Postdischarge Survival After Sepsis: A Cohort Study

Milo Engoren; Michael D Maile; Troy Seelhammer; Robert E Freundlich; Thomas A Schwann

doi:10.1213/ANE.0000000000006744

. Author manuscript; available in PMC: 2024 Dec 1.

Published in final edited form as: Anesth Analg. 2023 Oct 18;137(6):1216–1225. doi: 10.1213/ANE.0000000000006744

Postdischarge Survival After Sepsis: A Cohort Study

Milo Engoren ^*, Michael D Maile ^*, Troy Seelhammer ^†, Robert E Freundlich ^‡, Thomas A Schwann ^§

PMCID: PMC10842030 NIHMSID: NIHMS1940946 PMID: 37851899

Abstract

BACKGROUND:

After hospital discharge, patients who had sepsis have increased mortality. We sought to estimate factors associated with postdischarge mortality and how they vary with time after discharge.

METHODS:

This was a retrospective study of hospital survivors of sepsis using time-varying Cox proportional hazard models, which produce a baseline hazard ratio (HR) and a second number (δHR) that reflects the amount by which the baseline HR changes with time.

RESULTS:

Of the 32,244 patients who survived sepsis at hospital discharge, 13,565 patients (42%) died (mean ± standard deviation: 1.41 ± 1.87 years) after discharge from the index hospitalization, while 18,679 patients were still alive at follow-up (4.98 ± 2.86 years). The mortality rate decreased with time after discharge: approximately 8.7% of patients died during the first month after discharge, 1.1% of patients died during the 12th month after discharge, and 0.3%% died during the 60th month; after Kaplan-Meier analysis, survival was 91% (95% confidence interval [CI], 91%–92%) at 1 month, 76% (95% CI, 76%–77%) at 1 year, 57% (95% CI, 56%–58%) at 5 years, and 48% (95% CI, 47%–48%) at 10 years after discharge. Organ dysfunction at discharge was associated with worse survival. In particular, elevated urea nitrogen at discharge (HR, 1.10 per 10 mg/dL, 95% CI, 1.08–1.12, P < .001) was associated with increased mortality, but the HR decreased with time from discharge (δHR, 0.98 per 10 mg/dL per year, 95% CI, 0.98–0.99, P < .001). Higher hemoglobin levels were associated with lower mortality (HR, 0.92 per g/dL, 95% CI, 0.91–0.93, P < .001), but this association increased with increasing time after discharge (δHR, 1.02 per g/dL per year, 95% CI, 1.01–1.02, P < .001). Older age was associated with an increased risk of mortality (HR, 1.29 per decade of age, 95% CI, 1.27–1.31, P < .001) that grew with increasing time after discharge (δHR, 1.01 per year of follow-up per decade of age, 95% CI, 1.00–1.02, P < .001). Compared to private insurances Medicaid as primary insurance was associated with an increased risk of mortality (HR, 1.17, 95% CI, 1.10–1.25, P < .001) that did not change with time after discharge. In contrast, Medicare status was initially associated with a similar risk of mortality as private insurance at discharge (HR, 1), but was associated with greater risk as time after discharge increased (δHR, 1.04 per year of follow-up, 95% CI, 1.03–1.05, P < .001).

CONCLUSIONS:

Acute physiologic derangements and organ dysfunction were associated with post-discharge mortality with the associations decreasing over time.

Sepsis is a leading cause of death in the United States and worldwide.¹ Currently, the annual incidence is 1.7 million in the United States and 48.9 million worldwide.² In the United States, hospital mortality from sepsis ranges from 18% to 40% and increases with sepsis severity.^3–5 While mortality from sepsis has decreased in recent years, the incidence of sepsis is increasing, producing more hospital survivors.^5,6 Even after hospital discharge, mortality remains elevated (~15%) in the first year, then decreases to 5% to 6% in succeeding years.^7–11

Most studies evaluating factors associated with mortality in sepsis have focused on short-term outcomes––up to 6 months after onset of sepsis.^3–6 Fewer studies have examined risk factors associated with longer-term mortality. Comorbidities such as cancer, liver disease, lung disease, renal disease, and substance abuse may both predispose to sepsis and contribute to hospital and possibly postdischarge adverse outcomes. Socioeconomic factors such as income and insurance status have been associated with both hospitalization for and hospital mortality due to sepsis.^7,8 Patients with sepsis have higher rates of comorbidities and substance abuse than nonseptic patients, which may contribute to postdischarge mortality.^9–11 Finally, abnormalities in laboratory values are associated with postdischarge mortality in medical patients,^12,13 and may also correlate with postdischarge mortality in septic patients.

The purposes of this study were (1) to estimate postdischarge survival in hospital survivors of sepsis and (2) to identify potential associations between patient characteristics on hospital discharge and mortality. We hypothesized that easily identifiable patient characteristics on or before hospital discharge would be associated with increased postdischarge mortality.

METHODS

This study was deemed exempt by the Institutional Review Board of the School of Medicine “IRBMED” at the University of Michigan (approval No. HUM00184978, dated August 18, 2020), and the requirements for written informed consent were waived. This article was prepared in accordance with the standards set forth by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Using the institutional sepsis database, we created a separate database consisting of all adult Michigan resident patients (≥18 years) who had sepsis between October 7, 2009, and April 9, 2019, and were treated at the University of Michigan Medical Center. For patients with more than 1 sepsis diagnosis during this time, only the first episode was included. Sepsis was defined using Sepsis-3 criteria. Sepsis was suspected when blood cultures were obtained and antibiotics administered. If antibiotics were administered first, then cultures needed to be obtained within 24 hours. If cultures preceded antibiotic administration, then antibiotics needed to be given within the subsequent 72 hours to qualify. Onset of suspected sepsis was the earlier antibiotic administration or drawing of blood cultures. To confirm sepsis, the patient then needed to have an increase ≥2 Sequential Organ Failure Assessment (SOFA) points within the 48 hours before to 24 hours after sepsis onset.¹ Variable choice for data extraction was informed by expert opinion based on clinical relevance and previous studies. Elixhauser comorbidities, vital signs, laboratory values, cultures, demographics, and processes of care (time to initial antibiotic, blood transfusions, mechanical ventilation, and kidney replacement therapy) were extracted from the medical record using DataDirect (University of Michigan). To estimate patients’ adjusted gross income, we used the Internal Revenue Service (IRS) database using the patients’ recorded ZIP codes.¹⁴ We determined hospital survival of the index hospitalization from hospital records and postdischarge mortality from both hospital records and the Michigan Death Index. As treatment of sepsis may have varied over the 10 years of the study, the date of sepsis occurrence was divided into 3 discrete eras defined by the version of Surviving Sepsis Guidelines used at the time of sepsis: July 2009 to February 2013 used the 2008 version, March 2013 to February 2017 the 2012 version, and March 2017 to September 2019, the 2016 version.^15–17

