A machine learning model for diagnosing acute pulmonary embolism and comparison with Wells score, revised Geneva score, and Years algorithm

BACKGROUND

Acute pulmonary embolism (APE) is a fatal cardiovascular disease, yet missed diagnosis and misdiagnosis often occur due to non-specific symptoms and signs. A simple, objective technique will help clinicians make a quick and precise diagnosis. In population studies, machine learning (ML) plays a critical role in characterizing cardiovascular risks, predicting outcomes, and identifying biomarkers. This work sought to develop an ML model for helping APE diagnosis and compare it against current clinical probability assessment models.

METHODS

This is a single-center retrospective study. Patients with suspected APE were continuously enrolled and randomly divided into two groups including training and testing sets. A total of 8 ML models, including random forest (RF), Naïve Bayes, decision tree, K-nearest neighbors, logistic regression, multi-layer perceptron, support vector machine, and gradient boosting decision tree were developed based on the training set to diagnose APE. Thereafter, the model with the best diagnostic performance was selected and evaluated against the current clinical assessment strategies, including the Wells score, revised Geneva score, and Years algorithm. Eventually, the ML model was internally validated to assess the diagnostic performance using receiver operating characteristic (ROC) analysis.

RESULTS

The ML models were constructed using eight clinical features, including D-dimer, cardiac troponin T (cTNT), arterial oxygen saturation, heart rate, chest pain, lower limb pain, hemoptysis, and chronic heart failure. Among eight ML models, the RF model achieved the best performance with the highest area under the curve (AUC) (AUC = 0.774). Compared to the current clinical assessment strategies, the RF model outperformed the Wells score ( P = 0.030) and was not inferior to any other clinical probability assessment strategy. The AUC of the RF model for diagnosing APE onset in internal validation set was 0.726.

CONCLUSIONS

Based on RF algorithm, a novel prediction model was finally constructed for APE diagnosis. When compared to the current clinical assessment strategies, the RF model achieved better diagnostic efficacy and accuracy. Therefore, the ML algorithm can be a useful tool in assisting with the diagnosis of APE.

Diagnostic Yield, Radiation Exposure, and the Role of Clinical Decision Rules to Limit Computed Tomographic Pulmonary Angiography-Associated Complications

OBJECTIVES

Computed tomographic pulmonary angiography (CT-PA) is associated with significant cost, contrast, and radiation exposure. Clinical decision rules (CDRs) reduce the need for diagnostic imaging; however, their utility in the medical intensive care unit (MICU) remains unknown. We explored the diagnostic yield and complications associated with CT-PA (radiation exposure and contrast-induced acute kidney injury [AKI]) while investigating the efficacy of CDRs to reduce unnecessary testing.

METHODS

All CT-PAs performed in an academic MICU for 4 years were retrospectively reviewed. The Wells and revised Geneva scores (CDRs) and radiation dose per CT-PA were calculated, and the incidence of post-CT-PA AKI was recorded.

RESULTS

A total of 439 studies were analyzed; the diagnostic yield was 11% (48 PEs). Positive CT-PAs were associated with a higher Wells score (5.8 versus 3.2, P < 0.001), but similar revised Geneva scores (6.4 versus 6.0, P = 0.32). A Wells score of ≥4 had a positive likelihood ratio of 2.1 with a negative predictive value of 98.2. More than half (88.9%) of patients with a Wells score of ≤4 developed an AKI, with 55.6% of those having recovery of renal function.

CONCLUSIONS

There is overutilization of CT-PA in the MICU. The Wells score retains its negative predictive value in critically ill adult patients and may aid to limit radiation exposure and contrast-induced AKI in MICU.

