Objectives Pre-exposure prophylaxis (PrEP) users are routinely tested four times a year (3 monthly) for asymptomatic Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) infections on three anatomical locations. Given the high costs of this testing to the PrEP programme, we assessed the impact of 3 monthly screening(current practice), compared with 6 monthly on the disease burden. We quantified the difference in impact of these two testing frequencies on the prevalence of CT and NG among all men who have sex with men (MSM) who are at risk of an STI, and explored the cost-effectiveness of 3-monthly screening compared with a baseline scenario of 6-monthly screening.
Methods A dynamic infection model was developed to simulate the transmission of CT and NG among sexually active MSM (6500 MSM on PrEP and 29 531 MSM not on PrEP), and the impact of two different test frequencies over a 10-year period. The difference in number of averted infections was used to calculate incremental costs and quality-adjusted life-years (QALY) as well as an incremental cost-effectiveness ratio (ICER) from a societal perspective.
Results Compared with 6-monthly screening, 3-monthly screening of PrEP users for CT and NG cost an additional €46.8 million over a period of 10 years. Both screening frequencies would significantly reduce the prevalence of CT and NG, but 3-monthly screening would avert and extra ~18 250 CT and NG infections compared with 6-monthly screening, resulting in a gain of ~81 QALYs. The corresponding ICER was ~€430 000 per QALY gained, which exceeded the cost-effectiveness threshold of €20 000 per QALY.
Conclusions Three-monthly screening for CT and NG among MSM on PrEP is not cost-effective compared with 6-monthly screening. The ICER becomes more favourable when a smaller fraction of all MSM at risk for an STI are screened. Reducing the screening frequency could be considered when the PrEP programme is established and the prevalence of CT and NG decline.
- Chlamydia trachomatis
- Neisseria gonorrhoeae
- gay men
Data availability statement
Data sharing not applicable as no datasets generated and/or analysed for this study. All data relevant to the study are included in the article or uploaded as supplementary information.
Statistics from Altmetric.com
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.
Pre-exposure prophylaxis (PrEP) is used for HIV prevention, and the efficacy in preventing HIV acquisition among men who have sex with men (MSM) is high.1 2 Currently, health programmes providing PrEP are implemented worldwide and include regular STI testing.3 4 The overall costs of these programmes are determined by the costs of PrEP medication and related care. One approach to lower the total costs per individual on PrEP is less frequent STI testing, especially for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) infection. This can substantially lower the overall costs, due to the relative high prices of laboratory tests and testing samples from three anatomic locations (urethra, rectum and pharynx).5 NG infections are less likely to be asymptomatic, more likely to have sequelae compared with CT,6 and antimicrobial resistance (AMR) is an emerging threat.7 The costs of STI testing can be up to five times the costs of medical consultations as part of PrEP-related care.5
Less frequent screening will lengthen the infectious period of asymptomatic CT and NG infections. This might have minimal negative consequences for the male health, since complications are rare in males, but may allow for more onward transmission possibly increasing number of infections. The magnitude of this is not clear. To our knowledge, no randomised controlled trials have evaluated the impact of less frequent screening of MSM on the prevalence of STIs. Modelling studies evaluating the impact of different screening strategies for STIs concluded that frequent testing could substantially decrease the number of infections.8 9 More information is needed on optimal STI screening strategies for MSM using PrEP, including their cost-effectiveness and their effects on STI prevalence.
From August 2019, a national PrEP programme started in the Netherlands for MSM at high risk of acquiring HIV.10 In this programme, both PrEP medication (tenofovir disoproxil/emtricitabine) and related care, including counselling and 3-monthly testing for HIV, other STIs and renal function, is provided for up to 6500 persons for a period of 5 years.10 The current Dutch PrEP guidelines advise 3-monthly screening for CT and NG infections,11 while a policy paper recommends that sexually active MSM get tested for STIs at least every 6 months regardless of symptoms.12 However, there is an ongoing debate on optimal STI screening strategies in the context of PrEP.13
We compared, through mathematical modelling, a 3-monthly screening programme for CT and NG with a 6-monthly screening programme among MSM included in the national PrEP programme in the Netherlands. We aimed to quantify the difference in impact of these two possible programmes on the prevalence of CT and NG among all MSM who are at risk of an STI, and to explore the cost-effectiveness of 3-monthly screening versus 6-monthly screening.
