Sepsis is one of the leading causes of mortality and morbidity worldwide (1–8). Globally, an estimated 20 to 30 million patients develop sepsis annually, and around 8.7 million people die from sepsis each year (5, 6, 9). The Emergency Department (ED) is an important component of sepsis care because early treatment with antibiotics and intravenous fluids can reduce the number of avoidable deaths and has the strongest impact on sepsis survival (10–14).
A previous study that used the nationwide inpatient sample of the Untied States, a national probability sample, demonstrated that patients with sepsis who were admitted from the ED were associated with a lower likelihood of overall inpatient mortality compared with patients directly admitted to the hospital floor (15). The authors postulated that the improved mortality could be a result of the earlier and superior critical care provided in the ED. However, the observed differences in mortality associated with different routes of admission have not been validated in other cohorts.
In addition, variables such as patient demographics, sites of infection, outcomes, and the temporal trend of incidence and mortality may differ between patients with sepsis admitted from the ED and patients who are hospitalized through other venues (15). However, there is a paucity of literature on the epidemiology of sepsis in ED patients using population-based database (15, 16). Understanding the true burden of sepsis cases that present to the ED is of paramount importance. Data regarding the volume of patients, their characteristics, and subsequent outcomes could be beneficial in developing strategies that optimize the delivery of care for sepsis and guide the allocation of limited healthcare resources.
Thus, the aim of this work is 2-fold. First, we sought to study the temporal trend of incidence and mortality in hospitalized patients with sepsis admitted through the ED. Second, we compared short-term mortality in ED-admitted versus non-ED-admitted patients. Prior studies were probability sample of hospital discharge and required weighting of the selected sample, which is often complicated by the inaccuracy in the determination of the hospital service population due to the heterogeneity of service population across different participating hospitals. To avoid selection bias, we utilized a nationwide administrative database that was linked to a death certificate database to analyze the true population incidence and mortality of sepsis over a prolonged time period. Moreover, with a valid record of death, we could also avoid underestimating the mortality rate of sepsis, which is a limitation when determining mortality from the survival status at discharge from a discharge record database.
PATIENTS AND METHODS
This study used the entire health insurance claims data from the National Health Informatics Project, which was released by the Collaboration Center of Health Information Application of Taiwan. Taiwan's National Health Insurance is a single-payer health insurance program, in which 99.8% of the 23 million people in the Taiwanese population are enrolled. We used the nationwide database for all participants from January 2001 to December 2012 to identify patients with severe sepsis and septic shock. Data available for analysis include demographics (age, sex, area of residence, insurance premium); ICD-9-CM codes for diagnoses and procedures; admission source (emergency department, nursing home, or transfer from outside hospital); admission and discharge dates; discharge disposition (home, nursing home, transfer to another hospital, discharge against medical advice, death). Death data were obtained from the linked national death certificate registry database. National Taiwan University Hospital's institutional review board approved this study. Since this is an electronic database study using anonymous subjects, patient consent was not required.
We defined sepsis as the presence of diagnostic codes for both (1) infection and (2) evidence of organ dysfunction, which is compatible with the latest sepsis-3 definition (17). This approach has been widely used in sepsis epidemiology studies and has been shown to have the highest sensitivity among all published code abstraction methods (2, 3, 18). Previous validation study using chart review found that sepsis patients identified by this approach have a specificity of 83.3% and specificity of 98.9% in Taiwan (19). In general, we followed Angus’ ICD-9 code list to identify cases of sepsis by selecting all cases with ICD-9-CM codes for both a bacterial or fungal infectious process (Supplementary Appendix 1, https://links.lww.com/SHK/A785) and a diagnosis of acute organ dysfunction (Supplementary Appendix 2, https://links.lww.com/SHK/A785) (3). Patients with septic shock required an additional ICD-9-CM code for shock or hypotension. We defined mortality as 30-day or 90-day all-cause mortality, which was verified by a linked national death certificate database.
