Analysis of Definitions Used In Clinical Research on Invasive Fungal Infections (IFI): Consensus Proposal for New, Standardized Definitions

S. ASCIOGLU¹, B. DE PAUW¹, J.E. BENNETT², J. BILLE¹, F. CROKAERT¹, D.W. DENNING¹, P. DONNELLY¹, J.E. EDWARDS², Z. ERJAVEC¹, D. FIERE¹, O. LORTHOLARY¹, J. MAERTENS¹, J.F.G.M. MEIS¹, T. PATTERSON², J.H. REX², J. RITTER¹, D. SELLESLAG¹, P.M. SHAH¹, D.A. STEVENS², T.J. WALSH²

¹European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Co-operative Group (EORTC/IFICG), Brussels, Belgium and ²National Institute of Allergy and Infectious Diseases/Mycoses Study Group (NIAID/MSG), Bethesda, MD, USA.

ABSTRACT

Background: Clinical research on the epidemiology, diagnosis, and treatment of IFI in immunocompromised patients is accumulating in the medical literature. Unfortunately, differences in definitions, nomenclature, and diagnostic criteria complicate comparison of results and collaboration between centers. We sought to review these differences and to develop standard definitions for IFI.

Methods: A task force was created in 1997 to systematically review the literature, describe problems related to heterogeneity of study populations, and abstract previously used definitions of IFI. These definitions were applied to 367 patients from the EORTC database, and the resulting patient classifications were compared between the studies.

Results: Of 7086 articles on IFI published between 1985 and 1997, 173 were clinical studies in immunocompromised patients. The definitions and nomenclature used in these studies differed widely: considering only the category of proven IFI category for example, there were 76 different combinations of diagnostic criteria. When applied to the 367 test patients, the definitions produced very low levels of agreement (Kappa =0.253 95% CI 0.251-0.255). These results were reviewed by the task force and used as the basis for developing new definitions of IFI. This development process emphasized use of diagnostic criteria with a high degree of intrinsic reliability. Consensus was reached on definitions and classification of IFI via a series of meetings and reviews of draft documents.

Conclusion: Prior studies have used many different definitions for IFI in immunocompromised patients. While no set of definitions will have 100% sensitivity or specificity, standard definitions have been proposed that could promote study-to-study comparison and enhance efforts to pool data.

Background

• Opportunistic invasive fungal infections (IFI) are a major cause of significant morbidity and mortality in immunocompromised patients.

• In the last decade there has been considerable progress in diagnosing and treating IFI but there remains as much uncertainty as controversy in establishing the definite diagnosis.

• This uncertainty results in differences in nomenclature, diagnostic criteria and definitions of IFI.

• These lack of standardized definitions, are also reflected in all kinds of clinical research on IFI including epidemiological, empiric, therapeutic, prophylactic and diagnostic studies which complicates collaboration and comparison of results between centers, groups.

• We sought to review these differences and to develop standard definitions for IFI for use in clinical research.

Methods

In March 1997 EORTC/IFICG initiated the "Consensus on definitions of IFI in cancer patients" project and appointed a task force with 12 members from 8 different countries. Six members of NIAID/MSG joined this task force later and formed the "Consensus Committee".

First a systematic review of literature was done for an explicit identification of problems related to heterogeneity of study populations. We searched for all clinical research articles on IFI published between 1985 and 1997. After reviewing the abstracts of all retrieved papers, 173 were selected, provided that our inclusion criteria were met. These articles were analyzed for their methodology, study populations, terms and definitions used to define IFI. A search for agreement between these studies was done to investigate inter-study reliability by using kappa statistics. Each definition used to define an IFI episode from selected articles was applied to 367 patients from EORTC database and resulting patient classifications were compared.

These results and literature search were reviewed by the consensus committee and formed the basis of a new proposal for defining and classifying IFI. Consensus was reached on via a series of meetings and reviews of draft documents.

Results-I

Sytematic Literature Review

• Of the 7086 articles published between 1985 and 1997, 173 were reports of clinical research (clinical trials, observational and methodological studies) which included immunocompromised adults and investigated non-mucosal deep tissue infections.
  
  
• Nomenclature used in these studies differed widely. Authors used 25 different adjectives just to indicate degrees of certainty of their diagnoses: proven, probable, possible being the most commonly used (Fig 1).
  
  
• Diagnostic criteria and definitions of IFI were also highly variable among investigators. There were 76 different combinations of diagnostic criteria for the proven IFI category, 34 for probable and 13 for possible IFI. Although many different combinations were used, there were particular trends towards some combinations (Fig 2).
  
  
• These definitions were applied to 367 patients on EORTC database and each patient was evaluated as many times as the number of different definitions. This showed that both overall agreement and category specific (proven, probable, possible) agreement were very low among the published studies in literature (Table).

Results-II

Proposed Definitions of IFI

• The final proposal for standardized definitions splits IFI into two broad categories microbiologically, as yeasts (and yeast-like fungi) and moulds.
  
  
• Clinically IFI are classified into 3 categories based on the level of certainty in diagnosis: Proven IFI, probable IFI and possible IFI. This classification depends on 3 key elements, which are host factors, clinical features and mycological results.
  
  
• Since every case of IFI cannot be proved in daily clinical practice, flexible combination of these individual elements namely host, clinical and mycological allows classifying a patient into one of the probable or possible categories.
  
  
• These definitions are restricted to immunocompromised patients with cancer and recipients of hematopoietic stem cell transplants who are suspected of having an invasive fungal infective disease.

Discussion

• The information available in literature has shown that there is considerable inconsistency in defining IFI and therefore in recruiting patients for clinical studies.
  
  
• This inconsistency results in low level of agreement which is a sign of low inter-study reliability among published literature.
• This hinders comparison of studies, pooling of data and collaboration.
  
  
• On the basis of our literature review and joint efforts of the consensus committee, we propose definitions for a new classification based on the level of certainty for diagnosing IFI.
  
  
• This proposal includes both diagnostic criteria for proven IFI and classification criteria for probable and possible diseases that are intended to promote a more uniform description of the patients when various research are reported.