Systemic Inflammatory Response Syndrome (SIRS) as Predictor of Severe Sepsis (SS) in Hospitalized Patients (pts) with Hematologic Malignancies.

Pts with hematologic malignancies (HM) are at high risk of developing SS and sepsis-related mortality. The diagnosis of sepsis requires suspicion of infection accompanied by at least two signs of systemic inflammation. The SIRS criteria were devised to improve the early bedside detection of sepsis and provide objective inclusion criteria for clinical research. Cut-point values based on consensus opinion are: HR >90, Temp > 100.4 or <96.8, RR>20, and WBC >12k or <4k. SIRS is defined by the presence of ≥2 abnormalities. Although widely used, SIRS criteria have not been studied in pts with HM because factors such as immunosuppression and other disease characteristics result in clinical and laboratory changes which are thought to potentially invalidate the SIRS criteria. Since HM pts have been excluded from pivotal sepsis trials, clinicians must extrapolate data from other populations to make diagnostic and treatment decisions for HM patients with SS. The purpose of this study was to evaluate the SIRS criteria as predictors of development of SS in pts with HM.

Methods: The association between the SIRS and SS was evaluated in hospitalized adult pts (age 18–83) with HM in a prospective single-center, nested case-control study. The primary outcome was SS, defined as an infection resulting in cardiovascular or respiratory failure. Vital signs were performed every eight hours. Of the 547 pts who consented for the study, 54 developed SS (9.9% CI 7.7–12.7%). Using incidence density sampling, 211 controls were selected by matching on length of stay at time of SS. Evidence-based cut-points were then derived for the individual SIRS variables by examining risk estimates based on variable percentiles.

Results: At hospital admission, HR, RR, BP, WBC, ANC, age, diagnosis, and transplant status were similar between the groups. In univariable analysis (24° prior to SS), significance of the four individual SIRS variables were: HR (p=.001), RR (p<.001), Temp (p<.001) and WBC (p=.218). Sensitivity, specificity and LR(+) for SIRS scores (1–4) are reported in the Table. In multivariable logistic regression, HR (OR 2.0, CI 1.0–4.1 p=.047), RR (OR 8.3, CI 2.8–24.4 p<.001) and Temp (OR 3.8, CI 1.8–8.0 p=.001) remained significant (ROC area=0.75). Age, diagnosis and transplant status did not modify these risks. WBC and hypothermia did not contribute to the model. Empirically derived cut-points for HR, Temp and RR were: HR (≥109, OR 5.5 2.64–11.5 p <.001), Temp (≥100.2°F, OR 2.71, CI 1.2–6.2, P<.001), (RR> 20, OR 8.3, CI 2.8–24.4 p<.001), ROC area= 0.79 (24° prior to SS).

Conclusions: This study is the first to evaluate and define the test characteristics of the SIRS criteria in hospitalized pts with HM. The rationale for excluding HM patients from SS clinical trials based on diagnostic concerns should be revisited when planning future SS studies. Fever, tachycardia and tachypnea seem most predictive in this population. These results suggest that cut-points for significant SIRS variables can be redefined to improve the specificity of the SIRS score. Additional markers for the early detection of SS may further improve upon SIRS in HM pts.