Missing Data

Data were complete for demographics, comorbidities, processes of care, and cultures. For some patients, race or insurance status was listed as “unknown.” Similarly, a few culture results were labeled “unknown” site. For these categorical variables, “unknown” was treated as an extra level in the categorical variable. For laboratory data that were not obtained, we used the SPSS (IBM) multiple imputation procedure for a fully conditional specification, an iterative Markov chain Monte Carlo method that used all variables in Supplemental Digital Content 1 and 2, Supplemental Tables 1, http://links.lww.com/AA/E551 and 2, http://links.lww.com/AA/E552, creating 5 imputed datasets.¹⁸

Transformation

Laboratory variables were assessed for their univariate association with postdischarge mortality. Variables where both higher and lower values were associated with increased risk of death, for example, potassium, sodium, and platelet count, were transformed by taking the square root of the absolute value of the difference between the measured value and the midpoint of the reference range, for example, for sodium, |Na-140|^.5 This produces a U-shaped relationship with postdischarge mortality, because Na values both less than and greater than 140 will have the same transformed value. Other continuous variables where there were very few patients with values on 1 side of the reference range (eg, hemoglobin levels >17 g/dL) were analyzed without transformation.

Statistics

Univariate comparisons were made with standardized differences. Survival was estimated using Kaplan-Meier plots. We used Cox proportional hazard models where the factors’ coefficients vary with time from hospital discharge, that is, the strength of the association of that factor with mortality varies with time from discharge. Laboratory variables were the exposure of interest in the models, and we adjusted for potential confounding variables including comorbidities, processes of care, socioeconomic and demographic factors, and culture results. We first tested the proportional hazard assumption that the HR associated with each factor did not vary with time using Schoenfeld residuals using graphical interpretation and that the slope of the plot was not horizontal, that is, it was either >0 or <0 with 95% confidence.

As the assumption was violated for many factors, we created interaction terms between each factor and linear time, h_x(t) = h₀(t)·exp[β_x + γ_xt], where x is the factor, h is the hazard ratio (HR), β_x is the coefficient for factor X, γ_x is the coefficient of the time-factor X interaction, and t is time. If the 95% confidence interval (CI) for the interaction term included 0, then we concluded there was not sufficient evidence that the HR varied over time. Time was discretized into 120 months. Forward selection with P = .001 for entry and P = .005 for removal was used.

Our Cox proportional hazard model contains a baseline HR, which is the HR at time 0, and an amount by which the HR changes with time (δHR). At any time after discharge, the actual HR is the baseline HR multiplied by δHR raised to the follow-up time in years (HR × δHR^t). For example, with a baseline HR of 1.29 (per decade of age) and a time-varying coefficient =1.01 per year of follow-up, the HR for a patient 1 decade older than another patient at hospital discharge = 1.29 × (1.01)⁰ = 1.29. At 1 year after discharge, the HR is 1.29 × (1.01)¹ = 1.30. At 5 years after discharge, the HR is 1.29 × (1.01)⁵ = 1.36. Therefore, at time of discharge from hospital, the patient 10 years older would have a 29% higher hazard or risk of dying. At 1 and 5 years after discharge, the hazard or risk of dying would be 30% and 36% higher, respectively, compared to the younger patient.

As we used multiple imputation to impute missing data, we performed the regressions separately on each of the 5 imputed datasets, and then combined the resulting coefficients using Rubin’s Rules to estimate the final HRs and 95% CIs.¹⁹ The final variance incorporates both the within and the between imputation variability. All factors in Supplemental Digital Content 1 and 2, Supplemental Tables 1, http://links.lww.com/AA/E551, and 2, http://links.lww.com/AA/E552, were initially considered for model entry. Final models required a factor to have either its baseline HR or its δHR, or both, be significant at the P < .001 level. We present only the baseline HR and δHR that are statistically significant. For clinical significance, we chose δHR >1.04 or <0.96, as this produces a cumulative HR >1.5 or <0.67 over the 10 years of follow-up. Some factors, for example, creatinine, were present both as solo factors and as part of composite variables (SOFA score and Acute Physiology and Chronic Health Evaluation II [APACHE II]). To avoid collinearity effects, variables that were part of composite variables (SOFA and APACHE II) were entered into the model only as solo factors. The SOFA and APACHE II scores were not used. All statistics were done in SPSS 27.0.

Sample Size Justification

Computer modeling has shown that Events Per Variable between 10 and 20 produce an acceptable bias of the coefficient values at the P = .05 significance level.^20,21 In this study, given >30,000 patients with >12,000 deaths, we chose a much higher Events Per Variable (200) in conjunction with a lower P value threshold of 0.001. This would allow 60 factors in the final model.

