Objectives To explore patterns of repeat chlamydia testing among young people in England and factors associated with testing positive at repeat test.
Methods We analysed chlamydia testing among 15 to 24-year-olds in England in a single calendar year (2010) using data from the genitourinary medicine clinic activity dataset (GUMCAD) and tests reported through the National Chlamydia Screening Programme (NCSP). Case records were linked using patient clinic numbers (GUMCAD), or by matching date of birth, gender and postcode (NCSP). Individuals could not be linked between datasets. The incidence of repeat testing was estimated using survival analysis. Risk factors for testing positive at repeat test were explored using multivariable logistic regression.
Results 1 235 058 tests in the NCSP dataset and 502 095 in GUMCAD were included. The incidence of repeat testing was 18.4 and 26.1 per 100 person years in the NCSP dataset and GUMCAD respectively. Among NCSP repeat tests, the proportion testing positive was higher in those reporting recent change of sexual partner (adjusted OR males 1.44; females 1.52), and among those with a positive compared to a negative baseline test (adjusted OR males 2.57; females 1.95).
Conclusions We observed moderate levels of repeat testing within a year. Considering the frequency of partner change among young people, more could be done to encourage re-testing upon change of sexual partner. Increasing re-testing following a positive test could probably identify unresolved or repeat infections that may otherwise go untreated. Work to establish the optimum approach to repeat testing in England is now warranted.
- Chlamydia Trachomatis
Statistics from Altmetric.com
The National Chlamydia Screening Programme (NCSP) in England recommends that sexually active young people under the age of 25 should be encouraged to test annually for Chlamydia trachomatis (‘chlamydia’), or whenever there is a change in sexual partner.1 The substantial incidence of chlamydia among young people and the increased risk of infection among those with a new sexual partner provide the rationale for this recommendation.2 Guidelines from the British Association of Sexual Health and HIV recommend test of cure for pregnant women at 5–6 weeks after treatment, but no other recommendation is made on the frequency of testing.3 Rates of partner change are relatively high among young people.4 Thus a high proportion of individuals tested for chlamydia in England are likely to be eligible for re-testing within 1 year under current policy recommendations. Several studies have shown the higher risk of subsequently testing positive following a positive test,2 ,5 ,6 and a number of countries including USA,7 Canada,8 Australia9 and the Netherlands10 ,11 recommend repeat testing for individuals following a positive chlamydia test.
We analysed chlamydia testing data for England in 2010 to identify repeat testing patterns among young people and to explore the factors associated with testing positive at repeat test.
We used data on chlamydia tests in England from two Health Protection Agency maintained datasets. Tests performed in genitourinary medicine (GUM) clinics are reported through the GUM clinic activity dataset (GUMCAD12). Tests performed outside of GUM clinics in a range of settings including general practice, community sexual and reproductive health services, community pharmacies, education and youth venues and tests accessed via the internet are reported via the NCSP (herein referred to as the ‘NCSP dataset’).
Analyses were carried out separately in the two datasets, as it was not possible to link individuals between them. For both datasets, individuals with at least one test recorded in 2010, aged 15 to 24 at their first test in 2010, were included. The first recorded test for each individual in 2010 was considered to be their ‘baseline test’. Subsequent tests in the same individual identified in the 2010 datasets were classed as ‘repeat tests’. Test records within 6 weeks of previous test were considered to be the same testing episode and were excluded.
In GUMCAD, the case records of individuals testing more than once were linked using their clinic number, which is unique within a clinic. In this dataset chlamydia tests and diagnoses are recorded separately, and may be recorded on different dates. All recorded sexual health screens were considered to include a chlamydia test (as specified in the coding guidelines13). This test was considered to be positive if there was a chlamydia diagnosis recorded on the same day or within the following 6 weeks. A chlamydia diagnosis reported without an accompanying sexual health screen was included as a positive test. Men who reported ever having sex with a man at any clinic attendance in GUMCAD were excluded from the dataset, to avoid overestimating repeat testing as men who reported ever having sex with a man are likely to attend GUM clinics more frequently for testing (see online supplementary figure S1).
Within the NCSP dataset, test records with a matching postcode of residence, date of birth and gender were considered to refer to the same person (88% with valid data included, online supplementary figure S2). Data on sexual orientation are not collected in the NCSP dataset.
The number of tests conducted in 2010 and the proportion testing positive was calculated for: baseline and repeat tests; repeat tests following a positive or a negative baseline test; and baseline and repeat tests with and without a new sexual partner reported in the previous 3 months (available in the NCSP dataset only). Proportions testing positive within subgroups were compared using the χ2 test.
The incidence of repeat testing (per 100 person years) was calculated using survival analysis to allow for the different follow-up time available for each individual. Each person contributed follow-up time from the point of their baseline test until their next test, or until the end of 2010 if only one test was recorded in the dataset. Individuals with baseline tests within 6 weeks of the end of the 2010 were excluded, as these individuals could not have a repeat test in the available follow-up time. As the NCSP dataset does not collect information on those over 25 years old, individuals were right censored on the date of their 25th birthday as they could not contribute any further follow-up time.
