Influence of an Alternative Diagnosis on the Diagnosis of Pulmonary Thromboembolism

1. Introduction

Pulmonary Embolism (PE) is a prevalent condition, with an annual incidence ranging from 39 to 115 cases per 100,000 inhabitants [1,2,3,4]. It is the third most common cardiovascular disease, following acute myocardial infarction and cerebrovascular accident [5,6,7], with morbidity and mortality rates reaching up to 30% in some reports [2].

The clinical presentation of PE is highly variable and nonspecific [8,9,10,11]. Most commonly, it presents with dyspnea, cough, or chest pain [10,12], symptoms that overlap with other pulmonary and cardiovascular conditions such as pneumonia, exacerbation of COPD, or decompensated congestive heart failure, among others [9,11,13]. Additionally, the average age of PE presentation is between 60 and 70 years, a period when these other pathologies are also prevalent [1,13,14,15,16,17,18,19]. Consequently, establishing a presumptive diagnosis of PE based solely on clinical and epidemiological data can be challenging [8,20,21]. Several risk factors for venous thromboembolic disease (VTE) can modify the probability that a patient's symptoms are due to PE. Notably, cancer, recent surgery, immobility, and estrogen use are associated with an increased risk of PE [22,23,24,25].

Clinical prediction scales have been developed to stratify the risk of PE in patients. Among these, the Wells score is widely used and has demonstrated strong external validation, serving as a primary tool for initial PE risk assessment [7,26]. This scale evaluates signs, symptoms, and risk factors, categorizing patients into low, medium, or high risk [7,26]. Based on this assessment, patients are directed through clinical algorithms to undergo further tests for diagnosis or exclusion of PE [21,27,28,29,30]. However, the Wells score has limitations, including its exclusion of hospitalized patients, lack of a clearly defined target population, inclusion of subjective elements, and dependence on a hospital setting capable of performing necessary tests and procedures [31].

More recently, alternative approaches to diagnosing VTE have emerged. The Pulmonary Embolism Rule-Out Criteria (PERC) is one such model, including eight factors with a high negative predictive value for PE [32]. This model can rule out PE in patients with very low probability, achieving near 100% sensitivity [33,34,35]. Despite its usefulness, PERC's representativeness is limited, affecting approximately 13% of the at-risk population and thereby being inadequate for many patients with suspected PE [36].

In summary, the epidemiological, clinical, and analytical heterogeneity of PE patients, combined with the complexity of diagnostic tests, complicates standardized clinical judgment. While it seems reasonable to rule out PE in patients with a clear alternative diagnosis, such as pneumonia, COPD exacerbation, or congestive heart failure, or those on anticoagulant therapy for other reasons [37], definitive evidence supporting this approach is lacking. Therefore, this study aims to evaluate the discrimination capability of clinical factors with a high negative predictive value for PE and compare them to the PERC scale in terms of screening efficacy and reduction in unnecessary testing.

2. Materials and Methods

This observational, retrospective cohort study included all patients who underwent chest Computed Tomography Angiography (CT Angiography) between 2008 and 2017 at HM Sanchinarro Hospital, a 200-bed university hospital. The study was approved by the center's ethics committee, and informed consent was waived (HM Hospitals CEIm Code: 21.03.1792-GHM).

Patients were identified through a review of digitized medical records. Data collection included demographic information, risk factors, comorbidities, clinical variables, vital signs at the time of the test, and analytical and radiographic data. Variables corresponding to the Wells and PERC scales (Table 1 and Table 2) [26,32] were recorded. The presence of PE on CT Angiography was considered the outcome variable. Exclusion criteria were being under 18 years of age, undergoing CT Angiography for reasons other than suspected PE, having an acute PE diagnosis at the time of the test, and insufficient data for calculating the Wells or PERC scales. This resulted in a final study cohort of 399 patients.

Objective variables with a high negative predictive value were defined based on diagnostic guidelines [38,39,40]: (i) Pneumonia: fever and radiographic consolidation as assessed by the physician; (ii) COPD Exacerbation: smoking habit, cough, and fever; (iii) Congestive Heart Failure (CHF): radiological pattern consistent with heart failure and dyspnea or orthopnea; (iv) Previous Anticoagulation in Non-Oncological Patients: anticoagulated patients without a prior oncological diagnosis at the time of CT Angiography (Table 3).

Statistical analysis employed standard methods. Qualitative variables were analyzed using the chi-square test, and quantitative variables were assessed with t-tests and ANOVA. Sensitivity, specificity, and positive and negative predictive values were calculated for each study variable and the PERC scale. A new scale, PERC plus, incorporating all variables with a 100% negative predictive value, was created. The ability of each criterion and the PERC plus set to reduce diagnostic tests was assessed by calculating the incidence of at-risk patients where CT Angiography could have been avoided. Differences with p < 0.05 were considered statistically significant. Statistical analysis was performed using SPSS version 27.0 (Armonk, NY, 2020).

