The diagnostic yield for computed tomography pulmonary angiography in patients with anticoagulation

Abstract

BACKGROUND:

Patients who present to the emergency department (ED) for suspected pulmonary embolism (PE) are often on active oral anticoagulation (AC). However, the diagnostic yield of computed tomography pulmonary angiography (CTPA) in screening for PE in patients who present on AC has not been well characterized. We aim to investigate the diagnostic yield of CTPA in diagnosing PE depending on AC status.

METHODS:

We reviewed and analyzed the electronic medical records of patients who underwent CTPA for PE at a university hospital ED from June 1, 2019, to March 25, 2022. Primary outcome was the incidence of PE on CTPA depending on baseline AC status and indication for AC.

RESULTS:

Of 2,846 patients, 242 were on AC for a history of venous thromboembolism (VTE), 210 were on AC for other indications, and 2,394 were not on AC. The incidence of PE on CTPA was significantly lower in patients on AC for other indications (5.7%) when compared to patients on AC for prior VTE (24.3%) and patients not on AC at presentation (9.8%) (P<0.001). In multivariable analysis among the whole cohort, AC was associated with a positive CTPA (odds ratio [OR] 0.26, 95% confidence interval [CI]: 0.15–0.45, P<0.001).

CONCLUSION:

The incidence of PE among patients undergoing CTPA in the ED is lower in patients previously on AC for indications other than VTE when compared to those not on AC or those on AC for history of VTE. AC status and indication for AC may affect pre-test probability of a positive CTPA, and AC status therefore warrants consideration as part of future diagnostic algorithms among patients with suspected PE.

INTRODUCTION

The diagnosis of pulmonary embolism (PE) has a widespread medical impact, affecting approximately 39–115 per 100,000 population worldwide.[1] The differential diagnosis for dyspnea, chest pain, and arrhythmias among patients presenting to the emergency department (ED) is broad and includes PE among other conditions. Therefore, assessing who to screen for acute PE remains a difficult task.

The tools available for screening patients for PE include Pulmonary Embolism Rule-out Criteria (PERC), Wells score, and the revised Geneva score.[2,3] However, practical utilization of these scores may be difficult and they may not outperform clinical gestalt.[4] Among imaging modalities, computed tomography pulmonary angiography (CTPA) remains a first-line diagnostic approach in patients with suspected PE. While CTPA is a sensitive and specific imaging modality, over testing for PE remains a concern[5,6] particularly since CTPA exposes patients to radiation and contrast, and is associated with a financial burden.[7,8] Therefore, additional patient characteristics which may affect PE diagnostic yield are needed to better risk stratify patients presenting with suspected PE. One potential factor affecting the diagnostic yield for PE is active anticoagulation (AC), as patients receiving AC therapy may have a lower likelihood of developing PE. However, we are unaware of any published data regarding this association. To better address the relationship between active AC and the incidence of a PE diagnosis, we considered AC status when evaluating the diagnostic yield of CTPA in diagnosing PE in patients with suspected PE who presented to the ED.

METHODS

We performed a retrospective cohort study of consecutive adults (aged > 18 years) who underwent a CTPA at Loma Linda University Medical Center ED, a tertiary teaching hospital, between June 1, 2019 and March 25, 2022. Consecutive individuals who met the inclusion criteria (i.e., patients who underwent CTPA for the evaluation of PE) were included. If a patient had multiple CTPA studies during the study period, only the first study was included. We queried the electronic medical records system for key associated demographics, use of AC, laboratory values and vital signs, and key comorbidities as coded on presentation based on defined fields, with comorbidities based on ICD-10 codes. AC was defined as pre-admission use of apixaban, rivaroxaban, dabigatran, or warfarin, and venous thrombosis embolism (VTE) was defined as a history of venous thrombotic disease or PE. To increase the accuracy of the data without relying solely on ICD codes, the medical record of patients receiving AC were reviewed to further clarify indication for AC and to identify potential changes in the management of patients diagnosed with a PE who were on pre-existing AC. If a patient had multiple indications for AC, the patient was included in the VTE group if VTE was one of the indications. The presence of PE was determined based on the radiologist’s reading in the electronic medical records.

The primary outcome was the incidence of PE on CTPA between patients who were on AC and who were not, specifically based on AC indication.

