Health Articles

S101980100071.qxd

2001 · 2:142–149 Springer-Verlag 2001
J. Rovira1 · R.Tremosa1 · A. Gilabert2 · M.Torralba2
1 Grup de Recerca en Economia de la Politica Social,Universitat de Barcelona,Spain
2 Servei Català de la Salut,Unitat de Planificació Farmacèutica,Generalitat de Catalunya,Spain
The role of prices in drug
expenditure analysis
An application of price indexes for two therapeutic
groups in the Catalan Health Service (1991–1999)
The increase in pharmaceutical expen- sumption towards presentations or
diture is one of the main concerns of products of a different price.
The objective of this article is to develop and health managers, given both its relative The first factor, a change in the price apply several types of price indexes to the importance in total expenditure and the of the products, is not that relevant in analysis of pharmaceutical expenditure to difficulty in controlling its growth. Ex- Spain and, thus, cannot explain the in- improve the way traditional indexes adjust penditure on personnel or equipment is crease in pharmaceutical expenditure in for innovation in drug supply.The Laspeyres easier to control, as the availability of re- recent years. It should be borne in mind and Paasche indexes in the fixed and linked sources for making that expenditure is that all drugs financed by the Spanish modalities were used. Price is defined as av- subject to administrative procedures National Health System are subject to a erage expenditure at consumer price per that require prior authorization. The use price control system. This means that daily defined dose (DDD).The empirical ap- of a drug depends on a long list of deci- the price of the drug at the time of its plication is restricted to two therapeutic sions (registration, price fixing, inclu- launch, and any subsequent change, groups, antipsychotics and antidepressants, sion in public financing, inclusion in for- must be authorized by the health au- and to the pharmaceutical expenditure of mularies of specific institutions, consid- thorities. Revision of the price of a single the Catalan Health Service in the period eration in clinical guidelines, etc.).
drug hardly every happens. Changes are 1991–1999.The indexes can be computed However, once those decisions have been usually applied generally to the whole from the information currently recorded in made, supply is very elastic, in other market. General increases in authorized the administrative procedures of health care words the laboratories will supply at a prices have been infrequent and, in any insurers, or from sample information provid- given price, the maximum authorized case, small, and have not even kept pace ed by regular market surveys.The proposed price for example, practically any quan- with inflation. In some cases, general method allows for controlling the effect of tity that doctors decide to prescribe.
price reductions have even occurred.
several factors on average pharmaceutical The increase and, in general, the This pricing policy has led the phar- price changes, specifically, the introduction variation in pharmaceutical expenditure maceutical industry to develop certain of new drugs that show some degree of on a product are a result of the simulta- strategies to maintain or increase their neous variation in two factors: the quan- income and profitability. First, encour- tity of units sold and the price of that aging consumption, to increase the product. When the expenditure relates number of units sold. Second, to by-pass to a heterogeneous group of products, as the control of authorized prices, the Pharmaceutical expenditure · Prices · when the total pharmaceutical expendi- manufacturers opted for the introduc- Laspeyres indexes · Paasche indexes ture or that of a therapeutic group are tion and promotion of new products, being considered, the variation is thevariation in units sold multiplied by thevariation in the average price of those Joan Rovira
units.As regards average price, its varia- The World Bank, HDNHE, 1818 H Street NW, tion may result from: (a) a change in the Washington DC 20433, USA, price of existing drugs, (b) a shift in con- 142 HEPAC 4•2001

HEPAC
2001 · 2:142–149 Springer-Verlag 2001
J. Rovira · R.Tremosa · A. Gilabert
that are strongly promoted to make c. between different active sub- consumers switch from the older ones stances, albeit therapeutically to the newer and more expensive prod- equivalent, of a chemical family El papel de los precios en el análisis del
ucts. This constitutes an indirect mech- (or therapeutic subgroup); gasto en medicamentos. Una aplicación
anism for achieving higher sale prices.
