Although the use of clinical signs and symptoms is less likely than laboratory testing to identify women with cervical infections, it leads to treatment of more infected women because treatment can be provided immediately.¹ According to a study of female sex workers attending a clinic in Cotonou, Benin, researchers relying on clinical assessment correctly identified and treated 48% of women with gonococcal or chlamydial cervical infection. In contrast, use of locally available laboratory tests led to the correct identification of 75% of women with these infections. However, not all infected women returned to the clinic for their laboratory test results; consequently, only 29% of them received treatment in the laboratory-based approach.

To compare the performance of clinic- and laboratory-based methods of diagnosing and treating gonococcal or chlamydial cervical infection, researchers recruited 481 female sex workers attending a sexually transmitted infection (STI) clinic in Cotonou between November 1997 and October 1999. A physician administered a structured questionnaire to obtain demographic data, as well as information on the participants' sexual behavior and condom use. Participants then received HIV pretest counseling, education on STIs, and a gynecologic examination that included abdominal and pelvic palpation. The physician also took a venous blood sample for HIV testing and three cervical swabs, two of which were sent to a local laboratory to screen for Neisseria gonorrhoeae by culture and for Chlamydia trachomatis by enzyme immunoassay; a laboratory at the Institute of Tropical Medicine in Antwerp, Belgium, assayed the third sample for these two organisms using polymerase chain reaction (PCR) analysis, which was used as the "gold standard."

A clinical diagnosis of cervical infection was made if a woman had one or more of the following signs: mucopus, or yellow discharge or blood on a swab of endocervical secretions; cervical tenderness or lower abdominal pain on palpation (an indication of pelvic inflammatory disease); or cellular evidence of inflammation on a vaginal swab. Cases of cervical infection identified in this way were treated immediately according to national guidelines, and all women were asked to return within 10 days for laboratory test results and treatment, if needed.

As an alternative, the researchers devised a summary measure based on the duration of prostitution, the number of clients per working day, the presence of vaginal discharge, and any cervical sign of infection; scores above a specified cutoff value indicated a cervical infection. The performances of the two diagnostic methods and that of the summary measure were then evaluated against the PCR results.

The mean age of the participants was 29; 30% were younger than 25, 48% were aged 25-34 and the remainder were 35 or older. Roughly four in five were temporary immigrants from nearby African countries; one in five were from Benin. On average, the women reported 2.7 years of prostitution, during which they had seen four clients a day. About 14% reported no condom use in the previous week, whereas 40% said that they had used a condom with all clients. Roughly one-half were using antibiotics daily for STI prevention.

According to the PCR results, one in four participants had a gonococcal or chlamydial infection; four in 10 tested positive for HIV. Of the 118 women with cervical infections, 86% had gonorrhea and 21% had a chlamydial infection. Multivariate logistic regression showed that these infections were associated with having been a sex worker for less than two years, having at least four clients a day, observation of vaginal discharge during the gynecologic examination, any sign of cervical infection, cellular evidence of inflammation and HIV infection (odds ratios, 1.8-2.6).

Compared with PCR testing, use of clinical signs and symptoms correctly identified 48% of women who had a gonococcal or chlamydial infection and 75% of those who were not infected. Furthermore, only 38% of participants who had received a diagnosis of cervical infection based on clinical assessment actually had a cervical infection, and 82% of those with negative clinical results were actually disease free. When vaginal discharge was also included as a clinical sign of cervical infection, the accuracy of a positive clinical diagnosis rose to 66%, but the accuracy of negative findings dropped to 59%.

Compared with clinical assessment, the summary measure based on clinical and behavioral information correctly identified a higher proportion of infected women (64%) but a lower proportion of uninfected women (68%). The performance of the summary measure was not improved by including pelvic inflammatory disease or cellular evidence of inflammation.

The local laboratory tests were more accurate than either clinical diagnosis or the summary measure: They correctly identified 75% and nearly 100% of individuals with and without cervical infection, respectively. In addition, nearly all participants who had been identified as being infected on the basis of results from the local laboratory tests actually had a cervical infection (98%), and the majority of those with negative laboratory results were, in fact, disease free (92%).

Because women with a cervical infection had to return to the clinic within 10 days for their laboratory test results, the investigators assessed the usefulness of laboratory testing in this context by calculating the proportion of infected women who returned and received treatment within this period. Of the 118 infected women, 50 came back to the clinic, of whom 16 had been treated at their first visit on the basis of clinical assessment. Therefore, only 34 women who had positive results from the laboratory tests were treated (29%). However, clinical assessment and laboratory screening together resulted in treatment of 62% of infected women within 10 days.

The researchers comment that although laboratory analysis was more accurate in detecting infection than clinic-based screening, the practical difficulties and cost of laboratory testing in developing countries limit the usefulness of this method. Furthermore, after considering the follow-up rate among female sex workers, the analysts conclude that clinical assessment was superior to the local laboratory tests. The use of clinical signs and symptoms, they add, "still remains very useful and should be applied in these populations, even when valid laboratory tests are available." Alternatively, despite the likelihood of overtreatment, they suggest that "when [female sex workers] from a high-prevalence setting attend STI clinical services for the first time, presumptive treatment of cervical infection could be considered." --T. Lane

REFERENCE

1. Mukenge-Tshibaka L et al., Syndromic versus laboratory-based diagnosis of cervical infections among female sex workers in Benin: implications of nonattendance for return visits, Sexually Transmitted Diseases, 2002, 29(6):324-330.