Effect of a Multidisciplinary Pulmonary Embolism Response Team on Patient Mortality

Abstract

Multidisciplinary Pulmonary Embolism Response Teams (PERTs) may improve the care of patients with high risk pulmonary embolism (PE). The impact of a PERT on long-term mortality has never been evaluated. An observational analysis was conducted of patients, 137 before PERT (between 2014 and 2015) and 231 after PERT implementation (between 2016 to 2019), presenting to the emergency department of an academic medical center with submassive and massive PE. The primary outcome was six-month mortality, evaluated by univariate and multivariate analyses. PERT was associated with a sustained reduction in mortality through six-months (six-month mortality rates of 14% post-PERT vs. 24% pre-PERT, unadjusted HR of 0.57, RRR of 43%, p=0.025). There was a reduced length of stay following PERT implemenation (9.1 vs. 6.5 days, P=0.007). Time from triage to diagnosis of PE was independently predictive of mortality, and the risk of mortality was reduced by 5% for each hour earlier that the diagnosis was made. In conclusion, this study is the first to demonstrate an association between PERT implementation and a sustained reduction in 6-month mortality for patients with high risk PE.

INTRODUCTION

Pulmonary embolism (PE) is the third leading cause of cardiovascular mortality, and its downstream consequences of morbidity and disability are a driver of health care expenses and social and economic consequences^1–5. The emergence of a vast array of new therapeutic options (catheter-based treatments and mechanical hemodynamic support) in the advent of a multidisciplinary Pulmonary Embolism Response Team (PERT) arose in an attempt to improve the care of patients with high risk PE by focusing on optimizing management strategies rather than a singular therapy^6–8. PERTs have influenced new treatment strategies and may have favorable effects on mortality, as well as improved efficiency of PE management including reduced time from PE diagnosis to anticoagulation, a metric associated with reduced mortality^9–12. The present study tested the hypothesis that the implementation of a multidisciplinary PERT would have a sustained improvement in mortality in patients high risk PE through an identifiable management strategy or treatment.

METHODS

We conducted an observational cohort study of adults with submassive and massive PE presenting to the ED of a tertiary care medical center (University of Rochester Medical Center/Strong Memorial Hospital). The study population is comprised of retrospective and prospective cohorts, before (pre-PERT) and after (post-PERT) the institution of the PERT in2016, respectively. The study was approved by the institution’s Research Subject Review Board.

Adults with acute massive PE were defined as those with new symptomatic PE with sustained hypotension (systolic blood pressure (SBP) <90 mmHg), cardiopulmonary arrest, or dependence on catecholamine use to maintain blood pressure. Submassive PE was defined as those with new symptomatic PE and evidence of RV strain by imaging (computed tomography aor echocardiography) or elevated cardiac biomarkers (N-terminal brain natriuretic peptide (NT-proBNP) or cardiac troponin T) in normotensive patients.

Patients in the pre-PERT cohort were retrospectively identified by International Classification of Diseases (ICD) 9 and 10 codes for PE. Patients were included if they met the above criteria for submassive or massive PE, and were excluded if they were low risk, if they had chronic venous thromboembolism (VTE) but the reason for presentation was unrelated, and transfers from an outside facility where the diagnosis of PE was previously established and pre-hospital treatment decisions were already made. Pre-PERT patients received the standard care at the discretion of the treating physicians. Patients in the post-PERT cohort were prospectively enrolled upon PERT activation with care determined using a multidisciplinary approach.

