Septic shock is the most severe form of sepsis and occurs as a result of both community-acquired and health care-associated infections. Pneumonia, followed by intraabdominal and urinary tract infections are the most common causes.1–4

Despite major technological advances, mortality of septic shock remains high, ranging from 20 to 50%.5–7 However, recent reports have shown decreases in mortality, associated to advances in training, better surveillance and monitoring, and prompt initiation of therapy to treat the underlying infection and support failing organs.8–12

Most studies in patients with septic shock have analyzed ICU and/or hospital mortality, but few studies have evaluated the time in which these patients die.13–15 The most recent study to analyze the time of death of patients with septic shock found that 32% of patients with septic shock died within 72h of admission to the ICU, and that these early deaths were mainly attributable to intractable multiple organ failure.13

The main objective of the present study was to estimate the annual incidence rates of mortality within 72h after ICU admission of community-acquired septic shock (CASS) through a secondary analysis of recently published data from a 14-year period.12 Secondary objectives included characterizing these patients and identifying risk factors associated with early mortality (EM) with the aim of improving survival in these patients.

During the study period, 625 patients were admitted to the ICU with CASS. The most common comorbidities were diabetes (29.1%) and cancer (20%), and 24% were immunocompromised. Infection was documented by a positive culture in 92.2% of patients. Blood cultures were positive in 52.5% of the patients. The most common site of infection causing sepsis was the lung (27%). The most frequent organ dysfunction associated with septic shock was acute kidney injury, which occurred in 555 (88.8%) patients, 132 (23.7%) of whom required continuous renal replacement. Thrombocytopenia was present in 390 (62.5%) patients. A total of 410 (65.7%) patients required invasive mechanical ventilation; ARDS was diagnosed in 142 (22.8%) patients.

During the study period, the mean mortality related to septic shock was 25.8% in the ICU and 29.6% in the hospital. The mean EM related to septic shock was 14.4%. The percentages of early and late septic shock related deaths did not vary significantly over the years (Fig. 1). There was a trend toward decreased rates of both early and late mortality over the study period (Fig. 2).

Figure 1.

Distribution of annual early and late deaths during the period of study (p=0.576).

(0.1MB).

Figure 2.

Incidence of annual early and late deaths during the period of study.

(0.1MB).

A total of 161 patients died in the ICU, 90 (55.9%) within the first 72h of ICU admission; all these deaths were attributed to intractable multiple organ failure related to the primary infection. Of the 90 patients who died within the first 72h, 49 (54.4%) died in the first 24h, 21 (23.3%) in the second 24h, and 20 (22.2%) in the third 24h (p<0.001).

Table 1 compares the characteristics of the patients who died in the first 72h versus those of the patients who survived more than 72h. Apache II and SOFA at admission were higher in patients who died in the first 72h. Among comorbidities, only cirrhosis was significantly more frequent in the EM group. ARDS, thrombocytopenia, mechanical ventilation, and bacteremia were also more predominant in EM group.

Table 2 reports the anatomic sites of infection considered the source of sepsis and microbiology findings in the two groups. The sources of infection differed significantly between the groups and the proportion of patients who underwent source control was significantly lower in the EM group (41% vs. 53.2% in the group of patients that survived>72h, p=0.019). Intraabdominal infections, skin and soft-tissue infections, and other sources that did not fit into the main predefined categories (mainly, endocarditis, meningitis, and unknown origin) were more common in the EM group, and the incidence of biliary and genitourinary infections was lower in this group. We observed no significant differences between the two groups in the pathogens responsible for the infections. There was no difference in the rate of appropriate antibiotic treatment between the two groups.

Source-control interventions were significantly less common in the EM group [37/90 (41.1%) patients vs. 285/585 (53.2%) patients who survived≥72h, p=0.019]. This difference was mainly due to a lower incidence of biliary and genitourinary tract infections in the EM group (17.7% vs. 37.4%, p<0.001). In the 254 cases in which the adequacy of source control could be evaluated, adequate source control was in a smaller proportion of patients in the EM group [12/30 (40%) vs. 210/224 (93.7%) in patients who survived≥72h, p<0.001]. Moreover, the proportion of patients who received adequate source control for different source categories was significantly lower in the EM group than in the group of patients who survived≥72h: for intraabdominal infections (38.4% vs. 92.3%; p<0.001), for skin and soft-tissue infections (22.2% vs. 80%; p<0.001), and for genitourinary infections (0% vs. 100%; p<0.001), except in biliary infections where all 5 patients in the EM group received effective source control but died nevertheless. The proportion of patients receiving source control within 12h of hospital admission did not differ between groups.

