Nads-sc.uiowa.edu

Effects of Fexofenadine, Diphenhydramine, and Alcohol on
Driving Performance
A Randomized, Placebo-Controlled Trial in the Iowa Driving Simulator
John M. Weiler, MD; John R. Bloomfield, PhD; George G. Woodworth, PhD; Angela R. Grant, BS; Teresa A. Layton, BSN; Timothy L. Brown, MS; David R. McKenzie, MS; Thomas W. Baker, MS; Background: Sedating antihistamines may impair driving
Allergic rhinitis afflicts more than 39 million performance as seriously as alcohol.
persons in the United States (1). Only about Objective: To compare the effects of fexofenadine, di-
4.8 million persons (12%) take prescription drugs phenhydramine, alcohol, and placebo on driving perfor- for this condition; most go without treatment or self-treat with over-the-counter medications, which Design: Randomized, double-blind, double-dummy,
generally contain a first-generation antihistamine.
four-treatment, four-period crossover trial.
These medications may be effective but carry poten- tial risks, including drowsiness and impairment in Setting: The Iowa Driving Simulator.
performing everyday tasks (2–6). These adverse Participants: 40 licensed drivers with seasonal allergic
events may be sufficient to dissuade some persons rhinitis who were 25 to 44 years of age.
from treating their symptoms. Other patients take Intervention: One dose of fexofenadine (60 mg), di-
these sedating drugs, become impaired, and try phenhydramine (50 mg), alcohol (approximately 0.1% nonetheless to perform complex tasks; as a result, blood alcohol concentration), or placebo, given at weekly they are more likely to be involved in collisions (2, intervals before participants drove for 1 hour in the Iowa Our goal was to examine automobile driving per- Measurements: The primary end point was coherence, a
formance, a complex multiaspect task requiring continuous measure of participants’ ability to match the mental alertness; visual, auditory, and kinesthetic varying speed of a vehicle that they were following. Sec- information processing; eye–hand coordination; and ondary end points were drowsiness and other driving mea- manual dexterity. By using the Iowa Driving Simu- sures, including lane keeping and response to a vehiclethat unexpectedly blocked the lane ahead.
lator, a unique state-of-the-art facility, we evaluated driving performance measures and self-ratings of Results: Participants had significantly better coherence
drowsiness to determine the effects of alcohol and after taking alcohol or fexofenadine than after taking first- and second-generation antihistamines on driv- diphenhydramine. Lane keeping (steering instability andcrossing the center line) was impaired after alcohol and ing performance. No previous study has compared diphenhydramine use compared with fexofenadine use.
the effects of these drugs in the highly controlled Mean response time to the blocking vehicle was slowest after alcohol use (2.21 seconds) compared with fexofena-dine use (1.95 seconds). Self-reported drowsiness did not predict lack of coherence and was weakly associated withminimum following distance, steering instability, and left- Study Design
During ragweed season, we compared the effects Conclusions: Participants had similar performance when
of fexofenadine (60 mg), a second-generation anti- treated with fexofenadine or placebo. After alcohol use, histamine; diphenhydramine (50 mg) (Benadryl, participants performed the primary task well but not the Warner-Lambert Co., Morris Plains, New Jersey), a secondary tasks; as a result, overall driving performance first-generation antihistamine; alcohol; and placebo was poorer. After participants took diphenhydramine, on driving performance and self-reported drowsi- driving performance was poorest, indicating that diphen-hydramine had a greater impact on driving than alcohol ness of persons who were allergic to ragweed. A did. Drowsiness ratings were not a good predictor of im- randomized, double-blind, double-dummy, crossover pairment, suggesting that drivers cannot use drowsiness to design was used (9). The University of Iowa Insti- indicate when they should not drive.
tutional Review Board approved the study, and all Ann Intern Med. 2000;132:354-363.
See editorial comment on pp 405-407.
For author affiliations, current addresses, and contributions, see 2000 American College of Physicians–American Society of Internal Medicine participants signed a consent form before participa- completed drowsiness scales. After the drive, study staff determined vital signs. Participants were ob- served until they were sober. To maintain the dou- Inclusion and Exclusion Criteria
ble-blinding of the alcohol treatment, participants Key inclusion criteria were ability to remain for 5 remained for 5 hours or until the blood alcohol hours after the drives, history of alcohol use and level was less than 0.03% after alcohol and after willingness to consume alcohol, age 25 to 45 years, one of the other treatments (selected randomly). An seasonal allergic rhinitis caused by ragweed pollen, unblinded Clinical Research Center nurse with no previous successful use of antihistamine to treat sea- other study role determined alcohol levels by using sonal allergic rhinitis, status as a currently licensed a breath analyzer (Alco-Sensor, Intoximeters, Inc., experienced driver who drove an average of at least three times a week for at least 3 years, and 20/20 corrected vision. Key exclusion criteria were medical Treatment Preparation and Randomization
conditions that might interfere with ability to per- Capsules (fexofenadine, diphenhydramine, and form the study, pregnancy or lactation, unusual placebo) were blinded and packaged by Hoechst sleep patterns (including those of third-shift work- Marion Roussel, Inc. (Kansas City, Missouri). The ers), excessive alcohol consumption, use of tobacco Division of Pharmaceutical Service, College of Phar- in the past year or excessive caffeine consumption, macy, University of Iowa, Iowa City, Iowa, prepared previous experience in the Iowa Driving Simulator, and a positive result on a drug screening test.
