International Journal of Cardiology 221 (2016) 794–799
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
The prognostic impact of chest pain in 1306 patients presenting with confirmed acute pulmonary embolism Christopher C.Y. Wong a,1, Austin C.C. Ng a,b,⁎,1, Jerrett K. Lau a,b,1, Vincent Chow a,b,1, Andrew P. Sindone a,1, Leonard Kritharides a,b,1 a b
Department of Cardiology, Concord Hospital, University of Sydney, Australia Vascular Biology Group, ANZAC Research Institute, Australia
a r t i c l e
i n f o
Article history: Received 21 May 2016 Accepted 8 July 2016 Available online 11 July 2016 Keywords: Pulmonary embolism Prognosis Chest pain
a b s t r a c t Background: The prognostic influence of chest pain in patients presenting with pulmonary embolism has not been well defined. We investigated whether the presence of chest pain at presentation affected the mortality of patients with acute pulmonary embolism. Methods: Retrospective cohort study of consecutive patients admitted to a tertiary hospital with confirmed acute pulmonary embolism from 2000 to 2012, with study outcomes tracked using a state-wide death registry. Results: Of the 1306 patients included in the study, 771 (59%) had chest pain at presentation. These patients were younger with fewer comorbidities, and had lower 6-month mortality compared to patients without chest pain (5% vs 15%, P b 0.001). Chest pain was consistently found to be an independent predictor of 6-month mortality in three separate multivariable models (range of hazard ratios 0.52–0.60, all with P b 0.05). The addition of chest pain to a multivariable model that included the simplified pulmonary embolism severity index, haemoglobin, and sodium led to a significant net reclassification improvement of 18% (P b 0.001). Conclusions: Chest pain is a novel, favourable prognostic marker in patients with acute pulmonary embolism. © 2016 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Despite advances in diagnosis and treatment, pulmonary embolism (PE) remains a deadly condition with only a minimal improvement in age-standardized death rates over the last decade [1]. The importance of prognostic assessment in these patients has been emphasized in major society guidelines [2], with recommendations for utilizing either the original pulmonary embolism severity index (PESI) [3] or the simplified pulmonary embolism severity index (sPESI) [4] to stratify risk. Both indices have been shown to predict mortality at three and six months post PE [5]. The economic burden of acute PE treatment and venous thromboembolism in general has escalated in recent decades, with hospitalization costs identified as the primary costs driver [6]. Contemporary new therapeutic approaches to the management of acute PE may allow more efficient models of care by incorporating outpatient anticoagulation for patients with smaller, lower risk PE [7]. The safe application of these strategies will require the availability of reliable risk stratification tools, and in particular the accurate recognition of low risk patients.
Traditionally, the clinical presentation of PE has been divided into the three distinct syndromes of circulatory collapse, isolated dyspnoea, and pulmonary infarction [8]. Pulmonary infarction is characterized by the presence of chest pain or haemoptysis or both, and is associated with lesser clot burden on pulmonary angiography [9,10]. It has been hypothesised that occlusion of smaller, peripheral pulmonary arteries more frequently lead to pulmonary infarction due to inability of the smaller pulmonary arterial bed to accommodate systemic arterial inflow from bronchial anastomoses, resulting in extravasation of red blood cells into the alveoli and subsequent infarction [10]. However, whether pulmonary infarction carries independent prognostic value has been disputed amongst different studies [11,12]. More specifically, the prognostic significance of the symptom of chest pain in acute PE has never been defined. The primary aim of this study was to delineate the prognostic value of chest pain in patients with acute PE. The secondary aim was to determine whether its inclusion in a validated risk model improves the prediction of mortality in these patients. 2. Methods
⁎ Corresponding author at: Cardiology Department, Concord Hospital, The University of Sydney, Hospital Road, Concord 2139, NSW, Australia. E-mail address:
[email protected] (A.C.C. Ng). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
http://dx.doi.org/10.1016/j.ijcard.2016.07.129 0167-5273/© 2016 Elsevier Ireland Ltd. All rights reserved.
