Journal of Critical Care xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Critical Care journal homepage: www.jccjournal.org

Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis☆ Yui-Rwei Young, MD a, b, Bor-Fuh Sheu, MD b, c, Wen-Cheng Li, MD b, d, Ting-Min Hsieh, MD b, e, Chi-Wei Hung, MD b, f, Shy-Shin Chang, MD b, g, h,⁎, Chien-Chang Lee, MD, ScD i, j,⁎⁎ a

Department of Emergency Medicine, Chang Gung Memorial Hospital, Chiayi Branch, Puzih City, Chiayi County, 613, Taiwan Chang Gung University College of Medicine, Gueishan Township, Taoyuan County, 333, Taiwan Department of Emergency Medicine, Chang Gung Memorial Hospital, Linkou Branch, Gueishan Township, Taoyuan County, 333, Taiwan d Department of Occupation Medicine, Chang-Gung Memorial Hospital, Chiayi Branch, Puzih City, Chiayi County, 613, Taiwan e Division of General Surgery, Department of Surgery, Chang Gung Memorial Hospital, Kaohsiung Branch, Kaohsiung County, 833, Taiwan f Department of Emergency Medicine, Kaohsiung Chang Gung Memorial Hospital, Kaohsiung County, 833, Taiwan g Department of Family Medicine, Chang Gung Memorial Hospital, Taoyuan, Taiwan h Graduate Institute of Clinical Medical Sciences, College of Medicine, Chang Gung University, Taoyuan, Taiwan i Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA j Department of Emergency Medicine, National Taiwan University Hospital, Douliou, Yunlin Branch, Taiwan and Department of Emergency Medicine, National Taiwan University Hospital, Taipei, Taiwan b c

a r t i c l e

i n f o

Keywords: B-type natriuretic peptide N-terminal pro–b-type natriuretic peptide Cardiac complication Surgery

a b s t r a c t Background: We aimed to undertake a systematic review and meta-analysis of studies addressing perioperative natriuretic peptide (NP) levels to predict postoperative major adverse cardiac events (MACE) after major surgery. Methods: We searched MEDLINE and Embase with no language restrictions up to May 2013. The end points were major cardiac complications. We summarized test performance characteristics with the use of forest plots, hierarchical summary receiver operating characteristic curves, and bivariate random effects models. Results: Of the 662 retrieved articles, 24 studies satisfied the predefined eligibility criteria, including 5438 patients along with 712 (13.1%) events. After major surgery, the diagnostic odds ratio (DOR) of NP in predicting postoperative MACE was 14.3 (95% confidence interval [CI], 9.87-20.7) for overall population, 13.9 (8.43-22.8) for patients undergoing cardiac surgery, and 15.0 (8.84-25.5) for patients undergoing noncardiac surgery. The pooled sensitivity was 0.84 (95% CI, 0.79-0.88) and specificity was 0.76 (95% CI, 0.71-0.81). Postoperative measurement (DOR, 18.9; 7.68-46.3) was associated with higher predictive value than preoperative measurement (DOR, 13.6; 7.68-46.3). Results were similar for a subgroup with the composite outcome including mortality (DOR, 16.4; 10.625.5). B-type natriuretic peptide was associated with higher predictive accuracy (area under the summary receiver operating characteristic, 0.84; 0.81-0.87) than N-terminal pro–b-type natriuretic peptide (area under the summary receiver operating characteristic, 0.90; 0.87-0.92). Conclusions: The existing literature suggests that perioperative NP testing have reasonable accuracy and can be useful in perioperative risk stratification. Natriuretic peptide testing has high rule-out value and low rule-in value for predicting postoperative MACE. Medical decisions should be made in the context of these characteristics. © 2014 Elsevier Inc. All rights reserved.

1. Introduction Patients with multiple risk factors undergoing major surgery carry high mortality and morbidity. Most patients did not die due to the operation itself but die due to the development of major cardiac complications associated in the perioperative period [1]. Early identification of patients at highest risk preoperatively allows early ☆ Conflict of interest: None declared. ⁎ Correspondence to: SS Chang, No. 5, Fu-Hsing St, Kuei Shan Hsiang, Taoyuan, Taiwan. Tel.: + 886 3 328 1200x2482; fax: +886 3 328 7715. ⁎⁎ Correspondence to: CC Lee, No. 579 Sec 2 Yunlin Road, Douliou, Yunlin County 63247, Taiwan. E-mail addresses: [email protected] (S.-S. Chang), [email protected] (C.-C. Lee).

close monitoring and early preventive medical intervention such as perioperative β blockade or hemodynamic optimization, which may result in improved outcome and reduced cardiac complications. Currently existing risk assessment instrument, such as the Revised Cardiac Risk Index, were developed in patients undergoing major elective noncardiac surgery and may not be generalized in many other high-risk surgeries [2]. American Heart Association and the American College of Cardiology recommend stress testing and/or echocardiography for preoperative risk assessment but may be impractical in emergency surgery [3]. A readily available “biomarker” that can assist with risk stratification would therefore be of particular value in this situation. B-type natriuretic peptide (BNP) or its inactive cleavage product N-terminal pro–b-type natriuretic peptide (NT-proBNP) is secreted

http://dx.doi.org/10.1016/j.jcrc.2014.03.022 0883-9441/© 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

