You are not currently logged in.

Access your personal account or get JSTOR access through your library or other institution:


Log in to your personal account or through your institution.

Predicting Bacteremia among Patients Hospitalized for Skin and Skin‐Structure Infections: Derivation and Validation of a Risk Score

Benjamin A. Lipsky , MD, Marin H. Kollef , MD, Loren G. Miller , MD, MPH, Xiaowu Sun , PhD, Richard S. Johannes , MD, MS and Ying P. Tabak , PhD
Infection Control and Hospital Epidemiology
Vol. 31, No. 8 (August 2010), pp. 828-837
DOI: 10.1086/654007
Stable URL:
Page Count: 10
  • Subscribe ($19.50)
  • Cite this Item
Item Type
Predicting Bacteremia among Patients Hospitalized for Skin and Skin‐Structure Infections: Derivation and Validation of a Risk Score
Preview not available


Objective.  Bacteremia is relatively common in patients with skin and skin‐structure infection (SSSI) severe enough to require hospitalization. We used selected demographic and clinical characteristics easily assessable at initial evaluation to develop a model for the early identification of patients with SSSI who are at higher risk for bacteremia. Participants.  A large database of adults hospitalized with SSSI at 97 hospitals in the United States during the period from 2003 through 2007 and from whom blood samples were obtained for culture at admission. Methods.  We compared selected candidate predictor variables for patients shown to have bacteremia and patients with no demonstrated bacteremia. Using stepwise logistic regression to identify independent risk factors for bacteremia, we derived a model by using 75% of a randomly split cohort, converted the model coefficients into a risk score system, and then we validated it by using the remaining 25% of the cohort. Results.  Bacteremia was documented in 1,021 (11.7%) of the 8,747 eligible patients. Independent predictors of bacteremia ( \documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland,xspace} \usepackage{amsmath,amsxtra} \usepackage[OT2,OT1]{fontenc} \newcommand\cyr{ \renewcommand\rmdefault{wncyr} \renewcommand\sfdefault{wncyss} \renewcommand\encodingdefault{OT2} \normalfont \selectfont} \DeclareTextFontCommand{\textcyr}{\cyr} \pagestyle{empty} \DeclareMathSizes{10}{9}{7}{6} \begin{document} \landscape $P< .001$ \end{document} ) were infected device or prosthesis, respiratory rate less than 10 or more than 29 breaths per minute, pulse rate less than 49 or more than 125 beats per minute, temperature less than 35.6°C or at least 38.0°C, white blood cell band percentage of 7% or more, white blood cell count greater than \documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland,xspace} \usepackage{amsmath,amsxtra} \usepackage[OT2,OT1]{fontenc} \newcommand\cyr{ \renewcommand\rmdefault{wncyr} \renewcommand\sfdefault{wncyss} \renewcommand\encodingdefault{OT2} \normalfont \selectfont} \DeclareTextFontCommand{\textcyr}{\cyr} \pagestyle{empty} \DeclareMathSizes{10}{9}{7}{6} \begin{document} \landscape $11\times 10^{9}$ \end{document} /L, healthcare‐associated infection, male sex, and older age. The bacteremia rates ranged from 3.7% (lowest decile) to 30.6% (highest decile) ( \documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland,xspace} \usepackage{amsmath,amsxtra} \usepackage[OT2,OT1]{fontenc} \newcommand\cyr{ \renewcommand\rmdefault{wncyr} \renewcommand\sfdefault{wncyss} \renewcommand\encodingdefault{OT2} \normalfont \selectfont} \DeclareTextFontCommand{\textcyr}{\cyr} \pagestyle{empty} \DeclareMathSizes{10}{9}{7}{6} \begin{document} \landscape $P< .001$ \end{document} ). The model C statistic was 0.71; the Hosmer‐Lemeshow test P value was .36, indicating excellent model calibration. Conclusions.  Using data available at hospital admission, we developed a risk score that differentiated SSSI patients at low risk for bacteremia from patients at high risk. This score may help clinicians identify patients who require more intensive monitoring or antimicrobial regimens appropriate for treating bacteremia.

Page Thumbnails