RESULTS

Of the 32,244 patients who had sepsis and survived to hospital discharge, 13,565 patients (42%) died (mean ± standard deviation, 1.41 ± 1.87 years) after discharge from the index hospitalization, while 18,679 patients were still alive at follow-up (4.98 ± 2.86 years). Mortality was initially high but decreased rapidly thereafter: 8.7% of patients died within 1 month of discharge; by month 12 postdischarge, the monthly mortality declined to 1.1%; by 5 years postdischarge, the monthly mortality fell to ~0.3%. By Kaplan-Meier analysis, survival was 91% (95% CI, 91%–92%) at 1 month, 76% (95% CI, 76%–77%) at 1 year, 57% (95% CI, 56%–58%) at 5 years, and 48% (95% CI, 47%–48%) at 10 years after discharge (Figure 1). Details of the patients are presented in Supplemental Digital Content 1 and 2, Supplemental Tables 1, http://links.lww.com/AA/E551, and 2, http://links.lww.com/AA/E552.

Figure 1. — Kaplan-Meier plot of survival after hospital discharge from sepsis. Vertical hash lines represent the time each patient was censored (time to follow-up).

Using a Cox proportional hazard model with time-varying coefficients to assess which factors had a temporal association with regard to mortality, we found that 29 of the 46 factors associated with mortality varied with time as estimated by P < .001 of the time interaction term (Table). Organ dysfunction as characterized by abnormal laboratory values, whether measured on admission, at sepsis onset, or at discharge, was associated with worse survival. In particular, transformed glucose levels (HR, 1.26, 95% CI, 1.20–1.33, P < .001) and renal dysfunction as measured by urea nitrogen at discharge (HR, 1.10 per 10 mg/dL, 95% CI, 1.09–1.12, P < .001) were associated with increased risk of postdischarge mortality. However, the risk associated with urea nitrogen decreased with time after discharge (δHR, 0.98 per mg/dL per year, 95% CI, 0.98–0.99, P < .001). Abnormalities in serum bicarbonate, potassium, and sodium levels were all associated with higher risk of mortality, with the risk associated with sodium decreasing with time after discharge.

Table.

Cox Time Regression With Time-Varying Coefficient

	Baseline hazard ratio				Amount by which hazard rato changes
Factor	HR	L95% CI	U95% CI	P	δHR/y	L95% CI	U95% CI	P
Age (10 y)	1.29	1.27	1.31	<.001	1.01	1.00	1.02	<.001
Race––compared to White
Asian-American	0.62	0.54	0.72	<.001
Insurance––compared to private
Medicaid	1.17	1.10	1.25	<.001
Medicare					1.04	1.03	1.05	<.001
Female					0.96	0.95	0.98	<.001
Marital status––compared to married
Unknown					1.11	1.08	1.14	<.001
Unmarried					1.04	1.02	1.05	<.001
Weight (10 kg)	0.98	0.97	0.99	<.001
Comorbidities
Cardiac arrhythmias	1.15	1.09	1.21	<.001	0.98	0.96	0.99	<.001
Chronic pulmonary disease	1.15	1.10	1.21	<.001	1.03	1.01	1.04	<.001
Congestive heart failure	1.21	1.16	1.26	<.001
Depression	1.12	1.08	1.16	<.001
Fluid electrolyte disorders	1.13	1.08	1.17	<.001
Hypertension complicated	1.15	1.09	1.21	<.001
Lymphoma					1.04	1.02	1.06	<.001
Metastatic cancer	1.11	1.07	1.16	<.001
Obesity	0.73	0.68	0.77	<.001	1.06	1.05	1.08	<.001
Paralysis	1.19	1.11	1.27	<.001
Peptic ulcer disease excluding bleeding	0.79	0.72	0.86	<.001	1.06	1.03	1.08	<.001
Peripheral vascular disorders	0.78	0.74	0.83	<.001	1.09	1.07	1.11	<.001
Renal failure	0.64	0.60	0.68	<.001	1.13	1.11	1.15	<.001
Rheumatoid arthritis collagen vascular diseases	0.72	0.67	0.77	<.001	1.06	1.04	1.09	<.001
Solid tumor without metastasis	1.58	1.50	1.67	<.001	0.96	0.94	0.97	<.001
Valvular disease	0.76	0.72	0.81	<.001	1.07	1.05	1.09	<.001
Weight loss	1.19	1.13	1.25	<.001	1.04	1.02	1.06	<.001
Admit laboratory values
Bicarbonate (1 meq/L)^a	1.04	1.02	1.07	<.001
Platelet count (10,000/µL)^a	1.23	1.17	1.29	<.001
Potassium (1 meq/L)^a					1.11	1.05	1.17	<.001
Presepsis processes of care
Operation	0.77	0.73	0.82	<.001	1.04	1.02	1.05	<.001
Mechanical ventilation	0.83	0.78	0.88	<.001
Sepsis laboratory values
Chloride (10 meq/L)^a	1.10	1.08	1.13	<.001	0.87	0.81	0.95	<.001
Red cell distribution width (1%)^a	1.09	1.07	1.10	<.001	0.99	0.98	0.99	<.001
White cell count (10,000/µL)^a					1.11	1.05	1.17	<.001
Surviving sepsis guidelines––compared to 2008 version
2012 version	1.12	1.06	1.17	<.001	0.99	0.99	0.99	<.001
2016 version					0.99	0.99	0.99	<.001
Culture results
Positive blood culture—after sepsis	1.26	1.19	1.34	<.001
Positive respiratory culture—after sepsis	1.15	1.07	1.25	<.001
Blood culture sent––post sepsis	1.29	1.23	1.35	<.001	0.95	0.93	0.96	<.001
Discharge laboratory values
Bicarbonate (1 meq/L)^a	1.08	1.04	1.12	<.001
Glucose (10 mg/dL)^a	1.26	1.20	1.33	<.001
Hemoglobin (1 g/dL)	0.92	0.91	0.93	<.001	1.02	1.01	1.02	<.001
Platelet count (100,000/µL)^a	0.88	0.87	0.90	<.001	1.01	1.01	1.02	<.001
Red cell distribution width (1%)^a	1.02	1.01	1.03	<.001
Sodium (1 meq/L)^a	1.13	1.10	1.17	<.001	0.97	0.96	0.98	<.001
Total bilirubin (10 mg/dL)	1.28	1.21	1.36	<.001	0.92	0.88	0.96	<.001
Urea nitrogen (10 mg/dL)	1.10	1.09	1.12	<.001	0.98	0.98	0.99	<.001

Open in a new tab

Hazard ratio at any time after discharge is computed as HR × δHR^t, where t is time (years) after discharge. For computation, blanks for HR or δHR should be treated as equal to 1.

