Letter

In Reply: In his thoughtful comments on our report of the experience of caregivers of CATIE participants at study entry, Dr. Smith identifies two ways in which he feels the study could have been improved. First, he suggests that we might have usefully employed the distinction between objective and subjective burden frequently made in studies of family burden. Second, he argues that the findings would have greater utility if we had broken down our analyses by kin relation of the caregiver, rather than using as covariates marital status and whether the caregiver resided with the relative with schizophrenia.

Although we agree with Dr. Smith that the distinction between the objective stresses that caregivers encounter (for example, financial expenditures and time lost from work) and the subjective strains (for example, level of distress) is conceptually an important one, in our experience such distinctions have proved elusive. For example, in two previous studies of caregivers of people with bipolar disorder, we found that the objective and subjective burden scales of a widely used instrument were so highly intercorrelated (.76 to .83) that we combined them to avoid statistical redundance ( 1 , 2 ). Such high intercorrelations likely reflect measurement error, because most measures evaluate objective burden through caregivers' reports of problem behaviors, use of financial resources, and so on, while operationalizing subjective burden as the degree of distress experienced in relation to each reported problem. Objective burden ratings based on self-report lack external validation, and subjective ratings based on these "objective" reports are not obtained independently. Devising measures that accurately capture objective dimensions of burden, such as the economic cost to caregivers, will require not only incorporating methods of external validation but adjusting for factors that may bias or mitigate the cost, such as the ordinary provision of supports from one family member to another or the economic contribution from the consumer to the total family income ( 3 ).

Indisputably, the experience of caregiving differs for relatives with different kin relationships. However, our preliminary bivariate analyses found that other sociodemographic factors, such as living in the same residence and marital status, covaried with, and explained more variance than, kin differences. For example, although spouses and parents had significantly higher levels of problem behavior burden (factor 1) than siblings, these differences were no longer significant after the analysis adjusted for living situation, because spouses and parents more often reside with the consumer. This pattern of results is consistent with data cited by Dr. Smith that the amount of instrumental support provided—rather than the kin relation per se—explained the experience of burden.

Dr. Smith's comments illustrate the complexities and challenges confronting research on caregiving for individuals with major mental disorders.