Article Text
PDF
Abstract
Objectives There has been considerable discussion about anorectal Chlamydia trachomatis (CT) in women, with some calling for anorectal CT screening, but little about anorectal Neisseria gonorrhoeae (NG). Given that urogenital NG is more strongly associated with pelvic inflammatory disease, this is an evidence gap. This systematic review and meta-analysis investigates the associations between anorectal CT in women and CT positivity at other sites (urogenital/oropharyngeal) and with anal intercourse, and compares these with anorectal NG within the same study populations.
Methods Electronic databases were searched for English-language studies published to October 2018 using the following terms: (“Chlamydia” OR “Chlamydia trachomatis”) AND ((“anal” OR “rect*” OR “anorect*”) OR (“extra?genital” OR “multi?site”)). Studies were included if anorectal NG data were available. Random-effects meta-analyses calculated pooled estimates; heterogeneity was investigated using meta-regression.
Results 25 studies were eligible. Anorectal CT positivity ranged from 0% to 17.5%, with a summary estimate of 8.0% (95% CI 7.0 to 9.1; I2=88.5%). Anorectal NG positivity ranged from 0% to 17.0%, with a summary estimate of 2.1% (95% CI 1.6 to 2.8; I2=92.7%). The association between urogenital and anorectal positivity was stronger for NG than CT (summary prevalence ratio (PR)=89.3 (95% CI 53.1 to 150.3; I2=80.1%), PR=32.2 (95% CI 25.6 to 40.7; I2=70.3%), respectively), and between oropharyngeal and anorectal positivity it was stronger for NG than CT (PR=34.8 (95% CI 10.2 to 118.2; I2=89.9%), PR=8.8 (95% CI 6.8 to 11.5; I2=58.1%), respectively). Anal intercourse was associated with anorectal NG (PR=4.3; 95% CI 2.2 to 8.6; I2=0.0%) but not with anorectal CT (PR=1.0; 95% CI 0.7 to 1.4; I2=0.0%).
Conclusions Anorectal CT is more common than anorectal NG, but anorectal NG is more strongly associated with anal intercourse, urogenital and oropharyngeal NG, suggesting that ongoing discussion about anorectal CT should also include NG. Longitudinal data are required to further understanding of the aetiology of anorectal STIs and assess whether anorectal screening is needed in women.
Trial registration number CRD42df017080188.
- chlamydia trachomatis
- neisseria gonorrhoeae
- women
- systematic review, meta-analysis
Statistics from Altmetric.com
Key messages
Anorectal chlamydia in women has been widely discussed because of its possible role in causing urogenital infection potentially leading to pelvic inflammatory disease (PID).
In contrast, little is published about anorectal gonorrhoea in women despite urogenital gonorrhoea being more strongly associated with PID.
Although rectal gonorrhoea is less common than rectal chlamydia, it is more strongly associated with urogenital infection, pharyngeal infection and anal intercourse.
Longitudinal data collection is urgently required to further our understanding of the aetiology of anorectal Chlamydia trachomatis and Neisseria gonorrhoeae and inform rectal screening guidelines for women.
Introduction
Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) are two commonly diagnosed bacterial STIs, with overall CT diagnosis rates usually fourfold to fivefold higher than NG in high-income countries.1 In the last 5 years, overall CT rates in the USA and Australia have increased marginally by up to 20% but have declined by 6% in England during this time.2–4 In contrast, overall NG rates increased between 2013 and 2017 by up to 77% in these countries.2–4
Both CT and NG can cause pelvic inflammatory disease (PID) in women, although data linkage studies suggest that the risk of PID is at least twofold higher with NG than CT infection.5 However, there has been much less published research investigating the upper genital tract consequences of NG and its role in PID.
