Highlights

  • sPESI score identifies a PE cohort with lower clinical and economic burden.

  • LRPE patients were less likely to have HACs and other PE-related outcomes than HRPE patients.

  • LRPE patients also had shorter inpatient length of stay and lower health care costs.

  • Prognostic risk stratification is fundamental in choosing appropriate PE treatment.

Introduction

Pulmonary embolism (PE) is a major health concern causing hospitalization in the United States, with 176,000 admissions annually [1, 2]. In the span of 20 years, the incidence of PE increased by 81%, from 62.1 to 112.3 cases per 100,000 people [3]. PE is a potentially fatal disease with a 90-day mortality rate of up to 37.2% and ~ 100,000 annuals deaths in the United States [4, 5].

The economic burden of PE is substantial, with estimated annual cumulative costs ranging from $8.5 to $19.8 billion in the United States [6]. Additionally, the annual cost per patient has been estimated at $13,300 to $31,000 in PE patients [7]. However, risk stratification remains a challenge in clinical practice. The American college of chest physicians recommends that low-risk PE (LRPE) patients can benefit from abbreviated hospital stays or outpatient therapy which could substantially reduce the clinical and economic disease burden [8, 9]. Several risk-stratification methods have been developed to identify LRPE patients, including the Geneva score, the Pulmonary Embolism Severity Index (PESI) score, the simplified PESI (sPESI) score, etc [10].

The European Society of Cardiology (ESC) advocates for the risk stratification of PE patients and the consideration of an outpatient management option for LRPE patients. However, physicians in the United States have not widely adopted an outpatient or observation management strategy [11, 12]. Also, the new recommendations of the ESC suggest using the sPESI score as the first step for prognostic assessment of acute hemodynamic stability of PE [13]. Therefore, we compared the PE-related outcomes, health care resource utilization (HRU), and costs in LRPE vs high-risk PE (HRPE) identified using the sPESI score.

Materials and methods

Data source

This was a longitudinal, retrospective cohort study that used data from the Veterans Health Administration (VHA) during October 1, 2010 to September 30, 2015 (study period). The VHA is the largest integrated health care system in the United States, providing care at 1245 health care facilities—including 170 US Department of Veterans Affairs medical centers and 1065 outpatient clinics—and serving > 9 million enrolled veterans across the country [14].

Institutional Review Board approval was not required because this study did not use individually identifiable data from the VHA.

Study population

Patients were included in the study if they had ≥ 1 inpatient diagnosis for PE (ICD-9-CM codes 415.1, 415.11, and 415.19) during the identification period (October 1, 2011–June 30, 2015), with the hospital discharge date designated as the index date. Patients were required to be aged ≥ 18 years, have a prescription claim for an anticoagulant [unfractionated heparin (UFH), low-molecular-weight heparin (LMWH), warfarin, or novel oral anticoagulants (NOACs)] during the index hospitalization, and be continuously enrolled in their health plan with medical and pharmacy benefits for ≥ 12 months prior to the index date (baseline period) until 3 months post-index date or death (follow-up period), whichever occurred first (Appendix in Supplementary material). Patients administered subcutaneous heparin during the hospital stay were not included, since many patients are given subcutaneous heparin as a prophylaxis for PE. Also, patients with a PE claim or any anticoagulant claim prior to the index hospitalization were excluded.

Eligible PE patients were stratified based on the presence of sPESI parameters during the index hospitalization into LRPE and HRPE cohorts. The sPESI is an extensively validated prognostic tool, encompassing 6 items—including age ≥ 80 years, history of cancer/chronic cardiopulmonary diseases, heart rate ≥ 110/min, systolic blood pressure < 100 mmHg, and arterial oxygen saturation < 90%—each burdening 1 point when present [15, 16]. Patients scoring 0 points are considered at low risk and all others are considered at high risk.

Baseline measures

Patient demographics including age, gender, race, and body mass index during the baseline period were assessed. In addition, clinical characteristics including Charlson Comorbidity Index (CCI) score, past medical history, and the administration of various diagnostic tests were recorded. We assessed the association of hospital-acquired complications (HACs) and clinical markers with the risk level of PE patients during the index hospitalization (Appendix in Supplementary material).