Multivariable logistic regression was used to explore the association between testing positive at repeat test in the NCSP dataset and (a) reporting a new sexual partner in the past 3 months and (b) test result at baseline. Unadjusted and adjusted OR were calculated, adjusting for age, ethnicity, venue type of test and reporting more than one sexual partner in the previous year. Records with missing data on any of the variables were excluded from the models. Separate models were fitted for men and women.
A total of 1 235 058 test episodes for 15 to 24-year-olds in the NCSP dataset and 502 095 in GUMCAD were included.
10% of tests in the NCSP dataset and 13% in GUMCAD were repeat tests within the calendar year of 2010 (table 1). The majority of individuals in the dataset had only one test recorded in 2010 (>85% in both datasets); 1% had three or more tests within 2010. Among those with available data on recent partner change in the NCSP dataset (53%), the proportion reporting a new sexual partner in the previous 3 months was higher among repeat tests compared to baseline tests (60% vs 48%; p<0.001).
Rates of repeat testing
1 017 252 individuals in the NCSP dataset and 399 392 in GUMCAD contributed a total of 556 629 and 212 090 person years follow-up time, respectively. The overall incidence of repeat testing was 18.4 per 100 person years (95% CI 18.3 to 18.5) in the NCSP dataset and 26.1 per 100 person years (95% CI 25.8 to 26.3) in GUMCAD (figure 1). Rates of repeat testing were higher among females than males, and following a positive rather than a negative baseline test.
Proportion testing positive
The proportion testing positive was higher among repeat tests compared to baseline tests in women tested through the NCSP (6.3% vs 5.7%, p<0.001). In men tested through the NCSP and males and females in GUMCAD, the proportion testing positive was slightly lower in repeat tests compared to baseline tests (4.1% vs 4.4%, p=0.03; 11.6% vs 12.8%, p<0.001; 9.5% vs 11.4%, p<0.001) (table 1).
The proportion of repeat tests that were positive was higher among those with an initial positive test result at baseline than those who had an initial negative test in both GUMCAD (17.0% vs 8.2%, p<0.001) and the NCSP dataset (12.5% vs 4.8%, p<0.001).
Repeat test episodes in the NCSP dataset with non-missing data on the variables of interest were included in the logistic regression analysis (n=50,819, 42% of all NCSP repeat tests). All variables remained significantly associated with testing positive at repeat test in the multivariable analysis (table 2). The adjusted OR of testing positive for individuals reporting a new sexual partner within the previous 3 months was 1.44 (95% CI 1.16 to 1.79) for men and 1.52 (95% CI 1.38 to 1.68) for women. Men with a positive test at baseline had over twice the odds of testing positive at repeat test compared to those with a negative test at baseline (adjusted OR 2.57, 95% CI 2.11 to 3.14). Women with a previous positive test had almost twice the odds of testing positive at repeat test (adjusted OR 1.95, 95% CI 1.76 to 2.16).
In this analysis, we provide the first investigation of patterns of repeat testing for chlamydia among young people using available national data sources in England. We observed moderate rates of repeat testing within 2010, ranging from 18 (for individuals tested through the NCSP) to 26 per 100 person years (for individuals tested in GUM clinics). The proportion testing positive at repeat test was higher among those with a previous positive baseline test compared to those with a previous negative test.
Reporting a change of sexual partner in the previous 3 months was significantly associated with a 40% to 50% increase in the odds of testing positive at repeat test. This is consistent with previous studies that have reported an increased risk of infection among those with a recent change of partner2 ,14 and supports the existing recommendation to encourage screening on change of sexual partner.1 However rates of repeat testing among young people with recent change of sexual partner could be higher. In the second National Survey of Sexual Attitudes and Lifestyles, conducted in 2000 in the UK, 39% of women aged 16 to 24 and 53% of same aged men reported having at least one new sexual partner in the previous 12 months. While partner change rates are likely to be higher among the tested population compared to a general population sample,15 this provides an indication of the minimum rate of repeat testing that would be expected if all young people were to be re-tested for chlamydia upon change of sexual partner.4 We estimated the overall incidence of repeat testing to be 20 to 29 per 100 person years among young women, and 15 to 21 per 100 person years among young men in the NCSP dataset and GUMCAD respectively. Thus the observed rates of repeat testing were at least 25% lower for women and 50% lower for men, than could be expected if everyone were to have a test each time they had a new sexual partner. This indicates that more could be done to encourage young people to take a chlamydia test when they change sexual partner.
