Guidelines for cost calculation in economic evaluations of healthcare programs (PROJECT)

The approaches to measurement and evaluation of costs vary along a spectrum of specificity. On one end of the spectrum there is the approach calls for the direct enumeration and costing out of every input consumed in the treatment of a particular patient. On the other end is such gross approach as estimating the cost of an event, for example hospitalisation for heart infarct - by assigning average tariff. The former approach refers to the micro-costing methodology and the more aggregative to the gross-costing methodology. The choice between gross-costing and micro-costing must balance the needs of the analysis - the sensitivity of the results to the bias and precision in the cost estimates - with the difficulty and expense of obtaining a cost estimate.

Micro-costing starts with the detailed inventory and measurement of all inputs consumed in a healthcare intervention. Once the resources consumed have been identified and quantified, they are then converted into values terms to produce a cost estimate. Micro-costing approach is frequently associated with primary data collection within randomised clinical trials or observational research studies.

Gross-costing bases cost estimates on more aggregated information on resource use. It starts with identification of a sequence of "economically significant" events associated with the intervention. These events may include one or more of the following: hospitalizations, physicians services, drugs. Gross cost estimation of the cost of the intervention requires estimating these component event costs, then summing. The processes of measurement and valuation of resources, which are reasonably distinct in micro-costing, are more blurred in gross-costing.

Often these techniques draw on readily available administrative prices. Frequently gross-cost estimates are to be input into larger decision model. The distinguishing features of gross-costing are its simplicity, practicality and its (intended) insensitivity to site-specific details and patient-level characteristics.

The choice of adopted methodology should relate to the type of trial to be conducted. Micro-costing and gross-costing can be used within a single analysis. In general, micro-costing will be more important for aspects of the alternatives under consideration that are likely to diverge in cost, and for intervention and events occurring in the present. Gross-costing is acceptable when using a more exact micro-cost estimate cost will not have an important effect on the analysis. Precision is generally less critical in estimating resources that will be consumed far in the future.

The selection of methodology depends on the individual objectives of the analysis and should be made by the researcher. Nevertheless the choice of a specific methodology should follow standard recommendations and the use of standard sources of data. It should be understood that standardisation of analysis does not apply to the choice of methodology but to the procedures of analysis once the choice is made.

Standard values define the parameters used in the calculation of unit costs. The standard values are for example: gross salary, working hours per year, the number of workable hours of medical staff per year, yearly income in healthcare sector, the average distance of a household to a hospital (to calculate costs of transport), discount rate, inflation rate.

The use of standard values results in the decrease of differences in unit costs estimates between studies. Use of standard values is recommended while presenting analyses for the purpose of reimbursement of new medicines.

The application of standard unit costs has a major role in standardising the economic evaluation of healthcare programs. Despite many well-known limitations, it leads to the elimination of certain differences between trials where different unit costs were used.

The use of standard unit costs is recommended in the case of formal appraisal studies for reimbursement purposes. However for the units with the largest contribution to total or incremental costs, a more detailed costing approach might still be necessary (i.e. direct calculation of unit costs).

The use of the list of standard unit costs is limited to procedures and services for which these costs are calculated. The list of standard unit costs should contain at least standard unit costs for resource items of:

1) out-patient care,

2) in-patient care,

3) pharmaceuticals and medical materials consumed in hospitals,

4) DRG-group (data obtained by large costing studies in which many hospitals take part)

This list will be published and up-dated every year in order to allow adjustments to changes and incorporation of new findings.

In the process of cost calculation the following 5 stages are defined:

1) definition of time horizon and perspective,

2) selection of cost categories,

3) identification of units

4) measuring resource use

5) valuation of resource use

At each stage the choices have to be made between proposed alternatives and these together clearly define the way of proceeding. These choices depend on the aim and specific setting of the study. Since the choices in later stages of the analysis depend on the choices made in previous stages, the stages mentioned above have to be passed in chronological order.

The perspective and time horizon of costing should be the same as the perspective and time horizon of economic evaluation perspective of which cost calculation remains a key part.

The choice of time horizon and perspective is strictly linked to further stages of the analysis in which cost categories selected for the trial are identified and the method of their measurement and evaluation are defined.

The costs are divided into following categories: direct medical and non-medical costs, indirect costs within and outside the healthcare. The selection of cost categories covered in the analysis depends on the perspective has been chosen. For example, a study prepared from a narrow public purchaser would not consider indirect costs or direct costs paid by patient, whereas a study prepared from a broader societal perspective would do so.

