Health-economic figures can sound precise while hiding the assumptions that produced them.
I built the Psychosis Impact Calculator to make a service-impact conversation interactive. It estimates relapses prevented, annual cost saving and bed days avoided for a psychosis or schizophrenia cohort.
Put the assumptions on screen
The calculator starts with four inputs:
- the number of service users with a diagnosis;
- the average annual relapse rate;
- the average cost of a relapse;
- the average length of an inpatient stay.
Its default scenario uses 800 service users, a 20% annual relapse rate, a £25,000 relapse cost and an 82-day stay. The model then applies a 50% reduction to expected relapses, based on the study assumption described in the interface.
With those defaults, the tool estimates 80 prevented relapses, £2 million in avoided cost and 6,560 bed days.
These are scenario outputs, not promises. Their value comes from exposing the relationship between each input and the result.
Let local knowledge replace generic defaults
Trusts do not all have the same caseload, relapse pattern, inpatient use or cost base. Sliders and numeric inputs allow a team to replace national assumptions with its own figures.
That changes the quality of the discussion. Instead of arguing about one headline saving, people can ask which value is credible, which source should be used and how sensitive the result is to uncertainty.
The calculation remains deliberately legible:
- Expected relapses equal cohort size multiplied by annual relapse rate.
- Prevented relapses apply the stated reduction assumption.
- Avoided cost multiplies prevented relapses by cost per relapse.
- Bed days multiply prevented relapses by average length of stay.
A calculator is a conversation device
The interface presents assumptions beside the result and gives each output a short explanation. It also rescales to fit the available viewport so the complete model remains visible during a meeting or embedded presentation.
I see the prototype less as a forecasting engine and more as a structured conversation. It helps clinical, operational and commercial colleagues identify where they agree, where evidence is weak and which local data would make the estimate more useful.