Among repeat testers, individuals who tested positive at their baseline test were more likely to be positive than those who tested negative at baseline. This is consistent with several other studies that have shown a higher risk of being diagnosed positive for chlamydia at follow-up following an initial positive test.2 ,5 ,6 This observation suggests that increasing re-testing after a positive chlamydia test could identify unresolved or repeat infections that may otherwise go untreated with yearly screening. However the optimum approach to repeat testing in England is yet to be established. The resources required and yield in terms of number of individuals having a repeat test are likely to vary according to the approach used (which might include recommendation by clinician at the time of treatment, telephone or text message reminders or posting home testing kits to consenting patients at an agreed interval).16 Rates of repeat testing in countries that already recommend repeat testing following a positive test vary. For example, 66% of 18 to 29-year-olds who tested positive as part of a chlamydia screening implementation project in the Netherlands returned mailed testing kits 6 months after their initial positive test.17 In Bowring et al's recent analysis of repeat testing in general practices, 25% of 16 to 29-year-olds who tested positive re-attended and were re-tested within the recommended 1.5 to 4 months after a positive test,18 which is more comparable to the rates we have reported here.
For this initial investigation, we restricted analyses to tests performed within a single calendar year. Although we used survival analyses when estimating rates of repeat testing to allow for the differential follow-up time available, limiting to a single calendar year may have led to misclassification of tests as baseline instead of repeats. The reported differences between baseline and repeat tests should therefore be interpreted with caution, and this potential misclassification may provide one explanation for the higher proportion testing positive seen in baseline tests compared to repeat tests in GUMCAD and males tested through the NCSP. The restricted analysis period also meant that we were unable to assess adherence to the recommendation for annual screening, which we plan to investigate further when more years of data are available.
Our analyses are subject to limitations arising from the observational nature of the data. Firstly, we could not link individuals re-testing in a different GUM clinic, or moving between GUM and NCSP sites for re-testing. This is likely to have led to underestimation of the rate of repeat testing in both datasets. The matching process that we used to identify individuals in the NCSP dataset could also have resulted in some misclassification of tests as baseline or repeats. This could therefore have led to either overestimation or underestimation of repeat testing rates in the NCSP dataset. It is also feasible that tests may have been recorded in both datasets, especially in the situation where someone who was initially tested through the NCSP attends a GUM clinic for treatment. While the overlap arising from this is likely to be minimal, this could not be determined with the current data systems. A single laboratory-based reporting system for chlamydia is currently being implemented, thus future analyses will be able to assess and eliminate any overlap. No data were available on treatment of the individual or their partner(s). As such we could not determine whether repeat diagnoses represented re-infection from the same partner, from a new partner, non-compliance with treatment or treatment failure. Our results may also be subject to bias as we could only assess the proportion testing positive among those who returned for testing. Those who initially tested positive and were not re-tested may have a different risk of testing positive at re-test. In addition, a substantial proportion of tests in the NCSP dataset had missing data on recent partner change and the other variables included in the logistic regression model. While our findings are consistent with previous analyses, the potential for response bias remains a limitation in our analysis.
Considering the frequency of partner change among young people, more could be done to encourage re-testing upon change of sexual partner. The higher proportions of repeat tests that were positive among those with an initial positive test provides further evidence to suggest that re-testing after a positive chlamydia test could identify unresolved or incident infections that may otherwise go untreated. Further work to establish the optimum approach to repeat testing in England is now needed.
In this analysis of chlamydia testing among 15 to 24-year-olds in 2010 in England, rates of repeat testing ranged from 18 (for individuals tested through the National Chlamydia Screening Programme) to 26 per 100 person years (for individuals tested in GUM clinics).
The proportion of repeat tests that were positive was higher in those reporting a recent change of sexual partner, and following a positive baseline test.
Given the frequency of partner change in this age group, higher rates of repeat testing within a year could be expected if chlamydia screening was always conducted after change of partner.
Increasing re-testing following a positive test could probably identify unresolved or repeat infections that may otherwise go untreated, and work to establish the optimum approach to repeat testing in England is now warranted.
The authors thank Alireza Talebi and Stephen Duffell for support with NCSP and GUMCAD data management, Sarika Desai for assistance with planning the analyses, Catherine Lowndes for comments on the manuscript, and all of the clinical testing venues that continue to contribute data to HPA surveillance.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online figures
Contributors SW, JA, KS, GH and NG conceived of and designed the study; SW carried out the analyses and drafted the first version of the manuscript; all authors contributed to the interpretation of the data and approved the final version of the paper for publication.
Funding This work was undertaken by the Health Protection Agency who received funding from the Department of Health; the views expressed in this publication are those of the authors and not necessarily those of the Department of Health (England).
Competing interests All authors work for the Health Protection Agency; SW, JA, KS and AB work for the National Chlamydia Screening Programme.
Ethics approval Ethics committee approval is not required, as the analyses are based on surveillance data held by the HPA. These datasets have approval for analyses for public health purposes.
Provenance and peer review Commissioned; externally peer reviewed.
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.