Current strategies to diagnose reproductive tract infections (RTIs) in resource-poor settings by relying on presumptive diagnoses may be leading to considerable overdiagnosis and overtreatment. According to a study of 600 clinic users in Hue, Vietnam, the percentage of women with any RTI based on clinicians' presumptive diagnosis reached 62%, whereas only 21% had an infection that was etiologically confirmed.1 Five percent of the women had a confirmed sexually transmitted infection (STI), and 1% a cervical infection. Applying the syndromic algorithm developed by the World Health Organization (WHO) may reduce the amount of potential overtreatment for vaginal infections, but the algorithm would need to be adapted to local epidemiologic conditions to increase the accuracy of diagnoses of cervical infection.
The data for the analysis come from a survey conducted among clients of the Maternal- Child Health and Family Planning (MCH-FP) Center in Hue, Vietnam. All nondelivery clients were eligible to participate and a total of 600 visitors to the clinic (e.g., the first five such clients of the day) from May through October 1996 were enrolled. Participants responded to a structured questionnaire that asked about their personal and medical histories. For the RTI diagnoses, clients underwent a thorough pelvic examination, during which vaginal and cervical specimens were taken. When possible, these specimens were tested not only at the MCH-FP Center (using simple techniques and materials appropriate for resource-poor settings) but also at the nearby Hue Central Hospital, which provided gold-standard testing and served as a reference laboratory. Clinicians were asked to make presumptive diagnoses of vaginal or cervical infection after they had physically examined their clients but before they performed any laboratory-based diagnostic tests.
The overwhelming majority of the study participants were married (97%). On average, these women were in their mid-30s (mean age, 36 years), had had 2.7 births and had been 24 years old at first sex. More than one-half (54%) had had a primary education or less. Some 87% were currently practicing contraception, including 31% who were using the condom, 26% withdrawal, 19% the IUD, 3% the pill and 8% other methods.
On the basis of etiologic diagnostic criteria, 21% of the clients had one or more RTIs; of these accurate diagnoses, roughly 75% were endogenous (naturally occurring) infections and the remainder were STIs. There was no significant difference between asymptomatic and symptomatic women in the proportion with an etiologically diagnosed RTI (19% vs. 21%). Overall, 17% of the study women, regardless of symptoms, were found to have endogenous infections (12% had candidiasis and 6% bacterial vaginosis) and 5% had an STI (trichomonas, syphilis, chlamydia or gonorrhea). Thirteen women (2%) had two RTIs. None of the clients' social and demographic characteristics were associated with the diagnosis of infection.
When the results from tests conducted at the MCH-FP center were evaluated against the results of the gold-standard tests performed at the hospital, the center proved better at detecting bacterial vaginosis and candidiasis than at diagnosing trichomonas. The center failed to detect only 13% and 8% of cases of bacterial vaginosis and candidiasis, respectively, but it missed 60% of the trichomonas cases that were identified by the reference laboratory.
Presumptive diagnoses led to substantial overdiagnosis and, presumably, to unnecessary treatment: Overall, 21% of women had etiologically diagnosed RTIs, but 62% received a presumptive RTI diagnosis. The levels of etiologic and presumptive diagnoses differed more for cervical infections (1% vs. 44%) than for vaginal infections (20% vs. 43%).
Three clinical signs were positively associated with the odds that a woman would receive a presumptive RTI diagnosis--vaginal discharge (odds ratio, 19.3), redness of the cervix (2.7) and bleeding of the cervix when touched (1.6). The only clinical sign significantly associated with actual infection, however, was the presence of vaginal discharge (8.3).
The researchers analyzed the usefulness of these clinical criteria in presumptive diagnoses among the 480 women (80%) who reported vaginal discharge. This symptom correctly identified only 35% of infected women, but its absence correctly ruled out infection in 98% of cases.
The researchers also assessed whether applying WHO's syndromic management guidelines would reduce the amount of overdiagnosis among women with vaginal discharge. Using the WHO guidelines instead of clinicians' presumptive diagnoses lowered the percentage of women with "diagnosed" vaginal infections, from 45% to 17%, close to the 21% with an etiologic diagnosis. However, both application of the WHO algorithm and use of clinical signs resulted in about half of women with vaginal discharge (49% and 48%, respectively) receiving a diagnosis of cervical infection, although only 1% actually had such infections.
The researchers assert that given the moderate rates of non-sexually transmitted RTIs and the especially low rates of vaginal and cervical STIs in Vietnam, "case-management efforts for symptomatic women should focus on the more common endogenous infections, as opposed to mounting more costly, difficult, and largely unsuccessful attempts to provide presumptive treatment for cervical infection." According to the researchers, the findings for vaginal infections suggest that MCH-FP clinicians could be trained to diagnose them with simple microscopy tests, provided that the necessary investment in training and diagnostic facilities is not prohibitively high. The investigators conclude that the current overdiagnosis and overtreatment of RTIs could be addressed by relying less on presumptive diagnoses and more on the standardized WHO guidelines for management of vaginal infections, and that the section of the guidelines dealing with cervical infections be dropped in areas with a low prevalence of STIs.--L. Remez
1. Lien PT et al., The prevalence of reproductive tract infections in Hue, Vietnam, Studies in Family Planning, 2002, 33(3):217-226.