Article Text

Download PDFPDF

Pitfalls of internet-accessible diagnostic tests: inadequate performance of a CE-marked Chlamydia test for home use
  1. C-E C Michel1,
  2. F G Saison2,
  3. H Joshi1,
  4. L M Mahilum-Tapay3,
  5. H H Lee1
  1. 1
    Diagnostics Development Unit, Department of Haematology, University of Cambridge, NHS Blood and Transplant Site, Cambridge, UK
  2. 2
    Department of Obstetrics-Gynecology, Western Visayas Medical Center, Iloilo City, Philippines
  3. 3
    Diagnostics for the Real World (Europe) Ltd, Cambridge Science Park, Cambridge, UK
  1. Dr H H Lee, Diagnostics Development Unit, Department of Haematology, University of Cambridge, NHS Blood and Transplant Site, Cambridge CB2 2PT, UK; hl207{at}cam.ac.uk

Abstract

Objectives: To evaluate the performance of a Conformitée Européenne (CE)-marked home test for Chlamydia trachomatis (CT) that is available over the internet.

Methods: A total of 231 eligible women attending the Social Hygiene Clinic (SHC) or Obstetrics–Gynecology (OB-GYN) Clinic in Iloilo City, Philippines were recruited to an evaluation of the HandiLab-C Chlamydia home test (HandiLab-C). One vaginal swab was tested with HandiLab-C on-site and the second in Cambridge, UK with two nucleic acid amplification tests (NAAT), the Roche Amplicor and Abbott m2000. The organism load of NAAT-positive swabs was quantified.

Results: Concordance between the NAATs was high (kappa agreement: 0.984). Using the Abbott assay as the gold standard, the sensitivity and specificity of the Roche assay were 97.4% and 100%, respectively. CT prevalence by Abbott was 8.0% (8/100) in the OB-GYN Clinic and 23.7% (31/131) at SHC. The sensitivity of HandiLab-C was 12.5% (1/8) and 19.4% (6/31) in OB-GYN and SHC respectively, with specificities of 93.5% (86/92) and 88% (88/100) respectively. Overall positive and negative predictive values of HandiLab-C were 28% and 84.5% respectively. No correlation between HandiLab-C performance and organism load (range: 1.3×102 to 1.4×107 bacteria/swab) was observed.

Conclusions: The performance of HandiLab-C is very poor, with the test yielding more false-positive (18/193) than true-positive (7/38) results. It remains accessible via the internet under various brand names and has retained its CE mark. This situation raises serious concerns about the regulation of diagnostic products available via the internet and the standards of certain Notified Bodies that issue the CE mark.

View Full Text

Statistics from Altmetric.com

Most testing for Chlamydia trachomatis (CT) infection is laboratory-based, with results provided in 1–2 weeks. Rapid tests that can be performed in clinics while the patient remains on-site are commercially available but show variability in both sensitivity and ease of use.14 Given the asymptomatic nature of most CT infections, the availability of a high-performance home test for CT would provide an important additional tool to increase coverage of testing.

A CT home-use test is currently marketed via the internet under various brand names, including HandiLab-C, SureScreen and SELFCheck. The test is based on the detection of an enzyme (Peptidase 123A) specific to CT.5 According to the manufacturer (Medical Packaging Corporation, Camarillo, California), this assay is easy to use and has a sensitivity of 98.2% and 96.3% for physician or self-collected vaginal swabs, respectively, and a specificity of 97.9% compared with endocervical samples tested with PCR.6 The test is Conformitée Européenne (CE)-marked as a stamp of quality. The CE marking is based on the In Vitro Medical Device Directive 98/79EC7 implemented by the European Commission. In vitro medical devices (IVDs) may be CE-marked following different routes based on the risk(s) associated with a particular IVD. Chlamydia tests are categorised as Class IIb products, which require the submission of a Technical File to a Notified Body that issues the CE mark. The Technical File gives a summary of the essential requirements according to the IVD Directive including the performance of the test based on the clinical evaluation data and risk analysis on the product. Notified Bodies are designated by the Competent Authority of each member state according to stringent criteria outlined in the directive.

