Article Text
PDF
Abstract
Objective The objective of this study was to estimate the cost and cost-effectiveness of opportunistic screening for Chlamydia trachomatis in Ireland.
Methods Prospective cost analysis of an opportunistic screening programme delivered jointly in three types of healthcare facility in Ireland. Incremental cost-effectiveness analysis was performed using an existing dynamic modelling framework to compare screening to a control of no organised screening. A healthcare provider perspective was adopted with respect to costs and included the costs of screening and the costs of complications arising from untreated infection. Two outcome measures were examined: major outcomes averted, comprising cases of pelvic inflammatory disease, ectopic pregnancy and tubal factor infertility in women, neonatal conjunctivitis and pneumonia, and epididymitis in men; and quality-adjusted life-years (QALY) gained. Uncertainty was explored using sensitivity analyses and cost-effectiveness acceptability curves.
Results The average cost per component of screening was estimated at €26 per offer, €66 per negative case, €152 per positive case and €74 per partner notified and treated. The modelled screening scenario was projected to be more effective and more costly than the control strategy. The incremental cost per major outcomes averted was €6093, and the incremental cost per QALY gained was €94 717. For cost-effectiveness threshold values of €45 000 per QALY gained and lower, the probability of the screening being cost effective was estimated at <1%.
Conclusions An opportunistic chlamydia screening programme, as modelled in this study, would be expensive to implement nationally and is unlikely to be judged cost effective by policy makers in Ireland.
- Cost-effectiveness
- economic analysis
- health service research
- chlamydia infection
- mathematical model
- cost-effectiveness
- genitourinary medicine services
- hepatitis B
- natural history
- antibiotic resistance
- sexual behaviour
- STD control
- Chlamydia trachomatis
Statistics from Altmetric.com
- Cost-effectiveness
- economic analysis
- health service research
- chlamydia infection
- mathematical model
- cost-effectiveness
- genitourinary medicine services
- hepatitis B
- natural history
- antibiotic resistance
- sexual behaviour
- STD control
- Chlamydia trachomatis
Introduction
National screening programmes to identify undiagnosed Chlamydia trachomatis infection have been introduced in a number of European countries.1 In Ireland, no such screening programme currently exists and debate surrounds the optimal design of and setting for such an initiative, were it to be implemented. Central to these discussions, given the increasingly resource constrained environment in which healthcare systems operate, are issues of cost and cost-effectiveness. The Chlamydia Screening in Ireland Pilot (CSIP) Study was a research programme established to explore the feasibility of chlamydia screening models in Ireland.2The operation of screening programmes in Ireland is complicated by the absence of unique patient identifiers. While these would allow a proactive approach to screen offer through patient registers, this is not possible for legal reasons in Ireland. The alternative is opportunistic approaches to screening. As part of the wider research programme, a pilot study of an opportunistic screening programme, approved by the ethics committees of the National University of Ireland and the Irish College of General Practitioners, was undertaken in Galway city and county over a period of 6 months in 2009.
In this study, we present the results of a prospective cost analysis conducted alongside the pilot study. In addition, we explore the cost-effectiveness of screening using an existing decision modelling framework.3 The model incorporates the dynamic effects of infectious disease transmission.4 In the absence of an Irish-specific equivalent, a dynamic model parameterised using data for sexual behaviour and chlamydia prevalence for the UK was used alongside an economic model adapted to the Irish setting. The use of a UK-based behavioural model may be viewed as problematic. As we argue, however, it provides a framework unlikely to differ materially from that in Ireland and permits the estimation of a range for cost-effectiveness likely to encompass that which exists in practice. An added benefit of this approach is that it allows for a comparison with the evaluation of screening in the National Health Service, which unlike Ireland provides universal access to free primary care at point of use. These findings will help inform the development of chlamydia screening policy in Ireland and add to the body of international evidence on the costs and cost-effectiveness of opportunistic chlamydia screening.
Methods
The process of the opportunistic screening programme, the cost analysis and the decision modelling analysis are detailed in the following sections.
