Recrudescence of Malignant Hyperthermia

RECRUDESCENCE of a malignant hyperthermia (MH) reaction has previously been reported anecdotally, but in this month’s issue of ANESTHESIOLOGY, Burkman et al. present the first detailed evaluation of its incidence and associated factors. Appreciation of the potential for recrudescence of MH reactions is important for the anesthesiologist managing a suspected case because it informs decisions regarding the level of care required after treatment of the acute event. If we accept the data provided by Burkman et al. to be reliable, the 20% incidence of recrudescence supports current advice that patients should be monitored for at least 24 h in a critical care environment after resolution of the signs of the acute MH reaction. Furthermore, the presence of certain characteristics might suggest an even longer period of close observation. Because the study of Burkman et al. may impact on postreaction management of possible MH cases, it is appropriate to examine how reliable the data are and how applicable they are to practice outside North America. The study was a retrospective cohort study, and the authors discuss its limitations. The most obvious of these is the subjectivity of the key diagnoses, first the diagnosis of the initial episode as MH and second that of recrudescence. The clinical diagnosis of MH is difficult because there is no single pathognomonic feature and there are multiple differential diagnoses. Without a definitive laboratory diagnosis, the cohort of Burkman et al. will include some non-MH cases, which may have reduced the apparent incidence of recrudescence. On the other hand, the diagnosis of recrudescence is subject to the same vagaries with the additional differential of an inflammatory response in a critically ill patient. It may well be that these presumed overdiagnoses on both numerator and denominator of the recrudescence incidence ratio have a net neutral effect. In some senses though, this is academic. Until we have a point-of-care diagnostic test for MH susceptibility, management decisions must be based on the presumed diagnosis of MH and, indeed, recrudescence. By restricting their analyses to cases deemed retrospectively to be “likely MH” on the basis of the MH clinical grading score, Burkman et al. acknowledge the introduction of selection bias. Inclusion of only 308 of 528 reports to the North American MH Registry may have resulted in underestimation of the true incidence of recrudescence. Even if there were no further cases from the 220 reports not included, however, the 95% confidence interval would be approximately 10–15%. Such an incidence, considering the consequences of not identifying recrudescence and managing it appropriately, would still warrant high dependency care of all patients after an MH reaction. Having estimated the incidence of recrudescence, Burkman et al. next sought to determine which, if any, clinical variables were associated with its development. Considerable caution must be exercised in the interpretation of such data, and the authors of this article are right to stress the difference between association and causation. Significant univariate associations with the occurrence of recrudescence were found for muscular (vs. normal/lean) body habitus, increasing duration of the interval between induction of anesthesia and onset of the MH reaction, and the development of a temperature increase (inappropriate rate of temperature rise or temperature 38.8°C) as part of the reaction. Examination of the data indicates, however, that the predictive value for recrudescence will be low for each of the identified variables. A multivariate logistic regression model for prediction of recrudescence using the same variables was also generated. Such analyses can moderate the influence of interdependent covariates, but in doing so, they may exclude the real predictor at the expense of a confounding factor. The model-generating capabilities of the statistical software packages, furthermore, are so powerful at fitting the data that a statistically significant model can be produced even though the contribution to the outcome variable of the selected predictor variables is small. This can be examined using the r value for the model, which estimates the proportion of the variance of the outcome variable that the model explains. The r value of 0.102 for the model of Burkman et al. suggests that the identified variables are unlikely to be useful predictors of recrudescence. I do not think we should ignore the associations identified by Burkman et al. altogether, however, because This Editorial View accompanies the following article: Burkman JM, Posner KL, Domino KB: Analysis of the clinical variables associated with recrudescence after malignant hyperthermia reactions. ANESTHESIOLOGY 2007; 106:901–6.