Drugs and Money - Prices, Affordability and Cost Containment
(2003; 158 pages) View the PDF document
Table of Contents
View the documentIntroduction
Close this folderPart I: Problems and approaches to a solution
View the documentChapter 1: Scope of the problem
View the documentChapter 2: Data needed for developing and monitoring policies
View the documentChapter 3: Policy options for cost containment of pharmaceuticals
View the documentChapter 4: Methods for monitoring and evaluating processes and outcomes
View the documentChapter 5: Making use of economic evaluation
Open this folder and view contentsPart II: Selected experiences with policy options
View the documentList of Contributors
View the documentBack cover

Chapter 2: Data needed for developing and monitoring policies

Elias Mossialos and Monique F. Mrazek

To develop and monitor any aspect of drug policies - which naturally include cost containment - one needs to collect reliable and valid data on processes and outcomes. This chapter focuses on the data needed to detect and evaluate the impact of a drug policy on the different elements of drug management and delivery. Important variables relate to the various aspects of prescribing, dispensing and consumption, but also to the ultimate consequences in terms of health and finance. The types of data discussed in this chapter therefore include facts and figures on pharmaceutical expenditure, utilization, price, health and economics outcomes, as well as data on the pharmaceutical industry.

Since drug expenditure is determined largely by price, patient need and prescribing choice, data on expenditure can only be realistically monitored by taking these elements into consideration. The data needed in order to develop and monitor drug policies must therefore relate to a whole series of issues and actors. One needs to consider how these data can be obtained and examine some methodological problems that can arise in collecting and using them.

Monitoring alone will not explain the trend in drug expenditures, nor will it answer the question as to whether the level of expenditure is appropriate to meet reasonable goals in terms of health. It is therefore necessary to look beyond expenditure data and examine the changes in underlying trends in patient needs, prescribing choice and pricing. If we are to understand what factors create and modify a trend in drug expenditure, we shall need to look both at drug utilization and at prices. Expenditure data alone cannot, for example, differentiate a country with high unit consumption but low drug prices from one with high prices and lower levels of drug consumption.

1. Drug expenditure data

Monitoring pharmaceutical expenditure over a period of time makes it possible to determine the overall effect of cost containment policies (or other concurrent influences) on the size and growth of drug spending. Expenditure data can be expressed as a proportion of overall health care costs, as a percentage of national economic output (Gross Domestic Product = GDP) or as average per capita spending. When comparing expenditure on drugs as a function of total health expenditure, or as a per capita figure ratio, one finds wide variations between countries (Table 1). These relative expenditure levels and trends are used to make cross-national comparisons so as to understand better the relationship between structure and performance in different drug financing systems. It is particularly important to consider drug expenditure data within the context of overall health care financing. Pharmaceutical services, like health care services generally, may be financed from public or private sources or from a combination of the two. Comparing trends in public versus private (including the over-the-counter market) drug expenditures gives an indication of the way in which the burden of drug financing is apportioned between individuals, society generally and different levels of government. Such differences in financing arrangements between countries should be taken into consideration when comparing trends; an apparent success in reducing public drug expenditure in a particular country is not necessarily praiseworthy if it proves to have placed an intolerable burden on individuals.

Table 1
Pharmaceutical expenditure in selected OECD countries, 1997

Total expenditure on pharmaceuticals and other medical non-durables


% Total healthcare expenditure

Per capita, US$ exchange rate





















Czech Rep.




















United States
















United Kingdom








New Zealand









































indicates 1996 data.

The consumption of pharmaceutical goods comprises both prescription medicines and self-medication products, often referred to as over-the-counter (OTC) medicines. The series includes the pharmacists’ remuneration when the latter is separate from the price of medicines. Pharmaceuticals consumed in hospitals are excluded. The expenditure includes VAT and sales taxes where applicable. The amount of consumption in hospitals is included under in-patient care.

Source: [16].

It is also useful for policy development to break down drug expenditure within the pharmaceutical service into its various components. Significant variations in drug spending between different groups within the service will sometimes (but not always) be attributable to population characteristics (services directed at caring for the elderly patients or the chronically ill will for example be expected to spend proportionally more on drugs). Again, different policy initiatives may have been directed to drug spending in in-patient care (hospitals) as compared with out-patient care (general practice physicians). Differences in the level of spending and rate of growth in each of these fields may be directly attributable to such specific policy initiatives [7] though one will always need to be alert to other concurrent influences.