We collected the following patient information for analysis: age, sex, source of infection, presence of key comorbid conditions, system or organ dysfunction, intensive care unit (ICU) and hospital length of stay, and mortality. Geographic regions were based upon a set of validated codes that are correlated with the degree of urbanization in Taiwan. All patient characteristics were collected on the indexed emergency department visit or hospitalization, except for the presence of key comorbid conditions. The presence of key comorbid conditions was collected 1 year before the index date and identified using algorithms developed by Charlson (20). We defined the index date as the first day of emergency department or hospital admission due to sepsis.
Characteristics and outcomes were compared between ED-admitted and non-ED-admitted patients with sepsis. Categorical variables were presented as frequency and percentage, and continuous variables were presented as a mean ± standard deviation. We compared continuous variables with Mann–Whitney U tests and categorical data with the chi-square or Fisher exact test as appropriate. Cox proportional hazard models were used to compare mortality between ED and non-ED-admitted patients with sepsis. Crude and adjusted hazard ratios (HRs) of 30-day or 90-day mortality were calculated comparing ED admission versus non-ED admission, and 95% CIs were estimated by Cox model. The HRs were adjusted for age, sex, comorbidity, and source of infection. We further performed propensity score (PS) matching and estimated PS-matched results. PS was defined as the conditional probability of hospitalization from sepsis, which was derived from the logistic regression model that included all potential predictors. To verify that the baseline covariates were balanced after PS matching, we ensured minimum differences in the baseline covariates between ED and non-ED-admitted patients by using a standardized difference plot (Supplementary efigure 1, https://links.lww.com/SHK/A785). We determined the annual incidence as the number of sepsis events divided by the number of enrollees in a given year. The temporal trends of sepsis incidence and mortality were presented with line graphs. Since our nationwide database allowed us to capture all cases of sepsis, confidence intervals that show the range of sampling variation were not thought to be necessary in all estimates. All analyses were performed using SAS Version 9.4 (Cary, NC). A 2-sided P value < 0.05 was viewed as significant.
During the 11-year study period, we identified 1,256,684 patients with sepsis. Of these patients, 493,397 (29.3%) were admitted from the ED, and 763,287 (70.7%) were admitted directly to the hospital floor (Fig. 1). Compared with patients admitted directly to the hospital floor, patients admitted from the ED were younger, more likely to be male, and had a higher prevalence of diabetes mellitus, dementia, psychosis, HIV infection, liver cirrhosis, and autoimmune disorders (Table 1). Patients admitted directly to the floor had a higher prevalence of hypertension, chronic pulmonary disease, stroke, congestive heart failure, chronic renal disease, and any tumor. When differentiating the source of infection, patients with sepsis admitted from the ED were more likely to have lower respiratory tract, genitourinary tract, intra-abdominal, skin and skin structures, biliary tract, musculoskeletal, and systemic fungal infections. Compared with non-ED admitted patients, ED-admitted patients had a lower prevalence of acute respiratory failure, lower 30-day mortality, and shorter length of hospital stay and ICU stay (Table 2). However, ED-admitted patients had higher prevalence of central nervous system dysfunction, hematologic system dysfunction, hepatic dysfunction, renal system dysfunction, and metabolic system dysfunction.
Route of admission in patients with sepsis
Supplementary eFigure 2 (https://links.lww.com/SHK/A785) shows the number of admissions for sepsis between 2002 and 2012 according to the source of admission. From 2002 to 2012, the incidence of ED-admitted patients with sepsis increased from 237.3 per 100,000 persons to 369.6 per 100,000 persons (5.06% annual increase), while that of non-ED-admitted patients remained relatively unchanged from 399.7 per 100,000 persons in 2002 to 402.5 per 100,000 persons in 2012 (0.06% annual increase). During the study period, patients with septic shock followed a similar incidence trend.
Mortality of sepsis and septic shock
The mortality rate from 2002 to 2012, stratified by source of admission, is shown in Supplementary eFigure 3 (https://links.lww.com/SHK/A785). In general, mortality in patients with sepsis decreased over the 11-year period. For ED-admitted patients with septic shock, there was a pronounced decline in mortality over time, with mortality decreasing from 34.8% in 2002 to 26.6% in 2012 (annual decrease: 2.14%). Non-ED admitted patients with septic shock, however, experienced a slower rate of reduction in mortality, with mortality decreasing from 45.4% in 2002 to 42.0% in 2012 (annual decrease: 0.69%).
