Article Text


It matters what you measure: a systematic literature review examining whether young people in poorer socioeconomic circumstances are more at risk of chlamydia
  1. Jessica Sheringham1,
  2. Sue Mann2,
  3. Ian Simms3,
  4. Mai Stafford4,
  5. Graham J Hart5,
  6. Rosalind Raine6
  1. 1Department of Applied Health Research, University College London, London, UK
  2. 2Sexual Health Consultant, King's College Hospital, London, UK
  3. 3Department of HIV & STI, Health Protection Services, Colindale, Health Protection Agency, London, UK
  4. 4Medical Research Council Unit for Lifelong Health and Ageing, University College London, London, UK
  5. 5Department of Infection & Population Health, Faculty of Population Health Sciences, Centre for Sexual Health, HIV Research, University College London Mortimer Market Centre, London, UK
  6. 6Department of Applied Health Research, University College London, London, UK
  1. Correspondence to Dr Jessica Sheringham, Department of Applied Health Research, University College London, 1-19 Torrington Place, London WC1E 6BT, UK; j.sheringham{at}


Background England has invested in chlamydia screening interventions for young people. It is not known whether young people in poorer socioeconomic circumstances (SEC) are at greater risk of chlamydia and therefore in greater need of screening.

Objective To conduct a systematic review examining socioeconomic variations in chlamydia prevalence or positivity in young people.

Data sources Eight bibliographic databases using terms related to chlamydia and SEC, supplemented by website and reference searches.

Eligibility Studies published 1999–2011 in North America, Western Europe, Australia or New Zealand, including populations aged 15–24 years, with chlamydia prevalence or positivity diagnosed by nucleic acid amplification testing.

Appraisal and synthesis Two reviewers independently screened references, extracted data, appraised studies meeting inclusion criteria and rated studies as high, medium or low according to their quality and relevance. Socioeconomic variations in chlamydia were synthesised for medium/high-rated studies only.

Results No high-rated studies were identified. Eight medium-rated studies reported variations in chlamydia prevalence by SEC. In 6/8 studies, prevalence was higher in people of poorer SEC. Associations were more often significant when measured by education than when using other indicators. All studies measuring positivity were rated low. Across all studies, methodological limitations in SEC measurement were identified.

Conclusions The current literature is limited in its capacity to describe associations between SEC and chlamydia risk. The choice of SEC measure may explain why some studies find higher chlamydia prevalence in young people in disadvantaged circumstances while others do not. Studies using appropriate SEC indicators (eg, education) are needed to inform decisions about targeting chlamydia screening.

  • Socioeconomic factors
  • Chlamydia trachomatis
  • chlamydia screening
  • young people
  • epidemiology

Statistics from


Chlamydia is the most common bacterial sexually transmitted infection (STI) in England. In 2011, the Health Protection Agency surveillance data reported over 350 diagnosed cases per 100 000 people, nearly 10 times higher than the rate for gonorrhoea, the second most prevalent bacterial STI. Rates in people under 25 years old were five times that of the general population.1

The implications of a high burden of infection are not well understood, but chlamydia may increase the risk of pelvic inflammatory disease, tubal factor infertility and ectopic pregnancy.2 The risk of long-term reproductive problems has been the driver for major chlamydia control efforts in several countries.3 In England, the National Chlamydia Screening Programme (NCSP) was introduced in the National Health Service in 2003. The NCSP targets asymptomatic, sexually active people aged 15–24 years because of the comparatively high burden of infection in this group.4 ,5 In contrast to population-based programmes where individuals selected from a population register are invited for screening, the NCSP is delivered opportunistically (ie, eligible individuals are offered a test when they use venues registered to deliver the programme).

Socioeconomic circumstances (SEC) have been considered as a criterion for targeting chlamydia screening.6–9 A screening programme trialled in the Netherlands used information on young people's SEC to exclude those with low or no risk of chlamydia. The researchers concluded that this selective approach achieved higher positivity in their population compared with screening all young people.10 However, a more recent systematic review to examine socioeconomic associations with chlamydia concluded that the relationship is unclear and called for more research to build a ‘deeper understanding’ of chlamydia's relationship with social factors.11 This deeper understanding is needed to inform decisions about targeting chlamydia screening and to inform the development of interventions to prevent young people acquiring or transmitting chlamydia.

