Annals of the American Thoracic Society

Rationale: Adult sepsis survivors have an increased risk of experiencing long-term cardiovascular events.

Objectives: To determine whether the cardiovascular risk after sepsis is mitigated by renin-angiotensin system inhibitors (RASi).

Methods: We conducted a population-based cohort study of adult sepsis survivors designed to emulate a target randomized trial with an active comparator and new-user design. We excluded patients with a first-line indication for prescription of RASi (e.g., coronary heart disease, heart failure, chronic kidney disease, and hypertension with diabetes mellitus). The main exposure of interest was a new prescription of a RASi within 30 days of hospital discharge. The active comparator was a new prescription of either a calcium channel blocker or a thiazide diuretic, also within 30 days of hospital discharge. The primary outcome of interest was the composite of myocardial infarction, stroke, and all-cause mortality during follow-up to 5 years. We used inverse probability weighting of a Cox proportional hazards model and reported results using hazard ratios with 95% confidence intervals.

Results: The cohort included 7,174 adult sepsis survivors, of whom 3,805 were new users of a RASi and 3,369 were new users of a calcium channel blocker or a thiazide diuretic. New users of a RASi experienced a lower hazard of major cardiovascular events than new users of a calcium channel blocker or a thiazide diuretic (hazard ratio, 0.93; 95% confidence interval, 0.87–0.99). This association was consistent across different follow-up intervals and multiple sensitivity analyses.

Conclusions: A new RASi prescription is associated with a reduction in major cardiovascular events after sepsis. A randomized controlled trial should be considered to confirm this finding.

Sepsis is a life-threatening acute organ dysfunction secondary to infection and a leading cause of morbidity and mortality worldwide (1, 2). It accounts for up to 15% of all critical care admissions and is associated with both acute and long-term consequences (3, 4). Recent studies suggest an increased risk of major cardiovascular outcomes after sepsis, with approximately 10% of sepsis survivors experiencing an event during long-term follow-up (58).

The mechanistic pathways underpinning the association between sepsis and cardiovascular disease may include persistent inflammation and immunomodulation, coagulation disturbances, and dysregulation of the renin-angiotensin-aldosterone system (5, 912). Renin-angiotensin system inhibitors (RASi), including angiotensin-converting enzyme inhibitors (ACEi) and angiotensin receptor blockers (ARBs), reduce adverse cardiovascular outcomes in a variety of conditions, including coronary heart disease, congestive heart failure, hypertension, and chronic kidney disease (1315). Moreover, observational cohort studies have shown the potential benefits of RASi use before hospitalization or their continuation after hospital discharge on all-cause mortality among patients with sepsis (1618). The long-term benefit of ACEi on major kidney outcomes after acute kidney injury requiring short-term dialysis has also been observed (19). However, whether initiation of a RASi after an episode of sepsis reduces the incidence of subsequent cardiovascular complications remains unknown. We therefore conducted a population-based cohort study to estimate the association between a new RASi prescription, compared with a new prescription for either calcium channel blockers or thiazide diuretics (active comparator), and major cardiovascular events during long-term follow-up after an episode of sepsis.

Study Design and Cohort Creation

We used databases contained at ICES to conduct an observational cohort study using population-based data from the province of Ontario, Canada. ICES is 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. The databases used included the Canadian Institute of Health Information Discharge Abstract Database, the National Ambulatory Care Reporting System database, and the Ontario Drug Benefit Program (ODB) claims database (see Table E1 in the data supplement). These datasets were linked using unique encoded identifiers and analyzed at ICES.

Our study was developed in accordance with the amended Declaration of Helsinki, and this report follows the 1) Strengthening the Reporting of Observational Studies in Epidemiology and 2) Reporting of Studies Conducted using Observational Routinely Collected Health Data Statement for Pharmacoepidemiology guidelines (20, 21). The design of this cohort study emulated a target randomized controlled trial; see Table E2 for details (22, 23). 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.

Study Population

Eligible patients were adults older than 65 years of age who were discharged after surviving a first sepsis hospitalization from April 2008 to April 2019. Sepsis was defined using International Classification of Diseases, 10th Revision, with Canadian Enhancements (ICD-10-CA) codes following the previously validated Jolley algorithm (24, 25). The age restriction was applied to ensure that all patients had information about outpatient prescriptions provided through the ODB. We excluded patients in whom a RASi may be considered first-line therapy (e.g., coronary artery disease, congestive heart failure, peripheral vascular disease, stroke, chronic kidney disease, or hypertension with diabetes mellitus), as recommended by current guidelines (26, 27). These baseline conditions were defined using ICD-10-CA codes (see different available databases in Table E1) during a 5-year look-back window (28). To maximize sensitivity while classifying prior cardiovascular disease, the presence of a single code was considered sufficient. We also excluded patients who required new renal replacement therapy during the sepsis hospitalization, patients who had a decreased estimated glomerular filtration rate (i.e., <30 ml/min) (29, 30), and patients without any serum creatinine measurements available. Detailed exclusion and inclusion criteria are presented in Table E3; a study design diagram is shown in Figure E1.

Exposure and Comparator Arms

The main exposure was defined as the dispensing of a new prescription of a RASi (either ACEi or ARBs) within the first 30 days after hospital discharge after a sepsis-related hospitalization. To reduce the magnitude of both unmeasured and residual confounding at baseline, we used an active control group as a comparator (31, 32). This active comparator group was composed of adult sepsis survivors who were dispensed a new prescription of calcium channel blockers or thiazide diuretics but not an ACEi or ARB within the first 30 days after hospital discharge. As such, the expectation was that the study would be composed of adult sepsis survivors with baseline hypertension (and no first-line indications for RASi) in whom the new prescription of a RASi, a calcium channel blocker, or a thiazide diuretic was (conditional on potential confounders) mostly driven by individual physician preferences.