CSIP Study: opportunistic screening programme
In the pilot study, individuals aged 18–29 years old who accessed one of three healthcare settings—general practices, family planning and student health clinics—were invited to participate in the study. Those who agreed were offered a screen. Men who agreed to participate were asked to provide a urine sample and women either urine or cervical swab. In general practice, both general practitioners (GPs) and practice nurses offered screening, while in family planning and student health clinics, offers were made by nurses. Urine or endocervical swab samples were tested for each participant by the virology laboratory at a local hospital. Specimens were batch tested with PCR testing technology. The test used was the COBAS® TaqMan® CT Test v2.0 manufactured by Roche Diagnostics, Rotkreuz, Switzerland. All individuals were notified of their test results by the healthcare provider, and those with positive results were invited to attend a consultation where they received treatment (with azithromycin or doxycycline), information and counselling from a GP. Individuals with positive results were recalled for re-testing at 6 months after treatment when the testing process was repeated. Notification of partners was undertaken by the patient, the healthcare provider or the health advisor who worked exclusively on the pilot study. Partner treatment and testing took place in general practice, family planning, student health and genitourinary medicine clinics.
Cost analysis
A prospective cost analysis was conducted alongside the pilot study to ascertain the healthcare resource use associated with the screening programme. Consistent with Irish guidelines, a publicly funded health system perspective was adopted with respect to costing.5 Resource items were identified and measured using resource-use record forms, healthcare provider questionnaires, interviews with research staff and directly from the study financial accounts. A resource-use record form completed by providers was used to prospectively record the time input requirements for each patient episode. A provider questionnaire was used to identify setting-specific resource use for the participating practices. Interviews with research staff and a detailed review of the study accounts provided further information on the resources required to implement the programme. Resource items were grouped into two cost categories: overheads and variable costs. Overheads included the costs of project management and coordination by a health advisor, stationary, computer equipment, advertising, printing, photocopying and packaging, charges and travel expenses. An overhead cost per screen was estimated by allocating total overhead costs on the basis of the total number of participants in the pilot study. Variables costs included provider and support staff time input, information leaflets, screening materials and consumables, laboratory materials and testing, antibiotic medications, referrals, and telephone, fax and postage charges. To complete the cost analysis, resource use was valued using unit cost data. Unit costs were obtained from national data sources and were transformed to Euros (€) in 2008 prices using appropriate inflation rate indices.6–8 Details on usage for specific resource items and their respective unit costs are reported in table 1. Cost results were calculated for each stage of the screening process and presented in terms of the average cost per offer, per negative case, per positive case and per partner notified and treated.
Opportunistic screening programme: resource use and unit cost estimates
Decision modelling analysis
The decision modelling framework adopted has been discussed extensively elsewhere.3 9 In brief, the model consists of a dynamic component, which is used to model the impact of screening on chlamydia prevalence, and an economic component, which is used to determine its impact on terms of health outcomes and healthcare costs. As noted, the dynamic component of the model and specifically the assumptions pertaining to sexual behaviour and chlamydia prevalence used reflect those in the UK and were not re-calibrated or validated for the Irish setting (see online appendix table A). The UK-based figures were used in the absence of Irish data in particular for chlamydia prevalence. That values from Britain may provide reasonable estimates for Ireland is supported by the Irish Study of Sexual Health and Relationships,10 which indicated that patterns of sexual behaviour among younger Irish people are very similar to those among their British peers. It is conceded, however, that this is a limitation of our approach. The extensive sensitivity analysis of the dynamic part of the model as well as the observed convergence in behaviours suggest that the range of estimates produced will include those likely to exist while providing useful comparators to estimates produced elsewhere.