2.1. Study population The study database consists of consecutive patients admitted to a tertiary institution (Concord Hospital, The University of Sydney) with confirmed PE and has been described
C.C.Y. Wong et al. / International Journal of Cardiology 221 (2016) 794–799 previously [13–16]. All patients' medical records were reviewed to confirm the diagnosis of acute PE, which required a documented clinical diagnosis in conjunction with a positive imaging study, as defined according to published guidelines [2]. Patients that presented with more than one episode of PE during the study period only had their initial presentation included. All patients who were not residents of the state of New South Wales during admission were excluded to minimize incomplete tracking of outcomes. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Concord Hospital Human Research Ethics Committee (CH62/612015-06). The Ethics Committee also granted a waiver of the usual requirement for consent of the individual to the use of their health information. All patients' data were de-identified and analysed anonymously. 2.2. Data sources The authors A.C.C.N. and V.C. were responsible for extracting the relevant data directly from each patient's medical records. Comorbidities based on the International Classification of Diseases 10th revision coding were recorded. Each patient had their sPESI and Charlson Comorbidity Index (CCI) scores calculated from the extracted data. The sPESI score was calculated by assigning a point to the following: age N 80, heart rate ≥ 110 beats/min, systolic BP b 100 mm Hg, oxyhaemoglobin saturation b 90%, history of cancer, and history of cardiopulmonary disease. The CCI score was calculated by assigning one point for a history of myocardial infarction, heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, peptic ulcer disease, mild liver disease, and diabetes without end-organ damage; two points for hemiplegia, moderate to severe renal disease, solid organ cancer, lymphoma, leukaemia, diabetes with end organ damage; three points for moderate to severe liver disease; and six points for metastatic malignancy and acquired immunodeficiency syndrome. For patients that underwent transthoracic echocardiography (TTE) during the admission, the estimated right ventricular systolic pressure (RVSP), left ventricular function, right ventricular size and function, right atrial size, and the presence of tricuspid regurgitation were extracted from the final echocardiographic report. The RVSP was derived from the modified Bernoulli equation using the peak velocity of tricuspid regurgitation obtained by continuous wave Doppler in the apical 4-chamber view [17]. All echocardiograms were performed by qualified sonographers and reported by experienced cardiologists. There was no separate attempt to verify the findings in the final echocardiography reports. 2.3. Study outcomes The primary outcome of the study was all-cause mortality at six months, which was tracked using a state-wide death registry database. Secondary outcomes include inhospital mortality, 1-month and 3-month mortality, and cause-specific mortality analyses. A censored date of October 31, 2013 was pre-determined to fulfil the minimum follow-up of six months for each patient. The use of a state-wide death registry to obtain our primary outcome is advantageous as non-captured deaths during this period were estimated to be only 0.6% based on known migration rates [18]. All death certificates were retrieved to ascertain cause-specific mortality outcomes. Each cause of death was coded independently by two authors (A.C.C.N., L.K. or J.K.L) according to the general principles set by the World Health Organization [19]. During the coding process of the death certificates, patients were de-identified before the coding so that researchers were blinded to the patients' background history and presentation. Disparities in cause of death were resolved by consensus.
795
(StataCorp LP, TX, USA). A two-tailed probability value b 0.05 was considered statistically significant.
