M. J. POSTMA
Groningen University Institute for Drug Exploration/Groningen Research Institute of Pharmacy, Groningen, the Netherlands
L. G. WIESSING
European Monitoring Centre for Drugs and Drug Addiction, Lisbon, Portugal
J. C. JAGER
National Institute of Public Health and the Environment, Bilthoven, the Netherlands
The objective of the research described in the present article was to estimate the costs of hepatitis C virus (HCV), hepatitis B virus (HBV) and human immunodeficiency virus (HIV) infection by intravenous drug use in countries of the European Union.
The pharmaco-economic model applied linked incidence to lifetime cost estimates (incidence-based model). Estimates are presented within a broader framework of social costs.
In the baseline, estimated costs of HCV, HBV and HIV infection related to drug addiction amount to 1.89 euros ( € ). In sensitivity analysis, costs up to € 2.57 are estimated (0.5 per cent of total European Union expenditures on health care).
A preliminary estimate for the costs of HCV, HBV and HIV infection related to drug addiction in the European Union is provided. Those costs reflect a relevant share of total costs related to drug addiction.
Pharmaco-economics covers the research field of allocating scarce resources to drugs and drug-related activities [1, 2]. The primary field of pharmaco-economical applications covers the medical applications of drugs, with relevant areas such as costing studies and cost-benefit analyses of pharmaceuticals. The methodology can however be directly applied to “unhealthy” applications of drugs as well. The scope of the present study is limited to frequent smoking and injecting drugs such as opiates, benzodiazepines, amphetamines and cocaine, referred to here in brief as “drug addiction”.
Pharmaco-economics has been applied to drug addiction previously in only a few studies. For example, Kim and colleagues analysed costs and benefits of drug prevention programmes for the United States of America over the period 1979-1992 . They estimated a favourable cost/benefit ratio: there is a saving in direct and indirect costs of $15 for each $1 spent in drug addiction prevention. Rice and colleagues estimated the costs of drug addiction-related illnesses within a broader social costing framework . That framework included both the indirect costs of production losses and the costs of crime.
As regards drug addiction, the pharmaco-economic concept of social costs provides a very suitable approach. Social costs reflect the total burden of costs for society. The societal perspective is much wider than the health-care perspective, which is limited to direct health-care costs. Since it includes indirect costs, the societal perspective is often the viewpoint that policy makers are most interested in, as such evaluations assist in deciding on the allocation of resources between different sectors and generic methods developed allow valid comparisons to be made between very different programmes . One field enjoying a growing interest involves the pharmaco-economics of prevention of infectious diseases. For example, in the Netherlands, investigations into the pharmaco-economics of hepatitis B virus (HBV) vaccination and human immunodeficiency virus (HIV) screening programmes are explicitly required by the Ministry of Health before implementation of a programme . Pharmaco-economic evaluation of drug addiction is closely related to prevention of infectious diseases, as transmission of blood-borne diseases is among the most serious health and economic consequences of injecting drug use. The authors of the present article consider the costs of drug-related acquisition of hepatitis C virus (HCV), HBV and HIV in member States of the European Union (EU).
A methodology derived from pharmaco-economic theory to classify the costs related to drug addiction and to provide partial estimates is presented below. The classification draws on general economic principles of costing and the estimation procedure has previously been applied to costs related to HIV/acquired immune deficiency syndrome (HIV/AIDS) at the EU level . The estimation is based on the direct medical costs of HIV/AIDS and hepatitis epidemics among drug users in EU. The preliminary estimate provided only covers part of all social costs for diseases related to drug addition. Indirect costs of the infectious diseases and the costs of other (non-infectious) diseases are not considered here. Furthermore, costs of addiction treatment, criminal acts and low performance in jobs and indirect costs of mortality due to overdoses are also left for further research. However, a crude estimate is given here for the costing of infectious diseases in EU through drug addiction that has not previously been attempted.
