Efficacy and safety evidence
The primary objective of this review was to assess the effects of palivizumab prophylaxis compared to placebo, or another type of prophylaxis (e.g. motavizumab), in reducing the risk of hospitalization due to respiratory syncytial virus (RSV) infection in high-risk infants and children.
Palivizumab prophylaxis was associated with a statistically significant reduction in RSV hospitalisations, when compared to placebo. The magnitude of this effect is considerable; palivizumab reduced the risk of RSV hospitalization by half. Children treated with palivizumab prophylaxis spent fewer days in hospital, were admitted to the intensive care unit (ICU) less often, and had fewer days of supplemental oxygen therapy for RSV infection than the placebo recipients. These results suggest a favorable effect of palivizumab prophylaxis on the incidence of serious lower respiratory tract RSV disease in children at high risk.
Results for the total days spent in the ICU and the incidence and total days of mechanical ventilation were inconsistent between the two trials, possibly due to different severity of underlying medical conditions in children included in the trials. Children with haemodynamically significant congenital heart disease experienced fewer days in the ICU, and lower incidence and fewer days of mechanical ventilation when treated with palivizumab prophylaxis compared with placebo, while children born preterm, with or without bronchopulmonary dysplasia, having less severe baseline risk factors for RSV disease, showed the opposite trend in results. It could be that a drug has a larger relative effect in sicker populations. But also, medical practices, guidelines and recommendations on when to discharge from an ICU, and when to initiate and wean mechanical ventilation, differ substantially in different settings.
When palivizumab was compared to motavizumab prophylaxis, there was an obvious trend of increase in the risk of acquiring severe lower respiratory tract RSV disease in patients receiving palivizumab. The risk of hospitalization due to RSV infection was increased by one-third in palivizumab patients. Children treated with palivizumab had more admissions to intensive care, a higher need for supplemental oxygen therapy and more instances of mechanical ventilation than the children treated with motavizumab prophylaxis.
However, the results of trials comparing palivizumab and motavizumab should be interpreted with caution. The US Food and Drug Administration (FDA) cited methodological concerns (among others) in its review of a licensing application for motavizumab, specifically with regards to the laboratory testing of RSV, which may have biased the results toward motavizumab over palivizumab. The FDA also expressed concerns regarding both motavizumab's safety and efficacy. The non-fatal hypersensitivity adverse events were found to be three times higher in the motavizumab group than in the palivizumab group. Also, motavizumab's non-inferiority results are largely driven by data obtained from southern hemisphere countries, representing only about 9% of the total patient population. Removing this population led the FDA to determine that in the US population motavizumab did not meet the non-inferiority criterion relative to palivizumab (FDA 2010). These issues have implications for assessing the risk of bias in Carbonell-Estrany 2010, and the results should be interpreted with this in mind.
Total days in the hospital due to RSV infection, days spent in the ICU, receiving mechanical ventilation or with supplemental oxygen therapy were all much higher in palivizumab than in motavizumab-treated patients.
Data on days were summarised and presented as days per 100 randomized children in a study arm, and not as days per one child. That is why they are expected to be proportional to the number (or better, to the rate) of patients hospitalised from that study arm. For example, if two children out of 100 are hospitalised in one group, and one child out of 100 is hospitalised in the other group, and if each child from both groups stays in the hospital for five days, in the first group we would have 10 hospital days per 100 randomized children, and in the second five hospital days per 100 randomized children. The larger the incidence of RSV hospitalisations, the bigger are the numbers of days. This is a common problem in almost all RCTs in this review: results on total days presented in this way could be misleading. We cannot interpret them as a measure of severity of the disease, once a child has acquired an RSV infection. For both clinicians and patients it would be more beneficial for study authors to express the mean number of days per one child in future studies.
Another common problem of almost all studies reporting data on days is their incomplete reporting. Measures of dispersion were not provided and we cannot be confident about the precision of the results.
RSV-specific outpatient medically attended lower respiratory tract infections were assessed for motavizumab and palivizumab patients in two studies. Motavizumab was associated with statistically significant reduction in RSV-specific outpatient MALRIs when compared to palivizumab; the risk in the palivizumab group was twice that of the motavizumab group. However, this outcome was assessed in just a subset of patients, either in patients from selected study sites or in patients in season two only, making the risk of bias in these results high.
Palivizumab-treated children had lower mortality rates than children treated with placebo or motavizumab. However, it was hard to draw any conclusions since all studies expressed mortality as deaths due to any cause, regardless of their relation to study drug or to RSV infection. We were surprised to find that within-study mortality rates differed substantially between the studies (e.g. Carbonell-Estrany 2010 reports an all-cause mortality rate 30 times that of Feltes 2003). The difference could be attributed to different sample sizes and different underlying medical conditions in patient populations in the two studies.