Prevalent users (i.e., within 90 days before hospital admission) of either RASi, calcium channel blockers, or thiazide diuretics were excluded (32, 33). Calcium channel blockers excluded nondihydropyridine compounds (i.e., verapamil and diltiazem), which are also used to treat cardiac arrhythmias. All prescriptions were identified by their corresponding drug identification number in the ODB dataset. Those patients with a first new combination prescription (e.g., RASi in addition to either a calcium channel blocker or a thiazide diuretic) were considered part of the exposed group. Finally, in accordance with the estimation of the observational analog of the intention-to-treat effect, treatment changes after initial prescription were disregarded (Table E2) (23).

Outcomes and Follow-Up

The primary composite outcome was myocardial infarction, stroke, and all-cause death, occurring after the prescription date (i.e., index date) and up to 5 years of follow-up or until the end of the study period (April 1, 2020). This outcome was also analyzed at 1, 2, and up to 5 years. The association of a new RASi prescription with the individual components of the primary outcome was also summarized. Recurrent sepsis during the first year after hospital discharge served as a secondary outcome of interest (34). All outcomes were defined using ICD-10-CA codes (28, 35, 36). Coding strategies and algorithms (alongside their diagnostic accuracy) used for the main variables of interest have been described elsewhere and are summarized in Table E4 (6, 28).

Statistical Analysis

Patients’ demographic, clinical, and hospital-level characteristics were summarized using proportions for categorical variables and mean and standard deviation or median and interquartile range for continuous variables, as appropriate. Crude and weighted baseline characteristics of exposed versus control patients were compared using standardized mean differences (37). Standardized mean differences greater than 10% were considered relevant (37).

We used inverse probability of treatment weighting of a marginal structural Cox proportional hazards model to estimate the association between a new prescription of RASi and the primary outcome (38, 39). Specifically, we created a propensity score by fitting a logistic regression model with the exposure as the dependent variable. This model included potential confounders based on subject matter knowledge and selected on the basis of a directed acyclic graph (40, 41) (Figure E2) and the following principles: 1) controlling for covariates that act as common causes of the exposure and the outcome, 2) excluding any variable acting as a potential instrument, and 3) including any proxy for an unmeasured confounder (42). The set of covariates included age, sex, Charlson comorbidity index (43), income, active malignancy, dementia, atrial fibrillation, dyslipidemia, chronic conditions such as liver or pulmonary disease, long-term care residency, intensity of interventions during the hospital stay (e.g., receipt of invasive mechanical ventilation, blood transfusion, tracheostomy, and length of stay), and additional pharmacological prescriptions before and within 30 days of hospital discharge (e.g., the prescription of β-blockers, statins, antiplatelets, anticoagulants either before or together with the RASi or active comparator). For this propensity score model, continuous covariates were modeled flexibly using, if needed, cubic restricted splines. We then constructed stabilized weights that were trimmed at the 1st and 99th percentiles (37). Finally, we performed inverse probability of treatment weighting of a marginal structural Cox proportional hazards model (37, 44) to estimate the association between receiving a new prescription of a RASi compared with receiving a new prescription for a calcium channel blocker or thiazide diuretic and the primary outcome. Confidence intervals were estimated using nonparametric bootstrapping with 2,000 samples (45). We reported results using hazard ratios (HRs) alongside 95% confidence intervals (CIs) and constructed weighted survival curves. For the secondary outcome of interest (i.e., recurrent sepsis during the first year), we used a weighted Poisson model to estimate incidence rate ratios alongside 95% CIs.

Sensitivity analysis

We performed several sensitivity analyses to evaluate the robustness of our findings. First, to estimate the impact of potential confounding, we calculated the E-value for both the point estimate and the upper bound of the 95% CI (46). Second, to assess the impact of our analytical model, we refitted our model using the sandwich estimator to construct confidence intervals (45) and using propensity score adjustment with both linear and quadratic terms. Third, to maximize interpretability, we used a Bayesian framework considering varying degrees of prior beliefs (47). For this model, we reported median HRs alongside 95% CIs as well as posterior probabilities (details shown in the data supplement). Fourth, because patients with abnormal kidney function may respond differently to treatment with RASi, we refitted our analysis while restricting the population to those patients with a creatinine lower than 1.5 mg/dl. Fifth, we restricted our analysis to those patients classified as having hypertension at baseline. Sixth, we further planned to adjust, if needed, for those baseline covariates that were imbalanced (i.e., standardized mean difference higher than 10%) in the weighted sample. Seventh, because the analysis of the individual components of the primary outcome (i.e., myocardial infarction and stroke) may be subject to the competing risk of all-cause death, we report subdistribution HRs and 95% CIs from Fine and Gray models. Eighth, to assess the impact of different exposure definitions, we refitted our analysis 1) using a 90-day window for prescriptions after hospital discharge, 2) excluding patients who received any combination treatment (i.e., a RASi and a calcium channel blocker or a thiazide), and 3) considering patients without an antihypertensive prescription as the control group. Ninth, because we expected a high risk of all-cause death during follow-up, we refitted our analysis for the secondary outcome of interest (i.e., recurrent sepsis) using a weighted zero-inflated Poisson model. Finally, we estimated the observational analog of a baseline-adjusted per-protocol effect for the primary outcome.

