High Performance and Acceptability of Self-Collected Rectal Swabs for Diagnosis of Chlamydia trachomatis and Neisseria gonorrhoeae in Men Who Have Sex With Men and Women

Abstract

Background: 

Identification of sexually transmitted infections (STI) is limited by the infrequent assessment of rectal STI. This study assesses usability of self-collected rectal swabs (SRS) in diagnosing rectal Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG).

Methods: 

In 2006 to 2007, clients of the Amsterdam and South Limburg STI outpatient clinics reporting receptive anal intercourse were asked to fill out a questionnaire and provide SRS. A standard provider-collected rectal swab (PRS) was also taken, and both were tested for CT and NG by a nucleic acid amplification tests. SRS performance was compared with PRS as to agreement, sensitivity, and specificity.

Results: 

Prevalence of rectal CT was 11% among the 1458 MSM and 9% among the 936 women. Rectal NG prevalence was 7% and 2%. In 98% of both MSM and women, SRS and PRS yielded concordant CT test results, for NG agreement was 98% for MSM and 99.4% for women. SRS performance for CT and NG diagnosis was good in both groups and was comparable for both study regions. Slightly more (57% of MSM, 62% of women) preferred SRS to PRS or had no preference; 97% would visit the STI clinic again if SRS was standard practice.

Conclusions: 

Because anal sex is a common practice for MSM and women, and anal STI are frequently present, rectal screening should be an essential part of an STI consultation. SRS is a feasible, valid, and acceptable alternative for MSM and women attending STI clinics, and hence should be considered for other settings as well.

In Brief

Self-collection of rectal specimens is a feasible, valid, and acceptable alternative compared with provider-collection, for men who have sex with men and women attending sexually transmitted infection clinics and should be considered for other settings as well.

Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) both cause widespread sexually transmitted infections (STI) with major public health consequences. Active case finding and early treatment of CT and NG are critical strategies for preventing sequelae and reducing transmission.^1–3

Although STI are commonly tested at endocervical, vaginal, and urethral sites, testing is not standard at rectal and oropharyngeal sites. Exclusion of these body locations can leave many infections unnoticed.^4,5 In women, the risk for rectal STI is often overlooked. Data from our Amsterdam and South Limburg STI clinics show that in 37% to 66% of women who have rectal CT or NG, the same STI was not present genitally (unpublished data). Heterosexual anal intercourse is often reported, while condom use is limited. In the United States, 40% of men and 35% of women aged 15 to 44 years old reported heterosexual anal intercourse in 2002.⁶ In the Netherlands, anal intercourse has doubled over the past decade, reaching 15% and 14% in women and men aged 12 to 25 years, respectively.^7–9 In 2007, a national Dutch STI clinic registration system (registering 78,062 consultations) showed the proportion of rectal infections in the total number of CT and NG infections has increased from 11% and 22% in 2005 to 12% and 30% in 2007.¹⁰

Considering that anal sex and rectal STI are not uncommon in both men who have sex with men (MSM) and women, there is a clear need for multisite testing for CT and NG when anal sex is reported.^11–13 Testing strategies based on the reporting of rectal complaints would probably not suffice to reduce the number of unnoticed rectal infections. Symptoms are partly nonspecific and often silent, with about 85% of rectal CT and NG infections being asymptomatic in MSM.⁴ Diagnostic means are needed that are minimally invasive, acceptable for clients, and easy to implement in both clinical and nonclinical settings. Because nucleic acid amplification tests (NAATs) are highly sensitive and specific for the detection of CT or NG in endocervical, vaginal and urethral swabs (patient- or provider-collected), and first-catch urine samples, the self-collection of rectal swabs may be an excellent alternative to those taken by providers.^14,15 This private and simple noninvasive procedure initiates an opportunity to enhance active case finding and case recognition in both clinical and nonclinical settings. Studies have shown that rectal swab specimens were valid for detection of CT and NG (although still not officially included in manufacturers’ protocols).^16–18 The aim of the current study is to evaluate the validity, feasibility, and acceptability of self-collected rectal swabs (SRS) compared with health care provider-collected rectal swabs (PRS) for the diagnosis of anal CT and NG in both MSM and women who report receptive anal intercourse.

