Sequential Organ Failure Assessment score in predicting the prognosis of necrotising fasciitis: a prospective study

INTRODUCTION Necrotising soft tissue infections represent a spectrum of severe skin and soft tissue infections that are characterised by extensive rapidly progressing soft tissue necrosis and systemic signs of sepsis. The estimated incidence of the disease is 0.3–15 cases per 100,000 population.[1,2] Necrotising fasciitis (NF) is a surgical challenge because of the difficulty in establishing a diagnosis and managing the associated morbidity and mortality. While the local infectious component requires incision, drainage and debridement of the tissue with resulting debilitation and tissue defects, the systemic component is more directly responsible for organ dysfunction and death. Despite advances in surgical and medical care, mortality still remains high at about 20%–30%.[3,4,5,6] To prognosticate patients of NF, numerous scoring systems have been studied, including Laboratory Risk Indicator for NF (LRINEC),[7] Acute Physiology and Chronic Health Evaluation II (APACHE II),[8] Sequential Organ Failure Assessment (SOFA), Simplified Acute Physiology Score II (SAPS II)[9] and Mortality in Emergency Department Sepsis (MEDS),[10] to name a few. The SOFA[11] score was initially designed to sequentially assess the severity of organ dysfunction in patients who were critically ill from sepsis, as multiple organ dysfunction is common in critically ill patients. This scoring system differs from other scoring systems in that it was designed to be assessed sequentially over the course of a patient's stay in the hospital. There is always a need to find a simple, validated and rapid bedside tool for early stratification of patients with this potentially life-threatening illness. The present study aimed to evaluate the SOFA score as a prognostic tool for predicting mortality in patients with NF. Although the SOFA score is designed for use in the intensive care unit (ICU) setting, our goal was to extrapolate its role in emergency and general wards. METHODS This was a prospective observational study conducted in the Department of General Surgery, Maulana Azad Medical College, New Delhi, India, over a period of 18 months from October 2017 to April 2019. A total of 50 adult patients who presented with a clinical diagnosis of NF confirmed by operative findings were included in the study. Patients who underwent any type of surgical intervention for the same condition before admission were excluded. Written informed consent was taken from all the included patients, and approval for the study was given by the institutional ethical committee. Data collected included demographics, comorbidities, precipitating and risk factors, clinical and laboratory parameters, treatment modalities and complications (including death). A clinical diagnosis of NF was made according to the guidelines of the Center for Disease Control and Prevention and the National NF Foundation, and this was further confirmed by operative findings.[12,13] Wound was debrided serially, and aseptic dressing was done. Empirically broad-spectrum intravenous antibiotics were started, and they were changed according to the sensitivity pattern or based on the merit of each case. When the need arose, inotropes (noradrenaline, dopamine), transfusion of blood and blood products and mechanical ventilation were provided. Any other comorbidities were managed accordingly. The decision for limb amputation was made by the treating surgeon. Clinical and laboratory data that were required for the calculation of SOFA score were collected. The SOFA score was calculated at admission and repeated on days 2, 4, 7 and 9. Individual day and mean SOFA scores were calculated and correlated with mortality and morbidity. Patients were discharged once the wound started granulating. Data was analysed using IBM Statistical Package for the Social Sciences (SPSS) version 21.0 (IBM Corp, Armonk, NY, USA). Categorical variables were presented as number and percentage, while continuous variables were presented as mean ± standard deviation and median. Normality of data was tested by the Kolmogorov–Smirnov test. If the normality was rejected, then a non-parametric test was used. Statistical tests were applied as follows: quantitative variables between groups were compared using the Mann–Whitney U test (as the data sets were not normally distributed); qualitative variables were correlated using the Chi-square test; Spearman's rank correlation coefficient was used to assess the correlation between quantitative parameters; and the receiver operating characteristic (ROC) curve was used to find out the cut-off point of parameters for predicting outcome. A P value of <0.05 was considered statistically