J Thromb Thrombolysis DOI 10.1007/s11239-014-1141-y

Outpatient or inpatient treatment for acute pulmonary embolism: a retrospective cohort study of 439 consecutive patients Sebastian Werth • Virginia Kamvissi • Thoralf Stange • Eberhard Kuhlisch • Norbert Weiss • Jan Beyer-Westendorf

Ó Springer Science+Business Media New York 2014

Abstract Current guidelines consider outpatient treatment as an option for low-risk pulmonary embolism (PE), and risk assessment tools such as the HESTIA criteria can be used to identify PE patients who could feasibly be treated in an outpatient setting. Little is known about what proportion of patients in daily care this would comprise, and, in these patients, outcome data outside of clinical trials are scarce. To assess the proportion of PE patients receiving outpatient early discharge or in-hospital therapy, evaluate differences in patient characteristics between these subgroups and to assess clinical outcomes at 6 months. Monocentric, retrospective cohort study in 439 consecutive patients undergoing outpatient, early-discharge or in-hospital treatment for PE. Outcome data on recurrent VTE, pulmonary hypertension or death were collected from routine follow-up visits 6 months after VTE diagnosis. PE patients were treated as outpatient (OP; n = 49; 11.2 %); early-discharge (ED; n = 62; 14.1 %) or in-hospital (IH; n = 328; 74.7 %). Median duration of hospital stay in the ED and IH groups were 1 (IQR: 1) day and 9 (IQR: 7) days, respectively. Outcome event rates at 6 months were 3.9 % for recurrent VTE (95 % CI 2.3–6.1, similar between groups), 5.2 % for pulmonary hypertension (95 % CI 3.3–7.8, similar between groups) and 10.7 % for mortality (95 % CI 8.0–14.0). Mortality was significantly higher in IH patients (14.0 %; 95 % CI

S. Werth (&)
V. Kamvissi
N. Weiss
J. Beyer-Westendorf Center for Vascular Medicine and Department of Medicine III, Division of Angiology, University Hospital ‘‘Carl Gustav Carus’’, Technical University Dresden, Fetscherstrasse 74, 01307 Dresden, Germany e-mail: [email protected] T. Stange
E. Kuhlisch Institute for Medical Informatics and Biometry, Technical University Dresden, Dresden, Germany

10.5–18.3) compared to OP (0 %; 95 % CI 0.0–7.3) or ED (1.6 %; 95 % CI 0.0–8.7) patients. Mortality risk factors were high-risk ESC category (OR: 5.7), paraneoplastic VTE (OR: 3.0), need for oxygen supplementation (OR: 5.2), diabetes (OR: 2.5), age (OR per additional year: 1.1) and elevated INR (OR per 0.1 point increase: 1.5). No difference in the treatment groups for pulmonary hypertension during follow-up was found. Independent risk factors were thrombophilia (OR: 8.43), signs of right ventricular strain in baseline ECG (OR: 6.64) or echocardiography (RVESP [ 40 mmHg OR: 2.99). 32 % of the OP or ED patients had at least one criterion of the HESTIA score that would have excluded them from outpatient treatment. In daily care, treating PE in an almost exclusively outpatient setting seems feasible and safe for up to 25 % of all PE patients. The HESTIA criteria seem to exclude up to 30 % of patients for whom outpatient or early-discharge treatment seems feasible and safe. Keywords Management
Mortality
Outpatient treatment
Pulmonary embolism
Pulmonary hypertension

Introduction Over the last two decades, outpatient treatment of patients with acute deep vein thrombosis (DVT) has become widely accepted and is recommended in current guidelines [1, 2], as it has been shown to be effective and safe. Potentially, outpatient treatment for patients with acute pulmonary embolism (PE) could also offer benefits; however, the feasibility of outpatient PE treatment is still a matter of debate, since PE patients have a higher risk for complications and fatal outcomes than DVT patients without PE [3–6].

123

S. Werth et al.

PE patients show a broad variety of symptoms, which sometimes makes diagnosis of PE challenging. Furthermore, mortality of acute PE ranges between \1 and [50 %, depending on the extent of pulmonary artery occlusion and cardiac compensation reserves [7, 8]. Detailed risk assessments have been developed to identify low-risk PE patients who could potentially be feasible for outpatient treatment. One of these risk assessment tools is the Pulmonary Embolism Severity Index (PESI) score: a low 30-day mortality rate of 0.6 % for low-risk PE is defined as a low PESI score, which shows that outpatient treatment is safe [9]. Due to the complexity of the PESI, a simplification of this score (the simplified PESI [sPESI] score) was suggested, and derivation and validation studies demonstrated a diagnostic accuracy comparable to PESI with similarly low mortality rates in low-risk PE patients [10, 11]. In the prospective HESTIA study published in 2011 [12], PE patients were discharged from hospital within 24 h of PE diagnosis if the patient did not fulfill prespecified criteria of medical or social indication for inhospital treatment. In these patients, the authors found low complication rates (1.0 % mortality, 2.0 % recurrent VTE and 0.7 % major bleeding). Later studies of the same group confirmed the value of HESTIA criteria [1–3, 13–15]. Based on these and other studies [9], current guidelines recommend early discharge or even outpatient treatment for patients at low risk for PE or treatment complications [1, 16]. At our center, outpatient treatment of DVT patients has been standard of care for more than 10 years [17, 18]. During this time a significant number of daily care patients with low-risk PE had also received ambulatory treatment, with treatment decisions based on clinical experience rather than on HESTIA or PESI criteria, which were introduced only recently and have strong selected trial cohorts. Based on a large cohort of consecutive patients with objectively diagnosed PE at our vascular center, we performed a retrospective evaluation of the efficacy and safety of outpatient treatment in daily care. Furthermore, we evaluated differences in patient characteristics of PE subgroups of outpatient, early-discharge or in-hospital treatment in unselected patients from daily care and the influence of these characteristics on the outcome.

outpatients with suspected VTE) using the ICD-10 search code ‘‘I26.x’’ or ‘‘pulmonary embolism’’. All patients were screened for inclusion and exclusion criteria. Inclusion criteria for our study cohort were: 1.