3. Results 3.1. Baseline characteristics Between January 2000 and December 2012, 1426 patients were admitted to Concord Hospital with a primary diagnosis of acute PE. There were 120 patients with missing documentation of their presenting symptoms necessitating exclusion from the study. The final studied cohort consisted of 1306 patients (Fig. 1). There were 771 patients (59%) with chest pain and 535 patients (41%) with no chest pain at presentation. The baseline characteristics of the two groups are summarized in Table 1. Patients with chest pain were significantly younger, had shorter length of stay, less syncope at presentation, higher oxyhaemoglobin saturation, and a lower mean sPESI score. They also had fewer comorbidities as shown by a lower mean CCI score, with less heart failure, atrial fibrillation/flutter, chronic kidney disease, malignancy, and neurodegenerative disease. Blood tests taken at the time of diagnosis revealed these patients had higher mean estimated glomerular filtration rate and serum haemoglobin when compared to patients without chest pain. There were no significant differences between the two groups in antiplatelet or anticoagulant use prior to admission. 3.2. Unadjusted survival outcomes Fig. 2 shows the unadjusted Kaplan–Meier survival curves of the entire cohort stratified into two groups based on the presence or absence of chest pain. Table 2 summarizes all-cause mortality for both groups at different time-points during follow-up. Patients with chest pain at presentation had significantly reduced mortality compared to patients without chest pain from the first month onwards (2% vs 5%, P = 0.001), with the biggest difference found at the 6-month followup time-point (5% vs 15%, P b 0.001). There was also a trend towards
2.4. Statistical analysis Patients were divided into two groups based on the presence of chest pain at the time of presentation. Unless otherwise stated, all continuous variables were expressed as mean ± standard deviation and all categorical variables were expressed as number of subjects with percentage of total numbers in brackets. Baseline characteristics of the two groups were compared using the independent-samples t test for continuous variables, and Pearson's chi square test for dichotomous variables. Unadjusted survival curves between the two groups were compared using the Kaplan–Meier method. Differences in in-hospital and 1-month mortality between both groups were compared using binary logistic regression analysis, while Cox proportional hazards regression analysis was used to compare 3-month and 6-month mortality. In addition, Cox regression analysis was used to identify predictors of mortality at 6 months and compare adjusted survival curves between the two groups. Univariables with P b 0.05 were included in the multivariable modelling to identify independent predictors of mortality, and separate confirmatory multivariable analyses were performed using the individual components of the sPESI score in addition to analyses using the aggregate scores of sPESI and CCI. To avoid significant colinearity, only univariable predictors with a correlation coefficient ≤ 0.7 were chosen for the multivariable modelling. All covariates (including chest pain) had b10% missing data, and were dealt with using the multiple imputation method with pooled estimates from 20 imputations in a separate sensitivity analysis. We also examined the incremental prognostic performance of adding chest pain to a model with sPESI, haemoglobin, and sodium, by using net reclassification and integrated discrimination improvement [20]. Analysis of covariance (ANCOVA) and binary logistic regression were used to examine the effects of chest pain on echocardiographic outcomes and to adjust for covariates. All analyses were performed using SPSS version 22 (IBM, USA) or Stata version 10.1
Fig. 1. Derivation of study cohort. The flow chart represents the derivation of the study cohort from the original population of 1426 patients.
796
C.C.Y. Wong et al. / International Journal of Cardiology 221 (2016) 794–799
Table 1 Baseline characteristics.
Parameters Age, years Males Concomitant DVT LOS, days Clinical presentation Dyspnoea Syncope Haemodynamic profile at admission Heart rate, bpm Systolic BP, mm Hg Oxyhaemoglobin saturation, % sPESI score Comorbidities CCI score Cardiovascular disease Myocardial infarction Prior CABG or PCI Heart failure AF/Atrial flutter Stroke Valvular heart disease Prosthetic valve Peripheral vascular disease Cardiac risk factors Hypertension Hyperlipidaemia Diabetes Smoker Non-cardiac comorbidities Chronic renal failure Chronic lung disease Malignancy Neurodegenerative disease Blood profile during admission Day-1 sodium, mmol/l Day-1 eGFR, ml/min/1.73 m2 Day-1 haemoglobin, g/l Admission medication use Aspirin Clopidogrel Thrombolysis Warfarin Enoxaparin NOACs
Study cohort (n = 1306)
Chest pain (n = 771)
No chest pain (n = 535)
P value
67 ± 16 569 (43.6%) 269 (21%) 7.9 ± 6.7
64 ± 17 326 (42%) 135 (18%) 7.1 ± 6.0
72 ± 14 243 (45%) 134 (25%) 9.0 ± 7.4
b0.001 0.26 0.001 b0.001
939 (72%) 82 (6%)
539 (70%) 31 (4%)
400 (75%) 51 (10%)
0.06 b0.001
88 ± 21 140 ± 25 95 ± 4 0.9 ± 0.9
88 ± 21 140 ± 23 96 ± 3 0.7 ± 0.8
89 ± 22 140 ± 26 94 ± 5 1.1 ± 1.0
0.26 0.69 b0.001 b0.001
1.5 ± 2.0
1.3 ± 1.8
1.9 ± 2.1
b0.001
208 (16%) 66 (5%) 140 (11%) 176 (14%) 37 (3%) 30 (2%) 12 (1%) 126 (10%)
