- Email: [email protected]
Pre-exposure rabies vaccination: strategies and cost-minimization study
Vaccine 19 (2001) 1416 – 1424 www.elsevier.com/locate/vaccine
Pre-exposure rabies vaccination: strategies and cost-minimization study Christrophe Strady a, Van Hung Nguyen b, Roland Jaussaud a, Jean Lang b, Michel Lienard c, Alain Strady a,* a
Centre Antirabique, Hoˆpital Robert Debre´, A6enue du Ge´ne´ral Kœnig Reims Cedex, 51092, France b Pasteur Me´rieux Connaught, 58 A6enue Leclerc, Lyon Cedex 07, 69348, France c Mutualite´ Sociale Agricole, Caisses centrales, 40 Rue Jean Jaures, Bagnolet, 93547, France Received 22 December 1999; received in revised form 13 September 2000
Abstract An alternative strategy for pre-exposure rabies vaccination to the institutional recommendations of the World Health Organization and the Centers for Disease Control and Prevention is proposed based on recent long-term follow-up of post-vaccinal seroconversion rates. The alternative strategy uses the same primary series (i.e. vaccination in the deltoid area on D0, D7, and D28), but is completed by a scheduled booster vaccination at D365. The frequency of recommended subsequent booster injections depends on the serological test results obtained by a RFFIT on D379 and 3 years later. The objective of this study was to compare the efficiency of the two pre-exposure strategies. A cost-minimization analysis was carried out to compare the two rabies pre-exposure vaccination and serological test strategies based on the data from two published studies on the long-term evolution of the immunity achieved using the different recommendations. For a theoretically equivalent immunogenicity, the cost of the alternative strategy ranged from 1.7 to 5.2 times lower than that of the institutional recommendations. A sensitivity analysis confirmed the robustness of the results. The alternative strategy should be validated externally under field conditions. This approach would compare its real efficiency to the institutional recommendations. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Rabies; Pre-exposure vaccination; Cost
1. Introduction The recommendations of the World Health Organization (WHO) [1] and Advisory Committee of Immunization Practices of the Centers for Disease Control and Prevention (CDC) [2] for pre-exposure rabies vaccination include the administration of three injections of a cell culture vaccine given intramuscularly in the deltoid area on days 0, 7, and 21 or 28. Titration of neutralizing antibodies by a standardized Rapid Fluorescent Focus Inhibition Test (RFFIT) is recommended to assess the seroconversion of the subjects at high risk * Corresponding author. Present address: Centre Hospitalier Universitaire de Reims, Hoˆpital Robert Debre´, avenue du Ge´ne´ral Kœnig, 51092 Reims Cedex France. Tel.: +33-3-26787185; fax: +33-3-26784090. E-mail address: [email protected] (A. Strady).
for exposure to the virus, with a frequency adapted to that risk. A subject is considered to have seroconverted when the antibody level is ] 0.50 IU/ml [1] or with complete virus neutralization at a 1:5 serum dilution [2]. For persons at continuous risk (e.g. workers in production or research laboratories in the field of rabies), antibody levels should be assessed every 6 months. In case of frequent risk (e.g. rabies diagnostic laboratory workers, spelunkers, veterinarians and staff, and animal-control and wildlife workers in rabies-enzootic areas), the WHO and the CDC, respectively recommend antibody level assessment every 1 and 2 years. In all cases, boosters should be given as soon as the subject is found to be seroreverted (i.e. when the titer falls below 0.50 IU/ml). If the risk is considered to be infrequent (e.g. veterinarians and animal-control and wildlife workers in areas with low rabies rates, veterinary students, travelers visiting areas where rabies is enzootic
0264-410X/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 4 - 4 1 0 X ( 0 0 ) 0 0 3 6 8 - 6
C. Strady et al. / Vaccine 19 (2001) 1416–1424
and immediate access to appropriate medical care including biologics is limited), the WHO does not issue any recommendation, while the CDC indicates that there is no need for serological tests or routine booster injections. We previously conducted a prospective study on the 10 year follow-up of seroconversion rates (SCR) and antibody levels obtained after pre-exposure rabies vaccination [3,4]. In particular, our last study demonstrated the long-term immunogenicity of the 3-injection regimen (D0, D7, and D28) completed by a scheduled booster injection at 1 year (D365) [4]. The cell-culture vaccines used during this study were the human diploid cell rabies vaccine (HDCV) [5,6], used in the USA, and the vaccine prepared on Vero cells (PVRV) [7,8], used in France and in 41 other countries in which 15 million doses have been distributed. Vaccines were administered intramuscularly in the deltoid. An alternative strategy to the institutional recommendations of the WHO and the CDC as regards antibody serological tests and booster injections was suggested based on the results of this study. The proposed alternative strategy is based on antibody level assessment at times that are particularly predictive of the evolution of the SCR and antibody titers over the 10 year follow-up period as determined by a multivariate analysis of our prospective study [3,4,9]. On D379 (i.e. 14 days after booster at 1 year), a titer ] 30 IU/ml had a positive, predictive seroconversion value of 100% at 10 years while a titer B30 IU/ml had a negative, predictive seroconversion value of 18%. In other words, a subject with a titer of at least 30 IU/ml on D379 would definitely still be seroconverted after 10 years. Conversely the likelihood of seroreversion in subjects with a titer of B30 IU/ml on D379 was 18% at 10 years. In addition, if subjects seroreverted they did so between the second and third year after D379. Thus, at D379+ 3 years, 100% of subjects who had a titer ] 0.5 IU/ml were still seroconverted at 10 years. In contrast, subjects with titers B 0.5 IU/ml at this time were considered to be low responders [10], for whom subsequent serial scheduled booster vaccinations could be justified. We thus identified these two preferred times (i.e. D379 and D379 +3 years) as being particularly indicative of an individual’s serological evolution, and these timepoints were chosen for the scheduled measurements of antibodies in the alternative strategy [9]. The neutralizing antibody titer, as an efficacy criterion for pre-exposure anti-rabies vaccination is a point worth discussing. Since rabies is a 100% fatal disease, any vaccination strategy has to guarantee total protection from exposure, and the efficacy of the program can not be sacrificed in any optic of saving costs. However, the paradox is that, due to the nature of the disease, ethically it is not possible to obtain any clinical evalua-
1417
tion of the efficacy of a rabies vaccination program. The only possible criterion available is this neutralizing antibody titer. This is the consensual criterion of immunogenicity in the clinical evaluation of rabies vaccines [1,2,11]. It is this criterion which has been used in the cost minimization analysis and the immunogenicity of a strategy was expressed by the SCR observed in the cohort of subjects submitted to this strategy. Although cost-minimization studies have been used to compare the economic values of various drug regimens, they have been infrequently employed to compare vaccination strategies. The objective of this present study was to use this pharmacoeconomic approach to compare the efficiency of the two pre-exposure strategies, which differ in their pre-exposure vaccination protocol and assessment of persistence of measurable immune response through antibody serological measurements.
2. Methods
2.1. Choices and assumptions The efficiency of the two strategies was assessed using a cost-minimization decisional analysis [11], assuming that immunogenicity was equivalent in both cases and considering the direct cost of vaccination, boosters, and antibody serological tests over a 10 year period. The results of our prospective study [4] suggested that at 10 years, the immunogenicity of pre-exposure vaccination obtained with the alternative strategy (SCR= 0.968) was higher than that achieved with the institutional recommendations, taking into account the results of the Briggs and Schwenke study obtained over 9 years (SCR=0.730) [13]. However, the periodic serological tests and the booster vaccinations implicated in both strategies obviously modify the interpretation of these results. The alternative strategy proposes that booster vaccination be carried out at least every 3 years in cases of seroreversion at D379 + 3 years. In our prospective study, the SCR at 3 years is estimated to be 0.993 with a 95% confidence interval (CI): 0.990–0.996 [4]. The institutional recommendations specify that booster vaccination is to be performed as soon as the subject is found to be seroreverted in serological tests carried out at least every 2 years. The Briggs and Schwenke data gave an estimated SCR at 2 years of 0.991 (95% CI: 0.982–1.000). Therefore, assuming that the immunogenicity of vaccination and boosters are identical and constant in time for both strategies, the alternative strategy and the institutional recommendations are likely to be equivalently immunogenic over a 10 year period, guaranteeing a minimum SCR ranging from 0.982 to 1.000 at any time during the follow-up period.
