Evaluation of endocervical, first-void urine and self-administered vulval swabs for the detection of Chlamydia trachomatis in a miscarriage population

Authors


Dr S. Logan, Department of Obstetrics and Gynaecology, Aberdeen Maternity Hospital, Foresterhill, Aberdeen, AB252ZD UK.

Abstract

Objective  To compare, in parallel, different approaches of opportunistically screening women with bleeding in early pregnancy for Chlamydia trachomatis.

Design  Cross-sectional study.

Setting  Early pregnancy assessment unit, University hospital, Scotland, UK.

Population  Over 200 consecutive women admitted to an early pregnancy assessment unit were recruited. All had a positive pregnancy test, a history of vaginal bleeding and were less than 24 weeks of gestation. Women with recent antibiotic use, heavy vaginal bleeding and cervical shock excluded.

Methods  Each women provided two or more of the following specimens: a self-administered vulval swab, first-void urine and/or endocervical swab. Following screening, each completed a semi-structured questionnaire assessing the acceptability of each method undertaken.

Main outcome measure  Subjective rating of the screening methods; prevalence; method performance, including proportion requiring repeat testing.

Results  The majority accepted screening, with moderate prevalence rates (95% CI) 3.9% (2.0–7.4%) identified. All positive women were less than 30 years of age. Parallel screening exposed the potential of reduced test performance with urine. Non-invasive sampling was more acceptable, but more likely to require repeat testing.

Conclusion  Both acceptability and the effect of bleeding on test performance need further assessment before a particular specimen can be recommended for screening this population of women for C. trachomatis.

INTRODUCTION

Chlamydia trachomatis is the most common bacterial sexually transmitted infection in the UK and Europe. Most female infections are asymptomatic and recognised sequelae include pelvic inflammatory disease, which can lead to ectopic pregnancy, chronic pelvic pain and tubal factor infertility. Miscarriage affects 12–20% of pregnancies. Women with unrecognised chlamydial infection, who experience miscarriage, run the risk of ascending infection, irrespective of whether medical or surgical evacuation is required. For those who do not miscarry, infection can predispose to postpartum pelvic inflammatory disease and the neonatal complications of conjunctivitis and pneumonia.

Nucleic acid amplification assays have revolutionised screening for Chlamydia, enabling the use of non-invasive sampling such as first-void urine and vulval swabs, in addition to traditional endocervical sampling. Patients presenting with bleeding in early pregnancy are ideally placed for opportunistic screening for C. trachomatis infection. An earlier study undertaken in our department found high sensitivities and specificities using these specimens to screen antenatal patients.1 However, these various specimen options have not been assessed in a miscarriage population.

METHODS

Between September and December 2001, consecutive women admitted to an early pregnancy assessment unit were invited to participate as part of a BSc (Medical Science) project. Inclusion criteria included a positive pregnancy test, a history of vaginal bleeding and less than 24 weeks of gestation. The use of antibiotics in the previous four weeks and signs and symptoms of heavy vaginal bleeding and/or cervical shock resulted in exclusion. The study was approved by the local ethics committee. For those who declined to participate, age and reason for refusal were recorded.

Each woman was asked to provide two or more of the following specimens: a self-administered vulval swab, first-void urine and/or an endocervical swab. The vulval and endocervical specimens were collected using a polyurethane-tipped swab (CULTURETTE DIRECT; Becton Dickinson). On viewing the cervix with a speculum, a cleaning swab was first used to remove excess blood/mucus from the endocervix. A second swab was rotated firmly against the wall of the endocervix. The vulval swab was inserted into the vaginal introitus and rotated firmly against the vaginal wall. Women were instructed to collect first-void urine directly into clean universal containers. Blood contamination of the swabs and urine was assessed visually and by using a Multistix dipstick, respectively, as blood may cause both indeterminate and false negative results in both urine and swabs (BD ProbeTec ET system laboratory manual). Samples were either transported to the microbiology laboratory immediately or stored overnight in a refrigerator at 4°C. Each of the samples was tested according to the manufacturer's instructions. Results (positive, negative, equivocal or indeterminate) were issued for each specimen. A specimen recorded as negative on initial testing was reported as negative. All positive results were re-tested and in view of the published data reporting the reliable specificity of the BD Probe Tec assay,2 a confirmed positive result was considered to be a true positive, that is, a woman was reported to have a C. trachomatis infection on the basis of a single confirmed positive specimen, as done in routine practice. Non-confirmation resulted in the issue of an equivocal report. The test incorporated an amplification control for each sample and an indeterminate result was reported if there were substances within a sample that inhibited amplification.

Following screening, each woman completed a semi-structured questionnaire pertaining to acceptability of the different screening methods.

