Literature review > Issue_2 > Review Marrazzo et al. 

Selective screening of women for chlamydial and gonococcal genitourinary infections has been widely debated as a less expensive alternative to universal screening. Numerous studies over the past two decades have examined markers for gonococcal or chlamydial cervicitis [1]. Of these, age and cervical findings have been among the most extensively evaluated. The literature generally concurs that women less than 25 years of age are at greatest risk and the Centers for Disease Control and Prevention, in their 1993 guidelines, recommend as equivalent alternatives a risk-based screening approach and universal screening for women aged <30 [2].

Cost-effectiveness studies have concluded that universal screening for Chlamydia trachomatis is more cost-beneficial than selective screening in populations of women with a disease prevalence of 3-6% [3]. These models are affected by the sensitivity and cost of the screening assay, a point highlighted by Tao et al. in an analysis comparing universal and selective screening using various tests within U.S. family planning clinics [4]. They concluded that universal screening by DNA probe was the optimal cost-effective strategy when budgets were limited to $6 per visit, although using this strategy, only 61% of women would be cured. On an unlimited budget, they showed that universal screening of all women by DNA amplification testing would optimize the effectiveness of screening and would be expected to result in 89% cure rates. In comparing universal to selective screening, testing women under 25 years of age was less costly but would have missed 30% of infections.

Marrazzo et al. sought to determine the accuracy of cervical signs including induced bleeding, mucopus and leukorrhea (excessive numbers of leukocytes in endocervical secretions) in predicting gonococcal/chlamydial cervicitis among women in various age groups. In the overall population of women presenting to an STD clinic in Seattle, WA, 6.9% had an infection with C. trachomatis and 2.1% were infected by N. gonorrhoeae. The approach by Marrazzo et al. extends previous work by evaluating a combination of factors (age and cervical signs) in selective screening and was prompted, in part, by the recommendation by the Centers for Disease Control that the finding of mucopurulent cervicitis be used as a trigger to test for N. gonorrhoeae and C. trachomatis. They postulated and found that the positive predictive value for cervical signs would vary by age. This is not surprising since positive predictive value is influenced by prevalence and the prevalence of gonococcal/chlamydial cervicitis was predictably greater in younger women. However, even among quite young women, infection was present among less than half of women with any cervical sign. That is, having a cervical sign was no guarantee of having an infection.

Furthermore, the data presented in the Marrazzo et al. study demonstrate the well-known trade-off between positive predictive value (PPV) and negative predictive value (NPV). Using a given test, the downside to a higher PPV is a lower NPV. My own calculation from their data reveals a NPV for having any cervical sign among less than 19 year olds of 85%. That is, in 15% of less than 19 year old women, an infection was present despite cervical signs being absent. In contrast, among 25-29 year olds, wherein the PPV for any cervical signs was only 15% (vs. 41% in less than 19 year olds), the NPV was 95%. That is, only 5% of infections would be missed among women with cervical signs in the age group.

One caveat in generalizing the results of this study is that design elements might have resulted in an underestimate of the positive predictive value. The gold standard used for diagnosis of N. gonorrhoeae or C. trachomatis was culture, rather than a more sensitive nucleic acid test. Furthermore, the population prevalence will affect PPV and NPV values (as noted above). Finally, in resource poor settings, selective screening, versus universal screening, may be warranted, despite the inherent under-identification of infected individuals

On the other hand, the Marrazzo et al. study does not provide a convincing argument for a more complex approach to selective screening. Their conclusion, that cervical signs are more predictive among younger women, is well-supported by their data. However, it is not clear that the use of a syndromic criterion in addition to an age criterion is warranted in resource restricted settings, as compared to an age criterion alone. Furthermore, in resource-rich settings, the bulk of the literature supports universal screening for populations or sub-populations (eg. younger) of women in whom the prevalence of infection exceeds 2-3%, since, with selective screening, some infections will be missed

about SDI | newsletters | grants | publications | literature reviews

WHO Home - WHO Search - TDR Home - SDI Home - SDI Contact us
(c) WHO/OMS 2001