2

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

mainly from the left ventricle in response to pressure and volume overload. Systemic physiologic effects of natriuretic peptides (NPs) include natriuresis, diuresis, and vasodilation [4]. Several studies have shown its ability to predict perioperative major adverse cardiovascular events (MACE) including atrial or ventricular arrhythmia, myocardial infarction, or sudden cardiac death [5–31]. Several meta-analyses have demonstrated that increased NT-proBNP levels predict short- and longer term cardiac complications [4,32,33]. However, these meta-analyses reported odds ratio (OR) as the primary effect measure and did not report pooled estimates of sensitivity, specificity, or likelihood ratio [34]. For a diagnostic test, likelihood ratio can be used to calculate posttest probability of an event, which is critical for clinical decision or explanation of the risk of surgery to the patients. In addition, none of these analyses provided comprehensive comparison of the predictive performance between BNP and NT-proBNP tests. Recent individual patient’s data analyses, though reporting sensitivity and specificity measures, could not include as many studies as done in an aggregate data meta-analysis [35]. The small number of studies available also did not allow subgroup analysis on multiple categories of clinical interest. We therefore carry out an updated systemic review and metaanalysis to summarize data on the association between NP values and cardiac complications in adult patients. 2. Methods 2.1. Data sources and searches We performed this meta-analysis in accordance with the Preferred Reporting Items for Systematic reviews and Meta-

Analyses guidelines. We searched MEDLINE and Embase for studies published through May 2013 with the following Medical Subject Headings terms and free text: “natriuretic peptide,” “surgery,” “cardiac complications,” “myocardial infarction,” “cardiac death,” “ventricular tachycardia,” “ventricular fibrillation,” and “major adverse cardiac event.” There were no restrictions on language, population, or publication year. We did not search abstracts from conferences, proceedings, or clinical trial registries. Instead, we manually reviewed the bibliographies of relevant studies, reviews, and meta-analyses to identify references we may have missed during our primary search. Selection was performed independently by 2 reviewers. The authors of the included trials were contacted for missing information if necessary. Discrepancies between the reviewers were resolved by a consensus meeting with a third reviewer. The studies were screened for title and abstract in the first round, and potentially relevant articles were retrieved for full-text review in the second round. To be eligible for inclusion, the studies had to fulfill the following criteria: (1) have a study population of consecutive patients (age N18 years) undergoing cardiac or noncardiac surgery, (2) assess a BNP or NT-proBNP test in the perioperative period, (3) use perioperative cardiac death, acute coronary syndrome, heart failure, arrhythmia, or cardiogenic shock as one of the primary end point and (4) include calculation of sensitivity, specificity, or OR or have sufficient data to construct a 2 × 2 contingency table. We excluded case reports, case series, review articles, editorials, and clinical guidelines. Full-text articles were retrieved if any of the reviewers considered the abstract suitable. The study inclusion and exclusion process is summarized in Fig. 1.

Fig. 1. Flow chart of study identification and inclusion.

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

3

2.2. Quality assessment

2.3. Data synthesis and analysis

The methodological quality of the selected studies was evaluated independently by 2 reviewers with a validated tool for the quality assessment of diagnostic accuracy studies [36]. Discrepancy was resolved by consensus meeting by inclusion of a third coauthor.

We calculated the pooled sensitivity, specificity, likelihood ratio as well as diagnostic odds ratio (DOR) for included studies. Given the inherent negative correlation between sensitivity and specificity, we estimated these parameters by using the bivariate model for

Table 1 Summary of the characteristics of the included studies Author, year, country Hutfless et al, 2004, USA [5] Kerbaul et al, 2004, France [6]

Mean Prevalence age (N)

Outcomes

Setting Type of surgery

63

0.33 (98)

Cardiac complications

Ward

Heart surgery

67.7

0.35 (60)

ICU

OPCAB

Yeh et al, 2005,Taiwan [7] Dernellis & Panaretou, 2006, Greece [9] Feringa et al, 2006, Netherlands [10] Berry et al, 2006, UK [8] Gibson et al, 2007, UK [13] Cuthbertson et al, 2007, UK [12] Cuthbertson et al, 2007, UK [11] Mahla et al, 2007, Germany [14] Yun et al, 2008, Korea [17] Leibowitz et al, 2008, Israel [15]

57

0.08 (190)

Ward

Thoracic, major vascular, abdominal noncardiac surgery

70

0.06 (1590)

Ward

59

0.08 (170)

Cardiovascular complications, MI, cardiogenic shock, arrhythmias, congestive heart failure, death Cardiac death, ACS, heart failure, arrhythmia Cardiac death, nonfatal MI, pulmonary edema, ventricular tachycardia Death, nonfatal MI

Ward

Abdominal, genitourinary, orthopedic, head and neck, noncardiac surgery Major vascular, abdominal aortic, aneurysm, or leg bypass

70 68

0.27 (41) 0.21 (190)

fatal or nonfatal MI Cardiac death, nonfatal MI

Ward Ward

Vascular surgery Thoracic, major vascular, abdominal, major vascular surgery

74

0.28 (40)

Cardiac death, MI

ED

Abdominal, orthopedic major noncardiac surgery

66

0.06 (204)

MI, death, arrhythmia

Ward

Major vascular, abdominal, genitourinary major noncardiac surgery

70

0.20 (218)

Ward

Major vascular surgery

68

0.09 (279)

Ward

77

0.34 (44)

Ward

Thoracic, abdominal, genitourinary, orthopedic, head and neck, vascular surgery Abdominal, orthopedic, noncardiac surgery

Goei et al, 2009, The Netherlands [19] Rajagopalan et al, 2008, UK [16] Oscarsson et al, 2009, Sweden [21] Choi et al, 2010, USA [23] Bolliger et al, 2009, Switzerland [18] Nozohoor et al, 2009, Sweden [20] Schutt et al, 2009, USA [22]

70

0.14 (592)

Nonfatal MI, acute coronary revascularization, cardiac death Cardiac death, nonfatal MI, pulmonary edema, nonfatal stroke Clinical end points were death, MI, or pulmonary congestion requiring intravenous diuretics Cardiovascular death, nonfatal MI

