Intended for healthcare professionals


Comparing bivalent and quadrivalent HPV vaccines

BMJ 2011; 343 doi: (Published 27 September 2011) Cite this as: BMJ 2011;343:d5720

This article has a correction. Please see:

  1. René H M Verheijen, professor of gynaecological oncology
  1. 1Division of Women and Baby, Gynaecological Oncology, University Medical Centre Utrecht, Utrecht 3584 CX, Netherlands
  1. r.verheijen{at}

Modelling can help, but the tender price determines cost

In September 2008, the Department of Health in the United Kingdom started a national vaccination programme of human papillomavirus (HPV) in 12-13 year old schoolgirls. A catch-up programme for girls up to the age of 18 was simultaneously started. At that time, Jit and colleagues reported cost comparisons of the two available vaccines.1 Their cost threshold analysis showed that if the bivalent vaccine, Cervarix, which protects against HPV types 16 and 18, cost £13 (€15; $21) to £21 less per dose than the quadrivalent vaccine, Gardasil, both vaccines would be equally cost effective, although the bivalent vaccine would be less effective because it does not prevent anogenital warts. In an accompanying editorial, Kim concluded that policy decisions would continue to benefit from such model based analysis.2

Now, three years later, a political decision is again needed on the continuation of the vaccination programme, and Jit and colleagues have re-evaluated the two vaccines, taking into account more potential differences between the vaccines (doi:10.1136/bmj.d5775).3 They conclude that at the same price level the quadrivalent vaccine is still more cost effective and that the price differential between both vaccines seems larger than stated in their previous analysis.

In 2008 it was unclear whether the target of 80% coverage would be reached. Now we know that coverage for all three doses ranges from 76.4% (England) to 86.9% (Scotland).4 This excellent result was due to an efficient and well planned launch, and it secured the effectiveness of the vaccination programme at a population level. This coverage compares well with other countries—in the Netherlands coverage was 51.9% at first call and is still only 56.4% after recall.5 The Dutch paid for ignoring the importance of modern communication through social media networks and the impact of negative, although unfounded, messages conveyed by them. The safety of both the bivalent vaccine, which was used in the UK national programme, and the quadrivalent vaccine were monitored weekly by the Medicines and Healthcare Product Regulatory Agency (MHRA) in more than four million doses distributed across the UK.6 Only recognised and listed side effects were reported, and other adverse findings could not be related to the vaccines.

How should we decide which vaccine to use for a national vaccination programme? An evidence based approach should ideally rely on more than one comparative study. In one observer blinded head to head study the bivalent vaccine induced a higher, more sustained immune response than the quadrivalent vaccine.7 However, a sustained rise in antibodies may not be needed to secure protection against a future infection.8 Conversely, antibodies other than those tested for may have a protective role.

Another selection criterion could be the duration of protection. Although reports on sustained antibody titres have a different time range for both vaccines, Jit and colleagues justifiably use three scenarios for duration of protection that are similar for both vaccines.

Protection against multiple cancers would also be an attractive argument in favour of a vaccine. In one of Jit and colleagues’ scenarios, “the pessimistic one,” the authors assume that the bivalent vaccine protects against cervical cancer alone, whereas the quadrivalent vaccine also protects against other types of HPV-16 and HPV-18 related cancers. Just because clinical trials with the bivalent vaccine used only cervical cancer as an end point does not mean that it will not also be efficacious against other types of cancer. It is unlikely that the vaccines differ in this respect.

Jit and colleagues also made assumptions about protection against respiratory papillomatosis, a rare HPV-6 and HPV-11 related disease that is life threatening in young children. With no evidence, they assumed that a reduction in this disease, even in vertical infection, would follow the same pattern as for genital warts. If this were true, even protection against (respiratory) warts could save lives.

Finally, we should not forget that the vaccination programme aims to save lives from cervical cancer. This implies that politicians should trade deaths from cancer against morbidity from warts. The bivalent vaccine is more effective in preventing death as a result of cancer and for possibly longer than the quadrivalent vaccine, as acknowledged and assumed in the model used by Jit and colleagues. It has been estimated that the Italian programme, which uses the bivalent vaccine, would prevent 295 more deaths from cancer but 25 848 fewer cases of genital warts than if it used the quadrivalent vaccine.9 Evaluation in terms of quality adjusted life years unfortunately does not reflect prevention of death, but it is necessary for the calculation of the incremental cost effectiveness ratio, which in turn defines the reasonable cost per dose.

Several models, mainly the Markow model and the transmission dynamic model, have been published to predict the cost effectiveness of various vaccination strategies. As Jit and colleagues also state, none of these economic evaluations has considered all the potential differences between the vaccines within the same model. In all models, assumptions are made on the basis of studies of the efficacy of the vaccines, measured by surrogate end points, and by immunological follow-up studies. And in every model evaluation, the bivalent vaccine would be cost effective only if it were a dozen or so pounds cheaper than the quadrivalent vaccine.

Evidence of this type and level of difference should help decision makers. In this light it is essential to acknowledge that in all these models that have been meticulously validated and precisely calculated, the most decisive variable of all, the tender price, is confidential and thus unknown and not taken into account. In the end then, the key determinant of cost effectiveness is the only factor that cannot be evaluated, even though it will be important when deciding on the vaccine to be used in a national prevention scheme. Unfortunately, in most countries such decisions are also confidential, so we will never know whether the model or the money mattered.


Cite this as: BMJ 2011;343:d5720


  • Research, doi:10.1136/bmj.d5775
  • Competing interests: The author has completed the ICMJE uniform disclosure form at (available on request from the corresponding author) and declares: no support from any organisation for the submitted work; RHMV has received speaker’s fees and travel fees from GSK; RHMV has been a principal investigator for a GSK clinical trial.

  • Provenance and peer review: Commissioned; not externally peer reviewed.


View Abstract

Log in

Log in through your institution


* For online subscription