Despite the introduction of HPV 16/18 vaccination in the Netherlands and elsewhere, cervical screening will remain the most important cervical cancer prevention method for the majority of women in the coming decades. Modern alternatives (i.e. the HPV test and LBC) to conventional cytology may improve the efficiency of cervical cancer screening. This article has presented a large microsimulation study to evaluate the cost-effectiveness of the available cervical cancer screening strategies: we focused on the cost-effectiveness of adopting the HPV test as the primary test. Our analysis should help determine the optimal screening programme in the Netherlands and in other countries as well.
We found that primary HPV screening with cytology triage is the most cost-effective screening strategy under the base-case assumptions. In the base-case analysis, the optimal programme consists of HPV screening with cytology triage with three (at a €20 000 per QALY cost-effectiveness threshold) or seven (at a €50 000 per QALY cost-effectiveness threshold) screening rounds in a woman’s lifetime. However, an additional analysis showed that primary HPV screening is not cost-effective for young women: a combination of primary cytology screening for women younger than 33 years of age and HPV screening for older women can be more cost-effective than a programme with HPV screening for women of all ages. This result arises from the relatively low positive predictive value of HPV testing for progressive cervical lesions among younger women. All cost-effective screening programmes in this analysis used conventional cytology instead of LBC: the advantages of LBC (e.g. the possibility to perform an HPV test on the same material) did not outweigh the higher evaluation costs.
Sensitivity analyses have shown that several model inputs can affect what the optimal screening programme is. The optimal number of screening rounds is sensitive to the background cervical cancer risk in women attending screening, to the discount rates used for costs and health effects and to the cost-effectiveness threshold. An increase in the laboratory costs of the HPV test (from €34 to €42) or an increase in the utility loss associated with time spent in triage (from 2.2 to 6 days of life per year spent in triage) would make primary cytology more cost-effective than primary HPV testing. The latter result arises from the fact that strategy I (primary cytology screening with immediate HPV triage) does not require waiting for triage tests, whereas the strategies with primary HPV testing usually require 6 or 18 months of follow-up after a positive primary test. Assuming a somewhat higher sensitivity of LBC compared with conventional cytology, as found in previous studies,47,48
could make LBC more cost-effective than conventional cytology; however, the HPV test would still be the most cost-effective primary test in that case.
Although HPV testing has been possible for many years, only recent evidence from randomised controlled trials in the Netherlands, the UK, Italy and Sweden has shown that HPV testing at baseline leads to a lower detection rate of cervical lesions at the next screening round than cytology-based screening.4–7
Because of the higher sensitivity of the HPV test for CIN and cervical cancer, using this test as the primary test can yield more health effects than a cytology-based programme, even with fewer screening rounds. Nevertheless, the HPV test has some disadvantages. Because of the lower specificity of the HPV test, many women who do not have CIN or cancer attend triage tests after a positive HPV test. The uncertainty caused by waiting for a diagnosis after a positive test could cause anxiety. In addition, the HPV test may not be cost-effective if the ratio of the prevalence of HPV infections to the prevalence of CIN is high. This may be the case in some countries, e.g. in Italy,50,51
and in young women.52
Finally, the HPV test is currently more expensive than cytology, although this may change in the future.
The MISCAN model and the assumptions for costs and utilities were based on the Dutch situation. For other countries, the optimal screening intensity in particular can differ, based on the background risk of cervical cancer and other factors: for example, countries with a higher background level of cervical cancer incidence, e.g. Denmark,53
may require more screening rounds than the Netherlands. Nevertheless, some results presented here can be generalised to other countries. The optimal primary test and triage schedule will probably be the same as in our analysis for countries that have similar treatment and testing costs, and a similar positive predictive value of HPV testing for cancer and for CIN that would eventually become cancer. The latter condition requires that the ratio of the prevalence of HPV infection to the incidence of cervical cancer is approximately the same as in the Netherlands: this condition appears to be met in most countries.54
Also, the results of sensitivity analyses showed that, under varying circumstances, only strategies C, D and E (all primary HPV screening) or strategy I (primary cytology) were among the most cost-effective ones. In most of these strategies, women are only referred for colposcopy after a positive HPV test and a positive cytology result in the same screening round, or a cytological result of at least HSIL. These screening strategies may provide an optimal combination of the high sensitivity of the HPV test and the high specificity of cytology.
Several cost-effectiveness analyses of cervical cancer screening have been published previously.18–32
The analyses show a large variation in terms of what programmes are evaluated, the model assumptions and the methodological approach (e.g. the type of model). The results of our analysis suggest fewer screening rounds than in previous analyses. Reasons for this are that the background cervical cancer incidence in the Netherlands is lower than in most other countries, and that we assumed that women attending screening have a lower background risk than non-attenders. Most of the previous cost-effectiveness analyses support introducing the HPV test as a primary test. For example, a cost-effectiveness analysis of HPV testing in the Netherlands found that primary HPV screening can be more cost-effective than the current screening programme in the Netherlands.21
In that study, it was also suggested to increase the interval between screening rounds if the HPV test is used as the primary test. In one analysis that did not support introducing the HPV test, women were directly referred for colposcopy after a positive primary HPV test.25
Our analysis shows the importance of specifying an effective triage schedule for women with a positive primary test, as referring all women with a positive primary HPV test will result in a large number of unnecessary colposcopies. Another cost-effectiveness analysis suggested a programme of primary cytology with HPV triage (strategy I), with an interval of 3 years for the Netherlands;18
however, using only the HPV test as a primary test was not considered in that analysis. The main contribution of the present study is that it is more comprehensive than previous cost-effectiveness analyses of cervical cancer, with a much larger number of evaluated screening programmes. To provide a fair comparison between cytology and the HPV test, screening programmes with different numbers of screening rounds and triage schedules must be considered. We have also shown the effects of different assumptions for costs and utilities, different discount rates, and assuming a loss to follow-up.
Like any study, this study has some limitations. First, the cost-effectiveness results are sensitive to several model inputs. Some of these model inputs (e.g. the natural history of the disease) are not known exactly, whereas other inputs can change over time (e.g. the relative costs of cytology and HPV testing). Another limitation is that we did not take HPV 16/18 vaccination into account. Although the present analysis is exclusively intended for the unvaccinated cohorts, the vaccination of younger women may yield a herd immunity effect that can lower the incidence of cervical cancer in unvaccinated women. This effect could further reduce the optimal intensity of screening. However, we believe that any such effect is likely to be small, as there is a large age difference between the vaccinated women and the unvaccinated women considered here. Future research should evaluate screening in HPV-vaccinated women.
In summary, this paper has presented a comprehensive cost-effectiveness analysis of adopting the HPV test as the primary cervical screening test in the Netherlands. By comparing a variety of screening strategies and policies, we have shown how to optimally combine the type of screening test, the triage schedule, the screening age range and the intensity of screening. Our results support introducing the HPV test as the primary test in the Netherlands and increasing the interval between screening rounds. We have also shown the cost and utility conditions for which the HPV test is most cost-effective as a primary test.