To simulate the effects of less frequent screening for CT and NG infections in Dutch MSM a dynamic infection model was developed. The model consists of three mutually exclusive infection states, that is, susceptible to infection, infected without symptoms and infected with symptoms. Individuals can become infected when susceptible and subsequently have a probability to develop symptoms, and after recovery (due to spontaneous clearance or treatment) become susceptible again. We assumed that symptomatic and asymptomatic infections are equally infectious. People can become infected at the urethra or the rectum. Pharyngeal infections were not included, since those infections often occur with a concurrent rectal or urethral infection.14 The model was run separately for CT and NG infections. The duration of infection and the prevalence of asymptomatic and symptomatic infections (table 1) were used to infer per day the probability to become infected (force of infection (FOI)) and a probability of being symptomatically infected such that the steady state (model results before implementation of the programme) reflected the data. At steady state screening every 6 months (180 days) or 3 months (90 days) was introduced in the model. We assumed that background testing for STIs was reflected in the ratio of asymptomatic and symptomatic infections. Asymptomatic infections would only be diagnosed at screening visits. Model equations are presented in online supplemental appendix I. We assumed that STIs transmitted across the groups of MSM on PrEP and not on PrEP.
The duration of asymptomatic infections, and the percentage of symptomatic infections were based on cross-sectional studies.15–17 For STI prevalence in steady state, by proxy, positivity data from the inclusion visits of the Dutch AMPrEP study were used, a PrEP demonstration project among 376 HIV-negative MSM and transgender persons.18
The actual size of the MSM population in the Netherlands is unclear. We used the number of MSM that visited a sexual health centre in 2017 (n=29 531), as an approximation of the number of MSM that contribute to the transmission of STIs in the MSM population.19 It was assumed that 6500 of these MSM will use PrEP, since that is the maximum capacity of the national PrEP programme.10 The MSM population was assumed to be closed.
For simplicity, we assumed that all MSM on PrEP return for their screening visits, and that all infections are identified and treated successfully. We further assumed that all PrEP users were tested and treated at the same instant, resulting in a prevalence of 0% among all PrEP users after screening. These assumptions were evaluated in the sensitivity analysis.
Health effects were measured as loss in quality-adjusted life-year (QALYs) associated with a health condition, which is calculated using the utility and duration of that health condition in years (online supplemental appendix II, table A1).20 The utility was based on the Health Utilities Index developed by a committee of the Institute of Medicine (US) in Vaccines for the 21st century.21 The probabilities to develop each health condition are presented in online supplemental table A1. We assumed that asymptomatic infections do not lead to loss in health-related quality of life.
An activity-based costing approach was used. The healthcare costs consisted of medical consultations and laboratory tests related to visits to the STI clinic, and the related treatment (online supplemental appendix II, table A2). No costs were assigned to the diagnosis and treatment of asymptomatic infections outside PrEP visits, since we assumed that individuals do not visit the STI clinic for asymptomatic infections. Unit prices for laboratory tests were obtained from the Dutch Healthcare Authority and medication costs were retrieved from the National Health Care Institute.22 23 Consultations performed by nurses at the STI clinic were valued using hourly personnel costs including overhead costs and a fixed duration per consultation.
Non-healthcare costs consisted of travel costs and productivity loss due to a visit to the STI clinic. They were valued using the Dutch reference prices and average working time in the Netherlands.24 25 In case of an epididymitis, a rare complication of CT and NG,20 the additional costs for a visit to the general practitioner and hospital stay were included. Productivity loss for a visit to the STI clinic was assumed to be half a workday and for an epididymitis resulting in hospitalisation 10 workdays.
All costs are presented in 2018 Euros; where needed, costs were converted using the Dutch Consumer Price Index.26
Differences in costs and effects, from a societal perspective, were calculated between 3-monthly and 6-monthly screening over a 10-year horizon. Costs were discounted with 4% and QALYs with 1.5%, according to the Dutch guidelines.25 The incremental cost-effectiveness ratio (ICER) was calculated by dividing the incremental costs by the incremental QALYs to display the additional costs per QALY gained. Since 3-monthly screening is current practice in the Netherlands, that would be the reference scenario and 6-monthly screening the alternative scenario. Since 3-monthly screening is the most costly scenario, the ICER would be negative, reflecting the savings per QALY lost. Therefore, we used 6-monthly screening as the reference scenario and 3-monthly screening as the alternative to obtain a positive ICER, reflecting the additional costs per QALY gained. An ICER threshold of €20 000 per QALY gained was used, which is the informal threshold for prevention interventions in the Netherlands.27
A sensitivity analysis was performed by varying the following: time horizon of the model; costs related to STI testing and PrEP consultations; utilities per health condition; discounting rates; number of MSM on PrEP and MSM at risk of an STI; prevalence and percentage symptomatic infections at initiation of PrEP; duration of asymptomatic and symptomatic infection. We assessed the presence of risk compensation by increasing the FOI for individuals on PrEP with 20%. Also, a cost-effectiveness analysis was performed from a healthcare payer perspective.
