Predictors of the Extent of Agreements between Terminally ill Cancer Patients and their Family Caregivers in Taiwan

Abstract

Difficulties with recruiting patients and severe attrition are common problems in studies of pal liative care and terminally ill patients. Proxies (e.g. family caregivers) have been suggested as an alternative source of information about the patients' health related quality of life (HRQOL). Such proxy assessments should be in agreement with the patients' own views. Varying degrees of agreement have been found across HRQOL do mains with the best agreement observed for more overt symptoms like dyspnoea while poorer agreement has been observed for more subjective symptoms like pain [1]. Investigation of factors affecting the level of agreement may identify subgroups of patients for whom proxy assessments are appropriate (i.e. with good agreement between patients and proxies) or inappropriate (poor agreement) and it may help improving the agreement between patients and proxies by identifying possible reasons for dis agreement. The study by Dr Tang [2] aims at identifying such factors affecting the agreement between Taiwanese terminally ill patients and their family caregivers. As noted by Dr Tang, limited information is available about factors affecting the agreement, particularly in non-Western cultures. Hence, such a study seems clearly relevant. The primary reservation regarding this article is the use of overall measures only. These overall measures are composed of items regarding several diverse symptoms and HRQL domains. The use of such overall measures assumes (implicitly if not explicitly) either similar agreement across the symptoms in the measure or that possible differ ences in agreement are unimportant. For a mea sure like the SDS comprising items about 11 important but very diverse symptoms neither of these assumptions seem reasonable a priori. For example, Dr Tang found significantly better agreement regarding the SDS for patients receiv ing pain treatment than for patients who did not [2]. Dr Tang hypothesized (based on Lobchuk arid Kristjanson [3]) that this may be because caregiv ers are better at rating symptoms requiring actions on the part of the caregiver. If this is the case, it seems quite likely that administrating pain medi cation affects the caregiver's ability to assess pain. But why should extra attention from the caregiver regarding pain improve the agreement for other symptoms like problems with 'concentration' or 'outlook'? Without further investigation of the agreement for individual symptoms verifying that administration of pain treatment actually has the same effect for all symptoms, it does not seem likely that this is the case. Hypotheses like this could be investigated much better if the agreement were assessed separately for each symptom; if the agreement regarding pain is significantly better for caregivers who administrate pain medication than for caregivers who do not, this would indicate that the hypothesis was correct. The interpretation for the overall measure seems less clear. For example, how should the finding for the overall measure be used in practice? Does it mean that caregiver reports can be safely used for all symptoms included in the SDS for patients receiving pain treatment, but that caregiver reports are imprecise for all of the symptoms and therefore cannot be used for patients who do not? Findings of no difference in agreement may also be difficult to interpret for an overall measure. Does such a finding mean that the agreement is equally good in the two groups compared for all symptoms or problems included in the measure? Not necessarily. For example, if no differences are found between male and female caregivers for an