Purpose: Validate a screening tool for the early identification of sepsis in a general surgical ward.

Background: Sepsis is the leading cause of mortality in non-coronary intensive care units (ICU). Recent studies have shown that early implementation of evidence based guidelines improves survival. We have developed a comprehensive logic-based sepsis protocol for our surgical ICU (SICU); however, we found that early recognition of sepsis was a major obstacle to protocol implementation. To improve this, we developed a three step sepsis screening tool with escalating levels of decision making. We previously validated that this tool was effective in identifying sepsis in the SICU; however, >50% of the septic patients (pts) were transferred from the surgical ward. Therefore, in order for our sepsis screening tool to be maximally effective it needed to be implemented & validated in the non-SICU setting. We hypothesized that our sepsis screening tool would identify sepsis on the surgical ward.

Methods: All pts admitted to the general surgical ward were screened twice daily by nursing staff. The initial screen assesses the systemic inflammatory response syndrome (SIRS) parameters (heart rate, temperature, white blood cell count, & respiratory rate) & assigns a numeric score (0 to 4) for each based on severity of derangement. Pts with summed SIRS scores ≥ 4 screened positive & proceeded to the 2nd step of the tool in which a mid level provider attempts to identify an infectious source. If the patient screens positive for both SIRS & an infection, the attending surgeon was notified & sepsis specific interventions were implemented. Pts demographic & outcome data including source of sepsis, SICU transfer status, & mortality were prospectively collected. After verifying satisfaction of logistic regression analysis assumptions & testing for interactions, logistic regression analyses quantified the predictive properties of the overall SIRS score, & hierarchically assessed predictive properties of each of its 4 subcomponent scores.

Results: Over 6 months ending 10/06/09, 9,332 screens were completed on 959 pts. The average age of the screened population was 56.8 ± 17.6 years & 61.5% were female. The prevalence of sepsis was 1.7%. The screening tool yielded a sensitivity of 99.9%, specificity of 91.3%, a positive predictive value of 16.3%, & a negative predictive value of 99.9%. The sepsis related mortality in those pts that screened positive for sepsis was 6.3%. Of the 16 pts that developed sepsis, 4(25%) required transfer to the SICU. Of the 16 true positive screens, 14 (87.5%) had sepsis, & 2 (12.5%) had severe sepsis at the time of the screen. The sources of infection were intrabdominal (62.5%), bloodstream (6.25%), urinary system (6.25%), skin/soft-tissue (6.25%) & 3 cases did not have a definitive source identified.

Conclusion: The three step sepsis screening tool is a valid method for the early identification of sepsis on a surgical ward. The vast majority of pts detected by the screening tool were in sepsis & had not progressed to severe sepsis/septic shock. The high sensitivity & negative predictive value of the screening tool make it a useful & safe screening tool for sepsis in the surgical population. Future studies evaluating the use of this screening tool in conjunction with early goal directed therapy for sepsis & the impact of this on SICU transfer rates & patient outcomes are planned.