1418
C. Strady et al. / Vaccine 19 (2001) 1416–1424
2.2. Construction of the decision trees Two decision trees were constructed and analyzed to compare the alternative strategy with the various recommendations of the WHO and the CDC, one corresponding to a continuous risk for exposure to rabies (Fig. 1), and the other corresponding to a frequent risk of exposure (Fig. 2). In the decision trees, behavior choices are represented by rectangles called decision nodes. The consequences of these choices are represented by circles called chance nodes that describe the probability of a given outcome. The branches of the tree represent all the possible outcomes and their probabilities (branches 1 – 11 of Fig. 1 and Fig. 2). The cumulated probabilities deriving from the same chance node must be equal to 1. The resulting cost at a particular chance node corresponds to the sum of the products of the cost calculated at the end of each branch originating from this node by the probability assigned to each of these branches. A simplified model was used to construct the branches of the decision tree representing institutional recommendations. According to these recommenda-
tions, a large number of titer measurements are recommended over 10 years (4–19); each measurement gives rise to two possible outcomes: titer B 0.5 IU/ml or titer B 0.5 IU/ml with immediate booster. Only one chance node is therefore shown to summarize this complex situation into two possibilities for each individual; that the individual would remain seroconverted over the 10 years (branches 6, 8 and 10) or that the patient would serorevert at some time during the 10 years (branches 7, 9 and 11). This simplification enabled the probabilities for branches concerned to be calculated and the number of boosters in the branches to be estimated from the data available. In each decision tree, the branches of the alternative strategy with no serological analysis were pruned to show, for information purposes only, the suggested frequency of booster when serological tests could not be carried out and/or when the risk of exposure is infrequent. In this setting of low exposure risk, neither the WHO nor CDC recommend antibody measurement and/or booster injections.
Fig. 1. Decision tree constructed for a continuous risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC). : decision node; : chance node; ": pruned branch showing, for information purposes, the frequency of booster vaccination without serological tests; test: serological test; P= 0.753: probability affected to the branches; US$ 155 = cost label; 1 = number of the branch.
C. Strady et al. / Vaccine 19 (2001) 1416–1424
1419
Fig. 2. Decision tree constructed for a frequent risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC); : decision node; : chance node; test: serological test; P = 0.730: probability affected to the branches; US$ 155 =cost label; 8 = number of the branch.
2.3. Data The data used to construct the decision trees were derived from the two studies published to-date with the longest follow-up of the evolution of antibody titers following pre-exposure rabies vaccination. The Briggs and Schwenke retrospective study over 9 years [13] was conducted based on the institutional recommendations. Our prospective study [4], conducted over 10 years, is the basis of our proposed alternative strategy. Both studies performed a survival analysis (the notion of seroconversion replacing the notion of survival), considering seroreversion as the event which replaced the notion of death. The immunogenicity criterion was the SCR, measured at completion of the two studies. The data of the Briggs and Schwenke study [13] that are utilized in our analysis of the institutional recommendations were derived from a subgroup of 180 subjects who received pre-exposure vaccination (D0, D7, and D21 or D28) with HDCV vaccine by the intramuscular route. This sub-group only contained those subjects who do not possess risk factors for immunosuppression. Of the total 875 subjects included in the study, it was noted that 41 had received post-exposure vaccination, although unfortunately no informa-
tion was given on how many of these (if any) were in the subgroup of 180 subjects whose data we use. The analysis of the alternative strategy is based on data from 101 subjects who were administered HDCV or PVRV intramuscularly in the deltoid as a pre-exposure vaccination series carried out on D0, D7, and D28 with scheduled booster injection on D365 [4]. No significant difference was reported between the group vaccinated with HDCV (n= 32) and the group receiving PVRV (n= 69) in terms of SCR at 10 years, enabling us to pool their data for this analysis. When using the data from these two studies, we must take account of the fact that in the alternative strategy, as in the institutional recommendations, serological tests are carried out periodically and that booster doses must be administered as soon as a subject is found to be seroreverted. These interventions had considerable effect on the estimation of the probabilities assigned to their branches and the evaluation of the number of booster doses given. The probabilities assigned to the branches deriving from the chance nodes of the decision trees are presented in Fig. 1 and Fig. 2. The probability that a subject has a titer ] 0.5 IU/ml for the institutional recommendations (branches 6, 8, and 10) corresponded
1420
C. Strady et al. / Vaccine 19 (2001) 1416–1424
to the SCR observed at 9 years of the Briggs and Schwenke study. This probability was 0.730 (95% CI: 0.638–0.822) [12]. For the alternative strategy, based on the SCR results from our 10 year study [4], the probability that a subject would have a titer ] 30 UI/ml on D379 (branch 2) was 0.753 (95% CI: 0.710 – 0.796), and the probability of a titer ] 0.5 UI/ml on D379 + 3 years (branch 4) was 0.870 (95% CI: 0.803 – 0.937). Estimates of the mean number of boosters that are needed over 10 years in subjects liable to serorevert were made from the findings of our prospective study. In this study, the subjects found to be seroreverted within the 10 year follow-up period had all seroreverted in the first 3 years following pre-exposure vaccination. These subjects then systematically seroreverted every 3 years following any booster. It was thus calculated that an average of 3 boosters per seroreverted subject would be given over the 10 year follow-up period. This estimation was used in our decision trees for the institutional recommendations in the case of subjects becoming seroreverted (branches 7, 9, and 11), and for the alternative strategy for subjects found to be seroreverted at D379 +3 years (branch 5).
2.4. Cost and sensiti6ity analysis Due to the difficulties in calculating indirect and intangible costs, and the lack of consensus of how this should be done, we limited our analysis to the direct costs engendered by vaccination (including boosters) and serological tests. Vaccinations and serological tests occurring at the end of the 10 year follow-up period were not taken into account in our cost-minimization analysis. The vaccine costs correspond to the French prices applied in 1998 by the reference institutions responsible for rabies prophylaxis. The cost of a single intramuscular vaccination was taken to be US$ 27 including the cost before tax of a vaccine dose and administrative expenses of the anti-rabies centers. Since these centers only use PVRV, a very well tolerated vaccine associated with a rare incidence of minor vaccine-related reactions, no direct medical costs associated with side-effects of vaccination were taken into account in our analysis. According to the institutional recommendations, three doses of vaccine were administered initially (on D0, D7, and D21 or D28) compared with four administrations (on D0, D7, D28, and D365) for the alternative strategy. As described above, we estimated that three booster doses would complete the 4-dose primary series of the alternative strategy or the 3-dose primary series of the institutional recommendations in the branches for subjects found to be seroreverted. The cost of the RFFIT was US$ 37, including the cost before tax of the serological test and the administrative expenses of the Centre National de Re´fe´rence
pour la Rage. We determined the number of serological tests to be carried out in each branch of the two decision trees on a logical basis depending on the strategy followed and whether a subject seroreverted. The costs used in our analysis were likely to vary according to such parameters as the different vaccination methods used, the type of cell culture vaccine administered, the laboratories in charge of serological tests, the possibility of carrying out serial serological tests with a cost reduction that could amount to 20%, or taxes. So as to be able to generalize our results, a sensitivity analysis was performed, taking upper and lower bounds of the costs, i.e. 1 US$ and 250 US$ for the administration of a vaccine dose and 1 US$ and 250 US$ for a serological test. This sensitivity analysis also took into account the confidence interval of the probabilities assigned to the branches emanating from the chance nodes of the decision trees. Furthermore, given that the costs analysis is carried out over a time horizon of many years, costs have to be discounted. Therefore they were discounted at 3, 5, and 10% in accordance with pharmaceutical economic evaluation international guidelines [14–16]. Since the pre-exposure rabies vaccination costs are not covered by French Social Security, it is most likely for persons at continuous or frequent risk of rabies exposure that those costs would be borne by their employers. Thus the cost analysis was performed from the payer’s (employer’s) perspective.
3. Results The average cost of the alternative strategy was US$ 155 per subject for 10 years (Fig. 1 and Fig. 2, branches 2, 4, and 5). Considering persons at continuous risk of exposure (Fig. 1), the cost of the alternative strategy was 5.1 times lower than that of the institutional recommendations (WHO and CDC) (US$ 798; branches 6 and 7). Likewise, for a frequent risk of exposure (Fig. 2) the cost was less than that of the institutional recommendations, although the difference was not as great: US$ 431 for the WHO recommendations (branches 8 and 9) against US$ 248 for the CDC recommendations with a serological test every 2 years (branches 10 and 11). This difference in cost was mainly attributable to the periodic and systematic serological tests carried out according to the institutional recommendations. Univariate and bivariate (vaccine costs, serological costs and probabilities) sensitivity analysis confirm the robustness of our results. All costs associated with the institutional recommendations for continuous risk of exposure were higher than those of the alternative strategy, irrespective of probability or variations in cost of vaccine or serological test (between 1 dollar and 250 dollars). The alternative strategy remains the most efficient choice.
