Intensive Care Med (2003) 29:333 DOI 10.1007/s00134-002-1595-9
Nathan Kuppermann Richard Malley W. Charles Huskins
A new prognostic scoring system for meningococcal septic shock in children: comparison with three other scoring systems Received: 12 May 2002 Accepted: 30 October 2002 Published online: 14 December 2002 © Springer-Verlag 2002 Sirs: Outcome prediction models serve several important purposes [1]. These models aid in clinical decision making regarding the use of resources, invasive procedures, and experimental therapies. Prediction models also allow readers of the medical literature to compare, and researchers to stratify, participants in clinical studies according to risk. Ideally, researchers developing prediction rules should aim to develop rules which are accurate yet parsimonious, to allow for easy implementation by the clinician [2]. Over the past four decades many investigators have developed prediction models of outcome of meningococcal disease in children. Meningococcal disease carries a substantial risk of morbidity and mortality and is unpredictable (i.e., disease can range from occult bacteremia to fulminant sepsis). Novel therapies (bactericidal/ permeability-increasing protein, extracorporeal membrane oxygenation, anticoagulants such as activated protein C, and thrombolytic therapy with recombinant tissue plasminogen activator) offer possible options for improving outcomes. Candidates for these new therapeutic modalities need to be selected carefully based on the likelihood of adverse outcomes.
CORRESPONDENCE
Since Stiehm and Damrosch’s [3] seminal article regarding the prediction of adverse outcomes for children with meningococcal disease many authors have used different approaches to study this topic. These models differ from each other in patient selection criteria, clinical settings, prediction variables evaluated and definitions of outcomes. Furthermore, different research design and analytic methods have been employed by different investigators. Therefore comparing the accuracies of these prediction models is difficult. In their study, Castellanos-Ortega et al. [4] retrospectively evaluated the records of children with meningococcal septic shock (a highly select population) from 14 pediatric intensive care units (PICU) in Spain. They evaluated many variables during the first 2 h of PICU stay and recorded the most abnormal values for these variables. They included therapeutic as well as monitoring (e.g., urine output) variables, and their primary outcome was hospital mortality. They attempted to compare the model derived in their study to previously derived models, including a model previously derived and validated by our group [5]. Because the methodology differed substantially, however, comparing their model to ours is unrealistic and misleading. The model developed in our study was based on the data obtained from patients at the time of initial presentation of meningococcal disease, regardless of whether they were in septic shock or admitted to a PICU. We did not evaluate any clinical monitoring or therapeutic variables because our goal was to develop a model in which the prediction of outcome could be made at the time of initial encounter, when decisions regarding transfers to tertiary care centers or use of experimental therapies are frequently made. Additionally, the definition of adverse outcome used in our analysis included both death and loss of limb. The prediction model resulting from our analyses was accurate, but aimed to be more parsimonious than that of Castellanos-Ortega et al. The models being compared, however, were derived using very different methods and are intended for use in different clinical scenarios (evaluation of patients at the time of initial presentation vs. evaluation of patients in the PICU). Therefore comparisons between these two models are misleading and should be avoided.
References 1. Stiell IG, Wells GA (1999) Methodologic standards for the development of clinical decision rules in emergency medicine. Ann Emerg Med 33:437–447 2. Kuppermann N, Willits N (2000) In response to “Statistical models and Occam’s razor.” Acad Emerg Med 7:100–101 3. Stiehm ER, Damrosch DS (1966) Factors in the prognosis of meningococcal infection: review of 63 cases with emphasis on recognition and management of the severely ill patient. J Pediatr 68:457–467 4. Castellanos-Ortega A, DelgadoRodriguez M, Llorca J, et al (2002) A new prognostic scoring system for meningococcal septic shock in children. Comparison with three other scoring systems. Intensive Care Med 2:341–351 5. Malley R, Huskins W, Kuppermann N (1996) Multivariable predictive models for adverse outcome of invasive meningococcal disease in children. J Pediatr 129:702–710 N. Kuppermann (✉) Department of Pediatrics, School of Medicine, University of California at Davis, 2315 Stockton Blvd., Sacramento, CA, 95817 USA e-mail:
[email protected] Tel.: +1-916-7341535 Fax: +1-916-7347950 N. Kuppermann Department of Internal Medicine, Division of Emergency Medicine, School of Medicine, University of California at Davis, 2315 Stockton Blvd., Sacramento, CA, 95817 USA R. Malley Divisions of Infectious Diseases and Emergency Medicine, Department of Medicine, Children’s Hospital, Harvard Medical School, Boston, Mass., USA W. C. Huskins Division of Pediatric Infectious Diseases, Mayo Clinic, Rochester, Minn., USA