Long-term mortality in patients with pulmonary embolism: results in a single-center registry

Background While numerous studies have investigated short-term outcomes after pulmonary embolism (PE), long-term mortality remains insufficiently studied. Objectives To investigate long-term outcomes in an unselected cohort of patients with PE. Methods A total of 896 consecutive patients with PE enrolled in a single-center registry between May 2005 and December 2017 were followed up for up to 14 years. The observed mortality rate was compared with the expected rate in the general population. Results The total follow-up time was 3908 patient-years (median, 3.1 years). The 1- and 5-year mortality rates were 19.7% (95% CI, 17.2%-22.4%) and 37.1% (95% CI, 33.6%-40.5%), respectively. The most frequent causes of death were cancer (28.5%), PE (19.4%), infections (13.9%), and cardiovascular events (11.6%). Late mortality (after >30 days) was more frequent than expected in the general population, a finding that was consistent in patients without cancer (the 5-year standardized mortality ratios were 2.77 [95% CI, 2.41-3.16] and 1.80 [95% CI, 1.50-2.14], respectively). Active cancer was the strongest risk factor for death between 30 days and 3 years (hazard ratio [HR], 6.51; 95% CI, 4.67-9.08) but was not associated with later mortality. Death after >3 years was predicted by age (HR, 1.86; 95% CI, 1.51-2.29 per decade), chronic heart failure (HR, 1.66; 95% CI, 1.02-2.70), and anemia (HR, 1.62; 95% CI, 1.09-2.41). Conclusion The risk of mortality in patients with PE remained elevated compared with that in the general population throughout the follow-up period. The main driver of long-term mortality during the first 3 years was cancer. After that, mortality was predicted by age, chronic heart failure, and anemia.

that has been attributed to the increasing proportion of older individuals in the general population [2] and improved detection due to more frequent use and higher sensitivity of computed tomographic pulmonary angiography [3].
A large body of research has investigated risk factors and therapeutic interventions associated with short-term mortality in patients with acute PE, and the results of these investigations have been transformed into detailed clinical practice guidelines [2,4,5]. It is likely due to these efforts that longitudinal studies of patients with acute PE demonstrated improved short-term mortality over time (assessed at hospital discharge or after 30 days) [6,7].
The high risk of mortality in patients with PE is not limited to the early phase. Long-term mortality rates of >30% after 5 years have been reported, corresponding to a 2.5-fold increase in the risk of mortality compared with that in age-and sex-matched individuals in the general population [8]. However, current knowledge of long-term prognosis of patients with acute PE is still limited. Several studies that investigated long-term outcomes focused on the effects of long-term anticoagulation or thrombolysis and, hence, included a selected study population rather than a cohort of "real-world patients" [9][10][11][12][13][14]. Other study cohorts were limited to survivors of the acute phase [10,15,16], patients with cancer [17], or elderly individuals aged ≥65 years [18,19]. Most reports that compared mortality rates with those in the general population were based on retrospective analyses of data from health care registries [20,21] or did not account for the high prevalence of cancer in patients with venous thromboembolism (VTE) [8,12].
Hence, the present study aimed to investigate long-term outcomes in an unselected "real-world" cohort of consecutive patients with acute PE, evaluate changes in risk factors and causes of death over the duration of the follow-up period, and provide a comparison with the general population.

| Study design and definition of outcomes
The Pulmonary Embolism Registry of Göttingen (PERGO) prospectively includes consecutive patients with objectively confirmed acute PE aged ≥18 years admitted to the University Medical Center Göttingen, Germany. The study protocol has been described in detail previously [22,23]. Briefly, patient recruitment is performed by daily screening of new admissions to the emergency department and reports of computed tomography pulmonary angiographies performed.
After obtaining informed consent for participation in PERGO, complete data on comorbidities, previous medication, symptoms, results of diagnostic tests, treatment, and clinical course are recorded using a standardized case report form. The modes of data collection did not change over the years. The present analysis included patients enrolled in PERGO between May 2005 and December 2017. We excluded patients with another acute cardiorespiratory illness, such as acute myocardial infarction, left-heart decompensation, or respiratory decompensation responsible for clinical presentation and symptoms as well as recurrent PE (only the first event was included in the analysis).
The study outcome was all-cause death during long-term followup. Early mortality was defined as death within 30 days after the diagnosis of PE, whereas late mortality was defined as death >30 days after PE.
Data on patient characteristics, comorbidities, and survival status at the time of hospital discharge were obtained from individually reviewed hospital patient records. The status of these variables (including age and cancer) was determined at the time of hospital discharge; thus, a time-varying analysis was not performed. In patients discharged from hospital alive, long-term survival status was assessed

