Previous publications have described the design and conduct of the RIETE registry (ClinicalTrials.gov identifier, NCT02832245).14 Briefly, at each participating RIETE site, investigators aimed to enroll consecutive patients who had acute objectively confirmed VTE. Confirmatory testing for PE consisted of high probability ventilation–perfusion (V/Q) scintigraphy,17 positive contrast-enhanced, PE-protocol, CT for PE,18 or lower limb venous compression ultrasonography positive for proximal deep vein thrombosis (DVT) in patients presenting with PE signs or symptoms.19 All patients provided informed consent for participation in the registry in accordance with local ethics committee requirements.

This study included patients who were enrolled in RIETE and had a diagnosis of acute symptomatic PE from January 1, 2001, through March 31, 2022. Patients with hemodynamic instability at presentation (i.e., systolic blood pressure <90mmHg), and those in whom the RV/LV ratio on CT could not be measured were excluded.

The sPESI score included the variables of age >80 years, history of cancer, history of chronic cardiopulmonary disease, heart rate ≥110beats/min, systolic blood pressure <100mmHg, and arterial oxygen saturation <90% at the time of diagnosis.6

For this study, we also employed a modified sPESI that used a cutoff ≥100beats/min for heart rate. The original and the modified sPESI categorized patients without any variable present as low-risk, and those with any of the variables present as high-risk.

Local radiologists measured the ratio of the right-to-left ventricular short axis diameters. For this study, we defined CT-assessed RV enlargement as a ratio of the RV to the LV short-axis diameters of greater than 1.20

This study used all-cause mortality through 30 days after initiation of treatment as the primary outcome. The secondary outcome was 30-day PE-related mortality. The RIETE investigators used medical record review to assess vital status. For patients who died, further medical record review, and proxy interviews, when necessary, and review of autopsy reports if available, assisted with determination of the date and cause of death. For deaths confirmed by autopsy or those following a clinically severe PE, either after initial PE or shortly after an objectively confirmed recurrent event, in the absence of any alternative diagnosis, the investigators were instructed to judge death as due to fatal PE.

To compare categorical variables, the analyses used chi-squared or Fisher's exact tests. Continuous data were tested for a normal distribution with the Kolmogorov–Smirnov test. Continuous data that did not follow a normal distribution were compared with the Mann–Whitney U test.

We calculated the proportion of patients designated as low-risk based on the sPESI, the modified sPESI, and the combination of the original and modified sPESI with CT-assessed RV/LV ratio. For the combination strategies, both tests had to be negative to classify patients as low-risk. We then determined the proportion of patients with 30-day death (all-cause mortality and PE-related mortality) among low-risk patients. To assess test prognostic characteristics for predicting all-cause and PE-related mortality, we estimated sensitivity, specificity, and positive and negative predictive values and likelihood ratios for each score. Sensitivity meant the percentage of patients who correctly tested positive with a certain strategy over those who died, while specificity meant the percentage of patients who correctly tested negative with a certain strategy over those who survived. Using the sPESI score as the reference, pairwise comparisons of the proportion of all patients who had low-risk PE, and sensitivity were performed using exact binomial testing, and 95% confidence intervals (CIs) for the absolute differences were calculated using the Agresti and Min approach.21

The prognostic relevance of each strategy as well as single predictors with regard to study outcomes was tested using univariable logistic regression analysis and presented as odds ratios with the corresponding 95% CIs. Potential candidate variables were selected a priori based on their biological plausibility at explaining risk for mortality. Specifically, we considered the following covariates: male gender, immobilization, syncope, myocardial injury, and renal insufficiency. Additionally, we constructed multivariable regression models for each strategy separately. For the manual backward stepwise multivariable logistic regression models, we assessed variables that had a significance level of P less than 0.1 in univariable analyses. Since the proportions of missing data (for covariates) were below 5% and it is implausible that certain patient groups specifically were lost to follow-up (i.e., missing completely at random), we analyzed observed data (complete case analysis) ignoring the missing data.

Comparisons were considered significant if the two-sided P-values were less than 0.05. All statistical analyses were performed with the use of the SPSS/PC software package (version 26, SPSS) and the DTComPair package in R version 3.2.3.

We identified 52,979 patients with objectively confirmed symptomatic PE enrolled in RIETE during the study period. After exclusion of patients who had a diagnosis of hemodynamically unstable acute PE (n=1741), and those who did not have the RV/LV ratio measured on CT scan (n=48,121), the study cohort consisted of 3117 normotensive patients with confirmed PE. Patients who had the RV/LV ratio measurements reported in RIETE on CT scan did not differ significantly from those who had not in preexisting medical conditions, and in relevant clinical, physiologic and laboratory parameters (Table S1 in Supplementary Appendix).

Table 1 shows the patients’ clinical symptoms, predisposing conditions, and relevant findings at presentation. Overall, the mean (SD) age was 67.8 (16.1) years, and 49% of the patients were women. The overall number of patients treated on an outpatient basis (i.e., discharged in the first 24h after diagnosis) was small (13.6%; 424 of 3117 patients).

