13.6.8 Appraising economic evaluations

The effectiveness estimate

• An economic evaluation is only as valid as the clinical comparison underneath it, because the model inherits any weakness in the effectiveness evidence.

• Randomised data are preferred: an effectiveness estimate taken from a non-randomised comparison, such as a cohort or before-and-after study, is weakened by confounding.

• NICE expects effects to be identified systematically and appraised, not selected for convenience.

Comparators and perspective

• The comparator should be relevant routine care, because placebo or an outdated regimen may not answer a commissioning question.

• The NICE reference case counts NHS and Personal Social Services costs and relevant patient and carer health effects, and generally discounts both at 3.5%.

Reading the result

• The incremental cost-effectiveness ratio (ICER) compares the extra cost with the extra effect between options; the average cost per quality-adjusted life year (QALY) for each option does not show what additional spending buys.

• A cheaper, more effective option dominates, and a negative ICER alone cannot tell dominating from being dominated.

• NICE's current technology appraisal threshold range is £25,000 to £35,000 per QALY.

• A favourable ICER can still need substantial funding, so budget impact is assessed separately.

Tap to enlarge

Uncertainty