Sepsis is defined as life-threatening organ dysfunction induced by the dysregulated host response to infection (1). In 2017 sepsis was the 12th leading cause of death in the United States (2) and the most expensive condition to treat (3). The enormous impact of sepsis on the U.S. healthcare system makes analyzing sepsis trends important.
Previous studies utilizing hospital administrative databases and electronic health records to evaluate the sepsis trends have been limited to a few hospital systems and geographic areas and therefore have not been representative of the whole population. These studies also did not include nonhospital settings or children and therefore could not provide true population-based estimates. Mortality trends and comparison of age-adjusted mortality rates by demographics and geography could also not be obtained from these studies.
To overcome these limitations, we chose to use the multiple causes of death (MCOD) database available through the Centers for Disease Control and Prevention (CDC). Previously, a study by Melamed and Sorvillo (4) using the MCOD database showed a slight decline in age-adjusted sepsis-related mortality rates from 1999 to 2005. Sepsis-related mortality rates in the United States since then have not been well characterized. The objective of this study is to analyze sepsis-related mortality trends in the United States from 2005 to 2018, stratified by demographics.
MATERIALS AND METHODS
This is a retrospective, population-based study using the National Center for Health Statistics (NCHS) MCOD database, which is available through the CDC Wide-ranging Online Data for Epidemiologic Research (CDC Wonder) website (5). Mortality information from all death certificates filed is collected by state registries and provided to NCHS. The CDC Wonder website compiles this data and publishes aggregate data and mortality rates. The database contains the following data: an underlying cause of death (UCD), up to 20 additional MCOD, and demographic data. The UCD is defined by the World Health Organization as “the disease or injury which initiated the train of events leading directly to death.” The diseases are coded using the International Classification of Diseases, 10th Revision (ICD-10). Additional information about the database is available in our previous publication (6). This study was reviewed and the need for informed consent was waived by the Loma Linda University Health Institutional Review Board (No. 5210024).
Patients with sepsis-related mortality were identified by using ICD-10 codes similar to prior studies (e-Table 1, http://links.lww.com/CCM/G305) (4,7) listed among any of the 20 causes of death irrespective of the UCD which could be sepsis or other conditions (malignancy, pneumonia, etc.). Sepsis as the UCD is a subset of sepsis-related deaths. With ICD-10 coding, there is no separate code for septicemia or sepsis. Instead, the appropriate code for the underlying systemic infection is used (e-Table 1, http://links.lww.com/CCM/G305). If the type of infection or causal organism is not further specified, code A41.9 (sepsis, unspecified organism) is assigned. The database does not include ICD-10 codes for severe sepsis (R65.20) or septic shock (R65.21); instead, these conditions are coded for the underlying systemic infection, or A41.9 if the organism is not documented (8). In our cohort, 94.7% of the decedents were coded with A41.9. All deaths in the United States from 2005 to 2018 meeting the above criteria were included. Mortality rates were stratified by year of death, sex, age, race-ethnicity (race), geography, and place of death. Race was categorized according to U.S. census standards as non-Hispanic White (White), Hispanic, Asian-Pacific Islander (Asian), non-Hispanic Black (Black), and American Indian-Alaska Native (Native American). The place of death was categorized as inpatient, outpatient/emergency department (outpatient), dead on arrival, home, hospice, nursing home-long term care (nursing home), other, and unknown.
Age-adjusted mortality rates were used to compare the groups over time instead of crude mortality rates to ensure that the differences in death rates were not due to the differences in the age distribution of the populations being compared. The mortality rates were calculated using the July 1 population projections from the U.S. Census Bureau for the corresponding year. Age-standardized mortality rates were calculated by standardizing to the 2000 U.S. standard population. Rates are expressed per 100,000 persons. Poisson regression modeling was used to calculate mortality rate ratios and to analyze temporal trends in sepsis-related mortality from 2005 to 2018. Negative binomial regression was used for overdispersed data (deviance/degrees of freedom > 1). All analysis was performed using SPSS Version 25 (IBM Corp, Armonk, NY).
