We assessed all adult patients hospitalized at the Aga Khan University Hospital, a 740 bed quaternary care center in Karachi, between October 2019 and December 2019 with the diagnosis of sepsis. The study was approved by the Institutional Review Board and informed consent was waived because of the retrospective nature of the study and that the analysis used anonymous clinical data. Patients were identified using the ICD-9 codes for sepsis (995.91), severe sepsis (995.92) or septic shock (785.52).

Data was collected from medical records of patients using a pre-approved pro forma. The extracted variables included demographics, patient and hospitalization characteristics, laboratory investigations and outcome. The first serum lactate reading during the admission was considered for analysis. The SOFA and qSOFA scores were calculated by our research team using the relevant clinical findings and laboratory readings documented for each patient. Outcome measures included in-hospital mortality and length of stay.

IBM SPSS 22 was used for data analysis. We compared clinical characteristics of the mortality and survival group using chi-square test for categorical variables, and Student t test for continuous variables. P-values ≤ 0.05 were considered significant. Serum lactate, qSOFA and SOFA scores were compared by assessing the area under the receiver operating curve (AUROC).

Three hundred and sixty-six patients were admitted during the study period with the diagnosis of sepsis, severe sepsis and/or septic shock. There were 208 (57%) males and 158 (43%) females; mean age was 59 ± 18 years. Patient demographics and hospitalization characteristics are presented in table 1. Overall mortality in sepsis was 25% and mean length of stay was 7.5 ± 6.2 days (Figure 1).

Figure 1: Outcome of sepsis patients (N = 366)

The AUROC of serum lactate for predicting mortality in these patients was 0.79 (95% CI; 0.72–0.86) with 81% sensitivity and 64% specificity. In comparison, the AUROC of SOFA for predicting mortality was 0.54 (95% CI; 0.47–0.62) with 60% sensitivity and 43% specificity versus the AUROC of qSOFA score which was 0.49 (95% CI; 0.42–0.58) with 54% sensitivity and 48% specificity (Figure 2).

Figure 2: AUROC for serum lactate, qSOFA score and SOFA score (N = 366)

Lactate, SOFA and qSOFA are widely used markers across hospitals for predicting the outcomes of sepsis, a life-threatening inflammatory response for which immediate screening and intervention is needed (2). We looked at the performance of serum lactate level, qSOFA and SOFA as predictors of mortality in septic patients with varying degrees of severity and found that serum lactate levels had the highest sensitivity and specificity in predicting sepsis-related mortality. This is an important finding in a resource constrained setting of LMIC.

Lactate remains the more reliable serum marker for sepsis, as is also shown in our studywith an AUROC of 0.79, comparable to 0.664-0.72; similar observations have been made in other studies (7, 22). Sepsis-3 has also recommended serum lactate level > 2 mmol/L as a major criterion for clinical identification of septic shock (23).

While lactate has been a consistent marker in terms of its sensitivity and specificity, the values of AUROC for SOFA and qSOFA have been in different ranges. SOFA scores are recognized to be a useful predictor of ICU mortality and have been seen to have much higher values of 0.753- 0.829 previously. Between SOFA and qSOFA, SOFA is also known to be more reliable (23-24). While our study reiterates what is already known about lactate levels in sepsis, the AUROC value of 0.54 for SOFA in our study is much lower than what has been seen in other studies (25-27). AUROC value of 0.49 for qSOFA in our patients is contrary to what has been reported previously from other parts of the world. Some studies have reported lower qSOFA AUROC values of 0.607 while others have shown a much higher AUROC (0.66-0.754). This probably suggests the presence of other variables that could be influencing sepsis outcome (23). Despite its variable performance, qSOFA remains a quick and simple method of screening patients with sepsis, especially in the ICU setting (24, 28-29). To increase qSOFA’s overall reliability, it has been suggested to combine it with lactate levels (23-29).

Presence of diabetes mellitus, tachypnea, lower mean arterial pressure and low score on Glasgow Coma Scale showed association with sepsis-related mortality but did not achieve significance on statistical analysis; higher FiO2 was significantly associated with mortality. Other findings in patients who died of severe sepsis included higher total bilirubin levels and lower platelet counts. Length of stay was observed to be longer in patients with septic shock and in those who required ICU admission.

In our sepsis patients, serum lactate levels correlated with mortality; qSOFA and SOFA scores had low sensitivity and specificity as mortality predictors. Further large-scale studies are needed in our population to determine the utility and reliability of these tools as markers of outcome in sepsis.

Source(s) of Support an Funding: None.

Conflict of Interest statement: The authors declare no conflict of interest.