All analyses were performed using R V.3.5.1 (packages ‘tidyverse’, ‘gridExtra’).
Base case analysis
Three-monthly screening of 6500 MSM for CT and NG during 10 years cost, after discounting, €46.8 million more than 6-monthly screening. Most additional costs were made for the STI tests (€23.7 million), followed by the costs for productivity loss (€17.8 million) and the costs for the consultations (€4.9 million).
Impact on the number of infections and cost and QALYs
Regular screening of 6500 MSM on PrEP out of a total population of 29 531 MSM at risk of an STI reduced the prevalence of CT and NG infections among users and in the total population, regardless of the screening interval (figure 1). Three-monthly screening would avert ~11 550 more CT and 6700 NG infections, compared with 6-monthly screening (table 2). Most averted infections were symptomatic urethral CT infections. More infections were averted in the population not on PrEP compared with the population on PrEP, respectively ~14 200 and ~4000, due to the larger population size of those not on PrEP. After 10 years, the total number of diagnosed asymptomatic infections, which were assumed to be diagnosed at screening visits only, was slightly higher with 6-monthly screening compared with 3-monthly screening, due to a smaller drop in FOI with 6-monthly screening. Averting infections would save, after discounting, €7.2 million and would gain 92.1 QALYs.
The discounted incremental costs and health effects of 3-monthly screening for CT and NG infections in MSM who are on PrEP vs 6-monthly screening are €39.6 million and 92.1 QALY, resulting in an ICER of ~€430 000 per QALY gained, which is not cost-effective under the Dutch threshold.
The results of the sensitivity analyses in which we varied several parameters to assess the range of variation in the model are shown in figure 2 and online supplemental appendix II, table A3. Parameters with a variation of <5% from the ICER of the base case are not shown in figure 2, including duration of symptomatic infection and duration of asymptomatic urethral NG infection.
A crucial parameter was the ratio of MSM on PrEP to all MSM who are at risk of an STI. If a smaller fraction of all MSM at risk of an STI were on PrEP, the impact of testing on the reduction of STI prevalence decreased, resulting in a more favourable ICER for 3-monthly testing.
The ICER increased when assigning higher utilities of the health states.21 28 Of all biological and epidemiological parameters, urethral CT infection parameters were most influential, although still not affecting the ICER substantially. Including risk compensation for MSM on PrEP improved the ICER with 32%. Reducing the costs for STI testing to 10% of the original costs improved the ICER with approximately 50%. Both improvements were not enough to reach the threshold for cost-effectiveness of €20 000.
Increasing the frequency of testing CT and NG among MSM enrolled in the Dutch national PrEP programme from every 6 months to every 3 months will cost €39.6 million, with limited positive health effects. We found that 3-monthly screening compared with a baseline of 6-monthly screening is not cost-effective. This is because regular testing and treating of STIs among 22% of all MSM who are at risk of an STI in a PrEP programme is projected to result in a significant reduction of the STI prevalence within 10 years. Increasing the screening frequency to every 3 months can only avert a small number of additional incident infections, resulting in a limited incremental health benefit. More frequent screening averted mostly symptomatic infections, which would also have been diagnosed without frequent screening because individuals will seek for care due to symptoms. Next to high costs of testing, frequent STI screening also has potential important negative consequences. The prevalence of AMR in NG is increasing and frequent screening could contribute to the emergence of AMR, by increased use of antibiotics.7 This also needs to be taken into account when determining the optimal screening frequency.
One of the parameters that strongly influenced the ICER was the ratio of MSM on PrEP to all MSM who are at risk of an STI, which can be interpreted as the size of the national PrEP programme in respect to the overall at-risk MSM population. With a small programme or a large overall at-risk MSM population, our model showed a limited decrease in prevalence of STIs in all MSM at risk of an STI. With a large programme, 6-monthly screening is already frequent enough to reduce STI prevalence substantial over 10 years. Thus, screening a small part of the total population of MSM who contribute to the transmission of STIs will not have a high impact on the overall transmission. Thereby, the overall incidence of STIs will remain stable over the modelled period of 10 years and more frequent screening can avert a large number of incident infections. Indeed, most PrEP studies, in which only a small number of individuals are included compared with the total population, report stable or increasing rather than decreasing trend in STI positivity.18 29 When the reduction in prevalence due to screening is limited, the overall incidence of STIs will be high and more frequent screening can avert a large number of incident infections. As a consequence, the difference in health effects between 3-monthly and 6-monthly screening will increase with a larger programme, resulting in a more favourable ICER. Since not all MSM at risk visit a sexual health centre, the overall at-risk MSM population used in our study might be underestimated. Thereby the cost-effectiveness of 3-monthly compared with 6-monthly screening may be more favourable.
