DISCUSSION

One important step toward understanding the illness trajectory of cancer patients is to better understand the predictors of place of death (Bruera et al., 2002), in order to deliver adequate holistic care. Unfortunately for our population, the information regarding children's and families' preference for place of death was not available, making it difficult to determine whether they died in a setting of their choice.

Adam (1997) described that during the final 48 hours of life, patients experience increasing weakness and immobility, loss of interest in food and drink, difficulty swallowing, and drowsiness. The priority is then given to the control of symptoms and family support. This is a time when levels of anxiety, stress, and emotion can be high for patients, families, and other caregivers. It is important that the healthcare team adopts a sensitive yet structured approach. Reassurance that help is available can assist the family in choosing a place of death.

Many authors have described that age and gender are significant factors associated with location of death (Grande et al., 1998; Bruera et al., 1999). According to these studies, female and older patients were less likely to die at home, because of difficulties in accessing home care and the suggestion that men were less efficient as carers. In our study, the risk of dying at home increased independently only in the male gender. A boy had a higher risk to die at home than a girl (OR = 4.25), independent of their health insurance category, educational background of the mother, or the educational background of the home care provider. We could suggest that this result may reflect our cultural beliefs that males are taught to be strong and fearless, even though they tend to avoid hospitalization and painful procedures. Girls seem to be less fearful about hospitalization in our experience.

The concepts and reactions to fear and death in children vary according to the age as they are in particular cognitive phase (Anonymous, 2000). In our study, age was not statistically significant, although we have identified a tendency for more children between 6 to 10 years old to die at home (78.6%) than did the other ages (p = 0.362), suggesting that in this category, children may be more likely to start participating in the decision-making process and wish to stay at home.

Family composition was not significantly associated with location of death, although our findings have shown an increased number of home deaths when the patient was a second child. It may be related to the involvement of the older brother/sister in care giving. Cantwell et al. (2000) showed that the presence of more than one caregiver was also predictive of home death.

In our population, factors such as pain and low play performance status were not significantly associated with hospital deaths, suggesting that pain was not responsible for hospitalization, probably due to adequate pain control. As Bruera et al. (2002) found in their study, we believe the establishment of an integrated care system with interdisciplinary team and community support more likely increases the proportion of patients whose place of death reflects their preference, and probably results in more deaths at home. Although we had increased the number of desired home deaths, it still remains a challenge to meet patients and families' requests (Kurashima & de Camargo, 2003).

A home death involves many economic, cultural, supportive, and family-centered issues. In our society, most of the patients are from public insurance, which does not include access to home care. Even though we are still not able to provide hospice care or home care for all children, neither private nor public health insurance was significantly associated with location of death. As Bruera (1993) stated in his study, we also suggest that the increased emphasis placed on utilizing a family member as caregiver, and local resources have been helped us offer our children feasible care. These findings also suggest that the patients' preference for a home or a hospital death might be realized on a more consistent basis if a palliative care program were available.

The interdisciplinary team should advise and assist families in responding to the child's needs. Early development of a shared decision making program might enable children to receive end-of-life care with parents' involvement, and ultimately die where they feel more comfortable. Sharing family support and communication have been associated with positive long-term bereavement adjustment (Anonymous, 2000).

Although we have no previous analysis regarding location of death before the creation of the palliative care team, our study shows that we had decreased the number of hospital deaths with long periods of hospitalization. This may reassure the importance of a team approach to empower the patient and the family to achieve their own wishes.

Johnston (1999) has shown that taking care of the dying patient and his family is primarily a responsibility of the nursing team. In our interdisciplinary group, the nursing role begins with the recognition of the child's needs, an attempt to respect the child's wishes, the coordination of efforts to maintain a level of dignity, quality of life, and quality of care wherever he decides to be. Future studies may also identify the patients' and families' wishes, other specific characteristics, and the likelihood of a home/hospital death, giving us the resources to better address their wishes.