1.1. Sources of data on expenditure

Ministries of Health or other government institutions in general report annually on national drug expenditure. Such sources, as well as those covering health expenditure generally, usually provide good coverage of the overall trend in public pharmaceutical spending. It is however often difficult to obtain complete data on drug expenditure which include (and differentiate between) in-patient and out-patient drug costs, prescription drugs and OTC, branded drugs and generic medicines, and spending in non-reimbursed and private pharmaceutical markets. National pharmaceutical industry associations often republish government data on expenditure in their annual reports along with other helpful figures. Commercial publications, such as Scrip and the Financial Times, occasionally publish national expenditure data.

One set of figures commonly used in making cross-national comparisons of drug expenditure in Europe is the OECD Health Database. The data generally cover both private and public pharmaceutical expenditure. However, as there is no cross-national agreement on statistical methods, the OECD data cannot be considered a standardised health statistic. The database is compiled from official government statistics. This limits its external comparability as variations between countries arise in terms of what is included in the calculations and how the categories are specified. Some countries are treated as special cases by the database because of variations in the characteristics of the population covered by the available statistics. In addition, limitations to the data arise because not all the information is up-to-date. At best, publication of annual expenditure figures usually takes a year. Figures from countries that have joined OECD only recently have not been fully integrated into the database. The quality and reliability of the data may also vary depending on what component of the data are being examined. For example, public expenditure data may be more accurate than information on private or out-of-pocket spending.

1.2. Methodological issues

The first methodological issue to consider when using expenditure data is how the figures have been calculated. Drug expenditure is dependent on both drug consumption and price. It is important to know whether expenditure figures include the consumption of both in-patient and out-patient prescription medicines, and whether they also include OTCs. Drug expenditure figures will also vary depending on where in the drug distribution chain the drug price and consumption volumes for the calculation were obtained; retail and wholesale prices are naturally higher than the manufacturer’s price because of the intermediate mark-ups.

Three methods are commonly used to calculate drug expenditure. The first method derives drug expenditures from the totality of receipts of pharmaceutical manufacturers or wholesalers (excluding exports). Secondly, pharmaceutical expenditure can also be estimated from a sample of prescriptions dispensed by pharmacies, provided we know the average price of a prescription. Finally, pharmaceutical expenditure can be estimated from the sales of prescription drugs by retail pharmacies to the public.

When comparing data over time, or between countries, it is important to make sure that they are measured in the same way. For example, data should be measured at the same point and time in the distribution chain, should comprise the same segments in terms of hospitals and out-patient care, prescription and self-medication, and should cover either total, public, private or out-patient expenditures. Again, when monitoring pharmaceutical expenditures over time, or between countries, the figures used should be real (constant-price) and not nominal (current-price). Studies of “real” pharmaceutical expenditure from year to year involve taking the costs for a baseline year and adjusting these for the subsequent effects of inflation on drug spending before making any comparisons with later years; this provides a true picture of the manner in which the cost burden on the community is changing. “Nominal” comparisons simply set the data for the baseline year alongside that for the years which follow, and therefore reflect both the effect of inflation and real changes in costs.

When comparing drug expenditure between countries a common measure needs to be used. One approach is to select a common currency unit. The problem with using a common currency unit is that bias may be introduced due to exchange rate fluctuations. What is more, exchange rates do not reflect the relative purchasing power between countries because they do not take account of the equalisation of prices of non-marketed commodities such as health care [12].

A similar degree of caution is needed when expressing pharmaceutical expenditure per capita using a constant dollar exchange rate. Here, although measuring pharmaceutical expenditure per capita provides a relative indication of consumption by adjusting for population size, the measure may be confounded by a number of factors including the purchasing power of the currency, and differences in the price at which a given product is sold in one country or another. Nor does this method necessarily provide a valid estimate of the changes in the burden borne by the individual as patients may have different combinations of private and public coverage, pay different amounts out-of-pocket, or may be exempt from all charges.

To eliminate differences in price levels between countries, pharmaceutical expenditure can be converted to a common currency using rates of currency conversion called purchasing power of parities (PPPs). PPPs convert currencies in such a way that the purchasing power in different countries is equalised to purchase the same basket of goods and services in all countries. The resultant comparisons provide a better picture of the real differences in the quantities of goods and services purchased. This is generally the recommended approach for cross-national comparisons. However, in using PPPs caution should be exercised as to their calculation: weak points are that the consumption functions for PPPs are in general only calculated every 5 years; what is more, health service costs are inevitably measured on the basis of a very small sample of prices and on weakly comparable volume indices [12]. Therefore, when using PPPs for international comparisons, not only may the composition of the baskets differ between each country particularly for health care, but in addition different values may be attached to the components of the basket in each country.

