KEY POINTS
Question: What are the risk factors for major cardiovascular events after sepsis?
Findings: In this population-based cohort study, classic cardiovascular risk factors, comorbid conditions, and characteristics of the sepsis episode were associated with a higher hazard of major cardiovascular events in adult sepsis survivors.
Meaning: This may represent a first step in identifying sepsis survivors who remain at high risk and may benefit from cardiac risk stratification or preventive strategies.
Sepsis is a leading cause of morbidity and mortality, with approximately 50 million cases per year worldwide (1–4). The overall mortality risk in the acute phase remains high despite advances in diagnostic tools and treatment approaches (1,5–8). Patients who survive an episode of sepsis may experience long-term sequelae, for example, recurrent sepsis, hospital readmission, and increased mortality risk (9–19).
Sepsis survivors are also at increased risk of subsequent cardiovascular complications (18,20–23). A recent population-based cohort study completed by our group showed that, compared to survivors of a nonsepsis hospitalization, sepsis survivors faced a higher risk of major cardiovascular events during long-term follow-up (20). However, the risk factors for these cardiovascular complications remain uncertain—and the extent to which subsequent cardiovascular events are due to classic pathways (e.g., baseline risk factors such as hypertension) or sepsis-related complications is unknown (24). Better understanding of associated risk factors could potentially help 1) generate hypothesis for translational research models and 2) identify subgroups of patients at low or high risk of cardiovascular disease (12,18,25–27). This could in turn inform the enrichment of future trials evaluating potential preventative strategies (e.g., statin use after discharge) (28).
We therefore sought to describe risk factors for the occurrence of major cardiovascular events during long-term follow-up in adult patients who survived a first sepsis hospitalization in the province of Ontario.
MATERIALS AND METHODS
Data Sources and Study Population
We created the study cohort using population-based provincial health administrative databases contained at ICES, an independent, nonprofit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyze healthcare and demographic data, without consent, for health system evaluation and improvement. See Table S1 (https://links.lww.com/CCM/H284) for details on databases used. These datasets were linked using unique encoded identifiers. Our study was developed in accordance with the amended Declaration of Helsinki, and this report follows the Strengthening the Reporting of Observational Studies in Epidemiology, The Treatment And Reporting of Missing data in Observational Studies framework, and the Prognosis Research Strategy guidelines (29–31). The use of data in this project was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require review by a Research Ethics Board.
Our cohort included all adults (age 18 yr or older) in the province of Ontario, Canada, without preexisting cardiovascular disease who survived a hospitalization episode for sepsis between April 2008 and January 2017. The study dates were chosen to optimize data completeness, including laboratory information, and to allow a minimum follow-up of 1 year for all patients (to December 2017). Preexisting cardiovascular disease was defined—using International Classification of Diseases and Related Health Problems, Tenth Revision, Canada (ICD-10-CA) codes (Table S2, https://links.lww.com/CCM/H284)—as the presence of any of the following during a 5-year lookback period prior to the index sepsis hospitalization: coronary heart disease, stroke or transient ischemic attack, peripheral vascular disease, and congestive heart failure (32,33). To maximize sensitivity while classifying prior cardiovascular disease, the presence of a single code during the lookback period was sufficient.
Hospitalizations due to sepsis were identified using a previously validated algorithm (34,35). In contrast to Sepsis-3 criteria, the Jolley algorithm (34) classifies patients as having sepsis (and septic shock) by searching through diagnosis coding fields for any of the codes listed in Table S2 (https://links.lww.com/CCM/H284). If patients survived more than one sepsis episode, only the first hospitalization was considered. To include information on laboratory values as potential risk factors for cardiovascular disease, we constructed a subcohort of patients who had at least one troponin measurement available during their sepsis hospitalization. For all patients, the index date was defined as the date of hospital discharge to represent the start of follow-up. Patients were followed from the index date until outcome occurrence up to a maximum of 5 years or end of the study period.
Main Outcomes of Interest and Risk Factors
The composite primary outcome of interest comprised any myocardial infarction, stroke, or cardiovascular death, defined using ICD-10-CA codes (33,36,37). Table S2 (https://links.lww.com/CCM/H284) describes specific coding strategies and databases used to define main variables of interest. Noncardiovascular mortality was considered as a competing risk during long-term follow-up (38).
Based on subject matter knowledge, prespecified potential risk factors included a) baseline demographics and comorbidities (e.g., age, sex, chronic kidney disease); b) preexisting classic cardiovascular risk factors (e.g., hypertension, atrial fibrillation, and diabetes); and c) characteristics of the sepsis hospitalization (e.g., ICU admission, septic shock, renal replacement therapy, elevated troponin). Full details for all prespecified potential risk factors and characteristics that were considered are shown in Tables S2 and S3 (https://links.lww.com/CCM/H284).
Statistical Analysis
Clinical and demographic characteristics were reported using proportions, means and sds, or medians and interquartile ranges (IQRs) as appropriate. Baseline characteristics of patients who experienced a major cardiovascular event during long-term follow-up were compared with patients without such events using standardized mean differences (SMDs).
