The PICNICC (Predicting Infectious Complications of Neutropenic sepsis In Children with Cancer) Study



The aim of this project is to develop and evaluate an optimal risk stratification regime for paediatric and young adult patients with febrile neutropenia (FNP).


  1. To form a collaborative group to centrally collect individual patient data (IPD) from cohort studies of febrile neutropenia in childhood
  2. To undertake an IPD pooled analysis, using regression models to quantify the risk of adverse clinical outcomes according to clinical variables
  3. To develop and explore issues around the use of pooled IPD analysis in the development of predictive models
  4. To develop and validate a new risk prediction model
  5. To undertake an analysis of the potential economic impact of applying the new model in clinical practice


Evidence synthesis of observational studies of febrile neutropenia in children using an IPD approach.


  1. IPD pooled analysis. An international collaboration will be established to facilitate the central collecting, validation and pooling of data from appropriate studies. The risk variables and outcomes selected for inclusion will be defined by literature review and consideration of the collaborative group. Should consensus fail to emerge, a formal process ( Delphi questionnaires) will be used. The primary analysis will be based on the first recorded episode per patient to predict an absence of adverse outcomes (e.g. death, intensive care, or bacteremia). Subsequent episode data will be analysed to assess the independence of these data. The assessment will be by multivariate regression analysis, with variables entered after consideration of their clinical and statistical significance. The model will be tested by cross validation of intrinsic prognostic performance.
  2. An analysis will be undertaken the potential economic impact of applying the new model in clinical practice.
  3. Methodological refinement of the techniques of IPD pooled analysis will be undertaken (subject to the data obtained). The analyses may address the analysis of missing data, using different imputation models, the relative merits of prospective and retrospectively collected information, the comparison between episodic and patient-centred analyses and the use of categorical outcome variables.

Scientific opportunities

The development the methodology of IPD pooled analysis for prognostic models.

Medical opportunities

The appropriate selection of individuals for who FNP therapy could be safely changed from lengthy in-patient spells to short out-patient based treatment, reducing costs and inconvenience.

Conducted by: Bob Phillips1 , Lesley Stewart1, Ros Wade1, Alex Sutton2

1. Centre for Reviews and Dissemination; 2. Department of Health Sciences, University of Leicester


Phillips B, Ranasinghe N, Stewart LA; on behalf of the PICNICC Collaboration. Ethical and regulatory considerations in the use of individual participant data for studies of disease prediction. Arch Dis Child 2013; [Epub ahead of print]


Phillips B, Ranasinghe N, Stewart L. Ethical implications of sharing individual patient data for prognostic meta-analysis. 3rd International Symposium for Methods for Evaluating Medical Tests and Biomarkers. Birmingham, England, 15 - 16 July 2013


Commissioned by the Medical Research Council