|Home | About | Journals | Submit | Contact Us | Français|
The objective of this study was to examine the effects of clinical factors and of the type and timing of a secondary test in improving the sensitivity of Trichomonas vaginalis detection in young women over that of a wet mount alone. For this purpose, sexually active adolescent women (n = 345) were recruited from a hospital teen clinic or emergency department. Following an interview and a pelvic exam, four primary T. vaginalis tests (wet mount, culture, a rapid test, and a nucleic acid amplification test [NAAT]) were performed on vaginal swabs. If the wet-mount result was negative, two secondary tests (culture and a rapid test) were performed on the used wet-mount swab and saline. A positive result by any of the four primary tests was considered a true T. vaginalis-positive result. The prevalence of T. vaginalis was 18.8% overall and 8.8% in the 307 wet-mount-negative women. There was 100% concordance between primary and secondary rapid tests. Secondary culture was 80% sensitive compared to primary culture. The likelihood of a positive rapid test increased with increasing time between specimen collection and testing. A wet mount followed by a rapid test was the most sensitive strategy using two tests (86.4%; confidence interval [CI], 75.3 to 93.4%). Limiting secondary testing to those with multiple partners resulted in a lower sensitivity (73.9%; CI, 61.5 to 84%) that was not significantly better than that of the wet mount alone (58.5%; CI, 45.6 to 70.6%). We conclude that a rapid test can be delayed or performed on a used swab with no loss of sensitivity. Until a NAAT for T. vaginalis is commercially available, a stepwise approach using an additional rapid test for wet-mount-negative women is recommended for adolescent women regardless of clinical factors.
Trichomonas vaginalis infection is estimated to be the most prevalent nonviral sexually transmitted infection (STI) in adolescents (15). Population-level data from the National Longitudinal Study of Adolescent Health have documented that T. vaginalis affects 2.8% of young adult women, a prevalence seven times higher than that of Neisseria gonorrhoeae in this age group (10). In some populations, the prevalence of T. vaginalis is even higher. T. vaginalis infection is not innocuous; it has been associated with several ominous health outcomes. For example, women infected with T. vaginalis are more likely to acquire human immunodeficiency virus (HIV) (9) and herpes simplex virus type 2 (4). T. vaginalis infection also doubles the risk of persistent human papillomavirus infection in women (12). Finally, in both women and men with HIV, coinfection with T. vaginalis increases HIV shedding (6, 14). Because T. vaginalis is highly prevalent and is associated with adverse outcomes, improved detection of T. vaginalis is needed.
Wet mount is the most widely used method for the detection of T. vaginalis, with a sensitivity of 51 to 66% and a specificity of 100% (16). According to the 2006 guidelines for the treatment of sexually transmitted diseases (STDs) published by the Centers for Disease Control and Prevention (CDC), a T. vaginalis culture is recommended when T. vaginalis is suspected but not seen on a wet mount (3). Culture has a higher sensitivity (75 to 85%) than the wet mount and a specificity of 100% (2). Therefore, some authors have proposed a stepwise approach to T. vaginalis diagnosis, using culture after a negative wet-mount result (11, 13). Swygard et al. suggest that in an STD clinic setting, the addition of culture be limited to a subset of women, such as African-Americans or those reporting a contact with T. vaginalis or a history of substance use (13).
In many clinical settings, culture is not available, and it requires additional resources. New test methods for the detection of T. vaginalis are now available. The rapid antigen test is a point-of-care, lateral-flow test strip device that detects T. vaginalis membrane proteins. It is cleared by the FDA for use on a direct vaginal swab or on used wet-mount saline. Its reported sensitivity (85 to 90%) and specificity (100%) are similar to those of culture (7, 8). In addition, transcription-mediated amplification is a nucleic acid amplification test (NAAT) method that uses analyte-specific reagents for T. vaginalis. It is commercially available but not yet FDA cleared for the detection of T. vaginalis; therefore, it is used mainly in research settings (Aptima; Gen-Probe, Inc., San Diego, CA). This NAAT has high sensitivity (96.7%) and specificity (97.5%) compared to those of in-house PCR detection of T. vaginalis (5).
With the advent of these newer tests, we wanted to discover whether cultures, rapid antigen tests, or NAATs could be used in a stepwise fashion after the initial wet mount to improve the sensitivity of T. vaginalis diagnosis for adolescent women. Therefore, the aims of this study were to determine (i) which type of secondary test improves the sensitivity of T. vaginalis detection in young women above that of a wet mount alone and (ii) whether clinical factors, such as the presence of white blood cells (WBCs) or the timing of the test, or demographic variables, such as patient history, can identify a subset of wet-mount-negative women who would benefit from further T. vaginalis testing.