To identify potential risk factors associated with the occurrence of major cardiovascular events in adult sepsis survivors during long-term follow-up, we fitted a multivariable cause-specific Cox proportional hazards model (38,39). Covariate selection was performed based on subject matter knowledge and a conceptual model to identify known risk factors for the outcome and potential confounders and was not statistically driven. In order to use the entire cohort of sepsis survivors, the main model was fitted using only clinical characteristics (i.e., no laboratory values included since these were not available for all sepsis survivors; see below). Estimates were reported as hazard ratios (HRs) and their associated 95% CIs. For this model, the linearity assumption for continuous covariates was assessed using higher order polynomials, and we used Schoenfeld residuals to assess the proportional hazards assumption.
Secondary and Sensitivity Analysis
The main secondary analysis considered available laboratory values. For this, we conducted the analysis using a subcohort of patients with at least one troponin measurement. We used multiple imputation with chained equations when information on other laboratory measurements (e.g., creatinine) was missing. Specifically, five imputation sets were used, where models used an identity link for all imputed continuous covariates and included a vector of baseline characteristics and the main outcome of interest (40,41). A cause-specific Cox proportional hazards model with covariate selection driven by subject matter knowledge was then fitted to each imputed dataset, and estimates were reported as HRs and their corresponding 95% CI. We defined a high troponin value (i.e., yes vs no) using the 99th percentile for every assay (Table S2, https://links.lww.com/CCM/H284) (42). To assess the robustness of our analyses that used multiple imputation, we also conducted a) a complete case analysis (only including patients with full information on all laboratory values) and b) using the entire cohort of sepsis survivors (i.e., using multiple imputation for all missing laboratory measures including troponin values).
Furthermore, to assess the impact of the imperfect classification for sepsis, we refitted our main analysis restricted to those patients with ICU admission or septic shock where the Jolley algorithm (34) has better sensitivity and specificity. To better assess whether the potential impact of renal replacement therapy on subsequent cardiovascular outcomes was due to baseline kidney disease, we refitted our models while excluding those patients with chronic kidney disease. Furthermore, to fully characterize both the potential impact of misclassification of baseline cardiovascular disease and prescription of secondary prevention strategies after hospital discharge, we refitted our analysis while: a) restricting to patients without preadmission cardiovascular prescriptions and b) incorporating postdischarge prescriptions in the cause-specific Cox model for cardiovascular disease. Prescriptions of interest (both predischarge and postdischarge) included antiplatelets (excluding aspirin that is also sold over the counter), statins, and anticoagulants. To account for potential differences in practice patterns, outcomes, and associations during the 10-year span, we refitted our main analysis restricting to either the first or last 5 years of the study period. To account for the potentially time-varying cardiovascular risk following sepsis, we refitted our main model restricting to the first 3 months of follow-up after hospital discharge. Finally, since in the presence of competing risks, relative changes in the hazard function cannot be extrapolated to changes in the cumulative incidence of the event of interest, we report estimates from a Fine and Gray model (which better depict the potential impact of the associated factors on the cumulative incidence of major cardiovascular events after sepsis) (38,39,43).
A threshold of less than 0.05 was used to denote statistical significance. All analyses were performed at ICES using SAS Enterprise Guide Version 7.1 (Cary, NC).
RESULTS
During the study period, 472,251 adult patients survived a first sepsis hospitalization in the province of Ontario. Out of these, 268,259 did not have preexisting cardiovascular disease and were included in the study (Fig. 1). Table 1 and Table S4 (https://links.lww.com/CCM/H284) summarize the baseline characteristics of sepsis survivors; median age was 72 years (IQR, 58–82 yr) and more than half of patients (57.7%) were women. The most common comorbidities were hypertension (64.5%), chronic obstructive pulmonary disease (32.1%), diabetes mellitus (31.1%), dementia (17.5%), and dyslipidemia (16.4%). The most common causes of infection were urosepsis (36.5%) and pneumonia (32.1%). The median length of hospital stay was 7 days (IQR, 4–15 d).