Driving Simulation
Procedures
The Iowa Driving Simulator allowed collection of At visit 1, participants were selected on the basis data on driving performance measures in a manner of inclusion and exclusion criteria. Qualified partic- not available with on-street driving (11, 12). Briefly, ipants drove in the Iowa Driving Simulator for 8 the simulator consists of a domed enclosure mounted minutes; those with a tendency to develop simulator on a hexapod motion platform. The inner walls of the dome act as a screen on which correlated images Visits 2, 3, 4, and 5 (treatment visits) occurred weekly on the same day at the same time. Partici- The experimental drive was conducted in dry pants avoided consuming food or beverages, except weather conditions, with good visibility, on a two- water, for 2 hours before these visits. Participants lane rural highway that was 72.4 km (45 miles) long.
completed the baseline drowsiness visual analogue The lane widths were standard (3.66 m [12 ft]) and scale immediately before taking a capsule of fexo- the road surface was standard blacktop. The posted fenadine, diphenhydramine, or placebo; the drive speed limit was 88.6 km/h (55 miles/h) for most of was scheduled to start 2.5 hours later to coincide the course. Vehicles in the oncoming lane simulated with peak levels of antihistamine. Both researchers low-density traffic. Participants practiced driving in and participants were blinded to the treatment the simulator for 8 to 10 minutes before each ex- given. After treatment, participants were permitted perimental drive. The experimental drive consisted to consume only fluids; caffeine, stimulants, and of two phases driven consecutively without interrup- depressants were excluded. Vital signs were deter- tion. In phase 1 (30% of the total driving distance), mined and participants completed the second the driver followed a Volkswagen Golf. Phase 2 drowsiness scale 1 hour after taking the capsule.
began when this lead vehicle turned off the main The study beverage was dispensed 60 minutes be- road and participants continued to drive “as you fore the scheduled drive and was consumed over 20 normally would” along the designated route. In the to 30 minutes with a light snack. The dose of alco- first three sessions, the experimental drive ended hol (or placebo alcohol) was derived by using an uneventfully. At the end of the fourth and final algorithm that included the participant’s sex and session, participants encountered a vehicle that un- weight to reach an estimated blood alcohol concen- expectedly pulled out from a driveway into the lane tration of 0.1% (21.7 mmol/L) (10). Male partici- of the experimental vehicle. A truck with trailer pants received the equivalent of 800 mg of absolute simultaneously occupied the oncoming lane.
alcohol per kg of body weight, and female partici- pants received 640 mg/kg. Ninety-five percent alco- Outcome Measures
hol (or placebo alcohol) was added to a glass, which During the first phase, participants were in- was filled with the participant’s choice of noncaf- structed to maintain a constant distance behind a feinated carbonated soda. Alcohol was swabbed on lead car, which had realistic random velocity fluctu- the rim of each glass to maintain blinding. Imme- ations. The primary end point was coherence—the diately before and after the drive, participants again correlation between the velocity of the participant’s 7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 vehicle and the velocity of the lead vehicle. Partic- the raw data were fed into the Scenario Authoring ipants with high coherence were able to maintain a Tool, which replayed the drive using animation.
relatively uniform distance from the lead vehicle, whereas those with low coherence had more vari- Statistical Analysis
ability in distance between their cars and the lead The experiment was run as a crossover design with four periods and four treatments so that each In both phases of the drive, we evaluated steering participant received all four treatments (alcohol, di- instability, the root-mean-square deviation (in meters), phenhydramine, fexofenadine, and placebo) on four of the participant’s car around the participant’s pre- successive sessions in the driving simulator. With ferred position in the lane. Participants with high few exceptions, the sessions were 1 week apart at instability wandered left and right within (and some- the same time of day. Treatments were presented in times out of) the lane. We measured deviations 24 different sequences (such as AFDP and FDAP).
from the preferred position rather than the geomet- To ensure that each treatment occurred equally of- ric center of the lane to avoid penalizing otherwise ten in each period, the sequences were arranged in steady drivers who simply preferred to be closer to six Latin squares (for example, ADFP/DFPA/FPDA/ the center line or to the right shoulder line. We also PDAF). Four of the squares were replicated twice, evaluated left-lane excursions—the total number of for a total of 40 participants. The design was bal- times the participant partially or totally crossed the anced so that each treatment preceded and followed center line during the second phase of the driving the others equally often. Each treatment effect was estimated with equal precision in a model with We measured participants’ responses to the treatment, period, and first-order carryover effects.
blocking vehicle (the last event on the final drive).