Ward

Abdominal aortic, peripheral, or carotid noncardiac surgery

69

0.21 (136)

MI

Ward

Major vascular surgery

80

0.22 (110)

MACE

Ward

68

0.14 (2054)

PMCE

Ward

68

0.14 (133)

MACE

Ward

Urology, abdominal surgery, hand and reconstructive surgery, gynecologic, neurosurgery, orthopedic, ophthalmological Thorax, abdomen, head and neck, orthopedic, prostate, neurosurgery, noncardiac surgery Major vascular surgery

70.4

0.23 (161)

Heart failure

ICU

Aortic valve replacement

69.5

0.33 (75)

Heart failure, new dysrhythmia, MIs, cardiac arrests

Ward

77.1

0.08 (208)

Ward

Chong et al, 2010, Australia [24]

79

0.26 (89)

ED

Lower limb orthopedic surgery

Nojiri et al, 2011, Japan [26] Chong et al, 2012, Australia [29]

78.4

0.34 (80)

Cardiac events, death, acute MI; unstable angina, atrial fibrillation, ventricular tachycardia, heart failure Cardiac complications, acute MI, congestive cardiac failure, atrial fibrillation, or major arrhythmia Cardiopulmonary complications

Emergency major operation, vascular surgery, orthopedic surgery, head and neck surgery, prolonged procedure/large fluid shift, intraperitoneal or intrathoracic surgery, prostate surgery, gynecologic, neurosurgery Orthopedic surgery

Ward

Thoracic surgery

76.7

0.11 (187)

Ward

Orthopedic surgery

Yang et al, 2012, Korea [31]

67

0.34 (365)

Ward

Vascular

Bryce et al, 2012, UK [28] Biccard et al, 2012, South Africa [27] Mercantini et al, 2012, Italy [30]

73

0.15 (106)

Ward

Abdominal aortic aneurysm

58.2

0.17 (788)

Acute MI, congestive cardiac failure, atrial fibrillation, or major arrhythmia and death. MI, development or aggravation of congestive heart failure, or primary cardiovascular death. Cardiac events, nonfatal MI, and cardiac death MACE

Ward

Vascular surgery

64

0.10 (205)

Ward

Hepatobiliary, coloproctological, esophagogastric, endocrinologic, gynecologic, urologic, abdominal wall abdominal surgery

Villacorta et al, 2010, Brazil [25]

Cardiac breathlessness, heart failure, cardiac exitus, angina pectoris, hypertensive

ED indicates emergency department; PMCE, perioperative major cardiovascular event; OPCAB, off-pump coronary artery bypass graft; CABG, coronary artery bypass graft; ACS, acute coronary syndrome; MI, myocardial infarction.

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

4

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

Table 2 Summarizes the accuracy measures, adjusted variables, timing of measurement, and cutoff values of the BNP, and NT-proBNP tests Author, year, country

Sensitivity, specificity, OR

Timing of Adjusted variables measurement

Hutfless et al, 2004, USA [5]

90% 50% NA 84% 89% NA 100% 83% 76.3 (8.8-661.8) 100%

Preoperative

NA

385

BNP

Postoperative

NA

491

NT-proBNP

Preoperative

ASA grade, preoperative cardiac impairment, age

450

NT-proBNP

Preoperative

DM, CAD, CHF, type of surgery, hypertension, dyslipidemia, smoking, aortic stenosis, arrhythmia, LVH on echo, LVEF, A:E ratio, drug use, Goldman index, signs of chronic liver disease, family history of CAD, sex, age

189

BNP

Preoperative

DM, renal failure, DSE WM abnormalities (both new and at rest)

533

NT-proBNP

Postoperative

NA

100

BNP

Preoperative

CAD, CHF, CVA/TIA, renal impairment, type of surgery, sex, β-blocker, and statin use

108.5

BNP

Preoperative

RCRI, ASA scoring system

170

BNP

Preoperative

RCRI;a age and prior use of cardiac medication in a second model

40

BNP

Kerbaul et al, 2004, France [6] Yeh et al, 2005,Taiwan [7] Dernellis & Panaretou, 2006, Greece [9]

75%

Feringa et al, 2006, Netherlands [10] Berry et al, 2006, UK [8]

Gibson et al, 2007, UK [13]

Cuthbertson et al, 2007, UK [12] Cuthbertson et al, 2007, UK [11] Mahla et al, 2007, Germany [14] Yun et al, 2008, Korea [17] Leibowitz et al, 2008, Israel [15] Goei et al, 2009, The Netherlands [19] Rajagopalan et al, 2008, UK [16] Oscarsson et al, 2009, Sweden [21] Choi et al, 2010, USA [23] Bolliger et al, 2009, Switzerland [18] Nozohoor et al, 2009, Sweden [20] Schutt et al, 2009, USA [22] Villacorta et al, 2010, Brazil [25] Chong et al, 2010, Australia [24] Nojiri et al, 2011, Japan [26] Chong et al, 2012, Australia [29] Yang et al, 2012, Korea [31]

34.52 (17.168.62) 85% 91% 17.2 (2.8-106.4) 100% 90% NA 87% 87% 104.0 (20.0540.0) 82% 79% 13.6 (1.9-97.8) 75% 70% 7.5 (1.9-29.4) 73% 71% 5.34 (1.04-27.5) 80% 81% 7.6 (2.2-26.6) 100% 70% NA 71% 72% 4.78 (2.71-8.42) 71% 65% 3.40 (1.27-9.09) 87% 51% NA NA NA 3.89 (3.15-4.14) NA NA 6.5 (1.4-29.5) 90% 40% 5.9 (1.7-20) 76% 75% 10.5 (1.9-56.6) 76% 79% 3.58 (1.74-4.06) 90% 74% 24.4 (5.1-117.7) 79% 83% 108 (1.02-1.14) 75% 74.5% 4.5 (1.3-15.2) 56% 85%