MATERIALS AND METHODS

Study Setting and Population

The STI outpatient clinics of the Public Health Services of Amsterdam (approximately 26,000 new consultations per year) and South Limburg (approximately 5000 new consultations per year) both offer free-of-charge examination and treatment for STI. All visitors reporting receptive anal intercourse within the past 6 months are routinely screened for rectal CT and NG, for which swabs are taken by a specialized STI nurse. At the Amsterdam STI clinic, proctoscopic examination is performed as well and swabs are taken during this examination. For the current study, we asked participation and verbal or written consent, of MSM and women who attended the clinics in 2006 (South Limburg) and 2007 (Amsterdam) and reported receptive anal intercourse in the past 6 months. The study was permitted by the Medical Ethical Committees of the University of Maastricht and of Academic Medical Center, Amsterdam, The Netherlands.

Specimen Collection, Diagnostic Tests, and Treatment

Participants provided a SRS in addition to the PRS. STI nurses gave them a diagram and oral instructions about how to take a SRS, i.e., to insert the swab 2.5 cm into the rectum, rotate for 5 to 10 seconds, and place the swab in a capped tube.

All matching swabs were tested for CT and NG by the same NAAT (Cobas Amplicor, Hoffman–La Roche, Basel, Switzerland, or SDA, Becton and Dickinson Diagnostics, NY). Because positive NG results by Cobas Amplicor may often include false-positives, we confirmed all positive NG results by a real-time PCR developed in-house.^19–22 For SDA confirmation, all negative results around the cut-off (for CT), and all positive results (for NG) were retested in the next test round on the original sample and internal controls were used. Specimen were tested within 36 hours. For Amsterdam testing occurred by Cobas Amplicor at the Public Health Laboratory, situated in the same building as the STI clinic. For South Limburg, testing occurred by Cobas Amplicor at the microbiologic laboratory of Atrium, Heerlen and testing by SDA at the microbiologic laboratory of the academic hospital, Maastricht. In South Limburg, none of the samples were inhibited. In Amsterdam, for the purpose of the study, when inhibition occurred, further diagnostic dilution procedures were not carried out to completion. Individual patients test results were communicated based on the standard method of material collection (PRS and culture). Participants who tested positive for rectal CT or NG were treated according to the standard treatment protocol of the STI clinic. Sexual partners of the last 6 months were treated as well, if partner notification was successful.

Questionnaire

To evaluate acceptability, all participants were asked to fill out a questionnaire concerning demographic data (e.g., sex, age, nationality), how they experienced use of both swab methods, and their preferred method. For non-Dutch participants in Amsterdam, the questionnaire was available in English.

Statistical Analyses

To evaluate validity, agreement, κ (>0.8 considered good test agreement), sensitivity, and specificity of SRS compared with PRS were calculated. Positive and negative predictive values (NPV) were calculated for SRS. Additionally, to examine the performance of SRS in a virtual situation in which both PRS and SRS were used, we constructed a virtual gold standard. Here, a result was considered positive when at least one of the NAAT tests was positive. Prevalences, including confidence intervals, were assessed for both PRS and SRS results. Questionnaires were analyzed by using Pearson χ² test for independence and, in case of small numbers with Fisher exact test. A P <0.05 was considered statistically significant. Analyses were performed with the SPSS package version 15.0 (SPSS Inc., Chicago, IL).

RESULTS

Study Population

Of 2432 MSM and 1939 women who were eligible for this study, 1458 (60.0%) MSM and 936 (48.3%) women participated (study population: N = 2394). Compared with those who did not participate, those who did were more often of Dutch nationality and, if women, more often had a current HIV test (P <0.05). We asked a small group of clients who were asked to participate but refused (N = 66), their reasons for not participating. Of those recorded, the most common was fear of taking the swab incorrectly by 56% (N = 37). Finding instructions unclear was mentioned by 9% (N = 6) and other reasons included current complaints of pain (N = 4), no time for the extra procedure (N = 3), and dislike for self-collection (N = 7). Characteristics of the study population are presented in Table 1.

TABLE 1: Characteristics of the Study Population (N = 2394)

Because of PCR inhibition at the first diagnostic step (N = 47 for CT MSM, N = 35 for CT women and N = 31 for NG MSM, N = 33 for NG women), and later start of NG PCR in Amsterdam, not all participants of the study had complete results for CT PCR and NG PCR. Therefore, for comparison of SRS and PRS, 1411 MSM and 901 women were included for CT, and 929 MSM and 697 women for NG.