2.

Patients who presented in an ambulatory setting with confirmed acute, symptomatic PE listed in the VTE database or the hospital database with an age of at least 18 years. PE that was objectively confirmed by: – – –

CT angiography V/Q-scan, or echocardiography with confirmed right heart strain in combination with confirmed DVT in complete compression ultrasound (cCUS).

Exclusion criteria of our study cohort were: 1.

2. 3. 4.

PE that was not objectively confirmed by CT angio graphy, V/Q-scan or echocardiography with confirmed right heart strain in combination with confirmed DVT. Asymptomatic PE (or symptoms persisting longer than 14 days). Inpatients hospitalized due to other preexisting comorbidity suffering PE. Pre-treatment in external hospital.

All patients fulfilling these criteria were included and classified according to their duration of hospital treatment. Subgroups were defined as: – – –

outpatient treatment (OP; in hospital for less than 24 h). early-discharge patients (ED; in hospital for 24–72 h) and. in-hospital treatment patients (IH; in hospital for [ 72 h).

Data collection The following outcome parameters within the follow-up of 6 months were registered: recurrent VTE (objectively confirmed DVT/PE), pulmonary hypertension (elevated RVESP [ 35 mmHg on echocardiography or symptomatic exertional dyspnea without any other known reason) and death. Diagnostic tests

Methods Patients All PE patients treated in our university hospital between 2000 and 2010 were identified via the hospital database and the VTE database of the Vascular Department (mainly

123

In this retrospective analysis, assessment of VTE recurrence was not standardized a priori. However, strict hospital guidelines are established that allowed accurate retrospective DVT and PE assessment: –

All patients with confirmed or suspected PE have to be examined for DVT by vascular physicians using cCUS

Outpatient or inpatient treatment for acute pulmonary embolism

Fig. 1 Flowchart of patient identification and study inclusion

–

following an established protocol [19]. Since in our hospital cCUS is available 24/7, it is recommended to be performed within 24 h after onset of PE symptoms. All patients with suspected PE undergo clinical pre-test probability (Wells score or revised Geneva score) and

D-Dimer testing according to guidelines. If indicated, these tests are followed by CT scan (recommended test) or V/Q scan (alternative test for patients with contraindication for contrast media) to confirm presence of PE.

123

S. Werth et al. Table 1 Baseline characteristics of the study cohort Criteria n (%)

Data available for: n = 439

Total cohort n = 439

Outpatient cohort n = 49 (11.2 %)

Early-discharge cohort n = 63 (14.4 %)

In-hospital cohort n = 327 (74.5 %)

Sex (male/ female)

439 (100)

243/196 (55.4/44.6)

26/23 (53.1/46.9)

43/20 (69.4/30.6)

VTE history (yes/ no)

439 (100)

123/316 (28.0/72.0)

22/27 (44.9/55.1)

Concomitant DVT (yes/no)

432 (98.2)

363/69 (84.0/16.0)

DVT symptoms (yes/no)

363 (82.7)

210/153 (57.9/42.1)

Mean age ± SD (years)

439 (100)

64.5 (16.7)

57.3 (16.9)

57.6 (14.5)

66.8 (16.4)

Mean BMI ± SD (kg/m2)

359 (81.8)

27.6 (5.5)

27.7 (4.8)

27.9 (5.9)

27.5 (5.5)

OP ? ED vs. IH p-value

OP vs. ED ? IH p-value

174/153 (53.0/47.0)

0.152

0.762

20/43 (32.3/67.7)

81/246 (24.7/75.3)

0.014

0.007

44/5 (89.8/10.2)

49/13 (79.0/21.0)

270/51 (84.1/15.9) [0.999

0.303

40/4 (90.9/9.1)

32/17 (65.3/34.7)

138/132 (51.1/49.9) \0.001

\0.001

\0.001

0.001

0.715

0.942

Statistically significant values with p-value below 0.05 are given in bold

–

–

–

If PE is confirmed, all patients need to undergo transthoracic echocardiography for risk stratification, performed by an experienced cardiologist for signs of right heart strain (RV dysfunction, kinetic dysfunction, tricuspid valve insufficiency with an elevated right ventricular end-systolic pressure [RVESP]). All patients with confirmed DVT are scheduled for a repeat cCUS procedure at 6 months to evaluate thrombus regression. All patients with confirmed PE and right heart strain are scheduled for repeated transthoracic echocardiographies before discharge and after 6 months to search for persistent or chronic pulmonary hypertension.