Pharmaceutical expenditure can be also expressed as a percentage of GDP, defined in the OECD Health database as the total domestic expenditure plus exports and less imports of goods and services. Monitoring pharmaceutical expenditure as a percentage of GDP provides a relative indication of how much a country is spending on pharmaceuticals as compared to its economic status. However, using GDP as a denominator introduces a bias because, as the figure is a ratio, there is a risk of confounding changes in pharmaceutical expenditure with fluctuations in economic growth. Again, GDP expressed in national currency units indicates only how much of the economy is spent on pharmaceutical goods, but says nothing about the purchasing power of a country.

Other common units of measure include drug spending as a percentage of total health expenditure; this gives a relative indication of resource use within the health sector. When examining this measure over time, it is important to distinguish rises and falls in health expenditure generally from rises and falls in pharmaceutical expenditure; the two may not run parallel.

2. Drug utilisation data

Drug utilisation research throws light on the medical, social and economic consequences of the marketing, distribution, prescribing and use of medicines [20]. Utilisation data are both quantitative and qualitative. Patterns of drug utilisation can be used to determine how variations in need, prescribing choice and price account for differences in drug use in different countries [10].

The consumption of prescription medicines in a given population is closely related to a number of factors: its demographic structure (notably because older people use more drugs), to the incidence and prevalence of disease, to its socio-economic structure, to the nature and extent of health care coverage, medical culture [17], marketing by the pharmaceutical industry, as well as prescribing and dispensing incentives and regulations. By examining population characteristics, epidemiological data, and figures on the frequency and severity of illness one can arrive at an estimate of drug consumption, and therefore what allowance should be made for drug expenditure in health planning. However, it is important to note that the calculated level of reasonable need for prescription medicines will not necessarily correspond to the actual level of demand. If a patient does not seek medical intervention, demand will be less than need, and the resources used will be less than those estimated in advance. Conversely, if people buy drugs for which there is no reasonable need, whether under the influence of commercial persuasion or any other factor, demand will exceed need.

2.1. Sources of utilisation data

In many situations, valid and reliable sources of utilisation data are not readily available. Sources of drug utilisation data include commercial and administrative databases, but only certain of these are designed in such a way that they capture the data needed for drug utilisation studies.

Data from commercial survey organisations such as Intercontinental Medical Statistics (IMS)1 rely on sales figures as indicators of drug use. Figures are collected from manufacturers’ and importers’ records providing for a relatively complete coverage of the market. IMS also provides data based on samples of prescriptions from panels of pharmacists and physicians in each country. Analyses of sales data of this type have however primarily been designed for commercial purposes; various variables that are essential for assessing the appropriateness of drug use in a population are missing.

1 IMS is an organization that collects data on pharmaceuticals, chemicals and other healthcare matters in sixty-two countries. IMS has a near monopoly in the collection and distribution of international pharmaceutical sales figures, which are sold commercially to the pharmaceutical industry.

Administrative databases collect information on users of a particular system or sets of systems in order to meet the needs of health service providers. Typical examples of administrative databases include those compiled by bodies dealing with prescription payments or reimbursement schemes. These are increasingly complemented by databases that link diagnostic data with prescribing data (such as the General Practice Research Database in the UK, for more info on the database see http://www.gprd.com). By using such sources one can obtain extensive and detailed information on prescribing, dispensing, administration and the management of drug treatment in the public health service. One will find detailed figures on the level of use of particular drugs and classes of drugs, the quantities and doses prescribed, the price paid, and as a rule some information on the type of patient or indication for which the treatment has been given. It is also possible to detect trends to switch from one drug to another within a given therapeutic category, such as may happen where a new compound has been marketed or where the recommendations in a formulary have been changed; as noted above, these can be matters of considerable financial significance.

There are, however, major caveats associated with the use of these large computerised administrative databases [18]. Data compiled on the basis of public insurance or reimbursement schemes often do not collect data on drug consumption funded by private insurance or by the individual. Problems may arise in the use of these large databases because of non-randomly collected comparison groups; they may also present problems because of incomplete or inconsistent reporting, under-coding or coding errors. As a result studies using these data must be critically analysed and cautiously reported.