In the design phase, we did an extensive Monte Videotapes and a computer-generated aerial view of Carlo investigation of the robustness of this design the driving course and vehicles (generated by using to random loss of participants (rows of data) and Scenario Authoring Tool software [National Ad- found that the selected design (along with several vanced Driving Simulator, Iowa City, Iowa]) were similar designs) was robust, much more so than a reviewed by three blinded investigators who evalu- design consisting of 10 replications of one Latin ated two aspects of the participants’ responses to the blocking vehicle. Response time was the time Crossover designs have advantages and draw- from the moment the blocking vehicle began to backs. With four treatments, a crossover design re- move until the instant the participant started to quires one fourth the number of participants re- respond. The blocking vehicle response category was quired by a completely randomized design in which based on whether the participant’s car came into each participant receives only one treatment. Fur- contact with the incoming car or approaching truck thermore, because each participant acts as his or (collision), stopped completely in the lane before her own control, it is, in theory, possible to compare passing the plane of the incoming car (clear avoid- treatments with much greater precision. One draw- ance), or either passed the plane of the incoming back of a crossover design is that early dropout of car or was more than a tire’s width out of lane participants complicates the analysis and may have a before stopping (potentially unsafe avoidance). Fi- comparatively greater impact on the precision of the nally, we evaluated drowsiness by using a visual results than the loss of a participant from a com- analogue scale (3, 4, 13–15) that asked participants pletely randomized trial. The most problematic as- to rate drowsiness from “I feel wide awake” to “I pect of crossover designs may be the effect of pre- vious experiences on a participant’s reaction to the current treatment. Such effects can be broadly clas- sified as period effects, such as learning or habitua- Data Capture, Reduction, and Management
tion, which are unrelated to previous treatments, Simulator data were collected in real time at 30 and carryover effects, which are related to previous Hz and were then reduced. During the data reduc- treatments. Although it is unlikely that any residual tion stage, checks were performed to ensure that study drug remained after an interval of 1 week, output was correct and meaningful. Data were visu- drug effects can carry over in other ways. For ex- ally inspected, sorted to identify extreme values, and ample, if a drug promoted simulator motion sick- plotted to ensure that all points were within natu- ness, the participant may have driven more cau- rally occurring boundaries. When extreme values tiously the week after receiving that drug. Without were found, operator and experimenter source doc- statistical adjustment, this drug-induced caution is uments were consulted to determine an explanation.
attributed to whichever drug was administered the Videotape records were inspected to establish the week after that drug. Statistical adjustment to re- origin of any anomalies in the data, and, if necessary, move period and carryover effects from the treat- 7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 Figure 1.
Maintenance of following distance for individual participants with near-best, near-worst, and near-median coherence scores. Initial
and final transients are removed. The lower the score, the more erratic the following distance. The best driver (coherence, 0.99) varied about Ϯ2.5 m, theworst driver (coherence, 0.54) varied about Ϯ35 m, and the median driver (coherence, 0.89) varied about Ϯ10 m.
ment means was accomplished by including variables Inc., Cary, North Carolina). The contrast tests were for these effects in the analysis of variance model.
two-sided, and an ␣ level of 0.05 was required.
Another complication of crossover designs is the Markov chain Monte Carlo computations were statistical relation among repeated measures in the made by using WinBUGS, version 1.2 (19). For the same participant. Participants’ performance in the primary and secondary outcome measures, we re- simulator is expected to be similar from week to port treatment means and differences; CIs are given week (that is, positively correlated), and variability for differences between treatment means.
may increase or decrease over time. Specifying the Response time to the blocking vehicle, which was form of the “covariance structure” of the data deals measured in the fourth driving session only (so that with such issues (16). For simplicity, we chose the only 25% of each treatment group was confronted most general possible covariance structure.
with this situation), was analyzed by using a general Finally, the statistical method we used (the mixed linear model with treatments as the only effect. Re- general linear model) requires that the data be ap- sponse to the blocking vehicle (clear avoidance, po- proximately normally distributed. Most of the out- tentially unsafe avoidance, or collision) was ana- comes we measured were significantly non-normally lyzed by using an exact permutation test (20).
