Evaluation of the National Cervical Screening Program by calculation of the positive predictive value and standardized mortality rate

Abstract

Cervical cancer is the 3rd commonest cancer occurring among women worldwide, behind breast and colorectal cancer. (Parkin et al. 1999) Screening via the use of cervical smears to detect preclinical disease has been successful in many countries in reducing mortality and incidence due to this malignancy. The purpose of this dissertation is to evaluate aspects of the New Zealand cervical screening program by: • Comparing trends in cervical cancer mortality before and after the introduction of screening by direct standardization of age-specific mortality rates • Calculating the positive predictive value of cervical cytology

A centrally organized national program was launched in 1990 but prior to this screening activity has resulted in reductions in mortality. Direct standardization of mortality rates enables comparison of mortality across different time periods. A 50% reduction in mortality has taken place for women in the 1994-1996 period compared to 1954-1958.

The use of the smear is not without some degree of inaccuracy. Errors occur as a result of human error and the imperfection of this instrument to adequately obtain specimens. Analysis of smear results using the national screening register correlating a cytological diagnosis with histology, proves this to be the case. However the absence oflarge amounts of histology give a biased estimate: the predictive value positive for high-grade lesions was 61.8-78%; and for low-grade lesions this was 25.3-62.1 %.

The pathway to improvement is unlikely to lie in newer technology. Greater care in obtaining samples and reading slides will produce gains in efficiency in a service that is already performing well.