The charts of all hospitalized patients with NF diagnosis (screening for ICD-10 NF related codes including m72.6) [20] between 2002 to 2019 at the Rabin Medical Center were reviewed.
The diagnosis of NF was confirmed for all patients with intraoperative clinical findings such as presence of necrotic fascia, purulent discharge and/or histopathologic verification supporting the diagnosis. Aerobic and anaerobic cultures were obtained for all patients from both the blood and the soft infected tissues prior to the antibiotic treatment.
All blood and tissue cultures were processed at the microbiology laboratory at Rabin Medical Center. Bacterial identification and susceptibility testing were done by routine methods. Until 2012 all isolations were performed by manual biochemical testing, followed by VITEK-II in inconclusive cases. As of 2012 all isolations were verified by either MALDI-TOF or VITEK-II. Susceptibility tests were performed using disk diffusion and E-test methods and interpreted according to CLSI criteria.
Positive culture was defined as the isolation of a bacterial pathogen from at least one blood culture and/or infected soft tissue (obtained intraoperative or by skin biopsy) and evidence of necrotic fascia during an operation or pathology features consistent with the diagnosis (extensive tissue destruction, thrombosis of blood vessels, abundant bacteria spreading along fascial planes, and infiltration of acute inflammatory cells) [21].
For patients who have not undergone surgery, a culture was diagnosed and considered positive with the isolation of a bacterial pathogen and high clinical suspicion based on systemic clinical signs, soft tissue involvement and clinical manifestation.
The patients were assigned into two groups according to the NF classification, Group I was consistent with Type 1 NF and included patients with Polymicrobial infection (Pm group) and group 2 included type II NF based on the mono-microbial (Mm group) growth. Although type 3 NF is considered to be caused by mono-microbial organisms, our study found that all the causative mono-microbial organisms to belonged to type II NF, hence group 2 is equal in our study to Type II NF.
Clinical and laboratory data collected for all patients at presentation included fever, blood pressure, heart rate, white blood cells count, platelet count, C-reactive protein, creatinine, serum glucose levels, lactate, creatinine kinase (CK), albumin, and sodium.
Imaging, clinical examination, number of surgeries, duration of hospitalization, intensive care unit (ICU), amputation, use of vasopressors, death date data and Charlson comorbidity index were also collected [22].
The modified Chralson comorbidity index (CCI) predicts 10 years survival in patients. This index is in practice to assess patient comorbidity risk and consists of 9 different comorbidities.
All patients had their medical record checked for surgeries, related anatomically to the infection site, which were performed in the last 30 days
In order to assess sepsis severity, quick SOFA (qSOFA) score was calculated upon admission to the hospital. qSOFA score was first introduced in 2016, and was validated internally and externally for predicting mortality in patients diagnosed with sepsis. qSOFA Score ≥ 2 suggests high risk of poor outcome (in a scale of 0–3) [23, 24].
In addition, immunosuppression status was collected for all patients. Positive immunosuppression status was determined when patients were receiving more the 20mg prednisone and or cytotoxic drug/chemotherapy.
LRINEC score is in use to distinguish patients with severe cellulitis or abscess vs necrotizing fasciitis. LRINEC score includes the following laboratory data: C-reactive protein, WBC, Hb, Na, creatinine and glucose, a score equal or above is the agreed cut off to rule-in NF. We used the LRINEC score, for laboratory risk assessment [25]. In order to assess the severity of the NF, we calculated in our study the LRINEC score at admission to the Emergency Department (ED) for all patients.
The study protocol was approved by the local ethics committee.
The primary outcome was all-cause mortality at 90 days. Secondary outcomes included length of hospitalization, intensive care unit (ICU) admission, LRINEC score and the need for vasopressor use, and amputation by 90 days. We compared for these outcomes the poly-microbial group with the mono-microbial one.
Statistics
Data are expressed as means ± standard deviations (SD), median ± interquartile range (IQR), or number and percentage. We compared patient characteristics between poly vs. mono bacterial groups using t-test, chi-square and non-parametric tests. To estimate the association of the study outcomes with type of bacterial growth, we conducted a forward stepwise logistic regression of the dependent variable. Each set of covariates (demographic, medical history, laboratory, etc.) was entered as separate block to the model. The final model was selected based on model goodness of fit using the c-statistic and plausible clinical explanation. For the primary outcome of 90 days mortality, we used Cox regression, using the same stepwise method described above. Data analysis was performed using SPSS version 25.0.