For patients with sepsis, mortality in ED-admitted patients was also substantially improved, as mortality decreased from 27.2% in 2002 to 21.1% in 2012 (annual decrease: 2.04%). Non-ED-admitted patients had a slower rate of reduction in mortality, with mortality decreasing from 35.3% in 2002 to 30.7% in 2012 (annual decrease: 1.16%).
Comparison of mortality between ED-admitted and non-ED-admitted patients
To confirm that the difference in mortality between ED-admitted and non-ED-admitted patients was associated with the route of admission rather than differences in demographic and comorbid characteristics, we performed multivariate and propensity score matched analyses (Table 3). ED-admitted patients were associated with a decreased risk of 90-day mortality in the unadjusted analysis (HR: 0.88, 95% CI: 0.85–0.91), and PS-matched analysis (HR: 0.93, 95% CI: 0.89–0.97). However, there was no significant difference in 30-day mortality between ED-admitted and non-ED-admitted in the PS-matched analysis (HR: 0.91, 95% CI: 0.87–0.95).
In this large, national study that gathered prospective information on 1,256,684 person-years, we confirmed that patients admitted through the ED were associated with a lower 90-day mortality (PS-matched HR: 0.93, 95% CI: 0.89, 0.97). In addition, we found that patients with sepsis were increasingly using the ED as the primary route of hospital admission. Over time, mortality due to sepsis decreased at a faster rate in patients admitted from the ED compared with patients admitted directly to the hospital floor.
Two ICD-9 code abstracting strategies are commonly used for the identification of patients with sepsis (3, 4). The single code abstraction strategy relies on the physician's clinical diagnosis of sepsis, which has been shown to identify patients with sepsis who were more severely ill (21–23). The combination code abstraction strategy requires a code for an infectious disease and a code suggesting organ dysfunction. This combination strategy has been shown to identify a greater number of patients with sepsis and also those with a lower average severity of disease (3, 21–23). We chose to use the combination code abstraction strategy since infection plus organ dysfunction is most compatible with the latest Sepsis-3 definition. The combination code strategy is also less affected by the ICD-9 CM code upcoding practice due to increased awareness of sepsis over time (24, 25). In addition to the original combination code strategy proposed by Angus et al., we also included patients with a sole diagnosis code of “septicemia” or “sepsis.”
In general, the demographic features, comorbidity profiles, and source of infection of our sepsis cohort resemble that of the US (16). Our sepsis cohort differs from a Mainland China sepsis cohort in the source of infection and the outcome (26). In Zhou report, the prevalence of the leading two infection sources, pneumonia and genitourinary tract infection accounted for 72.7% and 8.9% of the sepsis patients, respectively. This is in contrast to the 55% pneumonia and 37% genitourinary tract infection as observed in our cohort. In addition, the mortality rate of septic shock in Zhou study was 84.6%, in contrast to the 35% as observed in our cohort.
In addition, in accordance with a previous report, we observed an increasing trend of patients with sepsis who used the ED as their main route of admission (2). The frequency of direct admissions to the hospital floor for patients with sepsis, however, remained relatively stable during the 11-year study period (Supplementary Figure 2, https://links.lww.com/SHK/A785). Possible alternative explanations include the higher rate of community-onset sepsis compared to hospital-onset sepsis (27). Although community-onset sepsis may explain the increasing trend toward patients with sepsis presenting at the ED, epidemiology reports comparing the incidence of community-onset versus hospital-onset sepsis remain scarce (28). Therefore, further studies involving electronic medical records are needed to differentiate community-onset and hospital-onset sepsis, and whether this plays a role in the incidence trends related to route of admission.
