Human papillomavirus vaccination in the UKBMJ 2008; 337 doi: https://doi.org/10.1136/bmj.a842 (Published 17 July 2008) Cite this as: BMJ 2008;337:a842
- Jane J Kim, assistant professor of health decision science
- 1Program in Health Decision Science, Department of Health Policy and Management, Harvard School of Public Health, Boston, MA 02115, USA
This September, the Department of Health in the United Kingdom will begin a national programme of routine human papillomavirus immunisation of 12-13 year old schoolgirls, coupled with a two year catch-up campaign for those up to the age of 18 in 2009.1 The selected bivalent vaccine, Cervarix, protects against two of the most common human papillomavirus types that cause cervical cancer (types 16 and 18), whereas an available quadrivalent vaccine, Gardasil, also protects against two non-oncogenic types that cause genital warts (types 6 and 11). The linked study by Jit and colleagues from the Health Protection Agency (doi: 10.1136/bmj.a769) describes the mathematical modelling approach and results used to inform the Department of Health’s decision.2
Mathematical models are used to synthesise multiple data sources, to extrapolate short term clinical findings into long term outcomes of population level benefits and cost effectiveness, and to investigate the influence of uncertainties about data and alternative scenarios. The authors develop a dynamic model that reflects the sexual transmission of human papillomavirus infections (types 6, 11, 16, 18, and other high risk types). The model captures the direct benefits to girls who receive the vaccine and the indirect benefits to those who are not vaccinated, as a result of the reduced prevalence of human papillomavirus in the population—so called herd immunity. Unlike most other model based studies of human papillomavirus and cervical cancer, the authors analyse thousands of scenarios in which epidemiological and economic dimensions are varied simultaneously, which allows them to evaluate the uncertainty more comprehensively.
The model projects the effect of human papillomavirus vaccination on cervical dysplasia, squamous cell carcinoma and adenocarcinoma, as well as genital warts in men and women. Because of the uncertainty about the efficacy of the vaccine for other health conditions, the effect on non-cervical cancers associated with human papillomavirus is estimated separately and included in a secondary analysis.
Assuming a willingness to pay threshold of £30 000 (€37 000; $59 600) per quality adjusted life year (QALY) gained, Jit and colleagues find that vaccinating 12 year old girls against human papillomavirus is cost effective in the context of current screening practice in the UK, when vaccine uptake is high (≥80%) and as long as protection lasts longer than 10 years. This result is consistent with most other model based analyses under similar assumptions of vaccine efficacy, longevity, and coverage in settings with organised screening programmes.3 Not surprisingly, the probability of this strategy being cost effective increases under optimistic scenarios of vaccine duration, efficacy against non-cervical cancers, and cross protective effects.
Their results also indicate that a two year catch-up campaign for females up to age 18 would be cost effective. Because sexual activity—and therefore risk of previous exposure to vaccine targeted human papillomavirus infections—increases with age, studies have found that the health benefits of vaccination past the age of 18 are marginal; however, the optimal upper age limit for a catch-up programme has varied from 18 to 25.4 5 6 Jit and colleagues also conclude that including 12 year old boys in the immunisation programme is unlikely to be cost effective when compared with vaccinating girls only. Conflicting results of including boys have been reported previously, with cost effectiveness ratios ranging from $45 100 to $442 000 per QALY gained.5 7 Because data on vaccine efficacy in boys are not yet available, all studies have relied on assumptions that will need to be revisited when empirical data become available from trials. Model attributes and assumptions that may contribute to variations in results across different studies are discussed in several reviews.3 8 9
One of the study’s unique contributions is a cost threshold analysis that compares the bivalent and quadrivalent vaccines. Because the quadrivalent vaccine protects against genital warts caused by human papillomavirus types 6 and 11, to be equally cost effective the bivalent vaccine must be less expensive—the authors estimate that the bivalent vaccine must be £13 to £21 less expensive per dose than the current price of the quadrivalent vaccine. Assuming 80% coverage of current 12 year old girls in the UK with the full three dose vaccine series, this price differential translates to savings of £11.5m to £18.6m from the vaccine price alone in the first year of the programme, compared with adopting the quadrivalent vaccine. The decision to select the bivalent vaccine implies that the Department of Health is willing to accept foregone health benefits (and additional cost savings) from averting cases of genital warts for the reduced financial outlay, which may be allocated to other priority investments in health.
Despite the study’s findings, several important questions need to be considered.For example, the authors assume that coverage of 80% is achievable and vary this value within a limited range only. Although a study in the BMJ reported encouraging uptake rates of first and second doses of vaccine in schoolgirls,10 the uptake rate for the full three dose series is unknown, and this will affect the magnitude of direct and indirect benefits. Also, an important finding from that study was that uptake was lower in girls from minority groups and from less affluent backgrounds. The extent to which these girls receive less screening in adulthood—and consequently face a higher incidence of cervical cancer—will influence the overall success of the vaccination programme and may widen disparities in the risk of developing cervical cancer among socioeconomic groups. Although Jit and colleagues include a small subgroup of women who are unscreened, their model does not accommodate further heterogeneities in screening behaviour in women to explore this matter more thoroughly. Other analyses have reported that the cost effectiveness of human papillomavirus vaccination is compromised when vaccination uptake is higher in women who are screened frequently in adulthood, suggesting that equitable access to the vaccine should be a priority.6 11
Furthermore, because nearly a third of cases of cervical cancer are attributable to non-vaccine human papillomavirus types, cervical screening will continue to be a vital component of cancer prevention efforts in the UK. Several analyses have shown that human papillomavirus vaccination is more cost effective when followed by less frequent screening, starting at later ages and with newer screening technology, such as testing for human papillomavirus DNA.11 12 13 Although the current model cannot look at changes in screening practice, alternative policies that efficiently synergise vaccination with screening should be considered and evaluated carefully.
Policy decisions that are being made now will continue to benefit from model based analyses that aim to synthesise the best available data, as long as model inputs and assumptions are iteratively revised as new information becomes available. It may be decades before we see the true effect of human papillomavirus vaccination on cervical cancer, even though most studies indicate that vaccinating adolescent girls will provide benefit and be cost effective; Jit and colleagues’ study shows that this is likely to be the case even in the context of current screening practice in the UK. Better data on the natural history of human papillomavirus and the properties of the vaccine are needed to generate a similar consensus on other policy questions involving catch-up age limits and including boys in the immunisation programme.
Cite this as: BMJ 2008;337:a842
Competing interests: None declared.
Provenance and peer review: Commissioned; not externally peer reviewed.