In recent years, there has been considerable discussion about the prevalence and role of anorectal CT in women, with three reviews reporting high median anorectal CT test positivity estimates ranging from 6.0% to 9.2%.6–8 Further, there have been calls for anorectal CT screening in women.9 However, the clinical significance of anorectal CT in women is unclear, with some questioning whether it can cause urogenital infection by autoinoculation, thereby leading to upper genital tract infection.10 Some question whether anorectal CT test positivity in women represents true infection, or a false positive result due to contamination from the genitals, particularly as anal intercourse has not been shown to be associated with anorectal test positivity in women.8
In contrast, there has been very little published about anorectal NG in women, and given that urogenital NG is more strongly associated with PID this is a considerable evidence gap. We conducted a systematic review and meta-analysis to investigate whether anorectal CT positivity in women is associated with positivity at other infection sites (urogenital and oropharyngeal) or with anal intercourse, and to compare these with infection sites associated with anorectal NG in the same populations.
Methods
Search strategy
The study protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO), and the results are reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines11 (online supplementary material 1). We searched for peer-reviewed studies reporting on anorectal CT infection in women published up to the end of October 2018. The search was performed on PubMed, EMBASE and MEDLINE electronic databases. The search terms were (“Chlamydia” OR “Chlamydia trachomatis”) AND ((“anal” OR “rect*” OR “anorect*”) OR (“extra?genital” OR “multi?site”)) (online supplementary material 2). Medical subject headings were used where possible. Citation lists were hand-searched for additional references.
Supplemental material
Eligible studies were those in humans aged 15+ years old, published in English and providing original data on CT and NG anorectal test results for women. Ineligible studies were those reporting exclusively on urogenital infection, those reporting no anorectal NG data and those conducted in men only. Review and opinion pieces or when the sample size was under 10 were excluded. CT positivity was defined as a detection by nucleic acid amplification test (NAAT) or DNA hybridisation probe. Studies using only culture or immunofluorescent assay for CT were excluded. NG positivity was defined as detection by culture or NAAT. Two reviewers (AL, JSH) applied the inclusion criteria independently. Any discrepancies were adjudicated by a third reviewer (FYSK).
Data extraction
Data extracted included country of study, study design, final year of data collection (<2010, 2010–2012, 2013+), study population, anal intercourse profile of population (included only women who reported anal intercourse vs included women who did or did not report anal intercourse), and site-specific CT and NG test positivity (anorectal, urogenital or oropharyngeal) (online supplementary material 3).
The primary outcomes were anorectal CT and anorectal NG positivity among women tested. Prevalent-only estimates were included. The secondary outcomes were associations between anorectal positivity and positivity at the oropharyngeal and urogenital sites or association with anal intercourse. These were measured as prevalence ratios (PRs) where anorectal positivity was the outcome. Where these were not reported, the PR and 95% CI were calculated using available data.
One author (AL) extracted the data and a second author (JSH) checked the extracted data. Disagreement was resolved by discussion between the two authors and consultation with an additional author (FYSK) until consensus was reached.
Analysis
For meta-analysis, random-effects methods were used to calculate pooled estimates of test positivity and PR with the assumption that heterogeneity was not wholly due to sampling variation. The I2 test was used to calculate the proportion of total variability attributable to heterogeneity rather than chance alone, and was considered moderate or high if greater than 50% or 75%, respectively.12 If I2 test was >25%, factors contributing to heterogeneity were investigated through the following: (1) we calculated stratum-specific summary estimates across several different subgroups; and (2) we used meta-regression to estimate PRs (for positivity estimates) and ratios of PRs (for association estimates). The relative reduction of between-study variance (τ2) provided an indication of the factor’s contribution to heterogeneity.
A sensitivity analysis was conducted excluding two studies in which a considerably larger number of women were tested for NG than CT13 14 and excluding three studies which used culture for NG testing.13 15 16 All analyses were performed using the ‘metaprop’, ‘metan’ and ‘metareg’ commands in Stata V.13.
Assessment of study quality
Assessment of within-study bias was undertaken using a combination of the evaluation criteria adopted by Sanderson et al 17 and the critical appraisal tool for cross-sectional studies, AXIS,18 and assessed the following criteria: selection bias, measurement bias, confounding, statistical methods and ethical considerations.