121 (16%) 37 (5%) 58 (8%) 87 (11%) 17 (2%) 13 (2%) 6 (1%) 71 (9%)
87 (16%) 29 (5%) 82 (15%) 89 (17%) 20 (4%) 17 (3%) 6 (1%) 55 (10%)
0.78 0.61 b0.001 0.005 0.10 0.08 0.52 0.52
322 (25%) 141 (11%) 173 (13%) 335 (26%)
183 (24%) 93 (12%) 102 (13%) 196 (25%)
139 (26%) 48 (9%) 71 (13%) 139 (26%)
0.35 0.08 0.98 0.82
71 (5%) 154 (12%) 251 (19%) 73 (6%)
32 (4%) 82 (11%) 124 (16%) 26 (3%)
39 (7%) 172 (14%) 127 (24%) 47 (9%)
0.01 0.12 0.001 b0.001
139 ± 4 78 ± 32 130 ± 19
139 ± 4 80 ± 32 132 ± 18
139 ± 4 75 ± 33 127 ± 20
0.15 0.003 b0.001
303 (23%) 64 (5%) 6 (1%) 119 (9%) 40 (3%) 1 (0%)
184 (24%) 36 (5%) 4 (1%) 70 (9%) 25 (3%) 1 (0%)
119 (22%) 28 (5%) 2 (0%) 49 (9%) 15 (3%) 0 (0%)
0.50 0.64 0.70 0.97 0.65 0.41
Plus–minus value represents mean ± standard deviation; all others represent numbers of patients with values in brackets representing percentages. AF = atrial fibrillation, BP = blood pressure, CABG = coronary artery bypass grafting, CCI = Charlson Comorbidity Index, DVT = deep vein thrombosis, eGFR = estimated glomerular filtration rate, LOS = length of stay, n = number of patients, Neurodegenerative disease includes dementia and Parkinson's disease, NOACs = new oral anticoagulants including dabigatran, rivaroxaban, and apixaban, PCI = percutaneous coronary intervention, sPESI = simplified Pulmonary Embolism Severity (sPESI score incorporates age, history of malignancy, heart failure/chronic lung disease, heart rate ≥ 110 beats/min, systolic blood pressure b 100 mm Hg and oxyhaemoglobin saturation b 90%).
reduced mortality during the index PE admission for patients with chest pain compared to those without (1% vs 3%, P = 0.06). 3.3. Cause-specific mortality rates Cause-specific mortality for patients at discharge and 6 months is summarized in Table 3. PE accounted for the majority of in-hospital mortality for both groups. At 6 months following the initial event, the leading cause for mortality was malignancy for both groups. Data for 1-month and 3-month cause-specific mortality are presented separately in Supplementary Table 1. 3.4. Chest pain as a prognostic marker Univariable predictors of 6-month mortality for the entire cohort (n = 1306) are shown in Supplementary Table 2. Significant univariable predictors of mortality at 6-months following acute PE include age, chest pain, dyspnoea, heart rate, systolic BP, oxyhaemoglobin saturation, sPESI, CCI, stroke, malignancy, serum sodium, and serum haemoglobin. Three separate multivariable models were subsequently
created to assess the prognostic impact of chest pain on 6-month mortality after acute PE (Table 4 and Supplementary Table 3). In Model 1 incorporating sPESI, chest pain was associated with reduced mortality (hazard ratio [HR] 0.52, 95% confidence interval [CI] 0.34–0.77, P = 0.001) independent of sPESI, dyspnoea, stroke, day-1 serum sodium and haemoglobin (Table 4). In the model without sPESI (Model 2), chest pain was similarly associated with reduced mortality (HR 0.52, 95% CI 0.34–0.79, P = 0.002), this time independent of age, malignancy, heart rate, systolic BP, oxyhaemoglobin saturation, dyspnoea, stroke, day-1 serum sodium and haemoglobin. In the final model with CCI used as an aggregate variable for comorbidities (Model 3), chest pain was associated with reduced mortality (HR 0.60, 95% CI 0.39–0.91, P = 0.02) independent of CCI, age, heart rate, systolic BP, oxyhaemoglobin saturation, dyspnoea, day-1 serum sodium and haemoglobin. Adjusted survival curves based on the three multivariable models are shown in Fig. 3a, b, and c respectively. Patient reclassifications for 6-month mortality are summarized in Table 5. The addition of chest pain to a multivariable model including sPESI, haemoglobin, and sodium led to a statistically significant net reclassification improvement estimated at 18% (P b 0.001), and a
C.C.Y. Wong et al. / International Journal of Cardiology 221 (2016) 794–799
797
Table 3 Cause-specific mortality during admission and at 6 months post PE. In-hospital death
Cardiovascular causes Pulmonary embolism Cardiopulmonary arrest Acute myocardial infarct Heart failure Cardiac-relateda Stroke Non-cardiovascular causes Sepsis Malignancy Other Undefined
6-Month death
Chest pain (n = 9)
No chest pain Chest (n = 14) pain (n = 40)
No chest pain (n = 79)
7 (78%) 0 (0%) 0 (0%)
11 (79%) 0 (0%) 1 (7%)
8 (20%) 0 (0%) 2 (5%)
14 (18%) 2 (3%) 3 (4%)
0 (0%) 1 (11%) 0 (0%)
0 (0%) 0 (0%) 0 (0%)
2 (5%) 3 (8%) 1 (3%)
4 (5%) 1 (1%) 2 (3%)
0 (0%) 1 (11%) 0 (0%) 0 (0%)
1 (7%) 0 (0%) 1 (7%) 0 (0%)
1 (3%) 19 (48%) 4 (10%) 0 (0%)
12 (15%) 33 (42%) 7 (9%) 1 (1%)
a Cardiac-related cause of death is coded when more than one cardiac cause of death is recorded on the death certificate.