From 2005 to 2018, there were 36,067,309 deaths in the United States. Of these, 6.7% (2,427,907 patients) had a diagnosis of sepsis.
Table 1 shows the differences in sepsis-related mortality stratified by sex, age, and race. When adjusted for age, males had a higher mortality rate than females. When stratified by age groups, the lowest crude mortality rate was in the less than 45 years old group, which increased in every subsequent older age group. The lowest mortality rate was seen in Asians, followed by Whites and Hispanics, and the highest was seen in Native Americans and Blacks.
TABLE 1. -
Average Annual Sepsis
-Related Mortality Rates by Demographic Characteristics in the United States, 2005–2018
||No. of Deaths (%)
||Mortality Rate per 100,000 (95% CI)
||Mortality Rate Ratio (95% CI)
| < 45
| ≥ 85
| Native American
aMortality rates are reported as age-adjusted rates.
cp < 0.001 for trend.
dp < 0.001 for trend.
Mortality rate ratios are adjusted for age and sex.
Overall, the sepsis-related mortality rate was 50.2 per 100,000, and the rate remained stable during this period (p = 0.15). The percent rate change between 2005 and 2018 stratified by demographics is shown in Table 2. The mortality rate for males was 57.0 and for females was 45.1 per 100,000 persons. The rates remained relatively stable over the study period (p = 0.10 for males and p = 0.29 for females) (e-Fig. 1, http://links.lww.com/CCM/G306; legend, http://links.lww.com/CCM/G314).
TABLE 2. -
Percent Change in Sepsis
-Related Mortality Rates by Demographics in the United States, 2005–2018
|Mortality Rate 2005 per 100,000
||Mortality Rate 2018 per 100,000
||% Rate Change
||Mortality Rate 2005 per 100,000
||Mortality Rate 2018 per 100,000
||% Rate Change
| < 45
| ≥ 85
| Native American
aMortality rates are reported as age-adjusted rates.
Dashes denote that the regression analysis of the data did not indicate a significant change in trend during this period.
Mortality trends stratified by gender and age are shown in e-Figure 2 (http://links.lww.com/CCM/G307; legend, http://links.lww.com/CCM/G314). In females, the sepsis mortality rate increased significantly in age groups 45–54 years (p < 0.001) and 55–64 years (p < 0.001) and was unchanged in the other age groups (e-Fig. 2a, http://links.lww.com/CCM/G307; legend, http://links.lww.com/CCM/G314). In males, the sepsis mortality rate increased significantly in age groups 55–64 years (p < 0.001), 65–74 years (p = 0.001), and greater than 85 years (p < 0.001) and was unchanged in the other age groups (e-Fig. 2b, http://links.lww.com/CCM/G307; legend, http://links.lww.com/CCM/G314).
Mortality trends stratified by gender and race are shown in e-Figure 3 (http://links.lww.com/CCM/G308; legend, http://links.lww.com/CCM/G314). In females, mortality rate increased in Native Americans (p < 0.001) and Whites (p = 0.005) and decreased in Blacks (p < 0.001), Hispanics (p = 0.01), and Asians (p = 0.02) (e-Fig. 3a, http://links.lww.com/CCM/G308; legend, http://links.lww.com/CCM/G314). A similar trend was noted in males; mortality rate increased in Native Americans (p < 0.001) and Whites (p = 0.002) and decreased in Blacks (p < 0.001), Hispanics (p = 0.04), and Asians (p = 0.01) (e-Fig. 3b, http://links.lww.com/CCM/G308; legend, http://links.lww.com/CCM/G314).
The mortality rates varied by region with the highest risk in the South (56.3/100,000) and the lowest risk in the West (44.0/100,000) (Fig. 1; and e-Table 2, http://links.lww.com/CCM/G309). There was an increased variation in the mortality rates for sepsis among states (e-Fig. 4, http://links.lww.com/CCM/G310; legend, http://links.lww.com/CCM/G314). The highest mortality rates were in the District of Columbia and Mississippi (79.3 and 70.8/100,000, respectively), and the lowest mortality rates were in Vermont and Minnesota (31.7 and 28.8/100,000, respectively).