TABLE 1. -
Characteristics of Adult
Sepsis Survivors in Ontario (2008–2017)
| Baseline Characteristic |
Major Cardiovascular Event |
Absolute Standardized Mean Difference |
| Yes (
n
= 27,888) |
No (
n
= 240,371) |
| Demographic characteristics |
| Age (yr), median (IQR) |
80 (70–86) |
70 (56–82) |
0.59 |
| Female sex, % |
57.2 |
57.7 |
0.01 |
| Income quintile, %a
|
|
|
|
| 1 |
24.8 |
24.1 |
0.02 |
| 2 |
21.3 |
21.3 |
0.00 |
| 3 |
19.4 |
19.2 |
0.01 |
| 4 |
17.7 |
18.0 |
0.01 |
| 5 |
16.1 |
16.8 |
0.02 |
| Missing |
0.6 |
0.5 |
0.01 |
| Long-term care home resident, % |
25.9 |
16.0 |
0.25 |
| Previous hospitalizations, median (IQR) |
1 (0–2) |
1 (0–2) |
0.03 |
| Baseline comorbid conditions |
| Charlson Comorbidity Index score, median (IQR) |
3 (1–5) |
2 (0–4) |
0.43 |
| Hypertension, % |
80.8 |
62.6 |
0.41 |
| Diabetes mellitus, % |
38.6 |
30.3 |
0.18 |
| Dyslipidemia, % |
16.2 |
16.4 |
0.01 |
| Atrial fibrillation, % |
8.0 |
4.2 |
0.16 |
| Venous thromboembolism, % |
2.2 |
2.4 |
0.01 |
| Chronic liver disease, % |
2.3 |
3.8 |
0.08 |
| Chronic kidney disease, % |
8.3 |
8.2 |
0.00 |
| Chronic obstructive pulmonary disease, % |
38.7 |
31.3 |
0.15 |
| Dementia, % |
23.5 |
16.8 |
0.17 |
| Active malignancy, % |
24.2 |
28.4 |
0.10 |
| Sepsis episode characteristics |
| Septic shock, % |
23.9 |
22.9 |
0.02 |
| Source of infection |
|
|
|
| Pneumonia, % |
32.9 |
32.0 |
0.02 |
| Urosepsis, % |
40.1 |
36.1 |
0.08 |
| Acute kidney injury, % |
13.4 |
11.3 |
0.06 |
| High troponinb, % |
4.3 |
3.6 |
0.03 |
| Intensity of organ support |
| ICU admission, % |
15.2 |
17.9 |
0.07 |
| Transfusion, % |
13.6 |
16.4 |
0.08 |
| Dialysis, % |
2.3 |
1.8 |
0.03 |
| Invasive mechanical ventilation, % |
5.1 |
6.9 |
0.07 |
| Noninvasive ventilation, % |
6.8 |
8.9 |
0.08 |
| Tracheostomy, % |
0.9 |
1.4 |
0.05 |
| Length of hospital stay (d), median (IQR) |
8 (4–16) |
7 (4–15) |
0.07 |
IQR = interquartile range.
aIncome quintile is a patient-level characteristic. Income calculation is based on demographic and geographic information. Income quartile available for 266,777.
bHigh troponin defined as troponin > 99th percentile. Available for 26,400 patients.
Major cardiovascular event defined as the composite of myocardial infarction, stroke, or cardiovascular death. Index date defined as date of hospital discharge.
;)
Figure 1.: Flow chart of study patients.
Overall, 10.4% of sepsis survivors subsequently experienced a major cardiovascular event during a median follow-up of 3 years (IQR, 1–5 yr). The cumulative risk of myocardial infarction, stroke, and cardiovascular death was 3.7%, 3.9%, and 3.0%, respectively. These patients tended to be older (80 vs 70 yr; SMD 0.59) and had a higher burden of comorbidities (median Charlson Comorbidity Index [44] score 3 vs 2; SMD: 0.43). Patients who had subsequent cardiovascular events also were more likely to have chronic conditions such as hypertension, diabetes, atrial fibrillation, and higher severity of acute disease (e.g., more acute kidney injury and renal replacement therapy requirement [Table 1 and Table S4, https://links.lww.com/CCM/H284]).
Follow-up information for the overall cohort and those patients who experienced a major cardiovascular event is presented in Table 2. The cumulative incidence of all-cause death at 1 and 5 years for all sepsis survivors was 20.5% and 39.4%, respectively. Patients who presented with a major cardiovascular event also had higher risk of recurrent sepsis and all-cause mortality during long-term follow-up (Table 2). Table S5 (https://links.lww.com/CCM/H284) shows the overall proportion of cardiovascular events by follow-up time after hospital discharge.
TABLE 2. -
Follow-Up Information of
Sepsis Survivors in Ontario (2008–2017)
| Follow-Up |
All Sepsis Survivors (n = 268,259) |
Patients Experiencing a Major Cardiovascular Event (n = 27,888) |
Patients Without a Major Cardiovascular Event (n = 240,371) |
| Follow-up time, yr, median (interquartile range)a
|
3 (1–5) |
2 (1–3) |
3 (1–5) |
| Recurrent sepsis, % (95% CI) |
17.7 (17.5–17.8) |
24.1 (23.6–24.6) |
16.9 (16.8–17.1) |
| All-cause death at 1 yr, % (95% CI) |
20.5 (20.3–20.6) |
27.4 (26.9–27.9) |
19.7 (19.5–19.9) |
| All-cause death at 5 yr, % (95% CI) |
39.4 (39.2–39.6) |
63.9 (63.3–64.5) |
36.6 (36.3–36.7) |
aFollow-up from hospital discharge until major cardiovascular event, end of study period or a maximum of 5 yr. Overall, 10.4% of patients experienced a major cardiovascular event during long term follow-up.
Major cardiovascular event defined as the composite of myocardial infarction, stroke, or cardiovascular death.
Risk Factors for Subsequent Cardiovascular Events
Results from a multivariable regression suggested that age (HR, 1.53 every 10 yr increase; 95% CI, 1.51–1.55), male sex (HR, 1.23; 95% CI, 1.20–1.26), diabetes mellitus (HR, 1.24; 95% CI, 1.21–1.27), hypertension (HR, 1.34; 95% CI, 1.30–1.38), atrial fibrillation (HR, 1.46; 95% CI, 1.40–1.52), and chronic kidney disease (HR, 1.11; 95% CI, 1.06–1.16) were associated with major cardiovascular events during long-term follow-up (full model shown in Table 3). Notably, higher income quintile was associated with a reduced hazard of major cardiovascular events (HR, 0.97 for every 1 quintile increase; 95% CI, 0.96–0.98). Chronic conditions such as obstructive pulmonary disease (HR, 1.21; 95% CI, 1.18–1.24) and previous venous thromboembolic disease (HR, 1.13; 95% CI, 1.05–1.23) were also associated with a higher hazard of subsequent cardiovascular events (Table 3).