The authors' pharmaco-economic model was originally applied to 10 EU countries (Ireland and Luxembourg were left out because of lack of data). In that model, epidemiological information (incidence of infections) and economics (lifetime costs per infection) were linked to estimate the direct medical costs of drug-related HBV, HCV and HIV.
Traditionally, pharmaco-economic research uses two basic methods to express the impact of drugs and disease, known as incidence-based and prevalence-based methods. The first involves accounting lifetime costs to the year of incidence of disease, while the second method involves a breakdown of those lifetime costs over the relevant stages of disease into stage-specific prevalence estimates with a view to, for example, annual costing estimates for budgetary purposes . To cover the full pharmaco-economic impact of drug addiction, it is necessary to analyse the lifetime cost consequences incurred by a representative incident of infection of a drug user. For direct medical costs, this incidence-based approach does however assume that future health-care technologies will remain the same as in the analysis. In addition, all the other assumptions are extended up to the time when the last incident of infection of drug users has been dealt with.
If the assumption of everything staying equal holds true for the period of analysis and the full pharmaco-economic impact is what is being referred to in a decision context, the incidence-based approach is the more appropriate one. This could be the case if the decision context, for example, refers to investments and allocating budgets to prevent problems related to drug addiction. Therefore results using the incidence-based approach are reported here.
Epidemiological information refers to the mid-1990s (the most recent published information available). HIV incidence among drug users in EU is estimated by back calculation. Estimates correspond to a range from approximately 250 per million population in Spain and Portugal to less than 10 per million in Germany, Greece, the Netherlands and the United Kingdom of Great Britain and Northern Ireland .
Incidence of drug-related hepatitis was estimated using a review of European prevalence studies for HCV, HBV and HIV in specific groups of drug users (see table 1) . Excess prevalence rates for HCV and HBV were applied over the HIV prevalence rate to HIV incidence (for example, if the HBV antibody is twice as prevalent as that of HIV in a sample of Spanish drug users, Spanish HBV incidence was assumed to be twice that of Spanish HIV).a
Table 1. Prevalence rates of infections with hepatitis C virus, hepatitis B virus and human immunodeficiency virus infection applied in the pharmaco-economic model1
Prevalence rate (percentage)
|Hepatitis C virus||Hepatitis B virus||
|United Kingdom of Great Britain and Northern Ireland||38||19||0.6|
|1 Adapted from L. G. Wiessing and others, “Impact and control of AIDS, HIV, and hepatitis B and C among injecting drug users in Europe: an exploratory overview”, Inaugural Meeting Report by National Institute on Drug Abuse, National Institutes of Health, United States Department of Health and Human Services, Geneva, 25 and 26 June 1998.|
Hospital bed needs arising in relation to HIV/AIDS infection vary over the different stages of the disease. Furthermore, hospital costs and the costs of pharmaceuticals are the major components of health-care costs to combat HIV/AIDS. In particular, the terminal phase of AIDS is accompanied by a relatively high level of hospital bed needs. In this regard, two stages were previously differentiated for AIDS: a final stage of maximally six months before death (the late stage (LS)) and a stage for the foregoing period (the chronic stage (CS)). Application of the stage concept to France, Greece, Italy, the Netherlands, Spain and the United Kingdom has shown that per person-year inpatient days in LS are at least twice those in CS. Inpatient day needs in symptomatic pre-AIDS stages equal 20 per cent of those in CS. Information on inpatient days was available for the six countries mentioned .
To allow for recent advances in pharmaco-therapy, the previously published progression-of-disease model was extended with a stage reflecting the life-years gained on highly active anti-retroviral therapy (HAART), with hospital inpatient days needs similar to those in the symptomatic pre-AIDS stages . To summarize the full progression-of-disease model for HIV: on average six years of asymptomatic HIV infection, one extra year gained on HAART, six years of symptomatic pre-AIDS disease and two years of AIDS (CS and LS) [7, 10].
