Abstract

Objectives. Systematic review of point of care (POC) diagnostic tests for sexually transmitted infections: Chlamydia trachomatis (CT), Neisseria gonorrhoeae (NG), and Trichomonas vaginalis (TV). Methods. Literature search on PubMed for articles from January 2010 to August 2015, including original research in English on POC diagnostics for sexually transmitted CT, NG, and/or TV. Results. We identified 33 publications with original research on POC diagnostics for CT, NG, and/or TV. Thirteen articles evaluated test performance, yielding at least one test for each infection with sensitivity and specificity ≥90%. Each infection also had currently available tests with sensitivities <60%. Three articles analyzed cost effectiveness, and five publications discussed acceptability and feasibility. POC testing was acceptable to both providers and patients and was also demonstrated to be cost effective. Fourteen proof of concept articles introduced new tests. Conclusions. Highly sensitive and specific POC tests are available for CT, NG, and TV, but improvement is possible. Future research should focus on acceptability, feasibility, and cost of POC testing. While pregnant women specifically have not been studied, the results available in nonpregnant populations are encouraging for the ability to test and treat women in antenatal care to prevent adverse pregnancy and neonatal outcomes.

1. Introduction

Globally, Chlamydia trachomatis (CT), Neisseria gonorrhoeae (NG), and Trichomonas vaginalis (TV) are responsible for over 351.7 million infections per year [1]. Though curable, those infections can lead to long-term sequelae in both men and women, especially among pregnant women and their children. CT, NG, and TV can cause preterm birth, fetal growth retardation, low birth weight, pelvic inflammatory disease (PID) which can lead to ectopic pregnancy or infertility, perinatal diseases such as conjunctivitis and pneumonia, and infant death [25]. It has been demonstrated that up to 40% of women with STIs will develop PID, and these women are up to 10 times more likely to have an ectopic pregnancy [5], resulting in loss of pregnancy and even death. Maternal to child transmission of HIV has also been demonstrated to occur at higher rates among mothers who are coinfected with CT or NG [6]. Additionally, an estimated 4,000 infants globally are born blind each year due to maternal to child transmission of CT and NG [7]. Despite those sobering outcomes, the global response to controlling STIs has been inadequate. The World Health Organization (WHO) estimates that the global incidences of CT, NG, and TV have been steadily rising since 1995 and increased from 2005 to 2008 by 11.7% [1]. Antenatal care provides an excellent opportunity for screening for STIs, and this strategy has largely been successful in preventing mother to child transmission of HIV. An estimated 180,000 HIV infections in infants were averted globally between 2005 and 2008 due to screening and treatment of HIV-infected pregnant women [8]. International studies have shown antenatal screening for CT, NG, and TV to be feasible and acceptable as well as effective at reducing rates of infection and adverse outcomes [912]. Therefore, antenatal screening and treatment not only is important in preventing sequelae of CT, NG, and TV in women and neonates but also is a feasible intervention for infection control at the population level. More needs to be done to ensure pregnant women receive treatment for STIs starting with improving access to rapid, accurate, and inexpensive screening.

Globally, the standard of care increasingly focuses on point of care (POC) diagnostic tools for STIs. In 2002, the WHO Special Program for Research and Training in Tropical Disease identified POC testing as a key priority for controlling curable STIs [13]. POC tests can potentially be defined in multiple ways, and the WHO Special Program outlined the ASSURED criteria: affordable, sensitive, specific, user-friendly, rapid and robust, equipment-free, and deliverable. However, others also define a POC test as any diagnostic tool that can provide accurate results and facilitate treatment within the same clinical visit as testing [14]. That definition includes some nucleic acid amplification tests (NAATs), although they do not fit classic ASSURED criteria.