TABLE 3. -
Factors Associated With Major Cardiovascular Events During 5 Years of Follow-Up in Adult
Sepsis Survivors in Ontario (2008–2017)
| Factor |
Major Cardiovascular Event |
Competing Noncardiovascular Death |
Comment |
| Crude Hazard Ratio (95% CI)a
|
Adjusted Hazard Ratio (95% CI)b
|
Adjusted Hazard Ratio (95% CI)c
|
| Baseline characteristics and comorbid conditions |
| Age (yr) |
1.59 (1.57–1.60) |
1.53 (1.51–1.54) |
1.32 (1.31–1.32) |
Every 10 yr |
| Male sex |
1.09 (1.06–1.12) |
1.23 (1.20–1.26) |
1.20 (1.18–1.23) |
Yes vs no |
| Income (quintiles) |
0.98 (0.97–0.99) |
0.97 (0.96–0.98) |
0.99 (0.98–0.99) |
Every 1 quintile |
| Chronic kidney disease |
1.02 (0.98–1.07) |
1.11 (1.06–1.16) |
1.07 (1.04–1.09) |
Yes vs no |
| Chronic liver disease |
0.78 (0.72–0.84) |
1.21 (1.12–1.31) |
2.32 (2.25–2.39) |
Yes vs no |
| Chronic pulmonary disease |
1.46 (1.42–1.49) |
1.21 (1.18–1.24) |
1.19 (1.18–1.21) |
Yes vs no |
| Venous thromboembolic disease |
1.14 (1.06–1.24) |
1.13 (1.05–1.23) |
1.47 (1.42–1.53) |
Yes vs no |
| Dementia |
1.94 (1.89–1.99) |
1.04 (1.01–1.08) |
1.76 (1.73–1.79) |
Yes vs no |
| Active malignancy |
1.23 (1.19–1.26) |
1.03 (1.00–1.06) |
2.80 (2.76–2.83) |
Yes vs no |
| Long-term care home resident |
1.95 (1.90–2.01) |
1.20 (1.16–1.24) |
1.09 (1.07–1.11) |
Yes vs no |
| Classic cardiovascular risk factors |
| Diabetes mellitus |
1.46 (1.43–1.50) |
1.24 (1.21–1.27) |
0.99 (0.98–1.00) |
Yes vs no |
| Hypertension |
2.63 (2.55–2.70) |
1.34 (1.30–1.38) |
0.85 (0.83–0.86) |
Yes vs no |
| Atrial fibrillation |
2.11 (2.02–2.20) |
1.46 (1.40–1.52) |
1.01 (0.98–1.04) |
Yes vs no |
| Sepsis episode characteristics and intensity of support |
| Site of infectiond
|
|
|
|
|
| Pneumonia |
1.17 (1.14–1.21) |
1.09 (1.05–1.12) |
1.12 (1.09–1.14) |
Vs other infection |
| Urosepsis |
1.27 (1.23–1.31) |
1.02 (0.99–1.05) |
0.97 (0.96–0.99) |
Vs other infection |
| Septic shock |
1.17 (1.14–1.20) |
1.08 (1.05–1.11) |
1.14 (1.12–1.16) |
Yes vs no |
| ICU admission |
0.78 (0.75–0.80) |
1.05 (1.01–1.10) |
0.81 (0.79–0.83) |
Yes vs no |
| Invasive mechanical ventilation |
0.67 (0.63–0.71) |
0.90 (0.82–1.00) |
0.71 (0.68–0.75) |
Yes vs no |
| Noninvasive ventilation |
0.70 (0.67–0.74) |
0.96 (0.88–1.05) |
1.10 (1.04–1.15) |
Yes vs no |
| Tracheostomy placement |
0.61 (0.54–0.69) |
0.93 (0.81–1.06) |
1.06 (0.98–1.13) |
Yes vs no |
| Transfusions |
0.94 (0.90–0.97) |
1.05 (1.01–1.09) |
1.57 (1.54–1.60) |
Yes vs no |
| Renal replacement therapy |
1.20 (1.11–1.30) |
1.51 (1.38–1.64) |
1.05 (1.00–1.11) |
Yes vs no |
| Length of hospital stay (d) |
1.01 (1.00–1.02) |
1.00 (0.99–1.01) |
1.02 (1.01–1.02) |
Every 10 d |
aBased on a cause-specific Cox proportional hazards model only including the variable of interest.
bBased on a cause-specific Cox proportional hazards model. Full model (i.e., all included covariates) shown.
cBased on a cause-specific Cox proportional hazards model. Full model (i.e., all included covariates) shown.
dSite of infection modeled using the following categories: pneumonia, urosepsis, and other infections.
Major cardiovascular event defined as the composite of myocardial infarction, stroke, or cardiovascular death. Index time defined as date of hospital discharge. Duration of follow-up to 5 yr. Lookback period to define baseline characteristics of 5 yr.