Cutoff (pg/mL)

280 860 CAD, preoperative creatinine, high-risk surgery, age, preoperative fibrinogen, preoperative high sensitivity C-reactive protein, duration of surgery, surgical complications 201 Preoperative History of CAD, history of CHF, intermediate-risk surgery, RCRI, left atrial enlargement, hemoglobin level, LVEF, age, operation time 60 min, diastolic dysfunction, atrial fibrillation, transfusion, regional wall motion abnormality Postoperative NA 165

Postoperative

BNP, NTproBNP NT-proBNP

BNP

Preoperative

age ≥70, angina pectoris, MI, heart failure, stroke, DM, renal dysfunction, site of surgery, and type of surgery

350

NT-proBNP

Preoperative

Age, RCRI

308

NT-proBNP

Postoperative

NA

2017

NT-proBNP

Preoperative

age, sex, and traditional clinical risk factors

301

NT-proBNP

Preoperative

Age, sex, and the RCRI

50

BNP

Preoperative

NA

82

BNP

Preoperative

NA

457

NT-proBNP

Preoperative

Age, BNP

60

BNP

Postoperative

Age, postoperative troponin increase, preoperative NT-proBNP, premorbid congestive cardiac failure, and preoperative creatinine

1401

NT-proBNP

Preoperative

Sex, smoking, hypertension, BNP, FEV

30

BNP

Preoperative

Premorbid ischemic heart disease and preoperative ACE-2 level

741

NT-proBNP

Preoperative

NA

302

NT-proBNP

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

5

Table 2 (continued) Author, year, country

Bryce et al, 2012, UK [28] Biccard et al, 2012, South Africa [27] Mercantini et al, 2012, Italy [30]

Sensitivity, specificity, OR 4.5 (2.3-8.7) 88% 89% NA NA NA 4.6 (2.1-10) 80.6% 67.2% 4.65 (1.64-13.2)

Timing of Adjusted variables measurement

Cutoff (pg/mL)

Preoperative

NA

99.5

BNP

Preoperative

NA

48.1

BNP

Preoperative

Male sex; age, ≥69 y; BMI, ≥25; RCRI, 2-3 Creatinine, N0.9; Pa max; CAD; ASA status, N2

36

BNP

ASA indicates American Society of Anesthesiologists; ACE, angiotensin-converting enzyme; A:E ratio; atrial flow velocity: early transmittal flow velocity; BMI, body mass index; CAD, coronary artery disease (including past MI); CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; DM, diabetes mellitus; DSE, dobutamine stress echocardiography; ECG, electrocardiography; FEV, forced expiratory volume; LVEF, left ventricular ejection fraction; LVH, left ventricular hypertrophy; NA, not available, PCI, percutaneous coronary intervention; RCRI, Revised Cardiac Risk Index; TIA, transient ischemic attack; WM, wall motion; Pa Max, maximal arterial pressure. a Variables in the Revised Cardiac Risk Index include high-risk type of surgery; ischemic heart disease; history of congestive heart failure or cerebrovascular disease; insulin therapy for diabetes mellitus; and preoperative serum creatinine, 2.0 mg/dL.

diagnostic meta-analysis [37]. The bivariate approach assumes a bivariate distribution for the log-transformed sensitivity and specificity. In addition to accounting for study size, the bivariate model estimates and adjusts for the negative correlation between the sensitivity and specificity of the index test that may arise from the different thresholds used in different studies. To compare diagnostic performance between 2 biomarkers, we calculated the area under the summary receiver operating characteristic curve and DOR as a mechanism to summarize the true- and false-positive rates of different diagnostic studies, irrespective of the different cutoff points used in various studies. To deal with zero observations in 2 × 2 contingency tables, one-half was added to each cell, reducing performance in the small studies. Overall, sensitivity and specificity and their 95% confidence intervals (CIs) were calculated based on the binominal distributions of the true-positives and true-negatives. To quantify the extent of between-study variation (ie, heterogeneity), we calculated the inconsistency index (I 2), which represents the proportion of heterogeneity not explained by random variation. Statistically significant heterogeneity was considered present at I 2 more than 50% [38]. Summary diagnostic ORs were estimated by random (DerSimonian-Laird) or fixed (Mantel-Haenszel) effect models depending on whether I 2 was greater or lower than 50% [33,39]. We defined a priori the following clinical and design characteristics of a study as potential relevant covariates: use of BNP or NT-proBNP tests, the type of surgery, the age range, and the timing of the NP measurement. Statistical analyses were conducted using STATA 11.0 (Stata Corp, College Station, TX). All statistical tests were 2-sided, and statistical significance was defined as a P b .05.

participants, settings, and outcomes of the studies selected for meta-analyses are summarized in Table 1. The sensitivity, specificity, and DOR for the BNP and NT-proBNP tests and the timing of biomarker measurement are summarized in Table 2. The sensitivity

3. Results The initial search yielded 662 hits form MEDLINE and Embase. After screening of title and abstract, 611 citations were excluded. Fifty-one potentially related articles were retrieved for full-text review. Finally, a total of 27 primary studies met the inclusion criteria, which included 5438 patients undergoing major operations whose BNP or NT-proBNP levels were measured in the perioperative period (Fig. 1). Seven hundred twelve patients (13.1%) developed cardiac complications. Of these 27 studies, 24 studies provided sufficient data for reconstructing a 2 × 2 contingency table. Twelve studies measured BNP, and the other 12 studies measured NT-proBNP. Fourteen studies included patients undergoing cardiac valve surgery or coronary artery bypass surgery, whereas the remaining 10 included patients undergoing other type of surgery. The study settings included inpatient unit patients (23 studies) intensive care unit (ICU) patients (2 studies), and emergency department patients (2 studies). Details of the

Fig. 2. An overall picture of the methodological quality of studies as evaluated by QUADAS.