For both comparisons we analyzed the proportion of children with adverse events in four categories, depending on the seriousness of the adverse event (AE) and its relatedness to the study drug. Unfortunately, we could not analyse a specific adverse event or adverse events grouped by organ systems, due to different reporting methodologies in studies.
As we expected, considering the underlying medical conditions in this high-risk population, rates of AEs and SAEs were high in all treated patients. Palivizumab was associated with a statistically significant reduction in the proportion of children reporting any SAE compared to placebo. After having confirmed the efficacy of palivizumab prophylaxis, these results are self explanatory. Palivizumab reduces the risk of severe RSV disease after the RSV infection has occurred, and thereby minimises hospitalisations, or possibly some life-threatening conditions or significant disabilities, which are all considered serious adverse events. It should also be noted that this result came from one study only. The proportion of children reporting any adverse event related to study drug, or any adverse event at all, was similar in palivizumab and placebo patients. Post-marketing surveillance data included in the Synagis (palivizumab) product leaflet provide additional insight into potential adverse events encountered, specifically: severe acute hypersensitivity reactions and anaphylaxis, which are described as rare and very rare (respectively) (FDA 2009).
We did not find any differences in the proportion of children reporting AE when palivizumab was compared to motavizumab in any of the four categories assessed. The proportion of children reporting any AE was similar between the two groups, and additional analyses of other groupings of AEs did not demonstrate any significant difference.
In all included economic studies, a cost-effectiveness or a cost-utility analysis was conducted that compared the clinical and financial consequences of palivizumab prophylaxis and no prophylaxis in infants and children at high risk. In this section, the Drummond definitions of the types of economic evaluations were followed (Drummond 1996) and all studies were classified into a health sector (payer's) or a societal perspective.
In general, costs and outcomes can be combined in three different ways, resulting in three different types of analyses: cost–benefit analysis (where both inputs and outcomes are considered in monetary terms); cost–utility analysis (where inputs are considered in terms of costs, and outcomes are measured in utility measures, such as quality-adjusted life-years (QALYs)); and cost–effectiveness analysis (where inputs are measured in terms of costs, and outcomes are measured using measures specific to the disease). A QALY is estimated in terms of a year of life, adjusted by the amount of quality that the life is lived at. Therefore, one year lived at full quality is 1 QALY, but one year lived at half quality equates to 0.5 QALYs, and half a year at full quality is also 0.5 QALYs. Different diseases and conditions can be compared using the cost–utility analysis and, therefore, these types of analyses are especially used by governmental approval groups, such as the UK National Institute for Health and Clinical Excellence, which often sets a threshold of utility gains per cost for all drugs and health technologies. A cost–effectiveness analysis usually compares the costs and outcomes of similar treatments for specific conditions. However, it would not provide data on the incremental cost–effectiveness ratio (ICER) per QALY, and if such data are required, would need to be modelled from the cost–effectiveness data.
Whether an intervention is cost-effective or not, and whether it should be provided or not, depends on the cost-effectiveness threshold established by the decision makers in a particular country. Following the recommendations of the Commission on Macroeconomics and Health, the World Health Organization (WHO) has derived three categories of cost-effectiveness using the nominal gross domestic product (GDP) per capita as a measure:
highly cost-effective (ICER is less than one GDP per capita);
cost-effective (ICER is between one and three times GDP per capita); and
not cost-effective (ICER is more than three times GDP per capita).
The nominal GDP per capita for the European Union (EU) for year 2011, as calculated by the World Bank, was USD 34,848 (EUR 24,621.37) (available at http://data.worldbank.org/). Using this GDP to calculate the cost-effectiveness threshold, the immunoprophylaxis would be cost-effective for the EU countries if the ICER present value at 2011 EUR is lower than EUR 73,864.11 per QALY. However, this threshold is substantially higher than the thresholds established by particular EU countries, e.g. the United Kingdom's cost-effectiveness threshold has been in the range of GBP 20,000 to GBP 30,000 for over 10 years now (EUR 22,791.74 to EUR 34,187.61 respectively, using the 2011 exchange rates).
As Peter Jacobson (Jacobson 2001) stated: "Cost control is a primary objective of the managed care environment. It is no longer possible to provide health care without regard to cost or patient demand. The question is not whether there will be cost containment, but how to structure and oversee its implementation. The use of cost-effectiveness analysis (CEA) in making clinical and payment decisions has become a significant cost containment approach, however CEA should be treated as one piece of evidence to be considered by health care sector to define way of action rather than being used to determine the standard of care."