All analyses were performed within ICES using SAS Enterprise Guide version 7.1 (SAS Institute). STATA version 15.1 (StataCorp LLC) and R version 4.1.0 (R Foundation for Statistical Computing) were used for figure production.

Out of 428,930 adult patients older than 65 years of age who survived a first sepsis hospitalization in the province of Ontario (2008–2019), 7,174 met eligibility criteria and were new users of a RASi (3,805 exposed patients) or either a calcium channel blocker or thiazide diuretic (3,369 active comparator patients) within 30 days of hospital discharge (Figure 1). These were included in the primary analysis to emulate a target randomized controlled trial.

Table 1 shows the baseline characteristics of the study sample. Approximately two-thirds (65%) of patients were female, and their mean age was 80.4 years (standard deviation, 8.5). The most common comorbidities were hypertension (86%), active malignancy (28%), and dementia (27%). For those patients with hypertension, the most commonly used antihypertensive before hospitalization was a β-blocker (13%). The most common source of infection was urinary tract infection (43%) followed by pneumonia (32%). Almost 15% of patients required admission to the intensive care unit during their first sepsis hospitalization, and approximately 5% received invasive mechanical ventilation. All baseline characteristics, including demographics, comorbidities, characteristics of the sepsis episode and hospital stay, and prescription patterns were appropriately balanced between groups after weighting (Tables 1 and 2).

Table 1. Baseline characteristics of adult sepsis survivors with dispensing of a new renin-angiotensin system inhibitor (exposure) or calcium channel blocker or thiazide diuretic prescription (active control) after hospital discharge in Ontario, 2008–2019

 Unweighted SampleWeighted Pseudopopulation*
RASi (n = 3,805)CCB or Thiazide Diuretics (n = 3,369)SMDRASi (n = 3,756)CCB or Thiazide Diuretics (n = 3,178)SMD
Baseline demographics
 Age, yr, mean (SD)79.7 (8.4)80.8 (8.7)0.1380.3 (8.4)80.4 (8.5)0.02
 Female sex, %60.866.60.1263.965.40.03
 Income quintile, %
  121.821.60.0021.821.50.01
  220.622.20.0420.922.60.05
  319.319.10.0119.119.30.01
  419.517.70.0519.917.50.06
  518.318.90.0218.218.60.01
  Missing0.50.50.000.50.50.00
 Area based material deprivation, %
  Quintile 1 (least deprived) to 357.657.50.0057.657.10.01
  Quintile 4 to 5 (most deprived)41.641.70.0041.742.20.01
  Missing0.80.80.000.70.70.00
 Long-term care home resident, %18.323.50.1320.721.60.02
 Previous hospitalizations, median (IQR)0 (0–1)0 (0–1)0.100 (0–1)0 (0–1)0.03
Baseline comorbidities
 Charlson comorbidity index score, median (IQR)1 (0–3)1 (0–3)0.011 (0–3)1 (0–3)0.00
 Hypertension, %86.884.30.0786.285.80.01
 Dyslipidemia, %18.114.30.1016.615.60.03
 Atrial fibrillation, %7.15.20.086.35.80.02
 Venous thromboembolism, %2.32.50.022.32.40.01
 Chronic liver disease, %1.01.90.081.31.50.01
 Chronic obstructive pulmonary disease, %35.935.90.0035.936.30.01
 Dementia, %22.529.70.1725.927.10.03
 Active malignancy, %26.830.00.0727.828.60.02
 LDL cholesterol, mmol/L, mean (SD)2.6 (0.9)2.7 (0.9)0.122.6 (0.9)2.7 (0.9)0.08
 HDL cholesterol, mmol/L, mean (SD)1.5 (0.5)1.5 (0.5)0.041.5 (0.5)1.5 (0.5)0.01
 Triglycerides, mmol/L, mean (SD)1.4 (0.7)1.4 (0.8)0.041.4 (0.7)1.4 (0.8)0.05
Sepsis episode characteristics
 Pneumonia as source of infection, %33.030.90.0431.931.70.01
 Urosepsis, %40.743.50.0642.242.50.01
 Acute kidney injury, %7.310.20.108.48.90.02
 Septic shock, %20.825.30.1122.823.40.01
 Hemoglobin, g/L, mean (SD)110.6 (20.1)107.8 (20.5)0.14109.9 (20.3)108.3 (20.0)0.08
 White cell count, ×109/L, mean (SD)14.7 (9.7)14.4 (12.9)0.0314.6 (9.8)14.5 (12.5)0.01
 Platelet count, ×109/L, mean (SD)305.5 (142.6)314.6 (162.8)0.06306.3 (140.9)314.8 (158.5)0.06
 Creatinine, μmol/L, mean (SD)84.6 (25.8)84.0 (27.9)0.0284.3 (25.9)84.1 (27.2)0.01
 Bilirubin, μmol/L, mean (SD)15.4 (20.5)14.6 (18.7)0.0415.2 (20.1)14.4 (16.8)0.04
 High troponin,§ %5.14.30.044.74.60.01
 Intensive care unit admission, %17.612.80.1415.013.90.03
 Transfusion, %10.112.50.0710.911.30.01
 Invasive mechanical ventilation, %5.44.60.045.04.90.00
 Length of hospital stay, d, median (IQR)8 (5–16)9 (5–20)0.148 (5–17)9 (5–20)0.01

Definition of abbreviations: CCB = calcium channel blocker (excluding verapamil or diltiazem); HDL = high-density lipoprotein; IQR = interquartile range; LDL = low-density lipoprotein; RASi = renin-angiotensin system inhibitor (includes both the use of an angiotensin convertase enzyme inhibitor or an angiotensin receptor blocker); SD = standard deviation; SMD = weighted absolute standardized mean difference.

