Validity of caregiver-reported hospital admission in a study on the quality of care received by terminally ill cancer patients F. Chini a, * , P. Giorgi Rossi a , M. Costantini b , M. Beccaro b , P. Borgia a a Agency for Public Health, Lazio Region, via di S. Costanza 53, 00198 Rome, Italy b Unit of Clinical Epidemiology, National Cancer Institute, Genova, Italy Accepted 13 February 2009 Abstract Background: The clinical status of terminally ill patients often makes it impossible for them to report information directly, which in- dicates the need to rely on information from indirect sources, such as from caregivers. This information needs to be validated, and particular attention must be given to the accuracy of recall. Objective: The objective of this study is to evaluate the agreement between caregiver-reported hospital admissions with the data reported in the regional hospital information system. Methods: A two-level probabilistic sample of cancer deaths from the ISDOC (Italian Survey on Dying of Cancer). For the 2,000 deceased sampled, hospitalizations were identified from the administrative data and reported by the caregivers via a questionnaire. We calculated Cohen’s kappa, sensitivity and specificity using the regional archives as the gold standard. A multivariate analysis was performed to assess possible variables that may influence agreement. Results: We interviewed 1,271 caregivers. Sensitivity and specificity were, respectively, 82% (95% confidence interval [CI] 5 79e84) and 65% (95% CI 5 60e69). Kappa statistic was 0.46 (95% CI 5 0.40e0.52). Multivariate analysis showed that agreement increases with educational level and caregiver age. Conclusion: The validation of caregiver’s recall for medical procedures has important implications for research and care, because often it is the only information source we can rely on. The questionnaire showed good sensitivity and poor specificity concerning real hospital- izations, and had moderate degree of agreement with the data reported in the administrative data. Ó 2010 Elsevier Inc. All rights reserved. Keywords: Caregiver reports; Terminal care; Accuracy of recall; Patient admissions; Agreement; Medical records 1. Introduction The World Health Organization considers the problems that emerge in the last stages of life to be among the six biggest priorities in the field of heath care, and in Italy, the National Health Authority plans to invest resources in this sector of health care. To appropriately plan and evalu- ate health care interventions it is a priority to have a clear picture of the needs of patients and their families, and the type and quality of assistance available to them. The problems and needs of terminally ill patients and their families are particularly difficult to evaluate, because of the nature of the treatment objectives and the difficulty in the val- idity and trustworthiness of assessment tools and modalities. The clinical status of the patient often makes it impossible to gather information directly, giving rise to the need to rely on information from indirect sources, for example, from an informal caregiver, the person closest to and best informed on the last phase of the patient’s life. Interviewing the care- giver instead of the patient himself has advantages and disad- vantages. The advantages include the fact that often the person who cares for the daily needs of the terminally ill are better informed, both about the disease and the patient’s status; in addition, this makes it possible also to study the im- pact that the disease has on the caregiver personally. The most evident disadvantage besides the strong emotional involve- ment is the caregivers’ perspective: they do not physically ex- perience the disease and inevitably perceive it in a different way than the patient does. Several studies have shown that agreement is poorest for aspects of the patients’ experience that are more subjective in nature, such as, the patients’ pain, feelings, and thoughts [1]. In particular, pain and mood dis- turbance were overestimated by the caregiver. When information is obtained in this way it needs to be validated, and particular attention must be given to the accu- racy of recall. Many studies have evaluated recall accuracy by measuring the agreement between two independent * Corresponding author. Tel.: þ39 0683060307; fax: þ39 0683060463. E-mail address: chini@asplazio.it (F. Chini). 0895-4356/10/$ e see front matter Ó 2010 Elsevier Inc. All rights reserved. doi: 10.1016/j.jclinepi.2009.02.006 Journal of Clinical Epidemiology 63 (2010) 103e108