Evidence of improvement: A two-decade decline in pulmonary embolism procedural mortality Authors: Claire Ferguson, Danylo Orlov, Pablo Cuartas, Ron Shemtov, Solomon Seckler, Nishat Choudhury, Dewi Sihaloho, Sherene E. Sharath, Panos Kougias Published: 2026 ABSTRACT ObjectiveGiven the severity of pulmonary embolisms (PEs), investigating mortality trends is important to evaluate health care system progress. The primary objective of this study was to investigate the trend in US PE-related procedural mortality over the past two decades. The secondary objective was to investigate the variation in trends across major treatment modalities.MethodsWe performed a longitudinal retrospective multicenter cohort analysis of PE-related mortality using the National Inpatient Sample, from 1998 to 2021. Our analytic sample included adult individuals diagnosed with PE and treated with catheter-directed thrombolysis, catheter-directed embolectomy (data available since 2015), surgical embolectomy, or medical management. The exposure was the year of hospital admission. The primary outcome variable was inpatient, all-cause mortality.ResultsThe analysis sample included 1,338,332 admissions treated medically and 14,153 admissions treated procedurally. Of the procedure group, 59% received catheter-directed thrombolysis (n = 8363), 24% catheter-directed embolectomy (n = 3453), and 17% surgical embolectomy (n = 2337). Percent mortality for patients undergoing procedural treatment for PE decreased from 33.3% in 1998 to 6.0% in 2021 (Kendall's tau = −0.79; P < .001). Predictors of mortality included advancing age, increasing comorbidity burden, female sex, and race. Percent mortality for patients undergoing catheter-directed thrombolysis decreased from 14.3% in 1998 to 5.7% in 2021 (Kendall's tau = −0.58; P < .001), for patients undergoing catheter-directed embolectomy from 20.8% in 2015 to 6.0% in 2021 (Kendall's tau = −0.90; P < .01), and for patients undergoing surgical embolectomy from 37.1% in 1998 to 10.0% in 2021 (Kendall's tau = −0.67; P < .001). Adjusted mortality odds for the full procedural cohort fell 10% yearly (odds ratio [OR], 0.90; 95% confidence interval [CI], 0.89-0.91). When stratified by treatment, the adjusted mortality odds dropped 7.0% (OR, 0.93; 95% CI, 0.90-0.96) for catheter-directed thrombolysis and 5.0% (OR, 0.95; 95% CI, 0.93-0.97) for surgical embolectomy. The adjusted change in mortality odds for catheter-directed embolectomy was not significant (OR, 0.96; 95% CI, 0.81-1.15). Percent mortality for patients undergoing medical management decreased from 12.0% in 1998 to 9.9% in 2021, with a nadir of 5.9% in 2012 (Kendall's tau = −0.62; P < .001). The adjusted mortality odds for the medical management group increased 1.0% yearly (OR, 1.01; 95% CI, 1.01-1.02).ConclusionsPE-related mortality has decreased significantly among patients treated procedurally and those treated with catheter-directed thrombolysis and surgical embolectomy over the past two decades. These findings suggest that management and patient selection for procedural treatment of moderate to severe PEs may be improving over time. Article Highlights •Type of Research: Retrospective Cohort Study of National Inpatient Sample Data•Key Findings: In this retrospective cohort trend analysis of 14,153 adult patients diagnosed with pulmonary embolism (PE) undergoing procedural treatment, percent mortality decreased from 33.3% in 1998 to 6.0% in 2021 (Kendall's tau = −0.79; P < .001). Adjusted mortality odds for the full procedural cohort fell 10% yearly (odds ratio, 0.90; 95% confidence interval, 0.89-0.91).•Take Home Message: PE-related mortality has decreased significantly among patients treated procedurally; management and patient selection for procedural treatment of moderate to severe PEs may be improving over time. Article Highlights •Type of Research: Retrospective Cohort Study of National Inpatient Sample Data•Key Findings: In this retrospective cohort trend analysis of 14,153 adult patients diagnosed with pulmonary embolism (PE) undergoing procedural treatment, percent mortality decreased from 33.3% in 1998 to 6.0% in 2021 (Kendall's tau = −0.79; P < .001). Adjusted mortality odds for the full procedural cohort fell 10% yearly (odds ratio, 0.90; 95% confidence interval, 0.89-0.91).