The purposes of POC testing are to reduce the frequency of patients who do not receive results and improve the ability to treat cases in the same clinical visit as testing. In resource-poor environments, patients may not be able to return quickly or at all for treatment, leading to morbidity in the infected individual as well as potential transmission from mother to child or to sex partners. Mathematical models demonstrate that even a POC test with 63% sensitivity ensures greater rates of treatment than a test with high sensitivity and a poor patient return rate [15]. Furthermore, while syndromic management may allow treatment without a follow-up visit, syndromic management misses most infections with CT, NG, and TV, as most are asymptomatic [24, 16]. Therefore, although syndromic management may identify individual cases, it is unlikely to be an effective public health tool in reducing the burden of disease on population levels. One study of an emergency department in North Carolina, USA, found that while syndromic management led to the treatment for presumed CT or NG of 28% patients studied, only 7.6% of all patients were actually infected, and 63.6% of those infected went untreated [17]. Syndromic management and presumptive treatment have the additional drawback of causing inappropriate treatment or treatment of uninfected individuals, contributing to the growing global problem of antibiotic resistance. Therefore, accurate testing facilitating same-visit treatment is crucial to reducing the burden of disease, especially among pregnant women and their children.

This systematic review aims to evaluate the current status of POC diagnostics for CT, NG, and TV and makes recommendations for the future of POC testing in low resource settings.

2. Methods

We used PRISMA guidelines [18] to focus our systematic review of recent literature published between January 2010 and August 2015 that reported on POC diagnostics for NG, CT, and TV. A comprehensive search term using a compilation of medical subject headings, text words, and subheadings was used to search PubMed (Figure 1). Abstracts of all search results and the full text of all potentially eligible articles were reviewed. This search yielded sixty-one articles whose abstracts we evaluated to determine whether they fit inclusion criteria. The inclusion criteria were (1) publications including CT, NG, or TV as STIs; (2) publications that date from January 2010 through August 2015; (3) publications relating to diagnostics; (4) publications published in English; and (5) original research. The exclusion criteria were (1) publications not covering CT, NG, or TV; (2) publications including those infections but not in the sexually transmissible form; (3) publications published before 2010; and (4) publications not evaluating POC diagnostics. Articles were sorted into the following subject categories and stratified based on disease: (1) performance evaluations, (2) cost analyses, (3) acceptability and feasibility trials, and (4) proof of concept studies.

3. Results

We analyzed results from 61 publications on POC testing for CT, NG, and/or TV from 2010 to 2015. Twenty-eight articles were excluded from analysis, yielding 33 articles for the review (Figure 1).

3.1. Performance Evaluations

Our search yielded thirteen articles that evaluated the performance of POC tests [1931]. Eight reports evaluated ten CT tests [1926], four reports evaluated four NG tests [21, 25, 27, 28], and three reports evaluated six TV tests [2931]. Two reports evaluated tests for both CT and NG [21, 25]. We further stratified the results when available for different populations (i.e., male versus female and/or high versus low risk) and different collection sources (i.e., urine versus vaginal swab versus cervical swab). Results are summarized in Tables 13. For each disease, the sensitivities and specificities of various tests varied widely. However, for each STI there was at least one test which consistently had sensitivity and specificity of ≥90%.

3.1.1. Chlamydia trachomatis

For CT, the ten diagnostics evaluated were the Gram stained urethral smear [19, 20], GeneXpert® Xpert CT/NG (Xpert CT/NG) (Cepheid, Sunnyvale, CA, USA) [21], aQcare Chlamydia TRF kit (Medisensor, Inc., Daegu, Korea) [22], Chlamydia Rapid Test (CRT) (Diagnostics for the Real World (Europe), Cambridge, UK) [23, 24], ACON Plate CT Rapid Test (ACON CT) (ACON Laboratories, San Diego, CA, USA) [23, 25], ACON NG and CT Duo test combo (ACON Duo) (ACON Laboratories, San Diego, CA, USA) [25], QuickVue Chlamydia Rapid Test (QuickVue) (Quidel Corporation, San Diego, CA, USA) [25, 26], Automated Urine Flow Cytometry (AUFC) of first void urine [20], Handilab-C (Zonda, Dallas, TX, USA) [26], and Biorapid Chlamydia Ag test (Biokit, S.A., Barcelona, Spain) [26]. The performance results are summarized in Table 1. All evaluations used NAAT as a reference test. The best performing test was the Xpert CT/NG, an FDA cleared real-time PCR platform, which consistently had sensitivities and specificities >97% across sample types [21]. The aQcare Chlamydia TRF kit and AUFC also performed well, with sensitivities and specificities above 90%, although they were each evaluated in one study [20, 22]. The remainder of the diagnostics did not perform adequately, with either sensitivities or specificities <75% and often <50% [19, 20, 2326]. The highly sensitive Xpert CT/NG and aQcare Chlamydia TRF kit, a rapid NAAT and lateral flow immunoassay, respectively, were the only tests of their test type evaluated [20, 21]. The other test types did not perform as accurately.

