OBJECTIVES: Healthcare interventions may have indirect ‘carer’ or ‘spillover’ impacts on the quality of life of family and carers. Where interventions impact survival, an assumption, implicit or explicit, is made regarding carers quality of life post-patient’s death (pdQOL). This assumption can have a profound impact on cost-effectiveness and is contestable. Setting pdQOL to zero might be seen as double counting survival benefits whereas setting it to general population utility leads to increased survival reducing spillover effects.

METHODS: Criteria are suggested that might identify ‘useful’ estimators of incremental carer utility: Carer effects should be positive if patients spend in more time in better health states; Where there is no change in time spent in ‘better patient health states, carer effects should be zero, regardless of survival effects; Impact grief on carers should be considered separately. Using simulated data for various scenarios regarding survival and time spent in ‘responding’ health states, estimators of spillover effects where carer pdQOL is set to: Zero; General population utility; Worst patient state carer utility are assessed. The implementation of care giver and family utilities as absolute and relative dis-utilities is also considered.

RESULTS: Setting pdQOL to worst patient state carer utility leads to an increase in incremental carer utility where there is increased survival in ‘better’ patient health states. There is no change in incremental carer utility with decreased survival in the worst patient health state, therefore the estimator does not suggest, ceteris paribus, carer preference for reduced patient survival.

CONCLUSIONS: Setting pdQOL to Worst patient state carer utility might reflect likely societal or carer value judgements and is a useful estimator for multistate models; certainly where the majority of patients enter the worst health state before death. Further research on societal preferences with respect to estimation of spill-over effects would be useful.