This study used a prospective surgical database, to investigate the level of systolic blood pressure (SBP) at which the mortality rates begin to increase in septic surgical patients.

All acute, septic general surgical patients older than 15 years of age admitted between January 2012 and January 2015 were included in these analyses.

Of a total of 6,020 adult surgical patients on the database, 3,053 elective patients, 1,664 nonseptic, 52 duplicates, and 11 patients with missing SBP were excluded to leave a cohort of 1,232 acute, septic surgical patients. The median age (intraquartile range [IQR]): 48 (32 to 62) and roughly 50:50 sex ratio (620 female: 609 male). Most of the patients were African: 988 (80.2%) followed by Asians (128 or 10.4%). More than two-thirds (852 or 69.2%) of the patient cohort underwent some form of surgery, and 152 or 12.3% required intensive care unit (ICU) admission. The median length of ICU stay (IQR) was 2 (1 to 4.5) days. The median length of total hospital stay (IQR) was 4 (2 to 9) days. The median SBP (IQR) on admission was 122 (107 to 138). A total of 167 patients died (13.6%). Those that died did have a significantly lower mean SBP compared with the survivors (116 vs 125, P <. 001). Six of 10 patients (60%) with a SBP less than 70 died. The receiver operating characteristic analysis suggests an optimal SBP cut-off of 111 when predicting mortality (area under the receiver operating characteristic curve: .6 [.551, .65]). This cut-off yields a moderate sensitivity (70%), high positive predictive value (90%) but low specificity, and negative predictive value when predicting mortality. Based on this optimal cut-off, 388 or 31.5% of the patients would be classified as shocked. The inflection curve below with fitted nonlinear curve (95% confidence intervals) clearly shows the upward change in observed mortality frequency at lower systolic and base excess (ie base deficit) values. Shocked patients had a significantly higher frequency of mortality (20% vs 11%, P < .001), a significantly higher median lactate (1.9 vs 1.5, P < .001), and mean base deficit (−2.8 vs −1.0, P = .001). No significant difference in mean age, ICU admission, duration of ICU admission, and total length of hospital stay was observed by shocked status.

Our data suggest that patients who die have a significantly lower SBP and clinically significant hypotension in sepsis with regard to increased mortality risk begins at a level of ∼111-mm Hg. This finding needs to be incorporated into bundles of care for surgical sepsis.