Abbreviations: HR, hazard ratio, δHR, amount that the hazard ratio changes annually; L95% CI, lower 95% confidence interval; U95% CI, upper 95% confidence interval.

Laboratory values that have been transformed as the square root of the absolute value of the actual value minus the midpoint of the reference range: |bicarbonate––25.5|^.5, |chloride-103|^.5, |creatinine––1|^.5, |glucose––125|^.5, |mean platelet value––10.6|^.5, |platelet––275,000|^.5, |potassium––4.1|^.5, |platelet count––275,000|^.5, |sodium-140|², and |white cell count––7000|^.5. The HR, for example, for Na, 1.13, is for a 1-point change in the transformed value. So, a Na, 139 or 141 would both have an increased HR, 1.13 compared to Na, 140. Na, 130 or Na, 150 would have |130–140|^.5 = 10^.5 = 3.16 > HR, 1.13^3.16 = 1.47 compared to Na, 140.

Older age was associated with an increased hazard (HR, 1.29 per decade of age, 95% CI, 1.27–1.31, P < .001), which increased with time after discharge (δHR, 1.01 per year after discharge per decade of age, 95% CI, 1.00–1.02, P = .003) (Table and Figure 2). Having Medicaid as primary insurance was associated with an increased risk compared to private insurance (HR, 1.17, 95% CI, 1.10–1.25, P < .001), which did not vary with time from discharge. In contrast, the association between Medicare status and mortality, which was initially similar to that between private insurance and mortality, progressively increased with time (δHR, 1.04 per year of follow-up, 95% CI, 1.03–1.05, P < .001) and became equal to the risk associated with Medicaid after 4 years (Figure 3). The mortality risk for female patients was the same as for male patients immediately after discharge, but declined with increased time after discharge (δHR, 0.96 per year, 95% CI, 0.95–0.98, P < .001). After 5 years, the risk associated with being female was only 82% of the risk experienced by males.

Figure 2. — Plots showing time-varying hazard ratios for selected Elixhauser comorbidities that are associated with postdischarge mortality compared to not having that comorbidity. A, Cardiac arrhythmias. B, Obesity and weight loss. C, Chronic pulmonary disease. D, Peptic ulcer disease. E, Lymphoma. F, Peripheral vascular disease. G, Solid tumor without metastasis. H, Renal failure.

Figure 3. — Plots showing time-varying hazard ratios for selected Elixhauser comorbidities that are associated with postdischarge mortality. A, Surviving Sepsis Guidelines 2013 and 2017 versions compared to 2009 version. B, Medicare and Medicaid insurance compared to private insurance. C, Ten-year higher age. D, Change in chloride value at sepsis onset per 10 transformed units |value––103 meq/L|⁵. E, White cell count at sepsis onset per 10,000 transformed units |value––7000/µL|⁵. F, Hemoglobin at discharge per g/dL. G, Urea nitrogen at discharge per 10 mg/dL. H, Sodium at discharge per 1 transformed unit |value––140|⁵.

We found that most comorbidities, such as chronic pulmonary disease, complicated hypertension, and metastatic cancer were associated with an increased risk of mortality regardless of time after discharge. Other comorbidities such as congestive heart failure and solid tumor without metastasis were associated with an increased risk of mortality at discharge, but the risk gradually decreased with time after discharge. Several comorbidities such as obesity, peripheral vascular disorders, and renal failure were associated with a lower risk of mortality (adjusted HR <1) that then increased with time after discharge. We also found that positive blood and respiratory cultures after the initial episode of sepsis were associated with increased hazard of death, which remained constant with time after discharge (Table).

DISCUSSION

We found in a retrospective cohort of patients discharged alive from a hospital after admission for sepsis that mortality was 8.7% in the first month after discharge and 43% at 5 years. We also observed differences in the effect of patient comorbidities on mortality after a discharge for sepsis. Specifically, some comorbidities, such as chronic pulmonary disease and weight loss, were associated with a mortality that increased with time from discharge. Others, such as solid tumors without metastasis, were associated with an elevated mortality risk at discharge that then decreased with time from discharge. Finally, some, such as congestive heart failure, depression, and complicated hypertension, were associated with an elevated risk of death that did not change with time from discharge.

While we found many factors whose HR varies with time, some of these δHR are small and clinically insignificant, such as δHR, 1.01 per year of follow-up per 100,000 platelets/µL. At 10 years after discharge, this is only a 10% higher hazard of dying. Conversely, each 1 meq/L difference in transformed potassium at admission has δHR, 1.11 per year of follow-up or 180% higher hazard of dying 10 years after discharge.

Our findings differ somewhat from previous literature. Our mortality rate is considerably lower than that observed in a study of Veterans Administration hospitals from 1983 to 1986, which found a ~60% 1-year and ~79% 5-year mortality.⁹ Our 5-year mortality rate (43%) is lower than the 66% observed in a recent (2021) study. However, this study excluded patients who died in the first year after diagnosis.²² Likewise, a 2013 Scottish study found a 31% 5-year mortality in patients with severe sepsis who survived to hospital discharge, but >75% of patients in that study had no comorbidities.²³ In addition, our study differs from previously published observations in using the new Sepsis-3 definition, whereas the others used Sepsis-2 criteria or were based on administrative (International Classification of Diseases) data.

