Professional Documents
Culture Documents
com/viewarticle/771597_print
www.medscape.com
Previous Prescription of Betablockers Is Associated
With Reduced Mortality Among Patients Hospitalized in
Intensive Care Units for Sepsis
Alejandro Macchia, MD; Marilena Romero, PhD; Pablo Dino Comignani, MD; Javier Mariani, MD;
Antonio D'Ettorre, PhD; Nadia Prini, MD; Mariano Santopinto, MD; Gianni Tognoni, MD
Crit Care Med. 2012;40(10):27682772.
Abstract and Introduction
Abstract
Objectives: Results from basic science and narrative reviews suggest a potential role of βblockers in patients with sepsis.
Although the hypothesis is physiologically appealing, it could be seen as clinically counterintuitive. We sought to assess whether
patients previously prescribed chronic βblocker therapy had a different mortality rate than those who did not receive treatment.
Setting: Record linkage of administrative databases of Italian patients hospitalized for sepsis during years 2003–2008 were
identified and followed up for allcause mortality at 28 days.
Interventions: None.
Measurements and Main Results: We identified 9,465 patients aged >=40 yrs who were hospitalized in critical care units for
sepsis. Of these, 1,061 patients were on chronic prescription with βblockers and 8404 were not previously treated. Despite a
higher risk profile, patients previously prescribed with βblockers had lower mortality at 28 days (188/1061 [17.7%]) than those
previously untreated (1857/8404 [22.1%]) (odds ratio 0.78; 95% confidence interval 0.66–0.93; p = .005 for unadjusted analysis,
and odds ratio 0.81; 95% confidence interval 0.68–0.97; p = .025 for adjusted analyses). Sensitivity and pairmatched results
confirm the primary findings.
Conclusions: As far as we are aware, this pharmacoepidemiologic assessment is the largest to examine the potential
association of previous βblocker prescription and mortality in patients with sepsis. Chronic prescription of βblockers may confer
a survival advantage to patients who subsequently develop sepsis with organ dysfunction and who are admitted to an intensive
care unit. Prospective randomized clinical trials should formally test this hypothesis.
Introduction
Sepsis is a challenging and costly medical problem, frequently leading to intensive care unit (ICU) admissions and high in
hospital mortality rates. [1–4] Recently, myocardial dysfunction or myocardial injury in the course of sepsis has received much
attention as it is associated with a two to fourfold increase in mortality. [5–9] Two mechanisms proposed for myocardial injury in
sepsis, catecholaminergic overdrive and excess cytokine production, seem to play a central role. [5–8] Because βblockers
modulate both pathways, they emerge as potential agents that could reduce mortality. Several narrative reviews[5–8] suggest a
protective role of these agents in sepsis, and at least one clinical trial is currently testing the hypothesis. [10] However, preclinical
studies of βblockers in different models of sepsis provide conflicting evidence of their effectiveness. [11–16] and clinical data so
far are limited to small uncontrolled case series.
Using hospital discharge records, ambulatory prescription information, and vital statistics, we evaluated the association of
previous βblockers prescription on shortterm outcomes in patients admitted to ICU with a diagnosis of sepsis or septic shock.
Material and Methods
We conducted a record linkage analysis of administrative databases comprising hospital discharge records, prescription data of
all ambulatory treatments dispensed by the National Sanitary system and vital statistics data. These databases capture a
representative sample of the Italian population (36,328,015 people, ~12% of the population) distributed across 22 local health
areas (Piemonte, n = 1; Liguria, n = 7; Abruzzo, n = 2; Puglia, n = 12). Our cohort of patients was hospitalized for sepsis
between 2003 and 2008. Hospital discharge records included information on primary diagnoses and up to five coexisting clinical
diagnoses, source of sepsis, procedures performed, dates of admission and discharge (index date), and inhospital death.
