Comparison of QSOFA and SIRS Scores for the Prediction of Adverse Outcomes of Secondary Peritonitis in a Tertiary Teaching Hospital in Uganda: a Prospective Cohort Study

Background: SIRS and qSOFA are two ancillary scoring tools that have been used globally, inside and outside of ICU to predict adverse outcomes of infections such as secondary peritonitis. Mulago hospital uses SIRS outside the ICU to identify patients with secondary peritonitis, who are at risk of adverse outcomes. However it’s associated with delays in decision making given its partial reliance on laboratory parameters. In response to the practical limitations of SIRS, the sepsis-3 task force recommends qSOFA as a better tool, however its performance in patients with secondary peritonitis in comparison to that of SIRS has not been evaluated in Mulago hospital, Uganda. Objective: To compare the performance of qSOFA and SIRS scores in predicting adverse outcomes of secondary peritonitis in Mulago hospital, Uganda. Methods: This was a prospective cohort study of patients with clinically conrmed secondary peritonitis, from March 2018 to January 2019 at the A&E, Mulago hospital. QSOFA and SIRS scores were generated for each of the patient, with a score of ≥ 2 recorded as high risk, while a score of ≤ 2 recorded as low risk for the adverse outcome respectively. After surgery, patients were followed up until discharge or death. In-hospital mortality and prolonged hospital stay were the primary and secondary adverse outcomes, respectively. Sensitivity, specicity, PPV, NPV and accuracy at 95% condence interval were calculated for each of the scores using STATA v.13 Results: A total of 153 patients were enrolled. Of these, 151(M: F, 2.4:1) completed follow up and were analysed, 2 were excluded. Mortality rate was 11.9%. Fourty (26.5%) patients had a prolonged hospital stay. QSOFA predicted in-hospital mortality with AUROC of 0.52 versus 0.62, for SIRS. Similarly, qSOFA predicted prolonged hospital stay with AUROC of 0.54 versus 0.57, for SIRS. Conclusion: SIRS is superior to qSOFA in predicting both mortality and prolonged hospital stay among patients with secondary peritonitis. However, overall, both scores showed a poor discrimination for both adverse outcomes and therefore not ideal tools.


Background
The use of scoring systems for surgical risk assessment in clinical practice has been known since 1941 when the American society of Anesthesiologists (ASA) developed a physical status scoring system for patients undergoing surgery. Since then, various efforts to nd an ideal scoring system that correctly predicts the risk of mortality continues to occupy medical scientists (1).Finding an ideal scoring tool is key in accurately predicting outcomes and selection of treatment options (2).
Despite the global advances in surgical practice and care, intra-abdominal infections arising from secondary peritonitis still remain one of the most signi cant causes of morbidity and mortality world over (3).In the setting of septic shock, mortality of up to 30% has been reported (4).In order to accurately predict these adverse outcomes of secondary peritonitis, a number of risk assessment scoring tools have Page 3/15 been developed and used with various performance levels in different clinical settings. SIRS and qSOFA are two such ancillary scoring tools that have been used extensively inside and outside the Intensive Care Unit (ICU) setting globally.
Mulago hospital uses SIRS outside ICU to assess disease severity and identify patients at risk of sepsis and other adverse outcomes arising from secondary peritonitis. This practice is however not guided by any local empirical evidence. Available evidence from other settings indicates that SIRS is highly sensitive but with a low speci city (5).Furthermore, it is partly based on laboratory parameters which have a long turnaround time and cost implications causing untoward delays in decision making. These drawbacks have increased the need for a more simple, accurate and less expensive clinical tool for bedside use by clinicians at many centers including Mulago hospital.
In response to practical limitations of SIRS, the third international consensus conference of 2016 on sepsis(sepsis-3) introduced and recommended qSOFA as a better tool than SIRS for predicting adverse outcomes among patients with infections outside the ICU(5).QSOFA is a surrogate for SOFA and it assigns one point for each of its parameters which are; systolic blood pressure less than 100 mmHg, respiratory rate greater or equal to 22 breaths per minute and Glasgow coma scale of less than 15. On the other hand, a systolic blood pressure equal to or more than 100 mmHg; a respiratory rate of less than 22 breaths per minute and a Glasgow coma scale of 15 are each scored zero under qSOFA scoring system.
Since its inception, qSOFA has been validated in many centers and found to be a better predictor of adverse outcomes in patients with infections outside the ICU compared to SIRS(6, 7).All these comparison studies however, were conducted in centers with different settings compared to that of Mulago hospital. This study aimed to nd out whether the performance of qSOFA is superior to that of SIRS in predicting adverse outcomes of secondary peritonitis in Mulago hospital, Uganda.