We presented and discussed economic data separately according to age and subgrouped data according to underlying medical conditions because, clinically, these patients are likely to have different baseline risks for serious complications due to RSV infection. We further classified the economic evaluations by whether they adopted the payer's or the societal perspective. We also debated about the main economic results obtained from the included studies and about variations in methodological approaches among studies that may justify the differences in cost-effectiveness results.
Data on cost–effectiveness of RSV immunoprophylaxis with palivizumab versus no prophylaxis are based on simulation modelling, rather than the direct collection of costs and outcomes. Data for the evaluations were drawn from a wide variety of sources, including the palivizumab clinical trials, published literature, hospital databases, country-specific price/tariff lists and national population statistics. Country-specific data sources were also used for economic measures and information on therapeutic choices. Clinical events and utilities in the majority of analyses are not country-specific and therefore were derived from international studies.
The main outcomes considered for cost-effectiveness analyses in the included economic studies were hospitalization due to RSV infection (ordinary ward or ICU) and life-years gained (LYGs). For cost-utility analysis, outcomes considered were QALYs. Challenges in the cost–utility approach for this specific problem lie in modelling of costs and QALY gains in the lifetime follow-up period to capture the impact of palivizumab on long-term morbidity and mortality, resulting from severe RSV infection beyond the RSV hospitalization period. Under the assumption that RSV hospitalisations are associated with clinical and economic consequences beyond the clinical trial period; a proportion of children may develop long-term sequelae (e.g. wheezing or asthma) leading to a reduction of QALYs and additional medical costs. It is known that the rates used to populate the economic models will drive the final results of the analyses towards higher or lower ICER values. The reduction in RSV hospitalization rate due to palivizumab prophylaxis corresponds to data available from the palivizumab clinical trials (e.g. IMpact-RSV 1998) that considered only one season period of follow-up, which is 150 days from the point of randomization (30 days after the last scheduled palivizumab injection). Therefore, for the economic models that adopted the lifetime time horizon it was necessary to extrapolate the efficacy data from the palivizumab clinical trials (reduction in the rate of RSV hospitalisations) to calculate the likely number of LYGs and QALYs gained from the use of palivizumab prophylaxis. Regardless of the time horizon considered in the analysis, if authors assumed that differences in RSV hospitalization rates allow for differences in mortality rates between the palivizumab prophylaxis and non-prophylaxis group, and thus populated the models from the beginning with differential mortality rates, the final results will favour palivizumab use, particularly if the societal perspective was adopted.
Modelling costs depend on the perspective of the analysis. The analyses performed from the societal perspective included not only the direct medical costs, but also costs for management of wheezing or asthma, and future lost productivity of a child resulting from mortality (a small proportion of children will die, which will lead to a lifetime loss of productivity benefits). The analyses that adopted the payer's perspective considered only direct medical costs. Generally, analyses that included direct medical costs associated with asthma (i.e. when asthma was included into the disease pathway modelled) showed moderately more favorable ICERs for palivizumab prophylaxis, while analyses that included the long-term indirect costs due to lost lifetime productivity following childhood mortality, showed a substantial improvement in the cost-effectiveness of prophylaxis with palivizumab. It means that palivizumab prophylaxis is more cost-effective if it has a long-term effect on the incidence of asthma and mortality.
A very important consideration that should be taken into account while interpreting the economic results presented in this review is that effectiveness data used to populate the models come from follow-up studies performed in hospitalised children, RSV-infected or not, with or without the underlying medical conditions (such as bronchopulmonary dysplasia or congenital heart disease); none of the studies measured the long-term impact that palivizumab prophylaxis could have on asthma and mortality in these high-risk populations. So, data used by study authors to populate the economic models are based on unsupported assumptions. Whether or not these assumptions and modelling practices lead to underestimation or overestimation of the mortality rates in children born preterm or with underlying heart or lung disease, that have received immunoprophylaxis with palivizumab, is unclear.
Currently there are no longitudinal trials providing robust data on long-term effects of palivizumab prophylaxis on a child's morbidity and mortality beyond the standard follow-up period. The forthcoming results from one investigator-initiated RCT (NTR1023) that assessed the number of wheezing days in preterm children during the first year of life, and the quality of life and asthmatic symptoms up to six years of age, might be offering some answers to this question.