Costs for one year on HAART were assumed to total 10,000 euros ( € ) for the package of pharmaceuticals (for example, zidovudine, lamivudine and ritonavir). Country-specific costs per acute care day were used, varying from € 420 (United Kingdom) to € 600 (France).b Costs per acute care day include average outpatient costs. Outpatient costs were corrected for by lowering costs per acute care day by 30 per cent in the baseline (based on Netherlands data ) and 10 per cent in sensitivity analysis. Costs per outpatient contact were assumed to be 25 per cent of corrected unit costs per acute care inpatient day .
Monetary costs were discounted for time preference at 4 per cent , 1 per cent and 7 per cent  in sensitivity analysis. Finally hospital costs for HIV/AIDS were assumed to reflect 90 per cent of total costs, excluding the costs of HAART . As detailed information on health care for HIV was only available for six countries (France, Greece, Italy, the Netherlands, Spain and the United Kingdom), for the other countries the cost figures of neighbouring countries were applied (that is, the Netherlands estimate for Belgium, Denmark and Germany; the Spanish figure for Portugal).
Only crude lifetime cost estimates for incidence of HCV and HBV infection could be used as very little information was available [14-17]. That information suggests that average lifetime costs per HCV infection are in the order of magnitude of 10-20 per cent of lifetime HIV costs and that lifetime HBV costs are 10 per cent of lifetime HCV costs [18, 19]. In the baseline it was assumed that lifetime HCV costs were 15 per cent of lifetime HIV costs; in two alternative scenarios the lower and upper limits of 10 and 20 per cent, respectively, were investigated.
Figure I shows estimated lifetime costs of HIV infection for the six countries for which detailed information was available. These varied from € 42,500 for the United Kingdom to € 90,800 for France.b
|Figure I. Estimated lifetime costs of human
immunodeficiency virus infection in six member States of the European
In the baseline, estimated costs of drug-related HCV, HBV and HIV infection amount to € 1.89 billion for EU as a whole. HCV infection accounts for approximately 40 per cent of those costs (see figure II), HIV infection for the major proportion of the costs. Figure III indicates the distribution over the 10 countries considered: Spain and Italy are estimated to account for a major proportion of total EU costs (approximately € 1.4 billion). Both countries show relatively large shares for drug-related HIV infection costs. Furthermore, results indicate that in the United Kingdom and Germany costs arise primarily from hepatitis.
|Figure II. Distribution of estimated drug-related costs
for hepatitis C virus, hepatitis B virus and human immunodeficiency virus
infection for the European Union as a whole, excluding Ireland and
Figure III. Costs of hepatitis C virus, hepatitis B virus and human immunodeficiency virus infection for 10 member States of the European Union a,b
a White = HCV, Grey = HBV, Black = HIV
b These country-specific results can only be considered indicative; differences between countries depend to a crucial extent on the assumptions derived from the European prevalence studies (see table 1)
Figure IV. Sensitivity analysis for total costs of drug-related hepatitis C virus, hepatitis B virus and human immunodeficiency virus infection for the European Union as a whole, excluding Ireland and Luxembourg a
a Baseline, variation in the discount rate (4 per cent in the baseline), variation in the level of lifetime HCV costs as a percentage of lifetime HIV costs and correction for outpatient costs in costs of acute care day by 10 per cent (20 per cent in the baseline; see text)
Figure IV shows the results of one-way sensitivity analysis. Variation in the discount rate induces the smallest changes in total costs. Estimated costs range from € 1,630 to € 2,159 (minus 14 per cent and plus 16 per cent of the baseline, respectively). Multi-way sensitivity analysis indicates estimated costs up to € 2,570 mil lion (HCV costs at 20 per cent of HIV costs, 10 per cent lowering of acute care costs for outpatient costs and 1 per cent discount rate).
The types of estimate described above may serve several purposes. They can, for example, be used in budget analysis. Estimates of the costs of drug addiction are typically related to health-care budgets, at the national and EU levels. In 1995, the estimated costs of drug-related HCV, HBV and HIV infection as a percentage of expenditure on health in 1995 were as follows (in euros at purchasing power parities for gross domestic product):c
| United Kingdom of Great Britain
and Northern Ireland
Relatively high shares are estimated for the southern European countries Italy, Spain and Portugal, with significant impacts of HIV. As for northern European countries, Germany and the United Kingdom have relatively high shares, estimated to be caused primarily by HCV. The baseline estimate for drug-related costs of HCV, HBV and HIV amounts to approximately 0.4 per cent of expenditure on health in EU as a whole (up to 0.5 per cent in sensitivity analysis).
