Chlamydia screening, retesting and repeat diagnoses in Cornwall, UK 2003–2009
- 1School of Social and Community Medicine, University of Bristol, Bristol, UK
- 2Chlamydia Screening Service, Royal Cornwall Hospital Trust, Cornwall, UK
- Correspondence to Dr Katy M E Turner, School of Social and Community Medicine, University of Bristol, Canynge Hall, 39 Whatley Road, Bristol BS8 2PS, UK;
- Accepted 23 August 2012
- Published Online First 28 September 2012
Objectives This study aims to describe the patterns of testing and retesting for chlamydia in Cornwall during the first 5 years of the National Chlamydia Screening Programme. We evaluate the factors associated with retesting and estimate the incidence of chlamydia diagnosis and repeat diagnosis.
Methods Study design: Secondary database analysis. Selection criteria: men and women tested for chlamydia between March 2003 and January 2009 in Cornwall, aged ≥12 years and ≤25 years at the first test. The factors associated with retesting in those with at least one known test result and at least 14 days follow-up time were analysed using Cox regression and the incidence of diagnosis and repeat diagnosis were calculated.
Results The final dataset consisted of 71 066 records from 49 941 individuals; of whom 59.0% were female and 75.4% were only tested once. There were 48 375 individuals with at least one known test result (negative or positive) and at least 14 days follow-up, included in the Cox regression analysis. Factors associated with testing more than once were (adjusted HR, 95% CI): being female (2.24; 2.14 to 2.34) and initially testing positive (1.43; 1.35 to 1.51). The positivity at first episode declined from 13.2% (1077 cases) in 2003/2004 to 5.8% (843 cases) in 2008/2009. The incidence of diagnosis at the second test was 5.9 per 100 person years in those testing negative at the first test compared with 18.1 per 100 person years in those initially positive.
Discussion Most individuals in this analysis were tested only once, but the testing volume and proportion of repeat tests were highest at the end of the study period. As the testing rate stabilises to 30% coverage, maintaining retesting rates in those previously tested and especially in those previously diagnosed with chlamydia will be necessary for the sustainability of the screening programme.
Conclusions A key feature of the next 5 years of the screening programme will be to maintain screening and rescreening.
The national chlamydia screening programme (NCSP) was introduced in England in 2003 and rolled out nationally by 2008. The aims of the screening programme are to detect and treat infection, reduce onward transmission and prevent the consequences of untreated infection.1 The current recommendations are to screen sexually active individuals aged under 25 years annually and on change of sex partner. Nationally, the number of young people tested has increased from 19 000 in 2003/20004 to 1.4 million in 2010/2011 (over 2.1 million if all genitourinary medicine and other tests are included), achieving an average coverage of just over 30%.2 Coverage is calculated as the number of tests divided by the number of people in the target population. However there are regional variations and women are tested more often than men.3 ,4
As the screening programme matures to maintaining coverage rather than increasing screening volume, it becomes increasingly important to maintain levels of retesting in those already screened. Recently the NCSP has renewed the focus on effective management of individuals diagnosed with infection, including partner notification. Accordingly, a new outcome indicator has been introduced; the diagnostic rate, which measures the number of people diagnosed per 100 000 target population. Taken together with coverage of screening this will enable local areas to better monitor the effectiveness of their chlamydia control activities.1
Data submitted nationally did not allow identification of individuals to examine retest rates directly. In Cornwall, individuals are identifiable within the locally maintained database. Cornwall was also an early implementer site of the national screening programme and reached >20% coverage in 3 consecutive years from 2006 to 2008. It is mainly a rural area with a target population of about 63 200 young people. All tests were performed at a single microbiology laboratory and Cornwall only shares a border with Devon, so the population is fairly closed. The raw data consisted of over 100 000 test records from all tests conducted within Cornwall (not just NCSP). We analysed testing and retesting patterns and diagnosis and repeat diagnosis incidence in men and women tested between March 2003 and January 2009 in Cornwall, aged ≥12 years and ≤25 years at the first test.
Retrospective cohort study (secondary data analysis).
Men and women tested for chlamydia between March 2003 and January 2009 in Cornwall, aged ≥12 years and ≤25 years at the first test. The factors associated with retesting in those with at least one known test result and at least 14 days follow-up time were analysed using Cox regression and the incidence of diagnosis and repeat diagnosis were calculated. All tests were performed in the same laboratory.