C. Strady et al. / Vaccine 19 (2001) 1416–1424
Fig. 3 and Fig. 4 illustrate the sensitivity analysis for vaccine costs and for serological costs in the case of frequent risk exposure and demonstrates that the alternative strategy is not sensitive to changes in vaccine
Table 1 Average costs per subject in US dollars (A) with discounting at 3% (B), 5% (C), and 10% (D) at 10 years of the alternative strategy (AS), the recommendations of the World Health Organization (WHO), or those of the Centers for Disease Contol and Prevention (CDC)
Continuous risk (Fig. 1) AS (branches 2, 4, and 5) WHO and CDC (branches 6 and 7) Frequent risk (Fig. 2) AS (branches 2, 4, and 5) WHO (branches 8 and 9) CDC (branches 10 and 11)
Fig. 3. Univariate sensitivity analysis of the vaccine cost in case of frequent risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC); - -: AS; - -: WHO, serological test every year; --: CDC, serological test every 2 years.
1421
A
B
C
D
155 798
153.8 679.0
153.1 627.4
151.6 524.8
155 431 248
153.8 420.3 224.9
153.1 393.5 211.8
151.6 340.4 185.7
costs (between 1 and 250 dollars). In the presence of serological costs of less than 2.77 dollars, however, institutional strategies became dominant. However, this situation is very unlikely to occur in practice. Bivariate analyses demonstrate that only a combined analysis using vaccine costs and serological testing costs may reverse these results: in a situation of high vaccine costs (more than 250 dollars) combined with low titering costs (10 dollars) the model would come out in favor of the institutional strategies. Results of discounting costs at 3, 5, and 10% at 10 years are presented in Table 1. The changes in discounting rates had very little effect on the results obtained and therefore do not influence the choice of options.
4. Discussion
Fig. 4. Univariate sensitivity analysis of the serological cost in case of frequent risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC); Threshold values: serological cost = $2.77, CDC and AS strategy cost= $112.70; - -: AS; - -: WHO, serological test every year; --: CDC, serological test every 2 years.
Pre-exposure rabies vaccination has two perfectly complementary objectives. First, it must protect against unrecognized exposure, especially in the case of subjects whose occupations continually or frequently expose them to the virus. Second, it must anticipate exposure when isolated situations in the field do not permit post-exposure treatment to be administered under reliable conditions. In this latter respect, pre-exposure vaccination provides psychological security, enables reliable medical facilities to be reached before infection, and simplifies post-exposure treatment by cutting down the number of necessary vaccine doses and by avoiding the use of specific immunoglobulins [1,2]. Pre-exposure vaccination mainly concerns individuals who are continually or frequently at risk of exposure to rabies in the context of their professional occupations. The vaccine is rarely given to individuals likely to be exposed to the disease during their leisure time. Vaccination is not reimbursed by the French Social Security (French universal healthcare insurance system), the costs are the responsibility of public or private employers, or are paid for by subjects not at professional risk
1422
C. Strady et al. / Vaccine 19 (2001) 1416–1424
of exposure. In our cost-minimization analysis, we considered just the direct costs associated with vaccination and serological tests taken from the payer’s perspective. Contrary to post-exposure vaccination and serological tests that are exclusively carried out by the reference institutions in charge of rabies prophylaxis, there are no precise and homogenous data on the number of vaccinations performed and on the subsequent costs of the different methods of pre-exposure vaccination. This potential variability in vaccine costs was taken into account in an univariate and bivariate sensitivity analysis of the costs due to institutional recommendations and the alternative strategy. The choice of preferred strategy would only be reversed by extremely high vaccine costs (more than 250 US$) combined with low serological test costs; this is extremely unlikely to occur under current conditions. While no medical costs associated with adverse reactions to vaccination were used in our analysis, it is well known that PVRV, which was the vaccine considered in our analysis, is well tolerated with rare side-effects of minor intensity. The pharmacovigilance data covering 9 years of PVRV use (1988– 1997) show that the prevalence of the side-effects likely to induce a specific cost is 4/1 000 000 doses when side-effects are spontaneously reported and 8/10 000 doses in the case of prospective studies. Inclusion of these costs would have had minimal effect and would not have influenced choice of best strategy.
4.1. Ad6antages of the alternati6e strategy This is the first published study comparing the cost of an alternative strategy to that of the institutional recommendations for pre-exposure rabies vaccination. Since, according to the published data, the immunogenicity of the alternative strategy appears to be at least as good as that of the institutional recommendations, for subjects at continuous risk of exposure and those at frequent risk, the alternative strategy seems to be more efficient. To explain the lower cost of the alternative strategy, two particular features should be considered, the scheduled booster at 1 year, and the serological test 3 years after the booster for all those subjects with antibody titers B30 IU/ml at D379. The advantage of performing a booster vaccination scheduled at 1 year (D365) has been demonstrated [3,4]. This booster induces a significant anamnestic response with a SCR equal to 0.968 at 10 years. The institutional recommendations do not include this booster dose and the SCR only reaches 0.730 at 9 years [13]. This gain in immunogenicity enables the subsequent number of serological tests and booster vaccinations to be reduced for a great majority of vaccinated subjects (75.3%), i.e. those subjects with a titer ] 30 IU/ml on D379. The increased immunogenicity largely compensates for the additional cost of scheduled booster vaccination at 1 year in the alternative strategy.
The interest of performing a serological test at 3 years after the booster, for all those subjects with antibody titers B 30 IU/ml at D379, is that it identifies the majority, if not all, of the subjects who will serorevert over the 10 year period. Only the few subjects (3%) with titers B 0.5 UI/ml 3 years after booster will need additional serological tests and booster vaccinations. Moreover, it appears that the anamnestic response is poor in this subgroup and subjects serorevert again during the third year after any booster [4]. It is quite conceivable that such subjects may be considered unsuitable for occupations involving a continuous risk of exposure, and be excluded from such a risk group. Insofar as seroreversions always occurred during the third year after a booster, another alternative exists for those subjects whose titer is B 30 IU/ml on D379. This would involve giving a booster injection every two years without further titering. This strategy would have the advantage of providing better guarantees of maintaining seroconversion over 10 years even in low responders, at a lower cost, which can be calculated to be 154 US dollars. It would have the disadvantage of giving unnecessary boosters to the majority of subjects (87%) who, despite a titer B 30 IU/ml on D379, would have remained seroconverted (] 0.5 IU/ml) throughout the 10 year period. In addition, this alternative would not identify poor responders who should be removed from continuous risk exposure. This alternative in fact appears in the 1991 version of the CDC recommendations for all subjects at frequent exposure risk [17], but was removed from the 1999 version [2]. Our alternative strategy enables the serological profile of the vaccinated subjects to be assessed and permits immunization and serological follow-up to be conducted on a more customized and objective basis than when exclusively based on the degree of the risk of exposure. It also makes it possible to adapt the strategy to the epidemiological and medical care environment. For instance, based on the results of our prospective study [4], periodic boosters every 5 years would be advisable if serological tests cannot be carried out at a particular time due to cost restrictions or inaccessible medical facilities.