Over the study period, sepsis was listed as the UCD in 21% of patients with sepsis-related deaths. Malignant neoplasms, influenza and pneumonia, and chronic heart diseases were listed as the UCD in 12%, 8%, and 7% of patients, respectively. Chronic renal disease, chronic pulmonary disease, diabetes mellitus, cerebrovascular disease, Clostridium difficile enterocolitis, and aspiration pneumonitis were listed as the UCD in 2–4% of the patients (Fig. 2). Trends in mortality rates for UCD are shown in e-Figure 5 (http://links.lww.com/CCM/G311; legend, http://links.lww.com/CCM/G314). The mortality rate from sepsis was unchanged (mean annual death rate, 10.8; p = 0.34). The mortality rates from ischemic heart disease decreased (2.4 in 2005 to 1.5 in 2018/100,000; p = 0.04) and from pneumonia increased (3.5 in 2005 to 4.6 in 2018/100,000; p = 0.02).
Over the study period, the majority of deaths occurred in the inpatient setting (81.1%), followed by nursing home (7.5%), hospice (3.3%), and home (3.3%) (Fig. 3). The percentage of deaths decreased significantly at nursing homes (9.5% in 2005 to 6.4% in 2018; p = 0.001) and increased at home (2.9% in 2005 to 4.4% in 2018; p = 0.001) and hospice (0.5% in 2004 to 6.3% in 2018; p < 0.001) and was unchanged in the other settings (e-Fig. 6, http://links.lww.com/CCM/G312; legend, http://links.lww.com/CCM/G314).
Our study represents an important examination of the epidemiology of sepsis-related mortality in the United States from 2005 to 2018 (4). In our study, we noted that 6.7% of all deaths during this period were sepsis-related, which is consistent with prior reporting (4). The mortality rate remained relatively stable for this period with an average annual sepsis-related mortality rate of 50.2 per 100,000 population (57.0 in males and 45.1 in females). This trend is consistent with the modest decline in sepsis-related mortality rate (0.18% per year) noted by Melamed and Sorvillo (4) for the period 1999–2005. A recent study evaluating the global burden of sepsis also noted a decline in sepsis-related mortality worldwide (9). Of the sepsis-related deaths identified in our study, sepsis was listed as the UCD in only 21%. Changing our case definition of sepsis to include only cases in which sepsis was listed as the UCD also showed a stable mortality trend. Since most sepsis-related deaths occur in elderly patients who are prone to sepsis due to underlying comorbidities, the use of the MCOD database allowed us to capture all decedents with a sepsis diagnosis on the death certificate.
Previous studies using administrative databases have shown an increase in sepsis cases (10, 11). The widespread efforts to improve awareness of sepsis and increased reimbursements for sepsis have resulted in increased detection and labeling of patients with sepsis, particularly in those with milder disease. This is demonstrated in a study by Rhee et al (12) showing an increase in the cases of hospitalizations for sepsis but a stable number of hospitalizations with clinical markers of sepsis (positive blood cultures, vasopressor use, and elevated lactate). A large study supported by the CDC Prevention Epicenters evaluated the sepsis cases using a validated clinical surveillance definition using electronic health records data from 408 hospitals (13). This study demonstrated that the sepsis cases was stable from 2009 to 2014. The inhospital mortality rate declined, but there was no significant change in the combined outcomes of death or discharges to hospice.
In our study, there were several demographic differences noted in sepsis-related mortality. We noted that males had an approximately 26% higher mortality rate. Men were more likely to experience sepsis-related death than women in all races. Previously, the influence of gender in sepsis-related mortality was not clear with multiple studies (with small sample sizes) yielding conflicting results (14–17). In a recent study evaluating the global burden of sepsis, the estimated cases of sepsis was higher in females. However, consistent with our study, the sepsis-related mortality rates were higher in males. The gender difference in sepsis mortality is likely the result of a combination of gender differences in biology, comorbidities, sites of infection, and gender biases in assessment and care.