According to draft of Polish guidelines for conducting pharmacoeconomic analyses indirect costs outside the healthcare sector (future costs during life-years gained) should be considered in economic evaluation only when they result directly from original intervention. Intangible costs because of the lack of relevant methods are not considered or they are calculated within value of changes in quality of life

The identification of units of resources used raises two questions: what type of resource use are relevant for the disease or intervention studied and to what level of detail do they have to be measured and valued separately. If for example almost all patients who visit a specialist receive the same diagnostic tests, it is not necessary to cost these test separately, but it is better to incorporate the average costs of diagnostic tests into the unit price of an outpatient visit. Only in the case where diagnostic tests performed during consultations vary significantly among patients is it worth identifying these tests as separate cost units.

It is recommended to describe in detail a given procedure in order to define which resources units should be taken into account in the analysis. In order to define the units which have the largest contribution to total and incremental costs it is recommended to conduct a sensitivity analysis. Sensitivity analysis shall also be used to determine which costs should be measured and evaluated in details (using micro-costing methodology) from those which can be evaluated via gross-costing methodology.

There are many different ways for determining resource use: primary data collection within randomized clinical trials or observational research or secondary data such as from databases. The selection of data sources depends on the level of detail required and should be based on following criteria:

1) the perspective of study,

2) the contribution of units to total and incremental costs,

3) availability of data,

4) the balance between internal and external validity.

Internal validity refers in this case to the degree in which resources use measured reflects actual use in the population being studied. External validity is related to the generalisability of the results to the other settings, especially real clinical practice. Often an increase in internal data reliability is associated with a decrease of external reliability and vice versa.

The advantage of using primary data is the level of their precision, the disadvantage - fact, that collected within randomized clinical trials or observational research they would likely include protocol-induced resource consumption. Secondary data such as from national registries provide externally valid estimates of resource utilization, but can be incomplete because many resources may not have been tracked in a given database.

The selection of unit costs (prices) used in the analysis should be consistent with the study perspective.

There are three alternative ways to obtain valid unit costs:

1) use of standard costs,

2) use of values derived from national registries, previously published trials in health economics or derived from local lists of tariffs or charges,

3) direct calculation.

The selection of the valuation method is determined by the choice of methodology used in the measurement of resources used. For example there is little point in performing valuation of resources used based on direct calculation whenever the measurement of resources used was based on national registries.

The use of standard costs is described in more detail in section 1.3.

The availability of prices derived from national registries or previous research is limited and often there is insufficient information about the way these prices have been determined.

The use of local charges or tariffs can be indicated in particular when the therapy in question is available only in certain type of healthcare institution. The attractiveness of this method, as in the case of standard costs, lies in the extensive list of procedures and services for which charges or tariffs are available, in simplicity and in availability without additional expense and time to the analyst. However, the use of charges is a controversial area. On the one hand there are arguments for their adoption such as that they are often more appropriate and sometimes the only available method. On the other hand they do not always reflect actual unit cost of the procedure or service but are merely the vehicle for transferring money from payers to providers. Application of charges is a method of choice only in the case of analysis conducted from the healthcare payer perspective. In the other cases the analyst should try to get some insight into whether the charge is good estimate of the actual unit cost of a procedure and can be used in the economic evaluation.

Direct calculation of cost is much more laborious than other valuation method and requires a completely different methodology. It is used for the valuation of units which have a substantial impact on total or incremental cost, and for which no adequate unit cost estimate from other sources are available. Important choices regarding the direct calculation of cost concern:

1) the selection of a specific setting in which costs will be calculated,

2) the use of "top-down" or "bottom-up" methods,

3) the allocation of costs of supportive departments, buildings, general equipment, etc.

Since unit costs can differ between healthcare providers and consequently the choice of medical service centre can influence the cost calculation, it is recommended to collect data in more than one centre and vary unit costs in a sensitivity analysis based on differences in unit costs that have been found in centres or are obtained from other studies.

In the "top-down" methodology the cost of department are derived from cost-accounting data and assigned to the products and services produced by the department. In the "bottom-up" methodology the cost of each service in each product line is computed as the sum of the labour and non-labour inputs estimated to be used in that service´s production. It starts therefore with the detailed measuring the actual use of materials, equipment, personnel and time spent on a certain procedure for a single patient. Top-down calculation can be applied in the case of a department with relatively homogenous production, in other cases the bottom-up method is more appropriate. The disadvantage of bottom-up calculations is that they are usually very time consuming and a researcher will not always have the opportunity to perform such detailed measurements. In practice a combination of both mentioned above methodologies is applied.

There are several methods for the allocation of costs of supportive departments, buildings, equipment, etc. The method most often used in economic evaluation is direct allocation. In this method a distinction is made between departments that directly serve patients (i.e. surgery department) and supporting departments (i.e. kitchen, financial services).Cost of supporting departments are firstly allocated to the departments that directly serve patients, and then allocated among the products of these latter departments.

The trial report should contain a detailed description of the methodology used in cost calculation. The sources of resource and unit costs data should be clearly stated. Resources used and unit costs of resources should be reported separately in tabulated form. In addition to mean values, the variation around the mean has to be described and tested in sensitivity analysis.