distributed, and it was necessary to re-express (transform) them to achieve normality. We used Missing Data
Box-Cox analysis (17) to select an appropriate One participant fell asleep after receiving alcohol power transformation of each variable. Specifically, and could not be roused for a driving session. Data left-lane excursion counts were re-expressed as log- for four other participants were missing for the (count ϩ 1), coherence (c) was re-expressed as second half of phase 2 in one session because these (1 Ϫ ͌1 Ϫ c2)1.25, steering instability (s) was re-ex- participants had simulator sickness. Mechanical pressed as sϪ1, and minimum following distance (d) problems resulted in the loss of phase 1 data for was re-expressed as dϪ1/4. Statistics for re-expressed one participant in one session and data from the data are difficult to interpret; what does it mean, for second half of phase 2 for another participant in example, that “log crossing count plus one” is 2.1 one session. Thus, 2 of 160 sessions lacked phase 1 points higher for 1 drug than for placebo? To make data and 6 of 160 sessions lacked data from the our statistics interpretable, we converted all statisti- cal results—means, differences, and CIs—back to The theory of missing data distinguishes between the original, more interpretable measurement random and informative missing values (21). Ran- scales. Thus, crossing counts are reported as counts, domly missing data are those that are missing for steering instability and minimum following distance reasons unrelated to the participant’s response to are expressed in meters, and coherence is expressed the treatment; they are therefore distributed like in its original form as the correlation between the the observed data and can be predicted from the velocity of the lead car and that of the participant’s observed values of this participant and other partic- car. We used a Markov chain Monte Carlo proce- ipants. Informative missing data are missing for rea- dure (18, 19) to compute these statistics and CIs.
sons related to the participant or treatment and are All data were analyzed by using SAS software, likely to have been somewhat anomalous if ob- versions 6.12 and 7.0 for Windows (SAS Institute, served. Consequently, the fact that these data are 7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 missing gives some information about the unobserved ment. One participant elected to discontinue partic- value. For example, the participant who could not ipation during the first portion of her first drive and be woken up would probably have driven badly if was not included in the efficacy analysis. Fifteen she had been awakened, and the participants who men (37.5%) and 25 women (62.5%) were included developed simulator sickness would probably have in the analysis. The mean age was 31 years (range, driven badly if they had completed the session.
25 to 44 years); 37 were white. Participants had a The statistical software that we used imputes ran- mean duration of ragweed allergy of 20 years.
domly missing data with the predicted value but adjusts degrees of freedom and SEs to reflect the fact that these values are not real data. Regarding informative missing data, Chow and Liu (21) re- Coherence
mark that “There is no satisfactory, well-developed As explained above, coherence was a partici- methodology to account for missing values or inter- pant’s ability to maintain a constant distance from a mittent missing values.” We believed that we should lead car that varied its speed randomly. Figure 1
probe the sensitivity of the results to a range of provides a plot of distance fluctuations for three plausible imputed values of the missing data. There- representative participants, one each with near-best, fore, we did analyses to assess whether the results near-median, and near-worst coherence. Differences were sensitive to possible values for the informative in coherence (Table 1) were observed among the
missing data. In one analysis, we treated them as four treatments. Pairwise comparisons revealed that missing at random; in another (the worst-case analy- after taking diphenhydramine, participants per- sis), we imputed the nonrandomly missing values of formed car-following with significantly less coher- impairment measures (high ϭ bad) as the predicted ence than after taking alcohol, fexofenadine, or pla- value plus 2.5 SEs of the predicted value. We chose 2.5 SEs because it is pessimistic but does not distort the analysis by adding outliers. For performance Minimum Following Distance
measures (high ϭ good), we subtracted 2.5 SEs Significant differences in minimum following dis- from the predicted value. The results of the two tance (Table 2) were observed among the four
analyses did not differ substantively. In this article, treatments. Pairwise comparisons indicated that when we report the results of the second analysis.
participants performed car-following after consum- ing alcohol, they had significantly smaller minimum Role of the Study Sponsor
following distances (15.1 m) than they did after The industry sponsor had a consulting role in the taking fexofenadine (17.1 m) or placebo (17.4 m).
design, conduct, and reporting of the study. Deci- sions in all aspects of the study, including the deci- Steering Instability
sion to publish the results, were made by the authors.
Significant differences in steering instability (Ta-
ble 2) were observed among the four treatments.
Pairwise comparisons showed that after participants took fexofenadine, they had significantly less steer- Study Participants
ing instability than after taking diphenhydramine or Seventy-one participants were screened; 41 were alcohol (but not placebo). After participants took randomly assigned and received double-blind treat- placebo, they had significantly less steering instabil- ity than after consuming alcohol or diphenhydramine.
Primary End Point: Coherence*
After completing phase 1, participants drove the remaining 30 miles of the course “as you normally would drive.” Road signs and markings were the only guidance that they received in this phase.
Steering Instability
Significant differences in steering instability (Table 2) were again observed among the four
treatments. Pairwise comparisons demonstrated that after participants took fexofenadine, they had sig- nificantly less steering instability than after taking diphenhydramine or alcohol (but not placebo). Af- * Data are expressed as the mean Ϯ SD.
ter participants took placebo, they had significantly 7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 Secondary End Points*
4OOOOOOOOOOOOOOOOmOOOOOOOOOOOOOOOO3 * Data are expressed as the mean Ϯ SD. All data are based on 40 participants.
less steering instability than after consuming alcohol cantly more slowly (2.21 seconds) to the event than or diphenhydramine. After participants consumed after they took fexofenadine (1.95 seconds) (differ- alcohol, they had the same or less steering instabil- ence, 0.26 seconds [CI, 0.02 to 0.66 seconds]).
ity than after taking diphenhydramine.