It was anticipated that a total of 200 women would detect a minimum difference of 8% in acceptability among the three methods with 95% confidence, assuming at least 70% acceptance of all three methods of sampling. These numbers would also allow detection of a 5% difference in positive results among the sampling methods. Data analysis was carried out using the computer package SPSS 10.14 for Windows. For normally distributed data, means and standard deviations were calculated and compared using the Student's t test. Confidence intervals (95%) were reported where appropriate. Unpaired categorical variables were compared using the χ2 statistic with Yates' continuity correction or Fisher's exact test. Statistical significance was at the 5% level.

RESULTS

Three hundred and ten women were invited to participate. None were clinically unstable, but 21 had used antibiotics in the past four weeks and were excluded. Seventy-seven (27%) women declined to participate. Their reasons for declining included the following responses (number of women): ‘I have enough to deal with already’ (17); ‘I want to go home’ (9); ‘I would be willing to give one sample’ (5); ‘I have been tested before’ (4); ‘I do not think I have C. trachomatis’ (2); ‘I'm bleeding’ (1); ‘I don't want anything done’ (1); and ‘I would if I had to give the samples anyway’ (1). No reason was stated by 37 women. Three women were excluded as they did not leave appropriate samples, leaving a study population of 209 women. The mean ages (SD) of the participants and decliners were 29.3 (5.9) and 29.7 (6.1) years, respectively, which were not significantly different (P= 0.6). One hundred and three (49%) and 106 (51%) of the participants were nulliparous and parous, respectively. Only 11 (5%) women reported a past history of Chlamydia.

Three endocervical swabs were not taken by staff. Samples from two women (one set of endocervical, vulval and urine specimens and one set of vulval and urine specimens) leaked and could not be assayed. This left 207 women with two or more specimens available for assay. One hundred and thirty-nine (67%), 205 (99%) and 205 (99%) provided endocervical, self-administered vulval swabs and first-void urine specimens. Sixty-eight women (33%) agreed only to non-invasive screening.

Eight out of 207 women were confirmed positive for chlamydial infection. The overall prevalence (95% CI) was 3.9% (2.0–7.4%). The prevalence was higher in women under 25 years and teenagers −12.2% and 6.7%, respectively. All women with a positive test were under 30 years of age.

The endocervical swab, self-collected vulval swab and first-void urine detected three out of four (75%), eight out of eight (100%) and three out of eight (38%) confirmed positive women, respectively (Table 1). Blood contamination was reported in 72%, 48% and 55% of the endocervical, vulval and urine samples, respectively. Blood contamination of urine was found in four out of five negative first-void urine from women with positive results in other specimens. This compared with one out of three positive first-void urine from women with positive results in other specimens, but was not statistically significant. Blood contamination of urine specimens did not appear to inhibit the assay, as the urine from positive women were reported either as negative or positive, rather than indeterminate.

Table 1.  Results reported for each specimen type and number requiring re-testing. Values are presented as n (%).
Result reportedSpecimen type
Endocervical swab (n= 139)Vulval swab (n= 205)Urine (n= 205)
  1. Urine = first-void urine.

Negative135 (97)188 (92)191 (93)
Positive3 (2)8 (4)3 (2)
Equivocal1 (1)3 (1)2 (1)
Indeterminate6 (3)9 (4)
Samples requiring repeat testing1 (1)9 (4)11 (5)

From a total of 549 specimens tested, 21(4%) were reported as equivocal or indeterminate, with the laboratory requesting a repeat specimen (Table 1). An equivocal result was reported in 1% of each specimen type. Non-invasive (urine and vulval) specimens accounted for all of the indeterminate (inhibited) samples (P= 0.02). Overall, 10% of the women were requested to provide an additional specimen.

Two hundred and nine women completed the acceptability questionnaire. Opinions regarding endocervical, vulval and urine sampling were reported by 140, 207 and 207 women, respectively. When asked to state their preferred method for screening, 4/140 (3%), 61/207 (30%), 144/207 (70%) women chose the endocervical, vulval and urine method, respectively. Urine was significantly preferred compared with the vulval (P < 0.0001) and endocervical (P < 0.0001) methods and the vulval method was significantly preferred when compared with the endocervical swab (P < 0.0001).

The comments reported by the women who declined the endocervical method highlighted five themes—firstly, the physically negative aspects: ‘Unpleasant, irritating, uncomfortable. Tense at examination, hard for nurse’ and ‘midwife was unable to find the cervix’; secondly, the positive aspects of non-invasive testing: ‘Preferred to do testing myself. Other methods straight forward and simple to undertake’ and ‘Don't see the need if urine works’; thirdly, a recent internal examination which they did not wish it repeated: ‘Had extensive internal examinations recently, emotionally and physically prefer not to’; and fourthly, felt psychologically unable to cope with the procedure: ‘Too much for one day’. Finally, a minority were concerned that the pregnancy would be affected by the procedure: ‘After having brown staining in pregnancy do not want cervix tampered with’ and ‘…fear of damaging baby’.

