Effects of population based screening for Chlamydia infections in the Netherlands limited by declining participation rates

Abstract

Background: A large trial to investigate the effectiveness of population based screening for chlamydia infections was conducted in the Netherlands in 2008-2012. The trial was register based and consisted of four rounds of screening of women and men in the age groups 16-29 years in three regions in the Netherlands. Data were collected on participation rates and positivity rates per round. A modeling study was conducted to project screening effects for various screening strategies into the future.

Methods and findings: We used a stochastic network simulation model incorporating partnership formation and dissolution, aging and a sexual life course perspective. Trends in baseline rates of chlamydia testing and treatment were used to describe the epidemiological situation before the start of the screening program. Data on participation rates was used to describe screening uptake in rural and urban areas. Simulations were used to project the effectiveness of screening on chlamydia prevalence for a time period of 10 years. In addition, we tested alternative screening strategies, such as including only women, targeting different age groups, and biennial screening. Screening reduced prevalence by about 1% in the first two screening rounds and leveled off after that. Extrapolating observed participation rates into the future indicated very low participation in the long run. Alternative strategies only marginally changed the effectiveness of screening. Higher participation rates as originally foreseen in the program would have succeeded in reducing chlamydia prevalence to very low levels in the long run.

Conclusions: Decreasing participation rates over time profoundly impact the effectiveness of population based screening for chlamydia infections. Using data from several consecutive rounds of screening in a simulation model enabled us to assess the future effectiveness of screening on prevalence. If participation rates cannot be kept at a sufficient level, the effectiveness of screening on prevalence will remain limited.

Figure 1. Participation trees for women and men.

First-time participation was modelled to depend only on gender (panel A for women, panel B for men), and year since the start of the screening program. Subsequent participation depended solely on the previous choices made. All rates are based on observed participation rates in subsequent rounds of CSI. The extrapolation for years after the 4 years of CSI is based on results of the fourth round as described in the text.

Figure 2. Effect of CSI on population prevalence of Chlamydia infections.

The projected Chlamydia prevalence for men (solid lines) and women (dashed lines) for the baseline scenario, CSI screening, and alternative scenarios. Panel A shows the projected prevalence on the national level, Panel B shows the projected prevalence in urbanized areas.

Figure 3. Effect of CSI on population prevalence of Chlamydia infections by age category, on the national level.

Prevalence levels are shown for a scenario without screening implementation (“baseline 2011”), and after 3 (“CSI 2011”) and 10 years (“CSI 2018”) of screening, for (A) women and (B) men. Splitting the population into age-groups gives a detailed view on the effect of CSI screening in addition to baseline testing and treatment at GPs and STD clinics.

Figure 4. Effect of CSI on population prevalence of Chlamydia infections by age category, in urbanized areas.

Prevalence levels are shown for a scenario without screening implementation (“baseline 2011”), and after 3 (“CSI 2011”) and 10 years (“CSI 2018”) of screening, for (A) women and (B) men. The effect on Chlamydia prevalence is most visible in the age-groups 21–25 for both women and men.

Figure 5. Effect of a high CSI participation rates on population prevalence of Chlamydia infections.

In contrast to the reduction in Chlamydia prevalence achieved by screening with the observed participation rates, CSI screening with a stable participation rate of 30% on national level (A) and 25.6% in urbanized areas (B) would lead to a drastic reduction in Chlamydia prevalence in men (solid lines) and women (dashed lines). On the national level, closed populations of 50,000 individuals are frequently unable to maintain Chlamydia in the population, and the average Chlamydia prevalence reported in panel A is therefore a combination of simulated populations where Chlamydia has gone extinct, and where Chlamydia is maintained at low prevalence levels.