d. between different therapeutic de índices de precios para dos grupos
As a result , the market is characterized terapéuticos en la sanidad catalana
by a very high innovation rate (which of- ten contributes little or no additional Decision makers are usually interested therapeutic value over products already in determining the causes of the increas- existing on the market) and a marked es in expenditure or in average prices of tendency to encourage a shift in pre- drugs and the extent to which they are El objetivo de este trabajo es desarrollar scribing. This affects the second factor justified (e.g., whether it is due to an in- índices de precios de medicamentos que in the increased expenditure: the aver- crease in patients correctly treated or permitan separar el efecto de distintos whether it is a process of replacing factores, especialmente la comercialización This shift effect is evident if we ana- cheap and effective existing products de nuevos medicamentos que suponen lyze the consumption of certain thera- with newer, more expensive ones), so distintos grados de innovación terapéutica, peutic drug groups, such as antihyper- that they can take measures to prevent sobre los incrementos de precios y sobre el tensives or antidepressants. In these spe- non justified increases. The first step is gasto. Se han utilizado los índices de cific cases, consumption of new drugs is to identify expenditure growth factors Laspeyres y de Paasche en sus versiones fija steadily growing, while the consumption and the relative importance of each of y encadenada, tomando como precio el of older drugs is constant or even falling.
gasto medio a PVP (Precio de Venta al This situation in the pharmaceuti- The key question, then, is to what Público) por DDD (Dosis Diaria Definida). La cal industry has already been dealt with extent does the increased expenditure aplicación empírica se circunscribe a dos extensively in the literature. Unlike oth- reflect an increase in therapeutic benefit grupos terapéuticos, antipsicóticos y anti- er industrial sectors of rapid technolog- or in patient well-being generally. The depresivos, y al ámbito de la prestación ical progress, such as computers or pes- availability of suitable price indexes can farmacéutica del Servei Català de la Salut en ticides, where a generation of new prod- provide the decision-maker with tools to el periodo 1991–1999. Esta metodología ucts clearly displaces a previous obso- help him answer all these questions.
puede ser aplicada de forma rutinaria y lete generation [3]. Moreover, very often However, the preparation of suitable automática a cualquier grupo terapéutico o the growing number of drugs available price indexes is no easy task [1, 10].
a la totalidad de los mismos, utilizando la is not accompanied by real innovation The objective of this work is to de- información que recogen habitualmente las in terms of significant therapeutic ad- velop different types of prices indexes to administraciones sanitarias en el proceso de vance [6]. In fact, the government regu- apply to drug expenditure analysis, that facturación de los medicamentos o la lation of pharmaceutical prices has in- represent an advance over traditional in- información muestral sobre consumo y centivated companies in that sector to dexes as regards the treatment of inno- gasto de algunos estudios de mercado.
diversify their products, as a way to in- vation. It is not our aim to determine crease or, at least, maintain profits [2].
"the" correct index, but rather the most To summarize, the increase in the suitable one for each possible question average price of drugs may be the result or problem. More specifically, we want Prestación farmacéeutica · Precios · Indices of a number of factors: to develop a set of indexes that will en- de Laspeyres · Indices de Paasche able us to quantify what part of the 1. "Pure" increases in the prices of change in pharmaceutical expenditure existing products.
is attributable to each of the factors re- 2. Appearance and increased consump- sponsible for that change.
tion of new products with a higher To be able to identify and quantify than average price. These products the factors determining drug expendi- may involve a major or minor thera- ture growth we need to have analytical peutic innovation or may simply be tools and data that enable us to separate products equivalent to other exist- the role of each factor. Price indexes are ing, cheaper products.
the traditional tool for that purpose.
3. Shifts in consumption from low-cost A price index is a measure of the products to those of a higher price, central trend in price changes for a giv- which may or may not involve any en group of products. The concept is additional therapeutic benefit. These rather simple. Its application, however, shifts take place: is more complex. First, from a theoreti- a. between different presentations – cal point of view, price averages should route of administration, dose, relate to homogeneous products; sec- pack sizes of the same specialty; ond, there is no single formula for aggre- b. between brands of the same ac- gating individual products (pharmaceu- tive substance (active principle); tical specialties) or for weighting them.