2.2. Methodological issues

The reliability of comparisons made using drug utilisation data is dependent upon the use of a common drug classification scheme and unit of measurement. The WHO Regional Office for Europe has, since 1981, recommended the ATC/DDD system for international drug utilisation studies [19]. The Anatomical Therapeutic Chemical (ATC) system classifies drugs at five different levels of detail and provides for a unique identifier (Table 2); using this one can deal with a drug as an individual item, with all drugs having the same field of use, or with the entire therapeutic class to which they all belong. The Defined Daily Dose (DDD), which is the unit of measurement of drug use, is the assumed average dose per day for each active ingredient when it is used in its main indication for adults (it is based on an adult dose with the exception of preparations that are exclusive to children). This average dose selected for the DDD system is based on recommendations in the literature, the manufacturer’s advice and experience with the product. The DDD does not necessarily reflect the recommended or Prescribed Daily Dose (PDD) for any individual patient or patient group, but using it as a utilization unit does provide the best means available for expressing consumption levels, particularly when making comparisons between prescribers, regions or countries.

Measuring the volume and the costs of drug use in terms of DDDs solves the problem of allowing for differences in prescription, duration of treatment or the potency of individual preparations. For example, it is possible to compare diabetes treatments between two centres, even though one may be using primarily (injectable) insulin with its potency measured in units and the other an oral antidiabetic drug dosed in milligrams; both types of drug have their own Defined Daily Doses and the number of DDD’s used can be compared directly. It would not be possible to make that same comparison if the two drugs were to be measured in terms of common physical units (the number of packages, tablets, or injections) or the number of prescriptions written.

Table 2
Drug classification system. Ex ample: Classification of Diazepam






Central nervous system


Central nervous system











Benzodiazepine derivatives





Source: [2].

Consumption within a geographic region, expressed as DDDs/1000 inhabitants/day provides a rough estimate of how many patients receive a standard dose per day. Alternatively one can express annual consumption levels in an entire country in terms of DDD per 1000 inhabitants per year; Lecomte and Paris [13] used this measure to identify differences in overall pharmaceutical consumption in Germany, France, Italy and UK. Others have chosen to express their data in terms of DDD per average inhabitant per year, while when considering levels of in-patient drug use, DDDs prescribed per 100 bed days is a helpful measure. Provided one ensures that the same measures are used throughout a study, one can choose the form of DDD measurement that is most convenient.

Even when using the DDD, some caution may be needed in drawing conclusions regarding the data that emerge. A number of studies have for example found variations in the quantity and cost of prescribing between individual prescribers from different geographic areas (see review by Bradley [1]), and it is all too easy to assume that some of these physicians are over-prescribing or under-prescribing; in fact one will need to know more about the patient populations involved (e.g., their age and socio-economic breakdown) before jumping to such a conclusion. This particular issue of differences in use between age and socio-economic groups needs to be tackled critically. A crude means of standardising the populations concerned by giving weights to older versus younger patients has been proposed but it is not reliable. A more accurate adjustment for demographic differences may involve taking into account proven differences in drug needs between both the various age groups and the sexes [14]. The most important element to bear in mind however is that even such adjustments are only approximate, and that one should never draw conclusions as to the rationality of prescribing patterns from crude drug utilisation data.

3. Data on pharmaceutical prices

Drug prices differ, sometimes very substantially, between countries, even within the same economic area. Data on pharmaceutical prices can be compared both nationally and internationally. They serve both as a measure of the effect of different regulatory policies on the cost of pharmaceuticals, and as a reference for setting national drug prices. Here again however one has to be sure that the figures are reliable and that the comparisons made are valid.

The reliability of the absolute figures is generally not in doubt but one must ensure that they are measured at the same level and in a comparable manner (i.e. price may be expressed at the retail, wholesale or ex-manufacturers level, it may relate to smaller or to larger packages and in some cases a company may set one price for the private sector and another for a public health care system). Nor is it not unknown for the price charged at a particular level to be influenced downwards by a form of subsidy or upward by the imposition of taxes or duties.

The validity of any comparison will depend on one’s prices differ from one country to another. The manufacturer’s determinant, and differences in price may in that respect reflect overcharging in some countries. However, government and insurance policies also play a role. The aim in negotiation may be to secure prices which are the very lowest attainable, or there may be some willingness to allow leeway for adequate input into research and development or to ensure sufficient income for retail pharmacists or dispensing physicians.