Our observation that lower hemoglobin levels in patients at discharge are associated with increased postdischarge mortality is consistent with previous studies of both sepsis and other patient populations,^24–26 but differs from 2 other studies of postoperative patients, which found no association with mortality.^27,28 These differing results may be due to different reasons for hospitalization (sepsis versus postsurgery) or severity of illness.^27,28 Our study is also similar to two 2019 studies that found an association between discharge dysglycemia and increased 30- or 180-day mortality, but we show that this association persists for 10 years after discharge.^29,30 The association we observed between increased postdischarge mortality and electrolyte abnormalities on discharge is likewise similar to findings in other patient populations. Our data add to these observations by finding similar relationships in patients with sepsis.^12,13

We found that mortality varied by insurance status. Compared to private insurance, Medicaid was associated with an elevated HR. Medicaid, in Michigan, provides more complete coverage for its beneficiaries with much smaller or even no copays and deductibles than many private insurances.³¹ This increased coverage may make postdischarge care more affordable. However, lost and lapses in Medicaid coverage are common. The typical Medicaid beneficiary maintains coverage for <10 months per year.³² A quarter of beneficiaries change coverage within a year, and more than half of those have a gap in health care coverage,³³ leading to delays in care and in taking prescribed medications.³⁴ This leads to more emergency department visits and hospitalizations.³⁵

Many comorbidities were associated with mortality. One outlier was obesity, which was associated with a lower mortality early after hospital discharge, which increased with increasing time from discharge. Such paradoxically improved outcomes in obese patients have been previously described in septic patients and in those undergoing cardiac surgery.^36,37 After hospital discharge, obesity has been associated with both improved and worsened 1-year survival.³⁸ In our study, associations between comorbidities such as solid tumor without metastasis decreased with time, whereas for others, such as chronic pulmonary disease, diabetes, congestive heart failure, and peripheral vascular disease, the deleterious associations increased with time. Future studies are needed to evaluate which type of comorbidity during and after a sepsis hospitalization improves postdischarge outcomes.

There are several limitations to this study. First, this is a single-center study of a large academic and community hospital; as such, it may not be generalizable. Second, our dataset may not be complete as patients who died after leaving the state were not recorded by the Michigan Death Index. Our mortality rate was generally consistent with other papers, however, suggesting that this effect was likely small. Third, it is possible that unmeasured confounders may have biased the study in unknown ways. Fourth, we only tested 1 time interaction––linear time. Use of other time models or inclusion of multiple time models might produce different time interactions. Limitations in computer power limited our ability to assess different time models. Simple time models such as linear time or log time are preferred and usually work well.³⁹ Finally, because we do not have cause of death, our associations are with all-cause mortality and not sepsis-specific mortality.

There are several novelties and strengths of this study. First, we used Sepsis-3 criteria to define our population. Previous studies using Sepsis-2 or administrative codes produce different patient populations. Existing evidence suggests incomplete overlap between Sepsis-2 and Sepsis-3 criteria as 53% of patients who meet Sepsis-2 criteria also meet Sepsis-3 criteria, and 43% of Sepsis-3 patients also meet Sepsis-2 criteria.⁴⁰ Second, we used a large, curated dataset that allowed us to control for a large number of factors. Finally, we used Cox proportional hazard models with time-varying coefficients to estimate if and how factors’ associations with mortality change with time.

In conclusion, we found in a retrospective dataset analysis that postdischarge mortality after an intensive care unit (ICU) admission for sepsis was associated with several patient comorbidities and abnormalities in laboratory values measured at discharge. However, associations between comorbidities and mortality generally grew stronger with time after discharge, whereas associations between abnormal laboratory values and mortality grew weaker with time after discharge. Studies are needed to ascertain if improved treatment of comorbidities and laboratory derangements will improve postdischarge mortality after sepsis.

Supplementary Material

SDC 1

NIHMS1940946-supplement-SDC_1.docx^{(28.7KB, docx)}

SDC 2

NIHMS1940946-supplement-SDC_2.docx^{(23.7KB, docx)}

KEY POINTS.

Question: What are the factors associated with postdischarge mortality in hospital survivors of sepsis, and how do these associations with mortality vary with time from hospital discharge?
Findings: Abnormal laboratory values were associated with mortality. Several comorbidities had a hazard ratio that increased with time from discharge.
Meaning: Our results may inform prospective studies designed to evaluate the correction of abnormal laboratory values on postdischarge mortality.

Funding:

This study was conducted at the University of Michigan and was supported by departmental and institutional resources. R. Freundlich is supported by K23HL148640.

GLOSSARY

APACHE II: Acute Physiology and Chronic Health Evaluation II
CI: confidence interval
HR: hazard ratio
δHR: amount that the hazard ratio changes annually
ICU: intensive care unit
IRS: Internal Revenue Service
L95% CI: lower 95% confidence interval
SOFA: Sequential Organ Failure Assessment
STROBE: Strengthening the Reporting of Observational Studies in Epidemiology
U95% CI: upper 95% confidence interval

Footnotes

DISCLOSURES

Name: Milo Engoren, MD, FCCM.

Contribution: This author helped in the conception and design of the work, acquired and analyzed the data, helped interpret the data, drafted and helped revise the article, and had final approval of this version.

Name: Michael D. Maile, MD.

Contribution: This author helped in the conception and design of the work, helped interpret the data, helped revise the article, and had final approval of this version.

Name: Troy Seelhammer, MD.

Contribution: This author helped in the conception and design of the work, helped interpret the data, helped revise the article, and had final approval of this version.

Name: Robert E. Freundlich, MD, FCCM.

Contribution: This author helped in the conception and design of the work, helped interpret the data, helped revise the article, and had final approval of this version.

Name: Thomas A. Schwann, MD.

Contribution: This author helped in the conception and design of the work, helped interpret the data, helped revise the article, and had final approval of this version.

This manuscript was handled by: Avery Tung, MD, FCCM.

The authors declare no conflicts of interest.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (www.anesthesia-analgesia.org).

Reprints will not be available from the authors.

