Article Text
PDF
Abstract
Objective To assess the cost-effectiveness of three testing strategies with or without light microscopic Gram-stained smear (GSS) evaluation for the detection of anogenital gonorrhoea among men who have sex with men (MSM) at the Amsterdam STI clinic using a healthcare payer perspective.
Methods Three testing strategies for MSM were compared: (1) GSS in symptomatic MSM only (currently practised strategy), (2) no GSS and (3) GSS in symptomatic and asymptomatic MSM. The three testing protocols include testing with nucleic acid amplification test to verify the GSS results in (1) and (3), or as the only test in (2). A transmission model was employed to calculate the influence of the testing strategies on the prevalence of anogenital gonorrhoea over 10 years. An economic model combined cost data on medical consultations, tests and treatment and utility data to estimate the number of epididymitis cases and quality-adjusted life years (QALY) associated with gonorrhoea. Incremental cost-effectiveness ratios (ICERs) for the testing scenarios were estimated. Uncertainty and sensitivity analyses were performed.
Results No GSS testing compared with GSS in symptomatic MSM only (current strategy) resulted in nine extra epididymitis cases (95% uncertainty interval (UI): 2–22), 72 QALYs lost (95% UI: 59–187) and €7300 additional costs (95% UI: −€185 000 (i.e.cost-saving) to €407 000) over 10 years. GSS testing in both symptomatic and asymptomatic MSM compared with GSS in symptomatic MSM only resulted in one prevented epididymitis case (95% UI: 0–2), 1.1 QALY gained (95% UI: 0.1–3.3), €148 000 additional costs (95% UI: €86 000 to–€217 000) and an ICER of €177 000 (95% UI: €67 000–to €705 000) per QALY gained over 10 years. The results were robust in sensitivity analyses.
Conclusions GSS for symptomatic MSM only is cost-effective compared with no GSS for MSM and with GSS for both symptomatic and asymptomatic MSM.
- homosexuality
- bacterial infection
- neisseria gonorrhoea
- economic analysis
- testing
- Sexually Transmitted Diseases
- Male
- Quality-Adjusted Life Years
- Point-of-care testing
Statistics from Altmetric.com
- homosexuality
- bacterial infection
- neisseria gonorrhoea
- economic analysis
- testing
- Sexually Transmitted Diseases
- Male
- Quality-Adjusted Life Years
- Point-of-care testing
Introduction
Gonorrhoea is one of the most frequent STIs in Europe.1 2 Men who have sex with men (MSM) are the highest risk groupfor gonorrhoea infection. In the Netherlands, MSM accounted for 74.8% of the 6092 detected gonorrhoea infections at STI clinics in 2016.3
Neisseria gonorrhoeae can be detected in genital, rectal or pharyngeal secretions by nucleic acid amplification tests (NAAT) or culture.2 4–6 These tests have high sensitivity and specificity, but the results are available only days after testing.2 This potentially leads to ongoing transmission due to treatment delay or loss to follow-up.2 4 5 7–10 A presumptive gonorrhoea diagnosis can also be made based on light microscopic detection of the bacterium in Gram-stained smears (GSS). This enables immediate treatment, thus preventing ongoing transmission and/or loss to follow-up.4 5 11 12 GSS evaluation of urethral samples has high sensitivity and specificity in symptomatic men, but has low sensitivity in asymptomatic men (≤55%) and in men with rectal infections (≤40%).2 13 This can lead to missed infections and undertreatment. Therefore it is recommended to confirm testing outcomes based on GSS by NAAT and/or culture.
At the STI clinic in Amsterdam, GSS is used as a presumptive test and confirmed by NAAT. Earlier research at the STI clinic in Amsterdam has shown that GSS testing can result in considerable reductions in costs and in gonorrhoea prevalence.4 5 Furthermore, patients lost to follow-up with an untreated gonorrhoea infection might develop complications like epididymitis requiring medical treatment, resulting in lower health-related quality of life (HrQol) and higher costs.
