Pulmonary embolism severity index accurately predicts long-term mortality rate in patients hospitalized for acute pulmonary embolism

Transcription

1 Journal of Thrombosis and Haemostasis, 11: DOI: /jth ORIGINAL ARTICLE Pulmonary embolism severity index accurately predicts long-term mortality rate in patients hospitalized for acute pulmonary embolism F. DENTALI, N. RIVA, S. TURATO, S. GRAZIOLI, A. SQUIZZATO, L. STEIDL, L. GUASTI, A. M. GRANDI and W. AGENO Department of Clinical Medicine, Insubria University, Varese, Italy To cite this article: Dentali F, Riva N, Turato S, Grazioli S, Squizzato A, Steidl L, Guasti L, Grandi AM, Ageno W. Pulmonary embolism severity index accurately predicts long-term mortality rate in patients hospitalized for acute pulmonary embolism. J Thromb Haemost 2013; 11: Summary. Background: The Pulmonary Embolism (PE) Severity Index (PESI) is a clinical prognostic rule that accurately classifies PE patients into five risk classes with increasing mortality. PESI score has been validated in studies with a relatively short-term follow-up and its accuracy in predicting long-term prognosis has never been established. Methods: Consecutive patients admitted to the tertiary care hospital of Varese (Italy) with an objectively diagnosed PE between January 2005 and December 2009 were retrospectively included. Information on clinical presentation, diagnostic work-up, risk factors, treatment and mortality during a 1-year follow-up was collected. Results: Five hundred and thirty-eight patients were enrolled in this study. The mean age was 70.6 years ( SD 15.2), 44.4% of patients were male, and 27.9% had known cancer. One-year follow-up was available for 96.1% of patients. The overall mortality rate was 23.2% at 3 months, 30.2% at 6 months and 37.1% at 12 months. The discriminatory power of the PESI score to predict long-term mortality, expressed as the area under the ROC curve, was 0.77 (95%CI, ) at 3 months, 0.77 (95%CI, ) at 6 months and 0.79 (95%CI, ) at 12 months. The PESI score confirmed its accurate prediction in patients without cancer. Simplified PESI had a similar overall accuracy to the original PESI at 3 and 6 months, but this was significantly lower at 1 year. Conclusions: The results of this study suggest that PESI score may also be an accurate tool to Correspondence: Francesco Dentali, Unita Operativa Medicina I, Ospedale di Circolo, Viale Borri 57, Varese, Italy. Tel.: ; fax: fdentali@libero.it Received 24 April 2013 Manuscript handled by: I. Pabinger Final decision: F. R. Rosendaal, 22 September 2013 define the 6-month and 1-year mortality rates in PE patients. Keywords: clinical prediction rule; follow-up study; mortality; prognosis; pulmonary embolism. Introduction Pulmonary embolism (PE) is associated with a considerable mortality rate. A short-term adverse outcome at presentation is mainly related to the presence and severity of hemodynamic instability. In the Management Strategy and Prognosis of Pulmonary Embolism Registry, the inhospital mortality rate ranged from 8.1% in a group of stable patients with acute right heart failure, to 25% in patients with cardiogenic shock or even 65% in those requiring cardiopulmonary resuscitation [1]. On the other hand, a long-term adverse outcome mainly depends on the presence and severity of concomitant diseases. In the Prospective Investigation of Pulmonary Embolism Diagnosis Project, 23.8% of patients died within 1 year after the diagnosis of PE, mainly because of cancer, sepsis or underling cardiac diseases [2]. The Pulmonary Embolism Severity Index (PESI) is a practical clinical prediction rule (CPR), which has been derived and validated in patients admitted to hospital with PE [3]. PESI uses 11 predictors from the medical history and physical examination, without the need for laboratory parameters or imaging procedures (Table 1). This model reliably stratifies patients into five risk classes with increasing risk of short-term mortality, and is proposed as a potential tool to guide initial intensity of treatment. PESI class I and II patients have a very low risk of adverse outcomes, with an in-hospital mortality of less than 1%, and are potential candidates for outpatient treatment or early hospital discharge [4]. Recently, a simplified version of the original PESI, based on six variables