HEPAC 4•2001 143

These considerations are very impor- existing on a given year and changes the tant, especially in a market such as the weighting each year according to drug market with the high rate of intro- In order to control the therapeutic nov- current consumption. The fixed Laspey- duction of new products mentioned elty factor, we based ourselves not on the res index is the most commonly used in earlier, and where it is difficult to distin- prices of the products per unit of sale, the calculation of official price indexes guish the additional therapeutic benefit i.e., price per pack, but on the price per such as the consumer price index (CPI).
of new commercial brands or new forms DDD of each product, meaning each Separate price indexes were con- of presentation of existing products.
route of administration of an active structed for each of the two therapeutic (The question of how to account for principle. (Using indicators such as price groups chosen, antipsychotics and anti- quality changes in building price index- per pack poses the problem that in the depressants (groups N05A and N06A, es is addressed in the Appendix by case of a shift in sales towards higher respectively). It should also be noted that means of a non pharmaceutical exam- dose packs or packs with a larger num- thanks to the data bases used, the index- ber of tablets, the value of the indicator es are based not on a sample of goods In this work the empirical applica- increases, which may wrongly suggest but on all of them. This improves the tion is limited to two therapeutic that there has been a price increase.) Ac- representativeness of the indexes con- groups. The ultimate objective, however, cording to the method used here, all ex- siderably, whatever the type of index ap- is to develop a method that can be ap- penditures at consumer price on plied routinely and automatically to any products with the same active principle group or to all of them on the basis of and route of administration (distin- Laspeyres indexes:
the information on pharmaceutical pay- guishing, for example, between solid fixed Laspeyres index
ments habitually compiled by the health oral administration and oral in solution) authorities of the Spanish state as part are added together and divided by the The formula used to calculate the Las- of the invoicing process.
respective number of DDD.
peyres price index (IL), in its traditional The prices obtained are incorporat- version, is shown below: Materials and methods
ed into the index weighted by their sharein the expenditure of the corresponding I L = ∑ w i0 i0 anatomic-therapeutic group. The prod-  i0  uct has also been defined, as an alterna- i0 i0 A longitudinal descriptive study was tive, in terms of active principle (for all carried out of medical prescriptions routes of administration) and even where Pit is the price of the item – active charged to the Catalan Health Service chemical family. The greater the aggre- principles by routes of administration, (Servei Català de la Salut, SCS). The in- gation criterion, the less sensitive the in- active principles and therapeutic fami- formation source was FAR 200 and FAR dex is to changes in quality. The less dis- lies or subgroups, "i" in the period "t"; 413 lists from the IDMSIX program of aggregated it is, the more it tends to pre- Pi0 is the price of the item "i" in the ini- the data processing application for mon- suppose that any price change is attrib- tial period; n is the number of items con- itoring the pharmaceutical payments of utable to an improvement in quality.
sidered for each aggregation level; qi0 the SCS. This program compiles infor- The variables considered were ex- are the quantities of the initial period mation on all prescriptions invoiced to penditure at consumer price and con- (also called base period, which in the the SCS. Two drug groups were selected sumption in DDD. The price variable case of the Laspeyres price index is al- for study: N06, antidepressants and used was expenditure per DDD, the re- ways the first period) and wi0 are the N05A, antipsychotics.Annual consump- sult of dividing total expenditure at con- weightings of each item in the base peri- tion was studied for the period sumer price by consumption in DDD of od (this quotient shows the weight of total prescriptions for each level of ag- each item considered in relation to total The degree of therapeutic innova- gregation considered. The aggregation expenditure for the therapeutic group tion was determined according to the as- levels used in the analysis were: (a) ac- for the first period).
sessment published in the quasi official tive principle by route of administration; The calculation of the fixed Laspey- drug journal of the Spanish National (b) active principle; anatomic-therapeu- res index uses fixed weightings, which Health Service, Therapeutic Informa- tic subgroup (c) or substitutive sub- are those of the base year (in our case, tion; specialties for which consumption group (which ultimately corresponds to 1991). Thus, successive price changes cal- had already been recorded in 1991 (at the that of chemical family).
culated are always weighted by the per- start of the study) were assigned a spe- For the calculation of price indexes centage expenditure of the item in the cial code, since although they might we considered the Laspeyres and Paa- first year of the period in question.