Table 2. Distribution of estimated social costs of drug addiction inCanada  and France , various cost categories (Percentage)
|Direct health care1||6||12|
Costs of HIV/AIDS and Subutex treatment only.
2 Comprises costs for Governments and costs of prevention, research and law enforcement.
3 Indirect costs for absence due to sickness and for mortality, using the human-capital method of enumeration.
Two other studies provide a framework to compare with the estimate presented here (table 2) [20, 21]. Both studies distinguish between direct health-care costs, other direct costs such as government spending and indirect costs of production losses. The table illustrates that the authors' estimate of the most relevant share of direct health-care costs related to HCV, HBV and HIV constitutes a significant contribution to a total estimate of social costs. However, large shares also remain open to further research. In particular, further research should consider indirect costs at the EU level. In that respect the distinction between the human-capital method and friction-costing method for indirect costing is relevant . The latter is favoured in some European countries and significantly reduces indirect costs related to mortality compared with human-capital costing. Application of the friction-costing concept would raise the share of direct health-care costs in the study by Single and colleagues  from the current 6 per cent in the table to 10-12 per cent.
The present authors' results depend to a crucial extent on the epidemiological information used in the model. Estimated incidences reflect the situation in the mid-1990s and were based on modelling (HIV) and relative prevalences of HCV and HBV in samples. For example, given that in the United Kingdom samples considered HCV is approximately 100 times more prevalent than HIV, an estimated HCV incidence for the United Kingdom of more than 100 times HIV incidence results. The epidemiological information used is the only information available at the moment, but the representativity of the samples for the national situations are subject to question. Next to, for example, the setting of the sample, one criterion of representativity concerns the size of the samples: sample sizes differ from 140 in Greece to 67,000 in Italy (the United Kingdom is in between, with approximately 2,000 drug users in treatment centres).
The incidence-based approach has been described in detail here. Cost estimates derived with the incidence-based approach are the most suitable estimates for use in cost-effectiveness analysis.  If, for example, the aim is to evaluate the cost-effectiveness of needle exchange or methadone programmes with respect to infections averted, lifetime costs related to one incident case of infection represent the most appropriate type of information. If, on the other hand, the budget impact of drug addiction is the focus of analysis and if that impact is expected to change with new technologies, the prevalence-based approach—applied over a well-defined period of analysis—is more appropriate to support decisions from the pharmaco-economic point of view. However, with respect to the long-term social costing or full pharmaco-economic impact, incidence- and prevalence-based approaches give the same result in the end.
In conclusion, a preliminary cost estimate for incidence-based drug addiction- related costs for HCV, HBV and HIV infection in EU member States has been provided. Further development of the model will be directed to designing a pre valence-based approach and estimating other categories of social costs.
*The authors acknowledge the support of Fernando Antoñanzas (University of La Rioja, Logroño, Spain), Ardine de Wit (National Institute of Public Health and the Environment, Bilthoven, the Netherlands) and Eric Single (University of Toronto, Canada).
aThis straightforward application of prevalence relationships to incidence certainly neglects differences in hepatitis and HIV mortality rates, potentially disturbing similar patterns in prevalence and incidence. HIV prevalence estimates were available for all countries considered. For Portugal, in the absence of an estimate for HBV antibody prevalence, an assumption of 50 per cent of HCV antibody prevalence was made, as this crudely reflects the pattern of HCV/HBV prevalence in the other countries considered.
bBased on purchasing power parities for gross domestic product as estimated by the Organisation for Economic Cooperation and Development . For the two countries with missing cost data, Greece and Italy, e 500 was assumed.
cSource: Organisation for Economic Cooperation and Development, Health Data 99 (Paris, 1999).