Several characteristics of the sepsis episode such as site of infection (pneumonia vs other sites: HR, 1.09; 95% CI, 1.05–1.12), septic shock (HR, 1.08; 95% CI, 1.05–1.11), ICU admission (HR, 1.05; 95% CI, 1.01–1.10), and the need for renal replacement therapy during the index hospitalization (HR, 1.51; 95% CI, 1.38–1.64) were also associated with a higher hazard of subsequent cardiovascular events (Table 3). Finally, respiratory support either with invasive (HR, 0.90; 95% CI, 0.82–1.00) or noninvasive mechanical ventilation (HR, 0.96; 95% CI, 0.88–1.05) was not associated with a higher hazard of major cardiovascular events in adult sepsis survivors.
Competing Risk of Noncardiovascular Death
The cause-specific model for the competing risk of noncardiovascular death is shown in Table 3. For example, age (HR, 1.32 every 10 yr increase; 95% CI, 1.31–1.32), male sex (HR, 1.20; 95% CI, 1.18–1.23), chronic kidney disease (HR, 1.07; 95% CI, 1.04–1.09), chronic liver disease (HR, 2.32; 95% CI, 2.25–2.39), chronic pulmonary disease (HR, 1.19; 95% CI, 1.18–1.21), and venous thromboembolic disease (HR, 1.47; 95% CI, 1.42–1.53) were associated with noncardiovascular death during long-term follow-up. Characteristics of the sepsis episode such as site of infection (pneumonia vs other sites: HR, 1.12; 95% CI, 1.09–1.14), septic shock (HR, 1.14; 95% CI, 1.12–1.16), and the need for renal replacement therapy during the index hospitalization (HR, 1.05; 95% CI, 1.00–1.11) were also associated with a higher hazard of noncardiovascular death (Table 3).
Secondary and Sensitivity Analysis
Table S6 (https://links.lww.com/CCM/H284) depicts the risk factors associated with cardiovascular disease in sepsis survivors when restricting to the subcohort of patients with at least one troponin measurement and including additional laboratory measures as potential factors. Specifically, after adjustment for baseline comorbidities and classic cardiovascular risk factors, elevated serum creatinine (HR, 1.03 for every 100 μmol/L increase; 95% CI, 1.00–1.07) and a high troponin (HR, 1.23 vs a low troponin; 95% CI, 1.13–1.33) were also associated with an increased hazard of long-term major cardiovascular events (Table S6, https://links.lww.com/CCM/H284). Higher levels of high-density lipoprotein were associated with a lower hazard of major cardiovascular events (HR, 0.87 for every 1 mmol/L increase; 95% CI, 0.78–0.96).
The complete case analysis and the analysis that used multiple imputation in the entire cohort of sepsis survivors yielded similar estimates to the main secondary analysis (Table S6, https://links.lww.com/CCM/H284). The sensitivity analysis restricting to those patients a) with ICU admission or septic shock and b) without baseline kidney disease yielded similar estimates to our main results (Tables S7 and S8, https://links.lww.com/CCM/H284). Sensitivity analyses considering baseline and postdischarge prescription patterns are shown in Tables S9 and S10 (https://links.lww.com/CCM/H284). Notably, the use of statins after discharge was associated with a reduction in the hazard of major cardiovascular events (HR, 0.95; 95% CI, 0.92–0.98; Table S10, https://links.lww.com/CCM/H284). Changes in the subdistribution hazard of major cardiovascular outcomes for all considered risk factors are shown in Table S11 (https://links.lww.com/CCM/H284).
Associations for all considered risk factors while restricting to the first or last 5 years of the study period or the first 3 months of follow-up are shown in Tables S12–S14 (https://links.lww.com/CCM/H284).
DISCUSSION
Our study shows that classic cardiovascular risk factors and preexisting comorbidities (such as age, sex, diabetes mellitus, hypertension, atrial fibrillation, and cholesterol levels), in addition to sepsis-specific characteristics (such as site of infection, septic shock, acute kidney injury requiring renal replacement therapy, and a high troponin value) are associated with a higher hazard of experiencing major cardiovascular events during follow-up after an episode of sepsis. Furthermore, although exploratory, our study shows that specific prescription practices after discharge (e.g., statin use) may modify the cardiovascular risk of sepsis survivors. These findings should be considered hypothesis generating and a first step in 1) developing appropriate ways to enrich future studies looking at cardiovascular outcomes in sepsis survivors and 2) identifying potential therapeutic targets for further evaluation.
Our study builds upon previous research demonstrating the increased risk of subsequent cardiovascular disease in sepsis survivors (e.g., compared with nonsepsis survivors of a hospitalization) (18,20–23,45,46). Our results are consistent with findings from recent cohort studies highlighting that an elevated troponin value, sepsis-specific variables, and presepsis characteristics were associated with subsequent cardiovascular events (24,27,47). Our study provides additional information on preexisting comorbidities, cardiovascular and sepsis-specific characteristics that are likely associated with subsequent cardiovascular disease and assessing the risk over a longer follow-up and using population-based data. Our study also highlights potential therapeutic strategies such as the use of statins after hospital discharge in sepsis survivors. Importantly, our findings in this regard are hypothesis generating and may warrant further evaluation in future observational and randomized studies.