In the interests of brevity, we discuss only those aspects of the model salient to their application in the Irish context. The dynamic model estimates the annual number of acute cases of chlamydia infection (asymptomatic and symptomatic) in a hypothetical heterosexual population of 20 000 men and 20 000 women aged 16–45 years and the number of complications resulting from untreated infection for the simulated population. Complications modelled include symptomatic pelvic inflammatory disease, ectopic pregnancy and tubal factor infertility in women, neonatal conjunctivitis and pneumonia, and epididymitis in men (see online appendix figure X and table B). The output from the dynamic model is used in the economic model to estimate the healthcare costs and health outcomes associated with the predicted infections and complications. The modelling framework allows for incremental analysis11 to be undertaken whereby alternative screening strategies are evaluated in terms of their relative impact on the estimated number of infections and complications, the number of people screened and treated and the health outcomes and costs which result. With respect to costs, two components are included in the model: (1) the costs of screening and (2) the costs of complications arising from untreated infections. Two measures of health outcome are included in the model: the number of major outcomes (MOs) averted and quality-adjusted life-years (QALY) gained (see online appendix table C). The model is run for a time horizon of 10 years to observe the impact of screening on longer term complications, with all future costs and health outcomes discounted at an annual rate of 3.5%.12
An incremental analysis was undertaken to compare the opportunistic screening programme to a control of no organised screening, or in other words existing care in Ireland. The screening programme was modelled as an ongoing continuous process incorporating data from range of national and, where necessary, international sources (see table 2). It was assumed for the model that 80% of women and 50% of men attended a healthcare setting in a given year. This is based on data from a national population survey for a representative sample of the target population of 18–29-year-olds in Ireland.13 Of those who attended, it was assumed that 70% were offered a screening test by the resident healthcare provider. Given insufficient national data, this estimate was based on observed effective screening rates from a review of opportunistic screening strategies.14 Furthermore, it was assumed that doctors and nurses equally shared the work of offering screens in general practice. It was assumed that 85% of women and 64% of men accepted the screen offer, based on data collected in the pilot study. An effective partner notification and treatment rate of 20% was adopted. This was based on international evidence,3 given the lack of nationally available data. Finally, we assume for the model that once an individual is screened they will not be offered another screening test for 1 year. While an individual may request a test within the same year that a screening test has been offered, it is unlikely that a provider would be reimbursed or choose to offer repeated screens within the same calendar year to those who had declined a screen or been screened negative. These base-case assumptions resulted in an annual screening coverage rate, that is the overall fraction of the target population who are screened, of 48% for women and 22% for men.
Modelled screening scenario input parameters for base-case and sensitivity analyses
Uncertainty in the model can be explored using sensitivity analyses. A series of one-way sensitivity analyses were undertaken to examine the impact of varying the individual assumptions of the economic model (see table 2). In addition, a probabilistic sensitivity analysis15 was undertaken, whereby input parameters were assigned probability distributions and model results re-estimated multiple times based on simultaneous random draws from each distribution. The probabilistic results were used to construct cost-effectiveness acceptability curves, which present the likelihood of screening being cost effective for a range of potential cost-effectiveness threshold values per QALY gained.16 This technique is particularly useful as no single cost-effectiveness threshold has been identified for health technology assessment in Ireland.17 Notably, however, it was not possible to model the uncertainty surrounding the assumptions of the dynamic model relating to sexual behaviour and chlamydia prevalence.
Results
The results of the cost analysis are presented in table 3. In the base-case analysis, it was assumed that doctors and nurses in general practice equally shared the work of offering screens. This resulted in an average cost of €26 per offer, €66 per negative case, €152 per positive case and €74 per partner notified and treated. Given the cost differential between GP- and nurse-led care, results were also estimated for scenarios in which screening were to be offered solely by GPs and nurses. In a GP-led programme, the average cost per offer was €38, per negative case €91 and per positive case €177. In a nurse-led programme, the equivalent results were €15, €42 and €128. As partner notification continued to be conducted in the usual manner, costs did not vary.
Estimated average costs of screening
The results of the cost-effectiveness analysis are presented in table 4 and indicate that screening led to improved health outcomes but required additional healthcare resources even when programme outlays were set against projected savings from avoided infections and complications. Across the entire modelled population of 20 000 men and 20 000 women aged 16–45 years, the average pre-screening prevalence level projected by the dynamic model is 3%. Post screening implementation, the annual number of screens in the target population of 18–29-year-olds was 1960 for men and 4128 for women. Through its projected reduction in modelled prevalence for all age groups (see online appendix figure Y), screening was estimated to result in 699 MOs averted and 45 QALYs gained at an additional cost of €4 258 868 when compared to control. This translated into an incremental cost per MO averted of €6093 and an incremental cost per QALY gained of €94 717. The probabilistic sensitivity analysis results were used to estimate a cost-effectiveness acceptability curve, which indicated that for threshold values of €45 000 per QALY gained and lower, the probability of the screening being cost effective was <1% (see online appendix figure Z). This indicates that the opportunistic screening programme, as modelled, is unlikely to be considered cost effective by policy makers in Ireland. The results from the one-way sensitivity analyses, also presented in table 3, further confirm this conclusion as the estimated incremental cost-effectiveness ratios (ICERs) remained high in all cases.