Statistical analysis

For comparisons of baseline demographics, comorbidities, vital signs, and laboratory values, we used the Chi-square test to compare categorical variables and the analysis of variance (ANOVA) to compare continuous variables. SPSS Statistics 22 (IBM Corp., USA) was used for ANOVA and Chi-square tests. A P-value <0.05 was considered statistically significant for all analyses, although exact P values are provided to three decimal places. A multivariate analysis was also performed showing the association between variables and incidence of PE in the entire cohort and in those on AC. D-dimer, which is considered a useful diagnostic tool in the clinical assessment for deep venous thrombosis (DVT)/PE in low-risk patients,[9] was not evaluated as a risk stratifying factor in the multivariable analyses in our study due to its limited value in patients already on oral AC therapy, with studies suggesting omitting D-dimer testing in patients on pre-existing AC[10] because patients on AC have a reduced ability to produce thrombin, and as a result, D-dimer formation.[11]

RESULTS

A total of 2,846 patients were included in the study and final analysis. Among these patients, 242 (8.5%) were receiving AC for prior VTE, 210 (7.4%) were receiving AC for other indications, and 2,394 (84.1%) were not on AC at presentation. Of the patients with non-VTE indication for AC, the most common indications were atrial fibrillation in 157 patients and mechanical valves in 21 patients, while 32 patients had other indications. The characteristics of the included participants are shown in Table 1. Patients who were on AC for other indications were older, more likely to be male, and were more likely to have heart failure (HF), diabetes mellitus, hypertension, chronic obstructive pulmonary disease and chronic kidney disease than patients who were on AC for prior VTE and patients who were not on AC at presentation.

Table 1.

The characteristics of the included participants

The incidence of PE on CTPA was significantly lower in patients who were on AC for other indications (5.7%) than in patients who were on AC for prior VTE (24.3%) and patients who were not on AC at presentation (9.8%) (P<0.001).

Among the entire cohort, factors including history of heart failure (odds ratio [OR]: 0.56, 95% confidence interval [CI]: 0.32–0.98, P=0.043), AC use (OR: 0.26, 95% CI: 0.15–0.45, P<0.001), and systolic blood pressure (OR: 0.84, 95% CI: 0.72–0.96, P=0.014) were independently associated with a reduced incidence of PE, while history of VTE (OR: 98.7, 95% CI: 51.6–188.8, P<0.001) and diastolic blood pressure (OR: 1.45, 95% CI: 1.16–1.82, P=0.001) associated with an increased incidence of PE (Table 2).

Table 2.

Multivariate analysis of association between variables and incidence of pulmonary embolism on CTPA in whole cohort

In the group of patients who were previously on AC, the factors independently associated with increased occurrence of PE were history of VTE (OR: 3.19, 95% CI: 1.21–8.39, P=0.02) and NT-proBNP (OR: 1.01, 95% CI: 1.00–1.012, P=0.003) (Table 3).

Table 3.

Multivariate analysis of association between variables and incidence of pulmonary embolism on CTPA in patients with anticoagulation

Of the 71 patients with baseline AC who had a positive CTPA, two were managed with invasive procedures (one with mechanical thrombectomy and one with pulmonary endarterectomy), while the rest were managed medically.

DISCUSSION

In this study, we found that among patients being evaluated for PE with CTPA in the ED, the incidence of PE varies based on pre-existing AC status as well as indication for AC. Specifically, we found that patients with AC for indications other than VTE were approximately half as likely to be diagnosed with PE compared to patients who did not on AC. These results suggest that the readily available patient characteristics of AC status and indication for AC may be useful in determining pre-test probability for PE, which may guide decision making regarding the use of CTPA.

