LinkExchange Member

Free Home Pages at GeoCities

Article #74

The authors tested reaction time by measuring the time delay in response (pressing a button) to a visual stimulus (a bright red light) which was turned on at variable intervals from 2 to 10 seconds during a 10-minute task. Each test involved 80-90 stimuli and responses. reaction time to a visual stimulus.      They compared performance on this test in three groups of people: (1) 113 with mild to moderate sleep-disordered breathing; and (2) 80 healthy volunteers given a drink of vodka and orange juice sufficient to raise their blood alcohol levels to an average of 0.08 g/dL (exceeds the 0.04 g/dL level which makes it illegal to drive a commercial vehicle in California); and (3) 11 healthy volunteers given no alcohol, to control for the effects of practice, learning, or fatigue on the results of repeated tests.      Testing took place just prior to drinking alcohol and 20-30 minutes later, during the time of rising blood alcohol levels following the drink. During this period three blood alcohol samples were taken.      What they found has already gained some media publicity, so should come as no surprise: people with sleep disordered breathing responded more slowly than those intoxicated with alcohol.      The difference in average reaction time between the groups showed a statistically significant disadvantage for the patients with sleep-disordered breathing compared to the alcohol-intoxicated group. The basis for statistical comparisons involved first translating the reaction time in milliseconds into seconds (ie, multiplying the former by 1000), then taking the reciprocal of the latter value (ie, dividing it into 1). The authors justified this in terms of statistical considerations, but as a result the values given in their tables often do not represent the actual reaction times, but a derived number. Furthermore, there does not seem to be any display making direct numerical comparisons possible between the mean reaction times of the patients with sleep-disordered breathing and those healthy subjects, with or without alcohol.      Their are three tables representing the reaction time results. The first (Table IV) shows reaction times over time in alcohol-intoxicated subjects only, first considered as a group, then separately for men and women. As one would expect, reaction times became slower as blood alcohol levels rose. This effect was statistically significant for all measures on the large total group, occasionally significant for the smaller groups of men and women considered separately.      The absolute magnitude of this effect (that is, the slowing of reaction time attributable to alcohol) was on the order of an 8-15% increase in reaction time from baseline (before alcohol) to the half-hour point where alcohol blood levels were maximal and reactions were slowest.      The second table of results (Table V) shows the reaction times of control subjects (healthy subjects never receiving alcohol) over a similar time interval of testing. This data showed no significant change over time, but here one must bear in mind that the group consisted of only 11 subjects (versus 80 and 113 for the alcohol and sleep disordered group) and that the smaller the group of subjects, the less chance of achieving statistical significance with any differences. This table is meant to demonstrate the neglible effect of time and repeated testing on the reaction time test results, but this non-effect might include both the beneficial effects of practice on reaction time and the opposite, negative effects of boredom and fatigue.      The final table (Table VI) compares the results of the present study's 91 healthy subjects (without alcohol) with those of two other studies, one examining 21 NASA Flight Crew and the other, 16 healthy young adults. There appear to have been no statistically significant differences so the authors interpreted the results of their study as comparable. Furthermore, because average ages of subjects in these groups differed (the NASA flight crew being oldest, averaging 41 years, and the healthy young adults of the last-mentioned study the youngest, averaging 23 years of age. The authors interpreted this as showing no effect of age on reaction time.      This last point attempts to address a major defect of their own study: the alcohol-challenged healthy group of 80 showed an average age (29 years) considerably lower than that (47 years) of their 113 subjects with sleep-disordered breathing.      The single reference made by the authors to the absolute magnitude of reactions was admittedly chosen to illustrate best the magnitude of the effect. I quote the authors below:      "The most extreme illustration of this major impairment in sleep-disordered subjects is that in these subjects the maximum RT [reaction time] of 566 milliseconds [I omit their reference to standard deviations] was comparable to and slightly greater than the worst mean performance at any BrAC (blood alcohol) in the alcohol-intoxicated subjects (554 milliseconds)."

My initial reaction to reading this article was that it illustrated well the nearly insuperable difficulty most lay readers would have in understanding original research reports as published in professional journals.      My next reaction was that I had great difficulty understanding this particular report, not because of unfamiliarity with the terms of the design of the study, but because of the remarkably abstruse way it was presented. I doubt that most clinical physicians without a research background would find their way through the article either--at least not to the point of viewing it from a critical perspective.      When, after considerable rereading, I felt that I had a reasonable grasp of the material, I came to feel that the one reason for its being presented in such a confusing way was to avoid bringing out the many serious flaws in the study design and data.      I do not want to exhaust my audience with the kind of critique I would have given had I been one of the journal's reviewers asked to evaluate the manuscript submitted. I will merely point out some of the greater flaws, which I have already suggested in the Summary above:   The authors appear to base almost all of their conclusions on the finding of "comparable" lengthening of reaction time in alcohol-intoxicated healthy subjects versus people with sleep-disordered breathing; in statistical language, such an attempt to prove an absence of differences is problematic if not impossible Those significant differences which they did find have little to do with their main point The issue of age effects on reaction time is a real one, which is nowhere in the data offered addressed adequately The patient and healthy groups were not comparable as to age and may well have proven dissimilar if compared on other relevant characteristics, such as education or socioeconomic status The two healthy control groups, though dissimilar in age, are also dissimilar in numbers (11 versus 80) and therefore unlikely to show significant differences in reaction times as a result of limitation of statistics in comparing different groups of very different sizes In all likelihood, though never specified, the two control groups were drawn from different population, the large group of younger healthy controls probably representing students, the smaller group possibly representing people who worked around the laboratory, making their comparability all the less probable To the extent that one can puzzle out the absolute magnitude of differences in reaction times between healthy, unintoxicated subjects and sleep-disordered subjects, they appear remarkably modest in size, even though quite statistically significant Nowhere do the authors reference other research on the relationship of differences of this magnitude to other measures, whether on neuropsychological testing or in actual risk of accidents, which might validate the importance of their finding To my reading, the authors have deliberately set the article up with the objective of offering all but the small group of people able to read their article critically, a conclusion not supported by their data but certain to draw public and government (i.e. funding source) attention       Lest you think these issues are irrelevant complaints of an overly critical researcher, I quote from the authors' concluding sentences:        ". . .this investigation has presented a comparative model to support the realization that the consequences of driving or engaging in safety-sensitive activities when sleepy as a result of SDB [sleep-disordered breathing] may be the same as driving under the influence of alcohol or, in some cases, driving drunk. In addition, it is hoped that this may further stimulate an awareness of the potential for altered performances at home and at work as a result of SDB."       Does this frighten you? It should. The authors are among the most esteemed researchers in the field of sleep, so their findings and opinions carry weight. Not only do their concluding comments suggest that people with sleep-disordered breathing shouldn't drive, they hint at justifications for finding them unsatisfactory as employees, employers, professionals, or even spouses and fathers!

This page hosted by Get your own Free Home Page