Increased Risk of Sudden Death Among Youths and Stimulant Use: Determining Alternate Potential Factors

To the Editor: Dr. Gould et al. (1) retrospectively determined the stimulant use rate in children/adolescents who died suddenly without explanation and in a matched comparison group who died as passengers in motor vehicle accidents. They identified stimulant use in 1.8% of the group that died suddenly without explanation and in 0.4% of the motor vehicle accident group. They concluded that this difference was the result of an increased risk of sudden death associated with stimulant use. In order to determine the validity of this conclusion, however, additional questions need to be considered.

Although Dr. Gould et al. stated that these rates of stimulant use were comparable to the general population, this was true only for the group that died suddenly without explanation. Olfson et al. (2) reported that the general child/adolescent population rate of stimulant use during the time period assessed in the study ranged from 0.6% in 1986 to 2.4% in 1997. The 1.8% rate in the group that died suddenly without explanation is approximately what would have been found in a random population sample during this 10-year period. If stimulants significantly increase the risk of sudden death, wouldn't we expect the group that died suddenly without explanation to have had a higher than average rate of stimulant use?

Moreover, why was the detected rate of stimulant use in the comparison group (0.4%) so much lower than the population average? Could stimulant use have been underdetected in the motor vehicle accident group? If stimulant use was underdetected in this study, is it possible to know whether the two groups were subject to the same degree of underdetection?

The authors reported collecting stimulant use data from three sources (family surveys, toxicology reports, and medical examiner reports), but they acknowledged that all three sources of data were available for only 10% of the paired comparisons. Most of the comparisons seem to have been based on only a single data source. Is it possible that reliance on a single data source could have contributed to underdetection or even biased detection? The medical examiner reports appear to have been the primary basis for more than one-half of the comparisons. Is it realistic to expect that medical examiners would have been as likely to collect accurate information on psychotropic medication use in the motor vehicle accident victims as they would for children who died without any apparent cause?

Data from family surveys were used in 44% of the comparisons, but the limited detail provided about how surveys were conducted makes it difficult to assess the quality of these data. Were these direct interviews, or were families simply sent questionnaires? Were parents asked specifically about stimulant use? Finally, 10 to 12 years after the incident, is it realistic to expect that a parent whose child died in a motor vehicle accident would recall his or her child's medications as accurately as a parent whose child died suddenly with no explanation?