Our institutional PERT was directed and co-managed by the department of cardiology, but members of the team are trained in pulmonary / critical care medicine, cardiac surgery, interventional radiology, cardiovascular disease, vascular medicine, pharmacy, and emergency medicine (EM). When PERT is activated by EM physicians for a patient meeting criteria for acute submassive or massive PE, an immediate and shared evaluation by both the inpatient medical intensive care unit (MICU) and cardiac care unit (CCU) is performed, and a treatment plan is formulated within 30 minutes. Assessment of hemodynamics, respiratory status, and right ventricular (RV) function (including bedside echocardiography by a cardiologist) is performed to risk stratify patients and assess the risks and benefits of therapeutic options including anticoagulation and advanced therapies including systemic thrombolytics, catheter directed therapy, surgical thrombectomy, IVC filter placement, or mechanical hemodynamic assistance including extracorporeal membrane oxygenation (ECMO). The optimal therapy is then determined by a consensus of the multidisciplinary PERT. Low molecular weight heparin (1 mg/kg enoxaparin twice daily) was recommended as the initial anticoagulant unless contraindicated. Patients with submassive or massive PE are admitted to an intensive care unit or step-down unit for enhanced monitoring.. All patients meeting submassive or massive criteria follow-up with our outpatient post-PERT clinic.

Patient data were obtained by reviewing the electronic medical record for both pre-PERT and post-PERT cohorts. Baseline characteristics including provocation factors, comorbid conditions, use of antithrombotic therapy, presenting symptoms, vital signs, imaging characteristics, and cardiac biomarkers were recorded. Management decisions by the treating physicians were recorded, including the type of anticoagulant (both initially and at discharge), use of advanced therapies, disposition of patient to an admitting service are shown in Supplemental Fig. 1. The efficiency of care (including time from ED triage to diagnosis of PE, time from diagnosis to administration of anticoagulation, and time from triage to admission to the hospital). was recorded data using a customized HIPAA-compliant Research Electronic Data Capture (REDCap) tool¹³.

The primary outcome of this study is all-cause mortality through 6 months. Univariate and multivariate analyses were performed to determine the predictors of mortality, and the mortality analysis was subsequently adjusted for baseline variables. Secondary outcomes included length of hospital stay, in-hospital or 30-day major bleeding events, and hemodynamic decompensation. Major bleeding was defined as clinically overt bleeding accompanied by a decrease in blood hemoglobin of at least 2.0 g/dL or transfusion of at least 2 units of packed red cells, or bleeding in high risk anatomic compartments. Hemodynamic decompensation was defined as one or more of the following: systolic blood pressure < 90 mm Hg for at least 15 minutes, or systemic thrombolysis, endotracheal intubation, catecholamine administration, mechanical circulatory support, or cardiopulmonary resuscitation (CPR).

Data collected were extracted for statistical modeling and analysis¹³. Dichotomous variables are presented as frequencies and continuous variables as mean with standard deviation (SD) unless otherwise stated. For comparisons between pre- and post-PERT groups, the Wilcoxon rank-sum test was utilized for continuous measures and χ² or Fisher’s exact test for categorical data, as appropriate. The primary outcome, cumulative all-cause mortality at six months, was displayed using the Kaplan-Meier survival estimate. Multivariate regression analyses were performed using Cox proportional hazards models. The best subsets procedure was used as a variable reduction technique with the death endpoint to select variables utilized for multivariable adjustment. Additionally, the variables selected by the best subsets procedure also needed to be significant at p <0.05 to be included in the multivariable model. A Forest plot was used to graphically display hazard ratios, 95% confidence intervals (CI) and p-values. Because there was a violation of the proportional hazard’s assumption for the PERT designation, hazard ratios were estimated for the time-varying association of PERT with the risk of death within one month and after one month up to six months, as suggested by the Kaplan-Meier graph. All statistical tests were 2 sided, and a p-value of <0.05 was considered statistically significant. Analyses were carried out with SAS software (version 9.4, SAS institute, Cary, North Carolina).

RESULTS

In the pre-PERT cohort, there were 571 patients with ICD codes for PE that were screened for the study, 434 patients were excluded due to absence of acute PE, a diagnosis made at an outside facility, or if low-risk PE was found, and the remaining 137 patients met inclusion criteria. In the PERT cohort, the team was activated for 264 patients, 33 of whom were excluded (due to the absence of acute PE or low-risk PE), leaving 231 patients who met criteria. Thus, the total study population comprised 368 patients with massive or submassive PE.