The aim of this study is to evaluate the cost utility of three GSS testing strategies for the detection of anogenital gonorrhoea among MSM, taking potential sequelae into account and using a healthcare payer perspective.
Methods
Standard diagnostic strategy for gonorrhoea infection
As ‘standard strategy’ for gonorrhoea infections at the STI clinic, patients with suspected symptoms for anogenital gonorrhoea are tested with GSS at the initial visit and samples are obtained for NAAT. Opposite to older diagnostic procedures where epithelial cells were needed, the used GSS test works on a urine sample with similar sensitivity. If the GSS test is positive, a presumptive gonorrhoea treatment is given. Samples for culture are collected either at the initial visit, if GSS is positive or when treated. Culture tests are carried out to assess antibiotic resistance. The NAAT results are communicated via online services. Individuals with positive NAAT who have not received presumptive treatment are asked to return to the clinic for treatment; this is on average 1 week after the initial visit.4 If they do not return, they are again contacted by a nurse for a maximum of three attempts. Thereafter, the patient is considered as ‘lost to follow-up’ and remains untreated and infectious until natural recovery.
Testing strategies
The following three testing strategies for MSM were analysed:
GSS for symptomatic MSM (reference): the current strategy at the STI clinic in Amsterdam. Microscopic examination of GSS is carried out only for symptomatic MSM in addition to NAAT. A positive GSS result is followed by presumptive therapy awaiting the definite NAAT result. Asymptomatic MSM are tested with NAAT only.
No GSS: only NAAT is performed for all MSM. This is a hypothetical budget-saving testing strategy.
GSS for all: GSS and NAAT are performed in all MSM, both with and without symptoms. This is a hypothetical health-gaining strategy and was the standard strategy at the STI clinic in Amsterdam until 2010.
Transmission model
A transmission model was used to assess the impact of the three testing strategies on gonorrhoea transmission among MSM.5 The model has been earlier described in detail.5 Briefly, MSM are distinguished in the model according to whether they are infected or not. Infected MSM were classified into five states of gonorrhoea infection: untested symptomatic MSM; untested asymptomatic MSM; tested and untreated symptomatic MSM; tested and untreated asymptomatic MSM; and lost to follow-up. In the transmission model, a range was defined for the expected positivity rate among those tested (5.9%–7.3%) based on the anogenital gonorrhoea positivity rate of 6.6% among MSM at the STI clinic in Amsterdam.5 Uncertainty analysis was carried out, by defining ranges of values for the uncertain parameters (online supplementary table S1). The model was calibrated to the positivity rate of the STI clinic by selecting the parameter combinations resulting in the expected positivity rate. From the uncertainty analysis, 145 parameter sets were selected (see online supplementary section 1). The range of the 145 results obtained from the transmission model (expressing uncertainty around the transmission model parameters) was used as input for the economic model and is summarised in table 1.
Supplemental material
Total number of tests, cases, effects and costs (in €1000) with three testing strategies for MSM suspected of anogenital gonorrhoea over 10 years, with average and 95% uncertainty intervals* in parentheses
Economic model
An economic model was developed in MS Excel using the add-in software program @Risk V.7.5 (Palisade, Newfield, New Jersey, USA) to assess the costs and effects of the three testing strategies. A flow chart of the economic model is shown in online supplementary figures S1–S3. The simulated annual numbers of 17 health states (online supplementary section 2) for a 10-year period served as input for a disease progression tree, specifying disease consequences of an untreated anogenital gonorrhoea infection (online supplementary figure S3). Probabilities were derived from the literature.2 14 15 Disseminated gonococcal infection (DGI) was only considered in scenario analysis, as it is seldom seen in daily practice. All complications were counted in the year of infections. The simulated annual numbers of epididymitis cases were summed up over 10 years to obtain the total number of epididymitis cases per testing strategy (an outcome in the current study). The impact of disease on HrQol was expressed in quality-adjusted life years (QALYs), a metric unit with a scale from 0 (death) to 1 (perfect health). We retrieved the annual number of the different health states, including complications, and