The significant difference in mortality between ED and non-ED admitted patients with sepsis was previously noted in Powell et al.'s (15) study, which found that cases of severe sepsis/septic shock admitted through the ED were associated with an 8% decreased risk in early inpatient mortality and 13% decreased risk in overall inpatient mortality after adjusting for demographic, comorbid, and hospital characteristics. By PS-matching, we demonstrated that ED-admitted patients had a 2% lower risk of mortality at 30-days and 7% lower risk of mortality at 90-days. The stronger effect on 90-day mortality is likely because of more deceased outcomes.
The evidence-based Surviving Sepsis Campaign guideline promotes early recognition, early antibiotics and source control, and early fluid resuscitation to reduce mortality in sepsis (29). Recent statewide evaluations in the United States showed that the time taken for the completion of a 3-h bundle in the emergency department is highly correlated with in-hospital mortality from sepsis (30). In Canada, McColl et al. (31) found that the implementation of a multidisciplinary ED sepsis bundle, including improved early identification and protocolized medical care, was associated with improved time to achievement of key therapeutic interventions and a reduction in 30-day mortality. The implementation of these time-sensitive guidelines required high levels of resources and well trained staff, such as those found in the ED. This may explain why patients with sepsis who present to the ED are more likely to receive the evidence-based bundle of care within the critical time window. Nonetheless, this hypothesis needs to be verified in future studies that collect complete information on the timing of recognition and management of sepsis, which is currently not available in an administrative database.
In addition to benefitting from a higher likelihood of receiving an early diagnosis and subsequent treatment, ED-admitted patients with sepsis are more likely to have community-onset sepsis. Community-onset sepsis is less likely to be caused by drug resistant organisms and has been shown to have superior prognosis compared to hospital-onset sepsis (27, 32). To avoid the inclusion of nosocomial infection in directly hospitalized patients, we did not include transferred patients and used the admission diagnosis, rather than the discharge diagnosis, as the primary inclusion criteria for directly admitted patients. However, without microbiology data, it is impossible to verify the distribution of resistant bacteria in the two comparison cohorts. Further studies are needed to verify whether the outcome difference is due to difference in the initial resuscitation effort or difference in the bacteriology between ED-admitted or directly admitted patients. Before such study is available, our study, in line with previous observation, which show that sepsis patients with similar demographic features, comorbidity, and source of infection may have different outcome for different admission routes. The policy that encourage direct admission from outpatient clinic to alleviate the worsening ED crowding problems may need to consider the different outcomes of patients shown in our and previous study (15).
Strengths and limitations
Results of this study should be interpreted in light of both its strengths and weaknesses. There are some limitations to this epidemiology study. The records of fluid resuscitation and the timeliness of antibiotic administration are not available. In addition, our analysis focuses on the ED and cannot be extrapolated to other areas of hospital care. Numerous sepsis clinical initiatives, such as the international Surviving Sepsis Campaign, were developed during the time period of our study and may have influenced sepsis detection, increasing the total number of identified sepsis cases (29, 33). In spite of the above limitations, the main strengths of the study are as follows. As far as we are aware, this the first true nationwide population-based study comparing patients with sepsis admitted via different routes. The universal insurance coverage in Taiwan provides health care utilization information for nearly 100% of the population. We integrated an approach to comprehensively ascertain the incidence trend of sepsis cases and associated mortality within the context of the most recent definition of sepsis, which could provide a new code abstraction strategy for future studies. By using a linked death certificate database, we were able to avoid underestimating mortality from sepsis by using inpatient mortality as our primary endpoint.
Although the majority of sepsis patients are admitted through the ED, direct admissions constitute a significant proportion of hospital admissions. There has been report that increasing rates of direct admission among patients with access to outpatient care might be an effective strategy to reduce hospital costs and the volume of patients in the ED. Despite this, past studies have not compared the outcomes of patients admitted from these two routes. Additional research is needed to establish direct admission policies are safe. Our study shows, although ED-admitted sepsis patients had more organ dysfunction than directly admitted sepsis patients, they had more favorable mortality, length of ICU stay, and length of hospital stay. Results of our study were in line with previous observation. Further studies are needed to clarify the potential causes underlying the observed differences in mortality associated with different routes of admission.