Laboratory values (except for creatinine, which was required for inclusion) were not available for the entire cohort of patients and were not included in the propensity score model or other statistical analysis; they are shown here for overall baseline description.

*Numbers in the weighted sample are slightly different because of weighting and trimming.

Based on inverse probability of treatment weighting, using stabilized weights trimmed at the 1st and 99th percentiles. Propensity score model included age, female sex, Charlson comorbidity index score, previous hospitalizations, long-term residency, atrial fibrillation, venous thromboembolic disease, dyslipidemia, liver disease, chronic obstructive pulmonary disease, dementia, active malignancy, hypertension, septic shock, intensive care unit admission, site of infection, mechanical ventilation, renal replacement therapy, tracheostomy, transfusion requirements, respiratory failure, acute kidney injury, length of hospital stay, baseline creatinine, preadmission and postdischarge prescription patterns (including β-blockers, statins, antiplatelets, anticoagulants, antipsychotics, benzodiazepines).

The calculation of material deprivation factor scores includes variables for traditional socioeconomic status indicators, such as income, education, and employment, and has previously been shown as the Ontario Marginalization Index domain most strongly associated with health outcomes.

§High troponin was defined as troponin >99th percentile. A troponin measurement was available for 12% of the study sample.

Table 2. Cardiovascular prescription patterns for adult sepsis survivors with a new renin-angiotensin system inhibitor or calcium channel blocker or thiazide diuretic prescription after hospital discharge in Ontario, 2008–2019

 RASi
(n = 3,756)
CCB or Thiazide Diuretics
(n = 3,178)
SMD*
Prehospital prescriptions (within 90 d), %
 β-Blockers12.312.10.01
 Antiplatelets2.32.10.01
 Statins17.917.80.00
 Vitamin K antagonists3.23.10.00
 New oral anticoagulants1.51.60.01
Postdischarge prescriptions (within 30 d), %
 β-Blockers23.721.20.06
 Antiplatelets4.82.90.10
 Statins24.520.70.09
 Vitamin K antagonists5.85.10.03
 New oral anticoagulants3.63.40.01

Definition of abbreviations: CCB = calcium channel blocker (excluding verapamil or diltiazem); RASi = renin-angiotensin system inhibitor (includes both the use of an angiotensin convertase enzyme inhibitor or an angiotensin receptor blocker); SMD = weighted absolute standardized mean difference.

*Based on inverse probability of treatment weighting, using stabilized weights trimmed at the 1st and 99th percentiles.

ACEi were the main RASi prescribed (75%), whereas the active comparator arm was mostly composed of calcium channel blockers (81%) (Table E5). Thirteen percent of patients received combination treatment (mostly a RASi in addition to a calcium channel blocker), and nearly 3% of patients who initially received a prescription for a calcium channel blocker or a thiazide diuretic received a subsequent one for a RASi later (and up to 30 days after hospital discharge) (Table E5). The numbers of prescriptions per calendar year and by pharmacological group are shown in Figure E3. The main components of the propensity score model for a RASi after surviving sepsis are shown in Table E6; the distribution of propensity scores and description of weights by study arm is shown in Figure E4.

Association with Major Cardiovascular Events

Overall, the cumulative incidence rates (up to 5 years of follow-up) of the primary composite outcome in the weighted sample were 50.7% and 51.9% for those patients with a new prescription of a RASi or the active comparator, respectively. Follow-up information for both groups is shown in Table E7. Adult patients who survived a sepsis hospitalization and who received a prescription for a RASi within 30 days of hospital discharge experienced a lower hazard of cardiovascular events in up to 5 years of follow-up when compared with patients who received a prescription for either a calcium channel blocker or a thiazide diuretic (HR, 0.93; 95% CI, 0.87–0.99) (Table 3). Weighted survival curves for both groups are presented in Figure 2.

Table 3. Effect of a new renin-angiotensin system inhibitor prescription, compared with either a calcium channel blocker or a thiazide diuretic, within 30 days of hospital discharge on major cardiovascular events in adult sepsis survivors

Analytical StrategyHazard Ratio (95% CI)
Primary analysis
 Marginal structural Cox model with bootstrapping*0.93 (0.87–0.99)
Sensitivity analyses
 Changes in exposure definition
  Prescriptions up to 90 d of hospital discharge0.94 (0.88–0.99)
  Excluding patients with combination treatment0.93 (0.87–1.00)
  Comparison with patients with no antihypertensive therapy§0.89 (0.83–0.97)
 Changes in analytical strategy
  Marginal structural Cox model with robust standard errors0.93 (0.87–0.99)
  Propensity score adjustmentǁ0.93 (0.87–0.99)
  Bayesian reanalysis with strong pessimistic priors0.95 (0.89–1.01)
  Additional adjustment for cardiovascular prescriptions**0.93 (0.87–0.99)
  Baseline-adjusted per-protocol effect††0.89 (0.81–0.97)
 Changes in study population
  Restricting to patients with creatinine <1.5 mg/dl0.93 (0.86–0.99)
  Restricting to patients with coded baseline hypertension0.92 (0.85–0.99)
 E-value‡‡Risk Ratio
  For the point estimate1.28
  For the upper bound of confidence interval1.09

Definition of abbreviations: CCB = calcium channel blocker (excluding verapamil or diltiazem); CI = confidence interval; RASi = renin-angiotensin system inhibitor (includes both the use of an angiotensin converting enzyme inhibitor or an angiotensin receptor blocker).