REFERENCES

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. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.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. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Ranieri VM, Thompson BT, Barie PS, et al. ; PROWESS-SHOCK Study Group. Drotrecogin alfa (activated) in adults with septic shock. N Engl J Med 2012;366:2055–2064. [DOI] [PubMed] [Google Scholar]
4.Levy MM, Artigas A, Phillips GS, et al. Outcomes of the Surviving Sepsis Campaign in intensive care units in the USA and Europe: a prospective cohort study. Lancet Infect Dis 2012;12:919–924. [DOI] [PubMed] [Google Scholar]
5.Opal SM, Laterre PF, Francois B, et al. ; ACCESS Study Group. Effect of eritoran, an antagonist of MD2-TLR4, on mortality in patients with severe sepsis: the ACCESS randomized trial. JAMA 2013;309:1154–1162. [DOI] [PubMed] [Google Scholar]
6.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. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.O’Brien JM Jr, Lu B, Ali NA, Levine DA, Aberegg SK, Lemeshow S. Insurance type and sepsis-associated hospitalizations and sepsis-associated mortality among US adults: a retrospective cohort study. Crit Care 2011;15:R130. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Vazquez Guillamet MC, Dodda S, Liu L, Kollef MH, Micek ST. Race does not impact sepsis outcomes when considering socioeconomic factors in multilevel modeling. Crit Care Med 2022;50:410–417. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Quartin AA, Schein RM, Kett DH, Peduzzi PN. Magnitude and duration of the effect of sepsis on survival. Department of Veterans Affairs Systemic Sepsis Cooperative Studies Group. JAMA 1997;277:1058–1063. [PubMed] [Google Scholar]
10.Shankar-Hari M, Ambler M, Mahalingasivam V, Jones A, Rowan K, Rubenfeld GD. Evidence for a causal link between sepsis and long-term mortality: a systematic review of epidemiologic studies. Crit Care 2016;20:101. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Garland A, Olafson K, Ramsey CD, Yogendran M, Fransoo R. Distinct determinants of long-term and short-term survival in critical illness. Intensive Care Med 2014;40:1097–1105. [DOI] [PubMed] [Google Scholar]
12.Klausen HH, Petersen J, Bandholm T, et al. Association between routine laboratory tests and long-term mortality among acutely admitted older medical patients: a cohort study. BMC Geriatr 2017;17:62. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Faisal M, Howes R, Steyerberg EW, Richardson D, Mohammed MA. Using routine blood test results to predict the risk of death for emergency medical admissions to hospital: an external model validation study. QJM 2017;110:27–31. [DOI] [PubMed] [Google Scholar]
14.Internal Revenue Service. SOI Tax Stats––Individual Income Tax Statistics––ZIP Code Data (SOI) Accessed October 11, 2021. https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-statistics-zip-code-data-soi.
15.Dellinger RP, Levy MM, Carlet JM, et al. ; International Surviving Sepsis Campaign Guidelines Committee. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock: 2008. Crit Care Med 2008;36:296–327. [DOI] [PubMed] [Google Scholar]
16.Dellinger RP, Levy MM, Rhodes A, 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. Crit Care Med 2013;41: 580–637. [DOI] [PubMed] [Google Scholar]
17.Rhodes A, Evans LE, Alhazzani W, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock: 2016. Crit Care Med 2017;45:486–552. [DOI] [PubMed] [Google Scholar]
18.IBM. topic=imputation-overview-multiple-command Accessed June 20, 2023. https://www.ibm.com/docs/no/spss-statistics/26.0.0
19.Marshall A, Altman DG, Holder RL, Royston P. Combining estimates of interest in prognostic modelling studies after multiple imputation: current practice and guidelines. BMC Med Res Methodol 2009;9:57. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Peduzzi P, Concato J, Feinstein AR, Holford TR. Importance of events per independent variable in proportional hazards regression analysis. II. Accuracy and precision of regression estimates. J Clin Epidemiol 1995;48:1503–1510. [DOI] [PubMed] [Google Scholar]
21.Ogundimu EO, Altman DG, Collins GS. Adequate sample size for developing prediction models is not simply related to events per variable. J Clin Epidemiol 2016;76:175–182. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Song IA, Park HY, Oh TK. Sleep disorder and long-term mortality among sepsis survivors: a nationwide cohort study in South Korea. Nat Sci Sleep 2021;13:979–988. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Cuthbertson BH, Elders A, Hall S, et al. ; Scottish Critical Care Trials Group. Mortality and quality of life in the five years after severe sepsis. Crit Care 2013;17:R70. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Denstaedt SJ, Cano J, Wang XQ, Donnelly JP, Seelye S, Prescott HC. Blood count derangements after sepsis and association with post-hospital outcomes. Front Immunol 2023;14:1133351. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Shankar-Hari M, Rubenfeld GD, Ferrando-Vivas P, Harrison DA, Rowan K. Development, validation, and clinical utility assessment of a prognostic score for 1-year unplanned rehospitalization or death of adult sepsis survivors. JAMA Netw Open 2020;3:e2013580. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Foss NB, Kristensen MT, Kehlet H. Anaemia impedes functional mobility after hip fracture surgery. Age Ageing 2008;37:173–178. [DOI] [PubMed] [Google Scholar]
27.Su H, Aharonoff GB, Zuckerman JD, Egol KA, Koval KJ. The relation between discharge hemoglobin and outcome after hip fracture. Am J Orthop (Belle Mead NJ) 2004;33:576–580. [PubMed] [Google Scholar]
28.Kerfeld MJ, Kor DJ, Frank RD, Hanson AC, Passe MA, Warner MA. Hospital discharge hemoglobin values and posthospitalization clinical outcomes in transfused patients undergoing noncardiac surgery. Transfusion 2020;60:2250–2259. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Spanakis EK, Umpierrez GE, Siddiqui T, et al. Association of glucose concentrations at hospital discharge with readmissions and mortality: a nationwide cohort study. J Clin Endocrinol Metab 2019;104:3679–3691. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Sahni N, Simon G, Arora R. Finding the sweet spot: the last blood glucose measured in the hospital and 30-day outcomes-a retrospective study. J Gen Intern Med 2019;34:510–512. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Centers for Medicare and Medicaid Services. Cost sharing out of pocket costs Accessed June 28, 2023. https://www.medicaid.gov/medicaid/cost-sharing/cost-sharing-out-pocket-costs/index.html.
32.Ku L, Steinmetz E, Bysshe T. Continuity of Medicaid coverage in an era of transition Accessed June 28, 2023. http://www.communityplans.net/Portals/0/Policy/Medicaid/GW_ContinuityInAnEraOfTransition_11-01-15.pdf.
33.Sommers BD, Gourevitch R, Maylone B, Blendon RJ, Epstein AM. Insurance churning rates for low income adults under health reform: lower than expected but still harmful for many. Health Aff (Millwood) 2016;35:1816–1824. [DOI] [PubMed] [Google Scholar]
34.Sommers BD, Chen L, Blendon RJ, Orav EJ, Epstein AM. Medicaid work requirements in Arkansas: two-year impacts on coverage, employment, and affordability of care. Health Aff (Millwood) 2020;39:1522–1530. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Hill I, Burroughs E, Adams G. New Hampshire’s experience with Medicaid work requirements Accessed June 28, 2023. https://www.urban.org/sites/default/files/publi-cation/101657/new_hampshires_experience_with_medic-aid_work_requirements_v2_0_7.pdf.
36.Pepper DJ, Sun J, Welsh J, Cui X, Suffredini AF, Eichacker PQ. Increased body mass index and adjusted mortality in ICU patients with sepsis or septic shock: a systematic review and meta-analysis. Crit Care 2016;20:181. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Schwann TA, Ramia P, Engoren MC, et al. Evidence and temporality of the obesity paradox in coronary artery bypass surgery: an analysis of cause-specific mortality. Eur J Cardthorac Surg 2018;54:896–903. [DOI] [PubMed] [Google Scholar]
38.Prescott HC, Chang VW, O’Brien JM, Langa KM, Iwashyna TJ. Obesity and 1-year outcomes in older Americans with severe sepsis. Crit Care Med 2014;42: 1766–1774. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Schemper M Cox analysis of survival data with non-proportional hazard functions. J R Stat Soc Ser D (The Statistician) 1992;41:455–465. [Google Scholar]
40.Engoren M, Seelhammer T, Freundlich RE, Maile MD, Sigakis MJG, Schwann TA. A comparison of Sepsis-2 (systemic inflammatory response syndrome based) to Sepsis-3 (Sequential Organ Failure Assessment based) definitions: a multicenter retrospective study. Crit Care Med 2020;48:1258–1264. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SDC 1