have been innovative at the time they sche indexes, in two of their modalities: It should be borne in mind that the were introduced, that characteristic was fixed and chained index [5]. In the fixed Laspeyres index implies acceptance that not relevant for analyzing innovation modality, the Laspeyres index uses con- the distribution of consumption in the during the study period 1991–1999. Fi- sumption in the first year for the initial year is representative of the whole nally, the source for assigning the daily weightings, while the Paasche index us- period, a very restrictive assumption in defined dose (DDD), was the WHO "ATC es consumption in the last year of the the case of the pharmaceutical con- Index with DDD 1999." period. The chained modality, on the sumption analyzed here, in which con- other hand, includes all products siderable inward and outward flows take 144 HEPAC 4•2001

year considered, and is therefore moreappropriate in a consumption groupwith a high rate of innovation.
This index, then, pulls together both the items that already existed in 1991 andthose which in subsequent years havebeen created and incorporated into drugconsumption statistics. Figures 1 and 2show the results of calculating the Paa-sche index in its fixed version for bothdrug groups.
Calculating the Paasche price index presents one difficulty, namely, deciding Fig. 1  Fixed Paasche index: antipsychotics
what assumption to make on the pricesof the products that were not marketed place, especially in active principles and ically different from 1991, so it is point- after the beginning of the period, for the routes of administration (a simple com- less considering 1991 as the weighting years before their introduction. Of the parison between percentage expendi- two possible options, considering a price tures for these two categories between zero or projecting the entry price back the first and last year illustrates this to 1991, we leaned towards the second. If point well), resulting in a considerable we want to study price change, the sec- change in distribution of consumption Fixed Paasche index
ond option makes more sense than the between 1991 and 1999.
first: in the first case a substantial price So in 1991 some of the items that The formula used for calculating the increase is assumed in the year of ap- will have considerable weight in total ex- Paasche price index (IP), in its tradition- pearance, while in the second case it is penditure in 1999 do not even figure, as al version, is shown below: assumed that until the product appears, they had not yet been introduced. This there are no price increases.
is the case for some active principles in I P = ∑ w the antipsychotic drug group: Olanza- Antipsychotics. As regards antipsychotic
 i0  pine increases from 0% in 1991 to 51.7% drugs, we can see in the Paasche index, at in 1999, and Risperidone from 0% to the level of the replacement subgroup, 36.1%; as for antidepressants, Paroxetine where Pit is the price of the item "i"in the steep growths in the years 1994 to 1999, increases from 0% in 1991 to 31.5% in period "t"; Pi0 is the price of the item "i" which can be related clearly to the intro- 1999, and Venlafaxine from 0% to 8%.
in the initial period; n is the number of duction of new active principles (*Table The calculation assumption of the fixed items considered for each aggregation 1). The appearance on the market of Ris- Laspeyres index is certainly restrictive: level; qit are the quantities for the final peridone (tablets) in 1994 explains the not existing in the first year of the peri- period (also called base period, which in 190% growth in the price index for anti- od analyzed means not being considered the case of the Paasche price index is al- psychotics: it increases from a value of in any subsequent year.
ways the last period); and wit are the 106 to 308, the base value in 1991 being On the other hand some items ap- weightings of each item in the same pe- 100 (Risperidone will finally represent pear in 1991 with a large weighting that riod. This quotient shows the weight – 32% of total expenditure in this thera- will end up having a very small weight the relative importance – of each item peutic group in 1999). On the other hand in total expenditure for 1999. This is the considered in relation to the total expen- in 1997 the introduction of Olanzapine case for some active principles in the an- diture of the therapeutic group in the (tablets) will explain the 40% growth in tipsychotic drug group: Levopromazine last period.