3.1. Sources of data on pharmaceutical prices

Data on drug prices from country to country often prove surprisingly difficult to obtain, though some of the figures can be extracted laboriously from sources such as National Formularies and reimbursement tables. Detailed pricing data are not readily available from government sources for use by researchers or other outside agencies, though in general, governments do publish average price data. In 1988, recognising the limited availability of drug price data, the European Community attempted to compile a data bank of drug prices and other basic product information from Member States. The data bank was intended to interface with a variety of national databases. That initiative led to the development of the European Community Pharmaceutical Information Network (ECPHIN), set up by the European Commission’s Joint Research Centre. ECPHIN was medicinal products authorised by the European Union including price and reimbursement rates. This proved too ambitious, and by 2001 plans for ECPHIN had been abandoned.

Commercial database organisations such as IMS (see Section 2.1 above) collect detailed drug pricing data. Their information includes prices at ex-manufacturer, wholesale and retail price levels. IMS does, however, acknowledge that price data are audited and adjusted to represent the whole market. It is difficult to verify the reliability and accuracy of this database since it is one of the few sources of detailed drug price data.

3.2. Methodological issues

Any attempt to express the overall price level of drugs in a particular market must be based on figures for a sample of products. The validity of any overall conclusion or comparison regarding price levels is obviously dependent on how this “basket” of products is constituted. To be representative of the market as a whole, the basket should include a random selection of brand name, generic and OTC products. If international comparisons are being made, the samples for all the markets being compared should ideally be matched in terms of manufacturer, active ingredient, dosage form, strength, pack size and brand name. A standard unit of measurement and classification such as the ATC/DDD system described in the previous section should be employed.

Constructing a truly ideal product basket as a basis for international comparisons is unfortunately often impossible. The range of products available varies from one country to another. Even if the same product is available it may not be made by the same manufacturer in all the countries which are being compared, or it may not everwhere be supplied in the same pack size, dosage form or strength of active ingredient. It is also important to consider how well the products selected represent national consumption patterns in the countries concerned.

As noted earlier, it is also essential that comparisons be made between prices at the same level of distribution chain. As medicines move along the distribution chain from the factory, to the wholesaler, the pharmacist and finally to the consumer the price of the drug will rise to reflect the value added along the way. The consumer price of a pharmaceutical is generally composed of four parts: an ex-factory price paid to the manufacturer of the products; a wholesaler’s margin, a fixed tariff or margin for pharmacists; and whatever taxes may be applicable. Variation in the consumer price of medicines in between countries in Europe is attributable to differences in the ex-factory price (ranging from 87.5 per cent of the consumer price in the UK to 49.9 per cent in Greece), the wholesaler’s in the Netherlands but only 3.2 per cent in Sweden), the pharmacist’s no more than 5 per cent in the UK) and the value added tax (zero in Sweden, UK, Austria, Ireland but 20.3 per cent in Greece) [3]. These data illustrate how marked these variations from country to country are, even within this closely integrated economic area.

A further complication when one makes comparisons is that in particular situations the prices charged may differ from the norm. It has already been noted that a company may charge different prices in the public and private sectors. In addition special discounts may be offered to bulk drug purchasers such as hospitals and retail pharmacy chains, and such discounts may or may not be passed on to the ultimate payer. Where these practices are widespread they can significantly affect the actual figures, both as regards true margins and consumer prices.

As with comparisons of expenditure data, when comparing drug prices between countries the position of any one country relative to others is likely to depend in part on how the comparison is made. Again one approach is to compare the actual prices paid in each country, transposing them into a common currency using current bank exchange rates. Such comparisons using a common currency unit are subject to complications of exchange rate fluctuations. They have the additional disadvantage that they provide no indication of the extent of the burden which these prices represent for the communities or individual concerned. For example, even a high absolute price may be relatively affordable in a country where earnings are similarly high. For this reason it can be more informative to compare price levels per country in terms of the actual PPPs.

When such studies are conducted over a period of time, it may be possible to detect trends in prices and to compare these trends from one country to another, again using a standard “basket” of representative products and applying a price index to it. For each country the index should measure the changes in the expenditure required to obtain the standard range and quantity of the drugs in the sample [4]. Using indexes alone to make comparisons between countries avoids the complications introduced by exchange rate conversions, and the local levels of inflation do not distort the comparisons. Consequently, it may be possible to conclude for example, that the burden of drug costs has risen less rapidly in one country than another. Again one must be aware that such a comparison requires that the pharmaceutical basket remains representative as time goes by, and this is not easy to ensure. Similarly, when constructing indexes it is difficult to take account of the effect of new drug introductions, the sometimes rapid change in the mix of drugs actually being prescribed, and developments in drug quality [11].

4. Health outcomes data

Data on health outcomes is required to develop, monitor and evaluate pharmaceutical policy. The types of health outcomes data commonly collected includes mortality and morbidity data, as well as data on health related quality of life (HRQL). Morbidity data can include both measures of actual and perceived disability.