The aim of this systematic literature review was to inform decisions about provision of the NCSP in England by updating the previous review (which included papers up to 1999)11 and by examining whether young people in poorer SEC are more at risk of chlamydia.

A range of dimensions relating to an individual's education, income or employment or the area in which they live can be used to capture SEC. These dimensions of SEC, while often highly correlated, do not measure the same construct and therefore may affect chlamydia prevalence in different ways.12 Therefore, as part of this review, we also sought to examine the extent to which results varied when different measures of SEC were used.


Search strategy and eligibility

We searched the following library databases for peer-reviewed articles: Medline, Embase, Psychinfo, Social Policy and Practice, Web of Knowledge, International Bibliography of the Social Sciences, Cinahl and Scopus. The search terms related to SEC and chlamydia (supplementary data). Reference lists of included studies and selected websites (ie, Health Protection Agency, Centres for Disease Control and the South West Public Health Observatory because of its focus on STI monitoring) were hand-searched to ensure all relevant articles were identified. To ensure the search retrieved all relevant references, an expert in systematic reviews reviewed the search strategy. The study authors and an independent expert in chlamydia control reviewed the list of studies meeting the inclusion criteria.

We sought studies that examined the relative risk of chlamydia by SEC in populations with sufficient similarity to young people in England. Therefore, we included studies published in English conducted in the UK, North America, Australia, New Zealand and Western Europe that reported analyses of men and/or women aged 15–24 years. We sought literature published since the last systematic review, so we searched for papers published from January 1999 to October 2011.

We sought papers reporting valid measures of chlamydia by SEC. We therefore excluded studies where chlamydia was self-reported or was not diagnosed using nucleic acid amplification techniques. We sought studies comparing the extent to which chlamydia is present in populations of different SEC, so we included measures of chlamydia prevalence and positivity (positive tests as a proportion of all those screened). We excluded studies reporting diagnoses rates where there was no information on the number of tests performed. This is because variations in chlamydia diagnoses rates may be biased by differential testing and reporting practices.

We excluded studies that reported chlamydia aggregated with other STIs (eg, gonorrhoea). While the route of infection for other STIs is similar to chlamydia, the transmission characteristics of other STIs vary, thus distorting true associations with either infection. Finally, we excluded studies that reported reinfections only, because individuals who experience more than one chlamydia infection may be a distinct group from those who test positive just once.

We sought to capture papers that explicitly identified SEC characteristics and papers where associations between chlamydia prevalence or positivity and socioeconomic dimension were reported, even if not discussed elsewhere in the paper. Therefore, we included papers using a broad range of SEC terms, including dimensions of education, income and employment, as well as circumstances that could act as proxies of SEC (eg, school type for education). We excluded studies where it was not possible to compare prevalence or positivity in different socioeconomic groups, so we excluded studies where participants were selected because they were in a particular socioeconomic group (eg, university students or socially disadvantaged).

For quality assurance purposes, two authors (JS and SM) reviewed 15% of retrieved references against the inclusion/exclusion criteria. Where decisions were discordant, references were reread and the application of inclusion/exclusion criteria were discussed.

Data extraction and appraisal

JS and SM independently appraised and rated all studies meeting the inclusion criteria. We used published guidance13–16 to develop a rating system specifically for this study, incorporating assessments of generic quality and relevance to the research question. A summary of our scale is shown in table 1 with a full description of the rating scale and scoring system in the web appendix.

Table 1

Appraisal scoring system summary


Search and screening

We retrieved 7151 citations, of which 70 papers were subject to a full paper review (exclusions at each stage are shown in figure A1, web appendix).

We identified 28 studies which met the inclusion criteria. These comprised eight population prevalence surveys, 11 prevalence and/or positivity studies of screening interventions and nine studies reporting prevalence or odds ratios from specialist sexual or reproductive health clinic populations.