the Paasche index for the same antipsy- decreases from 9.8% in 1991 to 1.0% in The calculation of the Paasche in- chotics, from 373 in 1996 to 523 in 1997 1999; Thioridazine decreases from 19.7% dex also uses fixed weightings, which in (Olanzapine will finally represent 51% of to 1.5%; Haloperidol decreases from this case are those of the last year. Ac- total expenditure in this therapeutic 23.3% to 2.1% and Sulpiride from 16.5% cordingly, successive price changes are group in 1999, being the main active to 1.7%; and as regards the antidepres- always weighted by the percentage of the principal of antipsychotic drugs). As we sant drugs, Clomipramine decreases item in question referred to the last year said at the beginning, drug prices are from 13.7% in 1991 to 2.1% in 1999 and of the period considered. The fixed regulated or controlled by the Adminis- Fluoxetine decreases from 54.7% to Paasche index, although it also involves tration: this explains why, once they have accepting the very restrictive assump- been launched, their price is maintained For the two drug groups considered tion, as with the fixed Laspeyres index, or even falls, and we have very few active the fixed Laspeyres price index calculat- that the distribution over the period is principles that show a clear price in- ed, which shows little price change, is maintained and that, accordingly, the crease over the years. It is at the moment neither very relevant nor very appropri- weighting for 1 year is representative of of introduction, when a new active prin- ate: given the high rate of products go- the whole period, at least ends up pull- ciple enters the market, that the ratio be- ing in and out, the market in 1999 is rad- ing together all items existing in the last tween its total expenditure and quantity HEPAC 4•2001 145

to calculate the price indexes of somepharmaceutical products.
The formula used for the Paasche index (IP) in them is as follows: I P = ∑ w  , −1  , −1 = nt−1 Fig. 2  Fixed Paasche index: antidepressants
it is the price of the item "i" in the period "t"; Pi,t-1 is the price of the item"i" in the period prior to "t" (t-1); n is the (DDD) leaps upwards.And it is this quo- principle: Fluoxetine starts being sup- number of items considered for each ag- tient between expenditure and quantity plied in solution, despite the fact that it gregation level; qib are the quantities of that gives a higher price than that of the has been available in tablet form since the item "i" in the base period "b", and other existing active principles, which 1991; given the price at which it enters this "b" may be equal to the initial peri- brings about the spectacular growth in the market (which also falls over the od, the previous period (t-1) or any oth- the general level of prices of the thera- years), this innovation does not cause a er period; wi, t-1 indicates the weightings peutic group. Consequently, stimulated substantial growth in the index for anti- of each item in the previous period, t-1; by the introduction of these two active depressants either. Finally, it is the intro- and finally, IPt-1 is the value of the chained principles (Olanzapine and Risperi- duction of Venlafaxine in 1995 that caus- Paasche index also corresponding to the done), which between them represent es a small growth in the price index in previous period.
nearly 90% of pharmaceutical expendi- the next 2 years: although its price falls Tables 3 and 4 (*) show the result of ture of the whole antipsychotic drug in the end, the price level at which it en- these calculations, figures 3 and 4 repres- group in 1999, the fixed Paasche index ters the market pushes the index up by ent the trends in those indexes for the increases from a value of 100 in 1991 to three aggregation levels considered. For 537 in 1999, a combined growth of over antipsychotics (*Table 3), the chained in- 500% in just 8 years. Figure 1 shows the Chained Paasche index
dex increases from a value of 100 in 1991 Paasche index of the antipsychotic to 139 in 1999. The evolution of this index group at the levels of replacement sub- To overcome the limitations of price in- shows the same leaps as those recorded group, active principle and active princi- dexes with fixed weighting we use in Table 1, although in a smaller propor- ple by route of administration. It can be chained indexes. These are used, for ex- tion: the entry of Risperidone in 1994 seen that it is only at the first aggrega- ample, in cases where the items consid- and Olanzapine in 1997 are responsible tion level that we observe the large ered move in and out at high rate. The for the subsequent jumps in the price in- growth in price index: when a new ac- formula for this index is the one used by dex. It can be seen that in this chained tive principle or route of administration the Bureau of Labor Statistics (BLS) to calculation modality, unlike the fixed, the enters the market, its value is 100; and calculate the Medical Care Price Index price indexes at the level of active princi- given that the majority of prices are (MCPI). It is also the one used by the Na- ple and route of administration also re- maintained or tend to fall once they are tional Bureau of Economic Research flect the impact of the introduction of on the market, the price index does not (NBER), in its more recent publications, therapeutic novelty: the index value for pick up the effect of the entry of the newproduct on the average price of the itemfor the other two lower aggregationlevels, as it projected the entry priceback to 1991.