Deep Learning-Based Algorithm for Automatic Detection of Pulmonary Embolism in Chest CT Angiograms

Purpose: Since the prompt recognition of acute pulmonary embolism (PE) and the immediate initiation of treatment can significantly reduce the risk of death, we developed a deep learning (DL)-based application aimed to automatically detect PEs on chest computed tomography angiograms (CTAs) and alert radiologists for an urgent interpretation. Convolutional neural networks (CNNs) were used to design the application. The associated algorithm used a hybrid 3D/2D UNet topology. The training phase was performed on datasets adequately distributed in terms of vendors, patient age, slice thickness, and kVp. The objective of this study was to validate the performance of the algorithm in detecting suspected PEs on CTAs. Methods: The validation dataset included 387 anonymized real-world chest CTAs from multiple clinical sites (228 U.S. cities). The data were acquired on 41 different scanner models from five different scanner makers. The ground truth (presence or absence of PE on CTA images) was established by three independent U.S. board-certified radiologists. Results: The algorithm correctly identified 170 of 186 exams positive for PE (sensitivity 91.4% [95% CI: 86.4–95.0%]) and 184 of 201 exams negative for PE (specificity 91.5% [95% CI: 86.8–95.0%]), leading to an accuracy of 91.5%. False negative cases were either chronic PEs or PEs at the limit of subsegmental arteries and close to partial volume effect artifacts. Most of the false positive findings were due to contrast agent-related fluid artifacts, pulmonary veins, and lymph nodes. Conclusions: The DL-based algorithm has a high degree of diagnostic accuracy with balanced sensitivity and specificity for the detection of PE on CTAs.

Factors influencing clinicians' willingness to use an AI-based clinical decision support system

Background

Given the opportunities created by artificial intelligence (AI) based decision support systems in healthcare, the vital question is whether clinicians are willing to use this technology as an integral part of clinical workflow.

Purpose

This study leverages validated questions to formulate an online survey and consequently explore cognitive human factors influencing clinicians' intention to use an AI-based Blood Utilization Calculator (BUC), an AI system embedded in the electronic health record that delivers data-driven personalized recommendations for the number of packed red blood cells to transfuse for a given patient.

Method

A purposeful sampling strategy was used to exclusively include BUC users who are clinicians in a university hospital in Wisconsin. We recruited 119 BUC users who completed the entire survey. We leveraged structural equation modeling to capture the direct and indirect effects of “AI Perception” and “Expectancy” on clinicians' Intention to use the technology when mediated by “Perceived Risk”.

Results

The findings indicate a significant negative relationship concerning the direct impact of AI's perception on BUC Risk (ß = −0.23, p < 0.001). Similarly, Expectancy had a significant negative effect on Risk (ß = −0.49, p < 0.001). We also noted a significant negative impact of Risk on the Intent to use BUC (ß = −0.34, p < 0.001). Regarding the indirect effect of Expectancy on the Intent to Use BUC, the findings show a significant positive impact mediated by Risk (ß = 0.17, p = 0.004). The study noted a significant positive and indirect effect of AI Perception on the Intent to Use BUC when mediated by risk (ß = 0.08, p = 0.027). Overall, this study demonstrated the influences of expectancy, perceived risk, and perception of AI on clinicians' intent to use BUC (an AI system). AI developers need to emphasize the benefits of AI technology, ensure ease of use (effort expectancy), clarify the system's potential (performance expectancy), and minimize the risk perceptions by improving the overall design.

Conclusion

Identifying the factors that determine clinicians' intent to use AI-based decision support systems can help improve technology adoption and use in the healthcare domain. Enhanced and safe adoption of AI can uplift the overall care process and help standardize clinical decisions and procedures. An improved AI adoption in healthcare will help clinicians share their everyday clinical workload and make critical decisions.

Automated detection of pulmonary embolism from CT-angiograms using deep learning

Background

The aim of this study was to develop and evaluate a deep neural network model in the automated detection of pulmonary embolism (PE) from computed tomography pulmonary angiograms (CTPAs) using only weakly labelled training data.

Methods

We developed a deep neural network model consisting of two parts: a convolutional neural network architecture called InceptionResNet V2 and a long-short term memory network to process whole CTPA stacks as sequences of slices. Two versions of the model were created using either chest X-rays (Model A) or natural images (Model B) as pre-training data. We retrospectively collected 600 CTPAs to use in training and validation and 200 CTPAs to use in testing. CTPAs were annotated only with binary labels on both stack- and slice-based levels. Performance of the models was evaluated with ROC and precision–recall curves, specificity, sensitivity, accuracy, as well as positive and negative predictive values.