Although we identified six studies of high quality and another five studies of high relevance, we did not identify any studies of both high quality and relevance. Eight studies (six population surveys and two screening studies) were of both medium relevance and quality. We analysed their associations between SEC and prevalence or positivity, and report these results below. We do not report associations between SEC and chlamydia from the remaining 19 studies because we judged their quality or relevance to be too low to inform recommendations for delivering chlamydia testing and treatment in England. The web appendix contains tables displaying appraisal scores, with a commentary on the relevance and quality of all studies.

Associations between chlamydia and SEC in medium-rated studies (n=8)

Five population prevalence surveys and three screening intervention studies report ORs for testing positive for chlamydia between individuals of different SEC (table 2). In six studies, men and women were reported separately and two studies reported associations for more than one SEC measure. Thus in total, 17 analyses across eight studies have been reported.

Table 2

Associations between socioeconomic circumstances (SEC) and chlamydia prevalence in medium-rated studies

There was extensive heterogeneity in the SEC dimension chosen and the ways in which data were collected and analysed. Education was most commonly used. It was measured in three different ways: school type, past participation (using a dichotomous scale regarding whether the respondent graduated high school or not, or an ordinal scale: low, medium or high) and current educational status. Income was measured in two studies, using absolute income or a subjective assessment. One medium-rated study used an area-based measure.

Educational SEC measures

Studies using measures of education were most likely to report higher chlamydia prevalence associated with lower SEC. Thus ten of the 11 unadjusted analyses were statistically significant, with chlamydia prevalence ranging from 1.4 times to 13 times higher in groups with low education compared with groups of high education. The strongest association was reported in NatChla, a home-based chlamydia testing programme conducted in a subsample of Contexte de la Sexualité en France, a population-based survey of sexual behaviour. Women with low educational participation were 13 times more likely to test positive for chlamydia than those with high participation, but there was no association among men (table 2).17 This study was judged to have medium population relevance because, although it was conducted outside the UK, NatChla oversampled participants aged 18–24 years. It collected data using a telephone survey, in which 70% of the eligible population took part. Analyses were weighted for differential response rates and so this study scored highly on quality.17

Strong associations were also reported by school type in two studies of chlamydia screening interventions. In a Philadelphia-based study, adjusted prevalence in disciplinary schools (ie, for students excluded from mainstream education) was nearly nine times higher than in magnet schools (public schools offering specialised curriculum, often with high academic standards).18 In a Belgian study, unadjusted prevalence in technical and vocational schools were 10–13 times higher than in general/art schools (table 2).19 We judged these study populations to have medium relevance. However, both studies scored high on quality because of their high response rates. Thus, the American study included 19 394 young people, where 65% of students were tested; the Belgian study reported results from 2800 16–23-year-olds where 88% of young people who were offered a chlamydia test provided a specimen.19

Two US studies—the National Health and Nutrition Examination Survey (NHANES) and the Longitudinal Study of Adolescent Health (AddHealth)—used more than one SEC measure.7 ,20 In NHANES, adjusted and unadjusted prevalences were significantly higher (2.9 times and 3.3 times respectively) in those who had not graduated high school compared with those who had higher levels of education. In AddHealth, chlamydia prevalence was significantly higher in participants not in education compared with current students (1.6 times in men, 1.8 times in women in unadjusted analyses) and in men without a high school degree than men with higher levels of education (2.2 times in unadjusted analyses). When adjusted for ethnicity, sexual behaviour, perceived risk and healthcare use, significant associations were attenuated (table 2). NHANES used face-to-face interviews with participants aged 14–39 years, of whom 60% were aged 14–29 years. In AddHealth, students were initially selected from a random sample of secondary schools and surveyed at three different time points (waves). Data from Wave III, included here, comprise results from 18–26-year-olds. These two national samples had comparable ethnicity and age profiles to 15–24-year-olds in England, and so were judged to be of medium relevance. In both surveys, response rates were high and analyses were weighted for differential response rates, and thus were judged of high quality.

