Two New Burn Severity Indices

Predicting survival in thermal injury: A systematic review of methodology of composite prediction models

“…Patient factors such as obesity, medical co-morbidities, and alcohol/ substance abuse have been identified in numerous cohort studies to predict poor prognosis [120][121][122][123]. This has not been borne out by inclusion of these variables in prediction models, either because of their limited additive significance when considered in context of populations with limited numbers of patients with similar conditions [93], or failure to incorporate them into the analysis [73]. Given the reducing mortality associated with burns, a study attempting to incorporate the effect of co-morbid conditions requires a very large number of patients and probably a multi-centre or national design to avoid over-and under-fitting.…”

“…Of the 45 studies 3 described prediction models derived from ''national'' data sets [101,103,104], 1 utilised data from 28 North American burn centres comprising the American Burn Association (ABA) Patient Registry [4], whereas 2 models (Roi's Total Burn and Full Thickness Burn models) described in one study were developed from data from 12 North American burn centres constituting the National Burn Information Exchange (NBIE) [73]. The remaining models were developed from data derived from either single (36) …”

“…Age of the subject population was stated in all but one study [73], total body surface area burn (TBSA) was stated in all but three studies [72][73][74], where as 15/45 studies did not state the incidence of inhalation injury. Raw mortality in the studies ranged from 1.2% [98] to 56.32% [90].…”

“…Missing data was declared in 9 studies, with only one study using multiple imputations to derive missing data [101]. In 2 large studies, missing data for inhalation injury resulted in exclusion of the variable from the analysis and final model [73,97].…”

Development and validation of a model for prediction of mortality in patients with acute burn injury

“…Finally, as the objective of the present study was to finetune the model developed by Ryan and colleagues 13 , the database did not include the data necessary for comparison with more complex prediction models developed by other investigators 9,10,14,19 . However, the findings suggested the limited advantage of formulas that are more complex.…”