Appropriateness of end-of-life care in children with serious illness: The development and measurement of pediatric-specific population-level quality indicators for big data

INTRODUCTION
A quality evaluation of children’s end-of-life care is needed. Evaluations have already been performed for adult’s end-of-life care, using routinely collected administrative healthcare data for the whole population. Some international evaluations with routinely collected administrative data have been performed for children, yet these studies use quality measures constructed for adult end-of-life care for measurement. The adult-focused measures likely do not properly reflect the issues relevant for children’s end-of-life care. There are currently no pediatric- specific quality indicators for administrative healthcare data. Quality indicators tailored specifically to the child at the end of life are requested nationally as well as internationally. The healthcare use in children’s end-of-life has gained interest in recent decades. There is significant evidence that part of the children with serious illness, such as cancer and neurological conditions, suffer from heavy symptom burden in the last days of life. Concerns are present for overly aggressive and futile healthcare provision, while comfort measures are shown to be available to relieve child and family yet possibly not provided sufficiently. Current studies on the topic indicate that the quality of children’s end-of-life care may vary widely for clinical and sociodemographic characteristics of children and families. Differences in healthcare provision at the end of life have been identified previously, for example for age and nationality. An evaluation of the quality of care with quality measures tailored to children can guide future steps in practice, research and policy for children’s end-of-life care: administrative data can provide a bird’s-eye-view, that cannot be obtained from individual practice, as children’s deaths from serious illness are relatively rare and most healthcare providers only ever experience a handful of cases. Administrative data is well-positioned to analyze disparities, as the full population is included and subgroups that normally are hard to reach are included in the cohort, such as families with a lower income or migrant background.
RESEARCH AIMS
This dissertation had the overall goal to evaluate end-of-life care for children with a serious illness, such as cancer, neurological conditions, and genetic and congenital conditions, using quality indicators and administrative healthcare data.211
The specific goals of this dissertation were: 1. To develop population-level pediatric-specific quality indicators for appropriateness of end-of-life care for children dying with cancer, neurological conditions, and genetic and congenital conditions,2. To measure these indicators within Belgian administrative databases; and to look into possible sociodemographic, clinical, and regional differences for appropriateness and inappropriateness of end-of-life care. Due to the use of administrative databases, the focus was limited to medication, treatments, and care providers, as only these aspects are measurable within the data.
METHODS
For research aim 1, development of the pediatric-specific quality indicators, we performed a systematic literature review and a RAND/UCLA panel. The RAND/UCLA consensus method consisted of expert interviews, an electronic survey, a group discussion round, and a second electronic survey. Interviews and panels included pediatricians, nurses, psychologists, physiotherapists, pharmacologists, care coordinators, general practitioners, social workers from hospitals, care teams, and general practice. Three indicator sets were developed: one for children with cancer, one for children with neurological conditions, and one for children with genetic and congenital conditions. For research aim 2, measurement of the pediatric quality indicators within population- level administrative databases, we first linked several Belgian databases. Then, we calculated and described the quality indicator results. For the sociodemographic, clinical, and regional differences in appropriateness of children’s end-of-life care, we performed analyses of variance as well as logistic regressions.

MAIN FINDINGS
In Chapter 1, we presented the results of a systematic review, which identified the healthcare interventions that are associated with improved and/or reduced quality of life for children at the end of life. We found that the current evidence base is broad (20 interventions were studied), yet studies were limited in quality, and many showed bias for design and execution. Palliative care, certain comfort and pain medications and treatments, and symptom monitoring were associated with improved children’s symptoms and quality of life at the end of life. Curative therapies, such as chemotherapy and stem cell transplant, were associated with decreased quality of life at the end of life for children. In Chapter 2, we described a RAND/UCLA consensus method to develop the pediatric-specific indicators. Across all illness groups, the quality indicators fell into four themes: 1. Treatment medication and monitoring (with quality indicators such as physiotherapy and specialized comfort medication), 2. Place of care and death (containing quality indicators such as home death), 3. Care services and providers (containing quality indicators such as contact with a family physician or having continuous care relationships), and 4. Administrative measures (containing quality indicators such as receiving palliative status (and therefore being administratively entitled to higher reimbursements)).