Responses to the blocking vehicle were catego- rized as clear avoidance, potentially unsafe avoid- Lane Excursions
ance, or collision (Table 3). The overall differences
We determined the effect of treatment on the were not significant (P Ͼ 0.2, Fisher exact test).
probability that the participant’s vehicle moved to Pairwise comparisons, expressed as odds ratios, the right and partially or totally crossed the right- were also insignificant. However, because this event edge lane marker or moved to the left and partially occurred only during the fourth drive, there were or totally crossed the center line (Table 2). No sig-
only 9 to 11 participants in each group (rather than nificant differences for lane excursions to the right 40, as was the case for all of the other measures).
were noted among the four treatments. For excur- As a result, this analysis had far less power than the sions to the left, however, significant differences analyses of the other secondary measures.
were noted the four treatments. Pairwise compari- Crashes were evaluated for speed of the driver’s sons demonstrated that after participants took di- vehicle at the instant of the crash. For the 5 colli- phenhydramine, they crossed the center line signif- sions, the speed at impact was 46 and 14 miles per icantly more often than after taking fexofenadine or hour after alcohol, 37 and 8 miles per hour after placebo. After participants took alcohol, they crossed diphenhydramine, and 6 miles per hour after fexo- the center line significantly more often than after taking fexofenadine and placebo. Fexofenadine and placebo did not differ significantly; the 95% CIs Subjective Drowsiness Ratings
indicate that the difference is small (Table 2).
Drowsiness ratings were expressed as differences Response to Blocking Vehicle
between the drowsiness scales completed after treat- No significant main effect of treatment on the ments and the baseline scale (Figure 2). Scores on
response time to the blocking vehicle was observed, the second visual analogue scale, given 1 hour after although pairwise comparisons showed that after capsule administration, had small average differ- consuming alcohol, participants responded signifi- ences from baseline (Ͻ10 points), and no significant 7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 Clear Avoidance, Potentially Unsafe Avoidance, and Collision in the Final Driving Session
* The denominator of the odds ratio was zero. Only the lower limit of the CI is given; the upper limit was unbounded.
differences were seen among the treatment groups difference between the third and first self-ratings.
(the confidence limits for differences less than Ϯ10 The correlation between drowsiness and the primary points). At the time of the third visual analogue end point, coherence, was not statistically significant scale, just before the drive, participants were most (Table 4). Statistically significant but small correla-
drowsy after taking diphenhydramine and least tions were found between subjective drowsiness and drowsy after taking fexofenadine or placebo. The minimum following distance, steering instability, and differences between diphenhydramine and fexofena- left-lane excursions; no correlation was greater than dine or placebo were significant (confidence limits ranged from 5 to 27 points on the 100-point visual Although a significant correlation indicates some analogue scale). The difference between fexofena- relation between two variables, the size of the cor- dine and placebo was less than 1 point, with confi- relation coefficient is not a good indicator of the dence limits of Ϯ11 points. After the drive, partic- strength of that relation. To give an idea of the ipants were most drowsy with diphenhydramine and practical meaning of the correlations we observed, least drowsy with placebo. The difference between Table 4 shows mean driving performance values for
fexofenadine and placebo was insignificant (confi- participants who had increases in drowsiness scores dence limits were Ϫ7 to 19 points). With diphenhy- in the upper quartile and lower three quartiles.
dramine, participants reported significantly higher Clearly, drowsiness was a weak predictor of poor levels of drowsiness than with fexofenadine (confi- driving. Indeed, only one of the five collisions oc- dence limits of 9 to 35 points) and placebo (confi- curred among the participants who were in the drowsiest quartile (as measured before or after the We examined whether self-reported drowsiness drive). Thus, “grounding” the drowsiest 25% of immediately before driving predicted impaired driv- drivers would have prevented only 20% of the col- ing performance. Drowsiness was expressed as the lisions. In contrast, three of five collisions occurred in participants who had the lowest quartile of fol- lowing distances (following distance Ͻ 12.2 m), and four of five collisions occurred in participants who had the highest quartile of left-lane crossings (seven Adverse Events
No unusual or serious adverse events were ob- served in this study. Adverse events occurred with similar frequency after all four treatments, with no significant differences between any two treatments Discussion
Figure 2.