Results

Both models performed well on both stack- and slice-based levels. On the stack-based level, Model A reached specificity and sensitivity of 93.5% and 86.6%, respectively, outperforming Model B slightly (specificity 90.7% and sensitivity 83.5%). However, the difference between their ROC AUC scores was not statistically significant (0.94 vs 0.91, p = 0.07).

Conclusions

We show that a deep learning model trained with a relatively small, weakly annotated dataset can achieve excellent performance results in detecting PE from CTPAs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12880-022-00763-z.

Adherence to Best Practice Advice for Diagnosis of Pulmonary Embolism

PURPOSE

This study evaluated clinician adherence to the American College of Physicians Best Practice Advice for diagnosis of pulmonary embolism.

DESIGN

A prospective, single-center, descriptive design was utilized.

METHODS

A heterogeneous sample of 111 hemodynamically stable adult inpatients with a computed tomography pulmonary angiogram ordered was consented. Electronic medical records were reviewed for demographic and clinical variables to determine adherence. The 6 individual best practice statements and the overall adherence were evaluated by taking the sum of "yes" answers divided by the sample size.

RESULTS

Overall adherence was 0%. Partial adherence was observed with clinician-recorded clinical decisions rules and obtaining d-dimer (3.6% [4/111] and 10.2% [9/88], respectively) of low/intermediate probability scorers. Age adjustment of d-dimer was not recorded. Computed tomography pulmonary angiogram was the first diagnostic test in 89.7% (79/88) in low/intermediate probability patients.

CONCLUSION

In hemodynamically stable, hospitalized adults, adherence to best practice guidelines for diagnosis of pulmonary embolism was minimal. Clinical utility of the guidelines in hospitalized adults needs further evaluation. Systems problems (eg, lack of standardized orders, age-adjusted d-dimer values, information technology support) likely contributed to poor guideline adherence.

Serial use of existing clinical decisions aids can reduce computed tomography pulmonary angiography for pulmonary embolism

Pulmonary embolism (PE) remains a diagnostic challenge in emergency medicine. Clinical decision aids (CDAs) like the Pulmonary Embolism Rule-Out Criteria (PERC) are sensitive but poorly specific; serial CDA use may improve specificity. The goal of this before-and-after study was to determine if serial use of existing CDAs in a novel diagnostic algorithm safely decreases the use of CT pulmonary angiograms (CTPA). This was a retrospective before-and-after study conducted at an urban ED with 105,000 annual visits. Our algorithm uses PERC, Wells' score, and D-dimer in series, before moving to CTPA. The algorithm was introduced in January, 2017. Use of CDAs and D-dimer in the 24 months pre- and 12 months post-intervention were obtained by chart review. The algorithm's effect on CTPA ordering was assessed by comparing volume 5 years pre- and 3 years post-intervention, adjusted for ED volume. Mean CTPAs per 1000 adult ED visits was 11.1 in the 5 pre-intervention years and 9.9 in the 3 post-intervention years (p < 0.0001). Use of PERC, Wells' score and D-dimer increased from 1.1%, 1.1%, and 28% to 8.8% (p = 0.0002) 8.1% (p = 0.0005), and 35% (p = 0.0066), respectively. Pre-intervention, there were six potentially missed PEs compared to three in the post-intervention period. Introduction of our serial CDA diagnostic algorithm was associated with increased use of CDAs and D-dimer and reduced CTPA rate without an apparent increase in the number of missed PEs. Prospective validation is needed to confirm these results.

Development and validation of a constipation treatment toolkit for patients on hemodialysis

INTRODUCTION

The cause of constipation is multifactorial and common problem for patients on hemodialysis. A lack of strong evidence on suitable treatment strategies means there is an unorganized approach to selecting therapies, which can exacerbate constipation or worsen symptoms. Clinicians and patients would benefit from a content and face validated treatment algorithm for treating constipation. In this study, our objective was to develop and content and face validate a constipation treatment toolkit for patients on hemodialysis, consisting of treatment algorithm, and patient information tools (pamphlet and video).