•Take Home Message: PE-related mortality has decreased significantly among patients treated procedurally; management and patient selection for procedural treatment of moderate to severe PEs may be improving over time. •Type of Research: Retrospective Cohort Study of National Inpatient Sample Data Type of Research: Retrospective Cohort Study of National Inpatient Sample Data •Key Findings: In this retrospective cohort trend analysis of 14,153 adult patients diagnosed with pulmonary embolism (PE) undergoing procedural treatment, percent mortality decreased from 33.3% in 1998 to 6.0% in 2021 (Kendall's tau = −0.79; P < .001). Adjusted mortality odds for the full procedural cohort fell 10% yearly (odds ratio, 0.90; 95% confidence interval, 0.89-0.91). Key Findings: In this retrospective cohort trend analysis of 14,153 adult patients diagnosed with pulmonary embolism (PE) undergoing procedural treatment, percent mortality decreased from 33.3% in 1998 to 6.0% in 2021 (Kendall's tau = −0.79; P < .001). Adjusted mortality odds for the full procedural cohort fell 10% yearly (odds ratio, 0.90; 95% confidence interval, 0.89-0.91). •Take Home Message: PE-related mortality has decreased significantly among patients treated procedurally; management and patient selection for procedural treatment of moderate to severe PEs may be improving over time. Take Home Message: PE-related mortality has decreased significantly among patients treated procedurally; management and patient selection for procedural treatment of moderate to severe PEs may be improving over time. Acute pulmonary embolism (PE) is the third most prevalent cause of cardiovascular death in the United States.1 PE is reported to affect >100 per 100,000 persons annually in the United States2 and is a major cause of morbidity and mortality, accounting for an estimated 300,000 deaths per year.3 Examining trends in mortality is important to evaluate the health care system's progress in managing PE. First-line therapy for PE consists of anticoagulation, with higher-risk cases requiring escalated medical treatment with systemic thrombolysis or procedural intervention with catheter-directed thrombolysis, catheter-directed embolectomy, or surgical embolectomy.4 In regard to the trend in mortality by treatment, in a large meta-analysis, catheter-directed thrombolysis was associated with a significantly lower risk of mortality compared with anticoagulation5 and was reported to have a decreasing mortality risk over time.6 Catheter-directed embolectomy has been reported to have comparable mortality as catheter-directed thrombolysis, but few studies provide data on trends of mortality over time.7 For surgical embolectomy, several studies have reported an improvement in the trend of mortality over time.8,9 As trends in interventional procedures shift toward targeted, minimally invasive procedures, it is important to understand how patient outcomes evolve with these trends.10 However, scarce contemporary studies have comparatively evaluated long-term mortality trends across multiple interventional treatment modalities in patients hospitalized for PE. In this study, the primary objective was to examine trends and predictors of overall mortality in a nationally representative cohort of US inpatients undergoing procedural interventions for PE. The secondary objective was to examine trends in treatment-stratified mortality, including catheter-directed thrombolysis, catheter-directed embolectomy, surgical embolectomy, and medical management. Methods Study design and data source Using a trend analysis, we analyzed inpatient admissions from 1998 to 2021 using the National Inpatient Sample (NIS). The NIS contains a sample of US hospital stay records from hospitals participating in the Healthcare Cost and Utilization Project. The database provides sample weights that can be applied to obtain population-level estimates representative of 97% of the US population. Diagnoses and procedures are coded using the International Classification of Diseases, Clinical Modification and Procedure Coding System (ICD-CM/PCS). ICD-9-CM/PCS codes were used up to 2015, with a switch to ICD-10-CM/PCS in the years following.11 This study was exempt from human subject research review as determined by the SUNY Downstate Health Sciences University's Institutional Review Board. The study was conducted and reported following the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines. Analytic sample Our sample included 1,375,419 adult patients diagnosed with PE between 1998 and 2021. Because patients undergoing procedural interventions likely had an elevated mortality risk, to reduce selection bias, patients were classified as having received medical management alone or procedural intervention. Patients who had systemic thrombolysis were excluded (n = 22,934). The procedural intervention group was further categorized by procedure code to one of three treatments: catheter-directed thrombolysis, catheter-directed embolectomy, or surgical embolectomy. For catheter-directed embolectomy, data were available after 2015. Patients with procedural codes for more than one of the three treatments were categorized using a hierarchical approach: surgical embolectomy was given the highest priority, followed by catheter-directed