In this secondary analysis of the HYPRESS and SISPCT trials conducted in German ICUs, we observed a significantly lower SIC prevalence (16.9% at sepsis onset and 22.1% during the observation period) in patients with sepsis but not septic shock (at the time of inclusion) and 24.2% in patients with septic shock than had been reported in previous studies from Asia (40–60%) [4, 10, 14, 21]. Furthermore, SIC was either already present at sepsis onset or developed within the first 4 days. This observation is in line with the pathophysiology of SIC being most likely triggered by a dysregulated interaction between the innate immune response and the coagulation system [22]. To address these special features of SIC and to differentiate SIC from other DIC subtypes, Iba and colleagues introduced the SIC score [4, 21]. By excluding the parameters D-dimer and fibrinogen, the SIC score is more likely to detect sepsis-induced coagulopathy at an earlier stage than previous screening tools [23]. However, this also means that not all SIC positive patients suffer from overt DIC. The SIC score was developed “based on the results of logistic regression analyses” using data from 1,498 patients [4]. However, the included patients were highly preselected, since all patients suffered from “DIC according to the Japanese Ministry Health and Labor Welfare Diagnostic Criteria for DIC” and were treated with thrombomodulin alpha [4, 11]. In consequence, SIC prevalence was > 60%. [4]. Therefore, it is not surprising that the observed prevalences in our study are lower although the median SIC-adapted SOFA scores (at the time of inclusion) of these patients were comparable to those of the patients analyzed from the HYPRESS trial (5 [IQR 3–7] vs. 5 [IQR 4–7]) [4]. However, there are several more studies reporting high SIC prevalences. Another study from Japan reported a SIC prevalence of 61.4% [5], and Ding and colleagues [5, 24] even found a SIC prevalence of 67.9% in a Chinese cohort of SEPSIS-3 patients. The most recent validation of the SIC sore is from 2021 [10]. In their secondary analysis, Tanaka and colleagues included only patients fulfilling sepsis criteria according to SEPSIS-3 [10]. The reported SIC incidences were 42.2% in patients not needing vasopressor therapy during their ICU stay and 66.4% in patients requiring vasopressors [10]. In the only European study that calculated SIC prevalence in patients with sepsis and vasopressor requirements, Julie Helms and colleagues reported a positive SIC Score in even 84.2% of the cases [25]. However, the SIC prevalence could have been overestimated in the latter study, since the additional condition introduced by Iba and colleagues (PSSC + ISSC ≥ 3) was not taken into account by Helms et al. when calculating the SIC score [4, 25]. In summary, the prevalence of SIC seems to vary significantly not only depending on the sepsis definition used, but also depending on the composition of the underlying cohort. As the HYPRESS trial included patients with severe sepsis but not shock with only about one fifth of the patients developing septic shock during the observation period (median SOFA Score = 5) and the SISPCT trial included patients with sepsis and septic shock with a median SOFA score of 10 (including 86.7% of patients with shock [15]), our secondary analysis includes two cohorts with different disease severities. However, in both trials SIC prevalence is considerably lower than previously reported. This is surprising as most of the prevalence studies mentioned above referred to patients with severe sepsis and septic shock according to the SEPSIS-2 definition and the patient groups covered by the definitions of “severe sepsis” (according to SEPSIS-2) and “sepsis” (according to SEPSIS-3) overlap widely [26]. Against this background, it is remarkable that the prevalence observed in both of our cohorts are lower by a factor of 2–3.
In this context, it is striking that almost all of the studies on the SIC score mentioned (even the one published in 2021) included patients that were treated for sepsis between 2011 and 2014 [4, 5, 10, 24]. At the same time, it is important to know that the recommendation for routine VTE prophylaxis was not included in the Japanese Sepsis Guideline before 2016 [13, 27]. The chapter did not exist in the 2014 version [27]. It was not until 2012 that the SCC recommended routine VTE prophylaxis for patients with sepsis for the first time [28]. There is no published data on how and whether pharmaceutical VTE prophylaxis was administered during this period. It can be assumed that it was heterogeneous and this fact could have had an impact on the SIC prevalence.