Methods
This was a prospective cohort study conducted between March 2018 and January 2019. The aim of this study was to determine whether qSOFA was superior to SIRS in predicting adverse outcomes of peritonitis among patients at Mulago National Referral Hospital. Mulago National Referral Hospital is a 1500 bed tertiary teaching hospital located in Kampala which is the capital city of Uganda. This study was based in the Accident and Emergency (A&E) department which handles surgical emergencies like secondary peritonitis, among others. Here patients are resuscitated and stabilized before de nitive surgical interventions are undertaken. Emergency surgical operations are carried out in the A & E theatre with an average of two laparotomies due to secondary peritonitis being performed daily. After surgery, the patients are temporarily admitted to the Emergency ward before transfer to their de nitive general surgical wards.

Sampling
We sampled 153 patients who were 13 years and above and were diagnosed and admitted with secondary peritonitis in the A & E department of Mulago Hospital between March 2018 and January 2019. These patients were followed up until discharge from hospital or death. We excluded 2 patients who were lost to follow up from the analysis and therefore analysed data from 151 patients.

Data analysis
Descriptive statistics for the patients were summarized using proportions for categorical variables, whereas continuous and discrete variables were summarized as means (standard deviation), or median (Interquartile Range (IQR)) depending on the distribution of data. Level of signi cance was taken to be 5%.
The proportions of patients with secondary peritonitis were scored as high risk(a score greater or equal to two) or low risk(a score of less than two) using the qSOFA or SIRS scores. Patients falling in either category of high or low risk were recorded in dichotomous 2 × 2 contingency table against presence or absence of an adverse outcome. Adverse outcomes were prolonged hospital stay of more than 7 days, admission to ICU and death. The sensitivity, speci city, positive predictive value, negative predictive value and accuracy and their 95% con dence intervals (CI) were computed. True positives were those who scored high risk and actually got an adverse outcome, false positives were those who did not get an adverse outcome but were scored high risk, false negatives got the adverse outcome but were scored low risk and true negatives were those who were scored low risk and did not get the adverse outcome. Since none of the two scoring tools is the gold standard, the truth against which they were compared was determined clinically.
Accuracy was obtained by constructing the receiver characteristic operating curve. The tool with a wider area under the curve was considered to be superior to the other.
All the above parameters were compared for both qSOFA and SIRS, and the better performing tool of the two reported accordingly.

Results
Patients' demographic characteristics.
There were 106(70.2%) males and 45(29.8%) females, giving a M:F of 2.4:1. Majority of the patients (55%),were between 13 and 29 years, Table 1. Adverse outcomes and distribution of qSOFA and SIRS scores At the end of follow up, 18 patients were dead, giving an overall mortality of 11.9% for secondary peritonitis. Majority of the deaths (66.7%) occurred in the elderly patients (older than 50 years). Forty (26.5%) patients had a prolonged length of stay (Table 2). QSOFA was a more speci c but less sensitive tool, while SIRS was a far more sensitive but less speci c tool for the prediction of both mortality and prolonged hospital stay as a result of having secondary peritonitis outside the ICU setting in Mulago hospital (Table 3). SIRS had a superior predictive value, in comparison to qSOFA, for both mortality and prolonged hospital stay resulting from secondary peritonitis (Table 4).  Overall, SIRS predicted both mortality and prolonged hospital stay, resulting from secondary peritonitis outside the ICU setting in Mulago hospital more accurately than qSOFA.