Patients were identified on the basis of a discharge diagnosis of sepsis (International Classification of DiseasesClinical
Modification, Ninth Edition, codes 038 [septicemia], 020.0 [septicemic], 790.7 [bacteremia], 117.9 [disseminated fungal
infection], 112.5 [disseminated candida infection], and 112.81 [disseminated fungal endocarditis]) using a previously reported
and validated methodology. [1,2] Organ failure was defined by a combination of International Classification of DiseasesClinical
http://www.medscape.com/viewarticle/771597_print 1/8
9/7/2016 www.medscape.com/viewarticle/771597_print
Modification, Ninth Edition, and Current Procedural Terminology codes indicative of respiratory (518.81, 518.82, 518.85, 786.09,
799.1, 96.7), cardiovascular (458.0, 785.5, 785.51, 785.59, 458.0, 458.8, 458.9, 796.3), renal (584, 580, 585, 39.95), hepatic
(570, 572.2, 573.3), hematologic (286.2, 286.6, 286.9, 287.3–5), metabolic (276.2), or neurologic (293, 348.1, 348.3, 780.01,
780.09, 89.14) failure.
Using encrypted affiliation numbers, data for this cohort were linked with the local health authority drug claims database, which
provided prescription information: the Anatomical Therapeutic Chemical Classification, [17] dosages, date of first prescription,
and duration of exposure.
The analysis complied strictly with the national Italian regulations for the full protection of the privacy rights of the subjects
included in the databases. No ethical approval is required to perform this type of analysis.
Definition of Cohorts
We included consecutive patients aged ≥40 yrs admitted to ICU with a diagnosis of sepsis using the aforementioned codes. All
patients had to be admitted directly to the ICU (from the emergency department) or transferred from other hospital departments
to the ICU within 48 hrs of hospital admission.
Comorbid conditions were extracted from patient clinical history over the 12 months preceding the index hospitalization. Chronic
exposure to pharmacological treatments, and previous procedures and hospitalizations were assumed as identifiers of underlying
clinical conditions; in particular, cardiovascular comorbidities of interest here were previous hospitalizations (in the year before
index hospitalization) for stroke, transient ischemic attack, peripheral vascular disease, embolic episode, atrial fibrillation,
congestive heart failure, and myocardial infarction. Previous chronic, ambulatory treatments allowed identification of patients
with a history of arterial hypertension and diabetes mellitus. Noncardiovascular conditions included malignancy, hospitalization
for major bleeding, chronic obstructive pulmonary disease, and depression. We considered patients as previously exposed to β
blockers if they filled greater than three or more prescriptions for these agents during the 4month period before the index
hospitalization.
Outcomes
The primary objective was to evaluate the association between previous prescription with βblockers and survival up to 28 days
from ICU admission.
Statistical Analysis
Continuous data were described using means and SD or medians and interquartile range, if normally or not normally distributed,
respectively. Categorical data are expressed as counts and percentages. To compare baseline characteristics, we used Student's
t or WilcoxonMannWhitney tests for continuous data and chisquare test for categorical variables.
The association between prescription of βblockers and survival at 28 days was evaluated through a logistic regression (LR)
model. To account for confounders, the model included the following covariates: age, sex, history of hypertension, dyslipidemia,
diabetes mellitus, myocardial infarction, congestive heart failure, atrial fibrillation, chronic obstructive pulmonary disease,
depression, and malignancy. Results of multivariate LR are presented as odds ratios with corresponding 95% confidence
intervals, for both unadjusted and adjusted estimates. Mortality rates were expressed as KaplanMeier estimates through the 28
day period, and differences measured using the logrank test.
Sensitivity Analysis
Two sensitivity analyses were conducted to test the robustness of the results. First was a subgroup analysis using multivariate
LR models that included as interaction term the indicator variable of the subset (one of the covariates, e.g., hypertension) by the
βblocker exposure indicator variable. To further assess consistency and reduce bias in treatment comparisons, we conducted a
propensity score (PS) matching analysis. [18] Pairwise LR models are first used to predict the probability (PS) to be assigned to
the specific treatment group vs. controls (i.e., no βblockers). These models included all the aforementioned covariates. PS
pairwise logistic models were selected in a stepwise fashion, and model building stopped when adequate covariate balance
within PS quintiles was reached, as described elsewhere. Residual imbalances of covariates within PS quintiles were assessed
at each step with a twoway analysis of variance where each confounder was considered as outcome and PS quintiles and
treatment as factors. Overlapping of PS distribution between treatment and control groups was also checked. Finally, PS
http://www.medscape.com/viewarticle/771597_print 2/8
9/7/2016 www.medscape.com/viewarticle/771597_print
quintiles were introduced in the LR model to allow an adjusted comparison between patients in each treatment group and
controls for the two outcomes. The association between exposure and outcome in the propensitymatched pairs was assessed
using McNemar's test.