Incremental cost-effectiveness results (screening versus no screening)
Discussion
In this study, we provide the first estimates of the costs and cost-effectiveness of opportunistic chlamydia screening in Ireland. The results for the average cost per screen offer, per negative case, per positive case and per partner notified and treated were €26, €66, €152 and €74, respectively. The estimated costs of screening in Ireland are relatively high when compared to equivalent estimates for the UK.18–20 The divergence in costs between the two jurisdictions can be attributed to a range of factors. These include the appreciation of the euro relative to sterling and the impact on costs of the mixed healthcare system that operates in Ireland. That this is a pilot programme and may benefit from economies of scale were it to be adopted nationally should also be remembered.
Using the UK dynamic model, our results indicate that the screening approach modelled here is unlikely to be considered cost effective by policy makers in Ireland, given an ICER of €94 717 per QALY gained. The results from sensitivity analyses support this conclusion. The cost-effectiveness findings can be directly compared to those from the study by Adams et al,3 which applied the same modelling approach to evaluate opportunistic screening strategies in England. The ICER for the equivalent National Health Service screening scenario was €54 000 per QALY gained. The divergence in the reported ICERs again point to differences across the jurisdictions with respect to the costs of screening, healthcare unit costs and the overall organisation and financing of services. Our cost-effectiveness results can also be compared to those from other studies, which, using alternative dynamic modelling frameworks, reported conflicting results for the cost-effectiveness of chlamydia screening strategies in various countries. Andersen et al,21 Gift et al,22 Welte et al23 and deVries et al24 individually found that various forms of opportunistic and proactive screening were cost effective in Denmark, the USA and the Netherlands. Conversely, Roberts et al25 found that proactive register-based screening in the UK was not cost effective. Notwithstanding the differences in the screening strategies evaluated, the healthcare systems involved and the costing methods adopted, the divergence in results across the reported studies is likely in part to reflect key differences in the underlying dynamics and assumptions in the models adopted.4 For example, as in the studies by Roberts et al25 and Adams et al,3 we adopted a probability of 10% for the progression from undiagnosed female chlamydia infection to pelvic inflammatory disease. This is in contrast to the studies by Andersen et al,21 Gift et al,22 Welte et al23 and deVries et al,24 which adopted equivalent probabilities of 25%, 20%, 20% and 15%, respectively. In all studies, this parameter has been shown to be a major influence on cost-effectiveness, and recent evidence appears to support the more conservative approach used here in the modelling of this relationship.26 The impact of adopting a lower probability is to reduce the likelihood of screening being considered cost effective. More generally, this forms part of an ongoing literature, which compares the alternative dynamic model specifications for chlamydia transmission, including the approach adopted here.27
Given these findings, an important question emerges as to the future direction of chlamydia screening policy in Ireland. It is clear that a universal approach to opportunistic screening in clinical settings would be expensive and is unlikely to be considered cost effective. As a result, alternative less costly opportunistic screening strategies, which more effectively target at-risk individuals, may be more likely to be considered cost effective as might programmes that screen simultaneously for a range of sexually transmitted infections thereby spreading overheads more widely. The clinical and cost-effectiveness of these and other such initiatives should be explored. The absence of a unique patient identifier in the Irish setting remains an issue for screening generally in Ireland. The removal of this barrier would expand the policy options available with respect to chlamydia screening in Ireland and is an issue that warrants further consideration. Similarly, in this pilot study, screening was not offered to persons younger than 18 years based on legal advice received by the study research team. This will have implications for the cost-effectiveness of screening as well as equity and it an issue that warrants further consideration.
