[Socio-economic status and family profile indicators: methodological issues in studying the burden of caregiving for terminally ill cancer patients]

Abstract

Introduction: The quality of care received by terminally ill cancer patients depends heavily on socio-economic conditions and family resources, especially because of the current increase in home care assistance.

Objective: To validate the demographic information on educational level which is recorded in the registries of residents in Italian towns; to compare simple indicators of socio-economic status and family profiles with composite indicators in their ability to predict the economic and social impact of the disease.

Design: A two-level probabilistic sample of cancer deaths from the Italian Survey on Dying Of Cancer.

Participants: 2000 deaths were sampled; caregivers were identified and interviewed between 4 and 12 months after the patient's death.

Methods: We calculated Cohens kappa for educational level as reported in the registry, and in the questionnaire. We constructed a composite indicator of socio-economic status and family profile using a cluster analysis; its association with the impact on finances and quality of daily life was compared with that from the educational level reported in the questionnaire and with a previously derived indicator of family profile.

Results: The weighted kappa of the two sources used for educational level was 0.60 (CI 95% 0.55-0.64). Of the two indicators, educational level and socio-economic status (6 groups), only the later showed a significant association with the outcomes "difficulties" in sustaining the costs of treatment" and "use of entire savings for the illness". The composite indicator of family profile (7 groups) was significantly associated with all outcomes considered. Log-likelihood was significantly better with model using the composite and the prior indicators of family profile than in models without them. Models that included educational level and prior indicator of family profile (7 groups) were more adaptable than models with the composite indicators.

Conclusions: The good level of agreement between the two sources regarding educational level suggests that the registry is an adequate data source, when other information is lacking. Our study did not reveal which of the indicators we used is the best, at least with regard to the outcomes we considered.