For both the Xpert CT/NG and aQcare Chlamydia TRF kit, the authors did not report significant differences in accuracy depending on sample type [21, 22]. For the Xpert CT/NG, there were not any significant differences between detecting infections in men using urine and women using urine, vaginal swabs, or endocervical swabs [21]. However, when Hurly et al. compared men and women for the CRT and ACON CT test, they found lower sensitivities for men than women, for both tests [23]. The authors attribute this difference in accuracy to the difference in sample type—urine versus vaginal swabs—and the organism load in each type of sample [23].

3.1.2. Neisseria gonorrhoeae

For NG, the five diagnostics evaluated were the Xpert CT/NG [21], ACON Duo [25], NG ACON Plate (ACON NG) (ACON Laboratories, San Diego, CA, USA) [25], Gram stain urethral or cervical smear [27], and BioStar Optical ImmunoAssay-Gonorrhea (BioStar) (BioStar, Inc., Boulder, CO, USA) [28]. The performance results are summarized in Table 2. The Gram stain was compared to culture [27]; the BioStar assay to NAAT, culture, and microscopy [28]; and the remainder to NAAT as the reference standard. All five tests were each only evaluated in one study. The Xpert CT/NG had the highest sensitivities and specificities, consistently >95% between sample types for both men and women [21]. The BioStar assay had similar sensitivities and specificities [28], although with a much smaller sample size. Although the BioStar assay and the ACON NG and ACON Duo tests are all immunoassays, the BioStar assay was far more sensitive than the ACON tests [25, 28]. As with the CT studies, the ACON Duo and ACON NG tests had sensitivities <35% [25]. Gram staining had dramatically and significantly different sensitivities between urethral smears for men and cervical smears for women. Among men, the Gram stain urethral smear had sensitivities >95%, but for women, the sensitivities of Gram stain cervical smear were <35% [27].

3.1.3. Trichomonas vaginalis

For TV, the six diagnostics evaluated were the OSOM Trichomonas Rapid Test (Sekisui Diagnostics, Lexington, MA, USA) [2931], acridine orange staining [30], wet mount microscopy [30, 31], real-time in-house PCR [31], and culture [31]. The performance results are summarized in Table 3. All diagnostics were compared to culture [30] or a composite of multiple tests as the reference standard [29, 31]. Across three different studies, the OSOM Test performed well, with sensitivities and specificities >88% [2931]. This was in contrast to the traditionally used POC diagnostic, wet mount microscopy, which had sensitivities <60% in two studies [30, 31]. All the studies were conducted on women using vaginal swabs, so no differences between samples types were reported, although one article mentioned that the higher organism load required for detection by culture and microscopy may have reduced the accuracy of those two tests as compared to other tests using the same sample type [31].

3.2. Cost and Cost Effectiveness

Our search found three publications which analyzed the cost effectiveness of POC diagnostics for CT and/or NG [27, 32, 33]. One study was a retrospective analysis of results from actual patients [27], while the other two used mathematical models [32, 33]. All three found that POC tests were a cost effective strategy for diagnosing those STIs, supporting the conclusion of Gift et al. in 1999 that a reasonably accurate POC test can be a cost effective diagnostic by minimizing loss to follow-up [15].

In 2014, Bartelsman et al. performed a study on women attending an STI clinic in Amsterdam between 2008 and 2011 [27]. In 2008-2009, all “high risk” (as determined by physicians) patients at the clinic were offered Gram stains for NG, but in 2010-2011 this was limited to only women who had urogenital symptoms. That change saved 2.34 (US$2.54 using 2015 conversion rate) per correctly managed consultation without significantly changing loss to follow-up and while maintaining an equivalent diagnostic accuracy as the standard of care, culture.

