Micro-Costing

A detailed bottom-up costing method that identifies and values every individual resource consumed, such as each minute of staff time.

Source: Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. 4th ed. Oxford University Press; 2015.

Every resource consumed is identified, the quantity of each is measured, and a unit value is attached to it. For a procedure that means listing staff by grade with the minutes each contributes, consumables by item and quantity, equipment time with an apportioned capital charge, and the share of accommodation and support services attributable to the episode. The total is built from its components rather than derived by dividing an aggregate, which is what allows any component to be varied independently and the estimate to be re-priced elsewhere. Recording quantities separately from unit values is essential, since only the quantities transfer to another setting and a total expressed as a single figure cannot be re-priced.

Source: Drummond et al. 2015

Direct observation and timing gives the most reliable estimate of staff input and is the most expensive to obtain. Structured self-recording by staff is cheaper and subject to recall and rounding. Extraction from systems that timestamp events captures what happened rather than what was reported, and is increasingly viable, though it records the presence of an activity better than the effort it consumed. Expert estimation is the weakest option and is used where nothing else is available. Combining methods is common in practice, with observation used for the dominant resource and lighter methods for the rest, and the mix should be reported rather than described simply as micro-costing.

Source: Glick et al. 2014

Where the resource being examined is the one the decision changes, since a decision about staffing cannot be informed by a cost that assumed staffing away. It is warranted for interventions with no established unit cost, for comparisons where options differ in the composition of resources rather than only in volume, and where the decision is large enough that a wrong answer would be costly. It is not warranted for resources the decision leaves untouched, which can be taken from published unit costs. A useful test is whether the decision would change if the resource were costed less precisely, since where it would not, the additional effort produces no benefit.

Source: Drummond et al. 2015

The apparent precision extends only to the resources actually measured, so a total assembled to the minute for staff time and by apportionment for everything else is no more accurate overall than its weakest component. Estimates are specific to the site observed, and a service with different staffing or throughput will differ substantially. Fixed capacity is easily misrepresented, since attributing a share of a theatre or scanner to an episode implies a cost that varies with volume when it does not. Reporting the number of observations behind each estimate is also useful, since a mean staff time derived from a handful of cases carries very different weight from one derived from several hundred.

Source: Drummond et al. 2015

The usual arrangement applies detailed measurement only to the resources the intervention changes and standard published unit costs to everything else, which concentrates effort where it affects the answer. The boundary between the two should be stated so a reader can see which parts of the total were measured and which assumed. Where two options are being compared, only the resources differing between them require detailed treatment, since common elements cancel. This is also why comparative analyses need detailed treatment only of the resources that differ, which frequently reduces the work by more than analysts expect.

Source: Drummond et al. 2015