Although the overall sepsis-related mortality rates were stable, we noted an increase in mortality rates in certain age groups. The largest increase in sepsis-related deaths between 2005 and 2018 was noted in the age group 55–64 years in males (17.9%) and age group 45–64 years in females (16–17.9%). To determine if this rise could be due to increased sepsis awareness and coding, we evaluated the sepsis-related deaths in which sepsis was listed as the UCD in patients 45–64 years old. The mortality rates with sepsis as the UCD were stable during this period. Therefore, the increase in sepsis-related mortality in these age groups is likely the result of an increase in comorbidities such as obesity and malignancies (18). Malignancies as the UCD increased by 17.8% in patients 45–64 years old during this period. An increase in the deaths due to drug overdoses and alcohol abuse, which are risk factors for sepsis, has also been reported in this age group (19).
Our study also revealed significant racial disparities in sepsis-related mortality. Whites and Hispanics had similar mortality rates, and Asians had the lowest rates. Blacks and Native Americans experienced a higher mortality rate than Whites. These racial differences were consistent across all age groups (e-Fig. 7, http://links.lww.com/CCM/G313; legend, http://links.lww.com/CCM/G314). The differences are likely multifactorial, but disparities in primary care access, income, education, and lifestyle are likely major factors. Studies have reported that Blacks are disproportionally admitted to urban teaching hospitals with poorer quality measures and outcomes of care (20). Underlying genetics and comorbidities also likely influence the cases, biologic response to infection, and clinical response to different treatments. Blacks more commonly have inflammatory diseases such as systemic lupus erythematosus and sickle cell disease which increases their risk for sepsis. HIV/AIDS is also reported more commonly in Blacks and Hispanics (21). Compared with Whites, Blacks had a nine-fold higher age-adjusted mortality rate from one of these listed conditions and concomitant sepsis (in the analysis using the MCOD database), although this was a small percentage of the overall sepsis-related deaths. We also noted an increase in mortality rates in Native Americans and a significant decrease in Blacks, Hispanics, and Asians. The decreasing mortality seen in non-Native American minorities is likely the result of improved access, which may have been further facilitated by the Affordable Care Act (22,23). In addition, improved quality of medical care and innovations such as the development of highly effective antiretroviral therapy to fight HIV has likely contributed to this decrease (24). The higher mortality in Native Americans is likely due to the increased prevalence of certain comorbidities and social determinants. Native Americans continue to have difficulty obtaining high quality and timely healthcare due to predominantly rural living and limited preventive care (25).
There was significant variability noted in sepsis-related mortality rates between geographic locations, with the lowest rate of 28.8 per 100,000 population in Minnesota and the highest in the District of Columbia at 79.3 per 100,000 population. Overall, the mortality rates were higher in the Southern states and lower in the Western and Northwestern states (e-Fig. 4, http://links.lww.com/CCM/G310; legend, http://links.lww.com/CCM/G314). This database does not allow further investigation into the causes of these differences, but they are likely due to variations in comorbidities as well as differences in sepsis diagnosis and coding practices.
Our study found that the locations of sepsis-related death have significantly changed over the study period. While the majority of deaths still occur in the inpatient setting, the percentage of deaths occurring at home and hospice have significantly increased, and deaths in nursing homes have decreased. This demonstrates improvement in the quality of end-of-life care for patients with sepsis. Provisions established by the Affordable Care Act have also facilitated this change by improving both access and insurance coverage for hospice services at the end of life (26). Consistent with our data, other studies have also shown an increase in hospital discharges to hospice for patients with sepsis (27). Transition to hospice also benefits the hospitals when reporting hospital mortality data as hospice usage converts an inhospital death to a hospice death (27). Further studies should explore racial and geographic disparities in in-home death and hospice utilization in patients with sepsis.
Implications and Next Steps
Our findings of the substantial variation in sepsis mortality between races, age groups, and states have several key implications for policymakers, clinicians, and researchers. Further investigation to understand the reasons for these disparities and developing policies and practices to close these gaps are crucial. In patient groups and locations with high sepsis-related mortalities, measures proven to improve sepsis outcomes should be implemented.