Owing to particular problems described above, and due to the fact that mortality rates could drive the final cost-effectiveness ratios, it is important to discuss the methodological approaches used by the authors to model this outcome in their economic analyses. Methods that were used for reporting, calculating and adjusting the probabilities of death that entered the models differed considerably across studies. In most cases, the absolute values of probabilities of death were not reported and the exact organisation of decision tree models was not presented. Some study authors directly modelled different mortality rates for patients receiving palivizumab prophylaxis and no prophylaxis. In some of the studies, models included difference in life-years gained between the two intervention groups, and this fact necessarily implies that a difference in mortality was also allowed. Some study authors assumed different mortality rates for hospitalised versus non-hospitalised patients. In other instances, different mortality rates were assumed for patients hospitalised with or without the RSV infection. Again, some other studies assumed the same mortality rates for the two intervention groups, but calculated the probabilities of death according to the related hospitalization rates in the two groups. The bottom line is that all these studies took into account that palivizumab prophylaxis reduces the rate of RSV-related hospitalisations, and this directly translates into reduced mortality risk in palivizumab group compared to no intervention group. Rare authors did not model a mortality benefit associated to the use of prophylaxis, but this was the case only in studies with a short time horizon (one year), and with final costs expressed per hospitalization avoided (the exception being the ElHassan 2006 study).
Other important differences in economic models included in this review are the different total amount of the drug, different resources and services consumed (depending on a healthcare system in a particular country), different overall costs (dependent on the costs that specific services/resources have in a specific country), different time horizons and different discount rates.
Each of the factors described could easily account for large differences in cost-effectiveness results across studies. An additional aspect that we have studied, while interpreting the economic results presented in this review, is whether the analysis was funded by the drug manufacturing company or not. Almost all included studies that were sponsored by the industry supported the cost-effectiveness of palivizumab prophylaxis, while practically all included studies that were not sponsored by the industry suggested that palivizumab was not cost-effective.
We made attempts to classify studies according to all these differing assumptions included in economic models, in an effort to identify premises that would be necessary for palivizumab prophylaxis to be regarded as acceptably cost-effective. However, by analysing the information available from the study reports, it became obvious that a huge problem lies in the lack of standardisation of the modelling approaches adopted in economic studies, and these differences can easily lead to big variations in cost-effectiveness results, making them almost incomparable. The use of palivizumab prophylaxis for reducing the risk of severe RSV infection might not be cost-effective enough to be considered a standard healthcare policy in the majority of low- and middle-income countries, because of the high costs of the drug. However, patient needs and individual risks should be considered in each case that physicians encounter in their everyday clinical practice.
Economic impact of immunoprophylaxis given at neonatal period or within the first six months of life
In an attempt to find systematic differences that could explain the variations in results of the studies reporting on preterm infants without other comorbidity, we analyzed patient populations, effectiveness outcomes, perspective taken and other methodological parameters. The doses of palivizumab varied across studies (from 3 to 6 doses at 15 mg/kg); gestational ages of preterm infants entered into the models differed between the studies; incremental effectiveness of palivizumab prophylaxis varied substantially across studies (i.e. RSV hospitalisations avoided, risk of asthma included, lower mortality rates due to palivizumab use). Finally, the included studies reported significant differences in economic results, coming primarily from the consumption of resources taken into account, and from the modelling approaches adopted. Many analyses considered a lifetime follow-up period to capture the impact of palivizumab on a long-term morbidity and mortality resulting from severe RSV infection. Since the available data from palivizumab clinical trials are all limited to a single RSV season, the way of modelling the evaluations presents an important source of variations leading to such differences in ICERs.
Economic impact of immunoprophylaxis given to children aged six months and older
Two studies evaluated the economic impact of RSV immunoprophylaxis in preterm children without other co-morbidities. The ICER values expressed per QALYs varied across these two studies substantially, making it difficult for decision-makers to identify the real magnitude of the economic impact that the palivizumab prophylaxis has in this population.
In the studies evaluating the economic impact of passive immunisation given to children with congenital heart disease, substantially higher ICER values expressed per QALYs and LYGs were reported by Wang 2011 and Yount 2004. These studies had comparable methodological characteristics to other studies, and they both included mortality benefits and lower risk of long-term sequelae for children receiving palivizumab prophylaxis. We did not find any clear explanations for this variation, other than that these two studies were the only ones not funded by the drug manufacturer.
Results from studies performed in children with bronchopulmonary dysplasia (or chronic lung disease) aged six months and older are quite consistent and rather high. Whether palivizumab prophylaxis is a cost-effective alternative, and whether it should be adopted as part of routine care in this population, depends on the threshold value set by the decision-makers in a particular country.
Economic impact of immunoprophylaxis given to high-risk infants and children (born preterm, with or without bronchopulmonary dysplasia, or with congenital heart disease) up to five years of age
From the evidence presented in the three included studies, it is very difficult to define the real economic impact that the RSV prophylaxis strategy has in a mixed population of high-risk infants and children.