2 2104 F. Dentali et al Table 1 Prognostic variables and risk stratification of the Pulmonary Embolism Severity Index (PESI). Adapted from Aujesky [3] Predictors Demographic characteristics Age (years) Age Male sex +10 Co-morbid illnesses Cancer (previous or active) +30 Heart failure +10 Chronic lung disease +10 Clinical findings Pulse 110 min Systolic blood pressure < 100 mmhg +30 Respiratory rate 30 min Temperature < 36 C +20 Altered mental status (disorientation, lethargy, +60 stupor or coma) Arterial oxygen saturation < 90% (with or +20 without the administration of supplemental oxygen) only, has shown a similar prognostic accuracy in predicting 30-day mortality, when compared with the original PESI [5]. These two CPRs are revealed to be accurate also with a more extended, 90-day follow-up [6 8], but their ability in predicting longer term prognosis has never been established. The purpose of this study is to investigate the accuracy of the original and simplified PESI to predict the 6-month and 1-year mortality rate in PE patients. Methods Patient identification and eligibility Points assigned Risk classes Points Risk stratification Class I 65 Very low risk Class II Low risk Class III Intermediate risk Class IV High risk Class V > 125 Very high risk All patients with an objective diagnosis of PE at the tertiary care hospital of Varese, Italy, from January 2005 to December 2009, were potentially eligible for this study. Patients were identified using discharge codes according to the 9th Clinical Modification International Classification of Diseases (ICD-9-CM and ). Charts of all potentially eligible patients were retrieved for evaluation. Only adult patients with an objectively diagnosed first episode of acute PE were included for the purpose of the study. The criteria used to establish the diagnosis of PE were a positive spiral CT scan, pulmonary angiography, a high-probability perfusion lung scan, or intermediate probability perfusion lung scan with deep vein thrombosis (DVT) documented by compression ultrasonography. Baseline data collection Trained study personnel retrospectively recorded baseline patient characteristics, including the variables that comprise the original PESI (age, gender, cancer, heart failure, chronic lung disease, pulse 110 beats min 1, systolic blood pressure < 100 mmhg, respiratory rate 30 min 1, temperature < 36 C, altered mental status and arterial oxygen saturation < 90%). For the simplified PESI the following parameters were considered: age > 80 years, cancer, chronic cardiopulmonary disease, pulse 110 beats min 1, systolic blood pressure < 100 mmhg and arterial oxygen saturation < 90%. Data on the presence of concomitant DVT and medical history focusing on potential risk factors for thrombosis, treatment and clinical outcome were gathered. Furthermore, information on personal history of venous thromboembolism (VTE) was also collected. A positive personal history of VTE was established if the patient had a previous objectively assessed episode of DVT, splanchnic vein thrombosis or cerebral vein thrombosis. In the case of previous PE, the patient was not eligible for the study. PE was defined as provoked in the presence of one of the following risk factors: cancer, recent surgery, oral contraceptive (OC) use, pregnancy, puerperium, hormone replacement therapy (HRT), recent acute medical disease or recent confinement to bed for 72 h. In the absence of the aforementioned predisposing factors, PE was defined as unprovoked. Using the prognostic variables of the PESI score, we calculated the risk class for each patient, and the proportion of patients classified within each risk class. Missing values for all prognostic variables were assumed to be normal (a strategy used in the original derivation of the PESI) [3]. After discharge, most patients with PE are regularly followed by the local anticoagulation clinic. Information on clinical events during follow-up for these patients was first collected using the computerized database of the clinic. If patients were not followed by the local clinic, or if followup data were not available, patients were contacted by telephone or by a mailed questionnaire. At the time of contact, information on vital status was collected. If death occurred, family members or the general practitioner were asked about the possible cause of death. Death was judged to be related to PE if confirmed by autopsy, or if death followed a clinically severe, objectively diagnosed PE. Sudden or unexpected death was classified as a possible fatal PE. The Institutional Review Board approved the study, which was carried out and is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for observational studies [9].