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

In this meta-analysis consisting of 5438 patients, we showed that elevated plasma BNP or NT-proBNP levels obtained in the perioperative period were a strong independent predictor of the occurrence of postoperative MACE. B-type natriuretic peptide levels appeared to have better predictive power than NT-proBNP levels. Similarly, postoperative assessment had better predictive power than preoperative assessment. Type of surgery and age did not appear to greatly affect the predicative

.002 .014 .001 .000 (41.1-80.9) (0.00-69.1) (24.5-77.8) (37.0-76.7) 66.4 35.3 59.1 61.7 (8.43-22.8) (8.84-25.5) (8.62-24.8) (10.6-25.5) 13.9 15.0 14.6 16.4 (0.83-0.89) (0.80-0.87) (0.84-0.90) (0.85-0.90) 0.87 0.84 0.87 0.88 (0.17-0.34) (0.10-0.30) (0.12-0.29) (0.14-0.29) 0.24 0.18 0.18 0.20 (2.64-5.15) (2.64-3.89) (2.50-4.26) (3.03-4.75) 3.68 3.20 3.26 3.79 (0.70-0.84) (0.68-0.78) (0.66-0.80) (0.72-0.82)

(33.7-73.2) (7.14-75.1) (11.0-75.9) (33.6-76.3) (0.00-82.2) 57.8 51.9 53.7 60.4 55.8 (9.87-20.7) (7.02-16.4) (10.7-35.2) (8.89-20.5) (7.68-46.3) 14.3 10.7 19.4 13.6 18.9 0.21 0.29 0.15 0.22 0.18 (2.83-4.29) (2.61-4.46) (2.54-4.77) (2.77-4.45) (2.18-5.23) 3.49 3.41 3.48 3.51 3.38 (0.71-0.81) (0.70-0.84) (0.66-0.82) (0.70-0.82) (0.62-0.84) 0.76 0.77 0.75 0.76 0.75

0.81 0.87 0.87 0.85 14 10 13 19

(0.79-0.88) (0.70-0.84) (0.83-0.93) (0.76-0.88) (0.77-0.92) 0.84 0.78 0.89 0.83 0.86 24 12 12 18 6

Overall [5–17,19–22,24–26,28–31] NT-proBNP [6,7,10,14,16,17,19,21,22,24,29,31] BNP [5,8,9,11–13,15,20,25,26,28,30] Preoperative overall [5,7,9–13,16,17,19,20,22,25,26,28–31] Postoperative [6,8,14,15,21,24] Type of operation Cardiac surgery [5–8,10,12–14,16,17,19,20,28,31] Noncardiac surgery [9,11,15,21,22,24–26,29,30] Elderly population (mean age N70) [8,9,11,14,15,19–21,24–26,28,29] Death outcome [5–15,17,19,21,28,29,31]

4. Discussion

No. of studies

Comparing the 2 subtypes of NPs, BNP was associated with a superior DOR to NT-proBNP. Postoperative measurement of NPs (pooled DOR, 18.9; 95% CI, 7.68-46.3) was associated with a higher diagnostic value than preoperative measurement (pooled DOR, 13.6; 95% CI, 8.89-20.5). The type of surgery, either cardiac or noncardiac surgery, did not appear to affect the predicative accuracy of NPs. The pooled DOR was 13.9 (95% CI, 8.43-22.8) for patients undergoing cardiac surgery and 15.0 (95% CI, 8.84-25.5) for patients undergoing other noncardiac surgery. Thirteen studies studied on elderly population. The pooled accuracy estimates in these studies (DOR, 14.6; 95% CI, 8.62-24.8) was similar to the overall effect estimates (DOR, 14.3; 95% CI, 9.87-20.7). None of the subgroup analyses were significantly different from the overall analysis by meta-regression analysis.

Table 3 Summary of subgroup analysis of the included studies by different study characteristics

3.3. Subgroup analysis

Sensitivity (95% CI)

Specificity (95% CI)

Likelihood ratio+

LR−

(0.16-0.28) (0.21-0.39) (0.10-0.24) (0.16-0.31) (0.10-0.33)

We identified a total of 24 studies reporting estimated sensitivity and specificity of NPs in predicting postoperative cardiac complications. Overall, the pooled sensitivity was 84% (95% CI, 79%-88%) and specificity was 76% (95% CI, 71%-81%) (Table 3). The pooled positive likelihood ratio was 3.49 (95% CI, 1.87-7.26), and the negative likelihood ratio (LR −) was 0.21 (95%CI, 0.16-0.28). Twelve studies evaluated NT-proBNP, and the other 12 evaluated BNP. Detailed pooled effect estimates of 24 studies reporting sensitivity and specificity are summarized in Table 3. Fig. 3 shows the summary receiver operating characteristic (ROC) curve for overall studies, NTproBNP, and BNP, respectively. The area under the ROC curve was 0.87 (0.84-0.90) for overall studies, 0.84 (0.81-0.87) for NT-proBNP, and 0.90 (0.87-0.92) for BNP. The DOR for overall studies, NT-proBNP, and BNP were 14.3 (9.87-20.7), 10.7 (7.02-16.4), and 19.4 (10.7-35.2), respectively. The forest plots for these 3 summary estimates were shown in Fig. 4. We identified a total of 21 studies reporting the adjusted DOR. After adjustment of other clinical predictors, the adjusted DOR attenuated (adjusted DOR, 7.37; 95% CI, 4.41-12.3).

