Price and quantity measures

Calculating price and quantity indexes using the formulae and approaches discussed in Chapter 5 necessarily relies on matched observations for both price and quantity. With the data available for this thesis there are two ways of obtaining these price and quantity observations. The first relies on deriving prices given expenditure and quantity while the second derives quantity given expenditure and prices. Because actual prices are used in the second approach it produces indexes that are conceptually

closer to the usual type of price indexes produced by statistical agencies, such as the CPI.

The derived price approach - actual quantities and derived prices

It is possible to obtain a value for the price of a PBS medicine by dividing expenditure on that medicine by the number of scripts dispensed for it. The price derived in this way is a unit value rather than an observed price. Although this is a simple way of obtaining a price, there are drawbacks to using the number of scripts as the measure of quantity in this fashion. Firstly it assumes that the amount of medicine specified by a script is constant among prescribers and over time. However the doctor writing a script is not bound to prescribe the maximum quantity allowed for that PBS item code, so this means that the amount of medicine dispensed could vary from script to script. Unfortunately it is not possible within the information available to assess the extent to which this is the case. As noted earlier, although Medicare Australia records the number of units of a medicine dispensed by a script it has advised that this measure is often not reported by pharmacists and its accuracy is therefore suspect.

There are two main sources of data on the prescribing patterns of general practitioners. The BEACH dataset maintained by the Family Medicine Research Centre at the University of Sydney is based on paper records kept by a sample of GPs on patient encounters. The records note among other things the GP’s diagnosis and a description of the type and amount of medicine prescribed. A more detailed description of BEACH is provided in Britt et al (2007). Similarly the General Practice Research Network (GPRN) dataset maintained by Health Communications Network Pty Ltd records information from a sample of GPs and specialists using the Medical Director software package (Sayer et al 2003). The BEACH data starts in 1998 and the GPRN data in 1999. Both these data sources are used by the Department of Health and Ageing and others to assess how much is prescribed on average per script.

If the number of scripts is to be used as the quantity measure in index number calculations or any other analysis, it is therefore assumed that either the same quantity (say the maximum quantity allowable) is prescribed by all prescribers or that the

distribution of fractional prescribing is constant over time and hence does not introduce a bias in the calculations.

The second limitation on using scripts as the quantity measure is that the maximum quantity allowable for a prescription by the PBS can and does change over time. This means that, even assuming that a script represents the maximum amount at any one time, the amount of medicine per script may differ from one time period to another. This source of bias can be rectified by multiplying the number of scripts by the maximum quantity to obtain a measure of the number of units (such as a tablet or capsule) that have been dispensed and using this as the quantity measure. Because the maximum quantity is obtained from the monthly PBS Schedule dataset it is necessary to convert this to an annual series before it can be used to adjust the annual scripts data. For the purposes of the analysis reported in this thesis, a simple unweighted annual arithmetic average is calculated using the 12 monthly values for a particular financial year. Ideally the annual average would be calculated by weighting the monthly values by their share in annual expenditure but this information was not available. In practice maximum quantities change infrequently and only for a limited number of items so the bias arising from an unweighted mean is likely to be small.

In this derived price approach then, dividing total cost just by scripts gives a unit value that is the average dispensed price across all transactions for a particular combination of PBS item and manufacturer code in a particular year. The price so obtained is comparable to the annual average of the manufacturer’s dispensed price for maximum quantity (MDPMQ) quoted in the PBS Schedule.

If the total cost is divided by the number of scripts multiplied by the maximum quantity allowable the unit value obtained is the average dispensed price per unit (such as a tablet or capsule) and is comparable to MDPMQ divided by the maximum quantity (MQ).

In the discussion of the outlier problem in Appendix A, it is pointed out that the expenditure data sometimes contains outlier observations, ie there are transactions recorded for periods when the medicine is no longer listed on the PBS Schedule and therefore there is no Schedule information available for the medicine. Obtaining a

derived price by dividing total cost by scripts retains these outliers and therefore gives a price for them. On the other hand once any modification is made to the expenditure or script data by introducing a variable from the PBS Schedule, such as obtaining a derived price per unit by dividing total cost by scripts times maximum quantity, this necessarily suppresses these outlier observations.