Change from baseline in visual analogue drowsiness
scores. Participants rated drowsiness on a scale from “wide awake” to
First-generation antihistamines, such as diphen- “extremely drowsy,” which corresponded to a score of 1 to 100 on a159-mm scale.
hydramine, cause sedation (2–6), which Gengo and 7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 Gabos (22) have distinguished as impairment and ministered to participants who were engaged in drowsiness. Cognitive impairment refers to some complex tasks that required divided attention.
interference with the patient’s ability to perform Horne and Baumber (25) reported that drivers who tasks and is measured by objective tests; drowsiness, had consumed alcohol were able to maintain lateral which may or may not limit performance, is mea- position in wind gusts but did not perform well at sured subjectively. The least desirable condition following another vehicle. Landauer and Howat (10) would be impairment without drowsiness because a used a nondriving task involving reaction time and patient might have no subjective clues suggesting tracking accuracy and found that after participants consumed alcohol, reaction time improved slightly The second-generation antihistamines have diffi- but the number of tracking errors increased. Mos- culty crossing the blood–brain barrier and are be- kowitz (26) and Kerr and Hindmarch (27) reviewed lieved to cause little or no central nervous system studies of alcohol and divided attention and sug- depression. In previous studies, fexofenadine and its gested that if one part of a divided attention task is parent compound, terfenadine, did not impair the perceived to be primary and the other part second- performance of automobile drivers or airplane pi- ary, only the secondary task becomes impaired.
When we examined how participants performed In this study, participants in a driving simulator when driving “normally,” we found more steering were first instructed to match the speed of the car instability after participants took diphenhydramine they were following, then to drive “as you normally or alcohol than after they took fexofenadine or would.” Coherence was chosen as the primary end placebo. Participants with poorer steering were point because the added complexity of trying to more likely to drive with part of the vehicle out of match the variable speed of the lead car might lead the lane. Lane excursions over the center line (caus- to greater sensitivity if impairment did occur. Co- ing potential exposure to oncoming traffic) may se- herence was significantly better after participants riously affect safety. The numbers of lane excursions took alcohol or fexofenadine than after they took over the center line more than doubled after the diphenhydramine. The minimum following distance participants had taken diphenhydramine compared was slightly shorter than the recommended distance after all four treatments (15.1 m [49.4 ft] to 17.4 m We also examined drowsiness and found that [57.2 ft]). The mean minimum following distance participants were significantly drowsier after taking was about one-half car length longer (and safer) diphenhydramine than after taking any of the other after participants had taken fexofenadine or placebo treatments. However, we found that subjective than after they had consumed alcohol. The shorter drowsiness either did not predict driving perfor- following distance might also have contributed to mance measures (coherence) or was a relatively increased coherence. However, during the car-fol- weak predictor (for minimum following distance, lowing phase, steering instability scores were highest steering instability, and left-lane excursion). This after diphenhydramine or alcohol use, indicating suggests that drivers who use alcohol or diphenhy- dramine are probably mistaken if they believe that Thus, although participants under the influence lack of drowsiness means that they will be able to of alcohol did surprisingly well at matching the ve- locity of the lead car, they did so at the expense of The potential crash scenario on the last drive driving closer to that vehicle and having less control provided some additional evidence of impairment.
over steering. These findings agree with the results Participants had to react to a vehicle that unexpect- obtained in other studies in which alcohol was ad- edly pulled out of a driveway and blocked their lane.
Performance Measures according to Degree of Sleepiness before Drive 4
* Increase in drowsiness between baseline (first drowsiness evaluation) and the third evaluation (immediately before a drive).
† r ϭ 0.06; P Ͼ 0.2.
‡ r ϭ 0.20; P ϭ 0.01.
§ r ϭ 0.20; P ϭ 0.01.
࿣ r ϭ 0.21; P ϭ 0.01.
¶ Statistically significant (P ϭ 0.006).
7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 Participants responded significantly more slowly to Fossen, Dave Bronder, Shawn Allen, Rachel Nador, Steven the event after consuming alcohol than after taking Zellers, Ianos Schmidt, Paul Debbins, and Dave Muller.
fexofenadine. At the posted speed, this slower reac- Grant Support: By a grant from Hoechst Marion Roussel, Inc., tion time resulted in a stopping distance that was and by grant M01-RR-59 from the National Center for Research Resources, General Clinical Research Centers Program, National The observations reported here, combined with past reports, indicate that diphenhydramine clearly Requests for Single Reprints: John M. Weiler, MD, Department of Internal Medicine, University of Iowa, 200 Hawkins Drive, impairs driving performance, whereas the second- T307GH, Iowa City, IA 52242-1081; e-mail, john-weiler@uiowa generation antihistamine fexofenadine was indistin- guishable from placebo. Vermeeren and O’Hanlon Requests To Purchase Bulk Reprints (minimum, 100 copies): Bar- (24) studied one driving variable, lateral position, bara Hudson, Reprints Coordinator; phone, 215-351-2657; e-mail, and also reported that fexofenadine did not affect standard deviation of lateral position in an instru- Current Author Addresses: Dr. Weiler: University of Iowa Hospi- mented car used in an on-the-road study, nor did it tals and Clinics, T307, 200 Hawkins Drive, Iowa City, IA 52242.
affect various nondriving psychomotor tasks. In con- Dr. Woodworth: Department of Statistics, University of Iowa, trast, the first-generation antihistamine clemastine 241 Schaeffer Hall, Iowa City, IA 52242.
Ms. Grant: 1660 Fulmer Street, Ann Arbor, MI 48103.