The authors thank the staff of the Core Labs at the Department of Medical Research in National Taiwan University Hospital for technical support, Medical Wisdom Consulting Group for technical assistance in statistical analysis, and National Taiwan University Hospital Health Economics and Outcome Research Group for advice on study design.
1. Angus DC, van der Poll T. Severe sepsis
and septic shock. N Engl J Med
369 9:840–851, 2013.
2. Wang HE, Shapiro NI, Angus DC, Yealy DM. National estimates of severe sepsis
in United States emergency departments. Crit Care Med
35 8:1928–1936, 2007.
3. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR. Epidemiology of severe sepsis
in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med
29 7:1303–1310, 2001.
4. Martin GS, Mannino DM, Eaton S, Moss M. The epidemiology of sepsis
in the United States from 1979 through 2000. N Engl J Med
348 16:1546–1554, 2003.
5. Fleischmann C, Scherag A, Adhikari NK, Hartog CS, Tsaganos T, Schlattmann P, Angus DC, Reinhart K. International Forum of Acute Care Trialists. Assessment of global incidence and mortality of hospital-treated sepsis
. current estimates and limitations. Am J Respir Crit Care Med
193 3:259–272, 2016.
6. Jawad I, Luksic I, Rafnsson SB. Assessing available information on the burden of sepsis
: global estimates of incidence, prevalence and mortality. J Glob Health
2 1:010404, 2012.
7. Mikkelsen ME, Shah CV, Meyer NJ, Gaieski DF, Lyon S, Miltiades AN, Goyal M, Fuchs BD, Bellamy SL, Christie JD. The epidemiology of acute respiratory distress syndrome in patients presenting to the emergency department with severe sepsis
40 5:375–381, 2013.
8. Cheng B, Li Z, Wang J, Xie G, Liu X, Xu Z, Chu L, Zhao J, Yao Y, Fang X. Comparison of the performance between Sepsis
-1 and Sepsis
-3 in ICUs in China: a retrospective multicenter study. Shock
48 3:301–306, 2017.
9. Gobatto AL, Besen BA, Azevedo LC. How can we estimate sepsis
incidence and mortality? Shock
47 (1S):6–11, 2017.
10. Rivers E, Nguyen B, Havstad S, Ressler J, Muzzin A, Knoblich B, Peterson E, Tomlanovich M. Early Goal-Directed Therapy Collaborative Group. Early goal-directed therapy in the treatment of severe sepsis
and septic shock. N Engl J Med
345 19:1368–1377, 2001.
11. Gaieski DF, Mikkelsen ME, Band RA, Pines JM, Massone R, Furia FF, Shofer FS, Goyal M. Impact of time to antibiotics on survival in patients with severe sepsis
or septic shock in whom early goal-directed therapy was initiated in the emergency department. Crit Care Med
38 4:1045–1053, 2010.
12. Seymour CW, Gesten F, Prescott HC, Friedrich ME, Iwashyna TJ, Phillips GS, Lemeshow S, Osborn T, Terry KM, Levy MM. Time to treatment and mortality during mandated emergency care for sepsis
. N Engl J Med
376 23:2235–2244, 2017.
13. Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, Suppes R, Feinstein D, Zanotti S, Taiberg L, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med
34 6:1589–1596, 2006.
14. Holler JG, Jensen HK, Henriksen DP, Rasmussen LM, Mikkelsen S, Pedersen C, Lassen AT. Etiology of shock in the emergency department; a 12 year population based cohort study. Shock
51 1:60–67, 2019.
15. Powell ES, Khare RK, Courtney DM, Feinglass J. Lower mortality in sepsis
patients admitted through the ED vs direct admission. Am J Emerg Med
30 3:432–439, 2012.
16. Wang HE, Szychowski JM, Griffin R, Safford MM, Shapiro NI, Howard G. Long-term mortality after community-acquired sepsis
: a longitudinal population-based cohort study. BMJ Open
4 1:e004283, 2014.