This procedure can be extended to obtain unit value price series for other components of PBS expenditure. For instance, dividing the government cost by scripts times maximum quantity gives a price series showing the price paid by the Government for medicines, while dividing the patient cost by scripts times maximum quantity gives the price paid by patients. Similarly the different categories of patients can be considered separately so that for instance dividing total patient cost for general patients by the number of scripts for general patients times maximum quantity gives the average price paid by general patients.

The derived quantity approach - actual prices and derived quantities

An alternative approach to specifying prices and quantities is to start with the actual prices given in the PBS Schedule and use these to derive a quantity measure by dividing the total expenditure by this price. Because the total expenditure refers to the actual cost incurred at the pharmacist’s cash register, the appropriate price to use is the manufacturer’s dispensed price for maximum quantity (MDPMQ). However because the maximum quantity can vary over time the best price to use is the MDPMQ divided by the maximum quantity which is the manufacturer’s dispensed price for a unit of the medicine.

This approach as already noted suppresses any outlier observations.

In deriving quantities in this way, the prices used are those in the monthly PBS Schedule dataset. As was the case for the maximum quantity, annual prices are obtained by a simple unweighted average of the 12 monthly observations.

Once the quantity series has been derived it can be used in price index calculations not only for the dispensed price but for the other prices in the PBS Schedule. Thus

indexes can be calculated for the Commonwealth dispensed price for maximum quantity (CDPMQ), and the manufacturer’s and Commonwealth price to pharmacist for manufacturer’s pack (MPPMP and CPPMP). Again, the preferred price for the Commonwealth dispensed price for maximum quantity is actually CDPMQ divided by MQ. For the prices to pharmacist, the preferred prices are MPPMP and CPPMP divided by the size of the manufacturer’s pack (PS). It should be noted here that the manufacturer’s pack size can vary among different suppliers within a particular item, unlike the maximum quantity which is the same for all suppliers. Further to this, a price reflecting the pharmacist’s markup, ie the difference between the price paid by the pharmacist to the wholesaler and the dispensed price can be defined as the difference between MDPMQ divided by MQ and MPPMP divided by PS.

For the period over which the analysis of PBS expenditure is undertaken – 1991-92 to 2005-06 - 10% of the price paid by the pharmacist went to the wholesaler and 90% to the manufacturer. A price index calculated for the price received by the manufacturer is therefore the same as that for the price received by the wholesaler, ie for MPPMP.

The derived quantity approach using the information in the PBS Schedule therefore enables price indexes to be calculated for the different points in the supply chain: manufacturer, wholesaler, pharmacist and patient. In the derived price approach, indexes can only be calculated at the dispensed point.

However the dispensed price is essentially an administrative device and differs from the actual price faced by the two payers in the PBS - the patient and the government. The price faced by the patient is the relevant copayment if any plus the premium (or SPC) if the brand being purchased has one. The price paid by the Government is the dispensed price less the price paid by the patient.

The price paid by the patient differs by type of patient, because there are three levels of copayment - the general copayment (paid by general non-safety net patients), the concessional copayment (paid by concessional non-safety net and general safety net patients) and free (paid by concessional safety net patients).

The premium is simply the difference between MDPMQ and CDPMQ. The general and concessional copayments are known (and given in Chapter 3) and can be added to the premium to obtain the price paid by those that pay the general and concessional copayment. For those that do not pay a copayment, the price is simply the premium if there is one. In this latter case, most items will have a zero price (which makes calculating price indexes problematic), while for patients paying the general or concessional copayment the price will always be positive.

All these prices can then be divided by the maximum quantity to obtain the price per unit as before.

Indexes can be calculated for the prices paid by the different categories of patients by using the prices calculated in this way in conjunction with an appropriate derived quantity measure. For those patients paying the general copayment the quantity measure is the total PBS expenditure by these patients divided by MDPMQ per unit. Similarly for those paying the concessional copayment it is the total PBS expenditure by these patients divided by MDPMQ per unit. The price indexes for the price paid by the Government for these two types of patients are calculated using the same quantity measures and prices equal to MDPMQ minus both the premium and the appropriate copayment. For the patients with no copayment, the price index for the price paid by the Government is simply the one for CDPMQ per unit.

The two approaches to obtaining prices and quantities therefore throw up a somewhat complex set of possibilities for calculating price and quantity indexes for various groups of participants in the PBS. Not all of these are attempted within this thesis. In this chapter only overall results are presented using the following combinations of price and quantity.