Tables 3 and 4 compare the characteristics of patients with EM and survivors over 72h depending on whether they need a source control or not. Patients without source control have a significantly higher severity on admission, incidence of ARDS and more comorbidities, mainly immunosuppression, alcoholism and cirrhosis. Abdominal, biliary and skin and soft tissue infections are more frequent in the group of patients where a source control was performed, both in patients with EM and in those who survive more than 72h. The opposite occurs with respiratory infections.

Table 5 reports variables significantly associated with EM mortality in the multivariate analyses. Variables independently associated with worse prognosis in the first 72h of ICU admission adjusted for age, comorbidities, and severity at ICU admission were APACHE II, ARDS, bacteremia, and effective source control was independently associated with better survival.

We repeated a multivariate analysis in the group of patients without SC, and the APACHE II, ARDS and origin of sepsis other than respiratory source were the variables independently associated with an early poor prognosis, while the multivariate analysis in the group of patients susceptible to source control, showed that ARDS, bacteremia, and APACHE II were the variables associated with EM. However, adequate source control was associated with a better survival rate in the first 72h (Table 5).

Despite the trend toward reduced ICU mortality in patients with CASS during the period observed by the present study, half of all the patients who die in the ICU due to septic shock continue to die within the first 72h of admission and this has remained stable over time. Patients with infections of non-respiratory origin which are not susceptible to source control or those where source control is necessary but inadequate, have a higher probability of EM secondary to CASS.

Overall, 14.4% of all patients admitted with CASS died during the first 72h (more than half of these died in the first 24h). Patients who died within the first 72h accounted for 55.9% of all deaths due to CASS in the ICU. The only factor that could be modified to decrease EM was adequate source control in the subgroup of patients with infections amenable to source control.

In the 1990s, a French multicenter study found that 27% of patients with documented severe sepsis or septic shock died within 48h of onset of severe sepsis.14 Ever since, the few studies that have analyzed the time of death of patients with severe sepsis or septic shock have reported lower EM rates similar to the 14.4% found in our study. In 2008, Spanish researchers published the results of a multicenter observational study of cohort including both severe sepsis and septic shock patients in two three-month periods in 2002, reporting that 19.3% had died within 72h after onset of severe sepsis.15 Ten years later, in a new multicenter study to analyze the influence of Surviving Sepsis Campaign, the Spanish group found mortality in the first 48h had decreased from 14.8% to 7%.23 In a French multicenter study in 2004 (EPISEPSIS)24 analyzing the epidemiology and prognosis of severe sepsis, mortality within 72h after diagnosis was 13.3%. In an observational study in 2015 analyzing the timing and causes of death in patients with septic shock, Daviaud et al.13 observed a 72-hour mortality rate of 14.3%, and multiple organ failure accounted for most early deaths. More recently, another French study analyzing the dose of vasopressors and the risk of early mortality in 370 patients with septic shock reported 13% mortality within 72h following vasopressor initiation.25

The present study analyzed early mortality over a 14-year period in which the early identification and treatment of patients with septic shock have improved.18,26 Despite these improvements, the proportion of deaths directly attributable to septic shock and progressive multiorgan failure occurring within 72h of ICU admission remained stable over time. Moreover, 54.4% of early deaths occurred during the first 24h, suggesting possible delays in identifying patients with septic shock and admitting them to the ICU, fulminant sepsis or inability to perform proper source control which renders therapeutic efforts ineffective. Unfortunately, we have no data about the duration of symptoms before patients reach the hospital.

Although the percentage of patients who die early in our study is similar to other published studies, the distribution of patients dying between early and late is different. Indeed, in our study, 90 of the 166 patients who died in the ICU (55.9%) died early. In contrast, in the Daviaud study13 that analyzes the time that patients die in septic shock, they find only a 32% of early deaths. In other older studies, death within 48–72h accounted for 27%17 and 40%15 of ICU deaths secondary to septic shock. These differences could be partially explained by differences in the study populations. Considering the difficulty of applying the definitions of Sepsis-3 in a retrospective study and the possible biases in the analysis that this entails, the present study only included patients with community-acquired infections if they fulfilled the criteria outlined in the Sepsis-3 definitions.16 Other studies14,15 have included patients with both community-acquired and healthcare-associated infections as well as some patients who had severe sepsis without septic shock and used other criteria for a septic shock diagnosis. Although our study and Daviud et al.’s study13 both analyzed only patients with septic shock, there are other important differences between populations. Whereas all of our patients had community-acquired infections, more than half the patients in their study had healthcare-associated infections, and the incidence of bacteremia in their early mortality group (20%) was much lower than in ours (64.4%). Bruin-Buisson et al.14 found that bacteremia was associated with an increased risk of early mortality in patients with severe sepsis and septic shock, and our multivariate analysis of risk factors found that bacteremia was independently associated with EM, corroborating these findings.