embolectomy, and then catheter-directed thrombolysis. Because there is no code for the administration of anticoagulation, patients who had a diagnosis of PE who were not treated procedurally were assumed to have been managed medically. Diagnosis and procedure codes are presented in Supplementary Tables I and II. The final analytic sample included 1,338,332 inpatient admissions with a diagnosis of PE managed medically and 14,153 inpatient admissions with a diagnosis of PE managed procedurally. Measures The primary independent variable was the year of hospital admission. The primary outcome variable was inpatient, all-cause mortality. Covariates assessed included the patient's age, Elixhauser Comorbidity Index, sex, race/ethnicity, primary payer, income quartile, hospital region, hospital Medicaid burden quintile, and total procedural volume. The Elixhauser Comorbidity Index is a composite index of patient comorbid conditions developed for use in administrative databases, with increasing index representing increasing individual comorbidity burden.12 Medicaid burden was a calculated variable that served as a proxy for facility resources. The variable was created by grouping facilities into quintiles by the proportion of patients with Medicaid insurance and represents low to high Medicaid facility burden.13 Statistical analyses Descriptive data were summarized using frequencies and percentages, and associations tested with χ2 tests for categorical comparisons. For continuous covariates, means and standard deviations were used with analysis of variance for statistical significance. Trends in mortality Annual percent mortality, calculated as the number of deaths divided by the total number of admissions each year, for all patients undergoing any PE procedure was plotted between 1998 and 2021. Additionally, we plotted the annual percent mortality stratified by treatment type. Trend significance was assessed using the Mann-Kendall test. Predictors of procedural mortality We used sample-weighted multivariable logistic regressions to identify predictors of mortality after PE-related admissions for patients undergoing procedural interventions. To verify that missing covariate data did not significantly affect associations, multiple imputation analyses were performed for each regression. Because results were similar across all models, we proceeded with complete case analysis. Stata SE 18.0 (StataCorp) was used for data management and analysis at an α = 0.05 significance level. Results Patient characteristics We analyzed 14,153 patients who underwent procedural interventions and 1,338,332 patients who were medically managed for a diagnosis of PE between 1998 and 2021. Of the procedural group, 8363 patients (59.1%) were treated with catheter-directed thrombolysis, 3453 patients (24.4%) were treated with catheter-directed embolectomy, and 2337 patients (16.5%) underwent surgical embolectomy. Each of the three procedural subgroups were composed of majority male patients (catheter-directed thrombolysis, n = 4439 [53.1%]; catheter-directed embolectomy, n = 1837 [53.2%]; surgical embolectomy, n = 1203 [51.5%]). Among the procedural subgroups, catheter-directed embolectomy had the oldest average age at 61.0 ± 15.3 years, followed by catheter-directed thrombolysis at 59.7 ± 15.4 years, and surgical embolectomy at 55.5 ± 16.2 years. In the overall procedural sample, the most common racial and ethnicity demographic was White (n = 9396 [71.7%]), followed by Black (n = 2521 [19.2%]), Hispanic (n = 735 [5.6%]), Asian (n = 105 [0.8%]), and Native American (n = 33 [0.3%]). Patients treated with catheter-directed embolectomy had the highest average Elixhauser index (catheter-directed embolectomy, mean 2.0 ± 1.3 compared with catheter-directed thrombolysis, mean, 1.7 ± 1.2; and surgical embolectomy, mean, 1.7 ± 1.3). The medical management group was mostly female (n = 711,752 [53.2%]) with an average age of 63.0 ± 17.2 years. Patients who were managed medically had an average Elixhauser index of 1.4 ± 1.2. Race and ethnicity proportions were similar to the overall procedural group. Table I reports the results of the bivariate analyses in detail.Table IDemographic factors in mortality analysis with pulmonary embolism (PE) treatment modalitiesaVariableMedical management (n = 1,338,332)Total procedural interventionb (n = 14,153)Catheter-directed thrombolysis (n = 8363)Catheter-directed embolectomyc (n = 3453)Surgical embolectomy (n = 2337)P valueSex<.001 Male626,179 (46.8)7479 (52.9)4439 (53.1)1837 (53.2)1203 (51.5) Female711,752 (53.2)6672 (47.2)3922 (46.9)1616 (46.8)1134 (48.5)Race/ethnicity<.001 White853,910 (73.1)9396 (71.7)5703 (73.4)2290 (68.4)1403 (70.1) Black195,636 (16.7)2521 (19.2)1401 (18.0)745 (22.3)375 (18.7) Hispanic73,679 (6.3)735 (5.6)425 (5.5)199 (6.0)111 (5.6) Asian14,080 (1.2)105 (0.8)60 (0.8)25 (0.8)20 (1.0) Native American5116 (0.4)33 (0.3)18 (0.2)5 (0.2)10 (0.5) Otherd26,462 (2.3)324 (2.5)160 (2.1)82 (2.5)82 (4.1)Age, years63.0 ± 17.259.3 ± 15.659.7 ± 15.461.0 ± 15.355.5 ± 16.2<.001Patient income quartile<.001 0-25th percentile368,667 (28.1)4086 (29.4)2384 (29.0)1103 (32.4)599 (26.3) 26th-50th percentile342,546 (26.1)3600 (25.9)2105 (25.6)921 (27.1)574 (25.2) 51st-75th percentile318,899 (24.3)3331 (24.0)2005 (24.4)781 (23.0)545 (24.0) 76th-100th percentile281,411 (21.5)2878 (20.7)1726 (21.0)595 (17.5)557 (24.5)Primary payer<.001 Medicaid145,409 (10.9)1479 (10.5)776 (9.3)418 (12.2)285 (12.2) Medicare706,159 (52.9)6063 (42.9)3645 (43.6)1559 (45.3)859 (36.8) Private387,605 (29.0)5394 (38.2)3200 (38.3)1198 (34.8)996 (42.7) Other96,573 (7.2)1195 (8.5)734 (8.8)266 (7.7)195 (8.4)Elixhauser Comorbidity Index1.4 ± 1.21.8 ± 1.21.7 ± 1.22.0 ± 1.31.7 ± 1.3<.001Hospital Medicaid burden quintile<.001 Low144,636 (10.8)1132 (8.0)721 (8.6)193 (5.6)218 (9.3) Low to moderate258,228 (19.3)2599 (18.4)1561 (18.7)592 (17.1)446 (19.1) Moderate345,918 (25.9)4008 (28.3)2482 (29.7)1021 (29.6)505 (21.6) Moderate to high342,767 (25.6)3730 (26.4)2201 (26.3)958 (27.7)571 (24.4) High246,783 (18.4)2684 (19.0)1398 (16.7)689 (20.0)597 (25.5)Hospital region<.001 Northeast242,232 (18.1)2127 (15.0)1098 (13.1)413 (12.0)616 (26.4) Midwest325,888 (24.4)3791 (26.8)2393 (28.6)875 (25.3)523 (22.4) South517,577 (38.7)5932 (41.9)3542 (42.4)1555 (45.0)835 (35.7) West252,635 (18.9)2303 (16.3)1330 (15.9)610 (17.7)363 (15.5)Values are number (%) or mean ± standard deviation.aColumn totals for individual variables may not add up to the overall totals due to the exclusion of missing observations.bColumn represents aggregate data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy.cData available since 2015.dOther includes races not specified or multiracial classifications. Table IDemographic factors in mortality analysis with pulmonary embolism (PE) treatment modalitiesaVariableMedical management (n = 1,338,332)Total procedural interventionb (n = 14,153)Catheter-directed thrombolysis (n = 8363)Catheter-directed embolectomyc (n = 3453)Surgical embolectomy (n = 2337)P valueSex<.001 Male626,179 (46.8)7479 (52.9)4439 (53.1)1837 (53.2)1203 (51.5) Female711,752 (53.2)6672 (47.2)3922 (46.9)1616 (46.8)1134 (48.5)Race/ethnicity<.001 White853,910 (73.1)9396 (71.7)5703 (73.4)2290 (68.4)1403 (70.1) Black195,636 (16.7)2521 (19.2)1401 (18.0)745 (22.3)375 (18.7) Hispanic73,679 (6.3)735 (5.6)425 (5.5)199 (6.0)111 (5.6) Asian14,080 (1.2)105 (0.8)60 (0.8)25 (0.8)20 (1.0) Native American5116 (0.4)33 (0.3)18 (0.2)5 (0.2)10 (0.5) Otherd26,462 (2.3)324 (2.5)160 (2.1)82 (2.5)82 (4.1)Age, years63.0 ± 17.259.3 ± 15.659.7 ± 15.461.0 ± 15.355.5 ± 16.2<.001Patient income quartile<.001 0-25th percentile368,667 (28.1)4086 (29.4)2384 (29.0)1103 (32.4)599 (26.3) 26th-50th percentile342,546 (26.1)3600 (25.9)2105 (25.6)921 (27.1)574 (25.2) 51st-75th percentile318,899 (24.3)3331 (24.0)2005 (24.4)781 (23.0)545 (24.0) 76th-100th percentile281,411 (21.5)2878 (20.7)1726 (21.0)595 (17.5)557 (24.5)Primary payer<.001 Medicaid145,409 (10.9)1479 (10.5)776 (9.3)418 (12.2)285 (12.2) Medicare706,159 (52.9)6063 (42.9)3645 (43.6)1559 (45.3)859 (36.8) Private387,605 (29.0)5394 (38.2)3200 (38.3)1198 (34.8)996 (42.7) Other96,573 (7.2)1195 (8.5)734 (8.8)266 (7.7)195 (8.4)Elixhauser Comorbidity Index1.4 ± 1.21.8 ± 1.21.7 ± 1.22.0 ± 1.31.7 ± 1.3<.001Hospital Medicaid burden quintile<.001 Low144,636 (10.8)1132 (8.0)721 (8.6)193 (5.6)218 (9.3) Low to moderate258,228 (19.3)2599 (18.4)1561 (18.7)592 (17.1)446 (19.1) Moderate345,918 (25.9)4008 (28.3)2482 (29.7)1021 (29.6)505 (21.6) Moderate to high342,767 (25.6)3730 (26.4)2201 (26.3)958 (27.7)571 (24.4) High246,783 (18.4)2684 (19.0)1398 (16.7)689 (20.0)597 (25.5)Hospital region<.001 Northeast242,232 (18.1)2127 (15.0)1098 (13.1)413 (12.0)616 (26.4) Midwest325,888 (24.4)3791 (26.8)2393 (28.6)875 (25.3)523 (22.4) South517,577 (38.7)5932 (41.9)3542 (42.4)1555 (45.0)835 (35.7) West252,635 (18.9)2303 (16.3)1330 (15.9)610 (17.7)363 (15.5)Values are number (%) or mean ± standard deviation.aColumn totals for individual variables may not add up to the overall totals due to the exclusion of missing observations.bColumn represents aggregate data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy.cData available since 2015.dOther includes races not specified or multiracial classifications. Demographic factors in mortality analysis with pulmonary embolism (PE) treatment modalitiesa Values are number (%) or mean ± standard deviation. Column totals for individual variables may not add up to the overall totals due to the exclusion of missing observations. Column represents aggregate data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy. Data