A previous study of the HandiLab-C test with 157 women attending a sexual health clinic in Norway found that the test had a sensitivity of 25% (4/16), a positive predictive value (PPV) of 23.5%, and a negative predictive value (NPV) of 85%.8 In addition, half of the study participants found it difficult to interpret the test results. Its poor performance resulted in the product being taken off pharmacy shelves, first in Denmark and subsequently in Norway.9 10 However, the test is still readily available on the internet with its CE mark in place. Given the large difference between the sensitivity claimed by the manufacturer and that reported in the Norwegian study,8 we evaluated the performance of the HandiLab-C test at sites with a high or low prevalence of CT infection against two nucleic acid amplification test (NAATs): the Abbott M-2000 CT test (Abbott Molecular Division, Abbott Park, Illinois) and Roche Amplicor CT/NG (Roche Molecular Systems, Indianapolis, Indiana). The Abbott m2000 has claims for both self- and clinician-collected vaginal swabs as a sample type, showing sensitivity and specificity ranging from 92.5 to 94.7% and from 99 to 99.8% respectively.11 The Roche Amplicor CT/NG is not licensed for self-collected vaginal swab; however, a published report for this specimen type indicated the sensitivity and specificity to be 90.7–93.3% and 98.8–99.0%, respectively.12

METHODS

Participants

Consecutive women attending the Social Hygiene Clinic (SHC) or Obstetrics–Gynecology (OB-GYN) Clinic at the Western Visayas Medical Center in Iloilo City, Philippines, were recruited during the month of May 2008. Individuals were eligible if they were at least 18 years old, had not taken antibiotics during the previous month, had not used a vaginal cream or feminine wash during the previous 24 h, could understand the provided forms and were willing to participate in the study. Informed consent was obtained from each participant, and the study was approved by the Institutional Research Ethics Committee of the Western Visayas Medical Center.

Specimens and assays

The HandiLab-C is licensed for self-collected vaginal swabs; however, results from self-collected vaginal swab versus physician-collected swabs were presented by the manufacturer on their website,6 showing the latter to be slightly better in performance. To limit sampling variation or errors, HandiLab-C swabs were collected by the clinician according to the manufacturer’s instructions. Briefly, HandiLab-C samples were taken by inserting the swab provided by the manufacturer into the vaginal canal approximately 3–5 cm and completely rotating the swab five times firmly against the vaginal canal.

Two vaginal swabs were collected, the first being tested immediately on-site by the clinician with HandiLab-C strictly according to the manufacturer’s recommendations.5 The second swab was collected with the use of Culturette EZ (Becton Dickinson Biosciences, San Jose, California), frozen at −20°C within 8 h of collection and shipped to Cambridge, UK in dry ice at the end of the study for testing by NAATs. Randomisation of the two swabs for HandiLab-C or NAAT testing would have removed the bias in favour of the former; however, since we were evaluating the performance of the HandiLab-C test, it was decided that the specimens should be tested in the manner recommended by the manufacturer. Given that NAATs are generally acknowledged as far more sensitive than rapid tests, using the second swab for NAAT would have not severely affected the accuracy of its results.

For NAAT testing at Cambridge, frozen swabs were placed overnight in 3 ml of predispensed M4RT medium provided by the manufacturer as part of the Amplicor CT/NG assay kit and tested according to the manufacturer’s instructions. Another portion (200 μl) was transferred into the Abbott multi-Collect Specimen Collection Tube containing 1.2 ml of Specimen Transport Buffer (guandine thiocyanate in Tris buffer). The samples were extracted using the automated Abbott mSample Preparation SystemsDNA using magnetic particle technology and tested by the Abbott m2000rt instrument for PCR amplification and fluorescence detection. Sensitivity, specificity, PPV and NPV of HandiLab-C were determined using the Abbott m2000 as the gold standard comparator.

Organism loads of the specimens found positive by either NAAT were determined according to a previously described method13 14 using pCTL12A plasmid as a standard. Quantitative real-time PCR standard curves were constructed with eight serial tenfold dilutions of the pCTL12A plasmid ranging from 10 to 108 copies of pCTL12A per 200 μl of nuclease-free water transferred into the Abbott multi-Collect Specimen Collection Tube and extracted using the automated Abbott mSample Preparation SystemsDNA, with the decision cycle number provided by the Abbott m2000rt instrument. The standard curves between runs were highly reproducible (r2 = 0.998).

Treatment

NAAT results were made available within 3 weeks from the time of sampling. Participants who tested positive for CT by either NAAT were given a single dose (1 g) of azithromycin or, if pregnant, 500 mg of amoxicillin for 7 days according to the guidelines of the Philippine Department of Health.15 As previous studies have indicated that the results by HandiLab-C were unreliable, treatment decision was based only on NAAT.

Statistical analysis

The 95% CI was calculated by the Wald method, and differences in test performance were evaluated by the χ2 test. A p value of <0.05 was considered statistically significant.