Only two women commented on why they declined a non-invasive method. The vulval swab decliner felt psychologically unable to cope with the procedure: ‘Didn't feel it was appropriate. I think I had enough to deal with, sorry’. The woman who declined the urine specimen was ‘Not able to go to the toilet’.

DISCUSSION

This study provides much needed information on infection rates, test performance and acceptability of different screening approaches when undertaking opportunistic chlamydial screening in a miscarriage population. A DNA based test was used and results were reported as in routine clinical practice. The inclusion of refusal rates along with reasons and the acceptability assessment captured women's perceptions of the screening process and the acceptability of each individual screening method.

Regarding weaknesses, it would have been preferable to have only recruited women who consented to all three methods, but would have reduced the study population by a third, while increasing bias. The low number of positives reflected the study design, but meant that confidence intervals relating to test performance were wide and a statistically significant conclusion comparing these could not be made. Regarding acceptability, as the methods were evaluated in parallel, each would have influenced the others to some degree. The addition of ‘no preference’ might have changed the acceptability weighting. Finally, the acceptability of the endocervical method was biased favourably as only those who agreed to be screened by this method completed the acceptability assessment.

In comparison, Oakeshott et al.3 screened over 1200 pregnant women before 10 weeks of gestation by self-administered vulval swab and first-void urine. Prevalence overall and in women under 25 years was not significantly different. Their vulval swab and urine identified 97% and 90% of the positive women, respectively, but women were excluded if they had vaginal bleeding. Furthermore, their DNA based test did not report inhibition and positives were confirmed by a less sensitive antigen detection test. By not recording the proportion of unconfirmed specimens, avoiding blood contamination and using a test that did not detect the presence of inhibitors, equivocal and indeterminate results would have been eliminated from the test performance equation, resulting in an elevated reported sensitivity. Their acceptability evaluation found that 47% preferred urine, 5% preferred the self-collected vulval swab and 48% stated no preference. In terms of the effect of blood on test performance, Mahony et al.4 identified the presence of haemoglobin as inhibitory to transcription-mediated amplification, but the BD ProbeTec ET system was not assessed.

Women with bleeding in early pregnancy are at increased risk of ascending infection due to the therapeutic interventions used to evacuate the uterus and are ideally placed for opportunistic C. trachomatis screening. Despite recommendations to screen sexually active women under 25 years,1 a recent Scottish audit reported that only 20% of women younger than 25 years with miscarriage were screened for C. trachomatis.5 Furthermore, only 2 out of 15 hospitals had a local guideline for Chlamydia screening within their early pregnancy assessment unit.5 This study found moderate prevalence rates and a policy of screening only women under 30 years would have identified all positive women, screening only 49% of the population.

Despite possible anxieties of offering screening for a sexually transmitted infection during a distressing time, the majority of women accepted chlamydial screening, particularly by non-invasive approaches. For those who felt unable to cope with the initial offer of screening, it could be re-offered at a less emotive time or prophylactic antibiotic cover considered, particularly if surgery was contemplated.

The availability of nucleic acid amplification tests is now widespread, but their weaknesses are perhaps not recognised. This study revealed two potential problems with screening a miscarriage population by non-invasive means. Firstly, first-void urine compared poorly with both the endocervical and vulval specimens in terms of test performance. There is evidence that urine has the lowest bacterial load compared with endocervical and vulval specimens,6 and actions such as wearing sanitary protection, sampling midstream or recent urination and wiping would adversely affect bacterial load. Blood contamination did not appear to cause inhibition, but may have resulted in false negative reports, uncovered by the parallel study design. Exact conclusions, however, cannot be made, as the study was not powered to address this issue. Secondly, non-invasive samples accounted for all of the indeterminate specimens. Re-testing is both inconvenient for women and creates a significant amount of extra workload for both ward and laboratory staff.

Regarding acceptability, negative comments surrounded endocervical sampling suggest that Chlamydia screening in an early pregnancy assessment unit would have to have a non-invasive option for women not undergoing a speculum examination for clinical reasons. The overall preferred sample method was first-void urine, but the self-collected vulval swab identified 50% more infections than urine, with no specific concerns reported regarding this approach. A randomised comparison to urine would assess acceptability more accurately, but the issues of test performance and inhibition of non-invasive specimens need addressing first.

CONCLUSION

The majority of women attending an early pregnancy assessment unit accepted C. trachomatis screening. Moderate rates of infection were identified, with all positive cases found in women younger than 30 years of age. Parallel screening exposed the potential of reduced test performance with urine. Non-invasive sampling approaches were more acceptable, but were more likely to be inhibitory to the assay and require repeat testing. Both acceptability and the effect of bleeding on test performance need further assessment before a particular specimen can be recommended for screening this population of women.

Accepted 27 May 2004

Ancillary