Calcium channel blocker excludes verapamil or diltiazem. Major cardiovascular events were defined as the composite of myocardial infarction, stroke, and all-cause mortality.

*Inverse probability of treatment weighting of a marginal Cox proportional hazards model. Propensity score model included age, female sex, Charlson comorbidity index score, previous hospitalizations, long-term residency, atrial fibrillation, venous thromboembolic disease, dyslipidemia, liver disease, chronic obstructive pulmonary disease, dementia, active malignancy, hypertension, septic shock, intensive care unit admission, site of infection, mechanical ventilation, tracheostomy, transfusion requirements, respiratory failure, acute kidney injury, length of hospital stay, baseline creatinine, preadmission and postdischarge prescription patterns (including β-blockers, statins, antiplatelets, anticoagulants, antipsychotics, benzodiazepines). Confidence intervals are based on 2,000 bootstrapped samples.

Total of 9,028 patients; including 4,880 exposed to RASi and 4,148 exposed to either calcium channel blockers or thiazide diuretics.

Excluding from the exposed group those patients who had a prescription for both a RASi and a CCB or a thiazide diuretic.

§Based on a multivariable Cox model with a time-varying exposure (RASi yes vs. no).

ǁIncluding the propensity as a linear and quadratic term.

Prior defined as Ln(HR) ∼ N(0.4, 0.15). 95% credible interval shown.

**Including postdischarge prescriptions for β-blockers, antiplatelets, and statins.

††Observational analog of a baseline-adjusted per-protocol effect. Ninety-day grace period after last prescription of either a CCB/thiazide diuretic or a RASi. Based on a weighted Cox model with the same covariates as the main analysis. CIs are based on 2,000 bootstrapped samples.

‡‡The E-value represents the strength of association of an unmeasured confounder with both exposure and outcome that would explain away the estimated association between RASi and the outcome of interest.

The occurrence of the primary outcome was mostly composed of all-cause death (Table E8). The effect of RASi on individual components of the composite endpoint is shown in Table E8. HRs of myocardial infarction, stroke, and all-cause death for a new prescription of a RASi compared with either a calcium channel blocker or a thiazide diuretic were 1.17 (95% CI, 0.93–1.46), 0.90 (95% CI, 0.74–1.09), and 0.92 (95% CI, 0.86–0.99), respectively (Table E8 and Figure E5).

Association with Recurrent Sepsis

The incidence rates of recurrent sepsis during the first year after hospital discharge were 20.5 and 23.1 per 100 patient-years for the RASi and active comparator groups, respectively. Adult patients who survived a sepsis hospitalization and who received a prescription for a RASi within 30 days of hospital discharge experienced a lower incidence rate of recurrent sepsis during the first year of follow-up when compared with patients who received a prescription for either a calcium channel blocker or a thiazide diuretic (incidence rate ratio, 0.87; 95% CI, 0.77–0.98) (Table E8).

Sensitivity Analysis

Our results were robust across a variety of sensitivity analyses. The association between RASi therapy and reduced risk of cardiovascular events was consistent at different follow-up times (Table E9). Our analysis using either 1) robust standard errors or 2) propensity score adjustment yielded consistent estimates (Table 3). The estimates were also robust when further adjusting for postdischarge cardiovascular prescriptions, with different exposure definitions, when estimating a baseline-adjusted per-protocol effect, and when restricting to patients with normal serum creatinine or baseline hypertension (Table 3 and Figure E6). The results were consistent while using varying prior beliefs (Tables 3 and E10). For example, the probability of any benefit with the use of RASi when considering strong pessimistic beliefs was 95% (Tables 3 and E10). The calculated E-value for the point estimate was 1.28 (Table 3), suggesting that any unmeasured confounder would require at least a moderate association with both a new prescription of a RASi and major cardiovascular events to explain the main findings.

Our population-based cohort study shows that a new prescription of RASi after a first sepsis hospitalization is associated with a reduction in major cardiovascular events during long-term follow-up. This was mostly driven by a reduction in all-cause mortality and was present across different follow-up times and multiple sensitivity analyses. A new prescription of a RASi was also associated with a reduction in recurrent sepsis during the first year of follow-up.

Several observational cohorts and a recent systematic review have showcased the potential increased cardiovascular risk in adult sepsis survivors (58). The underlying mechanisms of this association have not been completely elucidated, but patients with a higher burden of comorbidities, who have an elevated troponin, or who need for life support during the sepsis episode appear to be at highest risk (48). Previous reports have indicated a potential beneficial effect of the prior use (before hospital admission) or continuation of antihypertensives and RASi among patients with sepsis (1618, 49), although such findings may be subject to so-called healthy user bias (50). In contrast, we excluded prevalent users and only included patients with a new pharmacological prescription after hospital discharge. RASi may be beneficial among sepsis survivors through the mitigation of changes in the renin-angiotensin-aldosterone pathway (51, 52), protective effects in patients who had acute kidney injury (19, 53), and a reduction in subsequent sepsis episodes (34). The results of our secondary endpoint of interest highlight the potential reduction in sepsis episodes associated with a new prescription for a RASi (which may in turn lie in the causal pathway toward reduced cardiovascular outcomes and all-cause mortality).

