YEARS algorithm for pulmonary embolism (PE)

Clinical background

Suspected acute pulmonary embolism is one of the commonest diagnostic problems in the emergency department, and the majority of patients investigated with CT pulmonary angiography do not have an embolism. The need for a simple, safe strategy to reduce unnecessary imaging is therefore considerable. Traditional algorithms rest on the Wells score or the revised Geneva score combined with a fixed D-dimer threshold of 500 ng/mL FEU, but these still result in a substantial proportion of patients being referred for CT pulmonary angiography unnecessarily. The YEARS algorithm was developed to address this by combining three simple clinical criteria with a differentiated D-dimer threshold: a higher threshold (1000 ng/mL FEU) for patients with no YEARS criteria and the conventional threshold (500 ng/mL FEU) for patients meeting one or more criteria. The aim is to maximise the proportion of patients in whom PE can safely be excluded without imaging, without compromising safety.

Applying the YEARS algorithm

The algorithm rests on three clinical criteria assessed at the bedside:

1. Clinical signs of deep vein thrombosis (leg swelling, asymmetric pain)
2. Haemoptysis
3. Pulmonary embolism is the most likely diagnosis (clinical probability judgement)

The number of criteria met (0 to 3) is combined with a single D-dimer measurement according to the following logic:

$$\text{Decision} = \begin{cases} \text{PE excluded if } n = 0 \text{ and D-dimer} < 1000 \text{ ng/mL FEU} \ \text{PE excluded if } n \geq 1 \text{ and D-dimer} < 500 \text{ ng/mL FEU} \ \text{CT pulmonary angiography required otherwise} \end{cases}$$

where $n$ is the number of YEARS criteria met.

The derivation study (the YEARS study) was a prospective multicentre cohort study in 12 hospitals in the Netherlands from October 2013 to July 2015 [1]. Of 3,616 patients screened, 151 (4%) were excluded, leaving 3,465 patients for evaluation. PE was diagnosed in 456 patients (13%) at baseline. Of the 2,946 patients (85%) in whom PE was excluded and who were not treated, 18 developed symptomatic venous thromboembolism during 3-month follow-up, corresponding to a failure rate of 0.61% (95% CI 0.36 to 0.96), of whom 6 had fatal PE (0.20%, 95% CI 0.07 to 0.44). CT pulmonary angiography could be avoided in 1,651 patients (48%) with the YEARS algorithm, compared with 1,174 patients (34%) had the Wells score with a fixed D-dimer threshold of 500 ng/mL FEU been applied, an absolute difference of 14 percentage points (95% CI 12 to 16) [1].

Interpretation in practice

The algorithm leads to one of two possible decisions: PE is excluded without imaging, or CT pulmonary angiography is performed. The decision is governed by the combination of the number of criteria met and the D-dimer value, as shown in the table below.

The clinical benefit lies mainly in the group with no criteria met. There, the raised D-dimer threshold of 1000 ng/mL FEU allows a considerably larger proportion of patients to avoid imaging than would a fixed threshold of 500 ng/mL FEU. For patients meeting one or more criteria, the algorithm functions in practice as a conventional Wells-based strategy with a threshold of 500 ng/mL FEU.

An important caveat concerns patients with no YEARS criteria and a D-dimer between 500 and 1000 ng/mL FEU. These patients are ruled out by the algorithm, but they fall in a zone where an age-adjusted D-dimer threshold would have required imaging. External validation (see below) shows that this group is precisely where the weakness lies, and the clinician should be particularly alert to patients in whom the clinical suspicion is higher than the three YEARS criteria capture, especially in the elderly.

Validation and performance

External validation in non-pregnant patients. Eddy et al. applied the YEARS algorithm retrospectively to 3,314 patients from three earlier prospective cohort studies of outpatients with suspected PE [2]. The prevalence of PE was 22.1%, higher than in the derivation cohort. With the YEARS algorithm, PE could be excluded without imaging in 1,423 patients (42.9%). Of these, 17 patients (1.2%, 95% CI 0.8 to 1.9) had PE at baseline investigation. All 17 had no YEARS criteria and a D-dimer below 1000 ng/mL FEU, but above their age-adjusted D-dimer threshold. Among the 272 patients with no YEARS criteria and a D-dimer below 1000 ng/mL FEU but above the age-adjusted threshold, PE was diagnosed in 6.3% (17/272, 95% CI 3.9 to 9.8) [2]. This indicates that the YEARS algorithm is safe overall, but that it may miss PE in older patients with no criteria and a D-dimer between the age-adjusted threshold and 1000 ng/mL FEU.