3.6. Sensitivity analysis Fig. 2. Unadjusted Kaplan–Meier survival curves of patients with chest pain versus those with no chest pain at the time of presentation. The broken line shows the survival curve of patients who did not have chest pain when they presented with PE, while the unbroken line represents patients who had chest pain when they presented with PE.
statistically significant integrated discrimination improvement (0.013, P b 0.001).
The effect of missing data (b 10% for all covariates) on the results of the study was examined using the multiple imputation method, with pooled estimates from 20 imputations. Chest pain remained a significant predictor of 6-month mortality in multiple models independent of sPESI, CCI, age, heart rate, systolic BP, oxyhaemoglobin saturation, dyspnoea, malignancy, stroke, day-1 serum sodium and haemoglobin (Supplementary Table 5).
3.5. Echocardiography subgroup exploratory analysis 4. Discussion For the entire study cohort, 515 patients (39%) underwent TTE during their admission (Supplementary Table 4): 310 patients (40%) in the chest pain group and 205 (38%) in the group without chest pain. Patients with chest pain had less right ventricular dilatation (29% vs 42%, P = 0.004), impaired right ventricular function (25% vs 40%, P b 0.001), right atrial dilatation (27% vs 37%, P = 0.01), moderate to severe tricuspid regurgitation (10% vs 20%, P = 0.002), and lower mean estimated RVSP (34 ± 12 mm Hg vs 39 ± 15 mm Hg, P = 0.001). There was no significant difference between the groups in the proportion of patients with impaired left ventricular function. Given the increased prevalence of heart failure in patients without chest pain, adjustments were performed using heart failure as a covariate. Chest pain continued to have an independent effect on RVSP (adjusted mean RVSP 34 mm Hg vs 39 mm Hg, P = 0.002), right ventricular dilatation (odds ratio [OR] = 0.60, 95% CI 0.41–0.87, P = 0.007), impaired right ventricular function (OR = 0.59, 95% CI 0.40– 0.88, P = 0.009), and moderate to severe tricuspid regurgitation (OR = 0.59, 95% CI 0.344–0.996, P = 0.048) after adjustment for heart failure. There was a trend observed for right atrial dilatation (OR = 0.68, 95% CI 0.46–1.00, P = 0.05).
Table 2 All-cause mortality after acute PE.
All-cause mortality In-hospital 1-month 3-month 6-month
The current study described the clinical characteristics and outcomes of patients presenting with acute PE and chest pain in a large contemporary PE cohort. The presence of chest pain was found to be independently associated with a favourable prognosis; furthermore, the addition of this easily obtainable variable significantly improved net reclassification of patients. Patients with PE can present with a variety of symptoms. In a cross sectional study of 800 PE patients, chest pain was found to be the second most common symptom, accounting for approximately half of all cases [21]. Our study found similar rates, with 59% of the cohort having chest pain. Patients in our chest pain group were significantly younger and had less comorbidities. In particular, they had less heart failure and malignancy. Assuming that chest pain in patients with PE reflects Table 4 Multivariable models examining the 6-month prognostic impact of chest pain in patients with PE. Multivariable models Model 1 (including sPESI)a Chest pain Model 2 (without sPESI)b Chest pain Model 3 (including CCI)c Chest pain a
Chest pain (n = 771)
No chest pain (n = 535)
P value
9 (1%) 12 (2%) 27 (4%) 40 (5%)
14 (3%) 26 (5%) 58 (11%) 79 (15%)
0.06 0.001 b0.001 b0.001
Hazard ratio
95% confidence interval
P value
0.52
0.34–0.77
0.001
0.52
0.34–0.79
0.002
0.60
0.39–0.91
0.02
Model 1 includes the following variables: chest pain, sPESI (age, history of malignancy, heart failure/chronic lung disease, heart rate ≥ 110 beats/min, systolic blood pressure b 100 mm Hg and oxyhaemoglobin saturation b 90%), dyspnoea, stroke, day-1 serum sodium and haemoglobin. b Model 2, without sPESI, includes the following variables: chest pain, age, history of malignancy, heart rate, systolic blood pressure, oxyhaemoglobin saturation, dyspnoea, stroke, day-1 serum sodium and haemoglobin. c Model 3 includes the following variables: chest pain, Charlson Comorbidity Index (CCI) score, age, heart rate, systolic blood pressure, oxyhaemoglobin saturation, dyspnoea, day-1 serum sodium and haemoglobin.
