The PICNICC (Predicting Infectious Complications of Neutropenic sepsis In Children with Cancer) Study
Optimizing risk predictive strategies in febrile neutropenic episodes in children and young people undergoing treatment for malignant disease
Febrile neutropenia (FNP) is a complication of cancer therapy. It is the occurrence of a fever in a patient who is immunosuppressed following cancer therapy, and is potentially an indication of serious infection. It is the second commonest reason for hospital admission among children & young adults with cancer, with approximately 4000 episodes of FNP occurring annually in the UK. Since this was recognised in the early 1960s, patients have been aggressively treated with hospitalisation and intravenous antibiotics. This policy has improved the mortality rate dramatically, from 30% in the 1970s to 1% in the late 1990’s. There remain many episodes of FNP, possibly two-thirds or more, in which no significant infection is identified, and in these patients this management approach is unnecessary. Clear risk stratification would enable us to differentiate between these extremes.
In the UK, the National Institute for Clinical Excellence (NICE) guidance document “Improving outcomes with children and young people with cancer (2005)” called for "the development of robust methods of risk stratification in the management of FNP". At present there are many different policies for the management of FNP in the UK with lack of agreement about how risk stratification, if any, is used.
There have been a number of studies in small numbers of patients trying to define which patients need the aggressive treatment, and which could be treated with much reduced intensity of therapy. These risk stratification models vary between the studies, and none have been tested extensively. Small numbers of cases lead to uncertainty about the results of studies: any conclusions drawn may be chance findings, rather than real differences, and this may explain why different models have been developed. Secondly, models which have not been tested may not be usable on a larger scale.
To develop our understanding of risk stratification of febrile neutropenia we have undertaken two systematic reviews with meta analyses of the previously undertaken studies. Meta-analysis is the technique of collecting and statistically synthesising data from multiple reports. In doing this, it allows larger numbers to be studied, and the data is drawn from a variety of sources. These reviews have demonstrated that a wide number of potential risk stratification models exist, and some research has shown that extra blood tests may be helpful, but we can’t yet conclude which of the options is most effective.
Our systematic reviews have demonstrated that the published data is incomplete (missing vital information for truly meaningful meta-analysis), and there remain concerns that the reports are subject to significant publication bias (where studies showing prognostic markers to have ‘highly significant’ responses being more likely to be published). There have been particular statistical problems, such as the categorization of continuous outcome variables. (This refers to the grouping of data that can be measured on a scale, such as age or weight. The groups have varied between reports (e.g. “less than 5 years old vs. more than 5y” in one study, and “less than 12 years old vs. more” used in another study), which means they cannot easily be synthesised, and the reasons why a particular cut-point has been chosen may themselves be biased. There is the potential that other simple simple flaws exist too (such as inconsistent reporting of apparently similar outcomes).
In situations similar to the problem of risk stratification, an approach has been developed which uses the raw data (“individual patient data” or IPD) pooled analysis for the synthesis. There are a range of theoretical reasons why this approach will be better, and overcome some of the problems present in using previously published data alone. This part of the PICNICC study will use this IPD approach, by creating an international collaborative of researchers in the area to combine and share the data, with the objective of creating and testing an effective risk stratification model.
As the techniques of IPD pooled analysis are still relatively new, part of the project is to test theories that exist about how to undertake the synthesis. The exact issues will depend on the data available, but may address issues regarding the development of clinical decision rules in a meta-analytic setting (rather than just using data from one source), the analysis of missing data using different statistical imputation models, the relative merits of information collected prospectively and information taken by trawling through from medical records, assessing data from people who have few or many episodes of FNP (to determine if there is a difference in how risk should be assessed) and the use of categorical rather than dichotomous outcome variables (for example “Needed ICU”, “Needed oxygen”, “Identified infection” and “Well - no complications” compared with “Well” vs. “Not well”).This project was conceived to to develop a clinically useful decision rule and to advance our understanding of the statistical techniques, and was awarded funding as part of a Medical Research Council (MRC) Research Training Fellowship in October 2008.
The PICNICC (Predicting Infectious Complications of Neutropenic sepsis In Children with Cancer) Study Clinical Advisory GroupThis group will provide advice throughout the project. This group will help:
- ground decisions on data in clinical practice
- explore the opportunities for family/patient centered analyses
- advise on methodological issues
- provide advice as to likely clinical uptake and implementation
- encourage networking and data sharing for success of the project
This group will benefit from the involvement of two patient or family representative members, and a statistician with expertise in prognostic models and/or clinical decision rule building .
The group is expected to meet between one and two times per year, with occasional contact by telephone call or email in between the meetings. Such meetings would be unlikely to be more than half-a-day in length. Pre-meeting and post- discussions with one of the PICNICC team will be offered to patient or family representative members in order to discuss any issues that arise during the meetings. Expenses to cover travel and associated expenses have been awarded. Members of the group will ideally attend the international collaborative group meetings, when this is formed. These may occur only 2-3 times over 5 years.
The structure of this group is:
Prof. Lesley Stewart, CRD, York (Expert in IPD methodologies & reviews)
Dr. Sue Picton, Paediatric Oncology, Leeds (Extensive experience in FNP)
Dr. Julia Chisholm, Paediatric Oncology, London (Chair of CCLG supportive care group)
Dr. Alex Sutton, Medical Statistics, Leicester (Expert in meta-analytic techniques)
Should you wish to find out more to consider becoming involved, please contact Dr Bob Phillips at firstname.lastname@example.org or phone (01904) 321099.