Clinical background
The Killip classification meets a specific need: to grade the severity of heart failure at the bedside in a patient with acute myocardial infarction, without laboratory values or imaging. The degree of heart failure at presentation is one of the strongest single prognostic factors in myocardial infarction, and it determines not only prognosis but also the choice of treatment strategy, the level of monitoring required and the timing of invasive intervention. Without a structured classification, clinical assessments of heart failure tend to be inconsistent, particularly in the acute situation with several simultaneous physiological disturbances.
The classification is moreover a component of several composite risk scores, among them TIMI and GRACE, in which it contributes the clinical haemodynamic assessment that complements ECG and biomarker data [3].
Applying the Killip classification
The Killip classification is a four-level ordinal scale based solely on clinical examination:
| Class | Clinical findings |
|---|---|
| I | No signs of heart failure |
| II | Crackles, S₃ or a raised JVP (jugular venous distension) |
| III | Frank pulmonary oedema |
| IV | Cardiogenic shock |
The classification is made at the first clinical assessment, before reperfusion therapy. No score is summed; the class itself is the measure. The derivation cohort consisted of 250 patients with suspected acute myocardial infarction admitted to a coronary care unit at an American university hospital in 1967 [1]. It was an observational study without systematic data collection, without adjustment for confounders and without independent validation. Mortality in the original cohort was: class I 6%, class II 17%, class III 38% and class IV 81% [1]. These figures concern in-hospital mortality in a pre-reperfusion cohort in which neither thrombolysis nor primary PCI was available, and in which beta blockers, ACE inhibitors and statins were not part of standard treatment.
Interpretation in practice
The Killip classification leads to concrete clinical actions:
| Class | Clinical meaning | Management |
|---|---|---|
| I | No heart failure. Low risk in the original cohort. | Standard monitoring. Reperfusion as indicated. |
| II | Mild to moderate heart failure. Increased risk. | Monitoring with haemodynamic observation. Oxygen where needed. Diuretics for fluid retention. Early invasive strategy. |
| III | Frank pulmonary oedema. High risk. | Intensive care. Active treatment of the pulmonary oedema. Early invasive strategy, often primary PCI in STEMI. |
| IV | Cardiogenic shock. Very high risk. | Intensive care with haemodynamic monitoring. Mechanical circulatory support may become necessary. Emergency coronary angiography and revascularisation according to current shock guidelines. |
Class IV is defined in the original description as cardiogenic shock with arterial hypotension (a systolic blood pressure below 90 mmHg) and signs of peripheral vasoconstriction (oliguria, cyanosis, diaphoresis) [1, 2]. It is important to distinguish class IV from transient hypotension without signs of shock, since management differs fundamentally.
Validation and performance
The most comprehensive modern external validation is the Brazilian cohort from the Dante Pazzanese Institute of Cardiology, with 1,906 patients enrolled between 1995 and 2011 and followed for a mean of five years [2]. Of these, 64% had STEMI and 36% NSTEMI. The distribution across Killip classes was: I 84.3%, II 11.0%, III 2.4% and IV 2.3%. Total mortality during follow-up was: class I 17.7%, class II 27.3%, class III 30.4% and class IV 48.8% (p < 0.0001) [2]. Mortality is thus considerably lower than in the original cohort, particularly for class IV (48.8% versus 81%), but the relative gradient between classes persists.
In Cox regression models, the Killip classification was an independent prognostic factor for mortality after adjustment for age, sex, cardiovascular risk factors, haemodynamic parameters and angiographic findings, both for STEMI (Wald χ² 11.9, p = 0.008) and for NSTEMI (Wald χ² 16.5, p = 0.001) [2]. This was the first validation of the classification in NSTEMI patients in the reperfusion era.
The role of the Killip classification as a component of the GRACE registry is well established. In the GRACE study, with 43,810 patients from 94 hospitals in 14 countries (1999 to 2005), Killip class was one of nine independent predictors of death and of the composite of death or myocardial infarction from admission to six months after discharge [3]. The simplified GRACE model achieved a c-statistic of 0.81 for death and 0.73 for death or myocardial infarction [3]. The Killip classification therefore contributes not as a standalone instrument with its own c-statistic, but as one of several variables in a composite model.
A small but relevant study from Lithuania, with 288 STEMI patients treated with primary PCI between 2018 and 2021, found that in-hospital mortality rose from 1.0% in class I to 69.6% in class IV [4]. In multivariable analysis, class IV was a strong independent predictor (OR 60.94, 95% CI 15.98 to 232.46), while the lower classes showed limited prognostic separation [4]. This suggests that the strength of the classification in modern practice lies chiefly in identifying patients in cardiogenic shock, rather than in separating class I from class II.
Limitations
The mortality figures from the original cohort (6%, 17%, 38% and 81% respectively) must not be applied unchanged to today's patients. Reperfusion therapy, beta blockers, ACE inhibitors, statins and modern intensive care have lowered mortality in all classes, and in class IV in particular mortality has almost halved in modern cohorts [2]. The relative gradient nonetheless persists: a higher Killip class carries a higher risk.
The classification is subjective and rests on clinical auscultation and inspection. Inter-observer variability has not been studied systematically in larger validation studies, but it is an obvious weakness, particularly at the boundary between class I and class II, where the findings may be subtle. A single basal crackle or an uncertain S₃ can move a patient from class I to class II with consequences for risk assessment.
The classification applies to acute myocardial infarction. It is not validated for heart failure of other causes, and it does not capture patients with subclinical heart failure in whom auscultation is normal but biomarkers or echocardiography show dysfunction. In modern practice the Killip classification is therefore routinely complemented with NT-proBNP and echocardiography.
The study from Lithuania showed that classes I and II in a modern PCI cohort have very low and almost identical mortality (1.0% and 3.0% respectively), which calls into question the ability of the classification to separate low-risk patients in modern practice [4]. Its clinical value lies primarily at classes III and IV, where mortality is many times higher and management changes immediately.
References
- Killip T, Kimball JT. Treatment of myocardial infarction in a coronary care unit. A two year experience with 250 patients. Am J Cardiol. 1967;20(4):457–64. PMID: 6059183
- Mello BH, Oliveira GB, Ramos RF et al. Validation of the Killip-Kimball classification and late mortality after acute myocardial infarction. Arq Bras Cardiol. 2014;103(2):242–35. PMID: 25014060
- Fox KA, Dabbous OH, Goldberg RJ et al. Prediction of risk of death and myocardial infarction in the six months after presentation with acute coronary syndrome: prospective multinational observational study (GRACE). BMJ. 2006;333(7578):1374–5. PMID: 17032691
- Speckauskiene V, Meilutyte-Lukauskiene D, Cerapaite-Trusinskiene R et al. Prognostic utility of Killip classification in acute myocardial infarction: a retrospective cohort analysis in contemporary clinical practice. Eur Heart J Open. 2026;6(2):oeag050. PMID: 41958748