Baseline clinical characteristics of pre-PERT and post-PERT patients are shown in Table 1. Table 2 describes presenting symptoms, physiological data, and cardiac biomarker concentration at the time of presentation with acute PE. Table 3 reports imaging characteristics and risk classification of patients. Characteristics of clinical interventions, including efficiency of care, admitting specialty service, and therapies administered are described in Table 4.

Table 1. Baseline Patient Characteristics:

Demographic and relevant baseline clinical characteristics in the pre-PERT and in the post-PERT eras. Continuous variables are presented as mean ± SD and differences between groups were evaluated by the student’s t-test. Dichotomous variables are presented as frequencies (% of population) and differences between groups were evaluated by Chi-square. COPD = Chronic obstructive pulmonary disease. VTE = Venous thromboembolism.

Variable	pre-PERT (N = 137)	post-PERT (N = 231)	p-Value
Age (years)	63.2 ±15.5	63.9 ±15.5	0.614
Female	66 (48%)	107 (46%)	0.730
Male	71 (52%)	124 (54%)	0.730
COPD	35 (26%)	46 (20%)	0.194
Heart Failure	19 (14%)	31 (13%)	0.903
Major Bleeding	11 (8%)	22 (10%)	0.610
Anticoagulant use	14 (10%)	20 (9%)	0.617
Antiplatelet use	35 (26%)	66 (29%)	0.530
Provoked	91 (66%)	119 (52%)	0.005
Provocation factors
Surgery	18 (13%)	26 (11%)	0.590
Cancer	38 (28%)	52 (23%)	0.259
Hypercoagulable	7 (5%)	4 (2%)	0.109
Immobilization	24 (18%)	25 (11%)	0.068
Prior VTE	22 (16%)	44 (19%)	0.470

Table 2. Presenting symptoms and clinical features.

Relevant clinical and laboratory data in the emergency department at the time of presentation with PE. Continuous variables are presented as mean ± SEM and differences between groups evaluated by the student’s t-test. Dichotomous variables are presented as frequencies (% of population) and differences between groups were evaluated by Chi-square. DVT = deep venous thrombosis. NTproBNP = N-terminal-pro brain natriuretic peptide.

Variable	pre-PERT (N = 137)	post-PERT (N = 231)	p-Value
Presenting Symptoms
Chest Pain	75 (55%)	54 (23%)	<0.001
Dyspnea	109 (80%)	175 (76%)	0.401
Syncope	17 (12%)	58 (25%)	0.003
DVT Symptoms	23 (17%)	18 (8%)	0.008
Hemoptysis	4 (3%)	8 (3%)	1.000
Vital Signs
Heart Rate (beats per min)	108.24 ±22.04	102.86 ±21.57	0.064
Systolic Blood Pressure (mmHg)	104.47 ±20.10	123.07 ±29.53	<0.001
Diastolic Blood Pressure (mmHg)	62.96 ±15.62	76.22 ±17.46	<0.001
Respiratory Rate (breaths per min)	24.31 ±6.12	21.40 ±5.85	<0.001
SpO2 (%)	90.00 ±7.24	93.18 ±6.04	<0.001
Plasma Cardiac Biomarkers
Troponin T (4th gen, ng/mL)	0.16 ±0.49	0.06 ±0.08	0.689
Troponin T (5th gen, high sensitivity, ng/L)	---	98.1 ±136.6	---
NT-proBNP (pg/mL)	3321 ±6599	3484 ± 5632	0.437

Table 3. Characteristics of PE at Presentation:

The location of thrombus in the pulmonary artery (PA), the presence of right ventricle (RV) enlargement or dysfunction, and the risk category of PE were determined in the emergency at the time of presentation. Variables are presented as frequencies (% of population) and differences between groups were evaluated by the Chi-squared test. AHA=American Heart Association. ACC=American College of Cardiology. ESC=European Society of Cardiology.