All reported p values are two tailed, and a p value <.05 indicates a significant difference. We did no adjustments for multiple
comparisons. All analyses were conducted using SPSS version 16.0 for Windows (SPSS, Chicago, IL).
Results
Baseline Characteristics
During the study period, 12,594 patients were admitted to ICUs with a diagnosis of sepsis; 9465 were ≥40 yrs and constituted
the population under study. A total of 1,061 patients (11.2%) were previously prescribed βblockers and 8,404 (88.8%) were
previously unexposed. shows demographics and clinical background of these patients. Patients previously prescribed with β
blockers had higher prevalence of risk factors and cardiovascular hospitalizations than their unexposed counterparts.
Table 1. Demographic and clinical characteristics of the study population
Demographic and baseline conditions
Inhospital characteristics and severity
The median time interval between last prescription of βblocker and the index hospitalization was 27 (interquartile range 13.0–
50.0) days, and the median number of refills during the year before hospitalization was ten. By 28 days postICU admission,
2,045 patients died. Patients previously prescribed βblockers had a lower mortality of 17.7% (188/1061) compared to 22.1%
(1857/8404) in patients not previously exposed (odds ratio 0.78; 95% confidence interval 0.66–0.93; p = .005 for unadjusted
analysis, and odds ratio 0.81; 95% confidence interval 0.68–0.97; p =.025 for adjusted analyses) (Fig. 1).
http://www.medscape.com/viewarticle/771597_print 3/8
9/7/2016 www.medscape.com/viewarticle/771597_print
Figure 1.
Survival plot for 28day mortality.
Sensitivity Analyses
The results of the sensitivity analysis were consistent with main results. The effect of previous prescription of βblockers on 28
day mortality was verified in all groups of patients (Fig. 2), including both sexes, elderly patients, those with previous
comorbidities, as well in those who developed organ dysfunction. The inclusion of other previous pharmacologic treatments
(calcium channel blockers, amiodarone, angiotensinconvertingenzyme inhibitors, diuretics, or any nonsteroidal anti
inflammatory drugs) into the correction model did not materially change these results.
http://www.medscape.com/viewarticle/771597_print 4/8
9/7/2016 www.medscape.com/viewarticle/771597_print
Figure 2.
A sensitivity analysis provides estimates of effects of previous prescription with βblockers on 28day mortality by demographic or
http://www.medscape.com/viewarticle/771597_print 5/8
9/7/2016 www.medscape.com/viewarticle/771597_print
clinical characteristics. CHF, congestive heart failure.
PSmatched Analysis
Of 1,061 patients prescribed with βblockers, 995 had a PSmatched control. As expected for this analysis, controls and
previously prescribed patients had similar demographics and clinical characteristics (). Results of mortality at 28 days were
similar to those described for the whole cohort. There were 182 (18.3%) and 233 (23.4%) fatal events for those previously and
not previously prescribed βblockers, respectively (odds ratio 0.72; 95% confidence interval 0.57–0.91; p = .004).
Table 2. Demographic and clinical characteristics of the propensity score–matched population
Discussion
The present study suggests that patients who receive chronic βblocker prescription may have a survival advantage if they
subsequently develop sepsis and are admitted to the ICU. Although there are many theoretical reasons βblockers may be
useful during early phases of sepsis, [5–8] physicians may also find this therapy counterintuitive. [19,20] Experiences coming from
the routine use of βblockers in noncardiac surgery[21] and during the acute phase of myocardial infarction[22] seem to counter
our hypothesis. Despite this, our results are in line with a substantial body of basic research, [11–16] and as far as we are aware,
this analysis is the largest to show a potential association of βblocker use and survival advantage in patients with sepsis. This
challenging hypothesis merits a prospective randomized controlled trial to formally test the concept.