computed the number of QALYs lost due to infections. By multiplying the difference between the health utility index of a healthy person and that of a particular health state (online supplementary table S2) with the duration of that health state (online supplementary table S3), we obtained the QALY loss per health state.14 Total QALY losses were obtained by summing up overall health states over 10-year per testing strategy. Healthcare costs consisted of costs for medical consultations, tests and treatment. Cost data (resources used and unit prices) of medical consultations at STI clinics, tests and medication administrated at the STI clinic were derived from the STI clinic in Amsterdam.5 Costs for medication administered by general practitioners (GPs) or at hospitals were based on prices of the Dutch Healthcare Institute.16 GP visits, outpatient and inpatient visits were valued using Dutch reference prices.17 Used cost prices, resources and assumptions made were summarised in online supplementary table S5 and figure S3. Total healthcare costs were obtained by multiplying the healthcare costs per health state with the annual number of cases in that particular health state, and summing up overall health states over 10 years per testing strategy. All costs were expressed for the year 2016, using the Dutch Consumer Price Index,18 where necessary. For each of the 145 simulations from the transmission model, 250 iterations were run, using Monte Carlo simulation technique. The 250 iterations reflect uncertainty around parameters relating to QALYs and costs. The results were presented as mean and uncertainty interval, which is the range between the 2.5th and 97.5th percentile of the simulated results.
Cost-effectiveness estimates
The analysis was based on data from the STI clinic in Amsterdam and was conducted from the healthcare payer perspective for a 10-year time horizon. The current testing strategy (GSS for symptomatic MSM) was compared with two alternatives: GSS for all MSM (expanding GSS) and no GSS for MSM (abandoning GSS). The incremental costs were calculated by subtracting the net costs of the alternative (abandoning GSS or expanding GSS) from the net costs of the current strategy. The incremental effects were calculated similarly. The incremental cost-effectiveness ratio (ICER) was calculated by dividing the incremental costs by the incremental health effects:
We calculated the ICER for two effects: QALY gained and prevented epididymitis cases. Costs and QALYs were discounted by 3%.19 We used the WHO threshold of one and three times the gross domestic product (GDP) per capita19 (ie, €41 258 and €123 774)20 to determine if an alternative is cost-effective or not.
Sensitivity and scenario analyses
To assess the robustness of the results from the economic model, sensitivity and scenario analyses were conducted (online supplementary table S6). We examined the impact of variations in the following parameters: discount rates for costs and effects; different time horizons; the sensitivity of GSS in anogenital gonorrhoea infections; the percentage of untreated gonorrhoea infections developing epididymitis; the percentage of untreated gonorrhoea infections developing DGI; the percentage of GP visits for epididymitis cases; the percentage of DGI inpatient visits; lower cost prices of NAAT and culture; and the maximum duration of the interval with complaints due to epididymitis after hospitalisation. To determine the most influential parameters from the transmission model, the estimated ICER values were plotted against the parameters of the transmission model.
Results
With the current testing strategy, there were 6111 new gonorrhoea infections, 73 616 MSM tested for gonorrhoea and 13 MSM developed epididymitis over 10 years. Of those tested, 72 MSM (with gonorrhoea infection) were lost to follow-up and 63 MSM (without gonorrhoea infection) were overtreated. These resulted in 18 QALYs lost and €2.5 million healthcare costs. The majority of the costs (97%) were due to testing (table 1).
Abandoning GSS (no GSS instead of GSS for symptomatics) resulted, on average, in 1899 additional tests, 2418 additional treated gonorrhoea infections, 63 averted overtreated MSM, 478 additional MSM lost to follow-up, 9 additional epididymitis cases, 72 additional QALYs lost and €7000 additional healthcare costs (table 1). Abandoning GSS resulted always in less MSM being lost to follow-up (on average 30) and in health losses (negative incremental effects, expressed as QALY loss); in 43% of the model iterations, no GSS testing was more costly than the current strategy (see figure 1), while in 57% of the iterations no GSS testing resulted in negative incremental costs or cost-savings.