METHODS

Literature searches were performed to develop an initial toolkit using Lynn's method for developing content-valid clinical tools. Content and face validity were evaluated as per Lynn's method for determining content validity; the algorithm was evaluated by Canadian nephrology clinicians, while patient information tools were evaluated by clinicians and patients. Components were rated on a Likert scale for content relevance and on a 5-point scale for face validity. After each round, the content validity index (CVI) score was calculated and revisions were made based on feedback.

FINDINGS

A total of 23 clinicians and 15 patients were interviewed across three validation rounds. After three rounds, the treatment algorithm achieved content (overall CVI = 0.93) and face (91% agreement) validity. Our patient information tools achieved content and face validity (pamphlet overall CVI = 0.99, 85.5% agreement; video overall CVI = 0.99, 90.5% agreement).

DISCUSSION

A treatment algorithm and patient information toolkit for the treatment of constipation in patients on hemodialysis were content and face validated via expert review. Further research will be needed to ascertain the effectiveness and implementation of this toolkit.

Electronic pulmonary embolism clinical decision support and effect on yield of computerized tomographic pulmonary angiography: ePE-A pragmatic prospective cohort study

OBJECTIVE

Multiple professional societies recommend pre-test probability (PTP) assessment prior to imaging in the evaluation of patients with suspected pulmonary embolism (PE), however, PTP testing remains uncommon, with imaging occurring frequently and rates of confirmed PE remaining low. The goal of this study was to assess the impact of a clinical decision support tool embedded into the electronic health record to improve the diagnostic yield of computerized tomography pulmonary angiography (CTPA) in suspected patients with PE in the emergency department (ED).

METHODS

Between July 24, 2014 and December 31, 2016, 4 hospitals from a healthcare system embedded an optional electronic clinical decision support system to assist in the diagnosis of pulmonary embolism (ePE). This system employs the Pulmonary Embolism Rule-out Criteria (PERC) and revised Geneva Score (RGS) in series prior to CT imaging. We compared the diagnostic yield of CTPA) among patients for whom the physician opted to use ePE versus the diagnostic yield of CTPA when ePE was not used.

RESULTS

During the 2.5-year study period, 37,288 adult patients were eligible and included for study evaluation. Of eligible patients, 1949 of 37,288 (5.2%) were enrolled by activation of the tool. A total of 16,526 CTPAs were performed system-wide. When ePE was not engaged, CTPA was positive for PE in 1556 of 15,546 scans for a positive yield of 10.0%. When ePE was used, CTPA identified PE in 211 of 980 scans (21.5% yield) (P < 0.001).

CONCLUSIONS

ePE significantly increased the diagnostic yield of CTPA without missing 30-day clinically overt PE.

Applying Wells score to inconclusive perfusion only modified PIOPED II (Prospective Investigation of Pulmonary Embolism Diagnosis II) readings in order to optimize the lung scintigraphy diagnostic yield in acute pulmonary embolism detection

OBJECTIVE

When using perfusion only modified PIOPED II criteria for PE detection, generated non-diagnostic scans are found to be the main diagnostic restriction. The objective of current study is to identify the role of Wells criteria added to inconclusive readings with the intent of enhancing the lung scintigraphy diagnostic yield.

METHODS

CTPA was performed in 34 suspected PE patients with inconclusive lung scintigraphy. They also were evaluated by Wells score and classified as low, intermediate and high probability. Overall prevalence and the rate of PE for each probability were calculated. Furthermore, NPV for scores < 2 and PPV for scores > 6 were computed.

RESULTS

Having a mean age of 59.75 ± 17.38 years, 7 (20.6%), 23 (67.6%) and 4 (11.8%) of cases had total criteria point count < 2, 2-6 and > 6, respectively. Using CTPA, 5 patients (14.7%) were diagnosed with PE. None of the patients with scores < 2 had PE with an associated NVP of 100%. Patients with scores 2-6 had a PE rate of 4.3% and 100% of patients with scores > 6 were diagnosed with PE, implying that the PPV of scores > 6 was 100%.