significant. In patients with NF, the SOFA score was used to predict the primary outcome of mortality and the secondary outcomes of limb amputation, requirement of mechanical ventilation, requirement of inotropes and duration of hospital stay. RESULTS There were 50 patients (mean age 53 years) with NF, of whom 37 (74%) were male [Supplementary Table 1, Appendix]. Of these 50 patients, 17 (34%) died, three (6%) required amputation of the involved limb and eight (16%) required mechanical ventilation. Diabetes mellitus (26%) and obesity (24%) were the most common associated comorbidities. There were higher mortality rates among patients with NF and associated chronic obstructive pulmonary disease (five out of eight, 62%) or liver disease (three out of four, 75%), although this finding was not statistically significant. Extremities were the most common site of NF, accounting for 74% of cases. NF involving the trunk and lower limb resulted in higher mortality – three out of six patients (50%) with trunk NF and 11 out of 28 patients (39%) with lower limb NF died. Of the 50 patients with NF, 26 (52%) had tissue or pus culture that yielded polymicrobial growth predominated by Gram-negative bacilli. Ten (20%) patients had monomicrobial culture, the majority of which yielded growth of Streptococcus pyogenes. Blood culture data was not available for all the patients. DISCUSSION NF is a life-threatening condition that requires early diagnosis with aggressive resuscitation, prompt and repeated surgical debridement,[14] physiological support, broad-spectrum antimicrobial therapy[15] and nutritional support. The sepsis component of the disease causes metabolic, haemodynamic, renal, coagulative and cardiorespiratory compromise. Therefore, a goal-directed treatment plan to address these issues is key to preventing mortality. However, despite the above treatment modalities, the prognosis for NF remains poor. Determining the prognosis of this detrimental disease is crucial in management. A number of scoring systems have been designed to prognosticate patients of NF, but none of these scores are widely accepted. Anaya et al.[16] retrospectively collected the data of 350 patients and created a simple bedside scoring system by using six criteria at admission: age, heart rate, temperature, white blood cell count, serum creatinine concentration and haematocrit. The accuracy of this model was reported to be 86.8%. In a retrospective study analysing the LRINEC score of 294 patients, El-Menyar et al.[17] concluded that the LRINEC score, in addition to its diagnostic role, could identify high-risk NF patients who are likely to have a worse outcome. Using the ROC curve, the cut-off LRINEC score for mortality in this study was 8.5 with area under the ROC curve (AUC) of 0.64. However, the authors concluded that the low specificity in predicting septic shock and mortality requires further prospective evaluation.[17] A retrospective study by Yaşar et al.[18] compared the predictive accuracy of four scoring systems, namely APACHE II, SOFA, SAPS II and MEDS, in estimating the prognosis of patients with NF and concluded that the estimated mortality rates of APACHE II and SAPS II were much closer to the actual mortality rates than the other two scoring systems. The SOFA score was developed by the European Society of Intensive Care Medicine in 1996 and has been validated as a useful scoring system to predict morbidity and mortality rates for a variety of critically ill patients. SOFA differs from other scoring systems in that it was designed to be assessed sequentially over the course of a patient's ICU stay, and changes in the SOFA score over time are able to better predict clinical outcome. In the present study, admission SOFA score with a cut-off of 4 predicted mortality with 100% specificity and 82.35% sensitivity and had a positive predictive value (PPV) of 100% and a negative predictive value (NPV) of 91.7% (95% confidence interval [CI] 0.825–0.985). Mean SOFA score ≥2 predicted mortality with 100% specificity and 94.12% specificity and had a PPV and NPV of 100% and 97.1%, respectively (AUROC 0.974, 95% CI 0.884–0.997) [Table 1]. Furthermore, the mortality rate increased over an increased range of admission and mean SOFA scores [Table 2; Supplementary Table 2, Appendix]. An admission SOFA score of 0–2 predicts a mortality of 4.2%, 3–4 a mortality of 33.3%, 5–6 a mortality of 66.7% and >6 predicts a mortality of 100% [Table 2; Supplementary Figure 2, Appendix].Table 1: Multivariate analysis of SOFA scores for mortality, and mechanical ventilation, inotrope and amputation requirement.Table 2: Incidence of death, mechanical ventilation, inotrope requirement and amputation over a range of admission SOFA scores.Bulger et al.,[19] in a retrospective study