Research Proposal and Specific Aims: Approximately half of pediatric oncology patients experience malnutrition as a direct result of cancer and its treatment. Malnutrition is associated with a twofold increase in mortality and a threefold increase in morbidity (e.g., infections and neutropenia). Previous studies of nutritional risk factors and outcomes have been limited by small sample sizes, an inability to stratify by malnutrition severity, and an inability to evaluate the impact of nutritional interventions such as enteral and parenteral calorie supplementation that are known to improve weight with better cancer outcomes. To improve nutritional outcomes for pediatric cancer patients, there is a critical need to determine which patients are at greatest risk for malnutrition and examine if current, standard-of-care interventions that increase weight also reduce morbidity and mortality.
Supportive care research in pediatric oncology lends itself to new data sources outside of the classic clinical trial model and fits with recent calls for evidence from pragmatic and alternative data sources. With the rapid adoption of electronic health records (EHR)over the past decade, unprecedented amounts of digitized health and healthcare data are now available for large-scale studies. Moreover, the aggregation of multi-institutional health data in large networks, such as PEDSnet (pedsnet.org) which has a repository of 25,000 cases of childhood cancer and over 1 billion data elements, enables rapid cycle, real-world evidence generation on pediatric cancer care, which complements evaluative clinical trials on innovative therapeutics and treatment strategies.5Our overall goal of this project is determine the patients at greatest risk for malnutrition and examine the effects of weight changes on important outcomes including morbidity and mortality. These are crucial first steps to the long-term goal of comparing the effectiveness of current supportive care interventions on reducing the impact of malnutrition on health outcomes for pediatric patients with cancer and testing these interventions as part of a clinical trial. For this initial research, we propose using innovative methodological approaches with a combination of classical statistical and machine learning methods that leverage multi-institutional EHR data in a national network of children’s hospitals. The specific aims of the project are to:
Aim 1: Describe cluster patterns of weight loss and growth in pediatric oncology patients as indicators of nutritional status. Hypothesis 1: Multiple distinct trajectories of nutritional status and weight loss exist for pediatric patients receiving cancer treatment Latent class and trajectory modeling will be used to characterize subgroups of patients based upon weight, body mass index and height patterns
Aim 2: Examine the effects of changes in weight on mortality and non-fatal, serious adverse events (delays in chemotherapy, sterile site infections, duration of cytopenias, acute organ toxicities, and healthcare utilization [days admitted to an intensive care unit, days admitted to a hospital]), controlling on cancer type and risk stratification. Hypothesis 2: At least one weight trajectory cohort has >5% increase in all-cause mortality compared to other cohorts and worse non-mortality outcomes and higher healthcare utilization.Longitudinal modeling compatible with clinical understanding of cancer treatment protocols will be employed. Outcomes will be evaluated using a combination of classical statistical and machine learning methods.
Significance: This work embodies T3 translational research generating real-world evidence for pediatric cancer research with a particular focus on outcomes of oncology supportive care. This work is independent of and complementary to current research performed through trial networks and registries and has unique potential to facilitate scientific investigations to improve survival for children, adolescents, and young adults with pediatric cancer. This project is the keystone in a set of planned research activities to generate new observational research, develop new interventional trials, and facilitate current best practices through informing both quality improvement and implementation science initiatives all of which target improvements in nutritional status and outcomes for pediatric cancer patients.
Innovation: Our systematic and hypothesis-driven approach based on our conceptual framework is unique and innovative. This project will add to our body of knowledge about one of the most common serious adverse effects of cancer therapy, malnutrition. Furthermore, the use of PEDSnet allows us to leverage resource utilization information to examine nutritional trajectories within the context of treatment differences and complications to generate real-world data.