The dataset was anonymised by the data holder in Cornwall, prior to sharing, but in such a way that the linkage of individual screening tests was retained. The matching of individuals was done by name and postcode. The data were cleaned (flow diagram shown in online supplementary figure S1), so that the initial dataset of 110 102 test records was reduced to 71 066 tests, with one test per day per person, age at first test ≥12 and ≤25. Age and gender were input if available from duplicate or multiple tests. If multiple tests were performed on the same day, results were retained preferentially in the order: positive, negative, other, to minimise the number of unknown results in the dataset. In the final dataset ‘Test records’ (n=71 066), all test records included patient ID, gender, age at first test, date of test and result of test (positive, negative or unknown). A related dataset was also produced containing one line per individual with information on up to six tests. ‘Individual records’ (n=49 941).
Calculation of a test episode
Test information was collated into a single episode, consisting of any tests performed within 14 days of the primary test. Results were preferentially assigned to the episode in the order: positive (1), negative (0), then unknown (inhibitory, equivocal or insufficient, coded as missing), and dated as the earliest test date. This gave a total of 48 375 individuals with at least one known episode result and at least 14 days of follow-up time and 11 550 (23%) of those had two or more known episode results more than 14 days apart. This population of individuals (n=48 375) formed the denominator population for calculation of incidence of diagnosis and repeat diagnosis.
We describe the patterns of testing and chlamydia diagnosis during the study period. We calculated positivity as the proportion of first episode results which were positive (positive/positive+negative), by age, gender and year of test. Follow-up time was calculated as time from first to second episode or censored at the last test date in the database. We did not censor by age, but included repeat tests in those aged over 25 (since all initial tests were in those aged <25). We estimated the incidence of diagnosis and repeat diagnosis. We assessed the factors associated with retesting using Cox regression, according to gender, result of first test, year of first test, age and sexual behaviour in the last 3 months. We also evaluated the timing retesting and repeat diagnosis for the first two episodes. All statistical analyses were performed in STATA V.12.
Results 1: description of the dataset
Overall, there were 71 066 tests, of which 65.4% (46 472) were in women. There were 49 941 individuals (59.0% women) in the database of whom 75.4% were tested only once (see also online supplementary figure S1). We defined positivity according to known positive or negative test results: positivity=positive/(positive+negative). The average positivity in all tests with known result was 9.2% (2186/23 886) in men and 8.8% (3904/44 379) in women (see online supplementary table S1). Figure 1 shows the increase in the volume of tests over time, from 6492 in 2004 to 21 584 in 2008 (see also online supplementary table S1). Online supplementary table S1 shows additional information about all tests by year, age, gender and test result as well as antimicrobial treatment used. Most individuals were tested only once. The maximum number of tests observed within the study period was 21, and less than 5% of individuals were tested more than three times.
First recorded test episode
There were 48 375 individuals with at least one known test result (negative or positive) and at least 14 days follow-up included in the Cox regression analysis. The positivity at first recorded test episode and absolute number of diagnoses decreased during the study period from 13.3% (1055) in 2003/2004 to 5.7% (827) in 2008/2009, shown in figure 1A and online supplementary table S1. The distribution of positive results by age is shown in figure 1B for six age classes, by three date groups (March 2003–December 2004, January 2005–December 2006, January 2007–January 2009). There was an average decline in positivity at first test of 18% per year (logistic regression, adjusted OR 0.82, 95% CI 0.80 to 0.84, p<0.001), adjusted for age and gender. The positivity in men and women was similar after the first year, but the age profile of positivity was different between men and women (figure 1B). The highest mean positivity was observed in women aged 19–20 (11.0% positive) and in men aged 21–22 (12.5% positive). The age profile remained similar, although the absolute positivity declined over time. Online supplementary tables S2 and S3 show details of the first recorded test episode for the 48 375 individuals with at least one known test and ≥14 days follow-up time.
We also present the change over time in testing volume and positivity in different types of venue and by reason for test (see online supplementary table S4). In general practice and genitourinary medicine clinics the volume of testing has stayed fairly constant over time, but positivity of first episode has declined: in general practice the positivity has declined from 12.5% in 2003/2004 to 9.0% in 2008/2009 and in genitourinary medicine the positivity declined from 15.6% to 10.0% in the same time frame. Genitourinary medicine clinics account for greatest proportion of tests (31% of all tests) and 48% of positive cases, followed by general practice, 14% of tests and 15% of cases, shown in figure 2.