Our results should be viewed within the context of pharmacoepidemiologic studies, which have their own strengths and
constitute an important source of information, particularly in the field of critical care. [1,2] There are also limitations; the data
sources used in this study lack important clinical information (risk stratification at baseline, source of infection, physiologic
measurements, and data on standard medical strategies) and while the codes identifying sepsis were used previously, [1,2] they
have not been validated in Italy. Perhaps the most significant limitation, owing to the nature of our data, is that previous
prescription of βblockers cannot be interpreted as certain therapeutic effect. We do not have heart rate data, but the median
time between last refill and hospitalization would suggest that most patients were under therapeutic βblockade. Although these
patients had ≥3 prescriptions of βblockers filled within the 4 months before hospitalization, it does not necessarily mean that
they took the medications; however, it is the most probable explanation. Despite these limitations, our results do have a strong
internal consistency showing that patients previously prescribed βblockers, although at higher risk than the controls, had
significantly lower mortality. This reduction was consistently confirmed across the sensitivity analyses helping to strengthen the
hypothesis.
A plausible explanation with a strong body of evidence[5–8,11–16] may be the role of βblockers related to inhibition of cardiac
depression, which begins long before the development of severe sepsis or septic shock. [23–29] Another hypothesis for
http://www.medscape.com/viewarticle/771597_print 6/8
9/7/2016 www.medscape.com/viewarticle/771597_print
consideration is that prior use of βblockers might identify patients who are well connected to the healthcare system or that could
have enhanced and earlier signs and symptoms of sepsis, so they sought medical attention earlier.
Conclusions
Although challenging, our data support the hypothesis that previous prescription of βblockers may confer a survival advantage
to patients who later develop sepsis. Robust epidemiologic data, such as ours, should guide clinical decision to move forward to
conduct welldesigned, prospective clinical trials testing the effectiveness of this strategy in clinical practice.
References
1. Martin GS, Mannino DM, Eaton S, et al: The epidemiology of sepsis in the United States from 1979 through 2000. N
Engl J Med 2003; 348:1546–1554
2. Angus DC, LindeZwirble WT, Lidicker J, et al: Epidemiology of severe sepsis in the United States: Analysis of incidence,
outcome, and associated costs of care. Crit Care Med 2001; 29:1303–1310
3. Angus DC, Wax RS: Epidemiology of sepsis: An update. Crit Care Med 2001; 29(Suppl 7): S109–S116
4. Vincent JL, Abraham E, Annane D, et al: Reducing mortality in sepsis: New directions. Crit Care 2002; 6 Suppl 3:S1–
S18
5. Rudiger A: Betablock the septic heart. Crit Care Med 2010; 38(10 Suppl):S608–S612
6. Rudiger A, Singer M: Mechanisms of sepsisinduced cardiac dysfunction. Crit Care Med 2007; 35:1599–1608
7. de Montmollin E, Aboab J, Mansart A, et al: Benchtobedside review: Betaadrenergic modulation in sepsis. Crit Care
2009; 13:230
8. Oberbeck R, Kobbe P: Betaadrenergic antagonists: Indications and potential immunomodulatory side effects in the
critically ill. Curr Med Chem 2009; 16:1082–1090
9. VieillardBaron A, Caille V, Charron C, et al: Actual incidence of global left ventricular hypokinesia in adult septic shock.
