Objectives Systematic review of the performance and operational characteristics of point-of-care (POC) tests for the diagnosis of Neisseria gonorrhoeae.
Methods We searched PubMed and Embase until August 2010 using variations of the terms: ‘rapid test’, ‘Neisseria gonorrhoeae’ and ‘evaluation’.
Results We identified 100 papers, 14 studies were included; nine evaluated leucocyte esterase (LE) dipsticks and three immunochromatographic strips, and two clinical audits of microscopy were identified. Of the field evaluations the gold standard was nucleic acid amplification technology in six studies and bacterial culture in the other six. In four studies, 50% or more of the patients were symptomatic. The median sensitivity of LE dipsticks was 71% (range 23–85%), median specificity was 70% (33–99%), median positive predictive value (PPV) was 19% (5–40%) and median negative predictive value (NPV) was 95% (56–99%). One LE study found a sensitivity of 23% overall, increasing to 75% in symptomatic women. LE dipsticks mostly involved three steps and took under 2 min. The median sensitivity of immunochromatographic tests (ICT) was 70% (60–94%), median specificity was 96% (89–97%), median PPV was 56% (55–97%) and median NPV was 93% (92–99%). Immunochromatic strips involved five to seven steps and took 15–30 min. Specificity of microscopy ranged from 38% to 89%.
Conclusions ICT and LE tests had similar sensitivities, but sensitivity results may be overestimated as largely symptomatic patients were included in some studies. ICT had a higher specificity in women than LE tests. The findings highlight the need for improved POC tests for diagnosis of N gonorrhoeae and more standardised evaluations.
- Systematic Reviews
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Gonococcal infection is an infectious sexually transmissible infection (STI) caused by Neisseria gonorrhoeae. Uncomplicated infection is characterised by urethritis in men, and lower genitourinary tract symptoms such as vaginal discharge and dysuria in women. Infection of the rectum and pharynx can occur in both sexes.1 More serious complications can occur such as pelvic inflammatory disease, infertility and ectopic pregnancy2 ,3 in women or epididymo-orchitis in men. In both sexes gonorrhoea has been associated with an increased risk of acquiring and transmitting HIV.4–6
In high-income settings the diagnosis of gonococcal infection is usually made using bacterial culture or nucleic acid amplification tests (NAAT). These tests are performed in controlled laboratory environments with results generally available in a few days. In most settings, clinicians then request that the patient attends for treatment and contact tracing is initiated.2 In urban clinics, most patients diagnosed with STI receive treatment in a timely manner.7
In resource-limited settings, there is often a lack of laboratory infrastructure8 so syndromic management is often employed. Syndromic management enables first-line healthcare providers to commence treatment immediately, based on the most common organisms locally that cause such symptoms.9 ,10 Flow charts and algorithms developed by the WHO are used as a guide.8 A limitation of syndromic management is the potential for unnecessary treatment of patients whose symptoms are due to other causes. Unnecessary treatment carries the risk of antibiotic resistance developing11 and potential harm resulting from drug side effects. In addition, false positive results may result in psychosocial harm, especially if contact tracing is initiated.
It is well known that a significant proportion of infected persons either have no symptoms or do not present for treatment (approximately 46% of gonococcal infections in men and 85% in women are asymptomatic).12–14 Therefore control programmes that rely only on detecting and treating symptomatic persons will have a limited impact. In recent years with the availability of NAAT in high-income countries, offering testing broadly to asymptomatic persons, particularly in remote communities, has become a viable strategy.
Rapid point-of-care (POC) testing has the potential to improve the detection and management of gonococcal infection in resource-limited settings with poor laboratory infrastructure, or remote areas where there are delays in the receipt of laboratory results and potential loss to follow-up of patients. These tests could provide an almost immediate aetiological diagnosis, enabling clinicians to offer treatment and begin the process of partner notification at the same consultation. This process could increase treatment rates and reduce the time to treatment of positive cases and their contacts, which may in turn reduce re-infections and community prevalence. The United Nations development programme, World Bank and WHO's sexually transmitted diseases diagnostics initiative ASSURED criteria enumerate priorities for the development of new STI diagnostics.15 These criteria propose that STI diagnostics should be affordable, sensitive, specific, user friendly (three to four steps, with minimal training), rapid and robust, equipment free using easily collected non-invasive specimens and deliverable to end users.15
Recognising the potential of POC testing, leucocyte esterase (LE) dipstick tests have been used in a number of countries for many years. LE is an enzyme found in certain white blood cells that are present as a result of the inflammation caused by STI such as gonorrhoea. There have recently been a number of publications on immunochromatographic tests (ICT), which are simple devices intended to detect N gonorrhoeae. They are a form of immunoassay in which the test sample flows along the nitrocellulose membrane via capillary action, and particles such as colloid gold enable a colour reaction. Given the potential benefits described in a number of recent publications and the urgent need to improve the timeliness of diagnosis and treatment, we thought it timely to conduct a review of these two approaches and compare their operational characteristics and performance.