3 PESI score and long term mortality for acute PE 2105 Statistical analyses Continuous variables were expressed as mean plus or minus the standard deviation (SD) or as median with minimum and maximum values when data did not have a normal distribution; categorical data are given as counts and percentages. Initially, patients were divided into five classes according to the original PESI score. Class-specific mortality at 3, 6 and 12 months was compared. To assess the discriminatory power of the PESI score to predict long-term mortality, we measured the area under the receiver operating characteristic (ROC) curves at 3, 6 and 12 months. The overall mortality rate of low- (risk classes I and II) vs. high-risk patients (risk classes III-V) at 3, 6 and 12 months was compared using Kaplan Meier analysis and the log-rank test. We evaluated the effect of each PESI predictor at 12- month follow-up, using a multivariable Cox model including all the predictors of the PESI score. Furthermore, to explore the potential role of other predictors of mortality in patients with PE, we performed a multivariable Cox regression analysis with backward elimination, including the risk categories of the PESI score (low vs. high-risk) together with the following variables: previous acute coronary syndrome, previous cerebrovascular accident, peripheral artery disease, autoimmune disease, diabetes mellitus, history of chronic kidney disease, atrial fibrillation, previous VTE (DVT, splanchnic or cerebral vein thrombosis), chronic liver disease and dyslipidemia. Variables with a P value < 0.05 were considered independent predictors of mortality. In a separate Cox regression analysis, duration of anticoagulant therapy (less or equal to 3 months vs. more than 3 months) has been evaluated with respect to the risk categories of the PESI score. To explore the role of PESI in predicting mortality in patients without cancer at baseline, we performed a subanalysis excluding patients with cancer and removing 30 points from each PESI class. To assess the accuracy of PESI to predict overall mortality, we estimated sensitivity, specificity, positive and negative predictive values and likelihood ratios for low vs. high-risk patients. A positive likelihood ratio indicates how much more likely it is that patients who die are classified into PESI risk classes III, IV and V relative to those who survive; a negative likelihood ratio indicates how much less likely it is that patients who die are classified in PESI risk classes I and II compared with those who survive. Afterwards, all the analyses were repeated dividing patients into two classes according to the simplified PESI (0 points at low risk of death, 1 or more at high risk of death). To examine the predictive validity of the original and simplified PESI, we compared the area under the ROC curves using the roccomp command in STATA [10]. All analyses were performed using SPSS 19.0 (SPSS Inc, Chicago, IL, USA) and STATA 12 (StataCorp LP, College Station, TX, USA). Results Baseline patient characteristics The charts of 555 patients with an objective diagnosis of PE were reviewed. Seventeen patients were excluded because of a previous episode of PE, leaving 538 patients diagnosed with a first episode of acute symptomatic PE (Table 2). The median age was 73 years, ranging from 18 to 100 years; 239 (44.4%) patients were male. Diagnosis of PE was obtained with spiral CT in 419 (77.9%) patients and perfusion lung scan in 119 (22.1%) patients. Concomitant DVT was present in 301 (55.9%) patients, 293 in the lower limbs, eight in the upper limbs. Moreover, six patients had a diagnosis of unusual site thrombosis, five in the splanchnic veins and one in the cerebral veins, during the same hospitalization. The high prevalence of concomitant DVT is partly explained by a systematic search for the origin of the clot in most PE patients at our institution. Sixty-one (11.3%) patients had a personal history of VTE. PE was unprovoked in 251 patients (46.7%) and cancer-related in 150 patients (27.9%). Risk factors for secondary events are listed in Table 2. Among the clinical variables of the PESI score, 134 (24.9%) patients had arterial oxygen saturation < 90% (with or without the administration of supplemental oxygen), 92 (17.1%) had a pulse 110 beats min 1, 60 (11.2%) had systolic blood pressure < 100 mmhg, 30 (5.6%) had a temperature < 36 C, 25 (4.6%) had altered mental status (including disorientation, lethargy, stupor or coma) and 12 (2.2%) had a respiratory rate 30 min 1. According to the PESI score, 172 patients were at low risk (classes I and II combined) and 366 patients at intermediate-high risk (classes III, IV and V combined). During the hospitalization, 32 (5.9%) patients were treated with thrombolysis, 76 (14.1%) with unfractionated heparin and 410 (76.2%) patients received low-molecularweight heparin (LMWH). One patient had an inferior vena cava filter placed and 19 (3.5%) patients were not treated. Follow-up and accuracy of original and simplified PESI After discharge, no information was available for 13 out of 538 patients. Data on 525 (97.6%) patients were therefore available for the 3-month follow-up, on 524 (97.4%) for the 6-month follow-up and on 517 (96.1%) for the 12-month follow-up. Characteristics of patients with a complete follow-up were not significantly different from those of the entire population (data not shown).