3. Results

The study evaluated 399 patients who met the inclusion criteria, with 139 (35%) diagnosed with PE by CT Angiography. The cohort had an average age of 65 ± 16 years, and 53% were female (n=214). The average number of comorbidities per patient was 0.8 ± 0.3, with oncological pathology being the most common (n=121, 30%). Risk stratification according to the Wells scale showed that 6 patients with PE (4%) were in the low-risk group, 95 patients (68%) were in the moderate-risk group, and 38 patients (27%) were in the high-risk group. CT Angiography was performed after Emergency Department assessment in 81.20% of patients.

The most frequent symptoms were dyspnea (n=269, 67%), pleuritic pain (n=154, 39%), cough (n=115, 29%), fever (n=66, 17%), lower limb pain (n=43, 11%), orthopnea (n=25, 6%), syncope (n=15, 4%), and hemoptysis (n=11, 3%). At diagnosis, 108 patients (27%) had hemodynamic instability, and 185 patients (46%) had acute respiratory failure. PERC criteria were met by 13 patients (3%), pneumonia criteria by 27 patients (7%), COPD exacerbation criteria by 40 patients (10%), CHF criteria by 18 patients (5%), and 29 patients (7%) were anticoagulated without oncological pathology. The distribution of these patients according to Wells scale risk is shown in Table 4.

Patients with PE (n=139) had an average age of 66 ± 17 years and 47% were female (n=65), while non-PE patients (n=260) had an average age of 64 ± 16 years and 57% were female (n=149), showing no statistically significant differences. The average number of comorbidities in PE patients was 0.9 ± 0.3, with a predominance of oncological pathology (n=45, 32%), similar to the non-PE group. PE patients were more frequently detected in the Emergency Department (n=124, 89.2%), and were associated with lower limb pain (22.30% vs. 4.60%, p <0.001), heart rate > 100 bpm (33.10% vs. 22.70%, p = 0.025), and oxygen saturation < 94% (56.80% vs. 40.80%, p = 0.002). Cough was associated with negative CT Angiography (15.10% vs. 36.20%, p<0.001). Hemodynamic instability and desaturation were more frequent in PE patients. Both the PERC scale and the presence of COPD exacerbation and anticoagulation in non-oncological patients showed discrimination for PE (p<0.05), while pneumonia criteria approached statistical significance (p=0.06). CHF criteria did not show a significant association with PE (p=0.71) (Table 5).

The sensitivity of the Wells scale for diagnosing PE in the low-risk group was 95%, decreasing to 74.68% in hospitalized patients. Evaluating the sensitivity and specificity of conditions with high negative predictive value revealed 100% sensitivity and NPV for patients meeting criteria for alternative diagnoses in low-risk Wells groups. The PERC plus scale, combining objective clinical criteria with PERC, also achieved 100% NPV and sensitivity in low-risk patients according to the Wells scale but showed an NPV of 46.67% and 0% in medium and high-risk groups, respectively (Table 6).

The reduction in CT Angiography use for conditions with high NPV was 9% for pneumonia, 12% for COPD, 5% for CHF, 9% for anticoagulated patients without oncological pathology, and 5% overall (Figure 1).

4. Discussion

The findings highlight the importance of a structured approach to diagnosing Pulmonary Embolism (PE) in acute care settings. The 35% prevalence of PE in our cohort underscores the need for effective and efficient diagnostic strategies. The Wells score remains a valuable tool in risk stratification but has limitations, especially when applied to hospitalized patients and subjective factors.

The PERC scale, effective in ruling out PE in very low-risk patients, has limited applicability due to its low representativeness in the broader at-risk population. Our study demonstrates that integrating PERC with additional clinical criteria, such as alternative diagnoses, creates a more robust tool—PERC plus—with a 100% negative predictive value (NPV) and sensitivity in low-risk groups. This combined approach can potentially reduce unnecessary CT Angiography without compromising diagnostic accuracy.

Implementing the PERC plus scale could significantly decrease unnecessary diagnostic tests, optimizing resource allocation while maintaining diagnostic accuracy. This is particularly relevant for enhancing efficiency in emergency settings.

5. Conclusions

The PERC plus scale shows potential in excluding PE in patients classified as low risk by the Wells scale, potentially reducing the need for further tests in both outpatient and inpatient settings. Combining objective clinical criteria with the PERC scale could improve diagnostic accuracy and reduce unnecessary testing, thus enhancing patient management and resource utilization in acute care environments.