available since 2015. Other includes races not specified or multiracial classifications. Trends in mortality Percent mortality for patients undergoing procedural treatment for PE decreased from 33.3% in 1998 to 6.0% in 2021. The overall 24-year trend showed a significant monotonic decrease (Kendall's tau = −0.79; P < .001) (Fig 1). Percent mortality among those treated with all modalities—catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy—demonstrated a significant monotonic decrease between 1998 and 2021 (catheter-directed thrombolysis, Kendall's tau = −0.58 [P < .001]; catheter-directed embolectomy, Kendall's tau −0.90 [P < .01]; surgical embolectomy, Kendall's tau = −0.67 [P < .001]) (Fig 2). Percent mortality for patients undergoing catheter-directed thrombolysis decreased from 14.3% in 1998 to 5.7% in 2021, for patients undergoing catheter-directed embolectomy from 20.8% to 6.0%, and for patients undergoing surgical embolectomy from 37.1% to 10.0%. There was a significant monotonic decrease for those medically managed (Kendall's tau = −0.62; P < .001) (Fig 3). Percent mortality decreased from 12.0% in 1998 to a nadir of 5.9% in 2012, stayed relatively stable through 2019, then increased rapidly to 9.9% in 2021.Fig 1Percent mortality among patients undergoing intervention for pulmonary embolism (PE), 1998 to 2021. Percents are calculated as the number of patients who died during the index admission who underwent catheter-directed thrombolysis, catheter-directed embolectomy, or surgical embolectomy for PE divided by the total number of patients who underwent procedural treatment for PE. ∗The displayed tau value corresponds with the Mann-Kendall test for significance of monotonic trends; statistical significance (P < .05) is denoted with an asterisk.Fig 2Percent mortality per pulmonary embolism (PE) treatment modality, 1998 to 2021. Percents are calculated as the number of patients who died during the index admission who underwent catheter-directed thrombolysis, catheter-directed embolectomy, or surgical embolectomy for PE divided by the total number of patients who underwent respective procedural treatment for PE. ∗Displayed tau values correspond with the Mann-Kendall tests for significance of monotonic trends; statistical significance (P < .05) is denoted with an asterisk.Fig 3Percent mortality among patients treated medically for pulmonary embolism (PE), 1998 to 2021. Percents are calculated as the number of patients who died during the index admission who were treated medically for PE divided by the total number of patients who were treated medically for PE. ∗The displayed tau value corresponds with the Mann-Kendall test for significance of monotonic trends; statistical significance (P < .05) is denoted with an asterisk. Percent mortality among patients undergoing intervention for pulmonary embolism (PE), 1998 to 2021. Percents are calculated as the number of patients who died during the index admission who underwent catheter-directed thrombolysis, catheter-directed embolectomy, or surgical embolectomy for PE divided by the total number of patients who underwent procedural treatment for PE. ∗The displayed tau value corresponds with the Mann-Kendall test for significance of monotonic trends; statistical significance (P < .05) is denoted with an asterisk. Percent mortality per pulmonary embolism (PE) treatment modality, 1998 to 2021. Percents are calculated as the number of patients who died during the index admission who underwent catheter-directed thrombolysis, catheter-directed embolectomy, or surgical embolectomy for PE divided by the total number of patients who underwent respective procedural treatment for PE. ∗Displayed tau values correspond with the Mann-Kendall tests for significance of monotonic trends; statistical significance (P < .05) is denoted with an asterisk. Percent mortality among patients treated medically for pulmonary embolism (PE), 1998 to 2021. Percents are calculated as the number of patients who died during the index admission who were treated medically for PE divided by the total number of patients who were treated medically for PE. ∗The displayed tau value corresponds with the Mann-Kendall test for significance of monotonic trends; statistical significance (P < .05) is denoted with an asterisk. Adjusted trends in mortality In the overall procedure group, the odds of all-cause mortality declined by 10% per year (odds ratio [OR], 0.90; 95% confidence interval (CI), 0.89-0.91) (Table II). For patients undergoing catheter-directed thrombolysis, the odds of mortality decreased 7.0% annually (OR, 0.93; 95% CI, 0.90-0.96). For patients undergoing catheter-directed embolectomy, the adjusted yearly trend in mortality odds was not significant (OR, 0.96; 95% CI, 0.81-1.15). For