A Dutch study reported a gradient in chlamydia prevalence in adjusted and unadjusted analyses with the prevalence in those who did not complete secondary school being three times higher than those who continued in education post-school (table 2).8 This study was judged to have medium population relevance because it comprised a high proportion of participants aged 15–24 years in a population outside the UK, with comparable ethnicity profiles to the UK. The authors sampled households in four health service areas but just 41% responded to the postal questionnaire, so the study's quality was medium.8

Income-based or employment-based SEC measures

In eight analyses using income or employment measures, there were no significant associations between SEC and chlamydia prevalence. The second UK National Survey of Sexual Attitudes and Lifestyles (Natsal-2) collected information on employment. A standardised scale, the Registrar General classification system, was used to assign social class to occupational group. Chlamydia prevalence was not higher in men or women of lower social class compared with those in higher social classes (table 2).21 Natsal-2 was conducted on a large, representative probability sample of the British population. It scored high on study quality: 71% of the eligible population took part in the survey of randomly sampled households, with analysis weighted for non-response and specifically for differential response by socioeconomic group.21 ,22 However, only 26% of the study population were aged 18–24 years, resulting in only medium population relevance.

Two US studies (NHANES and AddHealth) also examined associations of income or employment with chlamydia prevalence. NHANES compared those on annual salaries above and below $20 000, and found ORs that were slightly weaker in magnitude compared with those for education (2.43 vs 3.27). These results were non-significant after adjusting for other variables.7 ,20 AddHealth applied a subjective measure of poverty (ability to pay utility bills) and a dichotomous measure of employment (job or no job). Non-significant associations were found in both cases.

Area-based SEC measures

One study of postal screening in England used an area-based measure to assign SEC. It also reported no significant association between chlamydia prevalence and SEC (table 2).23 SEC was defined by the Index of Multiple Deprivation 2004, which combines local measures of income, education and employment with other local characteristics (health, access to services, crime and housing) to produce a single summary score for an area. Index of Multiple Deprivation 2004 scores were assigned on the basis of respondents' general practice areas. The study had high population relevance; it was conducted in the West Midlands and Avon in England, and 75% of respondents were aged 16–24 years. However, it scored medium on quality because just 32% of this age group provided a specimen for chlamydia testing and uptake rates were lower in deprived than in advantaged areas.23


Main findings

We found that the current published literature is limited in its capacity to draw conclusions about the relationship between SEC and chlamydia prevalence in young people. Significant associations were most commonly found when measures of education were used. A diverse range of SEC measures were used in the 28 studies examined. Just eight of these studies were of sufficient quality and relevance to allow us to undertake a narrative synthesis of associations between SEC and chlamydia.

Explanations for our findings

Educational circumstances

Measures of educational level were most likely to demonstrate an inverse relationship between chlamydia prevalence and SEC. However, the mechanism by which education may influence the acquisition and transmission of chlamydia in young people is not clear. One hypothesis is that educational circumstances shape young people's sexual behaviour because education promotes aspirations.24 To realise these aspirations, young people adhere to social expectations about delaying intercourse, limiting partner numbers and using contraception. If individual sexual behaviours do indeed explain higher chlamydia prevalence in people with lower education, then SEC associations would be expected to reduce or disappear after adjusting for this. Unfortunately, the heterogeneity of behaviours adjusted for in the studies included in this review makes it difficult to reach conclusions. However, in the three studies that adjusted for individual sexual behaviours, associations between chlamydia prevalence and education in fact became stronger, indicating that factors other than individual sexual behaviour mediate or confound the association between education and chlamydia prevalence.7 ,8 ,17

Goulet et al (2010) propose that differential access to healthcare may explain the associations between chlamydia prevalence and education observed in women in their study.17 This research was conducted in France, where there is no free opportunistic screening. Women may be offered chlamydia tests in private settings for a fee. In this context, inequalities in education may reflect income inequalities, which directly determine women's ability to pay for testing.17 This explanation cannot apply to the English context where screening is free and people in disadvantaged areas do not have reduced access to testing.25

A third hypothesis proposed is that the significant association between education and chlamydia prevalence reflects unmeasured ‘risks involved in sexual partner choice’.8 For example, educational settings may determine young people's sexual networks in the same way that they influence social networks, that is, through school and extracurricular activities.8 ,19 This explanation may account for why variations in prevalence were greater when measures of school type were used rather than measures of individual education. Although this hypothesis has not been tested empirically, our findings provide an argument for further exploration.