We believe the association between RASi and reduced risk of cardiovascular outcomes after sepsis suggests that this and other common secondary prevention strategies to reduce cardiac morbidity could be explored for sepsis survivors, with the overall goal of reducing such burden and improving clinical trajectories after hospital discharge (12). A recent report has suggested that the relative risk of cardiovascular disease in sepsis survivors (compared with survivors of a nonsepsis hospitalization) is highest in younger patients (6). Of note, we could not assess this subgroup, because our pharmacological data were limited to those patients older than 65 years of age. Future studies should specifically focus on the effect of RASi and other potential strategies among the younger subgroup of sepsis survivors.

Limitations

Our study has several limitations. First, our results are subject to residual and unmeasured confounding. However, the E-value associated with our primary analysis suggests that any such unmeasured confounder would need to have at least a moderate association with both the exposure and the outcome to explain our results. One plausible example is direct cardiovascular sequelae (e.g., systolic dysfunction) of the septic episode not captured by our database, which would be associated with both an increased likelihood of being prescribed a RASi after discharge and experiencing subsequent cardiovascular outcomes. However, the direction of this bias would be to increase the risk of subsequent cardiovascular events among RASi users. Second, we followed the observational analogue of the intention-to-treat effect and classified patients’ exposure to RASi according to their initial prescription (which considered combination therapy as part of the RASi group); however, this is a conservative approach because patients may not continue therapy or may change medications during long-term follow-up (e.g., crossover). Importantly, the estimated baseline adjusted per-protocol analysis was similar to our main estimate. Third, the selection bias introduced by the competing risk of all-cause death likely affects the analysis of the standalone individual endpoints (i.e., myocardial infarction and stroke). We used a primary composite outcome that included myocardial infarction, stroke, or death to address this issue. Because all-cause death was the main contributor to our primary composite endpoint, it is possible there are alternative explanations for our results, including a reduction in all-cause death not directly through cardiovascular disease but through other pathways that have not yet been elucidated. However, describing interventions that may affect such pathways to reduce all-cause mortality after discharge is still useful and clinically relevant. Fourth, the coding strategies for prior cardiovascular conditions that would mandate the use of RASi are expected to have imperfect sensitivity and specificity, thus introducing residual confounding; however, the direction of such bias would be opposite to the estimated beneficial effect. Fifth, we did not have access to information on self-reported race, which could have influenced both the prescription of a RASi or disparities in long-term outcomes (54). Sixth, we did not have information on discharge location after sepsis (e.g., long-term care residency), which could act as a surrogate marker of clinical condition or frailty and could be associated with both prescription patterns and subsequent clinical outcomes. Finally, because we followed strict inclusion and exclusion criteria and our population cohort had a high prevalence of dementia and malignancy and overall low acuity, our findings may not be generalizable to other settings, such as younger patients who survive an episode of sepsis. However, our study may also highlight the potential feasibility of future cohort and randomized studies, especially if they are able to include a wider range of sepsis survivors.

Conclusions

In conclusion, a new prescription for a RASi in adult patients who survive an episode of sepsis was associated with a reduction in major cardiovascular events and all-cause mortality during long-term follow-up. Future studies should seek to delineate whether this association is fundamentally a causal one. This may represent the first step in the deployment of specific mitigating strategies to help reduce the burden of cardiovascular disease among adult sepsis survivors.

Parts of this report are based on Ontario Registrar General information on deaths, the original source of which is ServiceOntario. The views expressed therein are those of the authors and do not necessarily reflect those of the Ontario Registrar General or the Ministry of Government and Consumer Services. The authors thank Dr. Bruno Ferreyro, Dr. Augusto Ferraris, and Dr. Alejandro Szmulewicz for the thoughtful discussions on the analysis and presentation of data. The authors also acknowledge the support of the Acute & Intensive Care Outcomes Research Network during the conduct of the present research.