In the sensitivity analysis, no realistic cost-saving measures were identified that would result in reaching the cost-effectiveness threshold. For example, reducing the STI testing costs to 10% of the original costs did not make 3-monthly compared with 6-monthly testing cost-effective.
A major strength of our analysis is the activity-based costing approach, which is one of the most accurate and detailed approaches for estimating the total costs. Another strength is that differences in biological and epidemiological parameters for the type of infection and anatomical site being infected was considered in the model. This is important because the results showed differences between the type and anatomical location of infections in FOI and impact of screening.
This study has some limitations. We developed a relatively simple model, because the aim of our study was to investigate the incremental benefit of 3-monthly screening over 6-monthly screening rather than the absolute impact of screening. As a consequence, we side-stepped some of the complexities of CT and NG transmission. For example, concurrent CT and NG infections or simultaneous infections at multiple anatomic locations were not possible in the model, even though such co-infections are common.14 We assumed that these complexities were equal for the group on PrEP and the group not on PrEP and for 3-monthly and 6-monthly testing, and would therefore not substantially impact our incremental results. A consequence of not including concurrent infections is that even more infections would be diagnosed and treated, resulting in a larger reduction of the STI prevalence, a smaller incremental health benefit of 3-monthly testing and a less favourable ICER. Furthermore, we did not include possible interactions with HIV transmission. We expect that this will minimally affect our results, because the incidence of HIV in MSM PrEP users is very low.18
Moreover, pharyngeal infections were not explicitly included in the dynamic model. Less frequent testing will lengthen the infectious period of asymptomatic, pharyngeal-only infections, and thereby result in a smaller reduction of the STI prevalence, which would increase the incremental health benefit of 3-monthly testing. However, they were included implicitly, as the FOI we estimated for each of the two locations was based on the prevalence of each location and the duration of carriage. This prevalence also includes the infections which originates from a pharyngeal infection. Thereby, we expect that explicitly including pharyngeal infections in the model would have limited effect on our results.
In addition, it was assumed that all MSM using PrEP were tested at the same day, which does not reflect reality. So the sudden drops, with a zig-zag pattern in STI prevalence at each testing event will not happen in reality. Although this slightly exaggerated the impact of testing, it does not exaggerate the difference between 6-monthly and 3-monthly testing, and therefore will not change our conclusion.
In conclusion, in this modelling study a profound impact on the transmission of CT and NG was seen with a reasonably large PrEP programme with both 3-monthly and 6-monthly screening of PrEP users. Three-monthly screening for CT and NG of all MSM included in a relatively large PrEP programme was not cost-effective compared with 6-monthly screening, because 6-monthly screening already reduced the prevalence of CT and NG among all MSM at risk of STI. Thus, attracting and retaining a substantial proportion of MSM at high risk for STI in PrEP programmes contributes to reducing HIV, and to reducing STI prevalence. Although our results are clear, and make a strong case for decision-makers to reconsider the 3-monthly testing frequency within a PrEP programme, we used a simple model with several assumptions. We therefore argue for a pragmatic interpretation: we recommend to monitor the prevalence of CT and NG in PrEP programmes for MSM; in case of decreasing STI prevalence, 6-monthly screening should be considered to save costs and reduce antibiotic use without negative consequences for either the individual or public health.
One approach to lower the total costs per individual on pre-exposure prophylaxis (PrEP) is less frequent testing for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) infections.
With screening of a reasonably large men who have sex with men (MSM) population, a profound impact on the transmission of CT and NG is projected.
Three-monthly screening is not cost-effective compared with 6-monthly screening, because only a small number of additional incident infections would be averted.
The prevalence of CT and NG in the MSM population should be monitored; in case of decreasing prevalence, 6-monthly screening of MSM on PrEP should be considered.
Data availability statement
Data sharing not applicable as no datasets generated and/or analysed for this study. All data relevant to the study are included in the article or uploaded as supplementary information.
Patient consent for publication
Handling editor Laith J Abu-Raddad
Contributors Study concept and design: AJvH, EH, MFSvdL, FvW. Model development: AJvH, FvW. Analysis and/or interpretation of the data: AJvH, EH, MFSvdL, FvW, JH. Drafting of the manuscript: FvW. Critical revision of the manuscript: all authors.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests EH obtained advisory board fees and a research grant from Gilead Sciences; both of which were paid to her institute.
Provenance and peer review Not commissioned; externally peer reviewed.