Crit Care Med 2008; 36:1701–1706
10. Heart rate control with esmolol in septic shock. ClinicalTrials.gov Identifier: NCT01231698. Available at:
http://clinicaltrials.gov/ct2/show/NCT01231698. Accessed June 26, 2012
11. Markley K, Smallman ET, Briggs LW: Early mortality and temperature regulation in burned mice following administration
of catecholamines and adrenergic receptor blocking drugs. J Trauma 1979; 19:512–518
12. Lee SC, Han JS, Seo JK, et al: Modulation of cyclooxygenase2 expression by phosphatidylcholine specific
phospholipase C and D in macrophages stimulated with lipopolysaccharide. Mol Cells 2003; 15:320–326
13. Suzuki T, Morisaki H, Serita R, et al: Infusion of the betaadrenergic blocker esmolol attenuates myocardial dysfunction
in septic rats. Crit Care Med 2005; 33:2294–2301
14. Schmitz D, Wilsenack K, Lendemanns S, et al: BetaAdrenergic blockade during systemic inflammation: Impact on
cellular immune functions and survival in a murine model of sepsis. Resuscitation 2007; 72:286–294
15. Muthu K, He LK, Szilagyi A, et al: Propranolol restores the tumor necrosis factoralpha response of circulating
inflammatory monocytes and granulocytes after burn injury and sepsis. J Burn Care Res 2009; 30:8–18
16. Ackland GL, Yao ST, Rudiger A, et al: Cardioprotection, attenuated systemic inflammation, and survival benefit of beta1
adrenoceptor blockade in severe sepsis in rats. Crit Care Med 2010; 38:388–394
17. US Department of Health and Human Services: International Classification of Diseases, 9th Revision (Clinical
http://www.medscape.com/viewarticle/771597_print 7/8
9/7/2016 www.medscape.com/viewarticle/771597_print
Modification). Washington, Public Health Service, 1998
18. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects.
Biometrika 1983; 70: 41–55
19. Wheeler AP, Bernard GR: Treating patients with severe sepsis. N Engl J Med 1999; 340:207–214
20. Meldrum DR: Betablockade during sepsis: Inspired or insane? Crit Care Med 2005; 33:2433–2434
21. Devereaux PJ, Yang H, Yusuf S, et al; POISE Study Group: Effects of extendedrelease metoprolol succinate in patients
undergoing noncardiac surgery (POISE trial): A randomised controlled trial. Lancet 2008; 371:1839–1847
22. Chen ZM, Pan HC, Chen YP, et al; COMMIT (ClOpidogrel and Metoprolol in Myocardial Infarction Trial) collaborative
group: Early intravenous then oral metoprolol in 45,852 patients with acute myocardial infarction: Randomised placebo
controlled trial. Lancet 2005; 366:1622–1632
23. Annane D, Trabold F, Sharshar T, et al: Inappropriate sympathetic activation at onset of septic shock: A spectral analysis
approach. Am J Respir Crit Care Med 1999; 160: 458–465
24. Kovarik MF, Jones SB, Romano FD: Plasma catecholamines following cecal ligation and puncture in the rat. Circ Shock
1987; 22:281–290
25. Brealey D, Brand M, Hargreaves I, et al: Association between mitochondrial dysfunction and severity and outcome of
septic shock. Lancet 2002; 360:219–223
26. Tang C, Liu MS: Initial externalization followed by internalization of betaadrenergic receptors in rat heart during sepsis.
Am J Physiol Integr Comp Physiol 1996; 270:254–263
27. Tang C, Yang J, Wu LL, et al: Phosphorylation of betaadrenergic receptor leads to its redistribution in rat heart during
sepsis. Am J Physiol 1998; 274:R1078–R1086
28. Böhm M, Kirchmayr R, Gierschik P, et al: Increase of myocardial inhibitory Gproteins in catecholaminerefractory septic
shock or in septic multiorgan failure. Am J Med 1995; 98:183–186
29. Bernardin G, Strosberg AD, Bernard A, et al: Betaadrenergic receptordependent and independent stimulation of
adenylate cyclase is impaired during severe sepsis in humans. Intensive Care Med 1998; 24: 1315–1322
Acknowledgments
Editing for English and to reduce word count was provided by Elizabeth Sesler, PhD.
Crit Care Med. 2012;40(10):27682772. © 2012 Lippincott Williams & Wilkins
This website uses cookies to deliver its services as described in our Cookie Policy. By using this website, you agree to the use of
cookies.
close
http://www.medscape.com/viewarticle/771597_print 8/8