0.87 0.73 0.74 0.78

3.2. Results of individual studies

(0.74-0.87) (0.79-0.93) (0.79-0.92) (0.78-0.90)

AUROC (95% CI)

Diagnostic OR (95% CI)

I2 (95% CI)

Of the included studies, all studies collected data prospectively and provided clear descriptions of patient selection criteria and index tests. Most studies used independent reference tests to verify cardiac complications with little possibility of differential outcome ascertainment. None of the included studies provided an explanation for uninterpretable test results or participants’ withdrawal from the study. None of the studies specified whether they were blinded to index testing in ascertaining outcome. Incorporation bias is likely because the diagnosis of cardiac complications, especially cardiac failure, may have been influenced by the knowledge of serum levels of NPs. Fig. 2 provides an overall picture of the methodological quality of studies as evaluated by the quality assessment of diagnostic accuracy studies tool.

(0.84-0.90) (0.81-0.87) (0.87-0.92) (0.83-0.89) (0.86-0.91)

3.1. Study characteristics

0.87 0.84 0.90 0.87 0.89

Publication bias (Egger test P)

of NPs in predicting MACE ranged from 56% to 100%, the specificities ranged from 40% to 91%, and the DOR ranged from 3.4 to 108.

.000 .001 .077 .001 .019

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

Variables

6

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

7

Fig. 3. Shows the summary ROC curves for overall studies (3a), NT-proBNP (3b), and BNP (3c), respectively.

accuracy in our analysis. After adjustment for multiple clinical predictors, serum level of NPs still showed strong association with the adverse cardiac outcomes. Therefore, we conclude that measurement of BNPs may provide important prognostic information beyond clinical evaluation for risk stratifying patients undergoing major surgery. B-type natriuretic peptide is synthesized and released by myocardium through cyclic guanosine monophosphate–mediated pathway in response to volume or pressure overload. With its natriuretic and vasodilator properties, BNP plays a major role in maintaining stable hemodynamic [4]. Perioperative serum NP levels may reflect the magnitude of volume and pressure change associated with blood loss and catecholamine surge in major surgery and therefore can be used to predict the postoperative cardiac outcomes. There is also evidence that NT-proBNP or BNP may be released from injured myocardium regardless of hemodynamic change [32]. Previous meta-analyses showed high accuracy of NP testing in predicting postoperative MACE. The DOR in predicting postoperative MACE in these reports ranged from 17.37 to 19.77 [32,33,40]. These reports reported the pooled DOR but did not report the measures of sensitivity or specificity. The DOR is a single indicator measure of the accuracy of a diagnostic test. It describes the odds of positive test results in individuals with the disease compared with the odds of positive results in those without disease and corresponds to particular pairings of sensitivity and specificity. Although DOR provide clinicians with a sense of global accuracy for a diagnostic test, it does not provide critical information for clinical decision. Clinicians may need individualized posttest probability for major medical decision. By applying the bivariate random effect model, our analysis provides the pooled measures of likelihood ratio, which can be used for calculation of posttest probability. The pooled positive and LR − for NP testing was 3.49 (95% CI, 2.83-4.29) and 0.21 (0.160.28), respectively. In a population with a 13.1% prevalence of postoperative MACE, a positive result of NP testing translates into a posttest probability of 34%, whereas a negative result of NP testing translates into a 3% of posttest probability for developing postoperative MACE. Compared with NT-proBNP, BNP has even lower LR − (0.15; 95% CI, 0.10-0.24) and can further decrease the posttest probability to 2%. These calculations show that the value of NP testing in relies with its capability to rule out rather than rule in postoperative MACE. In other words, NP testing may not identify high-risk patients, but it can early identify low-risk patients

accurately. Several antiarrhythmic agents such as β-blockers, amiodarone, and magnesium sulfate have been shown to prevent postoperative arrhythmias. However, this medication carries risk for iatrogenic adverse events. Low serum levels of NPs may help physician to select a group of patients that preventive use of antiarrthymics can be avoided. It can also help physicians to select a group of patients who can be safely discharged from ICU at an earlier stage. Recent individual patient data analyses have reported related predictive accuracy measures but could not include as many patients as done in this aggregate data meta-analysis [35]. Therefore, the precision of the effect estimate, subgroup analysis, and potential publication bias will be an issue in these individual patient data meta-analyses. To enhance the value of NP testing in identifying high-risk patients, further efforts to combine NP testing and the established risk factors for postoperative MACE, such as age, sex, cardiovascular disease history, and echocardiographic characteristics in the form of risk-prediction scoring system or algorithm may further enhance its performance in predicting patients at high risk for postoperative MACE [41]. Choi et al [23] combined NT-proBNP and the Revised Cardiac Risk Index and showed improved accuracy in risk prediction. The main strength of this meta-analysis is that it provides clinical relevant summary measures other than DOR and highlights the high negative predictive value but low positive predictive value for NP testing in predicting postoperative MACE. However, several limitations need to be addressed. First, a high degree of heterogeneity was noted for the overall and subgroup analyses. Potential source of heterogeneity included study settings, case mix, and cutoff levels for NP tests. High heterogeneity among the included studies may limit the generalizability of the studies. The studies we evaluated used different threshold value for BNP and NT-proBNP assays to represent an abnormal value. Because this study was not an individual patients’ data meta-analysis, it was not possible to determine an optimal cutoff value across studies. We listed the cutoff values used in individual studies as a reference information. Finally, significant publication bias was noted in our analysis. Although we performed a comprehensive literature without language restriction and include the largest number of studies up to date, we cannot excluded the possibility that the accuracy estimates was overevaluated due to the favorable publication of studies reporting positive results [42].

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

8

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

9

Fig. 4. Shows the forest plots for overall studies (4a), NT-proBNP (4b), and BNP (4c), respectively.