Prevalence of Rectal CT and NG

Prevalence of CT and NG showed no statistical difference when assessed with PRS or SRS. CT prevalence among MSM was 10.8% (95% CI: 9.3–12.6) and 10.5% (95% CI: 9.0–12.2) for PRS and SRS, respectively. Among women, CT prevalence was 9.4% (95% CI: 7.7–11.5) and 9.3% (95% CI: 7.6–11.4), respectively.

For NG, the overall prevalence among MSM was 7.1% (95% CI: 5.6–8.9) for PRS and 7.9% (95% CI: 6.3–9.8) for SRS; among women, it was 1.9% (95% CI: 1.1–3.2) for both PRS and SRS.

SRS Test Performance and Validity

Agreement between PRS and SRS was very good for both NG and CT, being higher than 97% and showing few discrepant results (Table 2). In addition, other measures of agreement, such as κ, revealed good agreement between PRS and SRS (all κ >0.8). To obtain more insight into test performance, we calculated sensitivity, specificity, NPV, and positive predictive value (PPV) for SRS compared with PRS (Table 2). Results show that, compared with the standard practice of care (PRS as gold standard), PPV is in all cases above 80%.

TABLE 2: Performance of Self-Collected Rectal Swabs in Detection of CT and NG in MSM and Women

Additionally, to examine the performance of SRS in a virtual situation where SRS and PRS are combined for diagnostics, the constructed virtual gold standard was used. Compared with this standard, SRS picked up 89% of all positive CT results in both MSM and women, whereas PRS picked up 92% and 90% of all positive CT results. For NG, SRS was positive in 90% and 87% of positive samples and PRS was positive in 81% and 87% samples among MSM and women, respectively. Of note, performance results were comparable for both study regions.

Almost all samples (98%) were tested by Cobas Amplicor, only 36 complete sets (28 MSM, 8 women) of PRS and SRS were tested by SDA. When only considering results of Cobas Amplicor, the findings in Table 2 did not change except for NPV in MSM tested for CT, which increased from 98% to 99% and PPV in MSM tested for NG, increased from 79% to 80%. When comparing the data obtained from sets that were tested with SDA, SRS, and PRS test agreement ranged from 93% to 100% for CT and NG testing in women and MSM (data not shown).

Questionnaire on Acceptability

Of all participants, 1151 MSM (80%) and 694 (74%) women responded to questions on SRS feasibility and acceptability. Questionnaire responders did not substantially differ from nonresponders, except that MSM who filled out the questionnaires were older (P <0.01) than MSM who did not.

Of the total, 94% of MSM and 95% of women would use SRS again, and 97% of both MSM and women would visit the STI clinic again if SRS were the standard test. The SRS procedure seemed easy to 94% of MSM and 95% of women, and comfortable to 91% of MSM and 87% of women (P = 0.03); 97% of both groups found test instructions clear.

If given a choice between SRS and PRS, 50% of MSM and 57% of women would prefer SRS, whereas 44% of MSM and 38% of women would prefer PRS (P <0.01). Of all respondents, 7% of MSM and 5% of women had no preference. Preference for either collection method did not differ by nationality or age. Those who preferred SRS cited it is being more private and physically comfortable (44% of MSM and 48% of women) or cited SRS is easier/quicker (52% of MSM and 49% of women). Those who preferred PRS mainly cited the feeling that a nurse or other healthcare provider would be more expert in taking samples (the reason cited by 74% of MSM and 76% of women).

DISCUSSION

This study demonstrated that self-collection of rectal specimens (SRS) is an effective method for the detection of rectal CT and NG in MSM and women. This is in accordance with a recent study among MSM, were the Gen-Probe Aptima Combo 2 is used for detecting rectal CT and NG.²³ We showed that SRS is an appropriate alternative to the standard approach of provider-collection of specimen (PRS) in terms of test performance, traced infections, feasibility, and acceptability. Agreement between SRS and PRS is excellent (>97% and all κ >0.8), although some constraints must be taken into account when interpreting test results from either approach. The absence of a gold standard impedes any definite conclusion on the performance of either test versus reality. This is a well known problem for studies comparing performance of different tests to reality, especially in the area of CT diagnosis.^18,24 However, absence of a gold standard is only a small obstacle in this study because we sought an alternative to the PRS procedure that is currently in use at our clinic. We compared 2 different methods of sample collection using otherwise exactly similar diagnostic procedures.