At the same time, mortality of SIC positive patients in our study was comparable to previous reports. SIC was associated with a 90-day mortality between 26.8% (HYPRESS) and 53.3% (SISPCT). This is comparable to the 28-day-mortality in SIC positive patients reported by Iba and collegues ranging between 30% in patients with a SIC score of 4–45% in patients with a SIC score of 6 [4]. Tanaka and colleagues also observed increased in-hospital-mortalities in SIC positive patients (requiring vasopressors: 35.8% vs. 27.9%; not requiring vasopressors: 15.6% vs. 12.2%) [10].
The fact that mortality of SIC positive patients in our observation was comparable to those of SIC positive patients in previous reports is remarkable knowing that the SSC guidelines and the Japanese Clinical Practice Guidelines for Management of Sepsis and Septic Shock differ substantially with regard to their recommendations for anticoagulation [13, 29]. While both the German guidelines and the SSC guidelines recommend pharmaceutical prophylaxis of venous thromboembolism (VTE) with unfractionated heparin or low molecular weight heparins, they do not provide any recommendation regarding the treatment of SIC or overt DIC [29, 30]. In contrast, the Japanese guidelines recommend screening for DIC and the replacement of antithrombin as well as the administration of thrombomodulin alpha in case of SIC [12]. Moreover, older versions of the Japanese guidelines for the management of sepsis suggested the use of protease inhibitors, or heparinoids at doses exceeding VTE prophylaxis [27]. Only since 2016 there has been a recommendation against the use of heparin as standard treatment for patients with sepsis and DIC and against the use of protease inhibitors [13]. As a result, about 50% of patients included in the major Japanese studies received at least one of the following medications: antithrombin, thrombomodulin alpha, protease inhibitors, or heparinoids at doses exceeding VTE prophylaxis [10]. In contrast, most patients in the HYPRESS and SISPCT trials received only pharmacological VTE prophylaxis, most likely because the German sepsis guidelines recommend against the use of antithrombin due to low evidence and augmented risk for severe bleeding events [30].
Our work has several strengths as well as limitations. Using two well-characterized cohorts of patients, which were included in two German multicenter RCTs within the SepNet Critical Care Trials network, strengthens the internal validity of our study. In both trials, patients were treated at more than 30 study sites, supporting the generalizability of our findings to other health care settings respecting the SSC guidelines. This is the first work to evaluate the performance of the SIC score in a population of sepsis patients receiving anticoagulation and prophylaxis of venous thromboembolism (VTE) according to the International Guidelines for Management of Sepsis and Septic Shock by the SSC [13, 16, 17]. Moreover, this is the first work that has been able to discuss the onset of SIC, because detailed daily data during the 14-day observation period were available.
One limitation is that cases of “late-onset” SIC with an onset after 14 days were not captured. However, our data highlight SIC rather as a complication of early sepsis, making the first 4 days after sepsis diagnosis the most important. Considering the described crosstalk between the innate immune response and the coagulation system, it seems reasonable that patients becoming SIC positive after 14 days might have had either a second infection or another medical condition (e.g. severe bleeding complication), which are accompanied by a drop in platelet count or a rise in INR. Moreover, such a hypothetical “late-onset” case of SIC after day 14 would have been without a major clinical impact as the median ICU-LOS in SIC negative patients was 8 days with a range of [5–15]. Another limitation is that we were unable to calculate the SIC prevalence in the SISPCT trial, because INR data had not been collected. The SIC score requires two conditions to be met to be considered positive: First, the total SIC score has to be ≥ 4 points, second, as an additional condition, the sum of PSSC and ISSC has to be ≥ 3 points. To fulfill the second condition, SIC positive patients must have a PSSC > 0. By counting only patients who had a PSSC of 2 at onset (while having a SOFA score ≥ 2), we only counted patients with a high probability of having or developing SIC. On the one hand, we might have underestimated the SIC prevalence by missing some patients with a PSSC of 1 and an ISSC of 2 during the course of the disease. On the other hand, SIC prevalence was likely overestimated because not all patients with a PSSC of 2 necessarily have an ISSC > 0. However, as both effects balance each other to a certain extent, our estimation seems to be quite exact.