4 2106 F. Dentali et al Table 2 Baseline patient characteristics and risk-class distribution Patients, n 538 Provoked/unprovoked PE, n (%) 287 (53.3%)/251, (46.7%) PESI predictors Age (years), mean SD median (range) (17 100) Male sex, n (%) 239 (44.4%) Cancer, n (%) 150 (27.9%) Heart failure, n (%) 28 (5.2%) Chronic lung disease, n (%) 87 (16.2%) Temperature < 36 C, n (%) 30 (5.6%) Pulse 110 min 1, n (%) 92 (17.1%) Systolic blood pressure < (11.2%) mmhg, n (%) Respiratory rate 30 min 1, n (%) 12 (2.2%) Altered mental status, n (%) 25 (4.6%) Arterial oxygen saturation < 90%, 134 (24.9%) n (%) Risk-class distribution (PESI score) Class I, n (%) 49 (9.1%) Class II, n (%) 123 (22.9%) Class III, n (%) 146 (27.1%) Class IV, n (%) 117 (21.7%) Class V, n (%) 103 (19.1%) Other co-morbid illnesses Concomitant DVT, n (%) 307 (57.1%) Previous VTE*, n (%) 61 (11.3%) Oral contraceptive treatment, n (%) 14 (2.6%) Recent bone fracture or 33 (6.1%) orthopedic surgery, n (%) Chronic venous insufficiency, n (%) 15 (2.8%) Myeloproliferative neoplasm, n (%) 6 (1.1%) Concomitant infection at 67 (12.5%) admission, n (%) Previous cerebrovascular 70 (13.0%) accident, n (%) Previous acute coronary 64 (11.9%) syndrome, n (%) Peripheral artery disease, n (%) 16 (3.0%) Anamnesis of chronic kidney 49 (9.1%) disease, n (%) Atrial fibrillation, n (%) 65 (12.1%) Autoimmune disease, n (%) 30 (5.6%) Diabetes mellitus, n (%) 61 (11.3%) Dyslipidemia, n (%) 15 (2.8%) Prosthetic cardiac valve, n (%) 3 (0.6%) Chronic hepatopathy, n (%) 8 (1.5%) Congenital thrombophilia, n (%) 3 (0.6%) *Previous VTE includes DVT, splanchnic or cerebral vein thrombosis, because patients with previous PE were not eligible for this study. DVT, deep vein thrombosis; PE, pulmonary embolism; VTE, venous thromboembolism. Overall, mortality rate was 23.2% (95% CI, ) at 3 months, 30.2% (95% CI, ) at 6 months and 37.1% (95% CI, ) at 12 months (Table 3). Of the 86 deaths that occurred during hospitalization, 63 were attributable to PE and 15 were possibly related to PE, while of the 106 deaths that occurred after discharge, five were fatal PE and seven were possible fatal PE. Cancer was responsible for 32.3% of the overall mortality (Table 3). Stratification into risk classes according to the original PESI score was significantly correlated with mortality rate at 3, 6 and 12 months (chi-square for trend P < for all of these). At 1 year, mortality was 2.2% (95% CI, ), 10.8% (95% CI, ), 33.1% (95% CI, ), 52.2% (95% CI, ) and 72.3% (95% CI, ), respectively, in the five risk classes (Table 3). Significant PESI predictors of 1-year mortality at multivariable Cox regression analysis were: age, altered mental status, cancer, pulse 110 min 1, systolic blood pressure < 100 mmhg and arterial oxygen saturation < 90% (Table 4). In a multivariable Cox regression analysis exploring the potential role of other predictors of mortality, a high-risk category according to the PESI score was significantly associated with 1-year mortality rate (HR, 7.66; 95% CI, ; P < 0.001), while previous VTE was significantly associated with 1-year survival (HR, 0.44; 95% CI, ; P = 0.010). Duration of anticoagulant treatment did not emerge as an independent prognostic predictor (HR, 0.52; 95% CI, ; P = 0.113). The Kaplan Meier curves for cumulative survival of low-risk (classes I and II combined) vs. high-risk patients (classes III, IV and V combined) were significantly different when compared at 3-, 6- and 12-month follow-up (log-rank test P < for these three comparisons) (Figure 1). We performed another survival analysis censoring patients with cancer-related mortality, in order to take competing risks into account, but the results did not change (data not shown). When dichotomized as low risk vs. high risk, the PESI score had a sensitivity of 92.7% (95% CI, ) and a specificity of 46.5% (95% CI, ), a negative predictive value of 91.5% (95% CI, ) and a positive predictive value of 50.6% (95% CI, ) for overall mortality at 1 year, with a negative likelihood ratio of 0.16 (95% CI, ) and a positive likelihood ratio of 1.73 (95% CI, ) (Table 5). The discriminatory power of the PESI score to predict long-term mortality, expressed as