Antidepressants. As regards antidepres-
sants, the introduction of new active
principles does not cause increases in
the Paasche price index (with one excep-
tion).As a whole, this fixed index for an-
tidepressants increases from a value of
100 in 1991 to 109 in 1999 (*Table 2). In
1992 the appearance of Paroxetine hard-
ly alters the general Paasche index of the
group. In 1994 there appears a new route
of administration of an existing active
Fig. 3  Linked Paasche index: antipsychotics
146 HEPAC 4•2001
from Table 8 that, after falling from a val-ue of 100 in 1991 to 98.9 in 1999, the priceindex now falls to 89.4.
As the prices of drugs, once introduced in the market, are controlledby the government and because of theway in which the industry (with the ad-ministration's approval) has respondedto that control, traditional price index-es, such as the fixed Lapeyres index, donot provide valid and relevant informa-tion. These indexes show small increasesor no increase at all, a totally different Fig. 4  Linked Paasche index: antidepressants
picture from the one drawn from thetrends in average price per prescription, 1 year is affected (corrected) by its value price index were, therefore, the active which habitually records a notable, sus- in the previous year. And as regards an- principles which involved a notable tained growth, particularly in our case tidepressants, we can see a slight fall in therapeutic innovation (in particular for antipsychotics (nearly 700%). By us- the chained price index, of barely 1% be- Olanzapine and Risperidone).When the ing a fixed Paasche index, on the other tween 1991 and 1999, after slight increas- active principles of levels B and C (Sert- hand, the growth of prices of antipsy- indol pastilles, Zuclopenthixol pastilles, chotic products shows an increase from solution and parenteral depot) are elim- a value of 100 in 1991 to 537 in 1999: stim- Accounting for therapeutic innovation
inated, the price index increases ulated by the introduction of two active in price indexes
somewhat less than in the previous case.
principles, Risperidone in 1994 and Table 7 shows an increase in value from Olanzapine in 1997 (which between In this section we attempt to analyze 100 in 1991 to 111 in 1999; so level C pro- them cornered nearly 90% of pharma- price trends eliminating the possible ef- ducts have a slight deflationary effect.
ceutical expenditure for the whole anti- fects of innovation. To do so price index- In the case of antidepressants, there psychotic therapeutic group in 1999), the es are calculated eliminating, first, prod- is a slight difference between the index overall growth of the therapeutic group ucts introduced over the period which for products as a whole and the index re- is over 500% in just 8 years. Finally, us- involve some therapeutic contribution, sulting after eliminating the products ing the Paasche index in its chained ver- and second, new products which do not classed as reasonable therapeutic inno- sion, it increases by nearly 40 %, and the involve any relevant therapeutic contri- vation (i.e., Moclobemide, the only level periods of price increase coincide with bution over existing products.
B product). The price index, which be- the years in which the above therapeu- Tables 5-8 (*) show the results of fore fell from 100 in 1991 to 98.9 in 1999, tic novelties were introduced.
considering two categories or levels of now falls to 96.9: this minimal difference This phenomenon is not observed innovation, B (reasonable therapeutic is because in 1999 Moclobemide, with antidepressants. In terms of aver- innovation) and C (practically no thera- launched in 1992, has a weight of less age price the increase is 84 % between peutic innovation), in accordance with than 1 % of total expenditure for its ther- 1991 and 1999, although calculating the the criteria and subsequent cataloguing apeutic group. On the other hand, it is fixed Paasche index for antidepressants by the Ministry of when the level C active principles (Cita- shows an increase from a value of 100 in lopram, Mirtazapine, Nefazodone, Re- 1991 to 109 in 1999. In this case, too, it is ic/documentos). Figures 5 and 6 repres- boxetine, Paroxetine, Sertraline y Venla- the introduction of new products (Par- ent the consideration of those levels for faxine) are eliminated that the price in- oxetine 1992; Fluoxetine sachets 1994; the aggregation level of chemical family.