The study had complete survival information for all patients at the end of the 30-day follow-up period. Regarding the primary outcome, 137 of the 3117 (4.4%; 95% CI, 3.7% to 5.2%) patients died during the 30-day follow-up period. Forty-one patients (1.3%; 95% CI, 1.0% to 1.8%) died from PE, and 96 (3.1%; 95% CI, 2.5% to 3.8%) died from other causes (cancer 35, infection 18, hemorrhage 9, heart failure 7, unknown 10, other 17).

Symptomatic recurrent VTE was objectively confirmed in 15 patients in the cohort (15 of 3117 patients; 0.5%; 95% CI, 0.3% to 0.8%). Ten (0.3%; 95% CI, 0.2% to 0.6%) patients had recurrent symptomatic PE (with or without DVT) and 5 (0.2%; 95% CI, 0.1% to 0.4%) had symptomatic isolated DVT. Major bleeding was objectively confirmed in 113 patients in the cohort (113 of 3117 patients; 3.6%; 95% CI, 3.0% to 4.3%).

Using patients in this study cohort, the sPESI classified a significantly higher proportion of patients as low-risk (32.5% [1012/3117]; 95% CI, 30.8% to 34.1%) than the modified sPESI (26.9% [839/3117]; 95% CI, 25.4% to 28.5%) (P<0.001), the combination of the original sPESI with CT-assessed RV/LV ratio (19.4% [606/3117]; 95% CI, 18.1% to 20.9%) (P<0.001), or the combination of the modified sPESI with CT-assessed RV/LV ratio (16.5% [514/3117]; 95% CI, 15.2% to 17.8%) (P<0.001) (Fig. 1).

Fig. 1.

Proportion of patients categorized as having low-risk and failure rates for mortality in the low-risk subgroup according to different prognostic strategies.

Abbreviations: sPESI, simplified Pulmonary Embolism Severity Index; CT, computed tomography; RV, right ventricle; LV, left ventricle; PE, pulmonary embolism.

(0.11MB).

Of 1012 patients with a sPESI of 0 points, 10 (1.0%, 95% CI, 0.5% to 1.8%) died during the 30-day follow-up. Of 839 (26.9%) patients with a modified sPESI of 0 points, 6 (0.7%; 95% CI, 0.3% to 1.6%) died during the 30-day follow-up (Fig. 1). Six hundred and six (19.4%) patients had a sPESI of 0 points and a CT-assessed RV/LV ratio ≤1, of whom 5 (0.8%; 95% CI, 0.3% to 1.9%) died during the 30-day follow-up. Five hundred and fourteen (16.5%) patients had a modified sPESI of 0 points and a negative CT for RV enlargement, of whom 2 (0.4%; 95% CI, 0.1% to 1.4%) died during the 30-day follow-up. Thirty-day PE-related mortality rates were 0.7%, 0.4%, 0.7%, and 0.2% among low-risk patients identified by the sPESI, modified sPESI, the combination of sPESI and CT-assessed RV/LV ratio, or the combination of modified sPESI and CT-assessed RV/LV ratio, respectively (Fig. 1).

Compared with the sPESI (sensitivity 92.7%), the modified sPESI (sensitivity 95.6%; difference 2.9%, 95% CI, −3.5% to 9.5%; P=0.44), and the combination of the sPESI (sensitivity 96.4%; difference 3.7%, 95% CI, −2.6% to 10.1%; P=0.29) or the modified sPESI with a CT-assessed RV/LV ratio ≤1 (sensitivity 98.5%; difference 5.8%, 95% CI, 0.3% to 12.0%; P=0.03) had higher sensitivities and negative predictive values for predicting all-cause mortality (Table 2). Since these models were designed for identification of low-risk PE patients, specificities, and positive predictive values for predicting mortality were low (Table 2).

For predicting 30-day PE-related mortality, compared with the sPESI (sensitivity 82.9%), the modified sPESI (sensitivity 92.7%; difference 9.8%, 95% CI, −6.8% to 26.2%; P=0.31), and the combination of the sPESI (sensitivity 90.2%; difference 7.3%, 95% CI, −9.8% to 24.2%; P=0.52) or the modified sPESI with a CT-assessed RV/LV ratio ≤1 (sensitivity 97.6%; difference 14.6%, 95% CI, 0% to 30.4%; P=0.05) had higher sensitivities and negative predictive values (Table 2).

In univariable analysis, all models were associated with a significantly increased risk for 30-day all-cause mortality (Table 3). The highest odds ratio was observed for the combination of the modified sPESI with CT-assessed RV/LV ratio. Additionally, as shown in Table 3, a number of variables were identified as predictors of mortality using univariable logistic regression analysis. After adjustment, the 4 models remained independent predictors of 30-day all-cause mortality (Table 3).

Since hypoxemic patients with acute PE might not qualify for outpatient therapy of their disease, we recalculated sensitivity for each strategy after exclusion of patients with oxygen saturation <90%. Compared with the sPESI (sensitivity 90.6%), the modified sPESI (sensitivity 94.3%), and the combination of the sPESI with a CT-assessed RV/LV ratio ≤1 (sensitivity 95.3%) the combination of the modified sPESI with a CT-assessed RV/LV ratio ≤1 showed the highest sensitivity (98.1%) for predicting all-cause mortality.