Cost-effectiveness planes of changing testing strategies for gonorrhoea infection among MSM in Amsterdam, assuming a 10-year horizon and taking a healthcare payer perspective. Note 1: The vertical axis shows the incremental costs and the horizontal axis shows the incremental effects in QALYs of 36 250 iterations. Incremental costs and effects are for (A) abandoning GSS, that is, no GSS for MSM compared with the current testing strategy (blue points), and (B) expanding GSS, that is, GSS for all MSM compared with the current testing strategy (red points). The large squares indicate the mean, the blue square when abandoning GSS and the red square when expanding GSS. The solid line indicates the willingness-to-pay threshold of three times the GDP per capita (€123 774); the dashed line indicates a threshold of 1 GDP per capita (€41 258),20 the WHO thresholds for cost-effectiveness.19 The proportions of points in each quadrant indicate the proportion of simulation results in that quadrant and is indicated in grey at each corner in the two cost-effectiveness planes. Point estimates in the NW quadrant are in favour of the current strategy; point estimates in the SE quadrant (there was none) would be in favour of the alternatives. Note 2: GSS for all MSM versus GSS for symptomatic only resulted in an ICER lower than €41 258/QALY gained (1× GDP per capita) in 3.4% of all simulations and an ICER lower than €123 774/QALY gained (3× GDP per capita) in 39.3% of all simulations. GDP, gross domestic product; GSS, light microscopic examination of Gram-stained smears from anogenital locations; ICER, incremental cost-effectiveness ratio; MSM, men who have sex with men; NE, north-east quadrant; NW, north-west quadrant; QALY, quality-adjusted life years; SE, south-east quadrant; SW, south-west quadrant.
Expanding GSS to all MSM (GSS for all instead of GSS for symptomatics) resulted in 195 less gonorrhoea tests, 341 less gonorrhoea infections and 257 less treated gonorrhoea infections, but in 80 additional overtreated MSM cases, over 10 years. Compared with the current testing strategy, GSS for all MSM resulted in 0.8 prevented epididymitis case, 1.1 QALY gained and €148 000 additional healthcare costs (table 1). More than 99% of the simulated results were in the north-east quadrant of figure 1 (GSS for all MSM resulted in additional costs and additional health gains), and 0.03% of the simulation results were in the north-west quadrant (the alternative strategy is less effective and more costly than the current strategy). In 3.4% and 39.3% of all simulations, the ICER was lower than €41 258/QALY gained (1× GDP per capita) and lower than €123 774/QALY gained (3× GDP per capita), respectively.
Figure 2 and online supplementary tables S7-S8 show the results of the sensitivity and scenario analyses for the economic model. Abandoning GSS was in all scenarios less effective than the current strategy. Furthermore, the average ICER was most sensitive to the probability of developing DGI (higher incremental costs and slightly higher QALY loss); the time horizon (the longer the time horizon, the worse the health outcomes); and assumed cost prices of NAAT and culture (lower cost prices resulted in lower incremental costs and a more favourable ICER) (see figure 2). Expanding GSS (GSS for all instead of GSS for symptomatics) resulted in positive health effects (ie, QALYs gained), but at higher costs in all scenarios. The highest variations in the average ICER (of €135 000/QALY gained in the baseline scenario) were observed with higher sensitivity of the GSS (average ICER, €61 000/QALY gained) and shorter time horizon of 5 years (average ICER, €158 000/QALY gained).