In 2014, Turner et al. created a simulation of 1.2 million attendees at a sexual health clinic in England to compare the use of a POC NAAT with the standard protocol for CT and NG [32]. Using the NAAT instead of standard protocol saved £11.7 million (US$17.7 million using 2015 conversion rate) and an additional 47 Quality-Adjusted Life Years over all patients. They therefore found the POC NAAT to be a cost effective diagnostic, improving both cost and health outcomes. In a 2013 model, Huang et al. created a decision tree for a scenario of 10,000 women visiting an STI clinic with a hypothetical CT POC test [33]. With a test sensitivity of 92.9%, test cost of $33.48, and 47.5% of women willing to wait the 40 minutes for a test result, the POC test saved $5,050 for each case of PID averted compared with a non-POC NAAT. They found the POC test would save money compared to a non-POC NAAT as long as the POC tests had a sensitivity of 87.1% or greater or if it cost less than $41.52.

3.3. Feasibility and Acceptability

Our search yielded five articles which discussed feasibility and acceptability of using POC diagnostics in clinical settings [24, 3437]. Three studies, by van der Helm et al. and Huppert et al. in 2010 and 2011, surveyed female patients about their preferences [24, 34, 35]. The other two, by Hsieh et al. in 2010 and 2011, instead surveyed health care providers for their opinions on averting potential challenges of POC testing [36, 37]. All of the referenced reports agreed that POC testing would be feasible and acceptable to perform in a clinical setting.

van der Helm et al. and Hsieh et al. in 2010 and 2011 focused on wait time as a potential challenge for POC testing. In 2012, van der Helm et al. interviewed women at an STI clinic in Maastricht, the Netherlands, regarding the acceptability of waiting for the results of a rapid CT test, in this case CRT [24]. They found that 98.7% of women agreed to wait for half an hour for results, and 26.7% of this group would be willing to wait for an hour. These results correspond with qualitative studies performed by Hsieh et al. in 2010 and 2011. In 2010, focus groups of clinicians and other professionals working in STIs highlighted long wait times as a major barrier to use of current POC tests, in addition to complex protocols and difficulty of interpreting results [36]. As a follow-up to these surveys, Hsieh et al. performed an online survey in 2011 for STI clinicians who also identified time as a barrier for current POC diagnostic use [37]. In the focus groups, themes for the ideal POC test prioritized fast turnaround time, ease of use, accuracy, and user-friendliness. The online surveys confirmed this result with accuracy as the highest priority for an ideal test, followed by low cost. Both groups selected CT as their priority organism to target with a new POC test.

In 2010 and 2011, Huppert et al. studied the feasibility and acceptability of self-testing for TV among urban teens in Cincinnati, OH, USA. In 2010, sexually active women aged 14–22 years old recruited at a teen health center or emergency department collected vaginal swabs and used the sample for the OSOM Test [34]. All participants correctly performed the test, and 99% interpreted the results correctly. The results agreed with the physician-collected and physician-performed tests in 95.7% of cases. As a follow-up to the earlier study with a similar group of participants in 2011, Huppert et al. surveyed women to evaluate their trust and comfort of self-collected versus clinician-collected samples before and after testing [35]. They found that women’s trust in their ability to perform a self-test increased after having done so. Their comfort was higher for self-testing than clinician-testing both before and after performing the test. These results were in agreement with women’s attitudes toward self-testing: both at baseline and after testing, 93% of women were definitely or probably willing to test themselves at home. As in the 2010 study, 99% of women were able to correctly perform the test and interpret the results.

3.4. Proof of Concept

Our search yielded fourteen proof of concept articles for new POC diagnostics [3851]. Eleven reports introduced CT diagnostics [3848], five introduced NG diagnostics [39, 4750], and two introduced TV diagnostics [47, 51]. The diagnostic methods and performance, if provided, for those novel tests are summarized in Tables 46. Those diagnostics varied widely in method and in performance. Some, such as the BioStar assay, have been the subject of later articles with encouraging results [28]. Whereas some, such as the Xpert CT/NG, are already commercially available, others do not appear to have been followed up with later published research or development.

4. Discussion

We reviewed 33 articles on POC testing for CT, NG, and/or TV from 2010 to 2015. We analyzed the publications to report results on (1) performance evaluations, (2) feasibility and acceptability analyses, (3) cost effectiveness, and (4) proof of concept reports.

