CDR Blog

Wells Criteria for PE

December 13, 2020
Vivian Au, Alexa Caldwell, Malini Hu, John Kim, Veronica Stewart

Discussion

The Wells’ score for pulmonary embolism (PE) is a clinical decision rule (CDR) used to determine the pre-test probability of PE and was outlined in a 2000 paper by Wells et al.¹ The Wells’ score is used to guide further testing and can be used to determine the need for D-dimer testing and CT pulmonary angiogram (CT-PA) imaging. This simplified scoring system builds upon previous clinical models developed by Wells to aid in diagnosis of suspected venous thrombosis, including both deep vein thrombosis (DVT) and PE, in 1995² and 1998³. The original 2000 article, which derives the Wells’ score from the clinical model outlined in the 1998 paper and validates it within the same paper using a prospective cohort study design.

In our critical appraisal of the Wells’ score for PE, based on the guidelines established by McGinn et al. (2000)⁴, we found that the Wells’ score is a well-derived CDR that can be used to efficiently predict the likelihood of PE based on a well-defined set of easily measured clinical criteria. The Wells’ score for PE satisfied all of the methodological standards for derivation of a CDR, except for the fact that it did not specify if and how the assessors were blinded. Therefore, we found that this CDR improves upon previously proposed algorithms for PE by incorporating only clinical variables to determine the need for D-dimer measurement or CT angiography. This significantly increases its utility in a clinical setting while also facilitating more efficient utilization of resources, such as imaging.

Next, we analyzed eight studies that aimed to validate the Wells’ score, including the original study from 2000, as that study also contained its own internal validation of the CDR using 20% of the study population (the other 80% of the study population having been used to derive the CDR). Our analysis of these studies is summarized in the table below.

All of the studies we analyzed found that the Wells’ score for PE showed significant accuracy in risk stratifying suspected PE patients. As a result, we can conclude that multiple studies have validated the Wells’ score, as well as variants of the CDR, as an effective tool in guiding management of patients with suspected PE. Overall, most prospective validation studies had large study populations across multiple centres, minimal loss to follow-up, and demonstrated group-stratified PE prevalence similar to that of the original Wells’ score. One flaw in these studies was a lack of information on the blinding used in the study. In addition, systematic reviews and meta-analyses examining the Wells’ score, as well as other CDRs used in the management of PE, have provided further evidence to support the Wells’ rule efficacy^12-13.

The Wells’ score for PE is a well-derived and well-validated CDR that can provide accurate prediction of PE risk in patients based on discrete clinical criteria, with studies showing equivalent or superior utility compared to other CDRs for assessment of PE. However, despite its efficacy, the Wells’ score is not without its flaws. One disadvantage of the Wells’ score that was consistently mentioned in several validation studies was the subjectiveness of one of the variables, “PE is the most likely diagnosis”, which can vary substantially based on a clinician’s individual assessment of a patient’s history, physical exam, and lab results. Another downside of these validation studies is that the majority of them involve patients in secondary or tertiary care settings; as a result, applicability of the Wells’ score in a primary care setting may be limited.

Several studies have examined the impact of implementing Wells’ criteria on the decision-making behaviour of physicians and on patient outcomes. Buntine et al. (2019) found that introducing a flowchart incorporating the Wells’ score, PERC rule, and age-adjusted D-dimer to emergency departments in three sites increased the yield rate of CT-PA and VQ scans from 9.9% to 16.5% while also decreasing the number of imaging tests ordered from 1815 to 1116, a 38.5% reduction¹⁴. Similarly, Walen et al. (2016) found that mandatory adherence to Wells’ score increased the diagnostic yield of CTPA scans, with a 6.6% increase in positive scans and 3.1% decrease in negative scans¹⁵. Another study by Sadeghi et al. (2016) at a single centre found that use of the Wells’ score could lead to a 20% reduction in CTPA scans¹⁶. In contrast, Rajagopalan et al. (2014) found that, in a single teaching hospital, excessive CTPA requests were still made despite the use of the Wells’ score to guide management¹⁷.

Studies have suggested that implementation of certain administrative interventions can lead to a reduction in imaging and increase in imaging yield. Deblois et al. (2017) found that the implementation of clinical decision support (CDS) and performance and feedback reports (PFRs) led to 8.3-25.4% and an increase in diagnostic yield of 3.4-4.4%¹⁸. Similarly, Yan et al. (2016) found that CDS could promote clinician adherence to the Wells’ score and nearly double CTPA yield¹⁹. Augmentation of the Wells’ score with additional bedside examinations may also increase its utility. A study by Nazerian et al. (2017) proposed that the diagnostic accuracy of the Wells’ score could be increased with the concurrent use of lung and venous ultrasound²⁰.

Based on our findings, we have decided to give the Wells’ criteria for PE an impact level of 1. Several studies have shown that the Wells’ score can be used in a variety of settings, guide patient management, increase imaging yield, and reduce the number of unnecessary tests ordered. However, further research is required to analyze the impact of the Wells’ score in reducing unnecessary imaging and improving patient outcomes.

Authors

Vivian Au, BHSc, MD candidate at McMaster University (c2021). She is interested in psychiatry and a notorious lover of comfy blankets.

Alexa Caldwell, (H)BSc, MD candidate at McMaster University (c2021). She is interested in anesthesiology, critical care, med ed, and salted caramel cookies.

Malini Hu, BSc, MD candidate at McMaster University (c2021). Her hobbies include painting landscapes, practicing her latte art skills and dreaming of the day she can watch Rafael Nadal play in the French Open clay courts in person.

John Kim, BSc MSc, MD candidate at McMaster University (c2021). He is interested in urology, surgery, and clinical research. When not playing the guitar, he can be found hitting the gym or watching classic sci fi movies.

Veronica Stewart, Hons BSc, MD candidate at McMaster University (c2021). She is interested in general surgery, emergency medicine and critical care. Outside of the hospital she can usually be found drinking tea and reading mystery novels.