RESULTS

A total of 231 eligible participants were recruited to the study. Most of the SHC participants were commercial sex workers who attended the clinic for a weekly health check, whereas participants at the OB-GYN Clinic were mostly pregnant women attending for antenatal care. The CT positivity rates by the Amplicor assay were 8.0% (8/100) for the OB-GYN Clinic and 22.9% (30/131) for the SHC, consistent with the Amplicor results of a previous study with the same populations at the same sites.15 Of the total of 231 samples from both sites, 38 samples were concordant-positive and 192 concordant-negative by both NAATs. One SHC sample with an organism load of 256 bacteria per swab tested negative by HandiLab-C and Amplicor but gave a repeat positive result with the Abbott assay. Concordance between the NAATs was high (kappa agreement: 0.984). Using the Abbott assay as the gold standard, the sensitivity and specificity of the Roche assay were 97.4% (38/39) and 100%, respectively.

On-site testing of the OB/GYN Clinic population (n = 100) with HandiLab-C detected only one of the eight NAAT-positive samples, giving a sensitivity of 12.5% (95% CI 0.1 to 35.4). For the 92 Abbott-negative samples, HandiLab-C yielded six false-positive results, resulting in a specificity of 93.5% (86/92; 95% CI 88.4 to 98.5). Compared with the OB-GYN Clinic population, the sensitivity of HandiLab-C was somewhat higher for the high-risk SHC population (n = 131), detecting six of the 31 Abbott-positive individuals (sensitivity: 19.4%; 95% CI 5.5 to 33.3, p = 0.70). However, the specificity was slightly lower at 88% (88/100; 95% CI 81.6 to 94.4, p = 0.19) with 12 false-positive results. The sensitivity and specificity of HandiLab-C in the total populations (n = 231) were 17.9% (7/39; 95% CI 5.9 to 30.0) and 90.6% (174/192; 95% CI 86.5 to 94.8), respectively (table 1). The overall PPV and NPV were 28% (95% CI 10.4 to 45.6) and 84.5% (95% CI 79.5 to 89.4) respectively. The kappa agreement between HandiLab-C home test and the Abbott or the Roche assay was poor, with the former at 0.100 (95% CI −0.121 to 0.321) and the latter at 0.105 (95% CI −0.117 to 0.328).

Table 1 HandiLab-C performance versus nucleic acid amplification tests

The organism load in the 39 NAAT-positive samples ranged from 1.3×102 to 1.4×107 bacteria per swab, on the basis of the assumption that each organism harbours seven plasmids.13 14 This finding is consistent with that of a previous study with a similar population.4 There was no apparent correlation between organism load and the HandiLab-C detection rate (table 2).

Table 2 HandiLab-C and positivity rate of nucleic acid amplification tests versus organism load

DISCUSSION

A home test based on easy-to-collect vaginal swabs and with a simple testing procedure has the potential to increase uptake of CT testing, especially among young adults who prefer collecting their own specimens and test accessibility via the internet.16 Consistent with the findings of a previous Norwegian study,8 our results show that the HandiLab-C test has a very low sensitivity and poor specificity in both low- and high-risk populations. Indeed, HandiLab-C yielded more false-positive results than true-positive results, making the test an unreliable tool for diagnosis of CT infection. The poor performance of the test was not due to an abnormally low organism load in the tested populations, with the observed load being in the expected range.4

Despite the previous study showing the poor performance of the test8 and its removal from Danish and Norwegian pharmacies,9 10 HandiLab-C remains available via the internet under various brand names, priced between £11.49 and £18.99, and it has retained its CE mark. This situation raises serious concerns about the regulation of diagnostic products available via the internet and the standards of certain Notified Bodies that issue the CE mark.

Acknowledgments

We thank the patients and staff of the SHC and OB-GYN Clinic of the Western Visayas Medical Center for their participation and assistance in this study, I Clarke for providing the pCTL12A plasmid, and J White and C Nadala for critical review of the manuscript.

REFERENCES

View Abstract

Footnotes

  • Funding: The study was funded by a grant from The Wellcome Trust (081371/Z/06/Z).

  • Competing interests: C-EM, LMM-T and HHL are equity holders of Diagnostics for the Real World Ltd, a spin-off company based on rapid test technologies developed at the University of Cambridge. The University of Cambridge and The Wellcome Trust are also equity holders of the company.

  • Ethics approval: Ethics approval was provided by the Institutional Research Ethics Committee of the Western Visayas Medical Center.

  • Patient consent: Obtained.

  • Contributors: C-EM organised study materials, took part in writing and editing the paper and supervised NAAT assays on matched specimens. FGS was the clinical site coordinator, collected specimens, performed HandiLab-C testing on-site and organised shipment of specimens from the Philippines to Cambridge. HJ performed blinded NAAT assays on matched specimens. LMM-T helped in the design and implementation of the study, was the overall study coordinator and took primary responsibility for writing the initial draft of the manuscript. HHL was the principal investigator of the project, conceptualised the study and contributed to writing and editing the paper.

Request Permissions

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.

Linked Articles