NIHMS1940946-supplement-SDC_1.docx^{(28.7KB, docx)}

SDC 2

NIHMS1940946-supplement-SDC_2.docx^{(23.7KB, docx)}

[R1] 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. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] 2.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. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] 3.Ranieri VM, Thompson BT, Barie PS, et al. ; PROWESS-SHOCK Study Group. Drotrecogin alfa (activated) in adults with septic shock. N Engl J Med 2012;366:2055–2064. [DOI] [PubMed] [Google Scholar]

[R4] 4.Levy MM, Artigas A, Phillips GS, et al. Outcomes of the Surviving Sepsis Campaign in intensive care units in the USA and Europe: a prospective cohort study. Lancet Infect Dis 2012;12:919–924. [DOI] [PubMed] [Google Scholar]

[R5] 5.Opal SM, Laterre PF, Francois B, et al. ; ACCESS Study Group. Effect of eritoran, an antagonist of MD2-TLR4, on mortality in patients with severe sepsis: the ACCESS randomized trial. JAMA 2013;309:1154–1162. [DOI] [PubMed] [Google Scholar]

[R6] 6.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. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.O’Brien JM Jr, Lu B, Ali NA, Levine DA, Aberegg SK, Lemeshow S. Insurance type and sepsis-associated hospitalizations and sepsis-associated mortality among US adults: a retrospective cohort study. Crit Care 2011;15:R130. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Vazquez Guillamet MC, Dodda S, Liu L, Kollef MH, Micek ST. Race does not impact sepsis outcomes when considering socioeconomic factors in multilevel modeling. Crit Care Med 2022;50:410–417. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Quartin AA, Schein RM, Kett DH, Peduzzi PN. Magnitude and duration of the effect of sepsis on survival. Department of Veterans Affairs Systemic Sepsis Cooperative Studies Group. JAMA 1997;277:1058–1063. [PubMed] [Google Scholar]

[R10] 10.Shankar-Hari M, Ambler M, Mahalingasivam V, Jones A, Rowan K, Rubenfeld GD. Evidence for a causal link between sepsis and long-term mortality: a systematic review of epidemiologic studies. Crit Care 2016;20:101. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] 11.Garland A, Olafson K, Ramsey CD, Yogendran M, Fransoo R. Distinct determinants of long-term and short-term survival in critical illness. Intensive Care Med 2014;40:1097–1105. [DOI] [PubMed] [Google Scholar]

[R12] 12.Klausen HH, Petersen J, Bandholm T, et al. Association between routine laboratory tests and long-term mortality among acutely admitted older medical patients: a cohort study. BMC Geriatr 2017;17:62. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Faisal M, Howes R, Steyerberg EW, Richardson D, Mohammed MA. Using routine blood test results to predict the risk of death for emergency medical admissions to hospital: an external model validation study. QJM 2017;110:27–31. [DOI] [PubMed] [Google Scholar]

[R14] 14.Internal Revenue Service. SOI Tax Stats––Individual Income Tax Statistics––ZIP Code Data (SOI) Accessed October 11, 2021. https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-statistics-zip-code-data-soi.