Discussion
The third international consensus on sepsis 2016, introduced qSOFA as a better tool than SIRS, for identifying patients with infections, at risk of adverse outcomes. We set out to compare the performance of qSOFA and SIRS scores for the prediction of in-hospital mortality as a primary adverse outcome, and prolonged length of hospital stay as a secondary adverse outcome, among patients with secondary peritonitis at Mulago National referral and teaching hospital. Majority of the patients in this study were young. This nding is consistent with what Wabwire and Saidi reported in their study on strati ed evaluation of secondary peritonitis (21). There were more males compared to females in this study. A similar observation has been reported by previous studies (3,8,(18)(19)(20)(21)(22)(23).
The overall mortality rate in this study, albeit comparable to what has been reported by other studies (17,21,23) is still quite high and underscores the need for early identi cation of at risk patients for prompt intervention. The higher mortality rate among elderly patients could be probably because the elderly are likely to have poor physiological reserves and or comorbidities and possibly late diagnosis and also late patient presentation.
In this study, prolonged hospital stay was attributed to the attendant complications of secondary peritonitis, including but not limited to surgical site infections, burst abdomen, and relaparotomy.
In this study, qSOFA was a more speci c, but less sensitive tool, while SIRS was a more sensitive, but less speci c tool, for the prediction of both mortality and prolonged hospital stay. Findings akin to these, have been reported by Singer et al in their sepsis 3 report(5), and subsequent studies by Finkelsztein et al, Freund et al and Churpek et al(6,7,13). SIRS had a superior predictive value for both mortality and prolonged length of hospital stay, compared to qSOFA. This nding contrasts with that from previous studies (7,15,16). SIRS was superior to qSOFA in predicting both mortality and prolonged hospital stay, in this study, consistent with the ndings of Askim et al (15).
The above ndings however, contrast with those reported by Churpek et al, Freund et al and Finkelsztein et al, who found qSOFA to be superior to SIRS (6, 7, 13). This disparity could be because a few notable differences between the aforementioned studies and this study. Churpek and Finkelsztein included patients being transferred from wards to the ICU, since they were assessing prediction of ICU stay by both scores as a secondary outcome (6, 13). Such patients are more likely to have high qSOFA scores since they are critically ill, with multiple organ dysfunction, compared to stable patients in the A&E. None of the patients in this study was transferred to ICU. Freund et al calculated qSOFA scores by collecting its parameters at their worst level during the patients' entire hospital stay, that is, the highest respiratory rate, lowest systolic blood pressure, and lowest Glasgow coma scale. It is not clear from their study though, whether the same was applied to the SIRS scores. This could have biased their results (7). In our study, both qSOFA and SIRS scores were calculated at admission only, during the patients' entire hospital stay, irrespective of whether there was or there was no change in the parameters, from which they are generated.

Limitations of the study
In this study, Clinical and laboratory parameters used to generate both qSOFA and SIRS scores, were collected at admission only, during the patients' entire hospital stay. Cognizant of the fact that, these keep changing from time to time especially when patients deteriorate, could have resulted in many patients not meeting the criteria for both scores and yet developed the adverse outcomes (High false negative rate).
Both qSOFA and SIRS scores do not consider parameters like patients' age, sex, and presence of comorbidities, all of which have the potential to modify the outcomes of interest in this study. However, since baseline risk associated with those parameters was considered during the development of qSOFA score, this may not have affected the results signi cantly.
Conclusions SIRS score is superior to qSOFA score in predicting both in-hospital mortality and prolonged hospital stay of patients with secondary peritonitis in Mulago hospital. Overall however, both qSOFA and SIRS scores showed poor discrimination of both adverse outcomes in this study and therefore not ideal for this purpose. SIRS score is more sensitive but less speci c tool, while qSOFA is a more speci c but less sensitive tool for predicting both in-hospital mortality and prolonged hospital stay of patients with secondary peritonitis in Mulago hospital. and Ethics Committee. Written informed consent/assent was obtained from each of the patients with the support of a health care worker. All data and results generated from this study were kept con dential.

Consent for publication
Not applicable. Consent for publication was covered in the informed consent forms obtained from each of the patients in the study.

Availability of data and material
The datasets used or analyzed during this study are not publicly available because this data was for academic research but are available from the primary author Dr Emmanuel Nkonge on a reasonable request.

Competing interests
The authors declare that they have no competing interests or other interests that might be perceived to in uence the results and/or discussion reported in this paper.

Funding
Funding for this study was obtained from authors' personal resources.
Authors' contributions EK -collected data, analyzed and interpreted it and wrote manuscript OK -supervised data collection, data analysis and interpretation of the results, wrote and edited manuscript WO -supervised study and data collection All authors reviewed the manuscript All authors read and approved the nal manuscript ROC curve for prediction of mortality by SIRS. Figure 3 ROC curve for prediction of prolonged hospital stay by qSOFA. Figure 4 ROC curve for prediction of prolonged hospital stay by SIRS.