patients treated with surgical embolectomy, the odds of mortality decreased 5.0% per year (OR, 0.95; 95% CI, 0.92-0.97). For patients undergoing medical management, the odds of mortality increased 1.0% per year (OR, 1.01; 95% CI, 1.01-1.02).Table IIYearly change in odds of pulmonary embolism (PE)-related mortality per treatment modality and period, 1998 to 2021a,bTreatment modalityOR95% CIP valueAll proceduresc0.900.89-0.91<.001Catheter-directed thrombolysis0.930.90-0.96<.001Catheter-directed embolectomy0.960.81-1.15.68Surgical embolectomy0.950.92-0.97<.001Medical management1.011.01-1.02<.001CI, Confidence interval; OR, odds ratio.aAll analyses are adjusted for patient age, sex, race, income quartile, primary payer, Elixhauser comorbidity index, hospital region, and hospital Medicaid burden quintile.bData for catheter-directed embolectomy available since 2015.cModel includes data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy. Table IIYearly change in odds of pulmonary embolism (PE)-related mortality per treatment modality and period, 1998 to 2021a,bTreatment modalityOR95% CIP valueAll proceduresc0.900.89-0.91<.001Catheter-directed thrombolysis0.930.90-0.96<.001Catheter-directed embolectomy0.960.81-1.15.68Surgical embolectomy0.950.92-0.97<.001Medical management1.011.01-1.02<.001CI, Confidence interval; OR, odds ratio.aAll analyses are adjusted for patient age, sex, race, income quartile, primary payer, Elixhauser comorbidity index, hospital region, and hospital Medicaid burden quintile.bData for catheter-directed embolectomy available since 2015.cModel includes data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy. Yearly change in odds of pulmonary embolism (PE)-related mortality per treatment modality and period, 1998 to 2021a,b CI, Confidence interval; OR, odds ratio. All analyses are adjusted for patient age, sex, race, income quartile, primary payer, Elixhauser comorbidity index, hospital region, and hospital Medicaid burden quintile. Data for catheter-directed embolectomy available since 2015. Model includes data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy. Predictors of mortality Predictors of mortality were analyzed for patients diagnosed with PE undergoing procedural intervention. Odds of mortality increased for patients with a higher Elixhauser index (OR, 1.23; 95% CI, 1.17-1.29) (Table III), advancing age (OR, 1.01; 95% CI, 1.01-1.02), and female sex (OR, 1.20; 95% CI, 1.06-1.36). Compared with White patients, Black (OR, 1.18; 95% CI, 1.01-1.38) and Native American (OR, 4.56; 95% CI, 2.03-10.22) patients had a significantly higher odds of mortality. Patients undergoing catheter-directed embolectomy (OR, 2.09; 95% CI, 1.76-2.49) and surgical embolectomy (OR, 2.34; 95% CI, 1.97-2.77) had a relatively higher odds of mortality compared with patients undergoing catheter-directed thrombolysis.Table IIIPredictors of change in odds of pulmonary embolism (PE)-related mortality for patients undergoing procedural intervention, 1998 to 2021a,bCovariateOR95% CIP valueYear0.900.89-0.91<.001Elixhauser Comorbidity Index1.231.17-1.29<.001Age1.011.01-1.02<.001Sex MaleRefRefRef Female1.201.06-1.36.003Race WhiteRefRefRef Black1.181.01-1.38.04 Hispanic1.060.80-1.40.69 Asian1.640.94-2.84.08 Native American4.562.03-10.22<.001 Otherc1.541.11-2.13.01Procedure Catheter-directed thrombolysisRefRefRef Catheter-directed embolectomy2.091.76-2.49<.001 Surgical embolectomy2.341.97-2.77<.001CI, Confidence interval; OR, odds ratio.aRef indicates reference variable.bModel includes data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy.cOther includes races not specified or multiracial classifications. Table IIIPredictors of change in odds of pulmonary embolism (PE)-related mortality for patients undergoing procedural intervention, 1998 to 2021a,bCovariateOR95% CIP valueYear0.900.89-0.91<.001Elixhauser Comorbidity Index1.231.17-1.29<.001Age1.011.01-1.02<.001Sex MaleRefRefRef Female1.201.06-1.36.003Race WhiteRefRefRef Black1.181.01-1.38.04 Hispanic1.060.80-1.40.69 Asian1.640.94-2.84.08 Native American4.562.03-10.22<.001 Otherc1.541.11-2.13.01Procedure Catheter-directed thrombolysisRefRefRef Catheter-directed embolectomy2.091.76-2.49<.001 Surgical embolectomy2.341.97-2.77<.001CI, Confidence interval; OR, odds ratio.aRef indicates reference variable.bModel includes data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy.cOther includes races not specified or multiracial classifications. Predictors of change in odds of pulmonary embolism (PE)-related mortality for patients undergoing procedural intervention, 1998 to 2021a,b CI, Confidence interval; OR, odds ratio. Ref indicates reference variable. Model includes data from patients treated with catheter-directed thrombolysis, catheter-directed embolectomy, and surgical