5. Conclusions Preoperative NP levels predict postoperative atrial fibrillation in patients undergoing cardiothoracic surgery. B-type natriuretic peptide levels could be used to better stratify patients in this respect. In addition, BNP activation should direct the attention of clinicians toward monitoring cardiac rhythm and hemodynamic variables more closely than as might be suggested by a cursory evaluation.

References [1] Biccard BM, Lurati Buse GA, Burkhart C, et al. The influence of clinical risk factors on pre-operative B-type natriuretic peptide risk stratification of vascular surgical patients. Anaesthesia 2012;67(1):55–9. [2] Ford MK, Beattie WS, Wijeysundera DN. Systematic review: prediction of perioperative cardiac complications and mortality by the revised cardiac risk index. Ann Intern Med 2010;152(1):26–35. [3] Wijeysundera DN, Beattie WS, Austin PC, Hux JE, Laupacis A. Non-invasive cardiac stress testing before elective major non-cardiac surgery: population based cohort study. BMJ 2010;340:b5526. [4] Gopal DJ, Iqbal MN, Maisel A. Updating the role of natriuretic peptide levels in cardiovascular disease. Postgrad Med 2011;123(6):102–13. [5] Hutfless R, Kazanegra R, Madani M, et al. Utility of B-type natriuretic peptide in predicting postoperative complications and outcomes in patients undergoing heart surgery. J Am Coll Cardiol 2004;43(10):1873–9. [6] Kerbaul F, Collart F, Giorgi R, et al. Increased plasma levels of pro-brain natriuretic peptide in patients with cardiovascular complications following off-pump coronary artery surgery. Intensive Care Med 2004;30(9):1799–806 [Epub 2004 Apr 1727]. [7] Yeh HM, Lau HP, Lin JM, Sun WZ, Wang MJ, Lai LP. Preoperative plasma N-terminal pro-brain natriuretic peptide as a marker of cardiac risk in patients undergoing elective non-cardiac surgery. Br J Surg 2005;92(8):1041–5. [8] Berry C, Kingsmore D, Gibson S, et al. Predictive value of plasma brain natriuretic peptide for cardiac outcome after vascular surgery. Heart 2006;92(3):401–2. [9] Dernellis J, Panaretou M. Assessment of cardiac risk before non-cardiac surgery: brain natriuretic peptide in 1590 patients. Heart 2006;92(11):1645–50 [Epub 2006 May 1618]. [10] Feringa HH, Bax JJ, Elhendy A, et al. Association of plasma N-terminal pro-B-type natriuretic peptide with postoperative cardiac events in patients undergoing surgery for abdominal aortic aneurysm or leg bypass. Am J Cardiol 2006;98 (1):111–5 [Epub 2006 May 2006]. [11] Cuthbertson BH, Amiri AR, Croal BL, et al. Utility of B-type natriuretic peptide in predicting perioperative cardiac events in patients undergoing major non-cardiac surgery. Br J Anaesth 2007;99(2):170–6 [Epub 2007 Jun 2015].

[12] Cuthbertson BH, Card G, Croal BL, McNeilly J, Hillis GS. The utility of B-type natriuretic peptide in predicting postoperative cardiac events and mortality in patients undergoing major emergency non-cardiac surgery. Anaesthesia 2007;62 (9):875–81. [13] Gibson SC, Payne CJ, Byrne DS, Berry C, Dargie HJ, Kingsmore DB. B-type natriuretic peptide predicts cardiac morbidity and mortality after major surgery. Br J Surg 2007;94(7):903–9. [14] Mahla E, Baumann A, Rehak P, et al. N-terminal pro-brain natriuretic peptide identifies patients at high risk for adverse cardiac outcome after vascular surgery. Anesthesiology 2007;106(6):1088–95. [15] Leibowitz D, Planer D, Rott D, Elitzur Y, Chajek-Shaul T, Weiss AT. Brain natriuretic peptide levels predict perioperative events in cardiac patients undergoing noncardiac surgery: a prospective study. Cardiology 2008;110(4):266–70 [Epub 2007 Dec 2012]. [16] Rajagopalan S, Croal BL, Bachoo P, Hillis GS, Cuthbertson BH, Brittenden J. N-terminal pro B-type natriuretic peptide is an independent predictor of postoperative myocardial injury in patients undergoing major vascular surgery. J Vasc Surg 2008;48(4):912–7 [discussion 917. doi: 910.1016/j.jvs.2008.1005.1015. Epub 2008 Jun 1030]. [17] Yun KH, Jeong MH, Oh SK, et al. Preoperative plasma N-terminal pro-brain natriuretic peptide concentration and perioperative cardiovascular risk in elderly patients. Circ J 2008;72(2):195–9. [18] Bolliger D, Seeberger MD, Lurati Buse GA, et al. A preliminary report on the prognostic significance of preoperative brain natriuretic peptide and postoperative cardiac troponin in patients undergoing major vascular surgery. Anesth Analg 2009;108 (4):1069–75. http://dx.doi.org/10.1213/ane.1060b1013e318194f318193e318196. [19] Goei D, Hoeks SE, Boersma E, et al. Incremental value of high-sensitivity C-reactive protein and N-terminal pro-B-type natriuretic peptide for the prediction of postoperative cardiac events in noncardiac vascular surgery patients. Coron Artery Dis 2009;20 (3):219–24. http://dx.doi.org/10.1097/MCA.1090b1013e3283219e3283247. [20] Nozohoor S, Nilsson J, Luhrs C, Roijer A, Algotsson L, Sjogren J. B-type natriuretic peptide as a predictor of postoperative heart failure after aortic valve replacement. J Cardiothorac Vasc Anesth 2009;23(2):161–5. http://dx.doi.org/10.1053/j. jvca.2008.1011.1006 [Epub 2009 Jan 1023]. [21] Oscarsson A, Fredrikson M, Sorliden M, Anskar S, Eintrei C. N-terminal fragment of proB-type natriuretic peptide is a predictor of cardiac events in high-risk patients undergoing acute hip fracture surgery. Br J Anaesth 2009;103(2):206–12. http://dx.doi. org/10.1093/bja/aep1139 [Epub 2009 Jun 1013]. [22] Schutt RC, Cevik C, Phy MP. Plasma N-terminal prohormone brain natriuretic peptide as a marker for postoperative cardiac events in high-risk patients undergoing noncardiac surgery. Am J Cardiol 2009;104(1):137–40. http://dx.doi. org/10.1016/j.amjcard.2009.1003.1005 [Epub 2009 May 1013]. [23] Choi JH, Cho DK, Song YB, et al. Preoperative NT-proBNP and CRP predict perioperative major cardiovascular events in non-cardiac surgery. Heart 2010;96(1):56–62. http://dx. doi.org/10.1136/hrt.2009.181388 [Epub 182009 Oct 181326]. [24] Chong CP, Ryan JE, van Gaal WJ, et al. Usefulness of N-terminal pro-brain natriuretic peptide to predict postoperative cardiac complications and long-term mortality after emergency lower limb orthopedic surgery. Am J Cardiol 2010;106:865–72. http://dx.doi.org/10.1016/j.amjcard.2010.1005.1012.

