To assess the accuracy of:

nucleic acid hybridisation or amplification tests, in comparison with culture, when used for the presumptive identification of Neisseria gonorrhoeae (N. gonorrhoeae) in endocervical or male urethral specimens;

amplification tests on urine specimens, in comparison with culture, and amplification tests on endocervical and male urethral specimens;

hybridisation and amplification tests used for specimens from non- urogenital sites, e.g. the pharynx or rectum.

Additional objectives were listed but were not investigated in this review.

MEDLINE was searched from 1992 to October 1997 using the search terms 'gonorrhea' and 'N gonorrhoeae', and the bibliographies of identified studies were scanned. For amplification tests, abstracts published since 1992 from meetings of the International Society for Sexually Transmitted Diseases Research and the American Society of Microbiology, from the International Conferences on Antimicrobial Agents and Chemotherapy, and from the National Sexually Transmitted Diseases Prevention Conferences, were examined. Where studies were reported in both abstract form and as full papers, the full papers were included rather than the abstracts.

Study designs of evaluations included in the review

No inclusion criteria relating to the study design were specified.

Specific interventions included in the review

Studies were eligible for inclusion if they evaluated a second generation nucleic acid hybridisation test (Pace 2 or Pace 2C), or nucleic acid amplification (ligase chain reaction or polymerase chain reaction; LCR and PCR, respectively).

Reference standard test against which the new test was compared

No inclusion criteria relating to the reference standard were specified. The included primary studies used culture as the reference standard test.

Participants included in the review

No inclusion criteria relating to the participants were specified. Studies in which laboratory specimens rather than patient specimens were used for evaluation were excluded.

Outcomes assessed in the review

No inclusion criteria relating to the outcome measures used in the primary studies were specified. Sensitivity and specificity by testing method, gender and anatomic site were the outcome measures calculated by the reviewers.

How were decisions on the relevance of primary studies made?

Two authors identified the studies independently and a third author reviewed the identified studies. Any disagreements were resolved by review and discussion and, when clarification was necessary, the authors were contacted.

The studies were assessed for quality on the basis of six criteria: sample size; test quality; the blinding of investigators to all available test results and clinical information; clinical description; assessment; and the consistent application of a reference test to the diseased and nondiseased populations. The criterion of test quality covered whether the performance of the evaluated test was described in sufficient detail to enable the method to be reproduced, whether the procedure for gonococcal culture was a generally accepted method including the use of selective media and confirmatory testing, and whether the assay included positive and negative controls. The maximum quality score available was 12.

If the gender of the participants was not identified, or there were fewer than five culture-positive participants, a sample-size score was not assigned and the performance results were neither abstracted nor combined, unless specimens were taken from the pharynx or rectum.Two authors scored the included studies independently and a third author reviewed the score results. Any disagreements were resolved by review and discussion and, when clarification was necessary, the authors were contacted.

Two authors extracted the study data independently and a third author reviewed the extraction results. Any disagreements were resolved by review and discussion and, when clarification was necessary, the authors were contacted.

How were the studies combined?

Sensitivity and discrepant analysis-based specificity (where the numerator includes the number negative by both the index and reference standard tests, and the denominator includes the number negative by the reference standard less the number positive by both tests) were calculated for endocervical, urethral and urine specimens, along with 95% confidence intervals. The combined estimates of these subgroups were calculated by summing the numerators and denominators across the studies (i.e. each study was weighted by sample size).

How were differences between studies investigated?

No method for investigating between-study heterogeneity was reported.

Twenty-one studies were included with 17,737 samples (11,619 female, 3,874 male, and 2,244 other or total samples). There were 572 culture-positive female samples, 901 culture-positive male samples, and 92 other or total culture-positive samples.

The maximum study quality score assigned to any included study was 6 out of 12 possible points; this was attained by 2 of the 21 included studies

The sensitivity and specificity of nucleic acid hybridisation (approximately 85% and approximately 98%, respectively) and amplification tests (approximately 95% and approximately 99%, respectively) were high and did not appear to differ substantially by gender or anatomic site.