Our creation of a virtual gold standard, in which at least one positive test result in SRS or PRS indicated infection, allowed us to obtain more insight into the performance of both collection methods. Again, since both methods may be slightly imperfect, using the virtual gold standard has the limitation of reducing all false-positives to zero; no conclusions can be drawn on specificity, and sensitivity may be overestimated.²⁴ Therefore, we compared the results of SRS and PRS only between each other in this virtual setting, where SRS picked up ≤3% CT positives in both risk groups, a similar number of NG positives in women, and 9% more NG positives in MSM. None of the differences were statistically significant, again suggesting that both collection methods perform equally.

Nevertheless, current procedures using NAATs are superior to those using culture, and NAATs can improve our ability to diagnose rectal infection with CT or NG, although not yet licensed for such use.^16,18 A problem with NAATs for detection of rectal NG is cross-reaction with related Neisseria species resulting in false-positive results.²⁵ Because we confirmed all positive Cobas Amplicor NG results by a real-time PCR developed in-house, this is of limited importance in our study. Furthermore, if there still is cross-reaction, presumably it would be equal for SRS and PRS and not influence the comparison between the 2 methods.

Our study regions differed in level of urbanization, risk group composition, and CT prevalence (according to PRS: 11% and 9% in Amsterdam and 8% and 6% in Limburg among MSM and women, respectively). Results indicated that self-collection can be used in both higher and lower prevalence settings, since SRS was in all settings feasible, well accepted, and test performance was similar in the 2 study regions (data not shown).

This study and others show that test instructions, especially when illustrated, can be made sufficiently clear for patients.²⁶ However, patients need to be clearly informed that SRS is an approved method in STI examination, since some of our study participants preferred PRS due to feeling that provider-collection would be more expert. Doubt of their own capability was also an important reason among those who refused to take part in the study. Still, was found that such doubt would not deter the majority of clients to come to the clinic if SRS were to become the standard of care. It should be noticed that the median age of the female population is 26 years old and may not reflect the young adolescent’s perception of SRS. However, we have no reason to assume that acceptance in this young group is lower because we did not find an age-effect and other methods of self-collection are very well accepted by adolescents.²⁷

The study population can be considered representative of all participants who reported receptive anal sex at the STI clinics in the last 6 months, except that women of non-Dutch nationality were more inclined not to participate. Of note, no determinants were found to be related to nonacceptance of SRS in questionnaire respondents. The relatively low participation rate of the study (60% MSM, 50% women) and doubts mentioned earlier might suggest a general unease with rectal STI examination which underlines the need for alternative noninvasive diagnostics. However, the participation rate may largely be explained by time-pressure at the STI clinic that prevented all visitors from being invited to take part. Unfortunately, we have no data on the number of eligible participants who were actually asked to participate.

SRS can play an important role in controlling CT and NG transmission, considering that anal sex and rectal STI are not uncommon in both MSM and women and considering, as well, the often asymptomatic nature of these infections.¹⁸ Of course, self-sampling cannot always replace nurse evaluation and proctoscopic examination. For specific risk groups, it is important to inspect for anal warts or inspect the anal mucosa for lesions, discharge, or abscesses.^28,29 However, application of self-collection methods is especially valuable in settings that include risk groups for rectal STI but do not routinely offer rectal screening for CT and NG (e.g., the general practitioner’s office; outreach activities).³⁰ The noninvasive character of SRS can overcome the reluctance of at-risk populations to seek appropriate care. Besides, SRS can be performed at home and might be a good addition to the novel STI prevention strategies that use the internet.^31,32

CONCLUSIONS

The SRS is a valid, feasible, and acceptable approach in sample collection. In a clinical setting, it can be an additional alternative for patients who would otherwise be reluctant to undergo rectal examination. In a nonclinical setting, it can be an important public health tool to enhance active case finding and case recognition where it can be used for routine screening.