These strict hospital recommendations assured that DVT and PE diagnosis at baseline were objectively confirmed and that outcome assessment was based on objectively confirmed outcome events. Based on chart review of all patients, the ESC and HESTIA scores were evaluated retrospectively by a VTE expert. Thrombophilia testing is not routinely performed in our institution and is reserved for those with high clinical risk (indicative personal or family history, recurrent VTE during treatment, unusual-site VTE). Therefore, any patient with a ‘‘positive’’ thrombophilia testing did not only have an ‘‘abnormal value’’ but also had a clinical risk profile which led to the test in the first place. Most patients were tested at the end of treatment and, therefore, presence of thrombophilia was not known at the time of PE diagnosis and did not influence the decision-making for or against outpatient treatment. Thrombophilia testing always includes genetic testing for prothrombin mutation and Factor V Leiden mutation, testing of the coagulation parameters (aPTT, PT, INR, fibrinogen, antithrombin, and

123

protein C and S) and lupus anticoagulant, anticardiolipins, and anti-b2-glycoproteins. Statistical analysis To test for differences between groups, the three treatment groups were compared directly. Furthermore, analyses were performed for pooled ED and IH versus OP patients as well as IH patients versus pooled OP and ED patients. Statistical analyses were performed with SPSS 21 and R. Differences between patient cohorts were assessed using Fisher’s exact test, Student’s t test or Mann–Whitney U test, as appropriate. The 95 % confidence intervals for prevalence were calculated by Clopper–Pearson. Patients included in uni- and multivariate analyses had to have at minimum complete data on initial blood pressure, heart rate, body weight and a documented 6-month follow-up. Missing values beside the above-mentioned parameters were replaced by the expectation–maximization method. Using a logistic regression model, uni- and multivariate analyses were performed and included the following parameters: age, gender, size, weight, VTE history, diabetes, hypertension, hyperlipidemia, coronary heart disease, apoplexy, nicotine use, immobility, surgery in past 3 months, documented thrombophilia, oxygen supplementation, right heart strain in ECG or echocardiography, symptoms for DVT, proximal DVT, ESC risk score, reason for DVT, thrombolytic therapy, HESTIA score, INR, and creatinine clearance. In multivariate analysis the method of forward stepwise selection was used. Ethics The study protocol was approved by the local ethics committee (Dresden University Hospital; EK367112010).

Outpatient or inpatient treatment for acute pulmonary embolism

Due to the non-interventional, retrospective nature of our study, patient informed consent was not required.

Results Using the specified search methods, 1,015 patients with acute PE diagnosed in our institution between 2000 and 2010 were identified. Of these, 576 did not fulfill the inclusion criteria (Fig. 1). Consequently, 439 patients were included in our analysis. Of these, 49 patients (11.2 %) were treated as OP, 62 patients (14.1 %) were initially hospitalized but discharged within 24–72 h (ED) and 328 patients (74.7 %) were hospitalized for longer than 72 h (IH). Patient demographic characteristics are presented in Table 1. The mean age of all patients was 64.5 ± 16.7 years and 60.4 % of patients were at least 65 years old. The mean BMI was 27.6 ± 5.5 kg/m2 and 55.4 % were male. Median duration of PE symptoms was 4.1 days. 28.0 % of patients had a positive history for VTE and 84.0 % were found to have concomitant DVT (symptomatic in 57.9 %) at diagnosis of the index PE. Cohort characteristics for outpatient (OP), early discharge (ED) and in-hospital (IH) treatment Parameters of the clinical presentation of PE are presented in Table 2. Hospitalized patients were significantly older than patients in the OP and ED groups (66.8 vs. 57.3 vs. 57.6 years). Among the three groups, no differences were found with regard to the initial systolic blood pressure but patients in the OP group had significantly higher diastolic blood pressures compared to ED and IH patients (87.1 vs. 78.3 vs. 77.4 mmHg). Furthermore, OP patients had a significantly lower mean initial heart rate (79.1 vs. 87.3 vs. 93.3 bpm). In contrast, IH patients had significantly lower hemoglobin levels (8.0 vs. 8.4 vs. 8.7 mmol/l) and higher INR values on admission than OP and ED patients (1.10 vs. 1.07 vs. 1.01). The need for oxygen supplementation increased from OP (2.0 %) to ED (32.3 %) and further to IH (61.9 %), as did positive findings for elevated troponin (2.0 vs. 19.4 vs. 38.4 %) and right ventricular dysfunction in echocardiography (10.2 vs. 14.5 vs. 39.0 %). Finally, the rates of patients with a positive HESTIA score or ESC high-risk category also increased significantly from OP (12.2 and 0.0 %, respectively) to ED (46.8 and 4.8 %, respectively) and IH (78.0 and 13.1 %, respectively). The HESTIA score was zero in 33.7 % of all patients and 87.8 % of the outpatient-treated patients. The median (range; IQR) duration of hospitalization was 0 (0) days for OP, 1 (0–3; 1) for ED and 9 (3–72; 7) days for IH (p \ 0.001).