The relationship between incidence of retesting and gender, result of first test, age at first test, sexual behaviour and year of first test were analysed using Cox regression and are presented as unadjusted and adjusted HRs in table 1. The population at risk were those with at least one known test episode and at least 14 days follow-up time (n=48 375). Women were more than twice as likely than men to retest (adjusted HR, 95% 2.24, 2.14 to 2.34) and those testing positive were also more likely to retest (adjusted HR 1.43, 1.35 to 1.51).
Results 2: incidence and reinfection
The incidence of diagnosis and repeat diagnosis was calculated for those with at least one known episode and ≥14 days follow-up (n=48 375) according to whether the first test was negative or positive. The overall incidence of repeat diagnosis at the second test was 7.3 per 100 per years and 5.9 per 100 person years in those negative at the first test compared with 18.1 per 100 person years in those initially positive. This gave an incidence rate ratio of 3.1. The pathway of testing from first to second episode with a known result is given in figure 3. Overall 8.7% of those tested for a second time were positive at the second test and 7.2% of those initially negative compared with 19.4% of those initially positive.
Summary of findings
The majority of individuals were tested only once during the study period, however in the final year (2008), nearly a third of the tests were in people who had been tested previously, largely as a result of the increases in screening volume during the study period. The positivity and absolute number of diagnoses declined over time in all age groups, while the screening volume more than doubled. At the first known test episode, the overall positivity was 9.0% and at the second known episode was 8.7% (19.4% in those positive and 7.2% in those negative at first known episode).
Strengths and limitations
The strength of the study is the large size of the dataset within a well-defined geographic region during the early stages of a national screening programme. This dataset has the advantage of uniquely identified individuals so that the rates of retesting, diagnosis and repeat diagnosis can be estimated directly.
The limitations of the dataset are that although the positivity of tests has declined it is not possible to directly estimate the underlying prevalence to evaluate the effectiveness of the screening programme. Data on sexual behaviour, which could have informed this issue, were collected initially but were poorly completed in later years (see online supplementary table S3), possibly due to time pressures and focus on coverage targets. We have also assumed that the cohort within Cornwall is closed but of course many young people leave home or go to university during the key age range for chlamydia screening (16–25) and may have been screened elsewhere. There was also some outreach testing, for example, at festivals in some years during which non-residents may have been screened. Finally, some individuals may have tested more than once but gave different names or could not be correctly matched. All these factors would tend to underestimate the amount of retesting in this population.
Context of other studies
Estimating the rate of repeat testing for chlamydia within the context of ongoing control programmes is challenging due to the need to link multiple records for individuals over long periods of time. In the UK, routine national data do not permit direct estimation of re-screening rates for chlamydia. However in other countries with mandatory health identification numbers this is possible for example, in the Uppsala Women's cohort in Sweden, cumulatively, 70.7% of women were screened at least once but only about half of those were screened two or more times (47.8%).5 In England annual coverage of all chlamydia tests has increased substantially from 2008/2009 to 2010/2011: in men from 13.2% to 22.6% and in women from 31.3% to 42.7% although this does not account for any repeat testing within the same year.6 The positivity in those screened has decreased over time in Cornwall, which has also been observed nationally.6
Reinfection may occur due to infection from an untreated partner, infection from a new partner or treatment failure. If any of these occur, the risk of treatment failure occurs soonest, followed by reinfection from an untreated current partner, then infection from new partners (durations of risk are variable and overlapping). Systematic reviews of reinfection in women7 and men8 have shown consistently high rates of reinfection with both chlamydia and gonorrhoea, which is borne out in our data. The Recall study in England also found high rates of reinfection with 18.7% (78/417) individuals testing positive at two tests.9