Results
Study selection and characteristics
Overall, 583 references were identified, of which 458 were unique papers. Hand-searching reference lists identified a further 6 papers, giving a total of 464 papers. Overall, 25 reporting on both CT and NG anorectal positivity were eligible (figure 1). Twelve (48%) studies were from North America,15 19–29 seven (28%) from the Netherlands,13 14 30–34 three (12%) from the UK,16 35 36 two (8%) from South America37 38 and one (4%) from South Africa.39
Flow chart of study selection. CT, Chlamydia trachomatis; IFA, immunofluorescent assay; NG, Neisseria gonorrhoeae.
Table 1 shows that 23 studies were cross-sectional,13–16 19–24 26–37 39 1 was a prospective cohort study25 and 1 was a case–control.38 The cohort study reported both prevalent and incident STI data.25 A total of 18 (72%) were set in STI/sexual health clinics,13 14 19–23 25 27 29–35 37 38 5 (20%) in hospital, genitourinary/obstetrics/gynaecology or primary health settings,15 16 26 36 39 and 2 (8%) were based on surveillance data from a laboratory28 or STI clinics.24 Overall, 25 studies reported anorectal CT positivity using NAAT, and 22 (88%) studies reported anorectal NG results using NAAT only, 2 (8%) studies used NAAT and culture,13 15 and 1 (4%) study reported culture only.16
Characteristics of included studies
Anorectal positivity
Anorectal CT positivity among women tested ranged from 0% to 17.5%, with a summary estimate of 8.0% with high heterogeneity (95% CI 7.0 to 9.1; I2=88.5%) (figure 2A).13–16 19–39 Meta-regression identified that whether or not the study population included only women reporting anal intercourse was an important contributor to heterogeneity (Δτ2=−16.9%), although when summary estimates were stratified by this variable there remained considerable heterogeneity in each stratum-specific summary estimate (online supplementary material 4).
Anorectal positivity among women. (A) Chlamydia trachomatis and (B) Neisseria gonorrhoeae. ES, rectal positivity among those tested (%).
Anorectal NG positivity ranged from 0% to 17.0%, with a summary estimate of 2.1% with high heterogeneity (95% CI 1.6 to 2.8; I2=92.7%) (figure 2B). Meta-regression identified that region of study was a major contributor to heterogeneity (Δτ2=−47.8%) (online supplementary material 4).
Sensitivity analysis showed negligible impact on anorectal CT and NG positivity summary estimates when studies that had disproportionately larger numbers of women tested for NG13 14 or studies using culture for NG anorectal results were excluded13 15 16 (online supplementary material 5).
Association of urogenital positivity with anorectal positivity
A total of 13 studies provided data on the association of urogenital positivity with anorectal positivity for CT and NG 13 16 19–24 26 31–33 (figure 3). The PR for the association of urogenital CT with anorectal CT positivity ranged from 15.4 to 196.2. The summary PR estimate for the association was 32.2 with moderate heterogeneity (95% CI 25.6 to 40.7; I2=70.3%). Meta-regression did not identify the source of heterogeneity (online supplementary material 4). Of those who tested for anorectal and urogenital CT, meta-analysis found that 22.6% (95% CI 19.7 to 25.7; I2=30.9%) tested positive for anorectal CT alone (ie, urogenital CT negative) (online supplementary material 6A).
Association of urogenital positivity with anorectal positivity. (A) Chlamydia trachomatis and (B) Neisseria gonorrhoeae. PR, prevalence ratio (proportion of those who are urogenital-positive who are also anorectal-positive relative to the proportion of those who are urogenital-negative who are also anorectal-positive). D+L, Der-Simonian and Laird; M-H, Mantel-Haenszel.
The PR for the association of urogenital NG with anorectal NG positivity ranged from 12.5 to 1123.5 and the summary estimate was 89.3 with high heterogeneity (95% CI 53.1 to 150.3; I2=80.1%). Meta-regression identified that region of study was an important contributor to heterogeneity (Δτ2=−44.3%), although there was no statistical difference between regions (online supplementary material 1). Of those who tested for anorectal and urogenital NG, meta-analysis found that 20.3% (95% CI 13.9 to 27.3; I2=55.7%) tested positive for anorectal NG alone (online supplementary material 6B).
Sensitivity analysis showed negligible impact on PR summary estimates when the study that had a disproportionately larger number of women tested for NG13 or studies using culture for NG anorectal results were excluded13 15 16 (online supplementary material 1).