Efficacy and safety outcomes for outpatient (OP), early discharge (ED) and in-hospital (IH) treatment The efficacy and safety outcomes for the entire cohort and the three subgroups are presented in Table 3. The incidence of recurrent VTE was 3.9 % (95 % CI 2.3–6.1) for the total cohort and not significantly different between the treatment groups (OP 6.1 %; 95 % CI 1.3–17.9; ED 4.8 %; 95 % CI 1.0–13.3; IH 3.4 %; 95 % CI 1.7–5.9). The prevalence of pulmonary hypertension at 6 months was 5.2 % for the total cohort (95 % CI 3.3–7.8)with comparable rates in all treatment groups(OP 6.1 %; 95 % CI 1.3–16.9; ED 4.8 %; 95 % CI 1.0–13.5; IH 5.2 %; 95 % CI 3.0–8.2). Forty-seven patients died within 6 months of the index PE (10.7 %; 95 % CI 8.0–14.0), consisting of zero patients in the outpatient treatment group (0 %; 95 % CI 0.0–7.3), one patient in the ED group (1.6 %; 95 % CI 0.0–8.7) and 46 patients in the IH group (14.0 %; 95 % CI 10.5–18.3). Mortality was significantly higher in patients in the inhospital treatment group if compared separately to the OP and ED groups or against the pooled OP ? ED group. Furthermore, short-term mortality at day 30 was assessed and was found to be 7.1 % (31/439; 95 % CI 4.8–9.9). One death occurred in the ED group (1/62; 1.6 %, 95 % CI 0.0–8.7), 30 deaths occurred in the IH group (30/328; 9.1 %, 95 % CI 6.3–12.8) and none in the outpatient group (0 %, 95 % CI 0.0–7.3). The main cause for mortality within the first 30 days were VTE-related (VTE 80.6 % (95 % CI 62.5–92.5), cancer-related 16.1 % (95 % CI 5.5–33.7) and other cause of death 3.2 % (95 % CI 0.0–16.7). In the subgroup of patients with paraneoplastic PE (n = 72), five patients were treated as outpatients, 15 were in the ED group and 52 patients were treated in hospital. In this subgroup, only one patient died within the six-month follow-up (5 %) in the OP ? ED group compared to 11 patients (21 %) in the IH group. Risk assessment for recurrent VTE, pulmonary hypertension or death at 6 months To assess the impact of potential risk factors on the rates of recurrent VTE, pulmonary hypertension or death during the 6-month follow-up, uni- and multivariate analyses were performed. Tables 4, 5 and 6 illustrate the parameters with significant p-values in univariate analysis. For recurrent VTE, the presence of concomitant symptomatic DVT (OR 2.69; 95 % CI 1.36–5.30; p = 0.004) was the only independent risk factor in uni- and multivariate analyses, whereas increasing age was found to be associated with a slight but significant risk decrease (OR per year 0.97; 95 % CI 0.96–0.99; p = 0.005).

123

123 0.99–1.04

439 (100)

High-risk ESC score

HESTIA score [ 0

Statistically significant values with p-value below 0.05 are given in bold

Values in g/dl are in italic

356 (81.1) 430 (97.9)

Right ventricular dysfunction on echocardiography

371 (84.5)

1.01

75.3–83.0

6

0

5

1

1

22

12.2

0

10.2

2.0

2.0

44.9

29

3

9

12

21

20

46.8

4.8

14.5

19.4

32.3

32.3

%

n

%

n

1.03–1.10

8.1–8.7 (13.0–14.0)

82.7–92.0

74.7–81.8

130.1–141.1

53.9–61.3

ED n = 62 (14.1 %)

1.07

8.39 (13.51)

87.33

78.25

135.61

57.63

OP n = 49 (11.2 %)

8.4–9.0 (13.5–14.5

Elevated troponin

417 (95.0)

INR

8.68 (13.98)

79.13

439 (100)

423 (96.4)

Hemoglobin (mmol/l - g/dl)

81.1–93.1

439 (100)

410 (93.4)

Pulse rate (bpm)

87.08

136.4–149.9

52.5–62.2

O2 supplementation

409 (93.2)

Diastolic BP (mmHg)

143.18

57.33

History of VTE

410 (93.4)

Systolic BP (mmHg)

Data availability n/439 (%)

439 (100)

Age (y)

95 % CI

Mean

Mean

95 %CI

ED n = 62 (14.1 %)

OP n = 49 (11.2 %)

Criteria

Data availability n/439 (%)

Criteria

1.08–1.12

256

43

128

126

203

81

n

78.0

13.1

39.0

38.4

61.9

24.7

%

0.027

\0.001

\0.001 \0.001

\0.001

0.016

\0.001

\0.001 0.003

0.021 \0.001

0.030

p-value

\0.001

p-value

OP vs. ED ? IH

0.018

0.012

\0.001

\0.001 0.003

\0.001 0.015

0.123

0.003

\0.001 0.498

p-value

OP vs. ED ? IH p-value

OP ? ED vs. IH

OP ? ED vs. IH

7.9–8.2 (12.7–13.2)

90.9–95.7

76.0–78.8

133.0–137.6

65.0–68.6

95 % CI

IH n = 328 (74.7 %)

1.10

8.03 (12.93)

93.30

77.41

134.82

66.80

Mean

IH n = 328 (74.7 %)

Table 2 Clinical presentation and lab parameters of the three subgroups of PE patients: outpatient treatment (OP), early discharge\72 h (ED) and hospital treatment[72 h (IH). p-values were corrected for multiple testing

S. Werth et al.

0.733

0.015

[0.999

\0.001 Statistically significant values with p-value below 0.05 are given in bold

0.734

0.503

0.009

[0.999 [0.999 3.0–8.2

10.5–18.3

17 (5.2)

46 (14.0) 0.0–8.7

1.0–13.5 3 (4.8)

Discussion

1 (1.6) 0–7.3

1.3–16.9 3 (6.1)

0 (0.0) 8.0–14.0

23 (5.2)

47 (10.7)

Pulmonary hypertension

Death during FU

3.3–7.8

For prevalence of pulmonary hypertension at 6-month follow-up, uni- and multivariate analyses confirmed known thrombophilia (OR 8.35; 95 % CI 1.13–62.7; p = 0.037), signs of right heart strain in baseline ECG (OR 6.6; 95 % CI 2.08–21.1; p = 0.001) and RVESP [ 40 mmHg in baseline echocardiography (OR 2.99; 95 % CI 1.01–8.82; p = 0.047) as independent risk factors. For all-cause mortality, uni- and multivariate analyses established the following as independent risk factors: highrisk ESC category (OR 5.71; 95 % CI 2.35–13.9; p = 0.001), the need for oxygen supplementation (OR 5.22; 95 % CI 1.86–14.6; p = 0.002), paraneoplastic DVT (OR 3.01; 95 % CI 1.29–7.01; p = 0,011), diabetes (OR 2.26; 95 % CI 1.04–4.94; p = 0.040), increasing age (OR per additional year 1.1; 95 % CI 1.05–1.15; p \ 0.001), and elevated INR values at time of PE diagnosis (OR per INR increase of 0.1 point 1.5; 95 % CI 1.23–1.85; p \ 0.001). In patients with a proximal DVT the risk for mortality is decreased (OR 0.47; 95 % CI 0.22–0.99).