Income and employment

The failure of income and employment measures to adequately capture SEC in young people may be because many 16–24-year-olds are still in full-time education and have yet to enter the labour market.26

Area-level disadvantage

International analyses of large surveillance datasets consistently report higher rates of chlamydia in disadvantaged areas.27–29 In the USA, for example, diagnoses rates are six times higher in deprived than in affluent areas.27 Although these studies suggest that young inhabitants of socioeconomically disadvantaged areas are at high risk of STIs, the current literature is of insufficient methodological quality to draw conclusions about the influence of area disadvantage on chlamydia prevalence. The single medium-rated area-based study included in our review found no significant association. However, it had lower uptake in deprived areas and SEC was assigned on the basis of patients' general practice postcode, which may underestimate the strength of a socioeconomic association. We found that other studies using area-based measures generally reported higher chlamydia positivity in deprived areas, albeit of weaker magnitude than those found in notification studies.25 However, these studies were of low quality because they often used routinely collected data where researchers had little control over the data collection. As a result, these studies, even when well conducted, are often prone to significant methodological limitations.

Methodological considerations


This is the first systematic review of the association between chlamydia prevalence or positivity and SEC. By adopting rigorous exclusion criteria, we excluded important sources of potential bias.


Although 28 studies met our inclusion criteria, none were of both high quality and high relevance to our research objective. This limits the extent to which we can draw conclusions about the relationship between SEC and chlamydia prevalence in young people. SEC was rarely the focus of studies examining the epidemiology of chlamydia. As a result, significant associations may have been missed because studies were not designed for this particular exposure. Alternatively, studies which examined socioeconomic associations with chlamydia may not have reported non-significant findings. If so, our results may be biased in favour of a significant association between chlamydia and SEC.

We were unable to produce a pooled estimate of the risk of chlamydia associated with SEC because of the heterogeneity of study designs and measures of socioeconomic exposure. However, it has been argued that meta-analysis should not be a prominent component of reviews of observational studies, partly because of the difficulties in controlling for potential confounders. This means that studies with large numbers of participants could lead to biased pooled estimates.30 Furthermore, a pooled estimate of risk across studies using different SEC indicators would mask variations that may contribute to our understanding of how SEC may affect young people's risk of chlamydia.

Conclusions and implications for practice

A relationship between SEC and the risk of chlamydia cannot be assumed to apply in young people. Furthermore, different conclusions may be reached depending on which SEC dimension is measured. There was a dearth of high-quality studies using area-based measures. Relationships between SEC and chlamydia prevalence were most consistent when measures of education were used. However, national population studies are required that analyse both area-level indicators of SEC and individual age-appropriate educational variables to confirm an association. Research is also needed to elucidate the pathway between educational circumstances and chlamydia prevalence more clearly. If clear pathways are identified, this could inform the development of targeted strategies for preventing chlamydia transmission and acquisition.

Key messages

  • Socioeconomic circumstances (SEC) have been considered as a criterion for targeting large-scale chlamydia screening programmes.

  • We found that a relationship between SEC and chlamydia cannot be assumed to apply in young people. This relationship may depend on how SEC is measured.

  • Relationships between SEC and chlamydia prevalence were most consistent when using measures of education, where higher education was associated with lower chlamydia prevalence.

  • National population studies which analyse both area-level SEC indicators and individual age-appropriate educational variables may clarify whether targeted screening would be beneficial for this population.


We thank Jackie Cassell, Professor of Primary Care Epidemiology, Brighton and Sussex Medical School, for her advice on the retrieved references and on the variations in the epidemiology of different sexually transmitted infections; Sara Clarke, Knowledge Resources Librarian, Royal Free Medical Library for her helpful advice on the search strategy; and Dr Cath Mercer, Senior Lecturer, Infection and Population Health, University College London, for her advice on Natsal-2 and her helpful comments on an earlier version of this paper.


View Abstract
  • Supplementary Data

    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:


  • Funding This work was supported by the Medical Research Council (grant number G0701660).

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

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.