Prior studies have demonstrated that community practice guidelines (CPGs) (i.e. PERC, Wells score, revised Geneva score) can help reduce the use of CTPAs without increasing the burden of missed diagnoses in the ED.[3] However, utilization of CTPA continues to play a predominant role in the diagnosis of patients presenting with signs and symptoms of PE. While CTPA provides rapid and reliable results, the associated risks of acute kidney injury from exposure to iodinated contrast, long-term malignancy risk due to increased radiation exposure, and financial burden also warrant consideration when considering test utilization.[8,12] Perelas et al [13] in 2015 demonstrated that approximately 49.5% of CTPAs performed in the ED were possibly avoidable, suggesting the need for ongoing efforts to optimize risk stratification for testing. Current literature, supports obtaining a D-dimer in low risk patients, utilizing CPGs, and obtaining diagnostic tests for PE only when warranted by a validated CPG.[14] While such risk stratifying tools have been validated and used for decades, they fail to take into consideration AC status. Although D-dimer is a common screening tool, it may have reduced sensitivity in patients on baseline AC, as AC with both warfarin and novel oral agents can reduce D-dimer concentration, as well as reduce specificity, due to elevation in the setting of concomitant comorbidities such as atrial fibrillation.[15] In fact, prior development and validation studies of thrombosis risk scores have purposefully excluded AC status to avoid concerns of unreliability of the D-dimer test in this setting,[16] which may explain why established risk stratification scores, including PERC, Wells score, and revised Geneva score, all consider a history of DVT/PE to be risk factors but exclude AC status. Recommendations for the care of acutely or critically ill, hospitalized patients recommend AC to prevent the risk of VTE[17] and AC has been associated with lower VTE rates in the current era.[18] This may imply that lower rates of VTE may therefore be expected in outpatient populations on pre-existing AC, and prior studies have postulated that pre-existing AC needs further consideration as part of VTE diagnostic algorithms.[9] However, there is a scarcity of data in the current literature that explore the diagnosis of VTE, such as PE, in non-hospitalized patients on prior AC. Therefore, other considerations in addition to the currently available risk scores may be of value to optimize the utilization of CTPA in the ED. Our group has previously demonstrated that a pre-existing heart failure diagnosis may also lower the yield of CTPA by potentially increasing the likelihood of a viable alternative diagnosis.[19] Likewise, the current study suggested that concurrent use of AC, especially for reasons other than prior VTE, may be considered as a protective factor for potentially decreased CTPA utilization. While the positive rate of a CTPA study of 5% among patients on AC for reasons other than pre-existing VTE suggests that pre-existing AC in itself is not sufficient to rule out PE, the significantly lower rate of positive PE studies in both univariate and multivariable analyses suggest that future studies on the role of AC as a protective factor are warranted. Future studies are needed to replicate these findings among multicenter and multiregional patient populations. Particularly, future study with designs similar to those leading to the development of existing PE risk prediction scores will be needed to better address the role of AC in pre-test probability determination for PE as well as to evaluate the potential role of AC as a protective factor when added to existing risk prediction scores and algorithms.

Limitations

Our retrospective study has limitations. The cohort under investigation only included patients from a single tertiary center ED. Accurate race and ethnicity data are not available for this cohort, although it includes consecutive patients in a diverse region of southern California. Furthermore, the use of available risk scores (i.e. Wells score, revised Geneva score, and/or PERC) and their impacts on ordering CTPA have not been well characterized. Indications for ordering the CTPA were not readily available and were at the discretion of the ED physician, although the patient likely complied with the local standard of care and represented contemporary practice at a large tertiary care ED. Potential bias in management decisions made by the treatment team could not be evaluated in this study, although the consecutive nature of patient selection for inclusion in this retrospective analysis reflects the standard of care at a large university ED. We were not able to determine whether AC status affected the decision to perform a CTPA, and patients who never underwent a CTPA were not included. Some of the positive CTPA studies, particularly in patients with a history of VTE, may have identified chronic PEs, and we were unable to accurately characterize the chronicity of the PEs. Compliance with AC was not evaluated and may have affected decision making regarding PE screening. Data on patient comorbidities were obtained from ICD-10 codes within the electronic medical records and relied on accurate documentation by the treatment team. Likewise, laboratory values and medication use data were also obtained from the electronic medical records and relied on proper documentation and ordering of laboratory studies.

CONCLUSION

The yield of CTPA among patients evaluated in the ED for suspected PE may be affected by pre-existing AC use, with a significantly lower incidence of positive CTPA among patients on AC for reasons other than VTE. Therefore, such patients may benefit from a higher threshold for CTPA evaluation, which may help reduce utilization of CTPA studies. Although additional data are needed to optimize the evaluation of patients who present to the ED with symptoms of suspected PE, especially among patients who already present with AC, we believe that a history of AC may be considered for inclusion in future PE evaluation algorithms.