Prognostic Scoring Systems for Burns: A Comparative Analysis of Their Predictive Accuracies for Mortality in Burn Patients
Summary: Published March 19, 2026 in the European Burn Journal (Vol. 7, No. 1; MDPI / European Burns Association), this retrospective cohort study from University Hospital Hamburg-Eppendorf and BG Klinikum Hamburg, led by Susanne Rein, Jule Schmiechen, Jochen Gille, and Thomas Kremer, compares the predictive accuracy for in-hospital mortality of five scoring systems applied to 644 adult burn patients treated at a single German burn center between September 2018 and May 2022. The five systems evaluated span different conceptual frameworks: burn-specific severity (ABSI and BABSI), perioperative physiological status (ASA classification), comorbidity burden (Charlson Comorbidity Index, CCI), and frailty (modified Frailty Index-5, mFI-5). The ABSI (Abbreviated Burn Severity Index), developed by Tobiasen et al. in 1982, scores age, sex, TBSA, full-thickness burn presence, and inhalation trauma. The BABSI (Bogenhausen ABSI) extends this by adding pre-existing conditions: cardiovascular, pulmonary, renal, gastrointestinal, and endocrinological diseases, plus substance use history. Patient cohort: 644 patients (441 male, 203 female); median age 44 years (range 18–93); 25 in-hospital deaths (3.9%); median TBSA 3.5%; 51.5% full-thickness burns; inhalation injury in 5.3%. All five scores significantly differentiated survivors from non-survivors (p < 0.001 for all). ROC curve analysis found excellent predictive accuracy for BABSI (AUC 0.963), ABSI (AUC 0.952), and ASA (AUC 0.916), with fair accuracy for CCI (AUC 0.851) and mFI-5 (AUC 0.760). Optimal cut-off values by Youden’s index: BABSI ≥ 8.25, ABSI ≥ 6.5, ASA ≥ 2.5, CCI ≥ 1.5, mFI-5 ≥ 1.5. Calibration (Hosmer-Lemeshow test): excellent for BABSI and ABSI; good for CCI and mFI-5; poor for ASA (which had the highest odds per category increase, OR 6.7, but poor alignment of predicted with actual outcomes). Logistic regression found no statistically significant independent association between mFI-5 and mortality, consistent with prior studies in burn populations. The authors recommend routine use of both BABSI and ABSI in daily burn clinical practice, while noting that comorbidity- and frailty-based scores offer complementary clinical context on patient vulnerability without replacing burn-specific prediction tools.
Key Highlights:
- ROC ranking: BABSI (AUC 0.963) > ABSI (AUC 0.952) > ASA (AUC 0.916) > CCI (AUC 0.851) > mFI-5 (AUC 0.760); all five significantly discriminated survivors from non-survivors, but burn-specific scores (BABSI and ABSI) outperformed comorbidity/frailty scores
- BABSI advantage: by incorporating pre-existing comorbidities (cardiovascular, pulmonary, renal, GI, endocrine, substance use) on top of ABSI’s burn-specific parameters, BABSI marginally outperformed ABSI in both discrimination and calibration — the authors recommend both be routinely applied
- ASA paradox: highest odds ratio per category (OR 6.7), suggesting each ASA grade increase confers a nearly 7-fold increase in mortality odds in burn patients — but poor Hosmer-Lemeshow calibration (Chi-square 81.1, p < 0.001) means it overestimates or misaligns predicted versus actual outcomes; useful for risk flagging but not reliable for probabilistic mortality estimation
- mFI-5 limitations: not a statistically significant independent predictor of burn mortality in multivariate analysis; while it captures frailty burden, it lacks the burn-specific parameters (TBSA, burn degree, inhalation injury) that dominate mortality risk in this population
- Clinical recommendation: implement both BABSI and ABSI routinely in burn centre daily practice; use CCI and mFI-5 as supplementary tools for contextualising comorbidity burden and frailty, rather than as primary mortality predictors
- Study limitations: single-centre retrospective design at one German burn center; relatively low mortality rate (3.9%, n=25 deaths); limited generalisability across health systems and care standards; in-hospital mortality only (no long-term functional outcomes); Hosmer-Lemeshow calibration is sensitive to small sample sizes, warranting cautious interpretation
Keywords: burn severity scoring mortality, ABSI burn prognosis, BABSI burn comorbidity, burn center outcomes prediction, Charlson comorbidity index burns, burn frailty index mortality
Susanne Rein, Jule Schmiechen, Jochen Gille, Thomas Kremer