embolectomy. Other includes races not specified or multiracial classifications. Discussion In this large retrospective analysis, we found an overall decrease in PE-related mortality among inpatients undergoing procedural interventions between 1998 and 2021. Predictors of PE-related mortality included advancing age, female sex, increasing comorbidity burden, and race. For patients receiving either catheter-directed thrombolysis or surgical embolectomy, mortality rates decreased steadily over the 24-year period, with these trends remaining significant even after adjusting for patient- and facility-level covariates, including procedural volume. Although the unadjusted trends in mortality for patients undergoing catheter-directed embolectomy decreased, in the adjusted analysis, this decrease was not significant. For patients undergoing medical management, we identified an annual increase in the adjusted odds of mortality for patients undergoing medical management. Earlier studies from the 1990s through the first decade of the 21st century reported a decrease in PE mortality for patients listed with PE as a cause of death,14 hospitalized high-risk patients with PE,15 and all-cause mortality in patients diagnosed with PE.16 The decrease in PE mortality was attributed to improvements in the management of shock, hospital quality improvement initiatives in administering deep venous thrombosis prophylaxis, the increasing sensitivity and use of computed tomography angiography for diagnosis, and the development of PE response teams.14, 15, 16, 17 More recent epidemiological studies on individuals with PE as a cause of death have raised concern that there has been a plateau in the rate of decrease in PE-related mortality or even an increase over the last decade.17, 18, 19, 20 The potential reversal in progress gained, especially in younger populations, could be due to shifting comorbidity profiles in the US population, including increasing obesity and health care inequities.17, 18, 19 Our results for the trend in mortality for patients managed medically for PE align with studies demonstrating a plateau and reversal. However, for patients managed procedurally, our study showed a consistent decrease in mortality across the whole period. The continued improvement in PE procedural mortality, in contrast with the stagnating/reversing decrease in overall PE-related mortality, supports the hypothesis that patient evaluation for procedural intervention candidacy and procedural management is improving over time. The indications for procedural intervention for PE have evolved over the last 15 years. Whereas initially the use of catheter-directed therapy and surgical embolectomy were primarily for patients with a contraindication to fibrinolysis or who remained unstable after fibrinolysis,21 more recent studies have demonstrated the safety and efficacy of procedural interventions as a first-line of therapy with anticoagulation.22,23 Increasing evidence led to the current American Heart Association guidelines recommending that patients with acute PE at increased risk of adverse outcomes should be evaluated by a multidisciplinary pulmonary response team. Catheter-directed therapy is one of multiple options that may be considered by the team to treat patients with symptomatic PE and elevated clinical severity score (class C), incipient cardiopulmonary failure (class D), or cardiopulmonary failure (class E). Surgical embolectomy may be considered for patients with incipient cardiopulmonary failure (class D) or cardiopulmonary failure (class E).24 Therefore, the expanded indications for procedural intervention as a first-line therapy in contrast with being limited to patients who have a morbid contraindication to thrombolysis or higher mortality risk after failing thrombolysis may have contributed to the improvement in mortality trends. Our findings suggesting that female sex, older age, and non-White race predict increased mortality for patients undergoing procedural intervention are consistent with a prior study of socioeconomic predictors of mortality in patients with PE undergoing treatment with catheter-directed thrombolysis, catheter-directed embolectomy, and systemic thrombolysis; however, in contrast with Breuer et al.’s results,25 in this study primary insurance type was not a significant predictor of mortality. Increasing comorbidity index and Black race have also been reported as significantly associated with mortality odds for patients undergoing surgical embolectomy; however, in Kilic et al’s study,26 age and sex were not significant. The larger sample size and longer analysis period of our study may explain the differences in the identified predictors, but an important conclusion is that our evidence supports prior literature suggesting that there