Dr. Gould Replies

To the Editor: The thoughtful letters of Drs. Blume, Rostain, Nadkarni, Smith, and Johnson underscore the importance of examining potential confounding factors other than stimulants that may have been responsible for our observed association between sudden unexplained death and stimulant medications.

Drs. Blume and Rostain cite evidence that stimulants may lower the risk of motor vehicle accidents among adolescents, thereby introducing a bias toward lower stimulant use in our comparison group. However, because we selected a comparison group of individuals who died as passengers in a motor vehicle traffic accident involving another motor vehicle, the benefit ascribed to stimulants would not affect our findings. Passenger comparison subjects were specifically selected because they have been found not to be at greater risk for hyperactivity and other deficits in vigilance, attention, and impulse control (1).

Dr. Johnson expresses concern that stimulant use in our case subjects (1.8%) resembles population rates during the study period and that the rate of stimulant use in our comparison subjects (0.4%) was much lower than population rates. He references rates of 0.6% in 1986 and 2.4% in 1997, from a report by Olfson and colleagues (2). This report, however, indicated that 0.7% of youths (age 3–18 years) in 1987 received at least one stimulant prescription for ADHD and 2.9% of youths (age 3–18 years) in 1997 received at least one stimulant prescription for ADHD. Because of the intermittent nature of stimulant therapy in the community (3), these 1-year treatment rates would be expected to be considerably higher than the point prevalence rates that were the focus of our study. The younger age of the sample reported by Olfson et al. (3–18 years old) would also be expected to yield substantially higher rates of stimulant use (4) than our older study group, 90% of whom were between the ages of 12 and 19 years.

Dr. Johnson further suggests that medical examiner reports may have biased detection. We cannot exclude the possibility that, compared with a passenger motor vehicle fatality, an "unexplained” death may have prompted medical personnel to ask more questions about medications at the time of death. However, the primary analysis remained significant (odds ratio=7.3, 95% confidence interval [CI]=1.4–74.8, p=0.015) following exclusion of the one case subject whose methylphenidate exposure was detected solely by medical examiner report.

Dr. Johnson's last concern focuses on the quality of our surveys. A comparable survey was sent to parents of both groups of children. As noted in our article, the survey included items assessing past medical history, medications taken at the time of death, a list of medical problems, and the use of over-the-counter and prescription medications. A history of sudden death among relatives was also assessed. The items assessing medication use were open-ended and did not ask about any one specific class of medication, including stimulants. The reasons for taking the medication(s), and the frequency and duration of use, were also asked. While we cannot rule out the possibility that parents of children in the sudden unexplained death group remembered stimulant medications more vividly than parents of children who died in accidents, we remain confident that the association of sudden death and stimulant medication use is valid because our sensitivity analyses suggest that the strength of the association was not sensitive to the source of stimulant measurement.

Drs. Nadkarni and Smith raise the possibility that group imbalances in age and year of death might have introduced bias toward lower rates of stimulant use in our comparison group. Yet our case (mean age: 15.76 years old) and comparison (mean age: 15.83 years old) subjects had nearly identical mean ages, and the group distributions did not differ in year of death. Drs. Nadkarni and Smith also hypothesize that unmeasured covariates, such as season of death, income, and rural places of residence, may have not been comparable among the case and comparison subjects and that these covariates may affect the likelihood of receiving a stimulant. We do not have information readily available to address these speculations, but we agree with their expectation that such factors might attenuate, but not likely eliminate, the relationship.

We made rigorous efforts to enhance the comparability of case and comparison subjects that could threaten the validity of our study's findings. Significant associations between sudden death and stimulant medication use in youths call attention to potential risks of stimulant medication and hopefully will increase clinical attention and motivate further study. At the same time, we appreciate the benefits of stimulant medication for youths with ADHD and share concerns that physicians will be discouraged from prescribing stimulants or parents will discontinue their child's medication on their own based on our study's findings.