4.2. Validity and limits of the alternati6e strategy The validity of the alternative strategy essentially hinges on the notion of seroconversion and its evaluation by neutralizing antibody titer. As stated in the introduction, seroconversion (i.e. a RFFIT titer ]0.50 IU/ml) is not, however, synonymous with protection in terms of clinical efficacy and the validity of any new strategy must be considered carefully. This is, for example, the one reason why known rabies exposure of a subject who received pre-exposure vaccination must always be followed by post-exposure injections of vac-
C. Strady et al. / Vaccine 19 (2001) 1416–1424
cine. From this point of view, the validity of the alternative strategy must be discussed with particular regards on the subjects at high risk of exposure. The SCRs calculated and used in the present article are estimations based on survival analysis of data obtained from two published long-term follow-up studies [3,4,9,13]. The database used for the alternative strategy, the possible biases of which have been discussed [3], was prospective on 101 subjects, of whom 35 were lost to follow-up within 10 years and were considered censured data. Furthermore, it is of note that the cohort used in this study were unvaccinated professional workers, aged from 18 to 65 years, recruited without reference to any predefined inclusion criterion, which increases the external validity of these results. The database used for the institutional recommendations data was retrospective over 9 years [13]. There were numerous censured data with a short follow-up period (113 out of 180 at 3 years). The data available in the institutional recommendations database may have resulted in an overestimation of the SCR. First, the follow-up period was slightly lower (i.e. 9 years instead of 10 years). Second, a low but non-specified number of subjects may have been given post-exposure treatment, and would have received more vaccine injections (5) than those who received pre-exposure vaccination only (3). The results of multivariate analysis of our prospective immunogenicity study data at 10 years follow-up led us to use 30 UI/ml as the neutralizing antibody threshold in the alternative strategy [9]. This analysis showed that this 30 UI/ml threshold has a 100% positive predictive value for immunization at 10 years. Assessment of the neutralizing antibody titers bearing proof to immunization is based on the RFFIT (method of reference) [1,2,11]. This requires access to a laboratory capable of performing this evaluation following rigorous standardization. Other tests, such as the ELISA are also used. Results found with ELISA are correlated with those found using RFFIT [18]. This method, however, is less expensive and its neutralizing antibody threshold for seroconversion is different (2 UI/ml). The alternative strategy can therefore not be transposed to this test unless a specific study is performed. The alternative strategy is subject to certain limitations since our prospective long-term study which serves as a database [4,9], did not cover all of the situations that may occur with pre-exposure vaccination and did not go beyond 10 years. It seems, however, that the results of the study can be applied to other cell culture vaccines as HDCV and PVRV in so far as comparative studies did not show different immunogenicity and tolerance results in the context of pre-exposure vaccination studies conducted under similar conditions [19,20]. In contrast, these results cannot be
1423
extrapolated to pre-exposure intradermal vaccination, for which the immunogenicity, in terms of seroconversion and antibody titers, seems slightly lower than that achieved with intramuscular injections [20–24]. The immunogenicity of the alternative strategy beyond 10 years and the way to assess its effectiveness are unknown. Nevertheless it has been demonstrated that booster vaccination at 10 years induces an increase in antibody titers to antibody levels that are at least equal to those obtained on D379 [4], except in low responders. Therefore, nothing prevents serological tests from being carried out over each period of 10 years, under the same conditions as those of the first 10 years studied, and with the same consequences concerning the administration of booster doses. Finally, we confined our analysis to the direct costs of pre-exposure vaccination and serological tests, and did not take into account the cost of travel to the vaccination centre, or the cost of a possible exposure and its treatment after pre-exposure vaccination. In our opinion, the inclusion of these costs would surely have increased the total cost of the institutional recommendations strategy and resulted in a further advantage to the alternative strategy. The principally medico-economic aims of the alternate strategy should not overshadow the fact that individuals’ protection against rabies by pre-exposure immunization should be considered on an individual basis. The alternate strategy is therefore not meant to replace some or all of the institutional recommendations without careful consideration. Moreover, it actually does not resolve certain issues such as the protection of individuals who are particularly vulnerable to contamination such as nonresponders or persons at a permanent risk of exposure. This vulnerability issue may be of concern during the time which follows the three-dose primo-immunization (D0, D7 and D28) and precedes the first serology. This first antibody assessment is performed at D379, following the booster dose (D365). In our experience, however, all subjects immunized using this strategy and either the HDCV or PVRP vaccines had antibody titers ] 3.8 UI/ml at D42 (geometric mean at 37 UI/ml) and still had protective antibody titers at D365 (] 0.5 UI/ml) [3]. The issue of this vulnerability is also a concern during the time intervals which precede ulterior serological assessments, especially between D379 and Y3. The duration of this vulnerability may be longer with the alternative strategy, since the negative serostatus (B 0.5 UI/ml) of these individuals is revealed by antibody testing performed after a longer time lapse. Serological control testing in compliance with institutional recommendations should remain standard practice in all such vulnerable individuals. Nevertheless, based on a long and novel study, we hope that some of our suggestions will contribute to
1424
C. Strady et al. / Vaccine 19 (2001) 1416–1424
studies in optimizing pre-exposure anti-rabies vaccination. In the first instance, the alternative strategy should be validated externally under field conditions, comparing the decision trees results prospectively with a cohort of subjects. This approach would determine the real efficiency of the institutional recommendations and the alternative strategy, with emphasis on the evolution of the immunity of this cohort under the influence of periodic serological tests and boosters in case of seroreversion.
Acknowledgements We thank S.A. Plotkin, F. Medina and S. Wood for their thoughtful comments and contributions. Financial support: Pasteur Me´rieux Connaught.
References [1] WHO. Eighth report of the Expert Committee on Rabies, WHO Tech Rep Ser 1992;824:24–25. [2] Recommendations of the Immunization Practices Advisory Committee (ACIP), Rabies Prevention — United States 1999. Morbid. Mortal. Week. Rep. 1999;48:1–21. [3] Strady A, Lienard M, Ajjan N. Vaccination antirabique pre´ventive en milieu professionnel expose´. E´tude prospective et comparative d’immunoge´nicite´ sur 5 ans. Presse Med 1993;22:572 – 5. [4] Strady A, Lang J, Lienard M, Blondeau C, Jaussaud R, Plotkin SA. Antibody persistence using pre-exposure regimens of cellculture rabies vaccines: a 10 year follow-up and proposal for a new booster policy. J Infect Dis 1998;177:1290–5. [5] Wiktor TJ, Plotkin SA, Grella DW. Human cell culture rabies vaccine. J Am Med Assoc 1973;224:1170–1. [6] Aoki FY, Tyrrell DA, Hill LE, Turner GS. Immunogenicity and acceptability of a human diploid cell rabies vaccine in volunteers. Lancet 1975;i:660 – 2. [7] Dureux B, Canton P, Gerard A, Strady A, Deville J, Ajjan N. Rabies vaccine for human use cultivated on Vero cells. Lancet 1986;ii:98. [8] Klietmann W, Klietmann B, Cox J, Charbonnier C. Efficacy and tolerance of pre and post-exposure treatment with purified inactivated rabies vaccine prepared on Vero cell line. Vaccine 1988;6:39 – 43. [9] Strady C, Jaussaud R, Beguinot I, Lienard M, Strady A. Predictive factors for the neutralizing antibody response following pre-exposure rabies immunization: validation of a new booster dose strategy. Vaccine 2000;18(24):2661–7.
.
[10] Kuwert E, Barsenbach C, Werner J, et al. Early/high and late/low responders among HDCS vaccinees? In: Kuwert EWT, Koprowski H, editors. Cell Culture Rabies Vaccines and their Protective Effect in Man. Geneva: International Green Cross, 1981:160 – 7. [11] Nicholson KG. Modern vaccines: rabies. Lancet 1990;335:1201– 5. [12] Thornton JG, Lilford RJ, Johnson N. Decison analysis in medicine. Brit Med J 1992;304:1099 – 103. [13] Briggs DJ, Schwenke JR. Longevity of rabies antibody titer in recipients of human diploid cell rabies vaccine. Vaccine 1992;10:125 – 9. [14] Colle`ge des Economiste de la Sante´. Evaluation economique des strate´gies the´rapeutiques, JEM 1998;16:235 – 351. [15] Canadian Coordinating Office for Health Technology Assessment (CCOHTA). Guidelines for economic evaluation of pharmaceuticals: Canada. 2nd ed. Ottawa;1997. [16] Commonwealth Departement of Human Services and Health. Guidelines for the pharmaceutical industry on preparation of submissions to the pharmaceutical benefit advisory committee: including major submissions involving economic analyses. Camberra, Australian Government Publishing Service. November 1995. [17] Recommendations of the Immunization Practices Advisory Committee (ACIP), Rabies Prevention-United States 1991 Morbid. Mortal. Week. Rep. 1991;40:11 – 13. [18] Sureau P, Rollin PE, Zeller H. Correlations between the immunoenzymatic test, seroneutralization test and fluorescent focus reduction test in rabies antibody titration. Comp Immunol Microbiol Infect 1982;5:143 – 50. [19] Nicholson KG, Farrow PR, Bijok U, Barth R. Pre-exposure studies with purified chick embryo cell culture rabies vaccine and human diploid cell vaccine: serological and clinical response in man. Vaccine 1987;5:208 – 10. [20] Dreesen DW, Fishbein DB, Kemp DT, Brown J. Two year comparative trial on the immunogenicity and adverse effects of purified chick embryo cell rabies vaccine for pre-exposure immunization. Vaccine 1989;7:397 – 400. [21] Dreesen DW, Brown WJ, Kemp DT, Brown J, Reid FL, Baer GM. Pre-exposure rabies prophylaxis: efficacy of a new packaging and delivery system for intradermal administration of human diploid cell vaccine. Vaccine 1984;2:185 – 8. [22] Bernard KW, Mallonee L, Wright JC, et al. Pre-exposure immunization with intradermal human diploid cell rabies vaccine.Risks and benefits of primary and booster vaccination. J Am Med Assoc 1987;257:1059– 63. [23] Morrison AJ, Jr., Hunt EH, Atuk NO, Schwartzman JD, Wenzel RP. Rabies pre-exposure prophylaxis using intradermal human diploid cell vaccine: immunologic efficacy and cost-efficacy in a university medical center and a review of selected literature. Am J Med 1987;293:293 – 7. [24] WHO. WHO Consultation on intradermal application of human rabies vaccines, Week Epidemiol. Rec. 1995;70:336 – 337.