CONCLUSION

Adding Wells score to non-diagnostic scans allowed identification of PE to be done reliably, and provided further insight into how lung scintigraphy in conjunction with clinical assessment is a practical strategy not only for the patients unfit for performing CTPA but also in all the patients referred for PE evaluation.

DOES THE USE OF CONTEMPORARY CT SCANNERS ALTER THE RADIATION DOSE DEBATE IN THE IMAGING WORK UP FOR PULMONARY EMBOLISM?

The aim of this study was to compare patient doses from ventilation perfusion single photon emission computed tomography (V/Q SPECT) and computed tomography pulmonary angiography (CTPA) performed on contemporary scanners. Effective dose (ED) for V/Q SPECT was calculated using organ doses per unit administered activity of the radiopharmaceuticals. Organ doses in CT were measured using nanoDot aluminium oxide optically stimulated dosemeters placed within a female adult anthropomorphic phantom. To simulate a larger patient, the phantom was wrapped in three layers of Superflab sheets. The V/Q SPECT resulted in ED of 2.82 mSv and a breast dose of 1.12 mGy. The CTPA dose was 1.82 ± 0.42 and 3.43 ± 0.91 mSv, whilst dose to the breast tissue was 2.86 ± 0.86 and 5.95 ± 0.44 mGy for small- and medium-sized patients, respectively.

Diagnosis of pulmonary embolism: Following the evidence from suspicion to certainty

Accurate, timely and cost-effective identification of pulmonary embolism remains a diagnostic challenge. This article reviews the pulmonary embolism diagnostic process with a focus on the best practice advice from the American College of Physicians. Benefits and risks of each diagnostic step are discussed. Emerging diagnostic tools, not included in the algorithm, are briefly reviewed.

Optimization of computed tomography pulmonary angiography protocols using 3D printed model with simulation of pulmonary embolism

Background

Three-dimensional (3D) printing has been shown to accurately replicate anatomical structures and pathologies in complex cardiovascular disease. Application of 3D printed models to simulate pulmonary arteries and pulmonary embolism (PE) could assist development of computed tomography pulmonary angiography (CTPA) protocols with low radiation dose, however, this has not been studied in the literature. The aim of this study was to investigate optimal CTPA protocols for detection of PE based on a 3D printed pulmonary model.

Methods

A patient-specific 3D printed pulmonary artery model was generated with thrombus placed in both main pulmonary arteries to represent PE. The model was scanned with 128-slice dual-source CT with slice thickness of 1 and 0.5 mm reconstruction interval. The tube voltage was selected to range from 70, 80, 100 to 120 kVp, and pitch value from 0.9 to 2.2 and 3.2. Quantitative assessment of image quality in terms of signal-to-noise ratio (SNR) was measured in the main pulmonary arteries and within the thrombus regions to determine the relationship between image quality and scanning protocols. Both two-dimensional (2D) and 3D virtual intravascular endoscopy (VIE) images were generated to demonstrate pulmonary artery and thrombus appearances.

Results

PE was successfully simulated in the 3D printed pulmonary artery model. There were no significant differences in SNR measured in the main pulmonary arteries with 100 and 120 kVp CTPA protocols (P>0.05), regardless of pitch value used. SNR was significantly lower in the high-pitch 3.2 protocols when compared to other protocols using 70 and 80 kVp (P<0.05). There were no significant differences in SNR measured within the thrombus among the 100 and 120 kVp protocols (P>0.05). For low dose 70 and 80 kVp protocols, SNR was significantly lower in the high-pitch of 3.2 protocols than that in other protocols with different pitch values (P<0.01). 2D images showed the pulmonary arteries and thrombus clearly, while 3D VIE demonstrated intraluminal appearances of pulmonary wall and thrombus in all protocols, except for the 70 kVp and pitch 3.2 protocol, with visualization of thrombus and pulmonary artery wall affected by artifact associated with high image noise. Radiation dose was reduced by up to 80% when lowering kVp from 120 to 100 and 80 kVp with use of 3.2 high-pitch protocol, without significantly affecting image quality.

Conclusions

Low-dose CT pulmonary angiography can be achieved with use of low kVp (80 and 100) and high-pitch protocol with significant reduction in radiation dose while maintaining diagnostic images of PE. Use of high pitch, 3.2 in 70 kVp protocol should be avoided due to high image noise and poorer quality.