of 198 patients with NF, predicted mortality using the modified SOFA (mSOFA) score, which excluded bilirubin in the calculation of the score. In this study, mortality was 2% among patients who had an admission mSOFA score <2 versus 13.3% in patients who had an mSOFA score >2 (significant at P = 0.006). Similarly, there was a significant difference in mortality between patients with mSOFA score <3 and those with mSOFA >3 (3.1% vs. 15.9%, P = 0.003). The SOFA score at admission has significant clinical importance, as this baseline score can predict mortality very early in the course and identify high-risk patients, who can then be referred to better-equipped centres for management. Similar to SOFA scores at admission, serial SOFA scores are a good modality to predict the need for mechanical ventilation and mortality. In our study, the serial SOFA score graph showed a defined trend – patients with high SOFA scores remained high on the curve, while those with low scores remained low on the curve or close to the baseline [Figure 1]. The individual day and mean SOFA scores were higher among non-survivors than survivors, and this was statistically significant (P < 0.05) [Supplementary Table 3, Appendix]. The individual laboratory or clinical parameters representing an organ function included in the SOFA score also showed statistically significant difference between survivors and non-survivors. These findings correlated with Bulger et al.'s[19] retrospective study, wherein Day 14 mSOFA score >1 was defined as persistent organ failure and patients with high initial SOFA scores continued to have high scores on days 1, 2, 4, 7 and 14. Bulger et al.'s[19] study also showed statistically significant difference between patients with admission SOFA scores >2 and >3 and those with scores <2 and <3, in that those who had higher values remained in the higher end of the curve subsequently.Figure 1: Graph shows the trend of SOFA scores between survivors and non-survivors. SOFA: Sequential Organ Failure Assessment.In the present study, a mean SOFA score with a cut-off of 6 and an admission SOFA score with a cut-off of 4 predicted the need for mechanical ventilation and inotrope and, hence, the need for ICU admission with reliable accuracy. A cut-off mean SOFA score of 1.4 predicted the need for amputation with less accuracy (AUC 0.67, 95% CI 0.523–0.797) [Table 1]. Amputation of limb was undertaken when radical surgical debridement failed to control the infection. Hypothetically, locally aggressive infection causes more systemic sepsis, leading to a higher SOFA score. Consequently, higher SOFA scores predicting high chances of amputation are to be expected; however, our study could not establish this. Similarly, the relationships between the mean SOFA scores and duration of hospital stay among survivors (correlation coefficient 0.041, P = 0.822) [Supplementary Figure 2, Appendix] and between the SOFA scores and the number of debridements (correlation coefficient = 0.0573, P = 0.751) were not strong and, therefore, no definitive conclusions could be drawn. This is a prospective study with clearly defined diagnostic criteria for NF, unlike previous studies where the retrospective collection of data resulted in missing data and the inability to definitively differentiate between NF and other complicated wound infections. In this study, we observed that mortality and morbidity were far worse for any given range of SOFA scores compared to any other illnesses that cause organ dysfunction, highlighting the gravity of this unforgiving infection. This study is limited by its relatively small sample size. Some important factors predicting mortality in NF, such as time of surgical intervention, age, comorbidities and virulence of the causing organism, were not considered in the SOFA score.[4,5,16] Long-term follow-up of patients was not conducted, and hence, long-term complications and their relation to the SOFA scores could not be studied. In conclusion, the SOFA score accurately predicts mortality and prognoses in patients with NF, in terms of the need for mechanical ventilation and inotrope and, consequently, the need for ICU admission. It is based on simple, readily available parameters that can be reproduced and compared across institutions. The SOFA score at admission can risk stratify patients with NF and can be used as a tool to refer high-risk patients presenting to primary healthcare centres. It can also reliably predict the need for ICU admission and assist the treating physician in taking preventive measures at an early stage. However, the SOFA score is a poor predictor of the need for amputation and duration of hospital stay in patients with NF. Supplementary materialFinancial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.