Outcome measures

PE-related clinical outcomes [recurrent venous thromboembolism (VTE), major bleeding, and death] and diagnostic tests, including computed tomography angiography (CTA), echocardiogram, lung ventilation/perfusion scan, and venous Doppler ultrasound, during the 90-day post-discharge period, were evaluated. Recurrent VTE was defined as having a hospitalization claim for deep vein thrombosis (DVT) or PE between 8 and 90 days after the index date. Major bleeding was defined using a previously validated algorithm developed by Cunningham et al. [17] (Appendix in Supplementary material). Additionally, PE-related clinical outcomes were reported during the 30-day post-discharge period. Health care resource utilization (HRU) including the percentage of patients with any (i.e., not disease-specific) inpatient hospitalization, and outpatient visit, as well as the mean number of visits per patient and associated health care costs during the 90-day follow-up period were reported.

Statistical analysis

Descriptive statistics were provided for all study variables, including baseline demographics, clinical characteristics, and outcome variables among the LRPE and HRPE cohorts. Statistical tests of significance (Chi square for categorical variables and Wilcoxon-rank sum test for continuous variables) were conducted to assess differences between the cohorts. All analyses were conducted using Statistical Analysis Software (SAS)® (Version 9.3, SAS Institute, Cary, North Carolina, 2012).

Results

The study included 6746 PE patients, of which 1918 (28.4%) were stratified as LRPE and 4828 (71.6%) as HRPE (Fig. 1).

Fig. 1
figure 1

Study population and cohorts. HRPE high-risk pulmonary embolism, ICD-9-CM International Classification of Diseases, 9th Revision, Clinical Modification, LRPE low-risk pulmonary embolism, PE pulmonary embolism

Full size image

Baseline characteristics

The demographic and clinical characteristics during the baseline period were summarized in Table 1.

Table 1 Demographic and clinical characteristics of LRPE versus HRPE patients
Full size table

During the index hospitalization, LRPE patients had a shorter average inpatient length of stay (LOS) (8.8 vs. 11.2, p < 0.0001) and fewer HACs (11.4% vs. 20.0%, p < 0.0001) and were also less likely to have clinical marker testing (Table 1).

90-day follow-up outcomes

During the 90-day follow-up period, LRPE patients had fewer PE-related outcomes, including recurrent VTE (4.4% vs. 6.0%, p = 0.0077), major bleeding (1.2% vs. 1.9%, p = 0.0382), or death (3.7% vs. 16.2%, p < 0.0001). Additionally, the clinical outcomes during the 30-day follow-up period were reported (Table 2).

Table 2 PE-related outcomes among LRPE versus HRPE patients
Full size table

90-day follow-up HRU and costs

As compared to HRPE patients, LRPE patients had a fewer number of hospital re-admissions (0.2 vs. 0.3, p < 0.0001) with a shorter average re-hospitalization LOS (1.9 vs. 2.5 days, p < 0.0001). However, LRPE patients had higher outpatient visits per patient (19.7 vs. 18.4, p = 0.0002) (Fig. 2). Also, LRPE patients were more likely to have a venous Doppler ultrasound (8.5% vs. 4.9%, p < 0.0001) during the 90-day follow-up than HRPE patients (data not shown). LRPE patients incurred lower inpatient ($4503 vs. $7332, p < 0.0001), pharmacy ($1294 vs. $2502, p = 0.0013), and total costs ($12,021 vs. $16,911, p < 0.0001) than HRPE patients (Fig. 2).

Fig. 2
figure 2

HRU and health care costs among LRPE versus HRPE patients during the 90-day follow-up period. HRPE high-risk pulmonary embolism, HRU health care resource utilization, LOS length of stay, LRPE low-risk pulmonary embolism

Full size image

Discussion

The results of our study showed that in a real-world clinical setting, sPESI identifies a subset of PE patients at a lower risk for adverse clinical outcomes, including mortality. These patients might benefit from early discharge or outpatient therapy, thus reducing their economic burden, especially considering the extended length of stay of LRPE in this VA cohort (mean 8.8 days).