Association of oropharyngeal positivity with anorectal positivity
Five studies provided data on the association of oropharyngeal positivity with anorectal positivity13 14 21 31 32 (online supplementary material 7). The PR for the association of oropharyngeal CT with anorectal CT ranged from 6.2 to 14.5 and the summary estimate was 8.8 (95% CI 6.8 to 11.5; I2=58.1%) with moderate heterogeneity. The PR for the association of oropharyngeal NG with anorectal NG ranged from 4.1 to 128.3 and the summary estimate was 34.8 (95% CI 10.2 to 118.2; I2=89.9%) with high heterogeneity. The small number of studies prevented meta-regression for investigating the association between oropharyngeal positivity and anorectal CT or NG.
Association of anal intercourse with anorectal positivity
Two studies provided data to calculate PR for the association of anal intercourse with anorectal positivity (online supplementary material 8).19 33 The summary estimates for CT and NG were 1.0 (95% CI 0.7 to 1.4; I2=0.0%) and 4.3 (95% CI 2.2 to 8.6; I2=0.0%), respectively.
Assessment of within-study bias
All studies were set in at-risk populations and vulnerable to selection bias. Overall, 14 (56%) studies specified consecutive recruitment of patients attending clinics or inclusion of all data reducing selection bias.13 16 19 20 22 23 26–31 33 35 Inclusion/exclusion criteria were specified for 22 (88%) studies.13–16 20–29 32–39 Measurement bias was considered low to moderate in 21 (84%) studies because the type of NAAT was described. However, the risk of bias was considered high in four (16%) studies where it was not described.13 24 28 38 Sample size calculations were not reported in any studies, making several studies at risk of reduced statistical power to measure associations (online supplementary material 9).
Discussion
This novel systematic review and meta-analysis investigated anorectal CT and NG positivity in the same female population and their association with detection at other sites. We found that anorectal CT estimates were higher than anorectal NG estimates, which is expected considering that CT is much more common in women than NG.1 However, we also found that anorectal NG was substantially more associated with urogenital and oropharyngeal detection than was anorectal CT. Furthermore, anorectal NG was associated with anal intercourse, but anorectal CT was not. Given these observations were made in the same populations, these data raise several important questions about the transmission dynamics for these two organisms.
There are several limitations to this review. First, as our aim was to investigate anorectal CT and NG in the same population, our search focused on papers that reported both CT and NG anorectal estimates, so any papers reporting only one infection were excluded; it is possible that findings in these papers were different from those in our review. However, systematic reviews of anorectal CT in women show similar results to ours.6–8 Second, not every woman was tested for both anorectal CT and NG in each study, potentially introducing selection bias into the comparison between the two organisms. This was the case with two studies13 14 where greater numbers of women were tested for anorectal NG than CT. However, our sensitivity analysis showed that excluding these studies did not impact summary estimates. Third, our investigation of heterogeneity was limited. Ideally, we would have investigated whether the type of NAAT (to account for test sensitivity) or clinician versus patient-collected swab (to investigate contamination due to sampling) played a role in heterogeneity, but it was not possible to extract these data. Further, we were unable to differentiate between culture and NAAT for NG results in studies using both methods, but our sensitivity analysis showed negligible impact of this on summary estimates. Finally, our review was limited to studies published in English. Including non-English studies may have altered our findings of heterogeneity associated with region of study.
There are several factors that may cause the detection of anorectal CT or NG in women, including anal intercourse; autoinoculation, where the infectious material from one site (eg, cervix) establishes infection in another anatomical site (eg, rectum)40; oral ingestion, causing infection in the rectum41; and contamination of the anal epithelium with CT or NG DNA from an infection at the cervix that may occur, for example, during toileting.42 The high sensitivity of NAATs means that contamination of the anal epithelium with CT or NG DNA could result in a positive test result even if infection is not established. However, determining which of these factors is responsible is extremely challenging.