the area under the ROC curve (AUC), was 0.77 (95% CI, ) at 3 months, 0.77 (95% CI, ) at 6 months and 0.79 (95% CI, ) at 12 months. On the other hand, the PESI score has a moderate discriminatory power in predicting cancer-related mortality (AUC, 0.74; 95% CI, ), not-cancer-related mortality (AUC, 0.73; 95% CI, ), overall PE mortality (AUC, 0.73; 95% CI, ) or definite PE mortality (AUC, 0.74; 95% CI, ) at 12 months. Of note, PESI confirmed the accurate prediction of overall mortality at 3 months (AUC, 0.74; 95% CI, ), 6 months (AUC, 0.74; 95% CI, ) and 12 months (AUC, 0.74; 95% CI, ) also in patients without cancer. According to the simplified PESI score, 137 patients (26.1%) were classified as low risk and 388 (73.9%) as

5 PESI score and long term mortality for acute PE 2107 Table 3 Causes of death and comparison of risk-class-specific mortality according to PESI score During hospital stay (N = 86), n (%) After discharge (N = 106), n (%) Total (N = 192), n (%) Causes of death PE 63 (73.3%) 5 (4.7%) 68 (35.4%) Possible fatal PE 15 (17.4%) 7 (6.6%) 22 (11.5%) Cancer 8 (9.3%)* 54 (50.9%) 62 (32.3%) Bleeding 2 (2.3%) 5 (4.7%) 7 (3.6%) Infection 1 (1.2%) 5 (4.7%) 6 (3.1%) Other cardiopathy 0 (0%) 5 (4.7%) 5 (2.6%) Cerebrovascular accident 0 (0%) 5 (4.7%) 5 (2.6%) Acute renal failure 1 (1.2%) 1 (0.9%) 2 (1.0%) Unknown 0 (0%) 19 (17.9%) 19 (9.9%) Mortality rate at 3 months, n/n (%, 95% CI) Mortality rate at 6 months, n/n (%, 95% CI) Mortality rate at 12 months, n/n (%, 95% CI) Risk-class-specific mortality Overall 122/525 (23.2%, %) 158/524 (30.2%, %) 192/517 (37.1%, %) Class I 0/48 (0%, 0 7.4%) 0/48 (0%, 0 7.4%) 1/45 (2.2%, %) Class II 9/121 (7.4%, %) 12/121 (9.9%, %) 13/120 (10.8%, %) Class III 25/139 (18.0%, %) 37/138 (26.8%, %) 45/136 (33.1%, %) Class IV 33/115 (28.7%, %) 44/115 (38.3%, %) 60/115 (52.2%, %) Class V 55/102 (53.9%, %) 65/102 (63.7%, %) 73/101 (72.3%, %) Chi-square for trend, P value < < < *In four of these eight patients the concomitant presence of PE and cancer has been reported as cause of death in the hospital charts. n, number of events; N, number of patients available for follow-up; CI, confidence interval; PE, pulmonary embolism. Table 4 Effect of each PESI predictor at 12-month follow-up using the multivariable Cox regression analysis HR (95% CI) Age* 1.03 ( ) Male sex 1.10 ( ) Cancer 4.41 ( ) Heart failure 1.52 ( ) Chronic lung disease 1.03 ( ) Temperature < 36 C 1.52 ( ) Pulse 110 min ( ) Systolic blood pressure < 100 mmhg 2.10 ( ) Respiratory rate 30 min ( ) Altered mental status 4.72 ( ) Arterial oxygen saturation < 90% 1.45 ( ) *The hazard ratio for age is for 1-year increase. HR, hazard ratio; CI, confidence interval. high risk. At 1 year the simplified PESI score had a sensitivity of 95.8% (95% CI, ) and a specificity of 38.2% (95% CI, ), a negative predictive value of 93.9% (95% CI, ) and a positive predictive value of 47.8% (95% CI, ) for overall mortality, with a negative likelihood ratio of 0.11 (95% CI, ) and a positive likelihood ratio of 1.55 (95% CI, ) (Table 5). The simplified PESI had a similar overall accuracy to the original PESI at 3-month (P = 0.22) and at 6-month follow-up (P = 0.40), whereas at 1 year the overall accuracy of the simplified PESI was significantly lower compared with the original PESI (AUC, 0.75; 95% CI, vs. AUC, 0.79; 95% CI, ; P = 0.011) (Table 5 and Figure 2). Discussion In this study, we assessed the prognostic performance of PESI in stratifying patients according to their risk of long-term adverse outcomes. In particular, the results of our study indicate that PESI may identify a subgroup of PE patients at low risk of overall mortality even at 12 months; that is < 5% in class I. Furthermore, we were able to identify a large group of patients with a very high risk of mortality at 12 months. Defining the prognosis of patients with an acute PE may have important clinical implications and may help clinicians to allocate adequate resources in the management of these patients. Despite a great amount of data available on short-term prognosis of PE patients, only a few studies have investigated possible predictors of long-term prognosis in these patients [4,11,12]. Most data are only on the general natural history of PE patients without stratifying for prognostic risk factors or, more frequently, follow-up is limited to the first 3 6 months of treatment [13,14]. However, few studies suggest that some prognostic factors are independently associated with long-term prognosis in PE patients. An association between patient-related factors (e.g. cancer and renal insufficiency) and PE-related factors on