There are a number of limitations in our study. First, as the analysis was undertaken from the health provider perspective, other resources implications such as costs incurred by patients or wider society were not considered. The approach we adopt though is in line with guidance in both the UK and Ireland in this regard. Second, the use of a dynamic model developed for the UK embeds a series of assumptions the validity of which could be questioned. While with respect to sexual behaviour, evidence indicates a convergence between the UK and Ireland, differences may remain. With respect to prevalence, the model projected levels ranging from 2.5% to 3.5%, with a mean of 3% for the combined modelled population, with higher levels projected for the younger age groups reflecting their higher turnover of partners relative to older age groups. In the absence of Irish data,28 we suggest that the use of the UK estimates is reasonable. Furthermore, on the basis of the extensive sensitivity analyses undertaken, we do not think that any actual differences would impact on our conclusions. Notably, given the design of the model, we were unable to model the uncertainty surrounding these and other dynamic model parameters in sensitivity analyses. Moreover, we were unable to adequately model the process of retesting of positive cases. Therefore, the model does not fully take account of those who were re-infected and re-treated within 6 months of initially being tested positive and treated. However, we do not think that this assumption impacts significantly on our conclusions. Finally, only those deemed unable to pay are eligible for free primary care in Ireland. In the analysis, we used an attendance rate that combined those who are eligible for free primary care and those who are not.13 We do not differentiate in results between those who must pay from those who have free access and are, therefore, more likely to attend a healthcare setting.29 Our analysis does not account for heterogeneity between these groups among whom attendance rates, disease prevalence and cost-effectiveness might vary. The issue of heterogeneity was though beyond the scope of our study.
Third, in a study comparing three alternative modelling frameworks, including the approach adopted here, Kretzschmar et al4 argue that that while there is no definitive evidence as to superiority across the alternatives, the model adopted here is the most optimistic in terms of the projected impact of screening on prevalence. While we would expect the cost of identifying a positive case in the years post-screening implementation to be reduced if a more pessimistic model was adopted (given that there would be more residual cases in the target population), we would not expect the resulting impact in terms of overall cost-effectiveness to be considerable. This conclusion is based on the results from the sensitivity analyses, which explored screening scenarios with lower coverage in the target population (based on lower offer and acceptance rates) than the base-case strategy. As would be the case with a more pessimistic model, the impact of screening in reducing prevalence is less for the lower coverage scenarios than the base-case strategy, resulting in more residual cases in the target population, which can be detected at a lower cost per case. Consequently, we believe our results to be robust to the choice of a more pessimistic model than that which was employed.
Finally, cost and cost-effectiveness analysis in Ireland is complicated by a paucity of resource utilisation, unit costs and utility data. In some cases, we adopted the UK resource data to detail the treatment process of chlamydia complications, as national data were not readily available. The assumption that the management of infection matches that in the UK was deemed acceptable by expert opinion from within our study group that included public health consultants as well as those involved in the treatment of chlamydia. Similarly, utility data were estimated based on external data sources. Indeed, utility data associated with chlamydia infection and complications are not widely available, and further research is required to improve the QALY estimates adopted in studies such as ours.
In conclusion, we examined the costs and cost-effectiveness of an opportunistic screening programme in Ireland and found that it would be expensive to implement nationally and is unlikely to be judged cost effective by policy makers. These results are relevant to those charged with the development of chlamydia screening policy in Ireland and elsewhere.
Key messages
The absence of unique patient identifiers in the Irish healthcare system necessitates opportunistic approaches to screening for chlamydia.
The cost of screening in Ireland was €26 per offer, €66 per negative case, €152 per positive case and €74 per partner notified and treated.
A nationalised opportunistic screening programme, as modelled using a UK decision modelling framework, is unlikely to be judged cost effective by policy makers in Ireland.
Acknowledgments
We wish to acknowledge the Medical Scientific staff of the Department of Virology at University College Hospital Galway for performance of diagnostic testing. We would also like to acknowledge the cooperation and support of the clinicians and individuals who participated in the study.
References
Supplementary materials
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:
- Download Supplementary Data (PDF) - Manuscript file of format pdf
Footnotes
Funding The research was funded by the Health Protection Surveillance Centre and Health Research Board in Ireland. The study sponsors had no part in the design of the study; the collection, analysis and interpretation of the data; the writing of the report and the decision to submit the article for publication.
Competing interests None.
Ethics approval The pilot study was approved by the ethics committee of the National University of Ireland Galway and by the ethics committee of the Irish College of General Practitioners.
Provenance and peer review Not commissioned; externally peer reviewed.