HRQL is in general measured using multi-dimensional health status instruments, either psychometric or utility/preference instruments. Psychometric health status instruments measure health status along multiple domains and are either disease-specific (e.g., Skindex for skin diseases) or generic measures (e.g., the Sickness Impact Profile). Alternatively, preference weighting can be assigned to health states that reflect individual and population preferences for different health states. Quality weights can be assigned to health states using several methods: the rating scale, the standard gamble, and the time trade-off (TTO) are the most common techniques. Each health state is then combined with a time score in order to determine the number of Quality-adjusted-life-years (QALYs).

Deciding on the health outcomes to be measured depends on several things: potential differences in patient populations related to the main effects of the intervention; any side-effects or unintended consequences; and outcomes of interest depending on perspective taken (e.g., patient, third-party payers, society) [9, p. 84]. There may also be differences in the selection of health outcomes depending on whether our interest is clinical or economic. For example, health outcomes that are needed for economic evaluations may be final outcomes (e.g., rapidly of cure), while those of clinical interest may be intermediate outcomes such as changes in blood pressure. It may also be important to collect data on the frequency or probability of given outcomes.

4.1. Sources and methodological considerations of outcomes data

Outcome data can be routinely monitored and collected as indicators of policy performance. If outcome data are to be used as part of an evaluation it has been recommended that the primary data be collected from randomly controlled trials (RCTs) [6], but this represents an ideal which is not always attainable. Although the most unbiased evidence on outcomes comes from RCTs, these may lack precise data on some of the clinical end-points that are relevant to a given situation because they were designed to answer clinical questions rather than economic ones (e.g., efficiency may be a very different matter from effectiveness when conducting economic evaluations). RCTs clearly have a high degree of internal validity but it may not be possible to generalize from them to real world settings. It may simply prove impossible to collect from an RCT the data which are needed for monitoring and evaluating health outcomes associated with policy changes.

Alternatively prospective observational and descriptive studies can be used to generate data on outcomes. Both observational cohort and case-control studies can generate probability data of particular outcomes associated with an intervention; however, both types of observational studies are more prone to bias than RCTs (e.g., where patients are not randomly assigned, or questions of recall bias arise in case-control studies) [15, pp. 146-147]. Health outcome data can also be generated from administrative databases but the same caveats discussed earlier in this chapter apply.

5. Cost data on programmes or treatments

There are different ways of considering cost. In its basic accounting form cost equals the number of resources used multiplied by the unit cost. Costing in this way requires that all resources used by a particular programme or treatment be identified and valued. However, to economists “cost refers to the sacrifice (of benefits) made when a given resource is consumed in a programme or treatment” or in other words the opportunity cost [5, p. 54]. The value of opportunity forgone in the next-best alternative use of the resources do not necessarily equate to the market prices of the resources used (e.g., patient time). The total costs therefore comprise the sum of all expenditures or opportunity costs during a given time frame. It is difficult to make adjustments to costs that reflect the opportunity cost of the resource used; for that reason the pragmatic approach often adopted is simply to use the market prices for these resources.

Costs which may be measured should cover the direct costs incurred by the health care provider and (or) the patient (i.e. costs of hospitalisation, physician services, pharmaceuticals, etc.). Indirect costs to society of the productivity lost (e.g., due to the patient seeking care or costs resulting from disability or premature death). Data collected may also include intangible costs borne by patients in terms of pain and suffering; their inclusion is subject to debate [8, p. 189]. There is an ongoing debate on how to measure direct costs (i.e. marginal, variable, fixed, average, capital and shared costs), whether to included indirect costs and what the best way may be to measure opportunity cost. (See Drummond et al., 1997, Chapter 4.) There is also debate over how to measure unrelated future costs [9, pp. 45-48]. The implications of these debates are discussed further in Section 2, Chapter 5.

The purpose and type of analysis undertaken will determine the categories for which cost data will need to be collected. For example, hospital costs may involve identifying costs of services, facilities and overheads, while community costs may involve costs associated with visits to GPs, nurses or other health professionals. From a third party perspective one may want to consider actual charges rather than costs because often third parties do not cover total health care costs. As regards patients it is important to place a value both on their time and that of their family, as well as to take into account any out-of-pocket expenses incurred. In such matters, however, it can be important to assess the relative importance of a cost item to the overall outcome since the cost of including some minor costs may not be justified by their significance in the total picture.