6 2108 F. Dentali et al Cumulative survival Months of follow-up Low risk patients High risk patients Fig. 1. Kaplan Meier curves for low-risk (classes I and II) vs. highrisk (classes III, IV and V) patients, according to PESI score. admission (i.e. severity of clinical presentation, electrocardiogram and echocardiographic parameters) with adverse outcomes both in the short term and after hospitalization was previously shown [2,15]. Carson and colleagues [2] prospectively followed 399 patients with PE for 1 year. Co-morbidities associated with long-term mortality were: the presence of cancer (relative risk [RR], 3.8; 95% CI, ), left-sided congestive heart failure (RR, 2.7; 95% CI, ) and chronic lung disease (RR, 2.2; 95% CI, ). Meneveau and colleagues [15] analyzed a registry of 249 PE patients treated with thrombolytic drugs who 9 12 were followed for a mean follow-up of 5.3 years. At the multivariable analysis, the following variables were associated with long-term mortality: age > 75 years (RR, 2.73; 95% CI, ), persistence of vascular pulmonary obstruction > 30% after thrombolytic treatment (RR, 2.22; 95% CI, ) and cancer (RR, 2.03; 95% CI, ). Moreover, Ribeiro and colleagues [16] have shown that PE patients with systolic pulmonary arterial pressure of more than 50 mmhg at admission had a higher risk of persistent pulmonary hypertension at 1 year, as well as an excess of mortality at 5 years. Unfortunately, most of these studies were not able to identify patients at low risk of death. To the best of our knowledge, only two studies have investigated prognostic clinical variables formally combined in a CPR to predict long-term prognosis in PE patients [4]. Subramanian and colleagues prospectively evaluated the performance of the Geneva prognostic CPR in 105 PE patients at 3 and 12 months [17]. At the 12-month follow-up, 5/88 patients (5.7%) with a score of two or less died and 8/17 patients (47.1%) with a score of three or more died (P < ). Yamaki and colleagues [18] investigated the accuracy in predicting overall mortality and recurrent venous thromboembolism at 12 months of their own CPR in 203 PE patients. The adverse event rates were 6.0% for the low-risk group and 59.5% for the high-risk group. Our study investigated the long-term prognostic accuracy of PESI. Several large cohorts have confirmed its ability to predict the short-term mortality [4]. The PESI was studied in 21 cohorts with a total of patients, has a Level 2 of evidence at McGinn s scale for quality of CPR development, and has been used for selecting patients eligible for home treatment in a randomized controlled trial [19]. Moreover, PESI identified 43% of PE patients with in-hospital mortality of less than 1% (i.e. the threshold proposed by both the European Society of Table 5 Accuracy of the original and simplified PESI prognostic models to predict mortality for low vs. high risk patients Original PESI score Simplified PESI score 3 months (N = 525) 6 months (N = 524) 12 months (N = 517) 3 months (N = 525) 6 months (N = 524) 12 months (N = 517) Sensitivity,% (95% CI) 92.6 ( ) 92.4 ( ) 92.7 ( ) 96.7 ( ) 96.8 ( ) 95.8 ( ) Specificity,% (95% CI) 39.7 ( ) 42.9 ( ) 46.5 ( ) 33.0 ( ) 35.8 ( ) 38.2 ( ) Positive predictive 31.7 ( ) 41.1 ( ) 50.6 ( ) 30.4 ( ) 39.4 ( ) 47.8 ( ) value,% (95% CI) Negative predictive 94.7 ( ) 92.9 ( ) 91.5 ( ) 97.1 ( ) 96.3 ( ) 93.9 ( ) value,% (95% CI) Positive likelihood 1.54 ( ) 1.62 ( ) 1.73 ( ) 1.44 ( ) 1.51 ( ) 1.55 ( ) ratio (95% CI) Negative likelihood 0.19 ( ) 0.18 ( ) 0.16 ( ) 0.10 ( ) 0.09 ( ) 0.11 ( ) ratio (95% CI) Area under the receiver operating characteristics curve (95% CI) 0.77 ( ) 0.77 ( ) 0.79 ( ) 0.75 ( ) 0.76 ( ) 0.75 ( ) N, number of patients available for follow-up; CI, confidence interval.