Ms. Layton: University of Iowa Hospitals and Clinics, 200 In the United States, diphenhydramine is the Mr. Brown: 4133 SC, University of Iowa, Iowa City, IA 52242.
top-selling over-the-counter medication sold for Mr. McKenzie: Quintiles, Inc., Box 9708, Kansas City, MO treatment of allergic rhinitis (28). It is estimated that 47% of persons with allergies treat themselves Mr. Baker: Aventis Pharmaceuticals, Inc., Box 9627, Kansas City, with over-the-counter products, most of which con- Dr. Watson: National Advanced Driving Simulator, 2401 Oak- tain a sedating antihistamine (29). Consequently, dale Boulevard, University of Iowa, Iowa City, IA 52242.
millions of patients use first-generation antihista- Author Contributions: Conception and design: J.M. Weiler, J.R.
Bloomfield, A.R. Grant, G.G. Woodworth, D.R. McKenzie, T.W.
Several health programs have been developed that Analysis and interpretation of the data: J.M. Weiler, J.R.
limit patient access to nonsedating antihistamines and Bloomfield, A.R. Grant, T.L. Brown, G.G. Woodworth, G.S.
emphasize the use of first-generation antihistamines (30, 31). The cost savings of these programs should Drafting of the article: J.M. Weiler, T.A. Layton, D.R. Mc- be weighed against the potential increased risk to Critical revision of the article for important intellectual con- the driving public and against the laws of 27 states tent: J.M. Weiler, J.R. Bloomfield, A.R. Grant, G.G. Wood- that prohibit driving under the influence of any drug worth, D.R. McKenzie, T.W. Baker, G.S. Watson.
Final approval of the article: J.M. Weiler, G.G. Woodworth, T.A. Layton, T.L. Brown, T.W. Baker, G.S. Watson.
We conclude that participants performed simi- Provision of study materials or patients: J.M. Weiler, T.W.
larly when treated with fexofenadine or placebo.
Statistical expertise: G.G. Woodworth, D.R. McKenzie.
Participants who consumed alcohol did well in per- Obtaining of funding: J.M. Weiler, J.R. Bloomfield, D.R. McKen- forming the primary driving task but not the sec- Administrative, technical, or logistic support: A.R. Grant, T.A.
ondary tasks, resulting in poorer overall driving per- formance. This study demonstrates that the first- Collection and assembly of data: J.M. Weiler, J.R. Bloomfield, generation antihistamine diphenhydramine may A.R. Grant, T.A. Layton, T.L. Brown.
have an even greater impact than does alcohol on the complex task of operating an automobile.
References
From the University of Iowa, Iowa City, Iowa; and Hoechst Marion Roussel, Inc., Kansas City, Missouri.
1. Malone DC, Lawson KA, Smith DH, Arrighi HM, Battista C. A cost of
illness study of allergic rhinitis in the United States. J Allergy Clin Immunol.
Disclosure: Dr. Weiler serves as a consultant and Dr. Woodworth has provided consulting services to Hoechst Marion Roussel, Inc.
2. Starmer G. Antihistamines and highway safety. Accid Anal Prev. 1985;17:
Acknowledgments: The authors thank the following for their as- 3. Gengo F, Gabos C, Miller JK. The pharmacodynamics of diphenhydramine-
induced drowsiness and changes in mental performance. Clin Pharmacol sistance in the conduct or analysis of the study: Susan Quinn, Sue Ellen Salisbury, Elizabeth Lawler, Cindy Mitchell, Emily Meis, 4. Gengo FM, Gabos C, Mechtler L. Quantitative effects of cetirizine and
Kathy Phipps, Suzanne Sack, Jagadish Boggavarapu, Dixie Eck- diphenhydramine on mental performance measured using an automobile lund and the Clinical Research Center staff, Twila Finkelstein, driving simulator. Ann Allergy. 1990;64:520-6.
5. Brookhuis KA, De Vries G, De Waard D. Acute and subchronic effects of
Julie Qidwai, Christopher Miller, Joss Nichols, Brent Caven, the H1-antihistamine receptor antagonist ebastine in 10, 20 and 30 mg dose, Mark Young, Dawn Kenyon, Kristen Rassbach, Brad Graham, and triprolidine 10 mg on car driving performance. Br J Clin Pharmacol.
Chris McMillan, Nick Taiber, Sneha Viratia, Srinivas Maddhi, Rohit Goal, Lucas Davisson, Brian Berentsen, Shaheen Bahaud- 6. O’Hanlon JF, Ramaekers JG. Antihistamine effects on actual driving perfor-
mance in a standard test: a summary of Dutch experience, 1989-94. Allergy.
din, Peter Grant, Katie Enstrom, Omar Ahmad, Imran Pirwani, Ludovic Moineau, Yiannis Papelis, Matthew Schikore, Tim Van 7. Cimbura G, Lucas DM, Bennett RC, Warren RA, Simpson HM. Incidence
7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5 and toxicological aspects of drugs detected in 484 fatally injured drivers and 20. The FREQ Procedure, Exact Statistics. In: SAS/STAT Software: Changes and
pedestrians in Ontario. J Forensic Sci. 1982;27:855-67.
