Pulmonary embolism (PE) is a potentially life-threatening disease and the third most common cause of cardiovascular death. However, risk of death differs substantially across patients with acute PE according to a continuous spectrum of disease severity that results from different combinations of clinical features, right ventricle overload, troponin levels, and other biomarker abnormalities (brain natriuretic peptide and others).[2–4] In this complex clinical setting, clinical prediction models based on vital parameters and medical history can accurately identify patients with acute PE at low risk for death. The Pulmonary Embolism Severity Index (PESI) and the Geneva score were the first to be proposed and validated for risk stratification of patients with acute PE. The simplified version of the PESI score (sPESI) based on the assessment of six items (age, systolic blood pressure, heart rate, oxygen saturation, cancer, chronic obstructive pulmonary disease, or heart failure) allows identification of patients with acute PE and 30-day mortality as low as ~1% (upper 95% confidence limit 2.1). However, this score may appear too strict to fully summarize a medical judgement. In fact, in sPESI, age, blood pressure, and heart rate are assessed by the use of cut-off values, and cancer qualifies for adverse outcome regardless of staging or ongoing anticancer treatment. A more pragmatic approach based on the assessment of 11 criteria—the Hestia criteria—has been proposed to select low-risk patients for outpatient management. In a prospective study, 90-day mortality was 1% in patients with acute PE and none of the Hestia criteria for hospitalization who were managed as outpatients (upper 95% confidence limit 2.9). Limited evidence exists on the use of sPESI to drive home treatment in patients with acute PE, while Hestia criteria have been derived and validated for this purpose. The sPESI score and the HESTIA criteria have been endorsed by the current guidelines from the European Society of Cardiology for risk assessment and decision-making in patients with acute PE. Based on the same bulk of evidence but on a different medical approach, North American guidelines suggest not to replace clinical judgement with the use of clinical scores to predict outcome and make decisions on the management of patients with acute PE. Prediction models are increasingly used to complement clinical reasoning and decision-making in a number of clinical conditions and in the cardiovascular domain in particular (e.g. acute chest pain, atrial fibrillation, or transient ischaemic attack). Due to their simple structure, clinical models allow rapid bedside risk stratification and can drive decision-making concerning treatment and patient allocation. Preliminary data suggest similar outcomes in patients with acute PE managed based on clinical gestalt or on PESI/sPESI scores or Hestia criteria. However, the best strategy for selection of candidates for home treatment among patients with acute PE is still undefined.
In this issue of the European Heart Journal, Roy and colleagues report on a management study on the use of sPESI or Hestia criteria to triage patients with acute symptomatic PE for home treatment. In four European countries, 1975 normotensive patients with acute PE were randomized to be designated for home treatment based on either sPESI or the Hestia criteria. A specific patient pathway was set up before study initiation at all study centres to ensure that a dedicated clinical team could provide 24-h care by telephone service for patients discharged home. The primary outcome of the study—the composite of recurrent venous thrombo-embolism, major bleeding, or all-cause death at 30 days—occurred in 3.82% and in 3.57% of the patients randomized to triaging by Hestia or sPESI, respectively. Hestia was non-inferior to sPESI for triaging normotensive patients with acute symptomatic PE.
The results of this study should be interpreted taking into account some methodological aspects. Physicians in charge were allowed to overrule the sPESI and Hestia qualification for home treatment; according to the open study design, attending physicians were not blinded concerning the results of patients’ assessment by the comparator strategy (Hestia in patients randomized to sPESI, and vice versa) and concerning other predictors such as right ventricular dysfunction or troponin assessment. These elements of the study design could have reduced the impact of the randomized strategy on the clinical outcome. In this view, HOME-PE should be seen as a comparison of the feasibility of Hestia-driven vs. sPESI-driven management rather than an impact study. Finally, the individual values of the components of the primary composite endpoint—recurrent thrombo-embolism, death, and major bleeding—are probably different in the short-term after acute PE.
