Reply

19 December 2000 Dear Editor REPLY Drs Mills and Dargaville point to a number of perceived limitations in my analysis of the costs of using two immunoprophylactic agents (RespiGam® (CSL Laboratories, Melbourne, Vic., Australia) and Synagis® (Abbott Australasia, Kurnell, NSW, Australia)) against respiratory syncytial virus (RSV). As RespiGam® is no longer available in Australia I will limit my comments to Synagis®. Drs Mills and Dargaville are concerned that the methods used in my study are dissimilar to those recommended by Breese Hall et al.1 Unfortunately, they appear not to have read Breese Hall's paper too closely, as one of several methods of analysis advocated in their paper is to use ‘an estimate from the literature’, which is precisely what I have done. In any event, the real issue should be whether or not my methodology is valid, not how closely it conforms to methods espoused by other groups. Surely the field of health economics has not yet reached the stage where there is no room for new methodologies and original ideas. In my analysis I used data regarding admission numbers and length of stay to obtain an actual cost of caring for infants with RSV at Sydney's Children's Hospital over a given period of time (three seasons). It is simple (and valid) to work backwards from this data in order to obtain an estimate of how many admissions we could prevent (and calculate the expected costs or savings) with prophylaxis. In order to obtain this estimate I relied on data collected from hospitals in the USA as part of the IMpact-RSV study.2 Although Drs Mills and Dargaville have concerns about the applicability of this data to the local population, it is highly likely that the US data will reflect our own experience; certainly there are data from studies in the USA and Canada which at least indicate that admission rates and length of stay in the Intensive Care Unit (ICU) are comparable to those observed locally.3,4 It is true I have made several assumptions in calculating costs; however, wherever an assumption has been made the underlying premise was to bias the analysis in favour of Synagis® rather than against it. For example, costs of hospital care were estimated using the assumption that every infant with bronchopulmonary dysplasia (BPD) who acquired RSV infection and required admission to hospital would be treated in the ICU, with a length of ICU and hospital stay (not ICU alone as claimed by Drs Mills and Dargaville) equal to the average observed in our retrospective cohort. This grossly exaggerates the true costs of hospital care, meaning that the drug costs are seen in a more beneficial light. However, even using the most generous assumptions in favour of prophylaxis, the numbers remain overwhelmingly against its use. The absolute accuracy of my assumptions might be more relevant if the cost issues were ‘line ball’ between using or not using the drug. However, as the analysis demonstrates, the difference in expenditure between prophylaxis and non-prophylaxis is measured in orders of magnitude, making any minor inaccuracies in underlying assumptions largely irrelevant. There is undoubtedly intrinsic value in knowing the local re-admission rates for various subgroups of high-risk infants who acquire RSV. Unfortunately however, such information is of little value in performing local cost–benefit assessments as at some stage any analysis must resort to using published rates of reduction in re-admission from the large randomized controlled trials (e.g. the IMpact-RSV. study2). The only alternative to doing so is to perform a randomized controlled trial of drug versus placebo to gauge the specific effect in the local population – an undertaking that is as impractical as it is unnecessary. Drs Mills and Dargaville state that I have made no attempt to look at infants ‘at the highest risk of re-admission; those less than 32 weeks gestation with prolonged oxygen dependency discharged just prior to the RSV season’. This is because there are no data in the literature to indicate the effect of Synagis® in this selected population. There is no point in identifying high-risk subgroups without also knowing the effect of Synagis® in each such subgroup. It is erroneous to seek to justify the use of the drug in a selected population of very high-risk infants with high admission rates by assuming that prophylaxis in this group will reduce admission rates to the same degree as observed in a broader (lower risk) group. In the IMpact-RSV study, infants without BPD experienced a 78% reduction in admission rate (from 8.1% to 1.8%) with prophylaxis, whereas those with BPD experienced only a 39% reduction (from 12.8% to 7.9%).2 This suggests the drug might be least effective in those who need it most. Selecting even higher risk subgroups may well demonstrate even smaller beneficial effects. I emphasise again that the beneficial effects (if any) in such high-risk groups are unknown at this time; indeed it is not impossible that the drug may be deleterious in some specific subgroups. The most detailed part of my analysis focused on infants with BPD (not specifically those born at less than 32 weeks gestation) as these were demonstrated in the IMpact-RSV study2 and the PREVENT study5 as being the subgroup with the highest rates of admission (and thus most likely to benefit from prophylaxis). As Drs Mills and Dargaville seem to have difficulty understanding the analysis of this group as presented in my paper, I have set out an even simpler estimate of costs below. Assuming that the cost of administering the drug is limited to drug cost alone (i.e. no staff or ancillary equipment costs), and all patients treated weigh less than 6.7 kg (i.e. one ampoule only per treatment), the cost of a 5-month course of treatment will be $1456 × 5 = $7280. This is the very lowest estimate of cost we can make for prophylaxis with Synagis®. Administering prophylaxis to 50% of a pool of 2000 hypothetical infants with BPD will cost $7.28 million. The IMpact-RSV study data suggest 128 of the 1000 non-prophylaxed infants will be admitted to hospital compared to only 79 of the 1000 treated infants (with no difference in length of stay). It thus costs $7.28 million to save 49 hospital admissions, or $148 571 for each hospital admission saved. This is enough to cover the cost of 87 days in ICU (at $1700 per day) or 175 days in the general wards (at $850 per day) per patient. I am sure that Drs Mills and Dargaville would concede this is far in excess of the requirement for hospitalization in any high-risk group of patients they would be able to identify. Finally, Drs Mills and Dargaville take issue with the use of the term cost–benefit analysis. According to the definition proposed by Zarnke and coworkers, I have indeed performed a cost–comparison rather than a cost–benefit analysis.6 However, I see little relevance in this rather pedantic complaint. I also suggest that a true cost–benefit analysis is impossible to carry out according to this definition, which maintains that the benefits to patients of receiving treatment must be measured in monetary terms. Zarnke et al. themselves state in their article, ‘Not surprisingly, the main controversy surrounding cost benefit analysis is how monetary values are to be attached to benefits such as health outcomes’. In a further discussion of this issue, they cite a commonly used method of looking at loss of production caused by the disease; however, ‘no value is attached to the health status of those who are not working such as the retired and unemployed’ (and of course, children). If Drs Mills and Dargaville are seriously suggesting that we apply such hard line economic rationalism to the care of children we will be scarcely able to justify anything we do as paediatricians. An alternative method of assigning a dollar value to life-years saved by treatment is also widely used, but this assumes, of course, that the treatment used actually saves lives – manifestly unproven in the case of Synagis®.2 Whether the analysis is termed cost–benefit or cost– comparison is somewhat immaterial to the underlying message of the paper – the drug is simply too expensive and not effective enough to justify its use at present. It may be possible to identify selected very high-risk subgroups of infants in whom the expenditure on Synagis® prophylaxis may be justifiable. It is important to be aware that such subgroups would not only have to have an extremely high likelihood of hospital admission if infected with RSV but, in addition, have a proven substantial reduction in admission rates following prophylaxis. Thus, although it is superficially appealing to selectively administer the drug to infants most at risk of needing hospitalisation, to do so without knowing the expected reduction in admission rates in this specific subgroup cannot be justified on economic grounds.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.