1. Rudd KE, Johnson SC, Agesa KM, Shackelford KA, Tsoi D, Kievlan DR, et al. Global, regional, and national sepsis incidence and mortality, 1990-2017: analysis for the Global Burden of Disease Study. Lancet 2020;395:200211.
2. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016;315:801810.
3. Shankar-Hari M, Harrison DA, Ferrando-Vivas P, Rubenfeld GD, Rowan K. Risk factors at index hospitalization associated with longer-term mortality in adult sepsis survivors. JAMA Netw Open 2019;2:e194900.
4. Prescott HC, Costa DK. Improving long-term outcomes after sepsis. Crit Care Clin 2018;34:175188.
5. Kosyakovsky LB, Angriman F, Katz E, Adhikari NK, Godoy LC, Marshall JC, et al. Association between sepsis survivorship and long-term cardiovascular outcomes in adults: a systematic review and meta-analysis. Intensive Care Med 2021;47:931942.
6. Angriman F, Rosella LC, Lawler PR, Ko DT, Wunsch H, Scales DC. Sepsis hospitalization and risk of subsequent cardiovascular events in adults: a population-based matched cohort study. Intensive Care Med 2022;48:448457.
7. Ou SM, Chu H, Chao PW, Lee YJ, Kuo SC, Chen TJ, 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;194:209217.
8. Yende S, Linde-Zwirble W, Mayr F, Weissfeld LA, Reis S, Angus DC. Risk of cardiovascular events in survivors of severe sepsis. Am J Respir Crit Care Med 2014;189:10651074.
9. Mankowski RT, Yende S, Angus DC. Long-term impact of sepsis on cardiovascular health. Intensive Care Med 2019;45:7881.
10. Yende S, D’Angelo G, Mayr F, Kellum JA, Weissfeld L, Kaynar AM, et al.; GenIMS Investigators. Elevated hemostasis markers after pneumonia increases one-year risk of all-cause and cardiovascular deaths. PLoS One 2011;6:e22847.
11. Merdji H, Schini-Kerth V, Meziani F, Toti F. Long-term cardiovascular complications following sepsis: is senescence the missing link? Ann Intensive Care 2021;11:166.
12. Prescott HC, Angus DC. Enhancing recovery from sepsis: a review. JAMA 2018;319:6275.
13. Heart Outcomes Prevention Evaluation Study Investigators. Effects of an angiotensin-converting-enzyme inhibitor, ramipril, on cardiovascular events in high-risk patients. N Engl J Med 2000;342:145153.
14. Yusuf S, Teo KK, Pogue J, Dyal L, Copland I, Schumacher H, et al.; ONTARGET Investigators. Telmisartan, ramipril, or both in patients at high risk for vascular events. N Engl J Med 2008;358:15471559.
15. McMurray JJV. ACE inhibitors in cardiovascular disease—unbeatable? N Engl J Med 2008;358:16151616.
16. Hsieh MS, How CK, Hsieh VCR, Chen PC. Preadmission antihypertensive drug use and sepsis outcome: impact of angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin receptor blockers (ARBs). Shock 2020;53:407415.
17. Kim J, Kim YA, Hwangbo B, Kim MJ, Cho H, Hwangbo Y, et al. Effect of antihypertensive medications on sepsis-related outcomes: a population-based cohort study. Crit Care Med 2019;47:e386e393.
18. Ou SY, Lee YJ, Lo Y, Chen CH, Huang YC, Kuo YT, et al. Effects of renin-angiotensin–aldosterone system inhibitors on long-term major adverse cardiovascular events in sepsis survivors. J Am Heart Assoc 2021;10:e022870.
19. Brar S, Ye F, James MT, Hemmelgarn B, Klarenbach S, Pannu N; Interdisciplinary Chronic Disease Collaboration. Association of angiotensin-converting enzyme inhibitor or angiotensin receptor blocker use with outcomes after acute kidney injury. JAMA Intern Med 2018;178:16811690.
20. Langan SM, Schmidt SA, Wing K, Ehrenstein V, Nicholls SG, Filion KB, et al. The reporting of studies conducted using observational routinely collected health data statement for pharmacoepidemiology (RECORD-PE). BMJ 2018;363:k3532.
21. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007;370:14531457.
22. Madenci AL, Wanis KN, Cooper Z, Haneuse S, Subramanian SV, Hofman A, et al. Strengthening health services research using target trial emulation: an application to volume-outcomes studies. Am J Epidemiol 2021;190:24532460.
23. Hernán MA, Robins JM. Using big data to emulate a target trial when a randomized trial is not available. Am J Epidemiol 2016;183:758764.
24. Jolley RJ, Quan H, Jetté N, Sawka KJ, Diep L, Goliath J, et al. Validation and optimisation of an ICD-10-coded case definition for sepsis using administrative health data. BMJ Open 2015;5:e009487.
25. Jolley RJ, Sawka KJ, Yergens DW, Quan H, Jetté N, Doig CJ. Validity of administrative data in recording sepsis: a systematic review. Crit Care 2015;19:139.
26. Dasgupta K, Quinn RR, Zarnke KB, Rabi DM, Ravani P, Daskalopoulou SS, et al.; Canadian Hypertension Education Program. The 2014 Canadian Hypertension Education Program recommendations for blood pressure measurement, diagnosis, assessment of risk, prevention, and treatment of hypertension. Can J Cardiol 2014;30: 485501.
27. James PA, Oparil S, Carter BL, Cushman WC, Dennison-Himmelfarb C, Handler J, et al. 2014 Evidence-based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8). JAMA 2014;311:507520.
28. Tu JV, Chu A, Donovan LR, Ko DT, Booth GL, Tu K, 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;8:204212.
29. McMurray JJV, Packer M, Desai AS, Gong J, Lefkowitz MP, Rizkala AR, et al.; PARADIGM-HF Investigators and Committees. Angiotensin-neprilysin inhibition versus enalapril in heart failure. N Engl J Med 2014;371:9931004.
30. Inker LA, Eneanya ND, Coresh J, Tighiouart H, Wang D, Sang Y, et al.; Chronic Kidney Disease Epidemiology Collaboration. New creatinine- and cystatin c-based equations to estimate GFR without race. N Engl J Med 2021;385:17371749.
31. Angriman F, Masse M-H, Adhikari NKJ. Defining standard of practice: pros and cons of the usual care arm. Curr Opin Crit Care 2019;25:498504.
32. Yoshida K, Solomon DH, Kim SC. Active-comparator design and new-user design in observational studies. Nat Rev Rheumatol 2015;11:437441.
33. Hernán MA, Sauer BC, Hernández-Díaz S, Platt R, Shrier I. Specifying a target trial prevents immortal time bias and other self-inflicted injuries in observational analyses. J Clin Epidemiol 2016;79:7075.
34. Caldeira D, Alarcão J, Vaz-Carneiro A, Costa J. Risk of pneumonia associated with use of angiotensin converting enzyme inhibitors and angiotensin receptor blockers: systematic review and meta-analysis. BMJ 2012;345:e4260.
35. Porter J, Mondor L, Kapral MK, Fang J, Hall RE. How reliable are administrative data for capturing stroke patients and their care. Cerebrovasc Dis Extra 2016;6:96106.
36. Austin PC, Daly PA, Tu JV. A multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J 2002;144:290296.
37. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399424.
38. Hernán MA. Methods of public health research — strengthening causal inference from observational data. N Engl J Med 2021;385:13451348.
39. Hernán , Brumback B, Robins JM. Marginal structural models to estimate the causal effect of zidovudine on the survival of HIV-positive men. Epidemiology 2000;11:561570.
40. VanderWeele TJ, Hernán MA, Robins JM. Causal directed acyclic graphs and the direction of unmeasured confounding bias. Epidemiology 2008;19:720728.
41. Lederer DJ, Bell SC, Branson RD, Chalmers JD, Marshall R, Maslove DM, et al. Control of confounding and reporting of results in causal inference studies. Guidance for authors from editors of respiratory, sleep, and critical care journals. Ann Am Thorac Soc 2019;16:2228.
42. VanderWeele TJ. Principles of confounder selection. Eur J Epidemiol 2019;34:211219.
43. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373383.
44. Cole SR, Hernán MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol 2008;168:656664.
45. Austin PC. Variance estimation when using inverse probability of treatment weighting (IPTW) with survival analysis. Stat Med 2016;35:56425655.
46. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med 2017;167:268274.
47. Zampieri FG, Casey JD, Shankar-Hari M, Harrell FE Jr, Harhay MO. Using Bayesian methods to augment the interpretation of critical care trials. an overview of theory and example reanalysis of the alveolar recruitment for acute respiratory distress syndrome trial. Am J Respir Crit Care Med 2021;203:543552.
48. Garcia MA, Rucci JM, Thai KK, Lu Y, Kipnis P, Go AS, et al. Association between troponin I levels during sepsis and postsepsis cardiovascular complications. Am J Respir Crit Care Med 2021;204:557565.
49. Hsu WT, Galm BP, Schrank G, Hsu TC, Lee SH, Park JY, et al. Effect of renin-angiotensin-aldosterone system inhibitors on short-term mortality after sepsis: a population-based cohort study. Hypertension 2020;75:483491.
50. Hernán MA, Hernández-Díaz S, Robins JM. A structural approach to selection bias. Epidemiology 2004;15:615625.
51. Zarbock A, Chawla L, Bellomo R. Why the renin-angiotensin-aldosterone system (RAAS) in critically ill patients can no longer be ignored. Crit Care 2021;25:389.
52. Lund DD, Brooks RM, Faraci FM, Heistad DD. Role of angiotensin II in endothelial dysfunction induced by lipopolysaccharide in mice. Am J Physiol Heart Circ Physiol 2007;293:H3726H3731.
53. Gayat E, Hollinger A, Cariou A, Deye N, Vieillard-Baron A, Jaber S, et al.; FROG-ICU investigators. Impact of angiotensin-converting enzyme inhibitors or receptor blockers on post-ICU discharge outcome in patients with acute kidney injury. Intensive Care Med 2018;44: 598605.
54. Cerdeña JP, Plaisime MV, Tsai J. From race-based to race-conscious medicine: how anti-racist uprisings call us to act. Lancet 2020;396:11251128.
Correspondence and requests for reprints should be addressed to Federico Angriman, M.D., M.P.H., Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, Interdepartmental Division of Critical Care Medicine, University of Toronto, 2075 Bayview Avenue, Room D108, Toronto, ON, M4N 3M5, Canada. E-mail: .

