Results

A total of 662 patients were admitted to the PICU during the nine month period, representing 3409 consecutive patient bed days of study. Mean age of the patients was 35.7 (SD 54.7) months (median 9 months) and mean length of stay was 5.2 (8.4) days (median three days). A total of 77.8% of the patients (515/662) required mechanical ventilation. The mean PRISM II score was 11 (median 10). The overall percentage of bed days utilised by patients fulfilling any one of the definitions of futility was for 104 (3%) of 3409 bed days. A total of 34/662 of the patients (5.1%) fulfilled at least one of the criteria for futility.

Thirty three (0.9%) of the patient bed days met the operational definition for imminent demise futility as measured by PRISM II with a mortality risk exceeding 90%. Eighteen of the bed days were used by eight patients who died; the remaining 15 bed days were by six patients who survived. The majority (27/33) were patient bed days with mortality risks between 90% and 95%. Only six patient bed days were utilised by patients with daily mortality risks between 96% and 99%. Sixteen patients satisfying criteria forlethal condition futility used a total of 60 bed days (1.7%); eight of these died following withdrawal of treatment. These were mainly patients with disseminated malignancy or malignancies unresponsive to first line therapy (table 1). In the qualitative futility group no patient was in a vegetative state before admission to PICU or during their PICU stay and only 16 bed days were occupied by patients fulfilling criteria for brain death, all of whom died.

Fifty one (7.7%) of the 662 patients admitted died in PICU; 37 (72%) of the deaths were associated with withdrawal or withholding of therapy, and 14 died despite aggressive resuscitative efforts. Of the 34 patients who fulfilled any of the criteria for medical futility, 19 (56%) died; of these 15 had some form of limitation of treatment (nine withdrawal or withholding therapy, six extubation after fulfilling brain death criteria), and only four died after failed full resuscitative efforts. Patients who fulfilled criteria for futility had a significantly higher mortality rate than non-futile patients (p < 0.013). Table 2 shows characteristics of patients who fulfilled criteria for futility and those that did not.

Discussion

The intensive care unit symbolises the dilemma of modern healthcare—good outcomes can be achieved in the most critically ill, but at a great expense. High and continuously escalating expenditures in the ICU make rigorous economic evaluations necessary to achieve maximum cost effectiveness.15-17 Research from adult intensive care has shown potential savings of $2–5 million (£1.2–3.0 million) per year by identification of care that is futile.8 Treating “futile” patients predicted to die in ICU may cost up to $210 423 (£129 651) per survivor.9 While accounting for only 13% of all patients, they consume up to 32% of total resources.8 Sachdevaet al, however, showed that these potential cost savings might not apply to the paediatric ICU population as only small amounts of resources were expended on futile care.10These issues have not been investigated in a PICU setting outside North America. Using similarly broad definitions for futility we found that relatively few PICU patient bed days were used for futile care. The strength of the current study was that it was prospectively collected and included a proportionally large number of patient bed days. It was also conducted over an extended duration, to account for a potential skew in data caused by seasonal admission trends in PICU. Our findings appear to agree with the conclusion of Sachdeva et al that futile care was not a source of considerable resource consumption in PICU.

The lack of agreement with the findings from the adult ICU population is probably multifactorial. Our study suggests that paediatricians are readily recognising “futile” care early and dealing with it in an ethically appropriate manner. The majority of deaths in the current study involved some form of withholding or withdrawal of therapy, similar to other paediatric studies on end of life decision making.18 ,19 A combination of worsening severity of illness, poor response to medical therapy, and poor future quality of life were often the reasons for this assumption. It is likely that the practices for treatment limitation in adult ICUs were much more limited during the period when the adult literature on medical futility was published, although recent reports suggest a substantial increase in these practices.20 The wide use of treatment limitation in the current study may actually limit the amount of care given to critically ill children, whom might have deteriorated further and gone on to fulfil any one of the broad definitions of medical futility employed. However, restricting care in the interests of cost containment would stretch the limits of ethical behaviour. These decisions should always be made in the “best interest” of the child.18

Prognostication systems like the PRISM score used in our study have been shown to reliably predict mortality in the PICU population.21 Additionally any other patient days associated with PRISM scores exceeding 90% were also included. Although values obtained from days other than the admission day have not been previously validated as indicators of mortality, Changet al have shown that changes in daily individual physiological data were able to predict death accurately in an adult ICU population.2 The cut off using mortality risk of greater than 90% has been described as too low in assessing medical futility10 as shown in our study, where 43% of these children with high mortality risk were able to survive. Improvement in survival probabilities in PICU may necessitate the use of higher values to define futility. These would further decrease the extent of bed days fulfilling the definition of imminent demise futility.

The definition of lethal condition futility captured the greatest number of patient bed days. This definition attempted to identify children that were likely to die from their underlying disease, although not necessarily in the current admission. These were mainly children with malignancies failing first line therapy. In contrast to adult cancer patients failing first line therapy, the outcome in children is better with rescue therapy such as bone marrow transplantation.22 ,23 It could be argued therefore that treating such children does not constitute futile care. The use of these definitions would therefore lead to an overestimation of the extent of futile care provided.

There are several limitations to this study. Firstly, it was carried out in a single institution and might not be representative of the remainder of the UK. The study unit, being a tertiary unit has active ethical discussions among the intensivists, nursing staff, and family, thus potentially limiting the amount of futile care provided to children who are deemed incurable by actively withholding or withdrawal of therapy. Secondly, the process of gatekeeping, where potentially futile cases are not admitted to the PICU may further lessen the estimate of medical futility. Although length of stay or bed days as used in our study has been shown to be a reliable indicator of resource utilisation,24 this was not corrected for severity of illness. An ill child who succumbs may use significant resources despite a short PICU stay, thus leading to possible underestimation.

In conclusion, despite using broad definitions that were likely to lead to overestimation of futility, only small amounts of resources were expended on futile care. Generalising potential cost savings from limitation of futile interventions in adult studies to the paediatric population is unwarranted. Ethically we are reassured that the majority of care provided to critically ill children appears to be appropriate. Paediatricians are recognising care that is futile and have taken steps that are ethically appropriate to limit it. Further potential cost savings to be achieved from identifying and terminating care that is futile would thus be negligible in a paediatric critical care setting.