Long-Term Experience With a Mandatory Clinical Decision Rule and Mandatory d-Dimer in the Evaluation of Suspected Pulmonary Embolism

PURPOSE

This study evaluated the long-term effectiveness of mandatory assignment of both a clinical decision rule (CDR) and highly sensitive d-dimer in the evaluation of patients with suspected pulmonary embolism (PE).

MATERIALS AND METHODS

Institutional guidelines with a CDR and highly sensitive d-dimer were embedded in an order entry menu with mandatory assignment of key components before ordering a CT pulmonary angiogram (CTPA). Data were retrospectively extracted from the electronic health record.

RESULTS

This was a retrospective review of 1,003 CTPA studies (905 patients, 845 male and 60 female patients, age 63.7 ± 13.5 years). CTPAs were positive for PE in 170 studies (17%), representing an average yield of 15% (year [average]; 2007 [15%], 2008 [18%], 2009 [15%], 2010 [15%], 2011 [17%], 2012 [15%], 2013 [23%]). The increased yield represented efforts of mandatory order entry assignments, educational sessions, and clinical champions. Different d-dimer thresholds with or without age adjustments in combination with the CDR identified about 10% of patients who may have been managed without CTPA.

CONCLUSION

Mandatory assignment of a CDR and highly sensitive d-dimer clinical decision pathway can be successfully incorporated into an order entry menu and produce a sustained increase in CTPA yield of patients with suspected PE.

Limitations of an automated embolism segmentation method in clinical practice

PURPOSE

Automated pulmonary embolism (PE) segmentation is frequently used as a preprocessing step in the quantitative analysis of pulmonary embolism. Objective of this study is to analyze the potential limitation in automated PE segmentation using clinical cases.

METHODS

A database of 304 computer tomography pulmonary angiography (CTPA) examinations was collected and confirmed to be PE. After processing using an automated scheme, two radiologists classified these cases into four groups of A, B, C and D, which represent 4 different segmentation results namely, (1) entire pulmonary artery identified without motivation artifacts, (2) entire pulmonary artery identified with motivation artifacts, (3) part of the pulmonary artery identified, and (4) none of the pulmonary artery identified. Then, the possible failed reasons in PE segmentation were analyzed and determined based on the image characterization of the diseases and the applied CTPA scanning protocols.

RESULTS

In the study, 143 (47.0%., 30 (9.9%., 110 (36.2%. and 21 (6.9%. examinations were classified into groups A, B, C and D, respectively. Group C and D included the cases with failed segmentation. Fifteen failure reasons, including intrapulmonary abnormalities, extra-pulmonary abnormalities, diffuse pulmonary diseases, enlarged heart, absolute occluded vessels, embolism attached to artery wall, delayed scan time, skewed location, low scan dose, obvious artifact of superior vena cava, previous chest surgery, congenital deformities of the chest, incorrect positioning, missed images and other unknown reasons, were determined with corresponding case percentages ranging from 0.3%.o 9.2%.

CONCLUSIONS

Automated segmentation failures were caused by specific lung diseases, anatomy varieties, improper scan time, improper scan dose, manual errors or other unknown reasons. Realization of those limitations is crucial for developing robust automated schemes to handle these issues in a single pass when a large number of CTPA examinations need to be analyzed.

Pulmonary embolism rule-out criteria (PERC) rule in European patients with low implicit clinical probability (PERCEPIC): a multicentre, prospective, observational study

BACKGROUND

The ability of the pulmonary embolism rule-out criteria (PERC) to exclude pulmonary embolism without further testing remains debated outside the USA, especially in the population with suspected pulmonary embolism who have a high prevalence of the condition. Our main objective was to prospectively assess the predictive value of negative PERC to rule out pulmonary embolism among European patients with low implicit clinical probability.