This systematic review was conducted according to the PRISMA statement.16 The PRISMA statement aims to advance the reporting of systematic reviews and meta-analyses by improving the transparency and completeness of information. PRISMA includes a checklist detailing the content to be reported in each section of a review or meta-analysis.
Rapid POC testing was defined as any system that provided rapid gonorrhoea diagnosis results at the POC and could be conducted with minimal operator skill and infrastructure.
PubMed and Embase bibliographic databases were searched to the end of August 2010. Only English language papers were included.
The following terms (and variations/alternative spelling of these terms) were used in the search:
Rapid test, or POC test, or POCT, or LE, or urine dipstick; and
Gonococcal, or gonorrhoea or N gonorrhoeae; and
Evaluation, or performance characteristics, or validation, or performance, or sensitivity, or specificity.
The papers were reviewed and information was extracted by two authors independently. Disagreements were resolved by discussion and consensus.
A paper was included in the review if the performance characteristic of the rapid POC test for N gonorrhoeae was assessed against NAAT or bacterial culture.
Papers were excluded if the rapid POC test was conducted on clinical isolates in a laboratory setting rather than on field samples (such as urine); the papers did not report primary data (reviews, editorials and commentaries); or the paper described the application of an assay without reporting on the performance characteristics of the test.
For each paper that met the inclusion criteria, information was extracted on the setting, participants (age, sex, genital symptoms), evaluation design (sample size, specimen type), rapid POC test (type, manufacturer, brand name, LE dipstick cut-off), gold standard test (NAAT, culture), rapid POC test performance (sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV)) and operational characteristics (number of steps, major equipment, time to result).
We calculated the median sensitivity (the percentage of true positives (as determined by the reference test), which are positive on the POC test), specificity (the percentage of true negatives (as determined by the reference test), which are negative on the POC test), PPV (percentage of positive POC test results that are true positives (as determined by the reference test)) and NPV (the percentage of negative POC results that are true negatives (as determined by the reference test)) for both LE dipsticks and ICT. When two results were available on the same patients due to different LE cut-offs being presented (trace, or +) or types of swabs collected (vaginal, or cervical), we used the 1+ result and the cervical swab results in the median calculations. One study provided a breakdown of results for 18–25-year-old and 26–39-year-old women but no overall result, in this instance we included the younger age group. The two studies of microscopy were clinical audits of confirmed cases and because of the retrospective audit study design only sensitivity is reported.
If any of these outcomes were not reported but sufficient information was available in the paper, we calculated the outcome using Microsoft Excel. When the CI were not reported but sufficient information was available, we calculated a 95% CI using STATA statistical software V.10.17
Our literature search identified 100 papers, 14 studies were included in this review. Papers were excluded for the following reasons:
Diagnosis was not at POC (n=24);
Review editorial or commentary with no original data (n=16);
Testing for N gonorrhoeae and Chlamydia trachomatis combined but did not separately report N gonorrhoeae, or organisms other than N gonorrhoeae or non-genital gonococcal infection (n=20);
Gonorrhoea resistance or a subtyping study (n=9);
Test based on the correct identification of N gonorrhoeae previously isolated by bacterial culture (n=2);
Insufficient information to calculate either the sensitivity or specificity of the POC test (n=12);
Language other than English (n=3).
Key characteristics of the 14 studies are presented in tables 1, 2 and 3. There were nine papers reporting LE dipsticks11 ,18–25 (table 1), three reporting ICT3 ,13 ,26 (table 2) and two studies reporting the use of microscopy27 ,28 (table 3).
There were nine papers reporting LE dipsticks,11 ,18–25 which were published between 1991 and 2010 (median1996). The nine studies represented a variety of countries, with four being conducted in the USA. Five studies were conducted in dedicated STI clinics, departments or genitourinary medicine (GUM) outpatient clinics, and the remaining studies were conducted in a variety of other settings. Five of the nine LE dipstick studies were conducted in male populations. In four of the nine LE dipstick studies, genitourinary symptoms were present in 50% or more of the patients (two studies in women and two in men).