dex falls considerably. It can be seen Venlafaxine 1995) that pushes this index It is in the antipsychotic group that we see the most notable difference be-tween the price index with total prod-ucts and those that are left after elimi-nating the three active principles con-sidered of level B. In Table 5 we see thatwhen those principles (Olanzapine, Ris-peridone and parenteral Zuclopenthix-ol) are eliminated the increase in pricesin this therapeutic group is considerablyreduced: with all products, between 1991and 1999, the index grew from 100 to 139,and without those products, only from100 to 109. The products responsible fora large part of the growth in the (total) Fig. 5  Antipsychotics, linked Paasche index, and innovation level
HEPAC 4•2001 147
Moreover, to enable us to separate the different factors responsible for theincrease in average prices and, in the lastinstance, expenditure on drugs, it wouldbe helpful to develop for each therapeu-tic group a set of indexes with the fol-lowing aggregation levels: (a) pharma-ceutical specialty, (b) active principle byroute of administration, (c) active prin-ciple, (d) chemical family. In future, im-provements in quality should be identi-fied and assigned monetary values on a Fig. 6  Antidepressants, linked Paasche index, and innovation level
product by product basis, so as to deter-mine what proportion of the price in-creases are attributable to those im- upwards, albeit to a far lesser extent). Fi- (fixed Paasche) destroys homogeneity provements or whether they are pure nally, using the Paasche index in its but achieves greater representativeness.
price increases. This information would chained version, the index in question Given the trends experienced be- enable us to diagnose the causes behind shows a slight price fall of 1% between tween 1991 and 1999 in the composition trends in expenditure and to create a ra- 1991 and 1999.
of the therapeutic groups considered tional basis for future policies.
An attempt was also made to relate (with some active principles, especially the increase observed in the fixed Paa- among the antipsychotics, that did not Acknowledgements. This work was awarded
sche indexes to the level of innovation exist in 1991, but which represent a very a research prize by the "Sociedad Españolade Farmacéuticos de Atención Primaria." The represented by the different therapeutic high percentage of total expenditure in authors are grateful for the initiative and sup- innovations considered. In the case of 1999), it makes more sense to try to base port of Josep Lluís Segú in the preparation antipsychotics there is a notable corre- on the Paasche price index any conclu- and development of the project.
lation between the growth in prices and sions about the evolution of the prices.
the active principles involving reason- However, using this index poses in its Appendix: price indexes and
able innovation: by eliminating the three turn several problems. For instance, active principles of level B only (Olanza- there is no clear criterion for treating the pine, Risperidone and parenteral Zuclo- prices of products that did not exist at The treatment of changes in the quality penthixol) the growth in prices in this the beginning of the period: if they are of products over time is a problem com- therapeutic group between 1991 and given a zero value, price growth will be mon to all price indexes. The more usu- 1999 is reduced from 40% to 10%. In the overestimated. If they are assigned the al price indexes, such as the CPI, are Las- case of antidepressants, however, there price of the first period in which they peyres indexes which reflect trends in is practically no difference between the appear, an underestimation is likely to the prices of a basket of goods represen- price indexes for the actives principles occur.Another practical problem is that tative of consumption in the initial base as a whole and the remainder once the using the Paasche index routinely and period. If the basket is not changed, it principle classed as a reasonable inno- continuously over time means that the gradually loses its representativeness vation (Moclobemide) is excluded; it is weightings will vary each year, and so over time and, as a result, so does the in- when the level C active principles (Par- the index should be recalculated annual- dex. But updating the basket by intro- oxetine, Sertraline and Venlafaxine) are ly for the whole period. The results of ducing new products and removing excluded that the price index for this any analysis could change if repeated in those products no longer consumed group falls nearly 8 percentage points a subsequent period.
poses its own problems.
between 1991 and 1999.