7 PESI score and long term mortality for acute PE 2109 Sensitivity month follow up Specificity Original PESI Simplified PESI Fig. 2. ROC curves of the original and simplified PESI for mortality at 12 months. missing. Because lacking information on vital parameters in clinical practice, such as respiratory or heart rate, usually means that the patient is not tachypnoic or tachycardic, these variables were assumed to be normal, a strategy previously used in the original derivation of the PESI score [3]. Moreover, given the retrospective design of our study, the causes of death were retrieved from the hospital records or other medical documentation, without performing an external independent adjudication. Second, we were not able to compare the accuracy of the PESI with other CPRs because not all the items of these scores were routinely collected in our hospital. In conclusion, PESI may be an optimal tool for stratifying PE patients according to both their short-term and long-term mortality risk. Before implementing the PESI as a long-term prognostic CPR in clinical practice, future studies should confirm our data. Cardiology and the American Heart Association for defining low-risk PE patients). In our study, stratification into risk classes according to PESI score was significantly correlated with mortality rate for up to 12 months, with a high sensitivity (92.7%) and high negative predictive value (91.5%) for overall mortality also at 1 year. Furthermore, apart from the high-risk category of the PESI score, no other variable emerged as an independent predictor of long-term mortality. Vice versa, previous VTE was significantly associated with 1-year survival, suggesting that these patients might obtain early diagnosis and premature treatment. The simplified PESI score appeared to have a similar accuracy to the original PESI at 3 and 6 months, whereas at 1 year its accuracy was significantly lower compared with the accuracy of the original PESI, suggesting caution if used in defining the long-term prognosis of patients with PE. Of clinical note, in our cohort a lower percentage of patients were classified as low risk according to the PESI score (only 32%) and we found a higher mortality rate compared with previous studies evaluating patients with acute PE. The high mortality rate could be related to the extensive prevalence of cancer and the older age of our population, compared with previous studies. However, because we enrolled all consecutive inpatients with an objective diagnosis of PE, our study is likely to reflect everyday clinical practice. The main strengths of our study include the largest cohort that has been investigated for the performance of a prognostic CPR for long-term outcomes and less than 5% of PE patients lost to follow-up. However, our paper has potential limitations. First, the retrospective design of the study means, per definition, lack of systematic reporting of all data and misdiagnosis. However, objective criteria were required to confirm the diagnosis of PE and less than 5% of data regarding the PESI prognostic variables were Addendum F. Dentali contributed to the conception and design of the study, analysis and interpretation of data and drafted the article. N. Riva contributed to analysis and interpretation of data and drafted the article. S. Turato and S. Grazioli contributed to acquisition, analysis and interpretation of data. A. Squizzato and W. Ageno drafted the article and contributed to interpretation of data and critical revision of the manuscript. L. Steidl, L. Guasti and A. M. Grandi contributed to interpretation and critical revision of the manuscript. All authors provided final approval of the manuscript. Disclosure of Conflict of Interests The authors state that they have no conflict of interests. References 1 Kasper W, Konstantinides S, Geibel A, Olschewski M, Heinrich F, Grosser KD, Rauber K, Iversen S, Redecker M, Kienast J. Management strategies and determinants of outcome in acute major pulmonary embolism: results of a multicenter registry. J Am Coll Cardiol 1997; 30: Carson JL, Kelley MA, Duff A, Weg JG, Fulkerson WJ, Palevsky HI, Schwartz JS, Thompson BT, Popovich J Jr, Hobbins TE, Spera MA, Alavi A, Terrin ML. The clinical course of pulmonary embolism. N Engl J Med 1992; 326: Aujesky D, Obrosky DS, Stone RA, Auble TE, Perrier A, Cornuz J, Roy PM, Fine MJ. Derivation and validation of a prognostic model for pulmonary embolism. Am J Respir Crit Care Med 2005; 172: Squizzato A, Donadini MP, Galli L, Dentali F, Aujesky D, Ageno W. Prognostic clinical prediction rules to identify a lowrisk pulmonary embolism: a systematic review and meta-analysis. J Thromb Haemost 2012; 10: Jimenez D, Aujesky D, Moores L, Gomez V, Lobo JL, Uresandi F, Otero R, Monreal M, Muriel A, Yusen RD; RIETE Investigators. Simplification of the pulmonary embo-

