PROSPECTIVE EVALUATION OF A RISK SCORING SYSTEM FOR CERVICAL NEOPLASIA IN PRIMARY-CARE

Background. Against a background of concern over the costs of the cerv ical screening programme in the United Kingdom, increased precision in targeting groups at high risk of having an abnormal cervical smear of fers a means of increasing efficiency. Previous papers have described the development of a risk scoring system and its feasibility and relia bility in primary care. Aim. A study was carried out to assess the val idity of the scoring system by testing its predictive ability on a pro spective data set. Method. Consecutive attenders for cervical smear te sts at seven practices and three clinics were recruited for the study. The women completed a questionnaire from which their risk scores coul d be calculated The scores were compared with cytology and histology r esults. Various performance statistics were obtained. Results. In term s of cervical intraepithelial neoplasia (CIN) 2 or 3, there was an 11 fold increased risk among the low risk group (scores of four or five) compared with the very low risk group (scores of three or less). The s ystem enabled the identification of 75% (95% confidence interval 62% t o 84%) of cases of CIN 2 or 3 among the 21% of the 3629 women with kno wn histology who had a score of four or five. Conclusion. Given the ea se with which risk status can be ascertained (a risk score could not b e calculated for only 23 of 3661 women) and the magnitude of differenc e in risk, the risk scoring system appears to have potential for assis ting the targeting of screening resources. Studies of risk perception and behaviour, and ultimately a randomized controlled trial, are requi red to assess the effectiveness and cost effectiveness of risk targeti ng.