I read the article by Loots FJ et al with great interest, and strongly appreciate the importance of the study.1 The authors selected rigors (yes/no) as one of the candidate clinical predictors. The item of rigors was excluded in the simplified model because univariable and multivariable regression analyses showed no significant relationship between rigors and the diagnosis of sepsis.
The authors judged whether a patient developed rigors or not in a dichotomous manner. However, the degree of chills is important in predicting bacteraemia. For example, shaking chills showed a specificity of 90.3% and a positive likelihood ratio (PLR) of 4.65 for bacteraemia, while mild chills showed a specificity of 51.6% and PLR of 1.81 (see Tokuda et al).2 Considering a higher specificity, shaking chills may be a more desirable candidate than rigors only.
- © British Journal of General Practice 2022