Of interest, we found that the association between urogenital and anorectal positivity was nearly threefold stronger for NG than for CT. These data suggest that NG may be easier to transmit and establish infection than CT through autoinoculation. CT is a slow-growing organism that has tissue tropism for columnar or squamous metaplastic epithelium that in the rectum is 4 cm above the sphincter.43 NG replicates more quickly than CT and commences replication once it adheres to epithelial cells before invading.44 While CT infecting squamous epithelium is rare,45 NG can invade the squamous epithelium of the cervix.46 Given this, it seems likely that NG would be able to establish infection in the squamous epithelium inside the anal canal. However, it is possible that differences in test performance may have contributed to observed differences, with several NAAT tests being more sensitive for NG than CT.47 Further, selection bias may have played a role because anogenital symptoms occur more commonly with NG,48 and this may have prompted women to seek healthcare and be tested.
Oropharyngeal detection was associated with both anorectal CT and NG, but this association was nearly four times greater for NG than for CT. It has been previously suggested from animal models that oral ingestion could cause anorectal detection of CT,41 and a recent study detected anorectal CT in heterosexual men reporting cunnilingus but no receptive anal exposure.49 However, we were unable to find any suggestion that this could occur for NG in the literature, and as the data in this review are cross-sectional we cannot establish whether the presence of the organism at the oropharynx came before anorectal detection. Rather than oral ingestion being a factor with NG, the stronger association with NG may reflect that oropharyngeal NG is far more common than oropharyngeal CT,6 more likely to involve multiple anatomical sites,50 and can be readily cultured from oropharyngeal swabs and saliva.51 Further, there is some evidence that saliva has anti-CT effects, potentially reducing the incidence of oropharyngeal CT.52
We found evidence that anorectal NG was associated with anal intercourse, but anorectal CT was not. This is not a new finding. The absence of an association between anal intercourse and anorectal CT was reported in early culture-based studies, with the authors suggesting that autoinoculation from the urogenitals may be the primary mechanism by which such infections are acquired.53 In contrast, an association between anal intercourse and anorectal NG was reported in the 1950s.54 However, NG is much less prevalent in women and strongly associated with increased sexual risk, including number of sex partners55; further anal intercourse is also associated with the number of sex partners.56 Given this, we cannot exclude selection bias and confounding in our results, particularly as most studies were set in STI clinics. Unfortunately, we were not able to investigate the association between anal intercourse and anorectal CT or NG in women who did not have urogenital infection, which would have helped further investigate this association.
So given these results, should we worry about anorectal CT or NG in women? Anorectal infections tend to be asymptomatic,48 are likely to clear over time if left untreated57 and are usually associated with a concurrent urogenital infection. However, anorectal infections are a concern if they can autoinoculate, causing urogenital infection creating the risk that the infection ascends causing PID. If autoinoculation between the cervix and rectum can occur, then this is an issue for CT because of the potential reduced efficacy of azithromycin for anorectal infection.58 However, this is less of an issue with NG, where the recommended treatment is effective for anorectal infections.59 While anorectal STI screening in women is likely to be acceptable to women and clinicians in specialist STI clinics,34 extragenital screening is much less common in other primary health clinics (eg, general practice) where sexual health is not the core focus.60
Although anorectal CT in women has generated considerable recent debate in the literature, with calls for anorectal CT screening, our review highlights that anorectal NG also occurs and is also associated with a concurrent urogenital infection. This review highlights the urgent need for quality research involving longitudinal data collection with sufficient sample size and in representative population groups accounting for selection bias and confounding, to further our understanding of anorectal CT and NG in women, particularly their aetiology and whether they cause urogenital infection. This information is urgently needed to inform anorectal CT and NG screening guidelines for women.
Acknowledgments
The authors would like to thank Professor Dallas English and Dr Emily Karahalios for their advice with meta-regression.
References
Footnotes
Handling editor Jackie A Cassell
Contributors AL and JSH designed the search strategy and reviewed the abstracts. AL extracted the data, and JSH reviewed the data. AL performed the meta-analysis and meta-regression and wrote the first draft of the paper. JSH, FYSK, WH, EPFC and CKF commented in detail on the drafts and approved the final version.
Funding AL is supported by a National Health and Medical Research Council (NHMRC) Postgraduate Scholarship (1133144). JSH is supported by an NHMRC Senior Research Fellowship (1136117).
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.