0.006

[0.999 [0.999 [0.999 1.7–5.9 11 (3.4) 1.0–13.5 3 (4.8) 1.3–17.9 3 (6.1) 17 (3.9) Recurrent VTE

2.3–6.1

n (%) 95 % CI n (%)

95 % CI

n (%)

95 % CI

[0.999

[0.999

p-value OP vs. ED ? IH p-value OP ? ED vs. IH p-value OP vs. IH p-value ED vs. IH p-value OP vs. ED IH ED OP 95 % CI Total cohort n (%) Outcome at 6 month

Table 3 Rates of outcome events for the entire cohort and treatment subgroups: outpatient treatment (OP), early discharge \72 h (ED) and hospital treatment [72 h (IH). p-values were corrected for multiple testing

Outpatient or inpatient treatment for acute pulmonary embolism

Our study represents one of the first evaluations of outpatient treatment in an unselected cohort of PE patients seen in daily care. We evaluated the clinical course of both in- and outpatients treated for PE in daily care; this is in contrast to the retrospective chart analysis of Jimenez et al. [4], which evaluated the risk stratification using PESI or the Geneva prediction rule in hospitalized patients to identify low-risk PE patients for potential outpatient treatment. Our study is also different from the HESTIA study [5], which prospectively followed low-risk PE patients who were treated as outpatients based on predefined inclusion and exclusion criteria. As a consequence, in the HESTIA study, only a subset of patients was studied, and [40 % of patients were not eligible for outpatient treatment and were not further characterized. The selection process in these previous studies has been suggested to limit the generalizability of the study findings, which is still a matter of concern [6]. Therefore, our data help to put existing data on outpatient PE treatment into perspective. The demographic characteristics in our cohort are consistent with other trial cohorts in the literature [5, 7–9]. However, we found higher rates of concomitant DVT (84 %) compared to the GENEVA or PESI cohorts (45–50 %) and also higher rates than those reported by other groups or meta-analyses [20, 21]. We believe that there are at least three possible explanations. First, as part of our institutional policy, virtually all PE patients (98.2 %) underwent cCUS for DVT. Second, a standardized cCUS protocol that includes bilateral examinations and the evaluation of calf veins was introduced in our

123

S. Werth et al. Table 4 Uni- and multivariate analyses for recurrent VTE at 6 month follow-up Parameter

Univariate analysis

Multivariate analysis

pvalue

OR (95 % CI)

pvalue

OR (95 % CI)

Symptomatic DVT

0.001

3.30 (1.72–6.35)

0.004

2.69 (1.36–5.30)

Proximal DVT

0.006

3.19 (1.39–7.31)

0.067

2.25 (0.95–5.36)

Statistically significant values with p-value below 0.05 are given in bold

institution more than 15 years ago and has been used extensively in clinical and scientific settings [10, 11]. As a result, we believe that our sensitivity to detect DVT may be higher than in centers with less experience or less extensive ultrasound protocols. Third, as a result of our 24/7 cCUS availability, almost all patients with suspected PE undergo cCUS screening within 24 h of PE suspicion. As a consequence, the rate of concomitant DVT can be expected to be higher than in cohorts that receive ultrasound screening sometime during the PE work-up, due to early clot resolution in some patients. Proportion of patients feasible for outpatient PE treatment

physician is at least as safe as an established risk score for the decision for outpatient treatment. Furthermore, our data indicate that the use of the HESTIA score seems to exclude as many as 30 % of patients from home treatment for whom outpatient or early discharge treatment seems feasible. Patients with a history of VTE seem to be more viable candidates for outpatient treatment. While the proportion of patients with a history of VTE was 45 % in the outpatient and early-discharge cohorts, it was only 25 % in the cohort of hospitalized patients. The reasons for this finding are unclear but it seems reasonable to conclude that in patients with a history of VTE the suspicion of recurrent PE is raised earlier, resulting in less severe PE at the time of diagnosis. As expected, patients requiring in-hospital treatment were almost a decade older and had more severe PE-related findings such as lower diastolic blood pressure, higher heart rate, a higher demand for oxygen supplementation and a higher proportion of positive troponin or right heart strain in echocardiography. This finding is in line with recent trials that indicated that the severity of PE may increase with age: in the PEITHO trial [12], patients with intermediate-risk PE were nearly a decade older than patients with less severe PE studied in other recent PE trials such as EINSTEIN PE (66.5 vs. 57.9 years) [13]. Outcome event rates and risk factors

In our institution as many as 11 % of patients with acute PE were treated as outpatients. Furthermore, another 15 % of acute PE patients were admitted to hospital but were discharged within 1–3 days (median 1 day), indicating logistic reasons rather than PE severity as the reason for hospitalization. Therefore, the total proportion of patients potentially eligible for outpatient PE treatment in daily care could reach up to 25 %. Interestingly, about 32 % of the patients in our outpatient or early-discharge cohorts had a positive HESTIA score and 3.7 % even fulfilled ‘‘intermediate risk’’ criteria of the ESC score (signs of right heart strain in echocardiography or troponin values slightly above the upper limit of normal). Despite this, the overall outcome of these cohorts was favorable, with a mortality rate below 1 %, indicating that the subjective assessment of an experienced attending Table 5 Uni- and multivariate analyses for pulmonary hypertension at 6 month followup