For performance evaluations, we found that six publications had tests with >80% sensitivity and specificity, at least one test for each CT, NG, and TV [21, 22, 2831]. However, many commercially available or commonly used diagnostics, while their specificities were >90%, had sensitivities <50%, such as the QuickVue Chlamydia Rapid Test (with sensitivities of 25.0 [26]–37.7% [25], specificities of 99.4 [25]–99.7% [26]), ACON NG test (with sensitivities of 0–12.5% [25], specificities of 97.2–99.8% [25]), and wet mount microscopy for TV (with sensitivities of 38.0 [31]–58.8% [30], specificities of 99.3 [30]–100% [31]). Furthermore, these tests were usually analyzed in high risk or symptomatic populations, potentially giving them an inflated accuracy compared to the general population, which has lower rates of infection. However, articles such as Gaydos et al. in 2013 [21], Hurly et al. in 2014 [23], and Hegazy et al. in 2012 [29], which evaluated both symptomatic and asymptomatic women at reproductive health clinics, showed very good accuracy for the Xpert CT/NG, CRT, and OSOM Test, respectively. Indeed, in 2012, van der Helm et al. did not find a significant difference in the performance of CRT between women attending an STI clinic (with sensitivity of 39.4%, specificity of 94.4%) and those at a sexual health and family planning clinic (with sensitivity of 42.0%, specificity of 95.8%) [24]. While that study only evaluated one relatively poorly sensitive test, it shows that a diagnostic test will not necessarily perform worse in a lower risk patient population.

Among the tests available for women that had sensitivities ≥90%, the Xpert CT/NG and aQcare Chlamydia TRF kit were the only two tested with multiple sample types and neither had significantly different performances between sample types [21, 22]. AUFC [20], the BioStar assay [28], and the OSOM Test [2931] were each only evaluated using one sample type. Gram stain urethral smear also had a sensitivity >90% for CT [19, 20] and NG [27], but this test is only available to men. Gram stained cervical smears were not as accurate for NG [27] and were never studied for CT. Other tests which were studied across sample types, such as CRT, ACON CT test, and QuickVue Chlamydia Rapid Test, had some variability in sensitivities based on sample type, but no sample type ever reached specificity >75%. [2326].

Those tests available for women with sensitivities ≥90% varied in their style of detection: a rapid NAAT (the Xpert CT/NG) [21], immunoassays (the aQcare Chlamydia TRF kit and BioStar assay) [22, 28], an antigen detection test (the OSOM Test) [2931], and AUFC [20]. Those data are encouraging for the continued pursuit of multiple types of test. However, other tests of the same types did not perform with adequate sensitivities for use: the ACON CT, ACON NG, ACON Duo, and QuickVue immunoassays and the Biorapid Chlamydia Ag antigen detection test all performed with sensitivities <70% [23, 25, 26]. Overall, the highly sensitive performance of some immunoassays, antigen detection tests, and rapid NAATs in comparison with the low sensitivities of traditional diagnostics such as Gram stains, culture, and microscopy [19, 20, 27, 30, 31] underline the importance of continuing development and improvement of POC tests.

The cost and cost effectiveness articles about POC testing for STIs were similarly encouraging. However, as with the acceptability and feasibility articles, the small number of total studies as well as the emphasis on modeling rather than observational data leaves room for additional research. Models are important to pave the way for a potential new strategy, but now that POC diagnostics are in use around the world, more work should be done on evaluating the actual costs of implementing POC tests. While Bartelsman et al.’s study in the Netherlands [27] is a good first example of that, more studies should be done in varied economic settings and with different tests.

The articles on acceptability and feasibility showed that POC testing is a priority for health care professionals and is feasible to implement for patients. However, the low number of total reports and focus on surveys about potential POC testing demonstrate a need for additional research studying the actual use of POC diagnostics and how patients and providers respond to them. The articles by Huppert et al., which evaluated the actual use of the OSOM Test, showed promising feasibility and acceptability of the self-test and yielded an important insight into women’s comfort with self-testing versus clinician-testing [34, 35]. Additional studies to evaluate the best way to optimize the use of POC diagnostics among different populations with varying prevalences are needed.

The proof of concept articles show a promising number of novel tests and methods from the past five years, some of which have already gone on to become commercially available tests. However, the small number of TV and NG tests compared to CT demonstrates that there is still room for more innovation. As more tests become available, the cost to produce, distribute, and use these tests will also decline, increasing availability in low resource settings, where STI prevalence is highest and the burden of adverse outcomes is greatest.