In detecting gonococcal infections of the endocervix, the sensitivity of nucleic acid hybridisation (Pace 2) was 92.1% and the specificity (after discrepant analysis) was 99.1%. For nucleic acid amplification (LCR), the sensitivity was 96.7% and the specificity (after discrepant analysis) was 99.1%.

In detecting gonococcal infections of the male urethra, the sensitivity of nucleic acid hybridisation (Pace 2) was 96.4% and the specificity (after discrepant analysis) was 98.8%. For nucleic acid amplification (LCR), the sensitivity was 98.6% and the specificity (after discrepant analysis) was 99.97%.

In detecting gonococcal infection in a urine specimen using LCR (nucleic acid amplification), the sensitivity for women was 96.2% and the specificity (after discrepant analysis) was 100.0%. For men, the sensitivity was 98.3% and the specificity (after discrepant analysis) was 100.0%.

Other results.

For vaginal nucleic acid amplification versus cervical nucleic acid amplification, the sensitivity was 100% and the specificity was 99.6%. For rectal nucleic acid hybridisation versus culture, the sensitivity was 96.4% and the specificity was 100%. For pharyngeal nucleic acid hybridisation versus culture, the sensitivity was 77.4% and the specificity was 99.9%.

One study evaluated cost and found that the cost and performance of a Pace 2 nucleic acid detection test was similar to that of culture. The cost of Pace 2 increased as the volume of tests went down and as the proportion of tests needing to be confirmed by probe competition assay went up. The greatest contributor to the total cost of culture was labour, and to that of Pace 2 was materials.

The results of the review suggest that nucleic acid hybridisation and amplification offer high-performance alternatives to gonococcal culture. Conversely, culture remains an excellent test for the diagnosis of gonorrhoea in comparison with newer tests. When proficiency in the performance of culture is high, the new tests are comparable to culture and may not offer a substantial advantage; in settings where optimisation of culture is difficult, nucleic acid amplification may detect more infections than nucleic acid probe or culture.

Whereas a variety of tests for gonorrhoea can be used to detect N. gonorrhoeae in various anatomic sites, only culture can be used for isolation of the agent for antimocrobial testing and retention for medicolegal purposes. The lack of routine antimicrobial susceptability testing may interfere with clinicians' and public health practitioners' abilities to detect and respond promptly to the emergence of antimicrobial-resistant N. gonorrhoeae.

The authors clearly stated their research questions. However, only limited inclusion and exclusion criteria were specified, and the study did not address all of the stated objectives. The literature search was limited in its use of keywords, and may have missed studies published outside the USA and other English speaking countries by focusing the search on MEDLINE. The authors did not report whether there were any further language restrictions on their search or whether they sought unpublished data. The quality of the included studies was assessed, and the authors reported on how the articles were selected and how many of the reviewers were involved in the data extraction. Some details of the included primary studies were tabulated.

The drawbacks of the quality and design of the included studies were discussed, but no formal heterogeneity assessment was reported. In the light of this, the calculation of pooled estimates for sensitivity and specificity by summing the numerator and denominator values across studies was particularly inappropriate. This approach to pooling is always questionable, even where between-study heterogeneity cannot be rejected, since it effectively treats the primary studies as a single study population.

The authors' conclusions should be viewed with considerable caution given the problems with this review outlined above.

Practice: The authors did not state any specific implications for practice.

Research: The authors state that further research is needed. Data are required on the performance of nucleic acid amplification in populations of women and men with a low prevalence and for asymptomatic women and men, in particular for the use of urine screening in asymptomatic men. The roles of inhibitors and sample adequacy also need further evaluation. The duration of positivity of nucleic acid tests and, therefore, their performance as a test for cure, are unknown and require evaluation. All of the tests evaluated in this review remain unapproved and unevaluated in children.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.