Although a few studies have evaluated the value of sPESI in identifying those PE patients at an early risk of adverse clinical events, our study is unique in evaluating the 90-day clinical events [16, 18]. The results of previous studies showed that about 26.1–30.7% of PE patients were stratified as low-risk using the sPESI score [18,19,20]. Although right ventricular dysfunction (RVD) was considered as a significant predictor of early prognosis in LRPE patients, we observed that very few patients (~ 5) had evidence of RVD during the index hospital stay [21]. The results of our study were consistent with these studies, as 28.4% of the PE patients were classified as low-risk. Additionally, the results of our study showed that LRPE patients had ~ 2 days shorter LOS than the HRPE patients. However, in a study conducted by Shafiq et al. using the PESI criteria, no significant differences were observed in the hospital LOS between the LRPE and HRPE patients [22]. The study also observed that only 9% of the LRPE patients were discharged early (≤ 3 days), albeit the American College of Chest Physicians guidelines recommending early discharge of LRPE patients [22]. Our study also showed that LRPE patients were less likely to have HACs, which could possibly be related to the shorter hospital LOS among these patients. However, as pointed out in earlier studies, we could not conclude if HACs cause a longer LOS or if a longer LOS is caused by the HACs [23, 24]. Therefore, further research is warranted to evaluate the hospital LOS among PE patients stratified using the sPESI criteria.

The results of our study showed that the mortality rate in LRPE patients was 4 times lower than the HRPE patients, while several previous studies observed a 10 times lower mortality rate in the sPESI LRPE patients [16, 18, 25, 26]. A possible explanation for the larger difference in the magnitude of the mortality risk is that these previous studies evaluated the early mortality risk (30-day mortality) while our study assessed the 90-day mortality in PE patients. Similarly, the rate of major bleeding was lower in the LRPE patients, which is similar to the findings from Masotti et al. [16], who observed an increase in the bleeding rate with an increase in the sPESI score. Additionally, the combined rate of non-fatal major bleeding and recurrent VTE was about 5% in our study, which is consistent with a study by Ozsu et al. [18].

There is no evidence in the literature to date that evaluates the HRU and the associated costs among PE patients stratified by using any of the risk-stratification tools. The results of our study showed that LRPE patients had fewer hospital re-admissions well as lower health care costs. Surprisingly, LRPE patients had higher outpatient utilization, which could be related to the higher proportion of LRPE patients with venous Doppler ultrasound imaging during the follow-up period. This may be due to the possibility that clinicians performed more tests to confirm that LRPE patients were at low-risk for the complications and could be safely managed as outpatients or due to a higher death rate in HRPE patients, thus less opportunity for visits and extra testing. However, previous studies validating the sPESI score suggested that PE patients with a sPESI score of 0 do not require further prognostic imaging procedures or laboratory biomarker testing to define the low risk [8, 16]. Therefore, the burden of diagnostic tests can be reduced in the LRPE patients. Several previous studies suggested that reducing the hospital LOS, which is an important cost driver among PE patients, would substantially reduce the costs [18, 27, 28]. Also, in a previous study using the In-hospital mortality for pulmonary using claims data (IMPACT) criteria for the risk stratification of PE patients, the LRPE patients discharged early (≤ 3 days) had lower health care costs than those who stayed longer in the hospital [29]. Previous studies have suggested that PE outpatient management has cost savings in the range of $500 to $2500 per patient [7, 9]. Our study showed that LRPE patients incurred $4890 lower health care costs than the HRPE patients. However, further research is warranted to support the findings of our study in using sPESI for the outpatient management of the LRPE patients.

The findings from our study should be viewed in the context of some study limitations. First, the study relied on retrospective claims data. While claims data are extremely valuable for the efficient and effective examination of health care outcomes, treatment patterns, and costs, they are collected for payment and not research. To ensure exclusion of any rule-out PE diagnosis, PE patients were required to have an anticoagulant claim during their hospital stay. The presence of a claim for a filled prescription does not indicate the medication was consumed or taken as prescribed. Thus, the true number of medications prescribed may not be accurately recorded. Third, certain clinical and disease-specific parameters are not readily available in claims data, which could influence study outcomes. Additionally, the risk stratification in our study did not consider right ventricular dysfunction or elevated cardiac biomarkers which are considered as two of the important prognostic values in PE, and further analysis using these measures is warranted. Also, our study only quantified the overall clinical burden in the index hospitalization, and further research aiming to understand the early versus late complications after admission would be more beneficial in clarifying that the early discharge would minimize the risk of these HACs. The current study also represented only US data from a specific subpopulation (VHA veterans) who were mostly elderly men. Therefore, the general applicability of our findings to young male patients or females requires further study.

Conclusions

In summary, the results of this study showed that LRPE patients stratified using the sPESI score were unlikely to have complications or adverse clinical outcomes; hence, they had a lower clinical and economic burden than HRPE patients. Therefore, prognostic risk stratification may be considered a fundamental tool in choosing appropriate treatment in PE patients, and may substantially reduce the economic burden in LRPE patients.