In HOME-PE, Hestia identified a lower proportion of patients as candidates for home treatment compared with sPESI (39.4% vs. 48.4%), but the proportion of patients actually managed at home was similar in the two triaging groups (38.4% vs. 36.3% in the Hestia and sPESI groups, respectively). The different performance can be explained by the different nature of the two strategies. sPESI is focused on the assessment of predictors of death, while Hestia includes assessment of predictors of death, bleeding, and feasibility of outpatient care (Graphical Abstract).
The incidence of recurrent venous thrombo-embolism, major bleeding, or death in patients who qualified for home treatment by the Hestia or sPESI strategy and were actually treated at home was as low as 1.3% and 1.1%, respectively. Death at 30 days occurred in 0.27% and 0.28% of the patients, respectively. Such low incidences may represent acceptable failure rates for clinicians.
Rates of the composite outcome and, more importantly, of death were ~50% lower in patients actually treated at home based on final medical judgement compared with those of patients who qualified for home treatment by the allocated risk stratification strategy. This 50% gap suggests the appropriateness of the final medical judgement over that of the rules. Whether other predictors (right ventricular dysfunction or injury) also influenced final judgement concerning home treatment remains undefined. In fact, in patients with acute PE at low risk according to clinical models, right ventricular dysfunction is associated with early mortality.
To date, two randomized clinical trials (445 patients) have compared hospitalization with home management in patients with PE deemed at low risk based on PESI or Hestia, one study for each. No significant difference was reported in all-cause mortality, recurrent PE, and major bleeding with hospitalization or home treatment, but the overall level of evidence is low. Recently, a large prospective cohort study confirmed the safety of home management in patients with acute PE selected on the basis of the Hestia criteria and on the absence of right ventricular dysfunction. However, in everyday clinical practice, only one in every 13 PE patients eligible for home treatment is treated at home and less than half of the low-risk population are hospitalized for 5 days or less. HOME-PE suggests that creating physician awareness and structural facilities makes home treatment feasible in ~35% of normotensive patients with acute PE. This finding mirrors the results obtained in the USA by Vinson and colleagues. These authors showed that the implementation and promotion of a clinical decision support system to aid physicians in decision-making for patients with acute PE safely increased outpatient management (17.4% pre- to 28.0% post-intervention). In clinical practice, hospitalization is chosen by clinicians to complete observation for the occurrence of clinical deterioration and education for anticoagulant treatment and by patients because hospitalization seems safe and protective. Indeed, ~5% of patients treated at home in the HOME-PE study required unscheduled hospitalization during the study period. Whether initial hospitalization would have avoided these events is unknown. However, physicians and patients should be aware of the risk and severity of hospital-acquired complications more than the costs of undue hospitalizations.
HOME-PE should be considered in the context of recent challenges in the management of patients with acute PE. The shift in the diagnostic paradigm from the intention of confirming PE in patients with high pre-test probability to that of excluding PE in patients with low pre-test probability led to an increased diagnosis of minor PEs in paucisymptomatic patients. The expanding armamentarium of anticoagulant agents moved from continuous infusion of unfractionated heparin with unpredictable effects to the use of low molecular weight heparin, and now to direct oral anticoagulants with predictable dose response, rapid onset of action, and no need for laboratory monitoring. These agents have improved the quality of anticoagulation in patients with acute PE. Both of these challenges probably contribute to make home treatment or short hospital stay feasible and safe in selected patients with acute PE.
In conclusion, HOME-PE is definitely the largest randomized study assessing home management in patients with acute PE and its results can contribute to the progressive refinement in risk stratification and definition of management pathways for patients with acute PE. Home treatment seems feasible in ~30% of the normotensive patients with acute PE, if adequate patient care is available as an outpatient service. Hestia and sPESI are both able to identify a subgroup of patients with acute PE at low risk for complications, but clinical judgement cannot be completely replaced.