This study represents further analysis of a cross-sectional study of T. vaginalis detection methods for adolescent women. We have previously reported the comparison of four primary test methods for T. vaginalis (8). To summarize, we recruited a clinical sample of sexually active adolescent women 14 to 21 years old, with and without genitourinary symptoms. Subjects were not eligible if they had received metronidazole within 2 weeks of enrollment. Following a confidential interview, a pelvic exam was performed and four vaginal swabs were obtained. Four primary T. vaginalis tests were performed directly on the vaginal swabs: a wet mount, culture (InPouchTV; Biomed Diagnostics, San Jose, CA), a rapid antigen test (OSOM TV; Genzyme Diagnostics, Boston, MA), and a NAAT (Aptima; Gen-Probe, Inc., San Diego, CA).
For those women with a negative wet-mount result, two secondary T. vaginalis tests were initiated: the residual wet-mount saline was used to inoculate a second T. vaginalis culture, and the original wet-mount swab was used to perform a second rapid test. In order to mirror clinical care, we delayed the initiation of secondary testing until the results of the wet mount were known, and we recorded the time interval between specimen collection and test initiation. For the present study, subjects were included if the wet-mount result was negative and results were recorded for primary and secondary rapid antigen and culture tests for T. vaginalis, or if the wet-mount result was positive and results were recorded for any T. vaginalis test.
All tests were performed according to the manufacturer's instructions, as described previously (8). In the Teen Health Center, the wet mount was read immediately after collection by a wet-mount-trained and certified clinician, and the presence or absence of motile trichomonads, WBCs, clue cells, and yeast forms was reported. For subjects recruited from the Emergency Department, a wet mount and a swab for a Gram stain were sent to the laboratory. Trained laboratory technicians read the wet mount for the presence of motile trichomonads, and the Gram stain was read for WBCs, clue cells, and yeast forms.
Data from the wet mount and Gram-stained smear were recorded as follows. WBCs were categorized as moderate if the sample contained ≥10 WBCs per field (at a magnification of ×400). Clue cells were considered to be present if >20% of epithelial cells in a field were classified as clue cells (1). Yeast was recorded as present if either buds or pseudohyphae were seen at any magnification.
The clinical data recorded for each participant included the following patient-reported variables: race, history of STIs, number of sexual partners in the past 3 months, number of new partners in the past 3 months, whether a condom was used at the last sexual intercourse, and the presence or absence of genitourinary symptoms (vaginal discharge, vaginal itching, abnormal vaginal bleeding, or pelvic pain). Institutional board review approval was granted for this study.
We determined the sensitivity of each secondary test compared to the corresponding primary test for culture and rapid antigen. We measured the concordance between the primary and secondary tests separately for culture and rapid antigen. Because of the high specificities of these T. vaginalis tests (8), we defined a true-positive T. vaginalis result as a positive result if any of the four primary tests gave a positive result. We then compared the performance of either of the secondary tests to a true-positive result. In order to assess whether any factors predicted when the rapid test would perform better than culture, we created a dichotomous outcome variable, “rapidbetter,” where “yes” represents a positive secondary rapid test result with a negative secondary culture result and “no” represents all other results. We evaluated predictors of true-positive results for T. vaginalis by using logistic regression, entering all clinical variables that were associated with positive T. vaginalis results by univariate testing at a P value of ≤0.1 and using backward stepwise elimination to arrive at the most parsimonious model. Using the entire sample, we generated point estimates and 95% confidence intervals (95% CI) for the sensitivity of each diagnostic strategy and compared them for a strategy using a wet mount alone.
Of 376 women recruited, 10 were missing T. vaginalis rapid antigen or culture test results from primary swabs and 21 were missing T. vaginalis test results from secondary swabs. Thus, data were available for 345 women; 307 of these were wet mount negative. The median age of the wet-mount-negative participants was 17.6 years (range, 14 to 21 years), with 82% identifying their race as black. There was a high prevalence of STI risk behaviors among them, with many reporting a prior STI (62%), two or more sexual partners in the past 90 days (25%), a new partner in the past 90 days (29%), and not using a condom at the last sexual contact (51%). See Table Table11 for the demographic and clinical variables of the wet-mount-negative women.
The prevalence of T. vaginalis true-positive results was 18.8% (n = 65) for the full sample of 345 women and 8.8% (n = 27) for the wet-mount-negative subset. Of the 27 wet-mount-negative, true T. vaginalis positive results, 20 were positive by NAAT and either culture or a rapid antigen test; 5 were positive by NAAT only; and NAAT results were missing for 2 (1 was positive by culture only, and 1 was positive by both the rapid antigen and culture tests). The prevalence of T. vaginalis in women who were negative by both the wet mount and the rapid test was 3.1% (9 of 289) (5 positive by NAAT only; 4 positive both by culture and by NAAT). Figure Figure11 gives the number of positive results for each test.
