Correspondence

Correction

Estimates of the Probability of Death from Burn Injuries

N Engl J Med 1998; 338:1848-1850June 18, 1998

Article

To the Editor:

The article by Ryan et al. (Feb. 5 issue)1 concerning estimates of the probability of death from burn injuries raises three issues. Assessing the probability of death is useful when mortality is a relevant end point. This is not the case in patients with burns covering 20 percent of body-surface area or less and no associated problems, such as inhalation injury. In this study, such patients constituted 77.5 percent of the total population of 1665 patients, and the mortality rate in this subgroup was 0.46 percent. Including these very-low-risk patients adds no useful information. The model should be restricted to patients with a greater risk of this end point, which could change the weight given to the variables. Furthermore, there were only 22 patients with a high probability of death (87 percent), with a lower boundary of the 95 percent confidence interval of 78 percent. This can be interpreted as equivalent to a probability of error of 22 percent if one assumes that these patients will die. We think that this number is too small to allow reliable estimates of mortality.

A model that gives individual estimates within a continuous range should be preferred to one that gives only four risk strata (0.3 percent, 3 percent, 33 percent, and 87 percent). Indeed, all tests used to assess the severity of illness in patients in the intensive care unit (the Acute Physiologic and Chronic Health Evaluation III and the Simplified Acute Physiologic Score II) provide probabilities within a continuous range. This allows one to evaluate models by calculating their discriminant power (the area under the receiver-operating-characteristic curve) and their calibration (goodness-of-fit test).2 Because these data are lacking in the study by Ryan et al., their model cannot be compared with other models proposed for use in patients with burn injuries, such as the one described by Smith et al.,3 which relies on the same variables and provides individual estimates of the probability of death within a continuous range.

Jean Edouard Rohan, M.D.
Cristina Esteban, M.D.
Philippe Loirat, M.D.
Hôpital Foch, 92151 Suresnes, France

3 References

1. 1

   Ryan CM, Schoenfeld DA, Thorpe WP, Sheridan RL, Cassem EH, Tompkins RG. Objective estimates of the probability of death from burn injuries. N Engl J Med 1998;338:362-366
   Full Text | Web of Science | Medline

2. 2

   Hosmer DW Jr, Lemeshow S. Applied logistic regression. New York: John Wiley, 1989.
   

3. 3

   Smith DL, Cairns BA, Ramadan F, et al. Effect of inhalation injury, burn size and age on mortality: a study of 1447 consecutive burn patients. J Trauma 1994;37:655-659
   CrossRef | Web of Science | Medline

To the Editor:

Ryan et al. did not discuss an important finding in their study: the 12-fold increase in the percentage of patients with do-not-resuscitate orders between the period from 1975 to 1984 and the period from 1990 to 1994. Could the authors discuss the reasons for this large increase and address the question of futility in the burn unit? The data appear to suggest that if a person is over the age of 41 years, has a burn size in excess of 61 percent of body-surface area, and has sustained an inhalation injury, then in all probability that person will die during hospitalization. What moral obligation is there to continue to treat if this prognostication is accurate?

Thomas G. Hooyman, Ph.D.
Catholic Health Initiatives, Denver, CO 80202

Author/Editor Response

The authors reply:

To the Editor: The system for objectively estimating the probability of death from burn injuries is designed to help health care providers furnish initial information on mortality to patients and families, assist medical personnel with triage at injury scenes, and help researchers plan inclusion criteria for clinical trials in which mortality is a relevant end point. For these purposes, the method should be easy to use and applicable to all patients with burn injuries who require hospital treatment. Our goal was to develop a method based on a limited set of clinically apparent risk factors that could easily be determined. This method is not meant to replace severity scores that use continuous variables, such as the Acute Physiologic and Chronic Health Evaluation III index, or burn formulas1,2 that were developed for other purposes and have been found to be impractical in these settings.

Our study identified a population (patients over 60 years of age with a burn size of more than 40 percent of body-surface area and inhalation injury) with a high risk of death (87 percent), as noted by Dr. Hooyman. There were only a small number of patients in this group, resulting in the lower boundary of the 95 percent confidence interval (78 percent) noted by Dr. Rohan and colleagues. We think that this level of precision is adequate for the above-mentioned purposes, especially given the complex nature of decisions about whether to provide resuscitation in situations in which the risk of death is high.

The increased incidence of patients with burn injuries who had do-not-resuscitate orders between the period from 1975 to 1984 and the period from 1990 to 1994 coincides with the growing emphasis on patient autonomy. Documentation of advance directives became mandatory for Medicare reimbursement in 1991.

Medical futility remains undefined. Patients and surrogates can refuse life-prolonging treatment regardless of the prognosis. Only when doctors judge that treatment is futile and oppose requests for treatment does the issue of futility become relevant. A doctor's refusal to accede to requests for treatment on the basis of the presence of the three risk factors from our formula, against the family's wishes, represents an overvaluation of the certitude provided by the formula.

Table 1Table 1Mortality among Patients According to Age, Burn Size, and Presence or Absence of Inhalation Injury. of our article contained some errors. We have provided a corrected version, and we apologize for any confusion that may have resulted from the errors.

Colleen M. Ryan, M.D.
David A. Schoenfeld, Ph.D.
Edwin H. Cassem, M.D.
Ronald G. Tompkins, M.D.
Harvard Medical School, Boston, MA 02114

2 References

1. 1

   Smith DL, Cairns BA, Ramadan F, et al. Effect of inhalation injury, burn size, and age on mortality: a study of 1447 consecutive burn patients. J Trauma 1994;37:655-659
   CrossRef | Web of Science | Medline

2. 2

   Tompkins RG, Burke JF, Schoenfeld DA, et al. Prompt eschar excision: a treatment system contributing to reduced burn mortality: a statistical evaluation of burn care at the Massachusetts General Hospital (1974-1984). Ann Surg 1986;204:272-281
   CrossRef | Web of Science | Medline

Citing Articles (6)

Citing Articles

1. 1

   P SHARMA, R BANG, A ALFADHLI, P SHARMA, S BANG, I GHONEIM. (2006) Paediatric burns in Kuwait: Incidence, causes and mortality. Burns 32:1, 104-111
   CrossRef

2. 2

   Nele Brusselaers, Eric A. J. Hoste, Stan Monstrey, Kirsten E. Colpaert, Jan J. De Waele, Koenraad H. Vandewoude, Stijn I. Blot. (2005) Outcome and changes over time in survival following severe burns from 1985 to 2004. Intensive Care Medicine 31:12, 1648-1653
   CrossRef

3. 3

   Prem N. Sharma, Rameshwar L. Bang, Ibrahim E. Ghoneim, Sarla Bang, Promila Sharma, Mohammed K. Ebrahim. (2005) Predicting factors influencing the fatal outcome of burns in Kuwait. Burns 31:2, 188-192
   CrossRef

4. 4

   H Carsin. (2001) Letter to the editor. Burns 27:4, 418-419
   CrossRef

5. 5

   Colleen M. Ryan, Robert L. Sheridan. (2000) Should seriously burned children who suffer cardiac arrest be subjected to cardiopulmonary resuscitation?. Critical Care Medicine 28:2, 592-593
   CrossRef

6. 6

   Manon Choinière, Marc Dumont, Jacques Papillon, Dominique R Garrel. (1999) Prediction of death in patients with burns. The Lancet 353:9171, 2211-2212
   CrossRef