[R15] 15.Dellinger RP, Levy MM, Carlet JM, et al. ; International Surviving Sepsis Campaign Guidelines Committee. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock: 2008. Crit Care Med 2008;36:296–327. [DOI] [PubMed] [Google Scholar]

[R16] 16.Dellinger RP, Levy MM, Rhodes A, 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. Crit Care Med 2013;41: 580–637. [DOI] [PubMed] [Google Scholar]

[R17] 17.Rhodes A, Evans LE, Alhazzani W, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock: 2016. Crit Care Med 2017;45:486–552. [DOI] [PubMed] [Google Scholar]

[R18] 18.IBM. topic=imputation-overview-multiple-command Accessed June 20, 2023. https://www.ibm.com/docs/no/spss-statistics/26.0.0

[R19] 19.Marshall A, Altman DG, Holder RL, Royston P. Combining estimates of interest in prognostic modelling studies after multiple imputation: current practice and guidelines. BMC Med Res Methodol 2009;9:57. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Peduzzi P, Concato J, Feinstein AR, Holford TR. Importance of events per independent variable in proportional hazards regression analysis. II. Accuracy and precision of regression estimates. J Clin Epidemiol 1995;48:1503–1510. [DOI] [PubMed] [Google Scholar]

[R21] 21.Ogundimu EO, Altman DG, Collins GS. Adequate sample size for developing prediction models is not simply related to events per variable. J Clin Epidemiol 2016;76:175–182. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Song IA, Park HY, Oh TK. Sleep disorder and long-term mortality among sepsis survivors: a nationwide cohort study in South Korea. Nat Sci Sleep 2021;13:979–988. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Cuthbertson BH, Elders A, Hall S, et al. ; Scottish Critical Care Trials Group. Mortality and quality of life in the five years after severe sepsis. Crit Care 2013;17:R70. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Denstaedt SJ, Cano J, Wang XQ, Donnelly JP, Seelye S, Prescott HC. Blood count derangements after sepsis and association with post-hospital outcomes. Front Immunol 2023;14:1133351. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Shankar-Hari M, Rubenfeld GD, Ferrando-Vivas P, Harrison DA, Rowan K. Development, validation, and clinical utility assessment of a prognostic score for 1-year unplanned rehospitalization or death of adult sepsis survivors. JAMA Netw Open 2020;3:e2013580. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Foss NB, Kristensen MT, Kehlet H. Anaemia impedes functional mobility after hip fracture surgery. Age Ageing 2008;37:173–178. [DOI] [PubMed] [Google Scholar]

[R27] 27.Su H, Aharonoff GB, Zuckerman JD, Egol KA, Koval KJ. The relation between discharge hemoglobin and outcome after hip fracture. Am J Orthop (Belle Mead NJ) 2004;33:576–580. [PubMed] [Google Scholar]

[R28] 28.Kerfeld MJ, Kor DJ, Frank RD, Hanson AC, Passe MA, Warner MA. Hospital discharge hemoglobin values and posthospitalization clinical outcomes in transfused patients undergoing noncardiac surgery. Transfusion 2020;60:2250–2259. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Spanakis EK, Umpierrez GE, Siddiqui T, et al. Association of glucose concentrations at hospital discharge with readmissions and mortality: a nationwide cohort study. J Clin Endocrinol Metab 2019;104:3679–3691. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Sahni N, Simon G, Arora R. Finding the sweet spot: the last blood glucose measured in the hospital and 30-day outcomes-a retrospective study. J Gen Intern Med 2019;34:510–512. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Centers for Medicare and Medicaid Services. Cost sharing out of pocket costs Accessed June 28, 2023. https://www.medicaid.gov/medicaid/cost-sharing/cost-sharing-out-pocket-costs/index.html.

[R32] 32.Ku L, Steinmetz E, Bysshe T. Continuity of Medicaid coverage in an era of transition Accessed June 28, 2023. http://www.communityplans.net/Portals/0/Policy/Medicaid/GW_ContinuityInAnEraOfTransition_11-01-15.pdf.

[R33] 33.Sommers BD, Gourevitch R, Maylone B, Blendon RJ, Epstein AM. Insurance churning rates for low income adults under health reform: lower than expected but still harmful for many. Health Aff (Millwood) 2016;35:1816–1824. [DOI] [PubMed] [Google Scholar]

[R34] 34.Sommers BD, Chen L, Blendon RJ, Orav EJ, Epstein AM. Medicaid work requirements in Arkansas: two-year impacts on coverage, employment, and affordability of care. Health Aff (Millwood) 2020;39:1522–1530. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] 35.Hill I, Burroughs E, Adams G. New Hampshire’s experience with Medicaid work requirements Accessed June 28, 2023. https://www.urban.org/sites/default/files/publi-cation/101657/new_hampshires_experience_with_medic-aid_work_requirements_v2_0_7.pdf.

[R36] 36.Pepper DJ, Sun J, Welsh J, Cui X, Suffredini AF, Eichacker PQ. Increased body mass index and adjusted mortality in ICU patients with sepsis or septic shock: a systematic review and meta-analysis. Crit Care 2016;20:181. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R37] 37.Schwann TA, Ramia P, Engoren MC, et al. Evidence and temporality of the obesity paradox in coronary artery bypass surgery: an analysis of cause-specific mortality. Eur J Cardthorac Surg 2018;54:896–903. [DOI] [PubMed] [Google Scholar]

[R38] 38.Prescott HC, Chang VW, O’Brien JM, Langa KM, Iwashyna TJ. Obesity and 1-year outcomes in older Americans with severe sepsis. Crit Care Med 2014;42: 1766–1774. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Schemper M Cox analysis of survival data with non-proportional hazard functions. J R Stat Soc Ser D (The Statistician) 1992;41:455–465. [Google Scholar]

[R40] 40.Engoren M, Seelhammer T, Freundlich RE, Maile MD, Sigakis MJG, Schwann TA. A comparison of Sepsis-2 (systemic inflammatory response syndrome based) to Sepsis-3 (Sequential Organ Failure Assessment based) definitions: a multicenter retrospective study. Crit Care Med 2020;48:1258–1264. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Postdischarge Survival After Sepsis: A Cohort Study

Milo Engoren, MD, FCCM

Michael D Maile, MD

Troy Seelhammer, MD

Robert E Freundlich, MD, FCCM

Thomas A Schwann, MD