798
C.C.Y. Wong et al. / International Journal of Cardiology 221 (2016) 794–799
pulmonary infarction, this observation contradicts the traditionally held view that pulmonary infarction occurs more frequently in patients with chronic cardiopulmonary diseases or malignancy [22,23], but supports the results of a more recent study evaluating a contemporary PE population [12].
Clinical symptoms at presentation may correlate with severity of the underlying PE [9,10]. Stein et al. demonstrated that patients presenting with chest pain have significantly less severe PE as assessed by a pulmonary arteriography objective index score when compared to patients with shock, syncope, or isolated dyspnoea [9]. Dalen et al. found that patients with clinical pulmonary infarction (defined as pleuritic chest pain in the presence of a corresponding pulmonary infiltrate on X-ray) had significantly less embolic burden as demonstrated by a higher proportion of unilateral and distal filling defects with less severe obstruction when compared to patients with no pulmonary infarction. An autopsy study supported these clinical observations by finding a strong correlation between pulmonary infarction and occlusion of pulmonary arteries with diameters less than 3 mm [24]. Although only 40% of our study cohort underwent echocardiography during their admission, analysis of this subgroup lends further support to these observations, as patients with chest pain appeared to have significantly less right ventricular dysfunction and lower RVSP on echocardiography. Given the correlation between symptoms and severity in patients with PE, it can be hypothesised that their clinical presentation may also impact on prognosis. Lobo et al. evaluated the prognostic influence of different clinical presentations of PE using data from the Registro Informatizado de la Enfermedad ThromboEmbolica (RIETE), and found that patients with clinically-defined pulmonary infarction (defined as chest pain or haemoptysis) had lower 15-day mortality rates (3%) compared to patients to presenting with isolated dyspnoea (6%) or circulatory collapse (7%) [11]. Ghuysen et al. found similar results in a small study of 82 patients, with the pulmonary infarction group having a strikingly lower in-hospital mortality rate (0%) compared to patients with isolated dyspnoea (14%) or circulatory failure (25%) [25]. On the other hand, Cha et al. investigated outcomes of radiologically-defined pulmonary infarction (defined as radiological evidence of consolidation in the region of the segmental or sub-segmental PE) in 509 patients with confirmed PE and found no significant association between pulmonary infarction and mortality [12]. Unexpectedly, they demonstrated that their pulmonary infarction group had more proximally located PE, which differed from previous studies [9,10]. One reason for the discrepancy in findings between the reported studies may be due to the different definitions of pulmonary infarction. Lobo et al. [11] and Ghuysen et al. [25] used clinical symptoms including chest pain or haemoptysis to define pulmonary infarction and had very high rates of chest pain (96% and 100% respectively) in their study groups. Cha et al. [12] used radiographic features to define pulmonary infarction and had only 27% of patients with chest pain in their study group. Our results reinforce the prognostic relevance of chest pain itself in patients with PE, and suggest that chest pain may be a more accessible and unambiguous prognostic indicator which avoids the variable definitions of pulmonary infarction. Contemporary outpatient approaches to the management of acute PE will require accurate identification patients with lower risk PE. Incorporation of chest pain in the risk stratification of patients with PE may identify patients who are of lower risk and who may be more safely treated in the ambulatory care setting.
4.1. Study limitations Our study has several limitations. The first lies in its single-centre retrospective design, which makes selection biases unavoidable. Fig. 3. Adjusted survival curves of patients with chest pain versus those with no chest pain at the time of presentation. The broken line shows the survival curve of patients who did not have chest pain when they presented with PE, while the unbroken line represents patients who had chest pain when they presented with PE. The curves in Fig. 3a are adjusted for the simplified pulmonary embolism severity index (sPESI), dyspnoea, stroke, day-1 serum sodium and haemoglobin. The curves in Fig. 3b are adjusted for age, heart rate, systolic blood pressure, oxyhaemoglobin saturation, dyspnoea, stroke, malignancy, day-1 serum sodium and haemoglobin. The curves in Fig. 3c are adjusted for Charlson Comorbidity Index, age, heart rate, systolic blood pressure, oxyhaemoglobin saturation, dyspnoea, day-1 serum sodium and haemoglobin.