Partially supported by the Canadian Institutes of Health Research and the Interdepartmental Division of Critical Care Medicine at the University of Toronto. The primary sources of support were a Vanier Graduate Scholarship from the Canadian Institutes of Health Research and a research award from the Interdepartmental Division of Critical Care Medicine at the University of Toronto. This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. Parts of this material are based on data and/or information compiled and provided by the Canadian Institute for Health Information. However, the analyses, conclusions, opinions, and statements expressed in the material are those of the author(s) and not necessarily those of the Canadian Institute for Health Information. F.A. is partially supported by a Vanier Canada Graduate Scholarship from the Canadian Institutes of Health Research and a research award from the Interdepartmental Division of Critical Care Medicine at the University of Toronto. H.W. is partially supported by a Canada Research Chair [Tier 2] in Critical Care Organization and Outcomes. D.C.S. holds operating grants from the Canadian Institutes of Health Research. P.R.L. is supported by a Heart and Stroke Foundation of Canada National New Investigator Award. D.T.K. is supported by the Jack Tu Chair in Outcomes Research, Sunnybrook Hospital. L.C.R. is supported by a Canada Research Chair in Population Health Analytics. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.

Author Contributions: All authors contributed to the study conception and design. Material preparation and data analysis were completed by F.A., L.C.R., H.W., and D.C.S. All authors contributed to data interpretation. The first draft of the manuscript was written by F.A. and D.C.S. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. F.A. takes responsibility for (is the guarantor of) the content of the article, including the data and analysis.

Data and Code Availability: The dataset from this study is held securely in coded form at ICES. Although data sharing agreements prohibit ICES from making the dataset publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS. The full dataset creation plan and underlying analytic code are available from the authors upon request, with the understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.

This article has a data supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.

Author disclosures are available with the text of this article at www.atsjournals.org.

Related

No related items
Comments Post a Comment




New User Registration

Not Yet Registered?
Benefits of Registration Include:
 •  A Unique User Profile that will allow you to manage your current subscriptions (including online access)
 •  The ability to create favorites lists down to the article level
 •  The ability to customize email alerts to receive specific notifications about the topics you care most about and special offers
Annals of the American Thoracic Society
20
3

Click to see any corrections or updates and to confirm this is the authentic version of record