In theory, cardiovascular risk after sepsis may be attributable to a combination of interacting processes, including 1) the antecedent health status and burden of comorbid disease, 2) classic cardiovascular risk factors, 3) the severity of the acute episode and intensity of life-support required during hospitalization, and 4) both short- and long-term changes in the immune system, coagulation cascade, endothelial function, and inflammation induced by sepsis (12,14,18,25,26,48,49). Furthermore, sepsis may contribute to long-term cardiovascular risk through the trigger of new cardiovascular disease or by accelerating chronic disease trajectories (13,18,24,50). The latter may represent a biological mechanism or a byproduct of reduced health seeking behaviors and attention to the control of classic risk factors after a patient survives sepsis. Our findings that classic cardiovascular risk factors, prevalent chronic conditions, and characteristics of the septic episode may all be associated with subsequent major cardiovascular events help to improve our understanding of the potential underlying mechanisms. Since we only included patients without documented cardiovascular disease at baseline, our results may support the notion that sepsis acts as a potential risk factor for incident cardiovascular disease. However, our study cohort may have included patients with either subclinical atherosclerosis or with established disease that was misclassified (due to the imperfect accuracy of coding algorithms). Hence, whether the impact of sepsis is independent of baseline atherosclerosis, thrombosis, and myocardial injury (that might be expected in patients with established cardiovascular disease) remains to be elucidated (21,23). Figure 2 describes potential pathways that may explain the relationship between sepsis and subsequent cardiovascular events. Importantly, it should be noted that several of the identified factors (e.g., the deployment of renal replacement therapy) may be associated with a higher hazard of long-term cardiovascular outcomes regardless of whether the hospitalization was due to sepsis or not.
Figure 2.: Pathways for cardiovascular outcomes after sepsis. The figure depicts potential pathways for cardiovascular outcomes after sepsis, including: 1) a direct effect (i.e., not related to initial severity and potentially explained by changes in immune, coagulation, or inflammation cascades); 2) mediated effect through organ failures and acute complications (e.g., renal failure); and 3) classic pathway that depends on baseline cardiovascular risk factors and atherosclerosis (that may be enhanced and modified by sepsis).
Additional insight into the mechanistic pathways of long-term outcomes after sepsis may be provided by analyzing the factors associated with major cardiovascular events while considering their association with noncardiovascular death simultaneously. Several factors have qualitatively similar associations (38) with both cardiovascular events and the competing risk of noncardiovascular death (e.g., sex, income, chronic kidney disease, chronic pulmonary disease, long-term care residency, pneumonia, and presence of septic shock). This may imply that these baseline conditions and characteristics of the sepsis episode affect both cardiovascular and noncardiovascular endpoints by similar mechanistic pathways (or are associated with a wide variety of downstream effects). Conversely, several factors have qualitatively different strengths of association with either major cardiovascular events or noncardiovascular death. For example, atrial fibrillation, diabetes, hypertension, ICU admission, and renal replacement therapy are mostly associated with a higher hazard of major cardiovascular events. On the contrary, chronic liver disease, active malignancy, and transfusions are mostly associated with a higher hazard of noncardiovascular death. It is important to note that the qualitatively different association with both outcomes may be due to either the existence of different mechanistic pathways or the depletion of susceptible individuals (e.g., a patient who dies of noncardiovascular causes cannot face a major cardiovascular event).
Our study has several limitations. The competing risk of death (i.e., either noncardiovascular or due to inadequate ascertainment) likely affects the overall reported estimates and the observed cumulative incidence of cardiovascular disease in sepsis survivors. Our main analysis considered this competing risk using accepted methods, but it is possible that our results still underestimate the actual incidence of subsequent cardiovascular disease (38). The subcohort of patients with at least one troponin value had missing information on additional laboratory measures that may not have occurred completely at random. To address this possibility, we used multiple imputation to deal with missingness and conducted several sensitivity analyses that all yielded consistent findings. Furthermore, we did not have data on several known risk factors such as cardiovascular family history and smoking. However, we did include a proxy for severe smoking by identifying patients with chronic obstructive pulmonary disease. The associations presented for the individual characteristics and risk factors are not corrected for multiple comparisons and are subject to selection bias (i.e., collider stratification) and residual confounding. Therefore, they should not be interpreted as causal mechanisms to avoid the so-called “table 2 fallacy” (51). Since the definition of sepsis using administrative data are imperfect, this misclassification may render a cohort that includes patients without sepsis (or just infection but without systemic response); however, our cohort restricted to those patients with severe sepsis or ICU admission yielded similar estimates. Additionally, since we relied on administrative datasets, the absence of established cardiovascular disease or other classic cardiovascular risk factors (e.g., chronic kidney disease) is subject to the imperfect accuracy of the coding algorithms. Therefore, the associations between sepsis-specific characteristics and subsequent cardiovascular outcomes may represent both a true causal pathway and confounding due to unmeasured baseline cardiovascular disease or classic risk factors. The association of age and subsequent cardiovascular events after sepsis could also be explained by undiagnosed age-related atherosclerosis. Finally, we did not have information on brain natriuretic peptide, echocardiographic results, vasopressor or inotropic requirements, presence or absence of bacteremia, the deployment of extracorporeal life support during index hospitalization, aspirin use, self-reported race and ethnicity, obesity, or body mass index, which may also be independent predictors of cardiovascular events in sepsis survivors (or modify the effect of sepsis on such long-term endpoints).