There is also a decision to be made about the precision of the cost data to be collected. Micro costing includes each component of resources used and a unit cost for each. Costs can also be collected according to the case-mix group (i.e. the type of case category or patient). Per diem costs can also be collected as an average of all categories of patients or for each disease category.

5.1. Sources and methodological considerations of cost data

The collection of cost data includes not only the collection the prices of the resources used but also the quantity of resources used. As with health outcome data, cost data can also be collected from Randomised Controlled Trials. The problem with obtaining costs from these RCTs is that, as noted above, they lack external validity and instead of reflecting costs associated with regular patient management and resource use, the costs obtained from RCTs may be protocol-driven.

However, in many cases cost data will be collected either from routine sources (such as administrative databases) or by acquiring specific primary data collection (i.e. in observational studies). Both data on prices and costs are often based on accounting costs or service prices. It is important to be aware of level data collection as for example, the retail price of medicines is often different than that obtained in a hospital setting. Cost data can be estimated from chart reviews, administrative databases or reviews of other hospital and medical records. Cost estimates for non-market items can be estimated using market wage rates (i.e. for volunteer time); for items which are difficult to value (such the leusure time enjoyed by the patient or his family) the estimate may be set at zero and then adjusted using sensitivity analysis (see below) to examine impact.

Alternatively cost data can be collected from other published literature. This is also useful for identifying key costs. The problem with these is that they can vary considerably between different settings. Practice variation in for example facilities or services used, nursing time etc. can account for a number of differences from place to place. Variability also arises in estimating future costs. To overcome such variability sensitivity analysis is used. Sensitivity analysis is the process of repeatedly using different values for probability and utility values in order to assess the degree of uncertainty associated with a result. Inferences are also made from other studies.

It may be necessary to ask patients directly for information (e.g., number and length of time of home visits by health professionals). Missing data can be estimated using meta-analysis to combine results from other studies. Alternatively assumptions can be based on expert opinion and then tested using sensitivity analysis. If data is collected from different sources then it will be important to use simulation models to combine the data and take account of the variation.

In collecting data it is important to define the time horizon (i.e. short-run constrained by fixed resources versus long run where all inputs are variable). Resources may change over time and data may be limited to certain time horizons. Often data is limited to the trial period and estimating future costs requires either calculating survival within a given period of the trial (usually set at 5 years or less) and then either discounting costs during the time of the trial or extrapolating events and costs beyond the period of the trial [8, p. 191].

6. Information relating to the pharmaceutical industry

Up to this point we have considered solely the data required to examine the efficient use of pharmaceutical resources within the health care system. It is however, also important to monitor and take into account data on the pharmaceutical industry, which in many countries forms a significant part of the manufacturing sector and as such contributes to a country’s and maintenance of a domestic pharmaceutical industry can have a very positive effect on a country’s trade surplus, particularly in terms of foreign earnings and balance of trade. The interest of the health sector in regulating pharmaceutical expenditure, prices and profits may therefore have to be balanced judiciously against national economic interests. The latter will comprise the need to provide sufficient incentives and opportunities to preserve and develop an industry with a solid financial base, capable of effective innovation and sustained growth in the long-term.

Data are therefore required to calculate the contribution of this industry to the economy as a whole. This will include monitoring the industry’s employed. It is also important to monitor the investment of the pharmaceutical industry in R&D projects both in its own laboratories and externally at universities and clinics, or through other joint ventures such as those with biotech companies. Successful innovation can be monitored through the level of patenting and the success in commercialising innovative technologies. Finally, data on the environment in which the pharmaceutical industry is operating can be relevant; they will include figures and information on the availability of resources, the presence of related and supporting industries, and the availability of skilled labour. All these various forms of information will contribute to a view on the achievements and prospects of the national pharmaceutical industry, and the extent to which its interests should counterbalance those of containing pharmaceutical expenditure.

One of the best sources of data on the national pharmaceutical industry is usually the national industry association. National associations of pharmaceutical manufacturers publish in their annual reports numerous facts and figures which are intended to promote the industry, but which can be valuable in assessing its situation. Commercial databases such as Datastream or those compiled by the IMS contain company-specific financial and commercial data. Annual reports from individual companies are also useful. Other industry specific information available in publications such as the Panorama of EU Industry published by the Eurostat, the statistical office of the European communities, and reports issued by the European and International Federations of Pharmaceutical Manufacturers’ Associations. However, with all of these aforementioned data sources, there is really no way to verify its validity.