Enhancements through Release 6.12. Cary, NC: SAS Institute; 1997:221-46.
8. Gengo FM, Manning C. A review of the effects of antihistamines on mental
21. Chow SC, Liu JP, eds. Issues in efficacy evaluation. In: Design and Analysis
processes related to automobile driving. J Allergy Clin Immunol. 1990;86: of Clinical Trials: Concept and Methodologies. New York: Wiley; 1998:531-2.
22. Gengo FM, Gabos C. Antihistamines, drowsiness, and psychomotor impair-
9. Chow SC, Liu JP. Section 5.3. Crossover designs. In: Design and Analysis of
ment: central nervous system effect of cetirizine. Ann Allergy. 1987;59:53-7.
Clinical Trials: Concept and Methodologies. New York: Wiley; 1998:183-90.
23. Philpot EE, Brooker AE, Biegalski CS. Effects of sedating and nonsedating
10. Landauer AA, Howat P. Low and moderate alcohol doses, psychomotor
antihistamines on flying performance. Mil Med. 1993;158:654-60.
performance and perceived drowsiness. Ergonomics. 1983;26:647-57.
24. Vermeeren A, O’Hanlon JF. Fexofenadine’s effects, alone and with alcohol,
11. Kuhl JG, Evans D, Papelis YE, Romano RA, Watson GS. The Iowa Driving
on actual driving and psychomotor performance. J Allergy Clin Immunol.
Simulator: an immersive research environment. IEEE Computer. 1995;28:35- 25. Horne JA, Baumber CJ. Time-of-day effects of alcohol intake on simulated
12. Freeman JS, Watson GS, Papelis YE, Lin TC, Tayyab A, Romano RA, et
driving performance in women. Ergonomics. 1991;34:1377-83.
al. The Iowa Driving Simulator: An Implementation and Application Overview.
26. Moskowitz H. Validity of driving simulator studies for predicting drug effects
Warrendale, PA: Soc of Automotive Engineers International; 1995. SAE Tech- in real driving situations. In: Israelstam S, Lambert S, eds. Alcohol, Drugs and Traffic Safety: Proceedings of the Sixth International Conference on Alcohol, 13. Witek TJ, Canestrari DA, Miller RD, Yang JY, Riker DK. Characterization
Drugs, and Traffic Safety. Toronto: Addiction Research Foundation of Ontario, of daytime sleepiness and psychomotor performance following H1 receptor antagonists. Ann Allergy Asthma Immunol. 1995;74:419-26.
27. Kerr JS, Hindmarch I. The effects of alcohol alone or in combination with
14. McCormack HM, Horne DJ, Sheather S. Clinical applications of visual
other drugs on information processing, task performance and subjective re- analogue scales: a critical review. Psychol Med. 1988;18:1007-19.
sponses. Human Psychopharmacology. 1998;13:1-9.
15. Chernik DA, Gilllings D, Laine H, Hendler J, Silver JM, Davidson AB, et
28. Red Book. Montvale, NJ, Medical Economics; 1998:609.
al. Validity and reliability of the Observer’s Assessment of Alertness/Sedation
29. Smith EA. Blame El Nin˜o. Drug Topics. 1998;142:60.
Scale: study with intravenous midazolam. J Clin Psychopharmacol. 1990;10: 30. Meszaros L. Allergy: old nemesis, new approaches. Managed Healthcare.
16. The MIXED procedure. In: SAS/STAT Software: Changes and Enhancements
31. Use of OTC antihistamines, NSAIDs saves 12% over brand names. Managed
through Release 6.12. Cary, NC: SAS Institute; 1997:571-702.
17. Weisberg S. Applied Linear Regression. 2d ed. New York: Wiley; 1985.
32. Digest of State Alcohol-Highway Safety Related Legislation. 14th ed. Wash-
18. Gelman A, Rubin DB. Markov chain Monte Carlo methods in biostatistics.
ington, DC: Office of Alcohol and State Programs, National Highway Traffic Stat Methods Med Res. 1996;5:339-55.
Safety Administration, U.S. Department of Transportation; 1996.
19. Spiegelhalter DJ, Thomas A, Best NG. WinBUGS Version 1.2. User Man-
33. Slavin RG. Occupational and allergic rhinitis: impact on worker productivity
ual. Cambridge, UK: MRC Biostatistics Unit; 1999.
and safety. Allergy Asthma Proc. 1998;19:277-84.
7 March 2000 • Annals of Internal Medicine • Volume 132 • Number 5

Source: http://www.nads-sc.uiowa.edu/bio_pub_pdf/Weiler_et_al_-_Effects_of_fexofenadine,_diphenhydramine,_and_alcohol.pdf