Burn mortality. Experience at a regional burn unit. Literature review.

The burn patient population at a regional burn unit in Central New York State, from 1974 through 1980, (N = 507) is described completely in terms of age, total burn size, full thickness burn size, sex, race, etiology of burn, inhalation injury, referral pattern, distance from burn unit, interval to admission, length of hospitalization and causes of death. The interrelationship of these factors is examined to define their contribution to injury severity and identify variables useful in predicting death. Demographic profiles are compared to the population base, and the case-mix of patients is contrasted with that of other burn units. A logistic model to predict death is developed using the variables of full-thickness burn size, age, age 2, and the presence of inhalation injury. The power of statistical methods to predict death is discussed in relation to the 37 patients who died with emphasis on the outliers present in unscreened patient populations. The point is made that any clinically useful index of burn severity must include a factor reflecting the physiologic stress imposed on the individual patient. Clinical issues pertaining to the treatment of nonsurvivors are discussed in relation to the physician's responsibility, euthanasia, and resource consumption. Results of treatment are difficult to evaluate because of the variables of patient selection, injury severity, time, and the absence of a satisfactory measure of morbidity. Effective prevention is the only way to eliminate the deaths of victims who do not survive to enter the health care system and the morbidity that inevitably results from tissue loss. From the standpoint of public health, burn morbidity may have an economic impact far different from that of burn mortality. Supraregional burn units with access limited to seriously but not hopelessly burned individuals are proposed. Supraregional burn units would make the delivery of burn care more efficient and allow the selection of a patient sample homogeneous in terms of severity, thus increasing the reliability of treatment evaluations.