In order to discern whether there was a subset of wet-mount-negative women who would benefit from additional T. vaginalis testing based on clinical and lab predictors, we used univariate and multivariable logistic regression comparing independent variables with any positive result on a T. vaginalis test. We found that among wet-mount-negative women, T. vaginalis was not significantly associated with race, vaginal discharge, condom use, prior T. vaginalis infection, or the presence of WBCs on the wet mount (Table (Table1).1). The prevalence of T. vaginalis was higher among those who reported ≥2 sexual partners than among those who reported <2 partners in the past 90 days (14.5% versus 6.5%; odds ratio [OR], 2.7; P = 0.02). The magnitude of this effect persisted even after adjusting for a history of prior T. vaginalis infection and black race; however, including other variables only decreased the sensitivity of the model.
In a comparison of secondary tests to primary tests and true positives (n = 27) among the wet-mount-negative women, there was 100% concordance between a primary rapid test performed on a direct vaginal swab and a secondary rapid test performed on a used wet-mount swab; all 18 of those who tested positive by the primary rapid test also tested positive by the secondary rapid test. Of the 15 who tested positive by primary culture of a direct vaginal swab, only 12 were detected as positive for T. vaginalis by the secondary culture of the used wet-mount solution (sensitivity, 80%; specificity, 99%; kappa = 0.85). Rapid tests detected T. vaginalis in 67% (95% CI, 48 to 88%) of true T. vaginalis positive, wet-mount-negative cases, a performance slightly better than those of primary culture (56% [95% CI, 35 to 74%]) and secondary culture (48% [95% CI, 27 to 68%]), though with overlapping confidence intervals.
Next we explored in which scenario the secondary rapid test would perform better than the secondary culture. We tested the outcome variable “rapidbetter” against the factors listed in Table Table1.1. However, the only significant finding was that rapid tests performed better than secondary culture for patients with a history of multiple sexual partners (OR, 6.5 [95% CI, 1.6 to 26]).
When evaluating the effect of time on test results, we found that as the time interval between specimen collection and testing increased, there was a trend toward decreased positivity for the secondary culture, but the secondary rapid test was more likely to be T. vaginalis positive. Analysis of the data suggested a natural cut point at 50 min. Of those subjects (82% of the sample) for whom the interval between specimen collection and testing was ≤50 min, 3.8% were positive by the secondary rapid test, whereas of those for whom the interval was >50 min, 19% were positive by the secondary rapid test (P = 0.001).
We examined the data to determine if there was an algorithm that would improve the detection of T. vaginalis over that by the wet mount alone, starting with the full data set of 345 women. We compared the sensitivities of eight diagnostic strategies, using true-positive infections (n = 65) as the comparator (Table (Table2).2). The lowest sensitivity (58.5%) is that of the usual standard of care: strategy 1, a wet mount only for all. Several alternative approaches were significantly more sensitive than the usual care: using only a rapid test for all (strategy 3); using a wet mount for all, followed by a rapid test for wet-mount-negative subjects (strategy 4); using a wet mount for all, followed by a rapid test for wet-mount-negative subjects and culture for rapid-test-negative subjects (strategy 7); and using a wet mount for all, followed by a NAAT for wet-mount-negative subjects (strategy 8).
Although a history of multiple sexual partners was associated with T. vaginalis in univariate analysis, the strategy that limits additional rapid testing for T. vaginalis to those subjects who are wet mount negative and report multiple partners (strategy 5) was not significantly more sensitive than the wet mount alone. A wet mount followed by a rapid test for wet-mount-negative-subjects, with a culture for those who are negative by the rapid test (strategy 7), is the most sensitive strategy using currently marketed tests available in clinical settings. This approach results in a sensitivity of 92.3%, similar to the sensitivity obtained by using a NAAT for those who are wet mount negative.
In this study, we found that the prevalence of T. vaginalis was high among adolescents who were wet mount negative (8.8%). Risk factors did not help to identify wet-mount-negative adolescent women who tested positive for T. vaginalis. This is in contrast to the findings reported by Swygard et al., who recommended performing T. vaginalis culture for wet-mount-negative women with one of three risk factors: black race, reported T. vaginalis contact, or any drug use (13). Adding culture to the evaluation of women with any of these three factors resulted in the detection of 97.3% of T. vaginalis infections. That study enrolled 2,194 adult women from an STD clinic setting where both a wet mount and T. vaginalis culture were routine. In contrast, in our study, neither race nor a history of prior T. vaginalis infection was a predictor of T. vaginalis infection in women who were wet mount negative. We did not assess participants for recent T. vaginalis contact or drug use. In both our study and that of Swygard et al., neither clinical laboratory data (such as the presence of WBCs or clue cells) nor patient-reported symptoms were reliably associated with a T. vaginalis diagnosis. In addition, we included more-sensitive test methods (the rapid test and NAAT) to establish our definition of a true-positive infection. This could result in a decrease in the sensitivity of culture in our study or an increase in the sensitivity of culture in the study of Swygard et al.