Tornado graph depicting the changes in ICER for various sensitivity analyses and scenario analyses for (A) no GSS testing for MSM versus GSS for symptomatic (abandoning GSS) and (B) GSS testing for all MSM versus GSS for symptomatic (expanding GSS). Note 1: The magnitude of the bar relative to the vertical line represents the magnitude of the effect of varying a given parameter on the outcome. The vertical axis shows the varied model parameters. The horizontal axis shows the ICER in euros per QALY gained. The bars indicate the variation in the ICER caused by varying the value of the indicated parameter holding all other parameters constant. All costs are in 2016 euros. Note 2: Full details are presented in online supplementary tables S6 and S7. Note 3: No GSS testing for MSM versus GSS for symptomatic (abandoning GSS) resulted for most scenarios, on average, in a negative value for the ICER. This negative ICER value is due to a negative incremental effect (QALY lost) and additional costs, and hence would be in the north-west quadrant when presented in a cost-effectiveness plane (for illustration, see online supplementary figure S4). Only lower cost prices for tests or a shorter time horizon (5 years) resulted in negative incremental costs (cost-saving) as well as a negative health effect (QALY lost), and hence a positive ICER value and hence would be in the south-west quadrant when presented in a cost-effectiveness plane (online supplementary figure S4). For full details see online supplementary table S7. DGI, disseminated gonococcal infection; GP, general practitioner; GSS, light microscopic examination of Gram-stained smear from anogenital locations; ICER, incremental cost-effectiveness ratio; MSM, men who have sex with men; QALY, quality-adjusted live years.
Among the parameters of the transmission model, the most influential for the value of the ICER were the percentage of gonorrhoea infections with symptoms and the frequency of acts of unprotected anal intercourse with casual partners (figure 3 and online supplementary figure S5).
The ICER for abandoning GSS (no GSS vs GSS for symptomatics) in relation to (A) the percentage of gonorrhoea infections with symptoms and (B) the frequency of acts of unprotected anal intercourse with casual partners. Note 1: The ICER (vertical axes) is plotted against epidemiological parameters from the transmission model: (A) the percentage of gonorrhoea infections with symptoms and (B) the frequency of acts of unprotected anal intercourse with casual partners. The ICER is presented as incremental costs in 2016 euros per quality-adjusted life years (QALY) gained. Costs and QALYs are cumulative over 10 years. GSS, light microscopic examination of Gram-stained smear from anogenital locations; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life years.
Discussion
Our findings indicate that abandoning GSS from the current test strategy is less effective and would result in negative health effects, both in terms of QALYs and in terms of epididymitis cases. The impact on healthcare costs is more uncertain, ranging from cost-saving (57% of simulation results) to additional costs when compared with the current strategy. Additional gonorrhoea infections as an effect of ongoing transmission resulted in higher costs. This had a negative impact on the potential savings associated with lower testing costs per tested MSM. The considerable health losses when abandoning GSS can be explained by three facts: immediate treatment with GSS resulted in shorter duration of symptoms; reduced number of MSM lost to follow-up; and less gonorrhoea complications. Moreover, immediate treatment and fewer cases of loss to follow-up prevented ongoing transmission, resulting in fewer infections.
Expanding GSS to all MSM, as opposed to GSS for symptomatic MSM only, would always result in a positive health effect (eg, QALYs gained)—although marginal—but at considerable costs. The obtained ICER of €135 000/QALY gained was higher than the recommended willingness-to-pay threshold recommended by the WHO of three times the GDP,19 which in 2016 would have been €123 774/QALY gained20; also it was much higher than the Dutch threshold of €20 000/QALY gained. The choice made by the STI clinic in Amsterdam to stop GSS in asymptomatic MSM back in 2010 was therefore supported by the current study as a cost-effective-wise decision.
As far as we know, this is the first economic evaluation of different testing strategies for the detection of anogenital gonorrhoea in MSM considering complications, measuring health effects in QALYs and evaluating the ICER between test strategies. The complications of gonorrhoea infection in men are much less severe than in women; therefore, little attention has been focused on potential complications in men. Turner and colleagues examined the cost-effectiveness of point-of-care (POC) tests for male and female visitors of genitourinary medicine clinics in England. Our findings are in agreement with those of Turner et al. 11 They reported that POC testing is £11.7 million cheaper and results in an increase of 46 QALYs compared with standard care (off-site laboratories) only. In our study, the GSS testing strategies resulted always in higher HrQol than NAAT testing only (no GSS testing strategy). However, Turner et al used NAAT as POC test (with a more favourable sensitivity and specificity) and a time horizon of 28 days, while we used GSS as POC and a time horizon of 10 years. This indicates that accelerated diagnosis and treatment can prevent ongoing transmission of gonorrhoea infections. Another strength is the use of bottom-up cost price calculations of the STI clinic in Amsterdam of true testing and treatment costs of gonorrhoea infections and own bottom-up cost price calculations for the complications. Finally, in addition to point estimates, stochastic estimating was used to model the uncertainty around the output of the transmission model (145 simulations) and the costs and QALYs (250 iterations per simulation).