C.C.Y. Wong et al. / International Journal of Cardiology 221 (2016) 794–799 Table 5 Reclassification of patients who died or who were alive at 6 months post-acute PE based on the presence of chest pain. Established multivariable model without chest paina Patients who died, % b2% 2–6% 6–10% N10% Total no.b Patients who were alive, % b2% 2–6% 6–10% N10% Total no.b
Established multivariable model with chest paina b2%
2–6%
6–10%
N10%
Total no.
0 1 0 0 1
2 8 4 0 14
0 4 12 2 18
0 0 10 69 79
2 13 26 71 112
83 115 0 0 198
38 372 81 0 491
0 54 73 59 186
0 0 55 211 266
121 541 209 270 1141
a The established multivariable model included sPESI, day-1 haemoglobin and day-1 serum sodium. The sPESI (simplified pulmonary embolism severity index) incorporates age, history of malignancy, heart failure/chronic lung disease, heart rate ≥ 110 beats/min, systolic blood pressure b 100 mm Hg and oxyhaemoglobin saturation b 90%. The chest pain variable is labelled as a categorical variable. The net reclassification improvement was estimated at 0.18 (P b 0.001). b The total number of patients (n = 1253) included in the reclassification analysis did not match the total studied cohort (n = 1306) due to missing day-1 serum sodium (n = 46) and haemoglobin (n = 47) data.
Although our study performed comprehensive multivariable analyses to adjust for differences between the chest pain and non-chest pain group, residual confounders may remain and influence outcomes for both groups. We also did not have sufficient data to adjust for the impact of other biomarkers such as troponin [26–28] or brain natriuretic peptide [29]. It is possible that the independent prognostic value of chest pain may become less significant when these biomarkers are added to the multivariable modelling. 5. Conclusions The presence of chest pain is an independent predictor of 6-month survival in acute PE. The addition of chest pain to an established risk model significantly improved net reclassification of patients, suggesting that this easily obtainable information may be useful in the prognostication of patients with acute PE. Financial support Professor Leonard Kritharides is supported by the NHMRC of Australia Program grant 1037903. Conflicts of interest Nil. Acknowledgements Nil. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijcard.2016.07.129. References [1] S. Laribi, A. Aouba, M. Resche-Rigon, et al., Trends in death attributed to myocardial infarction, heart failure and pulmonary embolism in Europe and Canada over the last decade, QJM 107 (2014) 813–820.
799
[2] S.V. Konstantinides, A. Torbicki, G. Agnelli, et al., 2014 ESC guidelines on the diagnosis and management of acute pulmonary embolism, Eur. Heart J. 35 (2014) 3033–3069 (3069a-3069 k). [3] D. Aujesky, D. Obrosky, R. Stone, et al., Derivation and validation of a prognostic model for pulmonary embolism, Am. J. Respir. Crit. Care Med. 172 (2005) 1041–1046. [4] D. Jimenez, D. Aujesky, L. Moores, et al., Simplification of the pulmonary embolism severity index for prognostication in patients with acute symptomatic pulmonary embolism, Arch. Intern. Med. 170 (2010) 1383–1389. [5] F. Dentali, N. Riva, S. Turato, et al., Pulmonary embolism severity index accurately predicts long-term mortality rate in patients hospitalized for acute pulmonary embolism, J. Thromb. Haemost. 11 (2013) 2103–2110. [6] M.M. Fernandez, S. Hogue, R. Preblick, et al., Review of the cost of venous thromboembolism, Clinicoecon. Outcomes Res. 7 (2015) 451–462. [7] X. Liu, M. Johnson, J. Mardekian, et al., Apixaban reduces hospitalizations in patients with venous thromboembolism: an analysis of the apixaban for the initial management of pulmonary embolism and deep-vein thrombosis as first-line therapy (AMPLIFY) trial, J. Am. Heart Assoc. 