Combination with PERC. van der Pol et al. evaluated the combination of YEARS and PERC (the Pulmonary Embolism Rule-out Criteria) in 1,316 consecutive patients [3]. YEARS alone gave a proportion of 52% managed without CT pulmonary angiography and a 3-month VTE failure rate of 0.44% (95% CI 0.19 to 1.0). Applying PERC before YEARS in all patients would have led to a failure rate of 1.42% (95% CI 0.87 to 2.3), an increase of 0.98 percentage points compared with YEARS alone. The authors concluded that the combination gave a modest gain in efficiency for patients without YEARS criteria but an unacceptably high failure rate for patients with one or more criteria [3].

Pregnant patients. Langlois et al. analysed data from the CT-PE-pregnancy study and found that the YEARS algorithm could have excluded PE in 77 of 371 pregnant women (21%) without CT pulmonary angiography, with a failure rate of 0/77 (95% CI 0.0 to 3.9) [4]. A US retrospective study of 74 pregnant patients confirmed this: 35 patients with no YEARS criteria and a D-dimer below 1000 ng/mL FEU could have avoided imaging, and none of the patients who did not undergo CT pulmonary angiography was diagnosed with VTE at 30-day follow-up [5]. A pregnancy-adapted version of the algorithm, in which bilateral compression ultrasound of the legs is added for patients who do not meet the exclusion criteria, has been validated in the prospective ARTEMIS study and is recommended by the 2019 ESC guidelines for pregnant women with suspected PE [6].

Comparison with newer prediction models. An individual patient data meta-analysis by van Es et al. comprising 28,305 patients from 16 studies developed a new prediction model for PE including age, sex, previous VTE, recent surgery or immobilisation, haemoptysis, cancer, clinical signs of DVT, inpatient status and D-dimer with an interaction term between age and D-dimer [7]. The model achieved a pooled c-statistic of 0.87 (95% CI 0.85 to 0.89) with good calibration (O:E ratio 0.99). Compared with the Wells score combined with an age-adjusted D-dimer (c-statistic 0.73, 95% CI 0.70 to 0.75) or with structured clinical probability assessment (c-statistic 0.79, 95% CI 0.76 to 0.81), the new model performed better [7]. This model has not yet been implemented in clinical routine, however, and requires a prospective management study before it can replace YEARS or Wells in practice.

Limitations

The YEARS algorithm is validated for adults with clinically suspected acute PE in emergency and outpatient care. It has not been systematically evaluated in inpatients with other serious concurrent illness, in whom the D-dimer is often raised for other reasons and in whom the prevalence of PE differs from the derivation cohort.

The subjective criterion "pulmonary embolism is the most likely diagnosis" is difficult to standardise and varies with the clinician's experience. In the derivation study all patients were assessed by experienced clinicians at university and district hospitals in the Netherlands, which may have contributed to more consistent assessment than is achieved in general clinical practice.

The greatest risk in application concerns patients with no YEARS criteria and a D-dimer between the age-adjusted threshold and 1000 ng/mL FEU. External validation showed that 6.3% of these patients had PE [2]. The clinician should therefore consider age-adjusted D-dimer interpretation as a complement in this group, particularly in older patients, and should not rely blindly on the higher threshold of 1000 if the clinical picture is concerning.

The algorithm does not apply to pregnant patients in its original form. A pregnancy-adapted version, adding leg ultrasound for patients who proceed to imaging, has been validated separately and should be used instead [4, 5, 6].

References

1. van der Hulle T, Cheung WY, Kooij S, et al. Simplified diagnostic management of suspected pulmonary embolism (the YEARS study): a prospective, multicentre, cohort study. Lancet. 2017;390(10091):289-297. PMID: 28549662
2. Eddy M, Robert-Ebadi H, Richardson L, et al. External validation of the YEARS diagnostic algorithm for suspected pulmonary embolism. J Thromb Haemost. 2020;18(12):3289-3295. PMID: 32869501
3. van der Pol LM, van der Hulle T, Mairuhu ATA, et al. Combination of Pulmonary Embolism Rule-out Criteria and YEARS Algorithm in a European Cohort of Patients with Suspected Pulmonary Embolism. Thromb Haemost. 2018;118(3):547-552. PMID: 29390226
4. Langlois E, Cusson-Dufour C, Moumneh T, et al. Could the YEARS algorithm be used to exclude pulmonary embolism during pregnancy? Data from the CT-PE-pregnancy study. J Thromb Haemost. 2019;17(8):1329-1334. PMID: 31108013
5. Mileto A, Rossi G, Krouse B, et al. Pregnancy-adapted YEARS Algorithm: A Retrospective Analysis. West J Emerg Med. 2024;25(1):136-143. PMID: 38205996
6. Konstantinides S, Meyer G. Management of acute pulmonary embolism 2019: what is new in the updated European guidelines? Intern Emerg Med. 2020;15(6):957-966. PMID: 32458205
7. van Es N, Takada T, Kraaijpoel N, et al. Diagnostic management of acute pulmonary embolism: a prediction model based on a patient data meta-analysis. Eur Heart J. 2023;44(32):3073-3081. PMID: 37452732