Variable	pre-PERT (N = 137)	post-PERT (N = 231)	p-Value
Thrombus Location in PA
Saddle	13 (10%)	76 (36%)	<0.001
Main	37 (30%)	50 (23%)	<0.001
Lobar	27 (22%)	51 (24%)	<0.001
Segmental	36 (29%)	31 (14%)	<0.001
Subsegmental	11 (9%)	6 (3%)	<0.001
RV Dysfunction
Enlargement and/or Dysfunction	90 (66%)	193 (84%)	<0.001
AHA/ACC Criteria
Massive PE	36 (26%)	38 (16%)	0.023
Submassive PE	101 (74%)	193 (84%)	0.023
ESC Criteria
Intermediate-Low Risk PE	62 (45%)	85 (37%)	0.002
Intermediate-High Risk PE	39 (28%)	108 (47%)	0.002
High Risk PE	36 (26%)	38 (16%)	0.002

Table 4. Emergency Department Management and Therapies Delivered for Acute PE:

Time for decision making and treatment in the ED are represented as as mean ± 95% C.I. and differences between groups evaluated by the student’s t-test. Patient disposition location and treatment are presented as frequencies (% of population) and differences between groups were evaluated by Chi-square. IVC = inferior vena cava. MICU = medical intensive care unit. CCU = cardiac care unit. SICU = surgical intensive care unit.

Variable	pre-PERT (N = 137)	post-PERT (N = 231)	p-Value
Efficiency of Care Metrics
Time Triage to Diagnosis (min)	283 (221; 455)	204 (117; 291)	<0.001
Time to Disposition (min)	360 (240; 490)	284 (166; 404)	<0.001
Time from Diagnosis to Anticoagulation (min)	90 (38; 173)	50 (18; 117)	<0.001
Patient Disposition
MICU	51 (38%)	143 (62%)	<0.001
Medicine	74 (56%)	45 (20%)	<0.001
CCU (medicine)	5 (4%)	28 (12%)	<0.001
Cardiac ICU (surgery)	1 (1%)	12 (5%)	<0.001
SICU	1 (1%)	1 (0%)	<0.001
ED observation only	1 (1%)	1 (0%)	<0.001
Therapies Delivered
Anticoagulation at diagnosis	134 (97.8%)	227 (98.3%)	0.756
Enoxaparin as initial anticoagulant	83 (60.6%)	157 (68.0%)	0.151
Anticoagulation at discharge	120 (88%)	202 (96%)	0.002
IVC filter	9 (7%)	14 (6%)	0.867
Systemic Thrombolysis	11 (8%)	23 (10%)	0.501
Catheter-based Therapy	1 (1%)	10 (4%)	0.058
Surgical Thrombectomy	0 (0%)	12 (5%)	0.006

Post-PERT, patients received more efficient care, with reduced time from triage to diagnosis of PE, from diagnosis to administration of anticoagulation, and from triage to disposition to an admitting service. Post-PERT, there was a significantly higher rate of patients admitted to the ICU or step-down setting (79% post-PERT compared to 44% pre-PERT, P<0.001) compared to a general medicine service. The number of patients undergoing surgical embolectomy was higher post-PERT (5% vs. 0%, P=0.006), and more patients were discharged on anticoagulation post-PERT.

A multivariate Cox proportional hazards regression model was developed to identify predictors of mortality. Of the baseline patient clinical characteristics, a history of heart failure, cancer, and/or major bleeding and presentation with chest pain, concomitant DVT, low systolic blood pressure (SBP, <94 mmHg), and/or cardiac arrest were identified as predictors of patient mortality (Figure 1). Of the characteristics of clinical interventions employed, time from triage to diagnosis of PE was the only independent variable predictive of mortality in the multivariate model (HR 1.05, 95% CI 1.00–1.09, p=0.034), and the relative risk of mortality through 6 months was reduced by 4.6% by each hour earlier the diagnosis of PE was made.

Figure 1. Baseline Characteristics and Clinical Features Predictive of Mortality in High Risk PE:

Cox multivariate regression analysis was employed to identity those clinical variables independently predicting patient mortality for submassive and massive PE. Data are represented as a Forest Plot with Hazard Ratios (HR) ± 95% C.I. with P values as noted. SBP = Systolic blood pressure.