123

Parameter

In the 111 patients of the outpatient and early-discharge cohorts we found the risk of recurrent VTE (5.4 %) or mortality (0.9 %) during 6-month follow-up to be acceptably low. These data are consistent with the findings from HESTIA (2 % recurrent VTE and 1 % mortality after 3 months) and the low-risk PESI group (0.9 % mortality in PESI groups I and II) and lower than in the derivation and validation cohorts of low-risk patients stratified by the Geneva score (mortality rate 5.6 % for low risk-patients at 30 days) [4, 5]. The prevalence of pulmonary hypertension at 6 months was 5.2 % in our cohort and did not differ between the treatment cohorts. This prevalence is higher than in the literature (0.4–4.4 %) [14–16]; this might be explained by the fact that most patients underwent screening Univariate analysis

Multivariate analysis

p-value

p-value

OR (95 % CI)

OR (95 % CI)

History of VTE

0.031

2.60 (1.09–6.20)

0.054

2.49 (0.99–6.29)

Known thrombophilia

0.076

4.31 (0.86–21.66)

0.037

8.42 (1.13–62.69)

Right heart strain on ECG RVESP [ 40 mmHg

0.001 0.002

7.41 (2.67–20.60) 5.05 (1.82–14.00)

0.001 0.047

6.64 (2.08–21.14) 2.99 (1.01–8.82)

Outpatient or inpatient treatment for acute pulmonary embolism Table 6 Uni- and multivariate analyses for death during 6-month follow-up Parameter

Univariate analysis

Multivariate analysis

p-value

OR (95 % CI)

p-value

OR (95 % CI)

GFR \ 50

0.002

2.78 (1.47–5.26)

0.040

2.26 (1.04–4.94)

Diabetes

0.001

3.26 (1.68–6.32)

0.025

2.48 (1.12–5.48)

Need for O2 supplementation

0.001

8.71 (3.35–22.62)

0.002

5.22 (1.86–14.63)

Troponin positive

0.001

4.06 (2.13–7.73)

0.064

2.02 (0.96–4.23)

Proximal DVT

0.039

0.51 (0.27–0.97)

0.048

0.47 (0.22–0.99)

High risk in ESC score

0.001

7.36 (3.50–15.48)

0.001

5.71 (2.35–13.86)

DVT with trigger Paraneoplastic DVT

0.052 0.005

0.30 (0.09–1.01) 2.72 (1.35–5.49)

0.084 0.011

0.29 (0.07–1.18) 3.01 (1.29–7.02)

echocardiography at 6 months in our institution, which is also a pulmonary hypertension reference center. Multivariate risk factor analysis identified different risk factor profiles for VTE recurrence, prevalence of pulmonary hypertension or all-cause mortality at 6 months. Since the number of variables included in such analyses should be limited, we did not include established risk factors such as cancer, COPD, congestive heart failure, troponin or BNP in our analysis, since much larger studies have clearly demonstrated their impact on poor VTE outcome. In contrast to this, we decided to include established cardiovascular risk factors such as hyperlipidemia or diabetes in our risk factor assessment, since recent studies have indicated their impact on VTE [22–24]. Furthermore, given the demographic characteristics of our patients we suspected that the cardiovascular risk factors could have an impact on all-cause mortality. From the results of the multivariate analysis of risk factors for pulmonary hypertension it can be concluded that patients with known thrombophilia have a higher risk to develop chronic thromboembolic pulmonary hypertension (CTEPH), probably due to a higher risk of recurrent VTE, which is in line with previous data [25]. Of note, in our institution patients are not routinely tested for thrombophilia and testing is reserved for those with high clinical risk (indicative personal or family history, recurrent VTE during treatment, unusual-site VTE). Therefore, any patient with a ‘‘positive’’ thrombophilia testing did not only have an ‘‘abnormal value’’ but also had a clinical risk profile which led to the test in the first place. Most patients were tested at the end of treatment and, therefore, presence of thrombophilia was not known at the time of PE diagnosis and did not influence the decisionmaking for or against outpatient treatment. However, the selective thrombophilia testing may have resulted in the comparatively high impact of ‘‘known thrombophilia’’ on the prevalence of PH at 6 months (OR 8.4). For mortality, our data are consistent with previous studies. Multivariate analysis showed—not surprisingly— an increased mortality in patients with high ESC risk score;

this association has also been demonstrated by other groups [26] and is reflected by guideline recommendations to use the ESC risk score for risk stratification [16]. Surprisingly the lower systolic blood pressure did not influence the risk of mortality, possibly due to the fact that, out of the 46 patients with a high ESC risk score, 21 had a systolic blood pressure greater than 100 mmHg but fulfilled other ESC ‘‘high risk’’ criteria. We also found higher mortality for patients with a coexisting malignant disease and a trend towards reduced mortality in patients with a provoked VTE, which has also been demonstrated by others [27, 28]. Another important finding of our study is the fact that that an INR value [2.0 at the time of PE diagnosis was associated with a 59-fold increase in mortality. This finding has never been described before. Two hypotheses are possible: first, that patients suffering another PE during anticoagulant treatment with vitamin K antagonists (VKAs) have a severe underlying procoagulant disease (i.e. paraneoplastic disease) or, second, that the elevated INR at the time of PE diagnosis could be a sign for a relevant hepatic disorder or early cardiogenic shock. Of the 12 patients with INR [1.5 at baseline, only two had a documented VKA intake at the time of PE diagnosis. Further eight patients had clinically relevant organ dysfunction due to acute infections or cardiac diseases and it seems reasonable to conclude that in those patients without VKA exposure, elevated INR was due to a severe organ dysfunction, which could have contributed to the excess mortality.

