This article is only available in the PDF format. Download the PDF to view the article, as well as its associated figures and tables.
.—We thank Dr Jeffs for detecting the typographical errors in the equations for customizing SAPS II and MPM II24. We regret that we did not detect these in the page proofs.Dr Rubenfeld raises a number of concerns regarding the use of severity systems such as MPM II and SAPS II in an ICU environment. All severity systems for ICU patients were developed and validated on databases that included a wide range of patient diagnoses and conditions. However, no statistical model can be guaranteed to perform well for groups of patients not included in the original study population or, for that matter, exclusively for specific subgroups of patients (eg, patients with sepsis or cancer). Good model performance cannot be taken for granted.If ICU care is either much better or much worse than what a statistical model would lead us to expect, this could be due to
Lemeshow S, Klar J, Gall JL, Teres D. Customized Probability Models for Severe Sepsis: Correction and Clarification-Reply. JAMA. 1995;274(11):872-873. doi:10.1001/jama.1995.03530110034026