CONCLUSIONS
In conclusion, both classic cardiovascular risk factors and sepsis-specific characteristics are associated with a higher hazard of experiencing subsequent major cardiovascular events after a hospitalization for sepsis. Although current practice should continue to focus on guidelines and clinical judgment to reduce the burden of long-term cardiovascular outcomes in sepsis survivors, knowledge of these associated factors may help to identify patients at higher (or lower) risk of cardiovascular complications. This may aid in deploying enrichment strategies for future studies. Future research should seek to evaluate potential mitigation strategies (e.g., statin therapy) in those patients deemed at highest risk, with the overall aim of reducing long-term cardiovascular outcomes in sepsis survivors.
ACKNOWLEDGMENTS
We thank Kednapa Thavorn and Allan Garland for their comments and review of a previous version of this article. We would also like to acknowledge the support of the Acute & Intensive Care Outcomes Research Network during the conduct of the present research.
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
2. Law AC, Stevens JP, Walkey AJ: National trends in timing of death among patients with septic shock, 1994-2014. Crit Care Med. 2019; 47:1493–1496
3. Fleischmann C, Scherag A, Adhikari NKJ, et al.: Assessment of global incidence and mortality of hospital-treated
sepsis. Current estimates and limitations. Am J Respir Crit Care Med. 2016; 193:259–272
4. 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
5. Seymour CW, Gesten F, Prescott HC, et al.: Time to treatment and mortality during mandated emergency care for
sepsis. N Engl J Med. 2017; 376:2235–2244
6. Fleuren LM, Klausch TLT, Zwager CL, et al.: Machine learning for the prediction of
sepsis: A systematic review and meta-analysis of diagnostic test accuracy. Intensive Care Med. 2020; 46:383–400
7. Angus DC, van der Poll T: Severe
sepsis and septic shock. N Engl J Med. 2013; 369:840–851
8. Levy MM, Evans LE, Rhodes A: The surviving
sepsis campaign bundle: 2018 update. Crit Care Med. 2018; 46:997–1000
9. Shankar-Hari M, Saha R, Wilson J, et al.: Rate and
risk factors for rehospitalisation in
sepsis survivors: Systematic review and meta-analysis. Intensive Care Med. 2020; 46:619–636
10. Prescott HC, Osterholzer JJ, Langa KM, et al.: Late mortality after
sepsis: Propensity matched cohort study. BMJ. 2016; 353:i2375
11. Dalager-Pedersen M, Søgaard M, Schønheyder HC, et al.: Risk for
myocardial infarction and
stroke after community-acquired bacteremia: A 20-year population-based cohort study. Circulation. 2014; 129:1387–1396
12. Prescott HC, Angus DC: Enhancing recovery from
sepsis: A review. JAMA. 2018; 319:62–75
13. Mankowski RT, Yende S, Angus DC: Long-term impact of
sepsis on cardiovascular health. Intensive Care Med. 2019; 45:78–81
14. Yende S, Kellum JA, Talisa VB, et al.: Long-term host immune response trajectories among hospitalized patients with
sepsis. JAMA Netw Open. 2019; 2:e198686
15. Lone NI, Gillies MA, Haddow C, et al.: Five-year mortality and hospital costs associated with surviving intensive care. Am J Respir Crit Care Med. 2016; 194:198–208
16. Shankar-Hari M, Harrison DA, Ferrando-Vivas P, et al.:
Risk Factors at index hospitalization associated with longer-term mortality in adult
sepsis survivors. JAMA Netw Open. 2019; 2:e194900
17. Prescott HC, Langa KM, Liu V, et al.: Increased 1-year healthcare use in survivors of severe
sepsis. Am J Respir Crit Care Med. 2014; 190:62–69
18. Kosyakovsky LB, Angriman F, Katz E, et al.: Association between
sepsis survivorship and long-term cardiovascular outcomes in adults: A systematic review and meta-analysis. Intensive Care Med. 2021; 47:931–942
19. Farrah K, McIntyre L, Doig CJ, et al.:
Sepsis-associated mortality, resource use, and healthcare costs: A propensity-matched cohort study*. Crit Care Med. 2021; 49:215–227
20. Angriman F, Rosella L, Lawler P, et al.:
Sepsis hospitalization and risk of subsequent cardiovascular events in adults: A population-based matched cohort study. Intensive Care Med. 2022; 48:448–457
21. Yende S, Linde-Zwirble W, Mayr F, et al.: Risk of cardiovascular events in survivors of severe
sepsis. Am J Respir Crit Care Med. 2014; 189:1065–1074
22. Corrales-Medina VF, Alvarez KN, Weissfeld LA, et al.: Association between hospitalization for pneumonia and subsequent risk of
cardiovascular disease. JAMA. 2015; 313:264–274
23. Ou S-MM, Chu H, Chao P-WW, et al.: Long-term mortality and major adverse cardiovascular events in
sepsis survivors a nationwide population-based study. Am J Respir Crit Care Med. 2016; 194:209–217
