What is the POMPE-C tool?

The POMPE-C tool (Pulmonary embolism Outcome with Malignancy Prognosis Estimator for Cancer) is a bedside logistic regression model that estimates 30-day all-cause mortality in adults with active cancer and a diagnosis of acute pulmonary embolism (PE). It was developed by Kline and colleagues to address a gap in PE risk stratification: general PE prognostic scores (such as PESI and sPESI) were not calibrated specifically for oncology patients, who often have higher baseline mortality, different comorbidity profiles, and distinct goals of care.

POMPE-C outputs a continuous estimated mortality probability (0–100%) from eight clinical predictors available at the time of evaluation. The score supports disposition decisions (outpatient versus inpatient management), intensity of monitoring, and structured goals-of-care discussions, especially at the high-mortality end of the spectrum. It is a prognostic aid, not a substitute for hemodynamic assessment, imaging of right ventricular strain, bleeding-risk evaluation, or guideline-based anticoagulation.

Clinical context: pulmonary embolism in patients with cancer

Cancer is an independent risk factor for venous thromboembolism. Patients with active malignancy have substantially higher rates of PE than the general population, driven by tumor procoagulant effects, chemotherapy, surgery, immobility, and indwelling catheters. When PE occurs in this group, short-term mortality reflects not only embolic burden and cardiopulmonary reserve but also cancer stage, treatment intent, infection, and pre-existing functional status.

Standard PE severity indices may underestimate or misclassify risk in oncology patients because they emphasize age, comorbidities, and vital signs without fully incorporating cancer-specific prognosis and resuscitation preferences. POMPE-C was derived in a cohort of patients with active cancer (not remote history alone) after non-incidental PE diagnosis, with the explicit goal of estimating 30-day mortality to inform management location and intensity.

PE management in cancer still requires concurrent assessment for massive or submassive PE (hypotension, shock, severe hypoxemia, right ventricular dysfunction), contraindications to anticoagulation, and whether thrombolysis or embolectomy is indicated. POMPE-C does not replace those acute pathways.

Study population and appropriate use

Apply POMPE-C only when all of the following are true:

• Active cancer at the time of presentation (not merely a remote cancer history without active disease).
• Acute pulmonary embolism confirmed by imaging or equivalent diagnostic standard (non-incidental finding).
• Clinical evaluation at presentation or early in the admission, using variables from the prior 6 hours where specified.

Do not use POMPE-C in patients without active cancer, for incidental subsegmental PE discovered on staging scans without acute clinical syndrome, or as the sole basis to withhold resuscitation in unstable patients. Do not delay treatment of high-risk PE while calculating the score.

Model structure: logistic regression

POMPE-C uses a multivariable logistic model. First compute the logit (linear predictor), then transform to a probability:

Estimated 30-day mortality (%) = 100 × (1 − 1 / (1 + exp(logit)))

logit = 3.718

• + 1.552 × do not resuscitate (1 if yes, 0 if no)
• + 0.800 × respiratory distress (1 if yes, 0 if no)
• + 0.734 × unilateral limb swelling (1 if yes, 0 if no)
• + 1.473 × altered mental status (1 if yes, 0 if no)
• + 1.028 × heart rate ≥100/min in past 6 hours (1 if yes, 0 if no)
• + 0.044 × highest respiratory rate in past 6 hours (breaths/min, continuous)
• − 0.063 × room-air oxygen saturation (%, continuous)
• − 0.012 × body weight (lb, continuous)

The original derivation recorded weight in pounds. This CalcMD calculator accepts weight in kilograms and converts to pounds internally (1 kg = 2.2046226218 lb) before applying the coefficient. Heart rate enters the model as a dichotomous variable (any documented rate ≥100/min in the past 6 hours), not as a continuous maximum heart rate.

Variable definitions

Predictor	Definition	Coding in model
Do not resuscitate (DNR)	New or existing verbal or written desire not to be resuscitated from death	Binary: 1 = yes, 0 = no
Respiratory distress	Obvious anxiety from dyspnea or increased work of breathing on examination	Binary: 1 = yes, 0 = no
Unilateral limb swelling	New, noticeable swelling of a leg or arm causing visible asymmetry compared with the contralateral limb	Binary: 1 = yes, 0 = no
Altered mental status	New or different from baseline: disorientation, delirium, stupor, or coma	Binary: 1 = yes, 0 = no
Heart rate	Documented heart rate ≥100/min at any time in the past 6 hours	Binary: 1 = yes, 0 = no
Respiratory rate	Highest respiratory rate (breaths/min) recorded in the past 6 hours	Continuous (per breath/min)
Oxygen saturation (SpO2)	Most recent room-air pulse oximetry	Continuous (per 1% on room air)
Body weight	Measured or patient-estimated weight (entered in kg, converted to lb for the equation)	Continuous (per pound)

Clinical rationale for selected predictors

• DNR status captures explicit goals of care and correlates strongly with short-term mortality independent of embolic anatomy.
• Respiratory distress and tachypnea reflect cardiopulmonary compromise from embolic load, underlying lung disease, or deconditioning.
• Altered mental status may indicate hypoperfusion, hypoxemia, metastatic disease, infection, or other acute encephalopathy; it was among the strongest binary predictors in the model.
• Unilateral limb swelling supports clinically suspected deep vein thrombosis in the cancer PE population studied.
• Tachycardia (≥100/min) is a simple marker of physiologic stress in the prior 6-hour window.
• Higher SpO₂ and greater weight are protective in the logistic equation (negative coefficients), consistent with better reserve and lower observed mortality in the derivation cohort.