Pre-exposure rabies vaccination: strategies and cost-minimization study Christrophe Strady a, Van Hung Nguyen b, Roland Jaussaud a, Jean Lang b, Michel Lienard c, Alain Strady a,* a
Centre Antirabique, Hoˆpital Robert Debre´, A6enue du Ge´ne´ral Kœnig Reims Cedex, 51092, France b Pasteur Me´rieux Connaught, 58 A6enue Leclerc, Lyon Cedex 07, 69348, France c Mutualite´ Sociale Agricole, Caisses centrales, 40 Rue Jean Jaures, Bagnolet, 93547, France Received 22 December 1999; received in revised form 13 September 2000
Abstract An alternative strategy for pre-exposure rabies vaccination to the institutional recommendations of the World Health Organization and the Centers for Disease Control and Prevention is proposed based on recent long-term follow-up of post-vaccinal seroconversion rates. The alternative strategy uses the same primary series (i.e. vaccination in the deltoid area on D0, D7, and D28), but is completed by a scheduled booster vaccination at D365. The frequency of recommended subsequent booster injections depends on the serological test results obtained by a RFFIT on D379 and 3 years later. The objective of this study was to compare the efficiency of the two pre-exposure strategies. A cost-minimization analysis was carried out to compare the two rabies pre-exposure vaccination and serological test strategies based on the data from two published studies on the long-term evolution of the immunity achieved using the different recommendations. For a theoretically equivalent immunogenicity, the cost of the alternative strategy ranged from 1.7 to 5.2 times lower than that of the institutional recommendations. A sensitivity analysis confirmed the robustness of the results. The alternative strategy should be validated externally under field conditions. This approach would compare its real efficiency to the institutional recommendations. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Rabies; Pre-exposure vaccination; Cost
1. Introduction The recommendations of the World Health Organization (WHO) [1] and Advisory Committee of Immunization Practices of the Centers for Disease Control and Prevention (CDC) [2] for pre-exposure rabies vaccination include the administration of three injections of a cell culture vaccine given intramuscularly in the deltoid area on days 0, 7, and 21 or 28. Titration of neutralizing antibodies by a standardized Rapid Fluorescent Focus Inhibition Test (RFFIT) is recommended to assess the seroconversion of the subjects at high risk * Corresponding author. Present address: Centre Hospitalier Universitaire de Reims, Hoˆpital Robert Debre´, avenue du Ge´ne´ral Kœnig, 51092 Reims Cedex France. Tel.: +33-3-26787185; fax: +33-3-26784090. E-mail address: [email protected] (A. Strady).
for exposure to the virus, with a frequency adapted to that risk. A subject is considered to have seroconverted when the antibody level is ] 0.50 IU/ml [1] or with complete virus neutralization at a 1:5 serum dilution [2]. For persons at continuous risk (e.g. workers in production or research laboratories in the field of rabies), antibody levels should be assessed every 6 months. In case of frequent risk (e.g. rabies diagnostic laboratory workers, spelunkers, veterinarians and staff, and animal-control and wildlife workers in rabies-enzootic areas), the WHO and the CDC, respectively recommend antibody level assessment every 1 and 2 years. In all cases, boosters should be given as soon as the subject is found to be seroreverted (i.e. when the titer falls below 0.50 IU/ml). If the risk is considered to be infrequent (e.g. veterinarians and animal-control and wildlife workers in areas with low rabies rates, veterinary students, travelers visiting areas where rabies is enzootic
0264-410X/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 4 - 4 1 0 X ( 0 0 ) 0 0 3 6 8 - 6
C. Strady et al. / Vaccine 19 (2001) 1416–1424
and immediate access to appropriate medical care including biologics is limited), the WHO does not issue any recommendation, while the CDC indicates that there is no need for serological tests or routine booster injections. We previously conducted a prospective study on the 10 year follow-up of seroconversion rates (SCR) and antibody levels obtained after pre-exposure rabies vaccination [3,4]. In particular, our last study demonstrated the long-term immunogenicity of the 3-injection regimen (D0, D7, and D28) completed by a scheduled booster injection at 1 year (D365) [4]. The cell-culture vaccines used during this study were the human diploid cell rabies vaccine (HDCV) [5,6], used in the USA, and the vaccine prepared on Vero cells (PVRV) [7,8], used in France and in 41 other countries in which 15 million doses have been distributed. Vaccines were administered intramuscularly in the deltoid. An alternative strategy to the institutional recommendations of the WHO and the CDC as regards antibody serological tests and booster injections was suggested based on the results of this study. The proposed alternative strategy is based on antibody level assessment at times that are particularly predictive of the evolution of the SCR and antibody titers over the 10 year follow-up period as determined by a multivariate analysis of our prospective study [3,4,9]. On D379 (i.e. 14 days after booster at 1 year), a titer ] 30 IU/ml had a positive, predictive seroconversion value of 100% at 10 years while a titer B30 IU/ml had a negative, predictive seroconversion value of 18%. In other words, a subject with a titer of at least 30 IU/ml on D379 would definitely still be seroconverted after 10 years. Conversely the likelihood of seroreversion in subjects with a titer of B30 IU/ml on D379 was 18% at 10 years. In addition, if subjects seroreverted they did so between the second and third year after D379. Thus, at D379+ 3 years, 100% of subjects who had a titer ] 0.5 IU/ml were still seroconverted at 10 years. In contrast, subjects with titers B 0.5 IU/ml at this time were considered to be low responders [10], for whom subsequent serial scheduled booster vaccinations could be justified. We thus identified these two preferred times (i.e. D379 and D379 +3 years) as being particularly indicative of an individual’s serological evolution, and these timepoints were chosen for the scheduled measurements of antibodies in the alternative strategy [9]. The neutralizing antibody titer, as an efficacy criterion for pre-exposure anti-rabies vaccination is a point worth discussing. Since rabies is a 100% fatal disease, any vaccination strategy has to guarantee total protection from exposure, and the efficacy of the program can not be sacrificed in any optic of saving costs. However, the paradox is that, due to the nature of the disease, ethically it is not possible to obtain any clinical evalua-
1417
tion of the efficacy of a rabies vaccination program. The only possible criterion available is this neutralizing antibody titer. This is the consensual criterion of immunogenicity in the clinical evaluation of rabies vaccines [1,2,11]. It is this criterion which has been used in the cost minimization analysis and the immunogenicity of a strategy was expressed by the SCR observed in the cohort of subjects submitted to this strategy. Although cost-minimization studies have been used to compare the economic values of various drug regimens, they have been infrequently employed to compare vaccination strategies. The objective of this present study was to use this pharmacoeconomic approach to compare the efficiency of the two pre-exposure strategies, which differ in their pre-exposure vaccination protocol and assessment of persistence of measurable immune response through antibody serological measurements.
2. Methods
2.1. Choices and assumptions The efficiency of the two strategies was assessed using a cost-minimization decisional analysis [11], assuming that immunogenicity was equivalent in both cases and considering the direct cost of vaccination, boosters, and antibody serological tests over a 10 year period. The results of our prospective study [4] suggested that at 10 years, the immunogenicity of pre-exposure vaccination obtained with the alternative strategy (SCR= 0.968) was higher than that achieved with the institutional recommendations, taking into account the results of the Briggs and Schwenke study obtained over 9 years (SCR=0.730) [13]. However, the periodic serological tests and the booster vaccinations implicated in both strategies obviously modify the interpretation of these results. The alternative strategy proposes that booster vaccination be carried out at least every 3 years in cases of seroreversion at D379 + 3 years. In our prospective study, the SCR at 3 years is estimated to be 0.993 with a 95% confidence interval (CI): 0.990–0.996 [4]. The institutional recommendations specify that booster vaccination is to be performed as soon as the subject is found to be seroreverted in serological tests carried out at least every 2 years. The Briggs and Schwenke data gave an estimated SCR at 2 years of 0.991 (95% CI: 0.982–1.000). Therefore, assuming that the immunogenicity of vaccination and boosters are identical and constant in time for both strategies, the alternative strategy and the institutional recommendations are likely to be equivalently immunogenic over a 10 year period, guaranteeing a minimum SCR ranging from 0.982 to 1.000 at any time during the follow-up period.