Limitations A few limitations of our study need to be addressed. Our data are derived from a retrospective cohort study. Consequently, our data collection was limited to the data documented in databases and patient charts. While only very few data were missing, these might have limited our

123

S. Werth et al.

evaluations. Due to German data protection laws, we were unable to access death certificates. As a consequence, the cause of death could not exactly be determined in some patients. However, since we assessed all-cause mortality, this limitation did not affect our analysis. We did not evaluate bleeding complications, which is one of the most common complications of anticoagulant or thrombolytic PE treatment. During chart review, we noted that a significant documentation bias between the subgroups exists: while patients in the in-hospital group were regularly seen (and documented) in the early phase, they did not undergo regular follow-up visits for several months and were only seen again at 6 months (apart from few patients that presented earlier with acute problems). In contrast, patients in the outpatient and early-discharge groups were only seen at the very beginning of their treatment (1–2 days) and not consistently over the first 7–10 days (as in-hospital patients were). Furthermore, large proportions of the outpatient and early-discharge groups participated in clinical trials and underwent monthly follow-up visits during the 6 months of follow-up. Therefore, we found documented bleeding events mostly for the first 10 days for the in-hospital patients (and few events after discharge) and, in contrast, few bleeding events in the first days in the outpatient and early-discharge group, but considerably more documented events during the monthly trial visits that are specifically designed to collect safety outcomes. In contrast to VTE recurrence of pulmonary hypertension (which are objectively documented by imaging procedures) or all-cause mortality (which is a robust endpoint hardly influenced by biases), bleeding in most cases does not lead to a contact with a specialized facility but is treated by the primary care physician or the patient himself. As a consequence of this, objective bleeding assessment and documentation is especially influenced by the frequency of follow-up contacts. Given that the observed difference in visit schedules resulted in a relevant documentation bias we excluded bleeding evaluation completely. A relevant proportion of patients had no oxygen saturation documented at the time of initial presentation. As a consequence, we were unable to calculate the PESI score in many patients despite the fact that all other score parameters were available. Therefore, we excluded PESI score analysis altogether to avoid potential bias. We accept that the PESI score is a valuable tool in the risk stratification and that this score would have been a useful criterion in our analyses. Interestingly, according to the original publication of the derivation and validation cohorts of the PESI score, arterial oxygen saturation was not documented in 45.2 and 64.6 % of all patients, respectively [29]. Therefore, the inclusion of arterial oxygen saturation in the final PESI score is not beyond question, which puts the lack of

123

arterial oxygen saturation values in some of our patients into perspective. Our data are derived from consecutive patients treated in a single academic hospital specializing in all aspects of modern VTE diagnosis and treatment. As a consequence, outpatient treatment has been performed in some low-risk patients based on physician experience and good clinical practice rather than scientific evidence and before HESTIA and PESI criteria were introduced. While our data indicate that this has been feasible, we accept that our data may not be generalizable to hospitals with less experience in VTE treatment. As a consequence, we strongly support the ongoing work of many groups to evaluate prognostic markers and risk-stratification tools for patients potentially feasible for outpatient PE treatment. Finally, the outcome parameters were only analyzed as outcome rate at the time point of the follow-up after 6 months. Unfortunately, the exact time point of the outcome event was not exactly determinable in some patients. Therefore, we were unable to perform time-to-first-event or Cox regression analyses, which would have been useful to further analyze our data. In contrast to the described limitations, the monocentric design of our study and the evaluation of an unselected cohort of consecutive patients, together with the fact that treatment decisions and follow-up examinations for almost all PE patients in our hospital have been made by the same group of physicians in the Vascular Department throughout the observational period, leads to a high degree of standardization in the risk stratification process, which is a significant strength of our study.

Conclusion We conclude that approximately 25 % of patients with acute PE seem to be candidates for outpatient treatment in daily care. While HESTIA criteria certainly identify patients feasible for outpatient treatment, they exclude up to 30 % of potentially eligible patients. Therefore, more prognostic markers and better defined risk-stratification tools are urgently needed to identify patients for whom routine outpatient PE treatment is effective and safe. From the viewpoint of the patient, but also for economic and medical considerations, the extension of outpatient treatment from DVT to PE patients is desirable. In addition to optimizing therapy, this can be achieved mainly by improving risk stratification. VTE specialists are more and more able to predict the prognoses of these patients by risk scores, whether established or in development. Our study hopefully adds to this knowledge and indicates the feasibility and safety of outpatient treatment for PE patients seen in daily care settings.

Outpatient or inpatient treatment for acute pulmonary embolism Conflict of interest None of the authors declared a conflict of interest with regard to the presented data. No financial or editorial support was obtained for this study or the writing of the manuscript.

15.

References

16.