Risk stratification and suggested management framing

The CalcMD implementation assigns three risk bands based on the estimated 30-day mortality percentage, aligned with thresholds reported in the original validation analyses:

Estimated 30-day mortality	Risk stratum	Observed outcomes (validation framing)	Management-oriented summary
≤ 5%	Low	No deaths in the ≤5% group in derivation/validation reporting (0/50; posterior probability 0%, 95% CI 0–7%)	Overall mortality risk is low. Outpatient management and anticoagulation may be considered when aligned with guidelines, bleeding risk, social support, and full clinical assessment.
> 5% and ≤ 50%	Intermediate	Not a low-risk group; many patients survived beyond 30 days, but mortality was materially higher than the ≤5% stratum (e.g. 106/156 survivors had estimates >5% in reported cohorts)	Favor hospital admission and standard inpatient PE management. Reassess hemodynamics, bleeding risk, oncology context, and need for advanced therapies.
> 50%	High	Very high observed mortality in validation (posterior probability approximately 73%, 95% CI 46–95%)	Favor admission and intensive monitoring as indicated. Goals-of-care and resuscitation preferences warrant explicit discussion. Do not delay resuscitation for risk stratification.

These thresholds are cohort-derived cut points for communication and disposition planning. They should not be applied as rigid rules in every individual, particularly when clinical instability, massive PE, or patient preference for aggressive care is present.

How POMPE-C differs from general PE severity scores

PESI and sPESI stratify mortality in broad PE populations using age, comorbidities, vital signs, and imaging findings. They are widely used for outpatient versus inpatient decisions in non-cancer patients. POMPE-C was built specifically for active cancer and incorporates DNR status and cancer-relevant bedside findings (respiratory distress pattern, limb swelling) in a single mortality-calibrated logistic model.

POMPE-C does not include right ventricular dysfunction on echocardiology, troponin, BNP, or CT-defined clot burden. Patients with intermediate POMPE-C estimates may still have submassive PE requiring closer monitoring or escalation per guideline algorithms. Conversely, a low POMPE-C estimate does not exclude hemodynamic compromise if vitals are unstable.

Anticoagulation, disposition, and oncology coordination

Anticoagulation remains cornerstone therapy for most acute PE in cancer when bleeding risk is acceptable. Choice of agent (low-molecular-weight heparin, direct oral anticoagulants where appropriate, or warfarin in selected scenarios) depends on tumor type, drug interactions, thrombocytopenia, renal function, and recent procedures. POMPE-C informs where and how intensively to manage the patient; it does not replace anticoagulation decisions.

For low estimated mortality (≤5%), shared decision-making may support outpatient anticoagulation when social support, follow-up, and oncology access are reliable and no high-risk features mandate admission. For intermediate and high estimates, inpatient management allows monitoring for clinical deterioration, management of hypoxemia, and coordination with oncology for treatment-modifying discussions when prognosis is poor.

Early involvement of oncology, palliative care, and thrombosis specialists is valuable when DNR status is present, mental status is altered, or estimated mortality exceeds 50%, so that PE treatment aligns with overall cancer care goals.

Acute high-risk PE: when POMPE-C is insufficient

Regardless of POMPE-C output, evaluate for features requiring immediate escalation:

• Hypotension or shock suggesting massive PE
• Severe hypoxemia requiring high-flow oxygen or mechanical ventilation
• Right ventricular dysfunction or elevated biomarkers suggesting submassive PE at risk of decompensation
• Contraindication to anticoagulation requiring alternative strategies

Thrombolysis, catheter-directed therapy, surgical embolectomy, and vena caval filters are considered by guideline pathways independent of POMPE-C mortality estimates. Risk stratification must never delay resuscitation or definitive treatment in unstable patients.

Important limitations and practice caveats

• Cancer-specific: Validated in active cancer; do not extrapolate to PE without malignancy or with only historical cancer.
• All-cause mortality: The endpoint is 30-day death from any cause, not PE-specific mortality; advanced cancer progression may dominate the outcome.
• Retrospective derivation: Larger prospective validation is needed before using the score alone to mandate outpatient care.
• Temporal window: Vital signs and heart rate use the past 6 hours; inconsistent documentation may misclassify patients.
• Room-air SpO₂: Supplemental oxygen changes interpretation; the model used room-air oximetry.
• DNR as predictor: DNR is both a prognostic variable and an expression of patient values; it should be documented accurately and revisited, not inferred.
• Does not quantify bleeding risk on anticoagulation or need for inferior vena cava filter.
• Modern therapy era: Outcomes may differ with contemporary oncology immunotherapies, ambulatory anticoagulation pathways, and improved supportive care compared with the 2012-era cohort.

Documentation and workflow tips

When applying POMPE-C, document each input explicitly (DNR status, presence of respiratory distress, limb swelling, mental status, whether HR ≥100/min occurred in the past 6 hours, peak respiratory rate, room-air SpO₂, weight source), the calculated logit if reviewing the breakdown, the estimated 30-day mortality percentage, and the assigned risk stratum. Record whether disposition and goals-of-care discussions incorporated the estimate alongside guideline-based PE assessment. This supports handoffs between emergency medicine, hospital medicine, oncology, and palliative care.

Using this CalcMD calculator

Enter yes or no for each binary criterion and numeric values for highest respiratory rate (breaths/min), body weight (kg), and room-air oxygen saturation (%). The tool converts weight to pounds, computes the logit and estimated 30-day mortality (%), displays a term-by-term logit breakdown, and assigns low (≤5%), intermediate (>5% to ≤50%), or high (>50%) risk with management-oriented messaging. Use the output for education, structured documentation, and shared decision-making alongside clinical judgment and institutional PE pathways.