are disparities in mortality by patient sociodemographic characteristics, and further research is needed to understand and mitigate these disparities. Although our study did not find a significant volume-unadjusted change in mortality for patients undergoing catheter-directed embolectomy, the unadjusted results demonstrate a decrease in mortality from 20.8% in 2015 to 6.0% in 2021. Catheter-directed embolectomy represents a relatively new therapy in PE management, but is rapidly increasing in utilization, making it important to understanding outcomes.27 Although the adjusted outcome trend analysis is limited by the relatively decreased data compared with other established PE treatments, the unadjusted results showing comparable mortality between catheter-directed thrombolysis and catheter-directed embolectomy are encouraging and consistent with the literature reporting no significant difference in mortality between the two treatments.28 A prospective, randomized control trial from 2022 to 2024 even demonstrated the superiority of catheter-directed embolectomy compared with catheter-directed thrombolysis with a composite end point of all-cause mortality, intracranial hemorrhage, major bleeding, clinical deterioration and/or escalation to bailout therapy, and postprocedural intensive care unit admission and length of stay.7 It is possible as more data become available for catheter-directed embolectomy that more definitive conclusions about mortality trends in catheter-directed embolectomy may become feasible. Strengths and limitations To our knowledge, this is one of the few studies to comprehensively examine trends in PE procedural mortality and treatment-stratified mortality trends among inpatients with PE cases in the United States. A major strength of this study is its use of the NIS dataset, which offers access to a large, representative sample of US inpatient hospital admissions, granting both sufficient statistical power and the ability to draw broad population-level inferences. In addition, the use of multiple imputation during the sensitivity analysis adds an additional layer of statistical rigor, ensuring that missingness among covariate variables did not alter the results. Despite its strengths, this study is limited by certain factors. First, due to its observational nature, causal inferences cannot be drawn from our study. Although we did identify time-dependent mortality trends in patients diagnosed with PE and across the examined treatment modalities, the NIS dataset does not indicate cause of death. Therefore, our study is limited to associated mortality as opposed to causal mortality. A second major limitation of this study is that we were unable to differentiate between low-risk, intermediate-risk, and high-risk PE, precluding us from stratifying mortality analyses by disease severity. A third limitation is the transition from ICD-9 to ICD-10 between 2015 and 2016, which introduced potential challenges for consistent coding of both PE diagnoses and treatments. Because there is no ICD-9 code for catheter-directed embolectomy, catheter-directed embolectomy analysis was limited to 2015 through 2021. A fourth limitation is that the NIS dataset does not include information about medications, so patients who did not undergo a procedure were assumed to have received medical management with anticoagulation. Conclusions PE-related mortality has decreased significantly among patients treated procedurally, specifically among those treated with catheter-directed thrombolysis and surgical embolectomy over the past two decades. Predictors of overall mortality included advancing age, increasing comorbidity burden, female sex, and non-White race. These findings may indicate that evaluation for intervention and management is improving over time. Future studies should investigate trends in mortality stratified by PE clinical category, trends in procedural complications such as bleeding, and outcomes stratified by specialty conducting the procedure or additional hospital characteristics such as teaching status. Author Contributions Conception and design: CF, DO, SES, PK Analysis and interpretation: CF, DO, PC, RS, SS, NS, DS, SES, PK Data collection: Not applicable Writing the article: CF, DO, PC, RS, SS, NS, DS, SES, PK Critical revision of the article: CF, DO, PC, RS, SS, NS, DS, SES, PK Final approval of the article: CF, DO, PC, RS, SS, NS, DS, SES, PK Statistical analysis: CF, DO, DS, SS, PK Obtained funding: Not applicable Overall responsibility: CF CF and DO contributed equally to this article and share co-first authorship. Funding None. Data availability statement Data are available through the Agency for Healthcare Research and Quality. The authors are not allowed to share data per the data use agreement. Disclosures None.