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022

10

Y-R. Young et al. / Journal of Critical Care xxx (2014) xxx–xxx

[25] Villacorta Junior H, Castro IS, Godinho M, et al. B-type natriuretic peptide is predictive of postoperative events in orthopedic surgery. Arq Bras Cardiol 2010;95 (6):743–8 [Epub 2010 Sep 2024]. [26] Nojiri T, Inoue M, Yamamoto K, et al. B-type natriuretic peptide as a predictor of postoperative cardiopulmonary complications in elderly patients undergoing pulmonary resection for lung cancer. Ann Thorac Surg 2011;92(3):1051–5. http:// dx.doi.org/10.1016/j.athoracsur.2011.1003.1085. [27] Biccard BM, Naidoo P, de Vasconcellos K. What is the best pre-operative risk stratification tool for major adverse cardiac events following elective vascular surgery? A prospective observational cohort study evaluating pre-operative myocardial ischaemia monitoring and biomarker analysis. Anaesthesia 2012;67 (4):389–95. http://dx.doi.org/10.1111/j.1365-2044.2011.07020.x [Epub 02012 Feb 07011]. [28] Bryce GJ, Payne CJ, Gibson SC, et al. B-type natriuretic peptide predicts postoperative cardiac events and mortality after elective open abdominal aortic aneurysm repair in the short term, intermediate term, and long term. J Vasc Surg 2012;8(12):345–53 [01746-01746]. [29] Chong CP, Lim WK, Velkoska E, et al. N-terminal pro-brain natriuretic peptide and angiotensin-converting enzyme-2 levels and their association with postoperative cardiac complications after emergency orthopedic surgery. Am J Cardiol 2012;109 (9):1365–73. http://dx.doi.org/10.1016/j.amjcard.2011.1312.1032 [Epub 2012 Feb 1328]. [30] Mercantini P, Di Somma S, Magrini L, et al. Preoperative brain natriuretic peptide (BNP) is a better predictor of adverse cardiac events compared to preoperative scoring system in patients who underwent abdominal surgery. World J Surg 2012;36(1):24–30. http://dx.doi.org/10.1007/s00268-00011-01354-00266. [31] Yang JH, Choi JH, Ki YW, et al. Plasma N-terminal pro-B-type natriuretic peptide is predictive of perioperative cardiac events in patients undergoing vascular surgery. Korean J Intern Med 2012;27(3):301–10 [Epub 2012 Sep 2011].

[32] Sabatine MS, Morrow DA, de Lemos JA, et al. Acute changes in circulating natriuretic peptide levels in relation to myocardial ischemia. J Am Coll Cardiol 2004;44(10):1988–95. [33] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7 (3):177–88. [34] Akobeng AK. Understanding diagnostic tests 2: likelihood ratios, pre- and post-test probabilities and their use in clinical practice. Acta Paediatr 2007;96(4):487–91. [35] Rodseth RN, Biccard BM, Chu R, et al. Postoperative B-type natriuretic peptide for prediction of major cardiac events in patients undergoing noncardiac surgery: systematic review and individual patient meta-analysis. Anesthesiology 2013;119(2):270–83. [36] Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003;3:25. [37] Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005;58(10):982–90. [38] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in metaanalyses. BMJ 2003;327(7414):557–60. [39] L’Abbe KA, Detsky AS, O’Rourke K. Meta-analysis in clinical research. Ann Intern Med 1987;107(2):224–33. [40] Harbord RM, Whiting P, Sterne JA, et al. An empirical comparison of methods for meta-analysis of diagnostic accuracy showed hierarchical models are necessary. J Clin Epidemiol 2008;61(11):1095–103. [41] Rodseth RN, Padayachee L, Biccard BM. A meta-analysis of the utility of preoperative brain natriuretic peptide in predicting early and intermediate-term mortality and major adverse cardiac events in vascular surgical patients. Anaesthesia 2008;63(11):1226–33. [42] Song F, Khan KS, Dinnes J, Sutton AJ. Asymmetric funnel plots and publication bias in meta-analyses of diagnostic accuracy. Int J Epidemiol 2002;31(1):88–95.

Please cite this article as: Young Y-R, et al, Predictive value of plasma brain natriuretic peptide for postoperative cardiac complications—A systemic review and meta-analysis, J Crit Care (2014), http://dx.doi.org/10.1016/j.jcrc.2014.03.022