Table 3.

Recommendations for future research

Future research recommendation	Proportion rating as ‘high priority’a (%)	Priority rank
Distributional consequences Exploration of ways of establishing, presenting and considering the distributional consequences of family and caregiver health spillovers	94	1
Longitudinal study designs Adoption of longitudinal study designs to assess outcome measure sensitivity to change and allow for more in-depth investigation of the directionality and reciprocity of spillover effects	76	2
Joint utility elicitation methods Further examination of joint patient-caregiver utility elicitation methods, and estimation of QALY gains/losses at the family or household level	76	2
Context sensitivity of spillover magnitude/scale, social value and double counting Investigation of how: (i) the magnitude/scale of health spillover; (ii) the social value of these spillover health impacts; (iii) the functional form of the distance-decay function (used to project total spillover); and (iv) the risk and/or extent of double counting, vary on the basis of health condition/intervention, sociodemographic, cultural and other contexts, as well as in relation to the utility elicitation technique	75b	4
Modelling caregiver and family member health spillover as a function of available data Further examination of the potential for projecting/modelling spillover as a function of data such as patient health status, measures of sociodemographic status or caregiver burden	71	5
Inclusion of family and caregivers in decision models In the context of decision models, examination of the implications of inclusion of caregivers and family members as separate individuals (with their own utilities) vs as disutilities/utility increments applied to patient outcomes (to account for the effect of the patient intervention on caregivers and/or others in the family network)	69b	6
Adjusting for double counting Develop methodological approaches to adjust for double counting when applying spillover estimates from the literature to analyses, or when quantified in the context of primary data collection	53	7
Displacement Calculation of the average family and caregiver health spillovers in relation to displaced/foregone activity (to, e.g., inform thresholds empirically set on the basis of opportunity cost)	53	7
Modelling total network effects as a function of partial network outcome data Further investigation of how total network effects might be estimated from partial network outcome data (i.e., when net spillover effects are to be estimated but only data on the patient and closest network member can be collected)	44b	9
Outcome measure appraisal Development of novel generic preference-based measures aimed at capturing spillovers (beyond those in current use or development), underpinned by psychometric assessment of the construct validity and responsiveness of the measure among family and/or caregiver populations.	35	10
Social value weight estimation Investigation of methods for elicitation (e.g., person trade-off, discrete choice experiment) of spillover equity weights, giving particular focus to how estimated weights may vary by method used	29	11
Reporting bias Investigate the extent of reporting bias either related to social desirability or when using the same preference-based HRQoL measure for caregiver/family member HRQoL and patient HRQoL (including self-report and proxy report)	24	12

DCE discrete choice experiment, HRQoL health-related quality of life, PTO person trade-off, QALY quality-adjusted life-year

^aRanked based on the proportion of member votes falling within a ‘high priority’ bracket (i.e., ratings 5, 6, 7) on an 8-point scale (0 = lowest priority, 7 = highest priority)

^b16 of 17 members voted on this recommendation