4 (2015). [8] P.D. Stein, M.L. Terrin, C.A. Hales, et al., Clinical, laboratory, roentgenographic, and electrocardiographic findings in patients with acute pulmonary embolism and no pre-existing cardiac or pulmonary disease, Chest 100 (1991) 598–603. [9] P.D. Stein, P.W. Willis III, D.L. DeMets, History and physical examination in acute pulmonary embolism in patients without preexisting cardiac or pulmonary disease, Am. J. Cardiol. 47 (1981) 218–223. [10] J.E. Dalen, C.I. Haffajee, J.S. Alpert 3rd, et al., Pulmonary embolism, pulmonary hemorrhage and pulmonary infarction, N. Engl. J. Med. 296 (1977) 1431–1435. [11] J.L. Lobo, V. Zorrilla, F. Aizpuru, et al., Clinical syndromes and clinical outcome in patients with pulmonary embolism: findings from the RIETE registry, Chest 130 (2006) 1817–1822. [12] S.I. Cha, K.M. Shin, J. Lee, et al., Clinical relevance of pulmonary infarction in patients with pulmonary embolism, Thromb. Res. 130 (2012) e1–e5. [13] A. Ng, T. Chung, A. Yong, et al., Long-term cardiovascular and noncardiovascular mortality of 1023 patients with confirmed acute pulmonary embolism, Circ. Cardiovasc. Qual. Outcomes 4 (2011) 122–128. [14] A. Ng, V. Chow, A. Yong, et al., Fluctuation of serum sodium and its impact on short and long-term mortality following acute pulmonary embolism, PLoS One 8 (4) (2013) e61966, http://dx.doi.org/10.1371/journal.pone.0061966. [15] J.P. Moutzouris, A.C. Ng, V. Chow, et al., Acute pulmonary embolism during warfarin therapy and long-term risk of recurrent fatal pulmonary embolism, Thromb. Haemost. 110 (2013) 523–533. [16] J. Moutzouris, V. Chow, A. Yong, et al., Acute pulmonary embolism in individuals aged 80 and older, J. Am. Geriatr. Soc. 62 (2014) 2004–2006. [17] M. Berger, A. Haimowitz, A. Van Tosh, et al., Quantitative assessment of pulmonary hypertension in patients with tricuspid regurgitation using continuous wave Doppler ultrasound, J. Am. Coll. Cardiol. 6 (1985) 359–365. [18] Australian Bureau of Statistics, http://www.abs.gov.au2007 (Accessed March, 2009). [19] National Center for Health Statistics, Instructions for Classifying the Underlying Cause-of-death, ICD-10, 2008 1–259 (http://www.cdc.gov/nchs/about/major/dvs/ im.htm Accessed June 210, 2009). [20] M.J. Pencina, R.B. D'Agostino Sr., R.B. D'Agostino Jr., et al., Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond, Stat. Med. 27 (2008) 157–172 (discussion 207-112). [21] M. Miniati, C. Cenci, S. Monti, et al., Clinical presentation of acute pulmonary embolism: survey of 800 cases, PLoS One 7 (2012) e30891. [22] B.M. Parker, J.R. Smith, Pulmonary embolism and infarction; a review of the physiologic consequences of pulmonary arterial obstruction, Am. J. Med. 24 (1958) 402–427. [23] J.E. Dalen, Pulmonary embolism: what have we learned since Virchow? Natural history, pathophysiology, and diagnosis, Chest 122 (2002) 1440–1456. [24] M.S. Tsao, D. Schraufnagel, N.S. Wang, Pathogenesis of pulmonary infarction, Am. J. Med. 72 (1982) 599–606. [25] A. Ghuysen, B. Ghaye, V. Willems, et al., Computed tomographic pulmonary angiography and prognostic significance in patients with acute pulmonary embolism, Thorax 60 (2005) 956–961. [26] J.D. Douketis, O. Leeuwenkamp, P. Grobara, et al., The incidence and prognostic significance of elevated cardiac troponins in patients with submassive pulmonary embolism, J. Thromb. Haemost. 3 (2005) 508–513. [27] P.D. Stein, F. Matta, M. Janjua, et al., Outcome in stable patients with acute pulmonary embolism who had right ventricular enlargement and/or elevated levels of troponin I, Am. J. Cardiol. 106 (2010) 558–563. [28] K.M. Janata, J.M. Leitner, N. Holzer-Richling, et al., Troponin T predicts in-hospital and 1-year mortality in patients with pulmonary embolism, Eur. Respir. J. 34 (2009) 1357–1363. [29] N. Vuilleumier, G. Le Gal, F. Verschuren, et al., Cardiac biomarkers for risk stratification in non-massive pulmonary embolism: a multicenter prospective study, J. Thromb. Haemost. 7 (2009) 391–398.