On the other hand, the incidence of cirrhosis in the group of patients with EM in our study was higher than in those who survived for more than 72hours (13.3% vs. 6.5%, p=0.024); of these, 41% had septic shock of abdominal origin, 58.3% had bacteremia, and 91% had acute kidney injury, all of which are associated with high mortality. Although the prognosis of patients with septic shock and cirrhosis has improved in recent years,27 our results show that cirrhosis continues to increase the risk of poor outcome in patients with septic shock.

An important finding in our study was that the proportion of patients who underwent source control was significantly lower in the EM group (41% vs. 53.2% in the group of patients that survived>72h, p=0.019). This difference is partly explained by the lower frequency of urinary and biliary tract infections in the group with EM which usually require source control. These results would confirm that septic shock due to infections not amenable to source control is associated with poor prognosis.28 On the other hand, although delaying source control increases mortality,29 we found no significant difference in the percentage of patients undergoing early (<12h) source control between the EM group (62.1%) and the group who survived more than 72h (73.3%).

It is important to note not only that the EM group had a lower proportion of patients who needed source control, but also that source control was inadquate in 60% of those who underwent source control interventions, mainly in urinary, abdominal and skin and soft tissue infections. In the multivariate analysis of risk factors for EM, adequate source control was an independent protective factor among the patients with infections amenable to source control, while in the group that did not require source control, non-respiratory infections (73% of cases) were associated with higher EM. These results support Kumar et al.’s30 theory that the total microbial load is what drives shock in septic shock, so reducing the microbial load early through effective source control and appropriate antimicrobial treatment is crucial for survival. Another study that analyzed autopsy findings in surgical patients who died of sepsis/septic shock found a persistent septic focus in approximately 75%,31 indicating that ineffective measures to eliminate the infection decrease the chances of survival.

Finally, among the organ failures associated with septic shock, ARDS was more common in patients who died early (40% vs. 19.8% in the group of survivors over 72h, p<0.001); furthermore, in the multivariate analysis ARDS was independently associated with increased risk of dying early. Zilberberg et al.32 reported that sepsis as the origin of ARDS was independently associated with worse prognosis. The greater proportion of patients with inadequate source control in the EM group could also explain the higher incidence of ARDS in this group. In an experimental model of peritonitis in rabbits, Matute-Bello G et al.33 showed that the ability to contain the infection at the primary site appear to be an important factor affecting the severity of septic shock and distant organ injury. Inadequate source control and failure to reduce the bacterial load at the site of infection could result in a more severe systemic response that could explain the refractoriness of septic shock and the persistency of lung injury.

This study has several limitations. It was a retrospective study, done at a single center, so the results cannot be extrapolated to other populations in hospitals with different characteristics. Moreover, although we adjusted for a number of predisposing patient factors, there may be other confounding factors that we did not measure. We analyzed only clinical and microbiological data, and we cannot rule out the possibility that other variables such as lactate level or lactate clearance might have affected our results. Furthermore, although nearly 90% of patients in both groups received appropriate empirical antibiotic treatment, we could not assess the time to initiation of antibiotic treatment, so we cannot rule out the possibility of greater delays in the early mortality group. Unfortunately, we have no data about the duration of symptoms before patients reach the hospital. We must also consider the definition of the adequacy of source control used in our study, which may lead to bias in the interpretation of the results. However, it is necessary to consider the difficulty of evaluating the response to source control in critically ill patients, as has been previously stated by other authors.34,35 However, our study also has its strengths. The data come from a homogeneous database collected over a long period of time with strict quality-control measures to ensure validity; thus, we were able to analyze a large uniform cohort of patients with septic shock according to Sepsis-3 definitions who were monitored until death or hospital discharge.

Our study confirms a trend over time in decreasing the ICU-mortality of patients admitted to the ICU with septic shock. However, mortality within the first 72h occurs in more than half of all patients who die in the ICU, mainly in those patients with non-respiratory infections who do not need source control or in patients with infections amenable to source control but this is inadequate.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Jordi Vallès, responsible for the design of the study, review of patient data and preparation of the article.

Andrey Rodriguez, Jaume Mestre, Consuelo Guía and Ana Navas, contributed to the acquisition of patient data included in the study.

Emili Diaz, Ana Ochagavia, Melcior Martinez, Guillem Gruartmoner, Candelaria de Haro and Jaume Mesquida review and elaboration of the document.

Joan Carles Oliva, review and statistical analysis of the study.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The authors confirm that they have no conflicts of interest in the elaboration of this article.