Risk targeting in cervical screening: a new look at an old problem

Abstract

In the face of continuing debate about the level of effectiveness of the United Kingdom cervical cytology screening programme in preventing cervical cancer, more precise targeting of high risk groups might offer a means of enhancing its efficiency. Broad risk targeting is already practised by screening only sexually active women aged 20 to 65 years. This paper describes a risk scoring system constructed from the available literature and designed to be used by primary care health professionals and patients. The system involves four independent risk factors: educational level, current smoking habit, years of oral contraceptive use and number of sexual partners. Since the objective is simply to identify women at relatively high risk, inclusion of a factor neither requires nor implies causality. The next steps are to study the feasibility of putting the scale to practical use and to investigate its predictive value in a prospective evaluation.

PIP: This paper proposes an individual risk scoring system to target cervical screening beyond the broad targeting already used in Britain, that is, screening only sexually active women aged 20-65 years. It may be practical to introduce risk-related screening schedules, with intervals graduated by risk status. 5 large well-controlled studies were selected from the literature to construct a risk-scoring system. From these studies several significant risk factors emerged: 5-7 or more years of oral contraceptive use, current smoking, number of sexual partners, pregnancy outside of marriage, and age at first intercourse. In all 5 studies current smoking and number of sexual partners were identified as risk factors. The 3 North American studies found educational level to be an independent risk factor. Also the woman's age and the time interval since her last smear were significant. 4 independent risk factors were chosen as best describing risk of cervical intraepithelial neoplasia or cervical cancer: woman's educational level, current smoking, 5 or more years of oral contraceptive use, and lifetime number of sexual partners. These risk factors are not intended to imply causality or a formal statistical meta-analysis. This risk scoring tool is being tested in a pilot study for feasibility.