Our study also had some limitations. First, only epidemiological data of the STI clinic in Amsterdam were used; therefore, the results may not be directly applicable to other populations and/or healthcare settings (eg, primary care). Furthermore, this study was conducted from the healthcare payer perspective, so only direct healthcare costs were included; patient costs and productivity losses were not included. If we had taken a societal perspective, the increased numbers of tests and treatments when abandoning GSS would have implicitly imposed higher societal costs for the no GSS strategy, resulting in even higher number of simulations whereby no GSS was dominated by the current strategy. Extending GSS could result in little additional savings due to the reduced number of tests and the reduced number of treatments when compared with the current situation. However, these savings could be over-ruled by costs due to work absence—costs not considered in a healthcare payer perspective—resulting in even less favourable ICERs. Also, partner notification, counselling and treatment were not taken into account in the model; therefore, costs and effects due to partner notification were not considered here. The number of MSM lost to follow-up might have been overestimated, and consequently also the ongoing transmission, as MSM might get treated for example by their GP rather than returning to the STI clinic. The benefits of immediate treatment due to GSS might have been overestimated, as treatment might be initiated immediately due to the symptoms presented without waiting for any kind of test results. This analysis is based on the assumption that the current facilities of the clinic do not need expansion (additional laboratories) in order to perform the extra GSS tests. Also, it assumes that GSS testing does not interfere with other work activities of the clinic, such that testing and treatment of other STIs will not be delayed due to the overwork of the clinic personnel carrying out GSS testing. Finally, this study focused only on gonorrhoea infections. We did not account for potential costs and health effects of other STIs that are tested and treated at the same test moment, such as HIV and chlamydia. Including the impact of/on other STIs is quite complicated. For instance, with decreasing gonorrhoea prevalence, fewer STI tests might be carried out, with as consequence an increased number of undetected other STIs and subsequently increases in STIs. On the other hand, lower gonorrhoea prevalence might result in reduced HIV transmission,21–23 whereas with an increased gonorrhoea prevalence we might expect the number of STI tests to rise, as will the detection of other STIs and the transmission of HIV.
In conclusion, not testing MSM with GSS would not be advisable, since it could result in considerable health losses. On the other hand, extending GSS testing to all MSM would result in marginal health gains with substantial additional costs and an ICER far above any accepted cost-effectiveness threshold. In an era of increasing STIs, offering GSS only to symptomatic MSM seems a wise option to find a good balance between health gains and costs.
Key messages
This is a cost-effectiveness analysis of three anogenital gonorrhoea nucleic acid amplification test-based strategies with or without light microscopic Gram-stained smear evaluation in men who have sex with men (MSM).
Microscopy using Gram staining for identification of diplococci in symptomatic MSM (the current strategy) results in considerable health gains and is a cost-effective testing strategy.
The extension of microscopy of Gram-stained smears to all MSM, irrespective of symptoms, results in marginal health gains and unfavourable incremental cost-effectiveness ratios.
In contrast, the abandonment of Gram-stained smear evaluation in MSM results in health-related quality-of-life losses and unfavourable cost-effectiveness ratios compared with the reference strategy.
References
Footnotes
JMZ and M-JJM contributed equally.
Handling editor Jackie A Cassell
Contributors M-JJM, MX and JMZ were responsible for the design of the study. MX designed the mathematical model. M-JJM and JMZ designed the economic model. MSvR and MB collected and interpreted the sexual behaviour data. JMZ, M-JJM and MX drafted the paper. M-JJM and MX supervised the overall study. All authors reviewed and approved the final article.
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 None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All relevant data are within the paper. Further details are available from the corresponding author on reasonable request.