In a multivariate model adjusted only for the presence of massive vs. submassive PE, anticoagulation use alone was associated with reduced mortality (HR 0.26, 95% CI 0.09–0.72). Similarly, enoxaparin as the initial anticoagulant of choice was also associated with reduced mortality (HR 0.58, 95% CI 0.35–0.97, p=0.038) when adjusted for the presence of massive vs. submassive PE. Neither ICU level care nor advanced interventions (surgical embolectomy, systemic thrombolysis, catheter-directed therapies, catecholamines) were statistically-significant independent predictors of mortality following PERT implementation (Figure 2), although, notably, there were no mortalities amoung patients who underwent surgical embolectomy.

Figure 2. Patient Management Strategies Predicting Mortality in High Risk PE:

Cox multivariate regression analysis was employed to identity patient management strategies or treatment decisions that independently predict patient mortality for submassive and massive PE. Data are represented as a Forest Plot with Hazard Ratios (HR) ± 95% C.I. and P values as noted.

Kaplan-Meier survival analyses were performed comparing pre-PERT vs. post-PERT mortality through 6 months after diagnosis. Mortality was significantly lower in the post-PERT cohort through 6 months (14% vs. 24% Pre-PERT, HR 0.583, RRR 41.7%, ARR 10%, log-rank p=0.025) (Figure 3). Adjusted hazard ratios revealed significant predictors of mortality by a multivariate Cox proportional hazards regression model (Figure 3). Testing the proportional hazard assumption revealed a significant interaction between PERT intervention and follow-up time and thus revealing a violation of the proportional hazards assumption. Therefore, hazard ratios were estimated through 1 month after presentation, which showed no difference in the risk of mortality (HR 1.11, 95% CI 0.55–2.26, p=0.766). However, between 1 month and 6 months after presentation with acute PE, the post-PERT cohort had a sustained reduction in mortality compared to pre-PERT (HR 0.42, 95% CI 0.19–0.95, p=0.037).

Figure 3. Effect of a PERT on Patient Survival with High Risk PE:

Kaplan-Meier survival estimates through 6 months after initial diagnosis with the difference between groups evaluated using the log rank test. The submassive and massive PE groups were combined with cumulative probability of mortality during the noted time periods. Hazard ratios (HR) ± 95% C.I. and P values are noted for specific time periods and derived from the multivariate Cox proportional hazards regression model with adjustments for the following variables: systolic blood pressure, chest pain, cardiac arrest, heart failure history, major bleeding, and DVT symptoms.

In a mortality analysis stratified separately by AHA/ACC risk classification of massive and submassive PE, there was a trend towards a reduction in the risk of mortality that did not reach statistical significance in patients with submassive PE (10% vs. 19% Pre-PERT, hazard ratio 0.56, p=0.071). There was no difference in the risk of mortality in patients with massive PE following implementation of a PERT (32% vs. 39% Pre-PERT, HR 0.85, p=0.672). However, this study was underpowered for examination of subgroup effects (p-value for PERT-by-PE classification interaction p=0.411; Supplemental Figure 2)

In regards to secondary and safety outcomes with PERT, there were no significant differences in hemodynamic collapse or major bleeding compared to pre-PERT. There was a significant increase in the use of catecholamines post-PERT (9% vs 4% pre-PERT, HR 2.25, p=0.049), and no differences in the rates of endotracheal intubation, emergency thrombolysis, cardiopulmonary resuscitation (CPR) or extracorporeal membrane oxygenation (ECMO) were observed. The use of a multi-disciplinary PERT led to a reduced length of hospital stay (6.5 ± 9.8 days post-PERT vs 9.1 ± 10.8 days pre-PERT, p=0.007 (Supplemental table 1).