METHODS

We did a multicentre, prospective, observational study in 12 emergency departments in France and Belgium. We included consecutive patients aged 18 years or older with suspected pulmonary embolism. Patients were excluded if they had already been hospitalised for more than 2 days, had curative anticoagulant therapy in progress for more than 48 h, or had a diagnosis of thromboembolic disease documented before admission to emergency department. Physicians completed a standardised case report form comprising implicit clinical probability assessment (low, moderate, or high) and a list of risk factors including criteria of the PERC rule. They were asked to follow international recommendations for diagnostic strategy, masked to PERC assessment. The primary endpoint was the proportion of patients with low implicit clinical probability and negative PERC who had venous thromboembolic events, diagnosed during initial diagnostic work-up or during 3-month follow-up, as externally adjudicated by an independent committee masked to the PERC and clinical probability assessment. The upper limit of the 95% CI around the 3-month thromboembolic risk was set at 3%. We did all analyses by intention to treat, including all patients with complete follow-up. This trial is registered with ClinicalTrials.gov, number NCT02360540.

FINDINGS

Between May 1, 2015, and April 30, 2016, 1773 consecutive patients with suspected pulmonary embolism were prospectively assessed for inclusion, of whom 1757 were included. 1052 (60%) patients were classed as having low clinical probability, 49 (4·7%, 95% CI 3·5-6·1) of whom had a venous thromboembolic event. In patients with a low implicit clinical probability, 337 (32%) patients had negative PERC, of whom four (1·2%; 95% CI 0·4-2·9) went on to have a pulmonary embolism.

INTERPRETATION

In European patients with low implicit clinical probability, PERC can exclude pulmonary embolism with a low percentage of false-negative results. The results of our prospective, observational study allow and justify an implementation study of the PERC rule in Europe.

FUNDING

French Ministry of Health.

Pulmonary Embolism Testing Among Emergency Department Patients Who Are Pulmonary Embolism Rule-out Criteria Negative

OBJECTIVE

Previous studies have demonstrated that rates of pulmonary embolism (PE) testing have increased without a concomitant decrease in PE-related mortality. The Pulmonary Embolism Rule-out Criteria (PERC) intend to reduce testing for PE in the emergency department (ED) by identifying low-risk patients ("PERC-negative") who do not require D-dimer, computed tomography pulmonary angiogram (CTPA), or ventilation/perfusion (VQ) scan for PE. This study assesses PE testing rates among PERC-negative patients presenting to an urban academic ED.

METHODS

We prospectively enrolled a convenience sample of ED patients with chest pain and/or shortness of breath presenting between June 2010 and December 2015. We recorded baseline variables at the time of ED presentation, information on testing performed in the ED, and the diagnosis of acute PE during the ED visit. We classified patients as PERC-positive or PERC-negative utilizing baseline variables and clinical characteristics.

RESULTS

Of the 3,024 study patients, 54.8% (95% confidence interval = 53%-56.5%) were female and the mean age was 51.7 (51.1-52.3) years. A total of 17.5% (16.2%-18.9%) of study patients were PERC-negative and 33.7% (32%-35.4%) of all patients underwent testing for PE. A total of 25.5% (22%-29.4%) of PERC-negative patients had PE testing compared to 35.4% (33.6%-37.3%) of PERC-positive patients (p < 0.001). A total of 7.2% (5.3%-9.7%) of PERC-negative patients had advanced imaging without a D-dimer compared to 19.2% (17.8%-20.8%) of PERC-positive patients (p < 0.001). In multivariate analysis, factors associated with PE testing in PERC-negative patients included age, white non-Hispanic race/ethnicity, pleuritic chest pain, and a complaint of both chest pain and shortness of breath. Two PERC-negative patients (0.4%) were diagnosed with an acute PE in the ED compared to 2.2% of PERC-positive patients (p = 0.008). The overall testing yield for PE was 1.6% (0.4%-9.2%) among PERC-negative patients versus 6.3% (4.9%-8.1%) among PERC-positive patients (p = 0.017).

CONCLUSION

In an academic ED, a significant proportion of PERC-negative patients underwent testing for PE, including CT or VQ scan without D-dimer risk stratification. Future areas of research may include evaluating factors that lead clinicians to pursue PE testing in PERC-negative patients and implementing clinical pathways to minimize practice variability among these patients.