The Recall study also showed the importance of partner treatment and condom use in preventing infection and reinfection.9 Effective partner notification has subsequently been demonstrated in high quality trials to reduce reinfection with gonorrhoea and chlamydia.10 ,11 In the Cornwall dataset, the proportion of those giving reason for test as ‘contact’ of an infected individual who were themselves infected was consistently around 40% throughout the study, although the numbers were small (see online supplementary table S4). Studies are currently ongoing to evaluate the best methods of achieving high levels of timely partner treatment within the UK, including a randomised controlled trial of Accelerated Partner Notification, which is similar to the US partner delivered therapy, but requires a consultation with a health professional, to meet English legal requirements.12 Even if partner notification is effective, repeat diagnosis rates are still higher than first diagnosis rates in those not previously infected.7 This may be in part due to treatment failure or risk from new partners within the same sexual network, associated with continued higher risk behaviour.13 ,14 Batteiger et al15 have used genotyping and detailed follow-up of a cohort of young women to tease out the relative contribution of current partners, new partners and treatment failure on rates of reinfection. They found that of 183 repeat episodes, 13.7% (25) were likely/probably treatment failure, 84.2% (156) were reinfections and 2.2% (4) had no documented treatment. Of the reinfections, two-thirds had different genotypes (100), of whom 74 reported sex with a different partner, suggesting that at least half of the reinfections could be due to new partners. The authors note that this was in a high prevalence population and the proportions may not be the same in populations with lower incidence such as in England.15
The reduction in the absolute number of diagnoses, in the context of large increases in test volume appears to point tentatively towards a true reduction in prevalence. It remains to be seen whether these reductions in positivity will continue as the coverage stabilises. The variation between English counties in their initial prevalence and the trajectory of screening coverage over the last 10 years means that detailed understanding of the local situation is needed to ensure appropriate evaluation of the effectiveness of efforts to control chlamydia.
It remains imperative to ensure the basics of sexually transmitted infection control are done well that is:
Ensure effective and timely treatment of positives
Ensure effective and timely management of their sexual partners (partner notification)
In order to maintain and improve the effectiveness of chlamydia screening and particularly increasing rates of retesting, the message of screening annually or on change of partner for those testing negative is important given that those testing negative remain at risk of future infection. Interventions such as text or email reminders at 1 year could be introduced at minimal cost and their effectiveness evaluated as has been trialled in Australia.16 For those testing positive we support the conclusion of Hosenfeld et al that retesting 3–6 months later would be advisable.7 Interventions to improve retesting rates in positives could also include text/email reminders, postal kits as well as enhanced sexual health advice including condom provision, other contraceptive information and brief behavioural interventions to reduce risk of infection and promote healthy sexual relationships.
A key feature of the next 5 years of the screening programme will be to maintain screening and rescreening. This study clearly shows the importance of ensuring that those who test positive are re-screened and potentially these individuals could be the focus of additional interventions for example, postal screening packs or active recall for retesting. The incidence of infection in those who test negative is still sufficiently high so annual testing is appropriate. As the screening programme matures and the novelty wears off, maintaining high rates of screening will mean screening an increasing proportion of people who have been screened before. This represents both a challenge and an opportunity.
Maintaining retesting rates will be important to the long term sustainability of chlamydia screening in the UK
Individuals who test positive are at high risk of repeat diagnosis
Recommend retesting positives at 3–6 months after initial screen
In Cornwall the positivity in those tested for the first time has decreased over time
We are very grateful to Pam Gates for initial discussions and allowing access to the Cornwall data, Jonathan Childs for anonymising and providing the dataset and Frances Keane for helpful comments on the paper and support for the study.
Contributors KT: conceived the study design, planned the statistical analysis, edited the final draft and supervised LT-L. PH: facilitated data collection, helped devise the study and contributed to the public health message and clinical relevance of the study. LT-L: undertook the statistical analysis and wrote the first draft, MM provided statistical advice and supervised the statistical analysis. MS: provided advice on the Cornwall data and helped with data interpretation from the point of view of the local screening office. All authors reviewed and commented on the final draft of the paper.
Funding KT is grateful to NIHR for providing funding for her fellowship award.
Competing interests Margaret May, Lea Trela-Larsen and Matt Sharp do not have any conflicts of interest to declare. Lea Trela-Larsen is funded by a MRC PhD studentship.
Data sharing statement For further information about the dataset and do files please contact Katy Turner email@example.com.
Provenance and peer review Commissioned; externally peer reviewed.
Correction This article has been corrected since it was published Online First. The Competing interests statement has been edited to include Lea Trela-Larsen's MRC PhD studentship; and Matt Sharp's affiliation has been updated to read, ‘Chlamydia Screening Service, Royal Cornwall Hospital Trust, Cornwall, UK’.