DISCUSSION

Multi-disciplinary management of patients with intermediate-risk and high-risk PE was associated with a clearly-sustained reduction in mortality at 6 months. Our aim was to identify a particular management strategy responsible for persistently-favorable patient outcomes that could translate to improved care. Features embedded in the structure of our multidisciplinary PERT may be responsible for the observed reduction in mortality. In the univariate analysis, with PERT, there were significantly reduced times from triage to diagnosis, diagnosis to anticoagulation, and ED triage to hospital admission, increased utilization of advanced therapies, increased use of enoxaparin as the initial anticoagulant, and higher rates of admission to an ICU or step-down level of care. With the exception of time from triage to hospital admission, each of these metrics improved with PERT when evaluated by multivariate analysis. However, the primary clinical intervention independently associated with mortality was time from patient triage to the diagnosis of PE. This is likely resulted in earlier initiation of treatment, and earlier admission to a setting with more intensive monitoring for hemodynamic consequences of PE. In the first half of the pre-PERT era, the CCU and cardiac ICU exclusively cared for all patients with massive and submassive PE which was a decision requested of those services and this may have lead to an emphasis on hemodynamic monitoring and management going forth that may be reflected in improved mortality observed.

Frequent educational sessions on high risk PE management in our institution were concurrent with implementation of PERT and favorable Likert scale responses to evaluate successful training¹⁴. Enhanced awareness and a positive cultural change toward the care of PE may be responsible the reduction in the time to diagnose submassive and massive PE following PERT implementation. Earlier identification of PE and expeditious treatment of PE using enoxaparin were striking features of our study and idenpendent preditors of survival following PERT implementation. This observation is analogous to the “door-to-balloon time” in patients with ST-segment elevation myocardial infarction, or a “door-to-needle” time in patients with acute ischemic stroke^15,16. Our PERT pharmacist recommended immediate enoxaparin administration unless contraindicated due to its favorable pharmacokinetic profile and expeditious therapeutic effect. There was a trend toward increased use of enoxaparin post-PERT, and in an analysis adjusted for differences in massive vs submassive PEs, enoxaparin use was associated with a 42% reduced risk of mortality.

Several other institutions have published outcomes comparing patients before and after PERT. One study at the Massachusetts General Hospital of 440 patients before and after PERT showed no mortality benefit⁹. Another study of 769 patients with PE before and after implementation of PERT at the Cleveland Clinic demonstrated a reduction in both inpatient and 30-day mortality following PERT implementation¹⁰. Both studies were limited to only short-term follow-up. A third analysis of 554 patients with PE of all risk-types at the University of Virginia demonstrated reduced mortality of patients in the PERT era at 6 months compared to the pre PERT era, although no difference within the PERT era between patients who received a PERT activation compared with those who were not evaluated by PERT, however there were significant differences between the groups¹¹. Unlike the present study, the latter two did not use multivariate analysis or adjust for baseline clinical characteristics to account for the possibility of counfounders.

Consistent with other studies, we observed that PERT implementation increased the use of advanced therapies in patients with submassive and massive PE¹². We report, however, that utilizing advanced therapies was not a variable independently associated with improved patient mortality. This contrasts a recent multivariate analysis in which use of advanced therapies was associated with a 61% reduction in short-term mortality¹⁷.

There are limitations to this study. As an observational analysis, there is a risk of inherent biases. The number of PEs diagnosed post-PERT implementation was significantly greater than that in pre-PERT, intuitively due to heightened diagnostic suspicion and testing for PE as a result of the educational influence of PERT in patients in the ED whose symptoms were previously attributed to other diagnoses. Greater use of echocardiography and cardiac biomarkers in the post-PERT era may have identified more patients with (submassive) PE, possibly explaining a higher rate of submassive PE post-PERT. Additionally, while in the pre-PERT cohort, all patients with ICD codes for PE were screened for inclusion, in the post-PERT group only patients with PE for whom PERT was activated were screened for inclusion. This raises the possibility that there were patients with submassive or massive PE for whom PERT was not activated that are missing from the analysis. In addition, we were unable to adjudicate the cause of death, and differences in cardiovascular mortality and complications from PE compared to mortality from associated comorbidities could not be delineated.

In conclusion, we demonstrated a sustained long-term mortality benefit of PERT by utilizing a multivariate model to adjust hazard ratio estimates by differences in baseline patient characteristics. More rapid diagnosis of PE may be a contributor to the marked and sustained reduction in patient mortality post-PERT. Future studies are required to confirm these observations and to elucidate how PERTs can improve the care of patients with submassive and massive PE.