Marginal Cost of Healthcare

The additional cost of providing one more unit of healthcare service, such as treating one further patient.

Source: Phelps 2017

Because a large share of what a health service spends is fixed within any relevant range: buildings, equipment, establishment staffing and the support functions behind them do not change when one more patient is treated. The additional patient consumes consumables and some staff time, and little else. Average cost divides the whole base across activity and therefore includes resources the additional patient does not cause, which is why using it for a decision about volume overstates the consequence. The gap between marginal and average is therefore widest in capital-intensive services and narrowest in those dominated by consumables, which is why the same reasoning error has very different consequences in different parts of a system.

Source: Phelps 2017

At capacity thresholds, since health service capacity comes in indivisible units. Within a band, additional patients are absorbed at low cost; crossing a threshold requires another member of staff on every shift, another theatre session or another bed, and the whole block enters at once. The marginal cost therefore alternates between low values and jumps, which means a single figure describes only the range within which it was measured. The size of the jump depends on how large the indivisible unit is relative to current volume, so small services face proportionally sharper discontinuities than large ones.

Source: Horngren, Datar & Rajan 2015

Because capacity is added when volume rises and is frequently not removed when volume falls. Growth crossing a threshold requires the whole block before the additional activity can be delivered, while contraction releases nothing until activity falls far enough to empty a block entirely, and often not even then, since staff are redeployed rather than released. A service can therefore expand and contract by the same amount and finish worse off. This asymmetry is one reason services that have absorbed repeated activity growth without additional funding find that the corresponding reduction releases nothing when volumes later fall.

Source: Phelps 2017

A tariff set at average cost overpays for activity absorbed within existing capacity and underpays for activity requiring new capacity, so providers gain from volume growth in some ranges and lose in others. This is one reason payment arrangements frequently combine a block element covering fixed capacity with a marginal rate for activity above an agreed level, which aligns payment more closely with the cost actually incurred. It also explains why providers respond differently to the same tariff depending on where they sit relative to their capacity thresholds, which a system-wide price cannot accommodate.

Source: Phelps 2017

Because it sets an upper bound on what reducing activity can release. A plan built on average cost per case assumes the whole unit cost disappears with the case, whereas only the variable element does unless the underlying capacity is actually removed. Savings projected this way routinely fail to materialise, and the shortfall is predictable from the ratio of variable to fixed cost in the service concerned. Estimating the variable share before a savings plan is agreed is therefore the single most useful check available, and it is rarely performed before the target is set.

Source: Horngren, Datar & Rajan 2015