The need to initiate PPC in a timely fashion through improved identification of children and adolescents who might benefit has been highlighted [20, 21]. This study establishes early face and content validity for a new instrument to facilitate this identification in clinical practice.
The PaPaS Scale builds on the taxonomy of Together for Short Lives, formerly ACT  and underscores that PPC is not limited to end-of-life care. Neonates and infants were excluded as their disease trajectories are mostly short, two thirds of them dying during the first weeks of life [29, 34] on the neonatal intensive care unit without episodes of being in their ‘natural’ environment, the family’s home. Even if neonates constitute an important and numerically large proportion of children who could benefit from PPC, they have different needs and we decided not to include them.
The indicators proposed in our instrument operationalize in more detail the approach suggested by Rushton et al. . These authors recommended triggers to identify patients for PPC-conferencing, including: 1) limited life span; 2) ‘surprise-question’ – sudden death within 6–12 months; 3) increase in hospitalisations during the past 6–12 months; 4) major clinical events; 5) symptoms that have changed the frequency of clinic visits; 6) change in response to treatment; 7) conflicts about goals of care. In our experts’ interviews three of these triggers (prognosis, events and symptoms) plus increasing needs of the child and its family were brought up in the first part of the interview, which included a spontaneous reflection on how they identify children with PPC needs. However, with respect to life expectancy (prognosis), the discussions were highly controversial. While several experts of Phase 1 suggested omitting this domain, participants of Phase 2 strongly voted to keep it as it seemed important that this question receives more attention in decision-making. This controversy is interesting; it may reflect cultural differences or the various stages of national PPC achievements, - particularly in Switzerland where PPC is only starting to develop and to be recognized. Thus, in Switzerland, life expectancy may play a stronger role in decision-making towards the introduction of palliative care as compared to UK, the US or Canada where PPC starts to be integrated into care earlier in the course of a disease [20, 27].
Predictive factors and events focusing on the diagnosis of dying have also been evaluated by others [34, 35]. Brook and Hain  proposed that the following candidate factors be included in future studies: frequency of hospital or intensive care admission, episodes of acute illness without recovery to the child’s usual best level of health and physiological changes such as decreased oral intake. Feudtner et al. developed a paediatric mortality prediction model to analyse the likelihood of death during hospitalisation or 1-year post discharge . Among several predictors, the frequency of hospitalisations (>3) during the year before the index hospitalisation and the risk of death were strongly associated. Increase of hospitalisations was also emphasised by some experts from our study and was therefore included in Version 3 of the instrument.
The suggestion by some participants to exclude the domain on preferences of the patient and the family was surprising. This may reflect the difficulties that clinicians (and parents) experience in halting treatment that is futile from a medical perspective. A recent study , for example, found that more than a third of children with cancer continued to receive cancer-directed treatment after the parents had realised that there was no realistic chance of cure. However, parents who felt that their child had suffered due to cancer treatment prior to death were particularly unlikely to recommend such a treatment to other families. Our instrument could facilitate such a discussion by a broader exploration of the child’s situation, suffering and needs.
Some reservations concerning the instrument were expressed. There was scepticism about the extent to which clinicians would use the instrument. The challenges in assessing presence and severity of symptoms in children, or estimating life expectancy were highlighted. Written instructions and appropriate training on how to use the instrument will solve some of these concerns. Notwithstanding these, it was felt that the instrument could be a useful clinical and educational tool, increasing earlier activation of PPC and heightening awareness of palliative care needs amongst health professionals. The educational aspect of the tool was particularly seen in the fact that the use of the instrument would evoke discussions which may clarify what palliative care can add in the care of an individual child and its family, and thus translate the definition of palliative care into daily clinical work with severely ill children. Our study suggests a potential role for developing a scoring system linked to this stepwise approach. The predictive validity and clinical effectiveness (impact on children’s quality of life) of such a system will need to be shown.
There are several limitations. This is only the first phase in the development of a new instrument. The interviewer (EB) was also the person who developed the model and analysed the data, thereby potentially introducing bias. The discussions with the supervisor (JP) during the development of the instrument’s model and following the interviews and focus group for checking final themes and sub-themes of data analysis should have reduced the impact of this. The focus group included only clinicians from a German-speaking Swiss hospital, which affects the generalizability of the results. However, testing in different settings and countries is planned. In addition, the perspective of affected families, particularly parents, has not been incorporated yet. This is planned for the following steps of the instrument’s validation.