This study has several strengths. The use of the MCOD database allowed us to provide true population-based estimates since every death in the United States is reported. Our findings indicate that 18.9% of sepsis-associated deaths occur in nonhospital settings which would not be accounted for in studies using hospital discharge databases. The death certificate documentation of sepsis is also less likely to be influenced by efforts and policies to label patients with sepsis for “benchmarking” or to optimize reimbursement.
The main limitation is that we cannot confirm the accuracy of the data. Misclassification of sepsis-related death can result from both errors in diagnosis and the completion of death certificates. Similar to our study, ICD-10 codes for sepsis have been used to study sepsis in administrative datasets (7,28). However, to the best of our knowledge, ICD-10 codes for sepsis have not been validated in U.S. studies. Previous studies have shown lower sensitivity for explicit sepsis codes compared with implicit codes (infection and organ dysfunction); however, the use of explicit codes have a low likelihood of false positives (29). In addition, the sensitivity of explicit codes increases with increased severity of sepsis and septic shock (12). Most decedents included in our study would be expected to have experienced severe sepsis even when organ failure is not listed on the death certificate because organ failure is the mechanism by which sepsis causes death. Rates calculated in our study may also be biased by errors in the census count.
From 2005 to 2018, 6.7% of the deaths in the United States were sepsis-related. Our study demonstrated that the overall sepsis-related mortality has been stable for this time-period in both males and females. However, there were significant differences in the mortality trends based on race, geographic location, and age group. We noted an increase in mortality rates in females 45–65 years old and males 55–65 years old. Sepsis-related mortality trends decreased significantly in Blacks, Hispanics, and Asians and increased in Whites and Native Americans. Further validation studies are warranted to examine possible genetic and environmental factors contributing to these differences.
1. Singer M, Deutschman CS, Seymour CW, et al. The third international consensus definitions for sepsis
and septic shock (Sepsis
-3). JAMA. 2016; 315:801–810
2. Heron M. Deaths: Leading causes for 2017. Natl Vital Stat Rep. 2019; 68:1–77
3. Liang L, Moore B, Soni A. National inpatient hospital costs: The most expensive conditions by payer, 2017. Available at: https://www.hcup-us.ahrq.gov/reports/statbriefs/sb261-Most-Expensive-Hospital-Conditions-2017.jsp
. Accessed September 23, 2020
4. Melamed A, Sorvillo FJ. The burden of sepsis
-associated mortality in the United States from 1999 to 2005: An analysis of multiple-cause-of-death data. Crit Care. 2009; 13:R28
5. Centers for Disease Control and Prevention Wide-Ranging Online Data for Epidemiologic Research. Multiple cause of death, 1999-2018 2020. Available at: https://wonder.cdc.gov/mcd-icd10.html
. Accessed June 19, 2020
6. Jeganathan N, Smith RA, Sathananthan M. Mortality trends of idiopathic pulmonary fibrosis in the United States from 2004 to 2017. Chest. 2021; 159:228–238
7. Paoli CJ, Reynolds MA, Sinha M, et al. Epidemiology and costs of sepsis
in the United States-an analysis based on timing of diagnosis and severity level. Crit Care Med. 2018; 46:1889–1897
8. Centers for Disease Control and Prevention: ICD-10-CM Official Guidelines for Coding and Reporting FY 2020. 2020. Available at: https://www.cdc.gov/nchs/data/icd/10cmguidelines-FY2020_final.pdf
. Accessed December 23, 2020
9. Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis
incidence and mortality, 1990-2017: Analysis for the Global Burden of Disease Study. Lancet. 2020; 395:200–211