Previous literature reviews of POC diagnostics for STIs, including CT, NG, and/or TV, demonstrate the growth of the field in the last five years. As with our findings, reviews in 2011 showed the consensus that traditional diagnostics or strategies such as microscopy, syndromic management, and culture are not sufficient tools for diagnosing STIs [4, 52]. However, in 2011, the accuracy of alternative rapid assays was still not as high as desired, and the authors called for continued research into more accurate immunoassays. In recent years, more assays have been developed and rapid NAATs such as the GeneXpert have entered the market. In 2013, Gaydos and Hardick found the Xpert CT/NG for CT and NG and the OSOM Test for TV to be promising new developments in an otherwise unsatisfactorily inaccurate POC testing scene [53]. The optimism for the OSOM Test was shared by McGowin et al. in their 2014 review of TV diagnostics [54], and we, too, have found that these tests perform well. With additional data from articles published in 2014-2015, our review has confirmed those results and found new tests that have been demonstrated to have high accuracies [21, 22, 28, 30, 31]. One previous review also commented on the patient populations studied in performance evaluations [55]. Watchirs Smith et al. found that, unsurprisingly, tests demonstrated more accuracy when studied in symptomatic populations [55]. However, they did not comment on the accuracy of results for low versus high risk patients in contexts such as STI clinics versus family planning clinics.

Just as we found few articles on acceptability, feasibility, and cost effectiveness of implementing POC diagnostics, other reviews have also called for more research on these subjects [56, 57]. Previous reviews also noted the disproportionate number of studies on and diagnostics available for CT compared to NG [57]. This is in agreement with our review, in which twice as many articles evaluated the performance of CT as NG or TV, and for which we found no articles on the cost or cost effectiveness of TV testing.

5. Limitations

Our review had several limitations. Firstly, we only used PubMed to find relevant articles. Secondly, there are some existing commercial tests, such as the GeneXpert TV test, which are approved for use but were not evaluated in any publications between January 2010 and August 2015, preventing their inclusion in our study. Due to the small number of articles on cost effectiveness, feasibility, and acceptability of implementing POC testing, it is difficult to draw strong conclusions. Similarly, the lack of studies on NG and TV demonstrates a need for more research and development in order to control these diseases. There were some diagnostics, such as the Xpert CT/NG, aQcare Chlamydia TRF, and BioStar assay which had good sensitivity and specificity but were each only evaluated in one trial among one group [21, 22, 28]. Continued evaluation should be done with those diagnostics to confirm their accuracy across settings and populations. For TV particularly, it has not been clearly demonstrated that screening pregnant women is beneficial, so additional research on this topic with highly sensitive and specific POC tests should be performed. Of the publications on TV we reviewed, all used culture or a composite of imperfect tests as the reference standard rather than NAAT, the usual comparator for all CT and most NG studies. Additional trials should more rigorously compare TV diagnostics to a more sensitive reference standard, NAAT. The use of different reference tests in various studies undermined our ability to compare accuracy across articles, as some reference tests such as culture and microscopy have been demonstrated to have low sensitivities. Finally, especially given the vulnerability among pregnant women and neonates, more research is needed on diagnostics in pregnant women. We did not identify any research which specifically studied pregnant women, and most studies excluded pregnant women.

6. Conclusion

Overall, this review demonstrates that recent progress has occurred for developing diagnostics for CT, NG, and TV that have high accuracy. However, we still need more studies of those tests for acceptability, feasibility, cost (especially for low and middle income countries), and sensitivity and specificity among populations not considered to be at risk (especially for NG and TV). While pregnant women specifically have not been studied, the results available in nonpregnant populations are encouraging for the ability to screen and treat women in antenatal care to prevent adverse pregnancy and neonatal outcomes.

Competing Interests

The authors declare that they have no competing interests.

Acknowledgments

Claire C. Bristow acknowledges funding from NIDA T32 DA023356 and NIDA R01 DA037773-01A1. Additional funding was received from UCLA Center for AIDS Research (CFAR) NIH/NIAID AI028697 and UCLA Center for HIV Identification, Prevention, and Treatment Services (CHIPTS) NIH P30MH058107.