Does the Informal Carer Identification Method Matter? Evidence from Self-Declaration and Time Diary Approaches.

Objectives: Impacts on informal carers are increasingly being incorporated into cost-of-illness and cost-effectiveness analyses. However, little is known about whether the method used to identify carers affects the estimated impacts. We compare a novel time diary technique to a common self-declaration question for identifying carers. We investigate whether it: (1) detects more and different carers, and (2) if carers across identification techniques have different mental health outcomes.

Methods: We use the Innovation Panel component of the UK Household Longitudinal Study, which records all activities performed in two 24-h periods and contains a rich set of individual characteristics. We use regression analysis to compare the number and characteristics of carers identified across the two methods. We then use the doubly robust approach of entropy balancing combined with regression adjustment to estimate the mental health impacts of caregiving across both methods.

Results: Among 1055 individuals, we identify 261 carers by at least one method. The self-declaration method fails to classify 16% of individuals identified as carers through time diary data. We find that carers identified by the time diary have a 1.24 (p < 0.05) higher score on the 36-point General Health Questionnaire (GHQ) scale in the subsequent survey wave compared with similar non-carers. For self-declared carers, the estimated difference in GHQ score is 0.36 (p > 0.1), a smaller and statistically non-significant association compared with that observed among time diary-identified carers CONCLUSIONS: The mental health impacts of caregiving may be underestimated when carers are identified by self-declaration. Supplementing self-declaration with time diaries may offer a means of including more carers. Future research, if only one method is applied, should more carefully consider the means of identifying informal carers and the implications that the use of one method may have on conclusions.