17. Shankar-Hari M, Phillips GS, Levy ML, Seymour CW, Liu VX, Deutschman CS, Angus DC, Rubenfeld GD, Singer M. Sepsis
Definitions Task Force. Developing a new definition and assessing new clinical criteria for septic shock: for the third international consensus definitions for sepsis
and septic shock (Sepsis
315 8:775–787, 2016.
18. Iwashyna TJ, Odden A, Rohde J, Bonham C, Kuhn L, Malani P, Chen L, Flanders S. Identifying patients with severe sepsis
using administrative claims: patient-level validation of the angus implementation of the international consensus conference definition of severe sepsis
. Med Care
52 6:e39–e43, 2014.
19. Shen HN, Lu CL, Yang HH. Epidemiologic trend of severe sepsis
in Taiwan from 1997 through 2006. Chest
138 2:298–304, 2010.
20. Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol
47 11:1245–1251, 1994.
21. Gaieski DF, Edwards JM, Kallan MJ, Carr BG. Benchmarking the incidence and mortality of severe sepsis
in the United States. Crit Care Med
41 5:1167–1174, 2013.
22. Wilhelms SB, Huss FR, Granath G, Sjöberg F. Assessment of incidence of severe sepsis
in Sweden using different ways of abstracting International Classification of Diseases codes: difficulties with methods and interpretation of results. Crit Care Med
38 6:1442–1449, 2010.
23. Stevenson EK, Rubenstein AR, Radin GT, Wiener RS, Walkey AJ. Two decades of mortality trends among patients with severe sepsis
: a comparative meta-analysis. Crit Care Med
42 3:625–631, 2014.
24. Walkey AJ, Lagu T, Lindenauer PK. Trends in sepsis
and infection sources in the United States. A population-based study. Ann Am Thorac Soc
12 2:216–220, 2015.
25. Rhee C, Gohil S, Klompas M. Regulatory mandates for sepsis
care—reasons for caution. N Engl J Med
370 18:1673–1676, 2014.
26. Zhou J, Tian H, Du X, Xi X, An Y, Duan M, Weng L, Du B. China Critical Care Clinical Trials Group (CCCCTG). Population-based epidemiology of sepsis
in a subdistrict of Beijing. Crit Care Med
45 7:1168–1176, 2017.
27. Page DB, Donnelly JP, Wang HE. Community-, healthcare-, and hospital-acquired severe sepsis
hospitalizations in the University HealthSystem Consortium. Crit Care Med
43 9:1945–1951, 2015.
28. Shih FY, Ma MH, Chen SC, Wang HP, Fang CC, Shyu RS, Huang GT, Wang SM. ED overcrowding in Taiwan: facts and strategies. Am J Emerg Med
17 2:198–202, 1999.
29. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, Sevransky JE, Sprung CL, Douglas IS, Jaeschke R, et al. Surviving Sepsis
Campaign Guidelines Committee including The Pediatric Subgroup. Surviving Sepsis
Campaign: international guidelines for management of severe sepsis
and septic shock, 2012. Intensive Care Med
39 2:165–228, 2013.
30. Hershey TB, Kahn JM. State sepsis
mandates—a new era for regulation of hospital quality. N Engl J Med
31. McColl T, Gatien M, Calder L, Yadav K, Tam R, Ong M, Taljaard M, Stiell I. Implementation of an emergency department sepsis
bundle and system redesign: a process improvement initiative. CJEM
19 2:112–121, 2017.
32. Jones SL, Ashton CM, Kiehne LB, Nicolas JC, Rose AL, Shirkey BA, Masud F, Wray NP. Outcomes and resource use of sepsis
-associated stays by presence on admission, severity, and hospital type. Med Care
54 3:303–310, 2016.
33. Levy MM, Dellinger RP, Townsend SR, Linde-Zwirble WT, Marshall JC, Bion J, Schorr C, Artigas A, Ramsay G, Beale R, et al. The surviving sepsis
campaign: results of an international guideline-based performance improvement program targeting severe sepsis
. Intensive Care Med
36 2:222–231, 2010.