Clinical background
The simplified PESI (sPESI) was constructed to simplify the original Pulmonary Embolism Severity Index (PESI), which comprises eleven variables with different weights and age divided into deciles. PESI is prognostically strong but cumbersome to calculate during an acute assessment. The need for a simpler instrument arose because risk stratification in pulmonary embolism must in practice be done quickly, often in the emergency department, and because the decision concerns whether the patient can be managed as an outpatient or requires admission.
sPESI retains six of the PESI variables, coded as binary, and uses a single threshold: zero points defines low risk. The instrument is therefore not intended to grade the severity of risk among high-risk patients, but to identify with high sensitivity those whose 30-day mortality is low enough for outpatient management to be considered.
Calculating the simplified PESI
Each variable scores 1 point if the criterion is met, otherwise 0. The score is the sum of the criteria met, giving a range from 0 to 6. A score of 0 defines low risk; a score of 1 or more defines elevated risk.
The derivation cohort consisted of 995 Spanish patients with acute symptomatic pulmonary embolism recruited in the outpatient setting [1]. Jiménez and co-workers applied univariate logistic regression to the eleven original PESI variables and eliminated those that did not reach statistical significance for predicting 30-day mortality. What remained were age over 80 years, previous cancer, chronic cardiopulmonary disease, heart rate, systolic blood pressure and oxygen saturation. In the derivation cohort, 30.7% were classified as low risk with a 30-day mortality of 1.0% (95% CI 0.0 to 2.1%), compared with 10.9% in the high-risk group. Discrimination measured as the AUC was 0.75 (95% CI 0.69 to 0.80), which did not differ from the original PESI [1].
External validation was performed immediately in the independent multinational RIETE cohort of 7,106 patients, in which 36.2% were classified as low risk with a 30-day mortality of 1.1% (95% CI 0.7 to 1.5%) compared with 8.9% in the high-risk group [1].
Interpretation in practice
| sPESI | Risk category | 30-day mortality (pooled) | Clinical action |
|---|---|---|---|
| 0 | Low risk | approx. 1.0 to 1.5% | Outpatient management may be considered if home circumstances are good, there is no serious comorbidity and reliable follow-up can be guaranteed |
| ≥1 | Elevated risk | approx. 8.9 to 10.7% | Admission indicated; consider further risk stratification with biomarkers and imaging of right ventricular function |
A patient with zero points has, in meta-analyses, a pooled 30-day mortality of 1.5% (95% CI 0.9 to 2.5%) based on eleven studies [2]. This lies below the threshold generally accepted for safe outpatient treatment. It is important, however, that the score is never the only factor: the patient's social situation, analgesia, ability to adhere to treatment and access to follow-up must all be taken into account.
For patients with one or more points, sPESI offers no finer grading. A patient with one point and a patient with four points fall into the same category. To differentiate among high-risk patients, supplementary assessment is needed, usually with biomarkers (troponin, NT-proBNP) and echocardiography to identify right ventricular strain, in line with the ESC risk stratification model [5].
Validation and performance
sPESI has been validated in several independent cohorts with consistent results. In a systematic review and meta-analysis by Elias and co-workers including 71 studies with a total of 44,298 patients, sPESI together with the original PESI were the most extensively validated models [2]. Pooled 30-day mortality in the low-risk group was 1.5% (95% CI 0.9 to 2.5%) and in the high-risk group 10.7% (95% CI 8.8 to 12.9%), based on eleven studies of sPESI [2].
In an external validation in a North American cohort of 807 patients at Hartford Hospital, sPESI showed a sensitivity of 100% for in-hospital mortality with a negative predictive value of 100% [3]. No low-risk patient died during the hospital stay. The c-statistic was 0.76 for in-hospital mortality and 0.73 for 30-day mortality [3]. Specificity was low, around 27%, reflecting the fact that the instrument is constructed for high sensitivity at the expense of specificity: a substantial proportion of patients who survive are nonetheless classified as high risk.
In a comparative study by Venetz and co-workers in 15,531 patients, sPESI classified 36.8% as low risk compared with 40.9% for the original PESI [4]. Low-risk mortality was 2.7% for sPESI compared with 2.3% for PESI. Sensitivity (89% versus 90%) and negative predictive value (97% versus 98%) were comparable. The AUC was, however, significantly lower for sPESI: 0.72 (95% CI 0.71 to 0.74) compared with 0.78 (95% CI 0.77 to 0.79) for PESI (p < 0.001) [4]. The original PESI therefore classified a larger proportion of patients as low risk and had better discrimination, but the clinical utility of sPESI, measured as the ability to rule out mortality, was in practice equivalent.
Limitations
sPESI applies only to patients with confirmed acute pulmonary embolism and a normal or only moderately reduced blood pressure. Patients with haemodynamic instability (systolic blood pressure <90 mmHg or a need for vasopressors) are by definition high risk and should not be risk-stratified with sPESI; they should be assessed for reperfusion treatment.
The instrument was derived in patients with symptomatic pulmonary embolism. It has not been validated for incidental pulmonary embolism, that is, embolism discovered incidentally on imaging performed for another reason. This population generally has a lower mortality, and sPESI may therefore overestimate the risk.
Patients with active cancer form a particular group. Previous cancer is one of the variables, but the instrument takes no account of the type or stage of the cancer or of ongoing chemotherapy, all of which markedly affect prognosis. In the meta-analysis, mortality in the low-risk group was somewhat higher for sPESI than for PESI, which may in part be explained by the fact that cancer patients systematically score a point and therefore rarely fall into the low-risk group; those who do so despite a cancer diagnosis may have an elevated risk that the instrument does not capture [2,4].
The low specificity (below 40% in most validation studies) means that a large proportion of patients who would manage outpatient treatment are nonetheless classified as high risk [3]. This is a deliberate design trade-off: the instrument prioritises not missing any patient who dies, but the price is that fewer patients become candidates for outpatient management than would be possible with a more specific instrument.
sPESI gives no information about bleeding risk, which is central to the decision on outpatient anticoagulation. A patient with an sPESI of 0 but a high bleeding risk should not be treated at home on the basis of the score alone.
References
- Jiménez D, Aujesky D, Moores L, et al. Simplification of the pulmonary embolism severity index for prognostication in patients with acute symptomatic pulmonary embolism. Arch Intern Med. 2010;170(15):1383–9. PMID: 20696966
- Elias A, Mallett S, Daoud-Elias M, et al. Prognostic models in acute pulmonary embolism: a systematic review and meta-analysis. BMJ Open. 2016;6(4):e010324. PMID: 27130162
- Weeda ER, Kohn CG, Fermann GJ, et al. External validation of prognostic rules for early post-pulmonary embolism mortality: assessment of a claims-based and three clinical-based approaches. Thromb J. 2016;14:14. PMID: 26977136
- Venetz C, Jiménez D, Mean M, et al. A comparison of the original and simplified Pulmonary Embolism Severity Index. Thromb Haemost. 2011;106(3):423–8. PMID: 21713328
- Konstantinides SV, Meyer G, Becattini C, et al. 2019 ESC Guidelines for the diagnosis and management of acute pulmonary embolism developed in collaboration with the European Respiratory Society (ERS). Eur Respir J. 2019;54(3):1901647. PMID: 31473594
