Annals of Tropical Paediatrics (2001) 21, 135–140 Comparison of two different severity scores (Paediatric Risk of Mortality [PRISM] and the Glasgow Meningococcal Sepsis Prognostic Score [GMSPS]) in meningococcal disease: preliminary analysis P. S. L. SILVA, M. C. M. FONSECA, S. B. O. IGLESIAS, W. B. CARVALHO, R. M. BUSSOLAN & I. W. FREITAS Paediatric Intensive Care Unit, Department of Paediatrics, Sa ˜o Paulo Federal University and Emõ ´lio Ribas Infectious Disease Institute, Sa ˜o Paulo, Brazil (Accepted January 2001) Summary Two different illness severity scores, Pediatric Risk of Mortality (PRISM) and the Glasgow Meningococcal Sepsis Prognostic Score (GMSPS), were evaluated and compared in meningococcal disease in two paediatric intensive care units. Forty-nine children with a median age of 36 months who had meningococcal sepsis con rmed by laboratory data were evaluated. Overall mortality was 18%. The median GMSPS was 3 in survivors and 8 in non-survivors. A GMSPS $ 8 was signi cantly associated with death (p 5 0.0001) with a mortality predictivity and speci city of 70% and 92.5%, respectively. The median PRISM score in survivors was 5.5 and 23 in non-survivors. A PRISM score of $ 11 was signi cantly related to death (p , 0.0001). The Kendal correlation co-ef cient between GMSPS and PRISM showed t 5 0.6859 (p 5 0.0000). It is concluded that GMSPS and PRISM are useful methods for identifying and classifying children into low and high risk categories. GMSPS $ 8 or a PRISM score $ 11 are signi cantly predictive of mortality. Introduction Meningococcal disease follows a fast and ful- minant course and mortality rates vary in dif- ferent series from 18 to 53% 1–7 or even 50 to 60%. 8,9 These rates have not changed signi cantly in recent years. Because of the high mortality rate, rapid diagnosis and im- mediate therapeutic procedures are funda- mental to a better outcome. Many scoring systems have been developed to predict mortality and morbidity in meningococcal disease. Among the scores with de ned parameters are those developed by Stiehm, 4 Niklasson, 5 Leclerc, 10 Garlund, 11 Tesoro, 3 Tu ¨ ysu ¨ z, 12 the MOC score, 13 the Glasgow Meningococcal Sepsis Prognostic Score (GMSPS) 14 and, more recently, the Barquet 15 (for adults) and Kornelisse scores. 16 The GMSPS, developed by Sinclair 17 and validated retrospectively by Thomsom, 14 is easy to use, fast to obtain and re-usable in cases of clinical deterioration. It consists of one laboratory (base excess) and six clinical parameters. However, since Sinclair’s initial report in 1987 which assessed the predictive value of a GMSPS . 8 to be 100%, lower positive predictive values have been reported, Reprint requests to: Dr P. S. L. Silva, Rua das Aroeiras 30, 22 Bairro Jardim, Santo Andre ´, Sa ˜o Paulo, Brazil, CEP 09090-000. Fax: 1 55 11 69142340; e-mail: psls.nat@terra.com.br ISSN 0272-4936 print/ISSN 1465-3281/online/01/020135-06 Ó 2001 The Liverpool School of Tropical Medicine Carfax Publishing DOI: 10.1080/02724930120058197