10. Gaieski DF, Edwards JM, Kallan MJ, et al. Benchmarking the incidence and mortality of severe sepsis
in the United States. Crit Care Med. 2013; 41:1167–1174
11. Kumar G, Kumar N, Taneja A, et al.; Milwaukee Initiative in Critical Care Outcomes Research (MICCOR) Group of Investigators. Nationwide trends of severe sepsis
in the 21st century (2000-2007). Chest. 2011; 140:1223–1231
12. Rhee C, Murphy MV, Li L, et al.; Centers for Disease Control and Prevention Epicenters Program. Comparison of trends in sepsis
incidence and coding using administrative claims versus objective clinical data. Clin Infect Dis. 2015; 60:88–95
13. Rhee C, Dantes R, Epstein L, et al.; CDC Prevention Epicenter Program. Incidence and trends of sepsis
in US hospitals using clinical vs claims data, 2009-2014. JAMA. 2017; 318:1241–1249
14. Adrie C, Azoulay E, Francais A, et al.; OutcomeRea Study Group. Influence of gender on the outcome of severe sepsis
: A reappraisal. Chest. 2007; 132:1786–1793
15. Madsen TE, Simmons J, Choo EK, et al. The DISPARITY Study: Do gender differences exist in Surviving Sepsis
Campaign resuscitation bundle completion, completion of individual bundle elements, or sepsis
mortality? J Crit Care. 2014; 29:473.e7–e11
16. Pietropaoli AP, Glance LG, Oakes D, et al. Gender differences in mortality in patients with severe sepsis
or septic shock. Gend Med. 2010; 7:422–437
17. Sakr Y, Elia C, Mascia L, et al. The influence of gender on the epidemiology of and outcome from severe sepsis
. Crit Care. 2013; 17:R50
18. Steele CB, Thomas CC, Henley SJ, et al. Vital signs: Trends in incidence of cancers associated with overweight and obesity - United States, 2005-2014. MMWR Morb Mortal Wkly Rep. 2017; 66:1052–1058
19. Woolf SH, Schoomaker H. Life expectancy and mortality rates in the United States, 1959-2017. JAMA. 2019; 322:1996–2016
20. Barnato AE, Alexander SL, Linde-Zwirble WT, et al. Racial variation in the incidence, care, and outcomes of severe sepsis
: Analysis of population, patient, and hospital characteristics. Am J Respir Crit Care Med. 2008; 177:279–284
21. Centers for Disease Control and Prevention. Estimated HIV incidence and prevalence in the United States, 2010–2016. HIV Surveillance Supplemental Report 2019; 24(No. 1). Available at: http://www.cdc.gov/hiv/library/reports/hiv-surveillance.html
. Accessed September 21, 2020
22. Chinai B, Gaughan J, Schorr C. Implementation of the affordable care act: A comparison of outcomes in patients with severe sepsis
and septic shock using the national inpatient sample. Crit Care Med. 2020; 48:783–789
23. Lipton BJ, Decker SL, Sommers BD. The affordable care act appears to have narrowed racial and ethnic disparities in insurance coverage and access to care among young adults. Med Care Res Rev. 2019; 76:32–55
24. Masters RK, Hummer RA, Powers DA, et al. Long-term trends in adult mortality for U.S. Blacks and Whites: An examination of period- and cohort-based changes. Demography. 2014; 51:2047–2073
25. Espey DK, Jim MA, Cobb N, et al. Leading causes of death and all-cause mortality in American Indians and Alaska Natives. Am J Public Health. 2014; 104Suppl 3S303–S311
26. Parikh RB, Wright AA. The affordable care act and end-of-life care for patients with cancer. Cancer J. 2017; 23:190–193
27. Schorr CA, Angelo M, John G, et al. The association of increasing hospice use with decreasing hospital mortality: An analysis of the national inpatient sample. J Healthc Manag. 2020; 65:107–120
28. Buchman TG, Simpson SQ, Sciarretta KL, et al. Sepsis
among Medicare beneficiaries: 1. The burdens of sepsis
, 2012-2018. Crit Care Med. 2020; 48:276–288
29. Fleischmann-Struzek C, Thomas-Rüddel DO, Schettler A, et al. Comparing the validity of different ICD coding abstraction strategies for sepsis
case identification in German claims data. PLoS One. 2018; 13:e0198847