8 2110 F. Dentali et al lism severity index for prognostication in patients with acute symptomatic pulmonary embolism. Arch Intern Med 2010; 170: Aujesky D, Roy PM, Le Manach CP, Verschuren F, Meyer G, Obrosky DS, Stone RA, Cornuz J, Fine MJ. Validation of a model to predict adverse outcomes in patients with pulmonary embolism. Eur Heart J 2006; 27: Aujesky D, Perrier A, Roy PM, Stone RA, Cornuz J, Meyer G, Obrosky DS, Fine MJ. Validation of a clinical prognostic model to identify low-risk patients with pulmonary embolism. J Intern Med 2007; 261: Righini M, Roy PM, Meyer G, Verschuren F, Aujesky D, Le Gal G. The Simplified Pulmonary Embolism Severity Index (PESI): validation of a clinical prognostic model for pulmonary embolism. J Thromb Haemost 2011; 9: von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med 2007; 147: Cleves MA. From the help desk: comparing areas under receiver operating characteristic curves from two or more probit or logit models. Stata J 2002; 2: Torbicki A, Perrier A, Konstantinides S, Agnelli G, Galie N, Pruszczyk P, Bengel F, Brady AJ, Ferreira D, Janssens U, Klepetko W, Mayer E, Remy-Jardin M, Bassand JP; ESC Committee for Practice Guidelines (CPG). Guidelines on the diagnosis and management of acute pulmonary embolism: the Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC). Eur Heart J 2008; 29: Jaff MR, McMurtry MS, Archer SL, Cushman M, Goldenberg N, Goldhaber SZ, Jenkins JS, Kline JA, Michaels AD, Thistlethwaite P, Vedantham S, White RJ, Zierler BK; on behalf of the American Heart Association Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation, Council on Peripheral Vascular Disease, and Council on Arteriosclerosis, Thrombosis and Vascular Biology. Management of massive and submassive pulmonary embolism, iliofemoral deep vein thrombosis, and chronic thromboembolic pulmonary hypertension: a Scientific Statement from the American Heart Association. Circulation 2011; 123: Nijkeuter M, S ohne M, Tick LW, Kamphuisen PW, Kramer MH, Laterveer L, van Houten AA, Kruip MJ, Leebeek FW, B uller HR, Huisman MV; Christopher Study Investigators. The natural course of hemodynamically stable pulmonary embolism: clinical outcome and risk factors in a large prospective cohort study. Chest 2007; 131: Klok FA, Zondag W, van Kralingen KW, van Dijk AP, Tamsma JT, Heyning FH, Vliegen HW, Huisman MV. Patient outcomes after acute pulmonary embolism. A pooled survival analysis of different adverse events. Am J Respir Crit Care Med 2010; 181: Meneveau N, Ming LP, Seronde MF, Mersin N, Schiele F, Caulfield F, Bernard Y, Bassand JP. In-hospital and long-term outcome after sub-massive and massive pulmonary embolism submitted to thrombolytic therapy. Eur Heart J 2003; 24: Ribeiro A, Lindmarker P, Johnsson H, Juhlin-Dannfelt A, Jorfeldt L. Pulmonary embolism: one-year follow-up with echocardiography doppler and five-year survival analysis. Circulation 1999; 99: Subramaniam RM, Mandrekar J, Blair D, Peller PJ, Karalus N. The Geneva prognostic score and mortality in patients diagnosed with pulmonary embolism by CT pulmonary angiogram. J Med Imaging Radiat Oncol 2009; 53: Yamaki T, Nozaki M, Sakurai H, Takeuchi M, Soejima K, Kono T. Presence of lower limb deep vein thrombosis and prognosis in patients with symptomatic pulmonary embolism: preliminary report. Eur J Vasc Endovasc Surg 2009; 37: Aujesky D, Roy PM, Verschuren F, Righini M, Osterwalder J, Egloff M, Renaud B, Verhamme P, Stone RA, Legall C, Sanchez O, Pugh NA, N gako A, Cornuz J, Hugli O, Beer HJ, Perrier A, Fine MJ, Yealy DM. Outpatient versus inpatient treatment for patients with acute pulmonary embolism: an international, open-label, randomised, non-inferiority trial. Lancet 2011; 378: 41 8.