The LE dipsticks included ChemStrip (n=5) (Bohhringer Mannheim, Indianapolis, Indiana, USA (two); Biodynamics Indianapolis, Indiana, USA (two); ChemStrip, Indianapolis, Indiana, USA (one)), Multistix (n=3) (Bayer New South Wales, Australia (one); Bayer Diagnostics, location not specified (one); Miles Laboratories, Elkhart, Indiana, USA (one)) and LE urine dipstick (n=1) (Qiagen, Melbourne, Australia).
In seven of the LE studies, testing was performed on a first void urine sample, into which the test strip was dipped and read by clinic staff. This procedure consisted of three steps: sample collection, dipping and removing the test strip and interpretation of the results. While clinic staff conducted the testing in eight studies, in one study20 an automated urine analyser (Clinitek 10, Bayer Pty. Ltd.) was used. In two other studies LE testing was performed on endocervical swab samples, either mixed with saline24 or directly on vaginal and cervical secretions.25
Of the nine LE dipstick studies, the median sensitivity was 71% (range 23–81%) (nine studies), the median specificity was 70% (range 33–99%) (nine studies), the median PPV was 19% (range5–40%) (seven studies) and the median NPV was 95% (range 56%–99%) (six studies).
Among men, the median sensitivity was 73% (range 71–78%) (five studies) and the median specificity was 93% (range 65–99%) (five studies) and in four studies in women (sample type: urine, cervical and vaginal swabs) the median sensitivity was 61% (range 23–68%) (four studies) and the median specificity was 53% (range 44–70%) (four studies). One study11 found the sensitivity was 23% overall increasing to 75% in a subset of the same women with symptoms.
There were variations among the LE dipstick studies as to the value chosen to be indicative of a positive result. Two studies19 ,20 compared the LE test's performance according to different positive cut-off points (trace, or 1+). Use of the trace cut-off point resulted in higher sensitivity and lower specificity, while the use of the positive (1+) cut-off point resulted in lower sensitivity and higher specificity.
The three studies of ICT were published between 2004 and 2010. The studies were conducted in Benin, Brazil and Japan. Two studies were conducted in STI clinics or urology departments and the other was conducted at a dedicated clinic for female commercial sex workers. Of the ICT studies, the exact percentage of symptomatic patients was provided by the authors of one study (5.4%), and the participants in the other two studies composed of STI clinic attendees who were classified as high risk or symptomatic.
In the three studies involving ICT, the gonococcal check (PATH) evaluation involved testing of cervical and vaginal swabs, required five steps, and results were available in 25–30 min. The NG Thermo Biostar (GC OIA) involved testing cervical swabs, and seven steps, and results were available within 30 min. The Binax NOW (Inverness) evaluation involved testing urine samples, and six steps, and results were available in 25 min.
Sensitivity for the ICT studies were 60%, 70% and 94%. Specificities were 89%, 96% and 97%. PPV were 55%, 56% and 97% and NPV were 92%, 93% and 99% (table 2). In two studies among women, the sensitivity was 70% and 60% and specificity was 97% and 89%, and in the one study in men, the sensitivity was 94% and specificity was 96%.
One ICT study13 compared the use of cervical and vaginal samples to evaluate the performance of the gonococcal check immunochromatographic strip and found the POC had a greater sensitivity (70%) when cervical swabs were tested, compared with vaginal swabs (54%).
Two studies were identified that retrospectively audited clinical records of patients for whom microscopy had been used. The two microscopy studies were conducted in Scotland and England. Both studies were conducted using records from GUM clinic attendees.
In both studies records of confirmed cases of gonorrhoea infection were reviewed to determine whether microscopy was used, and specificity was determined by comparing confirmed cases with microscopy results. In one study the gold standard was culture and the other study did not specify which gold standard was used.
In one study the sensitivity was 38% among female GUM attendees. The other study reported sensitivity separately for different study participants and specimen types. The sensitivity among women (cervical), men who have sex with men (MSM) (urethral), MSM (rectal) and heterosexual men (urethral) were 51%, 89%, 54% and 84%, respectively. Estimates of specificity, NPV and PPV were not applicable to this study design.
This systematic review shows that ICT and LE dipsticks for diagnosis of N gonorrhoeae had a similar sensitivity but results may be an over-estimation due to the inclusion of largely symptomatic patients. These conclusions are limited by the small number of evaluations identified for inclusion in this review.