1418
C. Strady et al. / Vaccine 19 (2001) 1416–1424
2.2. Construction of the decision trees Two decision trees were constructed and analyzed to compare the alternative strategy with the various recommendations of the WHO and the CDC, one corresponding to a continuous risk for exposure to rabies (Fig. 1), and the other corresponding to a frequent risk of exposure (Fig. 2). In the decision trees, behavior choices are represented by rectangles called decision nodes. The consequences of these choices are represented by circles called chance nodes that describe the probability of a given outcome. The branches of the tree represent all the possible outcomes and their probabilities (branches 1 – 11 of Fig. 1 and Fig. 2). The cumulated probabilities deriving from the same chance node must be equal to 1. The resulting cost at a particular chance node corresponds to the sum of the products of the cost calculated at the end of each branch originating from this node by the probability assigned to each of these branches. A simplified model was used to construct the branches of the decision tree representing institutional recommendations. According to these recommenda-
tions, a large number of titer measurements are recommended over 10 years (4–19); each measurement gives rise to two possible outcomes: titer B 0.5 IU/ml or titer B 0.5 IU/ml with immediate booster. Only one chance node is therefore shown to summarize this complex situation into two possibilities for each individual; that the individual would remain seroconverted over the 10 years (branches 6, 8 and 10) or that the patient would serorevert at some time during the 10 years (branches 7, 9 and 11). This simplification enabled the probabilities for branches concerned to be calculated and the number of boosters in the branches to be estimated from the data available. In each decision tree, the branches of the alternative strategy with no serological analysis were pruned to show, for information purposes only, the suggested frequency of booster when serological tests could not be carried out and/or when the risk of exposure is infrequent. In this setting of low exposure risk, neither the WHO nor CDC recommend antibody measurement and/or booster injections.
Fig. 1. Decision tree constructed for a continuous risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC). : decision node; : chance node; ": pruned branch showing, for information purposes, the frequency of booster vaccination without serological tests; test: serological test; P= 0.753: probability affected to the branches; US$ 155 = cost label; 1 = number of the branch.
C. Strady et al. / Vaccine 19 (2001) 1416–1424
1419
Fig. 2. Decision tree constructed for a frequent risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC); : decision node; : chance node; test: serological test; P = 0.730: probability affected to the branches; US$ 155 =cost label; 8 = number of the branch.
2.3. Data The data used to construct the decision trees were derived from the two studies published to-date with the longest follow-up of the evolution of antibody titers following pre-exposure rabies vaccination. The Briggs and Schwenke retrospective study over 9 years [13] was conducted based on the institutional recommendations. Our prospective study [4], conducted over 10 years, is the basis of our proposed alternative strategy. Both studies performed a survival analysis (the notion of seroconversion replacing the notion of survival), considering seroreversion as the event which replaced the notion of death. The immunogenicity criterion was the SCR, measured at completion of the two studies. The data of the Briggs and Schwenke study [13] that are utilized in our analysis of the institutional recommendations were derived from a subgroup of 180 subjects who received pre-exposure vaccination (D0, D7, and D21 or D28) with HDCV vaccine by the intramuscular route. This sub-group only contained those subjects who do not possess risk factors for immunosuppression. Of the total 875 subjects included in the study, it was noted that 41 had received post-exposure vaccination, although unfortunately no informa-
tion was given on how many of these (if any) were in the subgroup of 180 subjects whose data we use. The analysis of the alternative strategy is based on data from 101 subjects who were administered HDCV or PVRV intramuscularly in the deltoid as a pre-exposure vaccination series carried out on D0, D7, and D28 with scheduled booster injection on D365 [4]. No significant difference was reported between the group vaccinated with HDCV (n= 32) and the group receiving PVRV (n= 69) in terms of SCR at 10 years, enabling us to pool their data for this analysis. When using the data from these two studies, we must take account of the fact that in the alternative strategy, as in the institutional recommendations, serological tests are carried out periodically and that booster doses must be administered as soon as a subject is found to be seroreverted. These interventions had considerable effect on the estimation of the probabilities assigned to their branches and the evaluation of the number of booster doses given. The probabilities assigned to the branches deriving from the chance nodes of the decision trees are presented in Fig. 1 and Fig. 2. The probability that a subject has a titer ] 0.5 IU/ml for the institutional recommendations (branches 6, 8, and 10) corresponded
1420
C. Strady et al. / Vaccine 19 (2001) 1416–1424
to the SCR observed at 9 years of the Briggs and Schwenke study. This probability was 0.730 (95% CI: 0.638–0.822) [12]. For the alternative strategy, based on the SCR results from our 10 year study [4], the probability that a subject would have a titer ] 30 UI/ml on D379 (branch 2) was 0.753 (95% CI: 0.710 – 0.796), and the probability of a titer ] 0.5 UI/ml on D379 + 3 years (branch 4) was 0.870 (95% CI: 0.803 – 0.937). Estimates of the mean number of boosters that are needed over 10 years in subjects liable to serorevert were made from the findings of our prospective study. In this study, the subjects found to be seroreverted within the 10 year follow-up period had all seroreverted in the first 3 years following pre-exposure vaccination. These subjects then systematically seroreverted every 3 years following any booster. It was thus calculated that an average of 3 boosters per seroreverted subject would be given over the 10 year follow-up period. This estimation was used in our decision trees for the institutional recommendations in the case of subjects becoming seroreverted (branches 7, 9, and 11), and for the alternative strategy for subjects found to be seroreverted at D379 +3 years (branch 5).
2.4. Cost and sensiti6ity analysis Due to the difficulties in calculating indirect and intangible costs, and the lack of consensus of how this should be done, we limited our analysis to the direct costs engendered by vaccination (including boosters) and serological tests. Vaccinations and serological tests occurring at the end of the 10 year follow-up period were not taken into account in our cost-minimization analysis. The vaccine costs correspond to the French prices applied in 1998 by the reference institutions responsible for rabies prophylaxis. The cost of a single intramuscular vaccination was taken to be US$ 27 including the cost before tax of a vaccine dose and administrative expenses of the anti-rabies centers. Since these centers only use PVRV, a very well tolerated vaccine associated with a rare incidence of minor vaccine-related reactions, no direct medical costs associated with side-effects of vaccination were taken into account in our analysis. According to the institutional recommendations, three doses of vaccine were administered initially (on D0, D7, and D21 or D28) compared with four administrations (on D0, D7, D28, and D365) for the alternative strategy. As described above, we estimated that three booster doses would complete the 4-dose primary series of the alternative strategy or the 3-dose primary series of the institutional recommendations in the branches for subjects found to be seroreverted. The cost of the RFFIT was US$ 37, including the cost before tax of the serological test and the administrative expenses of the Centre National de Re´fe´rence
pour la Rage. We determined the number of serological tests to be carried out in each branch of the two decision trees on a logical basis depending on the strategy followed and whether a subject seroreverted. The costs used in our analysis were likely to vary according to such parameters as the different vaccination methods used, the type of cell culture vaccine administered, the laboratories in charge of serological tests, the possibility of carrying out serial serological tests with a cost reduction that could amount to 20%, or taxes. So as to be able to generalize our results, a sensitivity analysis was performed, taking upper and lower bounds of the costs, i.e. 1 US$ and 250 US$ for the administration of a vaccine dose and 1 US$ and 250 US$ for a serological test. This sensitivity analysis also took into account the confidence interval of the probabilities assigned to the branches emanating from the chance nodes of the decision trees. Furthermore, given that the costs analysis is carried out over a time horizon of many years, costs have to be discounted. Therefore they were discounted at 3, 5, and 10% in accordance with pharmaceutical economic evaluation international guidelines [14–16]. Since the pre-exposure rabies vaccination costs are not covered by French Social Security, it is most likely for persons at continuous or frequent risk of rabies exposure that those costs would be borne by their employers. Thus the cost analysis was performed from the payer’s (employer’s) perspective.
3. Results The average cost of the alternative strategy was US$ 155 per subject for 10 years (Fig. 1 and Fig. 2, branches 2, 4, and 5). Considering persons at continuous risk of exposure (Fig. 1), the cost of the alternative strategy was 5.1 times lower than that of the institutional recommendations (WHO and CDC) (US$ 798; branches 6 and 7). Likewise, for a frequent risk of exposure (Fig. 2) the cost was less than that of the institutional recommendations, although the difference was not as great: US$ 431 for the WHO recommendations (branches 8 and 9) against US$ 248 for the CDC recommendations with a serological test every 2 years (branches 10 and 11). This difference in cost was mainly attributable to the periodic and systematic serological tests carried out according to the institutional recommendations. Univariate and bivariate (vaccine costs, serological costs and probabilities) sensitivity analysis confirm the robustness of our results. All costs associated with the institutional recommendations for continuous risk of exposure were higher than those of the alternative strategy, irrespective of probability or variations in cost of vaccine or serological test (between 1 dollar and 250 dollars). The alternative strategy remains the most efficient choice.