Essentials
• The effects of pulmonary embolism on long-term mortality are insufficiently studied.
• We investigated long-term outcomes in 896 patients with pulmonary embolism followed up for up to 14 years.
• Mortality was higher than expected in the general population throughout the follow-up period.
• This finding was consistent in patients with and without cancer. by periodically contacting the responsible community registration offices (in Germany, every town and county has 1 or several registration offices, where every resident has to be registered). Time at risk was calculated from the time of PE diagnosis until death occurred or the last date for which survival status was known. The observation period ended in November 2019.
Participants who were no longer registered (eg, due to emigration) were considered lost to follow-up. In such cases, data were censored on the last date for which survival data were available.
For comparisons of the observed mortality rate in the cohort with the expected mortality rate in the general population, we obtained mortality data for the German reference population from the Human Mortality Database [24].
If death occurred, supplemental information on the cause of death Renal insufficiency was defined as a glomerular filtration rate of <60 mL/min/1.73 m 2 of body surface area. Active cancer was defined as a known malignant disease, treatment with antitumor therapy within the last 6 months, metastatic state, or hematologic cancer not in complete remission [27]. Hemodynamic instability and the risk of PE recurrence were assessed according to the criteria proposed by the 2019 European Society of Cardiology guidelines [2]. The prognostic relevance of baseline characteristics with regard to the study outcome was assessed using univariable Cox proportional hazards models and results expressed as hazard ratios with corresponding 95% CIs. The predictors of mortality identified in the univariable analyses were entered in multivariable Cox proportional hazards models, including all univariable predictors.

| Statistical analysis
Cumulative mortality was obtained using SAS PROC LIFETEST using the Nelson Aalen estimator. CIs are based on log-log transformation. We also computed the expected survival of patients with PE using the Ederer II method [28]. Standardized mortality ratios (SMRs) were calculated to compare the observed all-cause mortality rate in the study cohort with the expected all-cause mortality rate in the general population taking into account sex, age at the time of PE diagnosis, and year of birth. SMRs were calculated for all survivors of the acute phase and the subgroup of survivors of the acute phase without cancer.

| Causes of mortality
As presented in Table 2    patient-years vs 67.0/1000 patient-years, respectively; P < .001). As expected, patients with PE with advanced cancer were more likely to die during follow-up than patients without known metastases (P < .001; Figure 3).

| Predictors of long-term mortality after PE
Univariable and multivariable predictors of overall long-term mortality and late mortality are presented in Table 3. In multivariable analyses, late mortality was predicted by active cancer, chronic pulmonary disease, anemia, chronic heart failure, and age at the time of PE diagnosis, while obesity was associated with a reduced risk of long-term mortality. Similar results were observed in the subgroup of patients without cancer (Table 3, right column). A more detailed analysis of mortality predictors during different time periods after acute PE is provided in Table 4. Active cancer at the time of PE diagnosis presented the strongest risk factor for death during the first 3 years after PE but was not associated with increased risk of mortality during the later follow-up period.

| Long-term mortality compared with that in the general population
In survivors of the acute phase (initial 30 days after PE diagnosis), the rates of long-term mortality were higher than the expected mortality rate in the general population taking into account sex, age, and year of birth ( Figure 4A).

| Risk of long-term mortality in patients with PE
The reported long-term mortality rates in patients with PE vary widely between studies, likely due to multifactorial causes such as differences in age and cancer status between study cohorts, observation period, and performance characteristics of national health care systems. for the low mortality observed in this latter study was likely the study design, which excluded patients with in-hospital mortality during their PE-related stay. Furthermore, the authors reported a relatively young cohort (median age, 50 years vs 69 years in our cohort) with low rates of cancer and cardiovascular comorbidities [16]. In contrast, the 5- Patients who survived for 5 years after PE diagnosis were still 40% more likely to die during further follow-up than expected based on age, sex, and year of birth. This finding is in contrast to an earlier investigation by Naess et al. [30], who reported that the mortality rate in patients with non-cancer-associated VTE normalized after 3 years.
Not surprisingly and in accordance with previous studies [8,21], late mortality (occurring >30 days after PE) was associated with cancer, chronic cardiopulmonary disease, anemia, and age. There was an inverse correlation between obesity and late mortality, a finding in accordance with the "obesity paradox" described in different cardiovascular diseases, including PE [31,32]. Of note, this finding does not infer a causal relationship but rather reflects differences in baseline risk in those exposed to the respective risk factors. As reported previously [30], we could confirm that PE with a low risk of recurrence (eg, after major surgery or trauma [2]) was associated with significantly higher chances of long-term survival compared with other causes of PE.
In patients with cancer, death was not only twice as frequent as that in patients without malignancy but also occurred sooner after PE.

| Causes of long-term mortality
Not surprisingly, cancer was the main driver of overall long-term mortality in our cohort (28.5%), confirming the findings of earlier reports [16,19]. In contrast to our findings, Ng et al. [8] reported that cardiovascular diseases (including PE) are the most frequent cause of late (postdischarge) mortality in patients with PE, accounting for 36% of deaths. Despite a similar distribution of baseline rates of cardiovascular comorbidities and cancer in their study and our cohort, we observed a substantially lower rate of cardiovascular or PE-related mortality in our cohort (19.5% after >30 days). As discussed earlier, this lower rate in our more recent study was likely due to multifactorial causes but may also have been influenced by the declining cardiovascular death rate of the overall population [33]. Although detailed information on the duration of anticoagulation in individual patients was not available, this may suggest underuse or an insufficient duration of anticoagulation treatment in our cohort.

| Strengths and limitations
Our report adds substantially to previous studies. First, we reported a large cohort of well-characterized patients and provided longer follow-up data (up to 14 years after PE) compared with previous reports [8,19] Actelion, Germany.

ETHICS STATEMENT
The study was conducted in accordance with the amended Declaration of Helsinki and approved by the local independent Ethic Committee of the University Medical Center Göttingen, Germany (protocol number 14/6/10). All patients gave informed written consent for participation in the study.