To our knowledge this is the first systematic review undertaken on this topic, and is timely considering the rapid development of STI POC tests in recent years.15 This review has some limitations to be considered. First, we did not search the grey literature, so it is possible that some evaluations were not identified. Second, many studies did not present the necessary information to enable breakdowns according to age group and symptom status. Finally, due to the heterogeneity of the study populations and differing positive cut-off points, we did not conduct a meta-analysis.
The review showed that leucocyte LE studies had a much lower specificity (and PPV) in women than ICT, which could be due to increased contamination of urine samples in the women by vaginal secretions and/or women being more prone to urinary tract infections when the LE enzyme is a marker of pyuria.29
Based on only one study in the review, which stratified the sensitivity estimates according to the presence of symptoms,11 the LE dipsticks performed better in symptomatic populations than in asymptomatic populations. Although not demonstrated in this review, it is likely that a similar pattern would also be seen with immunochromatographic devices, as a similar phenomenon has been demonstrated for chlamydia ICT.30 The higher sensitivity of the tests in symptomatic patients is likely to be related to greater bacterial load in their specimens.30 This finding should be noted when drawing conclusions about the performance of LE dipsticks. It is likely that the sensitivity of the tests when used for screening would be much lower.
Immunochromatographic devices detect specific target antigens of N gonorrhoeae, LE dipsticks on the other hand detect inflammation as a surrogate marker of various infections, including gonorrhoea. Although a positive LE result may not indicate the presence of N gonorrhoeae, it is important to recognise that this result may signify the presence of other clinically relevant pathogens. In many countries a considerable proportion of patients are coinfected with both gonorrhoea and chlamydia,31–33 and the development of POC tests that detect causal organisms for both of these infections would be ideal. There has recently been progress in the development of POC tests based on nucleic acid amplified techniques, such as GeneXpert for the detection of chlamydia and gonorrhoea, which will soon be available commercially.34 Other tests in the pipeline include the Atlas Velox System, which uses microwave-accelerated metal-enhanced fluorescence technology35 provide inexpensive and fast lysing bacteria, extraction of DNA and quantification of bacteria.
Both the LE dipsticks and the ICT were found to have less than ideal sensitivity compared with the gold standard tests, and will miss many infections if used for screening asymptomatic patients. However, it has been demonstrated mathematically that to have public health benefit, rapid POC tests for detecting chlamydia and gonorrhoea in women do not require the same sensitivity as the gold standard test in identifying infections. This is because the number of individuals with infections that are identified by a rapid test and are successfully treated can potentially outnumber the number of individuals who are tested and return for treatment using a more sensitive culture or NAAT strategy.36 ,37 Furthermore, mathematical modelling has shown that this sensitivity requirement threshold lowers in settings in which there is high potential for onward transmission between testing and result notification, and in settings where failure to return to the clinic to collect results is common.37 Gift et al38 also analysed the potential impact of a two-test algorithm involving a rapid chlamydia POC test followed by a more sensitive laboratory test, and found the strategy to be cost-effective at chlamydia prevalences above 9%. All these models did not, however, account for the extent of syndromic management being undertaken, so the benefits may be over-estimated when syndromic management is systematically undertaken.
In conclusion, the review has highlighted that further development of POC tests for the detection of N gonorrhoeae remains a public health priority. The fact that we could only identify three evaluations of N gonorrhoeae immunochromatographic strip tests suggests there is a lack of suitable tests on the market for this purpose, which is in contrast to a recent systematic review of chlamydia POC tests when 13 studies were identified.39 Furthermore, a review of syphilis POC tests identified 15 studies.40 LE dipsticks are inexpensive and interpretation of results can be performed by minimally trained staff within a short time frame.24 However, they are limited by their lack of specificity in diagnosing N gonorrhoeae in women20 and poor PPV.13
ICT and LE dipsticks for the diagnosis of N gonorrhoeae had similar sensitivities.
ICT had a higher specificity in women than LE.
Despite limitations, both tests may still provide advantages over syndromic management in high-prevalence settings.
These findings highlight the need for improved POC tests for the diagnosis of N gonorrhoeae and more standardised evaluations.
Contributions RG devised the original idea for the study. LAWS and RG drafted the manuscript. RH provided supervision to the lead author and advice on study design and analysis and critical revisions during the drafting phase. DMW, JW, LC, SS, JK and BD all provided technical and clinical advice and critical revision.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.