C. Strady et al. / Vaccine 19 (2001) 1416–1424
Fig. 3 and Fig. 4 illustrate the sensitivity analysis for vaccine costs and for serological costs in the case of frequent risk exposure and demonstrates that the alternative strategy is not sensitive to changes in vaccine
Table 1 Average costs per subject in US dollars (A) with discounting at 3% (B), 5% (C), and 10% (D) at 10 years of the alternative strategy (AS), the recommendations of the World Health Organization (WHO), or those of the Centers for Disease Contol and Prevention (CDC)
Continuous risk (Fig. 1) AS (branches 2, 4, and 5) WHO and CDC (branches 6 and 7) Frequent risk (Fig. 2) AS (branches 2, 4, and 5) WHO (branches 8 and 9) CDC (branches 10 and 11)
Fig. 3. Univariate sensitivity analysis of the vaccine cost in case of frequent risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC); - -: AS; - -: WHO, serological test every year; --: CDC, serological test every 2 years.
1421
A
B
C
D
155 798
153.8 679.0
153.1 627.4
151.6 524.8
155 431 248
153.8 420.3 224.9
153.1 393.5 211.8
151.6 340.4 185.7
costs (between 1 and 250 dollars). In the presence of serological costs of less than 2.77 dollars, however, institutional strategies became dominant. However, this situation is very unlikely to occur in practice. Bivariate analyses demonstrate that only a combined analysis using vaccine costs and serological testing costs may reverse these results: in a situation of high vaccine costs (more than 250 dollars) combined with low titering costs (10 dollars) the model would come out in favor of the institutional strategies. Results of discounting costs at 3, 5, and 10% at 10 years are presented in Table 1. The changes in discounting rates had very little effect on the results obtained and therefore do not influence the choice of options.
4. Discussion
Fig. 4. Univariate sensitivity analysis of the serological cost in case of frequent risk of exposure based on the alternative strategy (AS), the recommendations of the World Health Organization (WHO), and those of the Centers for Disease Control and Prevention (CDC); Threshold values: serological cost = $2.77, CDC and AS strategy cost= $112.70; - -: AS; - -: WHO, serological test every year; --: CDC, serological test every 2 years.
Pre-exposure rabies vaccination has two perfectly complementary objectives. First, it must protect against unrecognized exposure, especially in the case of subjects whose occupations continually or frequently expose them to the virus. Second, it must anticipate exposure when isolated situations in the field do not permit post-exposure treatment to be administered under reliable conditions. In this latter respect, pre-exposure vaccination provides psychological security, enables reliable medical facilities to be reached before infection, and simplifies post-exposure treatment by cutting down the number of necessary vaccine doses and by avoiding the use of specific immunoglobulins [1,2]. Pre-exposure vaccination mainly concerns individuals who are continually or frequently at risk of exposure to rabies in the context of their professional occupations. The vaccine is rarely given to individuals likely to be exposed to the disease during their leisure time. Vaccination is not reimbursed by the French Social Security (French universal healthcare insurance system), the costs are the responsibility of public or private employers, or are paid for by subjects not at professional risk
1422
C. Strady et al. / Vaccine 19 (2001) 1416–1424
of exposure. In our cost-minimization analysis, we considered just the direct costs associated with vaccination and serological tests taken from the payer’s perspective. Contrary to post-exposure vaccination and serological tests that are exclusively carried out by the reference institutions in charge of rabies prophylaxis, there are no precise and homogenous data on the number of vaccinations performed and on the subsequent costs of the different methods of pre-exposure vaccination. This potential variability in vaccine costs was taken into account in an univariate and bivariate sensitivity analysis of the costs due to institutional recommendations and the alternative strategy. The choice of preferred strategy would only be reversed by extremely high vaccine costs (more than 250 US$) combined with low serological test costs; this is extremely unlikely to occur under current conditions. While no medical costs associated with adverse reactions to vaccination were used in our analysis, it is well known that PVRV, which was the vaccine considered in our analysis, is well tolerated with rare side-effects of minor intensity. The pharmacovigilance data covering 9 years of PVRV use (1988– 1997) show that the prevalence of the side-effects likely to induce a specific cost is 4/1 000 000 doses when side-effects are spontaneously reported and 8/10 000 doses in the case of prospective studies. Inclusion of these costs would have had minimal effect and would not have influenced choice of best strategy.
4.1. Ad6antages of the alternati6e strategy This is the first published study comparing the cost of an alternative strategy to that of the institutional recommendations for pre-exposure rabies vaccination. Since, according to the published data, the immunogenicity of the alternative strategy appears to be at least as good as that of the institutional recommendations, for subjects at continuous risk of exposure and those at frequent risk, the alternative strategy seems to be more efficient. To explain the lower cost of the alternative strategy, two particular features should be considered, the scheduled booster at 1 year, and the serological test 3 years after the booster for all those subjects with antibody titers B30 IU/ml at D379. The advantage of performing a booster vaccination scheduled at 1 year (D365) has been demonstrated [3,4]. This booster induces a significant anamnestic response with a SCR equal to 0.968 at 10 years. The institutional recommendations do not include this booster dose and the SCR only reaches 0.730 at 9 years [13]. This gain in immunogenicity enables the subsequent number of serological tests and booster vaccinations to be reduced for a great majority of vaccinated subjects (75.3%), i.e. those subjects with a titer ] 30 IU/ml on D379. The increased immunogenicity largely compensates for the additional cost of scheduled booster vaccination at 1 year in the alternative strategy.
The interest of performing a serological test at 3 years after the booster, for all those subjects with antibody titers B 30 IU/ml at D379, is that it identifies the majority, if not all, of the subjects who will serorevert over the 10 year period. Only the few subjects (3%) with titers B 0.5 UI/ml 3 years after booster will need additional serological tests and booster vaccinations. Moreover, it appears that the anamnestic response is poor in this subgroup and subjects serorevert again during the third year after any booster [4]. It is quite conceivable that such subjects may be considered unsuitable for occupations involving a continuous risk of exposure, and be excluded from such a risk group. Insofar as seroreversions always occurred during the third year after a booster, another alternative exists for those subjects whose titer is B 30 IU/ml on D379. This would involve giving a booster injection every two years without further titering. This strategy would have the advantage of providing better guarantees of maintaining seroconversion over 10 years even in low responders, at a lower cost, which can be calculated to be 154 US dollars. It would have the disadvantage of giving unnecessary boosters to the majority of subjects (87%) who, despite a titer B 30 IU/ml on D379, would have remained seroconverted (] 0.5 IU/ml) throughout the 10 year period. In addition, this alternative would not identify poor responders who should be removed from continuous risk exposure. This alternative in fact appears in the 1991 version of the CDC recommendations for all subjects at frequent exposure risk [17], but was removed from the 1999 version [2]. Our alternative strategy enables the serological profile of the vaccinated subjects to be assessed and permits immunization and serological follow-up to be conducted on a more customized and objective basis than when exclusively based on the degree of the risk of exposure. It also makes it possible to adapt the strategy to the epidemiological and medical care environment. For instance, based on the results of our prospective study [4], periodic boosters every 5 years would be advisable if serological tests cannot be carried out at a particular time due to cost restrictions or inaccessible medical facilities.
