Key findings
Our systematic review found strong evidence of an association between chlamydia infection and socioeconomic position in both men and women (Combined OR: 1.66, 95 % CI: 1.37 to 2.02). Pooled results were equivalent to a doubled risk of chlamydia infection for those with lower educational attainment (Combined OR: 1.94, 95 % CI: 1.52 to 2.47). Risk of infection was also greater in those with lower occupational class or unemployment and greater area deprivation. No association was found between chlamydia infection and parental or household measures of income, occupation or education.
There was considerable variability in the prevalence of chlamydia between population-based prevalence studies. This variation presumably in part reflects differences in populations and underlying prevalence between countries. However, there were also variations in estimated prevalence within populations. Evidence of heterogeneity remained strong, even after stratifying meta-analyses by gender, age and region. Prevalence estimates in young men were lower in those aged under 20 years than 20 to 24 year olds in some studies but tended to be similar in women aged under and over 20 years.
Meta-regression suggested that age and gender may contribute to the heterogeneity of prevalence estimates. Interpreted on the prevalence scale, multivariable meta-regression suggested that prevalence is on average 1.1 % (95 % CI 0.1 to 2.1 %) higher in women than men and 1.1 % (95 % CI 0.0 to 2.2 %) higher in 20 to 24 year olds compared to under 20 year olds. However, the variables included in the regression model explained only a modest amount of the between-observation variance (adjusted R2 = 21.5 %) and strong evidence of residual variation due to heterogeneity remained (I2 = 79.2 %). Other potential sources of heterogeneity include the residual influence of study characteristics such as non-response bias, sampling bias and differences in true prevalence between populations.
Strengths and limitations
Strengths of our review include use of a pre-specified protocol, a systematic and comprehensive search strategy tailored to each bibliographic database, inclusion only of population-based studies of prevalence using an objective diagnosis of chlamydia infection, duplicate eligibility screening and data entry, and the lack of exclusions based on language or publication date. It was also possible to explore sources of heterogeneity using meta-analysis and meta-regression.
A challenge experienced in our review was the strong evidence for heterogeneity between studies even after stratifying by gender, age, and geographic region, which lead us to conclude that pooled estimates of prevalence would not be valid. Potential sources of heterogeneity between studies included study design, measurement of socioeconomic position, categorisation of reference groups and analysis of confounding. Differences in age groups and other variables between studies also limited the comparability of observations and prevented inclusion of socioeconomic position in meta-regression analyses. Ethnicity and level of urbanisation are other factors that may contribute to variation in prevalence between populations [
4,
28], but were beyond the scope of the review.
The reliability of prevalence estimates in our review is limited by the risk of bias in the studies included. Overall, 95 % of the studies were assessed to be high or medium risk of important bias, particularly selection bias, in their estimates. Unfortunately, most studies included in this review had a lack of data on non-responders, which meant that it was not possible to adjust estimates for non-response using multiple imputation, inverse probability weighting or other statistical approaches to missing data [
32].
Meta-regression suggested an explanation for only a small proportion of the between study variation in prevalence estimates observed in this review. In addition this approach involves the implicit assumption that true prevalence is the same in different populations. In this context, meta-regression mainly serves to identify hypotheses to be explored in future studies [
49,
50]. Further, the measures of possible sources of between study variation used in meta-regression analyses in this review were relatively crude and likely to themselves be subject to measurement error. For example, response rates are only a proxy for the potential for selection bias, and whether or not data collection occurred before/after 2006 will only crudely index an influence of the introduction of Chlamydia control programmes. There were too few studies to robustly examine the influence of country where the study was carried out on variation in prevalence.
There were also limitations related to analysis of socioeconomic position in this review. Reporting bias may have led to overestimation of associations. At least one study [
26] did not report the results of analyses where no association was found. Other potential sources of bias could have worked in either direction, for example potential residual confounding and adjustment of socioeconomic position for variables that may be on the causal pathway between exposure and outcome or consequences of the outcome (such as early sexual debut, number of sexual partners and symptoms of infection). Five out of thirteen observations included in the meta-analysis of educational measures were not adjusted for potential confounders, because no positive association was found in unadjusted analyses. In one study, the association between parental socioeconomic disadvantage and chlamydia was substantially attenuated and reversed in direction after adjusting for drug taking, contraceptive use and exposure to passive smoke [
38]. In accordance with the study protocol, adjusted estimates were used in this review; however, both adjusted and unadjusted estimates may be subject to bias.
Some health services data and studies from some settings suggest that there are inequities in the burden of chlamydia infections between ethnic groups, with higher rates in some black ethnic groups than other black and non-black ethnic groups [
4,
8,
46,
51‐
53]. UK population-based surveys have been inconclusive in this regard, which may reflect issues related to sample size or other methodological challenges [
37,
51]. Chlamydia prevalence has also been found to vary by geographic location, including between countries, and according to urban/rural residence in some studies [
28,
54] and between regions and cities in the same country in others [
43,
46] Detailed consideration of these questions is beyond the scope of this paper. Our focus on variations in prevalence by age, gender and socioeconomic position in part reflected the potential importance of these demographic factors for the design of control interventions, and was because extant evidence of their association with chlamydia infection appeared inconclusive [
10,
55].
Research and policy implications
The present systematic review builds on two recent reviews [
10,
55] by including more recent studies and by using meta-analysis and meta-regression to pool results and explore heterogeneity. One of these previous reviews examined inequities in prevalence by socioeconomic position and found inconclusive evidence of an association between chlamydia infection and socioeconomic position, concluding that a relationship cannot be assumed [
10]. The present review provides new evidence that young people-specific and area-based measures of socioeconomic position are associated with chlamydia prevalence when pooled across studies, and finds similar inconclusive results for parental or household measures.
The second review examined risk difference in prevalence between the sexes for individual studies and found that any difference in prevalence between women and men is likely to be modest [
55]. Our systematic review found weak evidence of differences in chlamydia prevalence by gender and age (under and over 20 years). These findings were from meta-regression and are best interpreted as hypotheses for testing in future research.
There are several possible mechanisms for socioeconomic inequities in chlamydia infections. These include lower engagement with Chlamydia control activities amongst disadvantaged individuals. Young people from families with lower socio-economic position may also be at greater risk of having a chlamydia infection, because individual, family, interpersonal, community and structural factors reduce the perceived benefits of safe sex, reduce consistency of condom use, reduce sexual health service provision and use, and increase other risk factors for unsafe sex, such as substance use, and mental health problems [
10,
56‐
58]. Potential reasons for gender differences in risk for chlamydia include age differences in sexual partnerships [
59], biological differences, cervical ectopy and use of hormonal contraceptives in women, and circumcision in men [
55]. However, apparent differences by gender may in part be due to selection bias, which may operate in a different way in each sex [
55].
Our review points to the need to monitor and address social variation in the risk of infection, in order to avoid the potential for screening programmes to exacerbate inequalities [
9]. In the early years of England’s National Chlamydia Screening Programme, screening provision, coverage and positivity were all higher in socio-economically deprived areas than more affluent areas [
13,
60]. However, recent population-based evidence on reported tests in the past year suggest that screening uptake is similar across all levels of neighbourhood deprivation [
37]. Our findings indicate the need for greater uptake of screening in disadvantaged areas to adequately address increased risk of infection. Another potential challenge for screening programmes is ensuring sufficient uptake of tests among people aged 20 to 24 years. This age group is more difficult to reach than younger age groups who are more likely to be in full-time education. Other studies suggest that chlamydia screening uptake may be lower in men and in older age groups [
13,
37]. More may need to be done to meet the need for screening in older men.