Obviously, this is not ideal if the in- In theory, price indexes should re- The above results demonstrate that dex is to be used as a basis for adminis- flect trends of a set of goods that pro- the choice of price index is an essential trative decisions, negotiations, etc.
vides the consumer with a particular question in the analysis of trends in utility. Let us suppose that instant coffee prices and pharmaceutical expenditure.
replaces traditional coffee in the family The choice of a particular index implies shopping basket. The price of normal certain specific assumptions regarding The optimum solution to the selection of coffee is 500 Ptas/kg and the price of in- the representativeness of the products a valid and reliable price index, at least in stant coffee is 1000 Ptas. Let us suppose, chosen and the homogeneity of con- the context of our study, seems to point furthermore, that those prices have re- sumption over time. Basing the price in- towards the application of chained Las- mained unchanged for the whole of the dex on products existing at the begin- peyres indexes, which in the case of period in question. If at any given mo- ning of the period (fixed Laspeyres) im- pharmaceuticals does not pose the infor- ment the price of the former is simply proves homogeneity at the expense of mation problems that would arise with replaced by the price of the latter, one is representativeness, while taking the other goods, since data on consumption implicitly accepting that the two goods weightings of the last period considered are usually available, making it easy to are homogeneous (i.e., the same prod- calculate weightings year by year.
uct) and that their price has doubled.
148 HEPAC 4•2001
Obviously, the products are not homo- Unfortunately, in other cases the sit- geneous and consumers may prefer the uation is more confusing. Let us suppose advantages of instant coffee, so the price now that coffee beans at 500 Ptas/kg are Boskin MJ, et al (1998) Consumer prices, difference should be attributed to the replaced by ground coffee at 1000 the consumer price index and the cost of improved quality and should not be re- Pta/kg.Although ground coffee may im- living. J Econ Perspect 12:3–26 flected in the price index. In economic ply an improvement in quality, it would Borrell J-R (1999) Pharmaceutical price terms, the individual is obtaining in- be hard to accept that such an improve- regulation. A study of the impact of the creased utility from the new basket in ment could justify such a large price in- rate-of-return regulation in the U.K.
which traditional coffee has been re- crease. This would be even more doubt- Cockburn IM, Anis AH (1998) Hedonic analysis placed by instant. Consequently, a new ful if the conditions within the coffee of arthritis drugs. National Bureau of Economic basket should be created which includes market were not those of a competitive Research working paper 6574 instant coffee.
market. Part of the increase could be a Fernandez-Cornejo J, Jans S (1995) Quality The argument could be further pure price increase due, for example, to adjusted price and quantity indices for pesti- complicated if we suppose that making a changes in supply and demand. If we cides. Am J Agricultural Econ 77:645–659 cup of coffee requires 10 g of traditional substitute the new product in the con- Frank RG, Berndt ER, Busch SH (1998) Price coffee and only 5 g of instant. In that case sumption basket and start the index indexes for the treatment of depression.
National Bureau of Economic Research it could be argued that the relevant unit with a new base, we will have underval- working paper 6417 of product is not price per kilogram but ued the real price increase.
García MA, et al (2000) Adopción de los nuevos the cost of the coffee necessary to make There are, of course, theoretical medicamentos por los médicos prescriptores.
one cup. Under these suppositions, the methods for differentiating the pure El medico innovador. Rev Atención Primaria price of coffee would have remained con- price increase part from the part attrib- stant (20 Ptas per cup). In fact it could be utable to quality changes. For example, Gerdtham U, et al (1997) Price indices of drugs argued that the price had fallen, as part one could estimate the cost of grinding and the switching to new drugs.
of the new price corresponds to an im- the coffee or consumers' willingness to Griliches Z, Cockburn I (1994) Generic and provement in quality of the product.
pay more for this feature of the product.
new goods in pharmaceutical price indexes.
However, the adjustment methods are Am Econ Rev 5:1213–1232 not easy and will no doubt be subject to Ministerio de Sanidad y Consumo (1992) Información Terapéutica del Sistema Nacionalde Salud, 16:222 Ruiz-Castillo J, Ley E, e Izquierdo M (1999) *Tables 1 to 8 are available at URL:
La mesura de la inflació a Espanya. Colleció d'Estudis Econòmics. Caixa d'Estalvis i Pensions de Barcelona "la Caixa",no 17 HEPAC 4•2001 149

Source: http://www.eco.ub.es/~tremosa/HEPAC.pdf