DEVELOPING-COUNTRY

Quality control should be maintained in all disease registries by appr opriate statistical procedures to assure complete, accurate, reliable, and timely morbidity and mortality data. Shewhart control charts are recommended for evaluating item completeness. Accuracy and reliability can be evaluated with stringent diagnostic criteria, re-abstraction o f records, and statistical measures such as the kappa statistic. Discr epancies between observed and expected data in various time periods ca n be used to evaluate whether data entering a registry meet timeliness criteria. The reporting delay distribution can be used to examine tim eliness of reported data and to adjust incidence rates for potential l ack of timeliness in reporting. An application of these quality contro l procedures is demonstrated using two variables selected from a hospi tal-based study consisting of 119 breast cancer records collected from the Division of Medical Records, Dr Soetomo Hospital, Surabaya, Indon esia. The results indicate that the variable LASTNAME met the selected criteria for completeness and reliability. The variable ADDRESS met t he reliability but not the completeness criterion. All reported cases met the timeliness criterion. The demonstration confirmed the need for registries to establish formal quality control procedures.