24. Garcia MA, Rucci JM, Thai KK, et al.: Association between troponin I levels during
sepsis and postsepsis cardiovascular complications. Am J Respir Crit Care Med. 2021; 204:557–565
25. Mayr FB, Prescott HC: Identifying survivors of
sepsis at risk for adverse cardiovascular outcomes. Am J Respir Crit Care Med. 2021; 204:500–501
26. Merdji H, Schini-Kerth V, Meziani F, et al.: Long-term cardiovascular complications following
sepsis: Is senescence the missing link?. Ann Intensive Care. 2021; 11:166
27. Walkey AJ, Knox DB, Myers LC, et al.: Prognostic accuracy of presepsis and intrasepsis characteristics for prediction of cardiovascular events after a
sepsis hospitalization. Crit Care Explor. 2022; 4:e0674
28. Shankar-Hari M, Rubenfeld GD: Population enrichment for critical care trials: Phenotypes and differential outcomes. Curr Opin Crit Care. 2019; 25:489–497
29. Riley RD, Hayden JA, Steyerberg EW, et al.: Prognosis research strategy (PROGRESS) 2: Prognostic factor research. PLoS Med. 2013; 10:e1001380
30. von Elm E, Altman DG, Egger M, et al.: The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. Lancet. 2007; 370:1453–1457
31. Lee KJ, Tilling KM, Cornish RP, et al.: Framework for the treatment and reporting of missing data in observational studies: The treatment and reporting of missing data in observational studies framework. J Clin Epidemiol. 2021; 134:79–88
32. Chen G, Farris MS, Cowling T, et al.: Prevalence of atherosclerotic
cardiovascular disease and subsequent major adverse cardiovascular events in Alberta, Canada: A real-world evidence study. Clin Cardiol. 2021; 44:1613–1620
33. Tu J, Chu A, Donovan LR, et al.: The Cardiovascular Health in Ambulatory Care Research Team (CANHEART): Using big data to measure and improve cardiovascular health and healthcare services. Circ Cardiovasc Qual Outcomes. 2015; 8:2014–2012
34. Jolley RJ, Quan H, Jette N, et al.: Validation and optimisation of an ICD-10-coded case definition for
sepsis using administrative health data. BMJ Open. 2015; 5:e0094871–e0094810
35. Jolley RJ, Sawka KJ, Yergens DW, et al.: Validity of administrative data in recording
sepsis: A systematic review. Crit Care. 2015; 19:139
36. Austin PC, Daly PA, Tu JA multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J. 2002; 144:290–296
37. Porter J, Mondor L, Kapral MK, et al.: How reliable are administrative data for capturing
stroke patients and their care. Cerebrovasc Dis Extra. 2016; 6:96–106
38. Austin PC, Lee DS, Fine JP: Introduction to the analysis of survival data in the presence of competing risks. Circulation. 2016; 133:601–609
39. Fine JP, Gray RJ: A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999; 94:496509
40. Madley-Dowd P, Hughes R, Tilling K, et al.: The proportion of missing data should not be used to guide decisions on multiple imputation. J Clin Epidemiol. 2019; 110:63–73
41. He Y: Missing data analysis using multiple imputation: Getting to the heart of the matter. Circ Cardiovasc Qual Outcomes. 2010; 3:98–105
42. Sheyin O, Davies O, Duan W, et al.: The prognostic significance of troponin elevation in patients with
sepsis: A meta-analysis. Heart Lung. 2015; 44:75–81
43. Mccaw ZR, Claggett BL, Tian L, et al.: Practical recommendations on quantifying and interpreting treatment effects in the presence of terminal competing risks: A review. JAMA Cardiol. 2021; 7:450–456
44. Charlson ME, Pompei P, Ales KL, et al.: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987; 40:373–383
45. Musher DM, Abers MS, Corrales-Medina VF: Acute infection and
myocardial infarction. N Engl J Med. 2019; 380:171–176
46. Jafarzadeh SR, Thomas BS, Warren DK, et al.: Longitudinal study of the effects of bacteremia and
sepsis on 5-year risk of cardiovascular events. Clin Infect Dis. 2016; 63:495–500
47. Myers LC, Knox D, Thai KK, et al.: Predicting post-
sepsis cardiovascular events with death as a competing risk. Ann Am Thorac Soc. 2022; 1:14
48. Yende S, D’Angelo G, Mayr F, et al.: Elevated hemostasis markers after pneumonia increases one-year risk of all-cause and cardiovascular deaths. PLoS One. 2011; 6:e22847
49. Alnabelsi TS, Gupta VA, Su LC, et al.: Usefulness of findings by multimodality imaging to stratify risk of major adverse cardiac events after
sepsis at 1 and 12 months. Am J Cardiol. 2020; 125:1732–1737
50. Lai CC, Lee MTG, Lee WC, et al.: Susceptible period for cardiovascular complications in patients recovering from
sepsis. CMAJ. 2018; 190:E1062–E1069
51. Westreich D, Greenland S: The table 2 fallacy: Presenting and interpreting confounder and modifier coefficients. Am J Epidemiol. 2013; 177:292–298