1. Kearon C, Akl EA, Comerota AJ et al (2012) Antithrombotic therapy for VTE disease antithrombotic therapy and prevention of thrombosis, 9th Ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 141:e419S– e494S. doi:10.1378/chest.11-2301 2. Bla¨ttler W, Gerlach H, Hach-Wunderle et al (2010) Diagnostik und Therapie der Venenthrombose und der Lungenembolie. Vasa 39:1–39 3. Stein EM, Carrelli A, Young P et al (2013) Bariatric surgery results in cortical bone loss. J Clin Endocrinol Metab 98:541–549. doi:10.1210/jc.2012-2394 4. Lankeit M, Konstantinides S (2012) Is it time for home treatment of pulmonary embolism? Eur Respir J. doi:10.1183/09031936. 00216811 5. Husted SE, Nielsen HK (2010) Treatment of pulmonary embolism at home? Thromb Res 126:3–4. doi:10.1016/j.thromres. 2009.12.020 6. Tapson VF, Huisman MV (2007) Home at last? Early discharge for acute pulmonary embolism. Eur Respir J 30:613–615. doi:10. 1183/09031936.00098007 7. Konstantinides S, Geibel A, Olschewski M et al (1997) Association between thrombolytic treatment and the prognosis of hemodynamically stable patients with major pulmonary embolism: results of a multicenter registry. Circulation 96:882–888 8. Kucher N, Rossi E, De Rosa M, Goldhaber SZ (2005) Prognostic role of echocardiography among patients with acute pulmonary embolism and a systolic arterial pressure of 90 mm Hg or higher. Arch Intern Med 165:1777–1781. doi:10.1001/archinte.165.15. 1777 9. Aujesky D, Roy P-M, Verschuren F et al (2011) Outpatient versus inpatient treatment for patients with acute pulmonary embolism: an international, open-label, randomised, non-inferiority trial. Lancet 378:41–48. doi:10.1016/S01406736(11)60824-6 10. Jime´nez D, Aujesky D, Moores L et al (2010) Simplification of the pulmonary embolism severity index for prognostication in patients with acute symptomatic pulmonary embolism. Arch Intern Med 170:1383–1389. doi:10.1001/archinternmed.2010.199 11. Venetz C, Jime´nez D, Mean M, Aujesky D (2011) A comparison of the original and simplified Pulmonary Embolism Severity Index. Thromb Haemost 106:423–428. doi:10.1160/TH11-040263 12. Zondag W, Mos ICM, Creemers-Schild D et al (2011) Outpatient treatment in patients with acute pulmonary embolism: the Hestia Study. J Thromb Haemost 9:1500–1507. doi:10.1111/j.15387836.2011.04388.x 13. Zondag W, den Exter PL, Crobach MJT et al (2013) Comparison of two methods for selection of out of hospital treatment in patients with acute pulmonary embolism. Thromb Haemost 109:47–52. doi:10.1160/TH12-07-0466 14. Zondag W, Vingerhoets LMA, Durian MF et al (2013) Hestia criteria can safely select patients with pulmonary embolism for

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

outpatient treatment irrespective of right ventricular function. J Thromb Haemost 11:686–692. doi:10.1111/jth.12146 Zondag W, Hiddinga BI, Crobach MJT et al (2013) Hestia criteria can discriminate high- from low-risk patients with pulmonary embolism. Eur Respir J 41:588–592. doi:10.1183/ 09031936.00030412 Torbicki A, Perrier A, Konstantinides S et al (2008) Guidelines on the diagnosis and management of acute pulmonary embolism: the Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC). Eur Heart J 29:2276–2315. doi:10.1093/eurheartj/ehn310 Schwarz T, Schmidt B, Beyer J et al (2001) Eligibility for home treatment of deep vein thrombosis: a prospective study in 202 consecutive patients. J Vasc Surg 34:1065–1070. doi:10.1067/ mva.2001.118821 Schellong SM, Schwarz T, Kropp J et al (1999) Bed rest in deep vein thrombosis and the incidence of scintigraphic pulmonary embolism. Thromb Haemost 82(Suppl 1):127–129 Schwarz T, Schmidt B, Schmidt B, Schellong SM (2002) Interobserver agreement of complete compression ultrasound for clinically suspected deep vein thrombosis. Clin Appl Thromb Hemost 8:45–49 Van Rossum AB, van Houwelingen HC, Kieft GJ, Pattynama PM (1998) Prevalence of deep vein thrombosis in suspected and proven pulmonary embolism: a meta-analysis. Br J Radiol 71:1260–1265. doi:10.1259/bjr.71.852.10318998 Girard P, Musset D, Parent F et al (1999) High prevalence of detectable deep venous thrombosis in patients with acute pulmonary embolism. Chest 116:903–908 Bell EJ, Selvin E, Lutsey PL et al (2013) Glycemia (hemoglobin A1c) and incident venous thromboembolism in the Atherosclerosis Risk in Communities cohort study. Vasc Med 18:245–250. doi:10.1177/1358863X13506764 Glynn RJ, Danielson E, Fonseca FAH et al (2009) A randomized trial of rosuvastatin in the prevention of venous thromboembolism. N Engl J Med 360:1851–1861. doi:10.1056/ NEJMoa0900241 Oren E, Smith NL, Doggen CJM et al (2006) Body mass index and the risk of venous thrombosis among postmenopausal women. J Thromb Haemost 4:2273–2275. doi:10.1111/j.15387836.2006.02119.x Kim NH, Lang IM (2012) Risk factors for chronic thromboembolic pulmonary hypertension. Eur Respir Rev 21:27–31. doi:10. 1183/09059180.00009111 Vanni S, Nazerian P, Pepe G et al (2011) Comparison of two prognostic models for acute pulmonary embolism: clinical vs. right ventricular dysfunction-guided approach. J Thromb Haemost 9:1916–1923. doi:10.1111/j.1538-7836.2011.04459.x Levitan N, Dowlati A, Remick SC et al (1999) Rates of initial and recurrent thromboembolic disease among patients with malignancy versus those without malignancy. Risk analysis using Medicare claims data. Medicine (Baltimore) 78:285–291 Hogg K, Hinchliffe E, Haslam S et al (2013) Predicting short term mortality after investigation for venous thromboembolism. Thromb Res 131:e141–e146. doi:10.1016/j.thromres.2013.01.030 Aujesky D, Obrosky DS, Stone RA et al (2005) Derivation and validation of a prognostic model for pulmonary embolism. Am J Respir Crit Care Med 172:1041–1046. doi:10.1164/rccm.200506862OC

123