4.2. Validity and limits of the alternati6e strategy The validity of the alternative strategy essentially hinges on the notion of seroconversion and its evaluation by neutralizing antibody titer. As stated in the introduction, seroconversion (i.e. a RFFIT titer ]0.50 IU/ml) is not, however, synonymous with protection in terms of clinical efficacy and the validity of any new strategy must be considered carefully. This is, for example, the one reason why known rabies exposure of a subject who received pre-exposure vaccination must always be followed by post-exposure injections of vac-
C. Strady et al. / Vaccine 19 (2001) 1416–1424
cine. From this point of view, the validity of the alternative strategy must be discussed with particular regards on the subjects at high risk of exposure. The SCRs calculated and used in the present article are estimations based on survival analysis of data obtained from two published long-term follow-up studies [3,4,9,13]. The database used for the alternative strategy, the possible biases of which have been discussed [3], was prospective on 101 subjects, of whom 35 were lost to follow-up within 10 years and were considered censured data. Furthermore, it is of note that the cohort used in this study were unvaccinated professional workers, aged from 18 to 65 years, recruited without reference to any predefined inclusion criterion, which increases the external validity of these results. The database used for the institutional recommendations data was retrospective over 9 years [13]. There were numerous censured data with a short follow-up period (113 out of 180 at 3 years). The data available in the institutional recommendations database may have resulted in an overestimation of the SCR. First, the follow-up period was slightly lower (i.e. 9 years instead of 10 years). Second, a low but non-specified number of subjects may have been given post-exposure treatment, and would have received more vaccine injections (5) than those who received pre-exposure vaccination only (3). The results of multivariate analysis of our prospective immunogenicity study data at 10 years follow-up led us to use 30 UI/ml as the neutralizing antibody threshold in the alternative strategy [9]. This analysis showed that this 30 UI/ml threshold has a 100% positive predictive value for immunization at 10 years. Assessment of the neutralizing antibody titers bearing proof to immunization is based on the RFFIT (method of reference) [1,2,11]. This requires access to a laboratory capable of performing this evaluation following rigorous standardization. Other tests, such as the ELISA are also used. Results found with ELISA are correlated with those found using RFFIT [18]. This method, however, is less expensive and its neutralizing antibody threshold for seroconversion is different (2 UI/ml). The alternative strategy can therefore not be transposed to this test unless a specific study is performed. The alternative strategy is subject to certain limitations since our prospective long-term study which serves as a database [4,9], did not cover all of the situations that may occur with pre-exposure vaccination and did not go beyond 10 years. It seems, however, that the results of the study can be applied to other cell culture vaccines as HDCV and PVRV in so far as comparative studies did not show different immunogenicity and tolerance results in the context of pre-exposure vaccination studies conducted under similar conditions [19,20]. In contrast, these results cannot be
1423
extrapolated to pre-exposure intradermal vaccination, for which the immunogenicity, in terms of seroconversion and antibody titers, seems slightly lower than that achieved with intramuscular injections [20–24]. The immunogenicity of the alternative strategy beyond 10 years and the way to assess its effectiveness are unknown. Nevertheless it has been demonstrated that booster vaccination at 10 years induces an increase in antibody titers to antibody levels that are at least equal to those obtained on D379 [4], except in low responders. Therefore, nothing prevents serological tests from being carried out over each period of 10 years, under the same conditions as those of the first 10 years studied, and with the same consequences concerning the administration of booster doses. Finally, we confined our analysis to the direct costs of pre-exposure vaccination and serological tests, and did not take into account the cost of travel to the vaccination centre, or the cost of a possible exposure and its treatment after pre-exposure vaccination. In our opinion, the inclusion of these costs would surely have increased the total cost of the institutional recommendations strategy and resulted in a further advantage to the alternative strategy. The principally medico-economic aims of the alternate strategy should not overshadow the fact that individuals’ protection against rabies by pre-exposure immunization should be considered on an individual basis. The alternate strategy is therefore not meant to replace some or all of the institutional recommendations without careful consideration. Moreover, it actually does not resolve certain issues such as the protection of individuals who are particularly vulnerable to contamination such as nonresponders or persons at a permanent risk of exposure. This vulnerability issue may be of concern during the time which follows the three-dose primo-immunization (D0, D7 and D28) and precedes the first serology. This first antibody assessment is performed at D379, following the booster dose (D365). In our experience, however, all subjects immunized using this strategy and either the HDCV or PVRP vaccines had antibody titers ] 3.8 UI/ml at D42 (geometric mean at 37 UI/ml) and still had protective antibody titers at D365 (] 0.5 UI/ml) [3]. The issue of this vulnerability is also a concern during the time intervals which precede ulterior serological assessments, especially between D379 and Y3. The duration of this vulnerability may be longer with the alternative strategy, since the negative serostatus (B 0.5 UI/ml) of these individuals is revealed by antibody testing performed after a longer time lapse. Serological control testing in compliance with institutional recommendations should remain standard practice in all such vulnerable individuals. Nevertheless, based on a long and novel study, we hope that some of our suggestions will contribute to
1424
C. Strady et al. / Vaccine 19 (2001) 1416–1424
studies in optimizing pre-exposure anti-rabies vaccination. In the first instance, the alternative strategy should be validated externally under field conditions, comparing the decision trees results prospectively with a cohort of subjects. This approach would determine the real efficiency of the institutional recommendations and the alternative strategy, with emphasis on the evolution of the immunity of this cohort under the influence of periodic serological tests and boosters in case of seroreversion.
Acknowledgements We thank S.A. Plotkin, F. Medina and S. Wood for their thoughtful comments and contributions. Financial support: Pasteur Me´rieux Connaught.
References [1] WHO. Eighth report of the Expert Committee on Rabies, WHO Tech Rep Ser 1992;824:24–25. [2] Recommendations of the Immunization Practices Advisory Committee (ACIP), Rabies Prevention — United States 1999. Morbid. Mortal. Week. Rep. 1999;48:1–21. [3] Strady A, Lienard M, Ajjan N. Vaccination antirabique pre´ventive en milieu professionnel expose´. E´tude prospective et comparative d’immunoge´nicite´ sur 5 ans. Presse Med 1993;22:572 – 5. [4] Strady A, Lang J, Lienard M, Blondeau C, Jaussaud R, Plotkin SA. Antibody persistence using pre-exposure regimens of cellculture rabies vaccines: a 10 year follow-up and proposal for a new booster policy. J Infect Dis 1998;177:1290–5. [5] Wiktor TJ, Plotkin SA, Grella DW. Human cell culture rabies vaccine. J Am Med Assoc 1973;224:1170–1. [6] Aoki FY, Tyrrell DA, Hill LE, Turner GS. Immunogenicity and acceptability of a human diploid cell rabies vaccine in volunteers. Lancet 1975;i:660 – 2. [7] Dureux B, Canton P, Gerard A, Strady A, Deville J, Ajjan N. Rabies vaccine for human use cultivated on Vero cells. Lancet 1986;ii:98. [8] Klietmann W, Klietmann B, Cox J, Charbonnier C. Efficacy and tolerance of pre and post-exposure treatment with purified inactivated rabies vaccine prepared on Vero cell line. Vaccine 1988;6:39 – 43. [9] Strady C, Jaussaud R, Beguinot I, Lienard M, Strady A. Predictive factors for the neutralizing antibody response following pre-exposure rabies immunization: validation of a new booster dose strategy. Vaccine 2000;18(24):2661–7.
.
[10] Kuwert E, Barsenbach C, Werner J, et al. Early/high and late/low responders among HDCS vaccinees? In: Kuwert EWT, Koprowski H, editors. Cell Culture Rabies Vaccines and their Protective Effect in Man. Geneva: International Green Cross, 1981:160 – 7. [11] Nicholson KG. Modern vaccines: rabies. Lancet 1990;335:1201– 5. [12] Thornton JG, Lilford RJ, Johnson N. Decison analysis in medicine. Brit Med J 1992;304:1099 – 103. [13] Briggs DJ, Schwenke JR. Longevity of rabies antibody titer in recipients of human diploid cell rabies vaccine. Vaccine 1992;10:125 – 9. [14] Colle`ge des Economiste de la Sante´. Evaluation economique des strate´gies the´rapeutiques, JEM 1998;16:235 – 351. [15] Canadian Coordinating Office for Health Technology Assessment (CCOHTA). Guidelines for economic evaluation of pharmaceuticals: Canada. 2nd ed. Ottawa;1997. [16] Commonwealth Departement of Human Services and Health. Guidelines for the pharmaceutical industry on preparation of submissions to the pharmaceutical benefit advisory committee: including major submissions involving economic analyses. Camberra, Australian Government Publishing Service. November 1995. [17] Recommendations of the Immunization Practices Advisory Committee (ACIP), Rabies Prevention-United States 1991 Morbid. Mortal. Week. Rep. 1991;40:11 – 13. [18] Sureau P, Rollin PE, Zeller H. Correlations between the immunoenzymatic test, seroneutralization test and fluorescent focus reduction test in rabies antibody titration. Comp Immunol Microbiol Infect 1982;5:143 – 50. [19] Nicholson KG, Farrow PR, Bijok U, Barth R. Pre-exposure studies with purified chick embryo cell culture rabies vaccine and human diploid cell vaccine: serological and clinical response in man. Vaccine 1987;5:208 – 10. [20] Dreesen DW, Fishbein DB, Kemp DT, Brown J. Two year comparative trial on the immunogenicity and adverse effects of purified chick embryo cell rabies vaccine for pre-exposure immunization. Vaccine 1989;7:397 – 400. [21] Dreesen DW, Brown WJ, Kemp DT, Brown J, Reid FL, Baer GM. Pre-exposure rabies prophylaxis: efficacy of a new packaging and delivery system for intradermal administration of human diploid cell vaccine. Vaccine 1984;2:185 – 8. [22] Bernard KW, Mallonee L, Wright JC, et al. Pre-exposure immunization with intradermal human diploid cell rabies vaccine.Risks and benefits of primary and booster vaccination. J Am Med Assoc 1987;257:1059– 63. [23] Morrison AJ, Jr., Hunt EH, Atuk NO, Schwartzman JD, Wenzel RP. Rabies pre-exposure prophylaxis using intradermal human diploid cell vaccine: immunologic efficacy and cost-efficacy in a university medical center and a review of selected literature. Am J Med 1987;293:293 – 7. [24] WHO. WHO Consultation on intradermal application of human rabies vaccines, Week Epidemiol. Rec. 1995;70:336 – 337.