We also found that a delay of more than 50 min between specimen collection and testing was associated with an increased likelihood of a positive secondary rapid test. This may be explained by the difference in the test mechanism between the rapid test and culture. Culture requires live trichomonads, whereas the rapid test requires that the trichomonad be dead and its membrane disrupted. A lengthy stay in saline increases the likelihood of trichomonad death, releasing membrane proteins and thus decreasing the sensitivity of culture and increasing the sensitivity of the rapid test. In addition, other, unmeasured clinical factors may have contributed to this result. While we do not recommend delaying the reading for a point-of-care test, the finding that sensitivity does not decrease with time is reassuring in usual-care settings where a rapid test might be delayed until the wet-mount results are known.
Although we found that secondary culture of used wet-mount saline is the least sensitive method after a wet mount only, others have reported that delayed inoculation of a T. vaginalis culture pouch using a primary swab yields good results. Schwebke et al. studied 150 women using a wet mount, primary culture (inoculated at the bedside), and a delayed culture (a primary swab held until after the wet mount was completed). For 39 T. vaginalis culture-positive subjects, delayed culture was 100% sensitive compared to primary culture, regardless of whether the delayed swab was held in saline or held dry (11). The better results with a primary swab may be related to the increased organism load found on a primary swab relative to that in the residual wet-mount saline that we used for the secondary culture.
Because T. vaginalis infection is linked to serious health outcomes, such as the acquisition and shedding of HIV, our findings suggest that in similar settings, all adolescent women may benefit from additional T. vaginalis testing regardless of other risk factors. In the hands of experienced providers, wet-mount tests yield valuable information on other parameters, such as the presence of WBCs, clue cells, or yeast. Therefore, despite its moderate sensitivity for T. vaginalis, the wet mount is a reasonable first step in the evaluation of women at risk for T. vaginalis and other STIs. Point-of-care tests, such as wet-mount and rapid tests, are important STI diagnostic strategies, since they allow immediate counseling and treatment. Therefore, we propose using a stepwise diagnostic strategy, as follows. (i) Since the wet mount is a useful and inexpensive test, all women at risk for STIs should have a wet mount performed. When the wet-mount swab is obtained, an additional primary vaginal swab can be obtained and held either in saline or as a dry swab. (ii) If the wet mount is negative for trichomonads, a rapid antigen test can be performed using the saved primary swab or the used wet-mount swab with no loss of sensitivity. (iii) If both the wet-mount and rapid antigen tests are negative, the provider should consider inoculating a T. vaginalis culture using the saved primary swab if T. vaginalis is highly suspected or if the prevalence of T. vaginalis is high in the population. Because the NAAT for T. vaginalis is not yet commercially available, we cannot recommend its use as a primary or secondary screening method for T. vaginalis infections.
A limitation to this study is that it may not be generalized to other populations because of the high prevalence of T. vaginalis and of risk behaviors (i.e., multiple partners) associated with T. vaginalis infection in our sample. Also, we did not calculate the costs associated with the use of additional T. vaginalis tests. It is reassuring that the manufacturers' listed costs for culture and the rapid test are significantly less than those for other commonly used STI tests, such as NAATs for Chlamydia spp. The incremental cost, that is, the cost per additional T. vaginalis infection detected, will vary depending on the prevalence of T. vaginalis in the population tested. At present, we still do not have sufficient data on health outcomes related to T. vaginalis infection upon which to base estimates of the cost-effectiveness of T. vaginalis screening.
The main outcome of this study is that our findings support and expand upon the CDC's recommendation to provide additional T. vaginalis testing for wet-mount-negative women. First, when a wet mount is routinely performed, a rapid test for T. vaginalis can be delayed until after the wet mount is read, and it can be performed on the used wet mount swab with no loss of sensitivity. A stepwise approach using an additional T. vaginalis test for wet-mount-negative women will increase the rate of detection of T. vaginalis and is recommended for all adolescent women. Second, we emphasize that it is important from a public health perspective to screen asymptomatic adolescent women for T. vaginalis. Further study is needed to assess the degree to which the increased cost of additional T. vaginalis testing may be balanced by public health benefits.
J. Huppert was supported by National Institutes of Health/National Institute for Allergy and Infectious Diseases (NIH/NIAID) grant 5K23AI63182. Kits and reagents were provided by Genzyme Diagnostics and Gen-Probe, Inc.
No funding entity participated in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript. J.H. has received unrestricted research funds, in-kind test kits, and a speaker's honorarium from Genzyme Diagnostics, Inc. The other authors report no conflict of interest.
Thanks to Grace Kim and Jennifer Bishop for their diligence in subject recruitment and data management.
Published ahead of print on 5 November 2008.