Chapter 3 to 5 described the measurements of the pediatric-specific quality indicators for each of the three illness groups .For potentially appropriate care, we found that reimbursements for comfort treatments were often not provided to children at the end of life. For instance, palliative care reimbursements were provided to less than one fifth of children. Half of children with cancer died at home (47%).Continuous care was provided to over half of all seriously ill children at the end of life. Over two thirds of children with neurological conditions received reimbursements from specialist physicians in the last month before death. However, less than one fifth of seriously ill children at the end of life received reimbursements from a general physician in the last month before death. Multidisciplinary care in the last month before death was low. Administrative measures were not provided often for children across illness groups. Palliative status was provided to circa one fifth of children with serious illness at the end of life. Potentially inappropriate care were usually not frequent in seriously ill children at the end of life. No or fewer than 5% of children received a new dialysis, surgeries or old-generation reimbursements for nausea within the last month before death for any of the illness groups. Around one fourth to one third of children received 2 MRI’s, X-rays or CT scans at the end of life, and blood drawings were very frequent, that is they were present for half of children across illness groups. Admissions to the Intensive Care Unit were provided to up to one third of children.
For clinical, sociodemographic and regional differences, children with neurological conditions, showed differences for appropriateness for disease categories: disorders of the central nervous system and movement diseases showed lower scores for appropriate care. For children with cancer, there were differences for region and nationality: One rural healthcare region showed higher appropriateness, and children with a non-Belgian background received more inappropriate care. For children with genetic and congenital conditions, appropriateness differed for various regions.
DISCUSSION OF MAIN FINDINGS
Our discussion pointed out that there was frequent and continuous contact with physicians, but less with other care professionals, and little palliative care provision. We indicated that while the numbers may underestimate the actual care provision due to the partly philanthropic nature of children’s end-of-life care, these numbers may provide a starting point for further inquiry into the palliative care provision and involvement of other care providers. Additionally, we discussed that findings for inappropriate care paint a complex picture of intense care at the end of life in children. Treatments such as chemotherapy, commonly seen as futile treatment and a staple quality indicator for inappropriate care in adult indicator sets, were not seen as a proper indicator for inappropriate care in three recently developed quality indicators for children at the end of life. The treatments may not be provided, such as dialysis, as our numbers as well as international measurements show, or the treatments maybe used to provide long-term comfort, such as chemotherapy. Diagnostics and blood drawings were an exception to the rule and could possibly be decreased at the end of life due to futility. However, we stressed the limitation that our study does distinguish between children who died with a foreseen end of life and those without .Lastly, we pointed out that differences for region, disease group, and nationality may point to differences in evidence base, structural support and family preference. It was hypothesized that a difference in knowledge on symptomatology and treatment may underlie these differences, especially region and disease group. Differences for children with a non- Belgian background may be goal-concordant and require further insight.
CONCLUSION
This dissertation found that children with serious illness at the end of life may require amongst others more comfort care provision, increased care provider support and follow-up, decreased imaging and blood drawings, and increased administrative support for families as well as care providers. Differences in appropriateness and inappropriateness for disease category, and nationality are present and may be caused by underlying knowledge gaps.
IMPLICATIONS
Administrative and legal measures for the support of palliative care provision and pediatric liaison teams are encouraged. The development of interventions with attention for different subgroups, such as disease trajectories, is also advised. Learning and improvement strategies could be implemented using the quality indicators as flexible bench marks. Competences for children’s end-of-life care are best added to the current standard medicine curriculum or specialized palliative care trainings provided in Belgium. Further empirical research on pain and symptom management is best carried out for both non- cancer and cancer conditions. It is paramount to include the perspectives of child and family in further studies.