7. Conclusion

This chapter has focused on the types of data needed to develop and monitor drug policies. Whether using data on drug expenditure, utilisation, price, health outcomes or on the pharmaceutical industry there are a number of common issues which should be taken into consideration particularly when making regional or international comparisons. It is important to be on the lookout for those characteristics of the pharmaceutical scene that are inevitably unique to each country, and to take them into account so that they do not invalidate comparisons. As pointed out in Section 3 above, for example, comparisons should be made using a common unit of measurement, matched samples and data from the same point in the drug distribution chain. In the past, too, bias has been introduced into some comparisons because of the limited availability and dissemination of good quality data on prices, volumes and outcomes outcome data. It is important that these and other possible sources of error raised in this chapter be addressed if data are to be obtained and used effectively to monitor and inform future policy development.


[1] C.P. Bradley, Decision making and prescribing patterns - a literature review, Family Practice 8(3) (1991), 276-287.

[2] D. Capella, Descriptive tools and analysis, in: Drug Utilization Studies: Methods and Uses, M.N.G. Dukes, ed., World Health Organization, Copenhagen, 1993.

[3] Danish Association of the Pharmaceutical Industry, Facts and Figures, Danish Association of the Pharmaceutical Industry (LIF), Copenhagen, 2001.

[4] P.M. Danzon and J.D. Kim, International price comparisons for pharmaceuticals: Measurement and policy issues, Pharmacoeconomics 14(Suppl. 1) (1998), 115-128.

[5] M.F. Drummond, B. O’Brian, Methods for the Economic Evaluation of Health Care G.L. Stoddart and G.W. Torrance, Programmes, 2nd edn, Oxford Medical Publications, Oxford, 1997.

[6] M.F. Drummond, Experimental versus observational data in the economic evaluation of pharmaceuticals, Medical Decision Making 18(2) (Suppl. 1) (1998), S12-S18.

[7] A. Earl-Slater and C. Bradley, The Inexorable rise in the UK NHS drugs bill: Recent policies, future prospects, Public Administration 74 (1996), 393-411.

[8] A.M. Garber, Advances in CE analysis, in: Handbook of Health Economics, J.P. Newhouse and A.J. Culyer, eds, Vol. 1A, North-Holland, New York, 2000.

[9] M.R. Gold et al., Identifying and valuing outcomes, in: Cost-Effectiveness in Health and Medicine, M.R. Gold, J.E. Siegel, L.B. Russell and M.C. Weinstein, eds, Oxford University Press, Oxford, 1996.

[10] F.M. Haaijer-Ruskamp and L.T.W. de Jong-van den Berg, Drug utilisation studies and drug monitoring in the Netherlands, Annali Dell Instituto Superiore Di Sanita 27 (1991), 217-223.

[11] B. Jonsson, Pricing and reimbursement of pharmaceuticals in Sweden, Pharmacoeconomics 6(Suppl. 1) (1994), 51-60.

[12] P. Kanavos and E. Mossialos, International comparisons of health care expenditures: What we know and what we do not know, Journal of Health Services Research and Policy 4(2) (1999), 122-126.

[13] T. Lecomte and V. Paris, Consommation de pharmacie en Europe, 1992: Allemagne, France, Italie, Royaume-Uni, No 1048, Paris, Credes, 1994.

[14] D.C.E.F. Lloyd, C.M. Harris and D.J. Roberts, Specific therapeutic group age-sex related prescribing units (STAR-PUs): weightings for analysing general practices’ prescribing in England, British Medical Journal 311 (1995), 911–994.

[15] J.S. Mandelblatt et al., Assessing the effectiveness of health interventions, in: Cost-Effectiveness in Health and Medicine, M.R. Gold, J.E. Siegel, L.B. Russell and M.C. Weinstein, eds, Oxford University Press, Oxford, 1996.

[16] OECD, OECD Health Data 2000, OECD, Paris, 2000.

[17] L. Payer, Medicine and Culture, Henry Holt, New York, 1996.

[18] W.A. Ray, Policy and program analysis using administrative databases, Annals of Internal Medicine 127(8 Pt 2) (1997), 712-718.

[19] World Health Organization, Guidelines for ATC Classification and DDD Assignment, 2nd edn, WHO Collaborating Centre for Drug Statistic Methodology, Oslo, 1998.

[20] WHO Expert Committee, The Selection of Essential Drugs, Technical Report Series No. 615, World Health Organization, Geneva, from 1977, continuing.


to previous section
to next section
The WHO Essential Medicines and Health Products Information Portal was designed and is maintained by Human Info NGO. Last updated: December 6, 2017