Airline Passengers' Alcohol Use and Its Safety Implications

Authors

  • Deborah C. Girasek MPH, PhD,

    Corresponding author
    1. Uniformed Services University of the Health Sciences, Department of Preventive Medicine & Biometrics, Bethesda, MD, USA
    Search for more papers by this author
  • Cara H. Olsen MS, DrPH

    1. Uniformed Services University of the Health Sciences, Department of Preventive Medicine & Biometrics, Bethesda, MD, USA
    Search for more papers by this author

Deborah C. Girasek, MPH, PhD, Associate Professor, Uniformed Services University of the Health Sciences, Department of Preventive Medicine & Biometrics, 4301 Jones Bridge Road, Bethesda, MD 20814, USA. E-mail:DGirasek@usuhs.edu

Abstract

Background. This investigation was designed to identify individual and contextual factors associated with airline passengers' alcohol use, and to explore potentially dangerous in-flight alcohol consumption.

Methods. Passengers waiting to board 24 domestic flights at an international airport in the South Atlantic United States were asked to complete an anonymous questionnaire. Eighty percent of those approached agreed to participate. Our findings are based upon the responses of 1,548 adults.

Results. A majority (84%) of passengers indicated that they did not intend to consume alcoholic beverages on the plane they were waiting to board. Passengers who were more likely to report that they would drink were on longer flights, traveling with friends, and anticipating First or Business Class seating. Passengers who had already consumed alcohol that day and those who drank more often generally were also more likely to say that they intended to consume alcohol, as were people who thought that in-flight alcohol use was enjoyable, acceptable, and unlikely to make jet lag worse. Eighty-nine percent of the passengers who said they intended to drink reported that they would purchase one to two drinks.

Conclusions. Very few respondents reported intentions that would pose a risk to others. Future studies should validate alcohol consumption and sample passengers at multiple airports throughout the year.

There has been virtually no scientific scrutiny of alcohol use among airline passengers. 1,2 This topic has received some attention in the lay press, usually when incidents of “air rage” are attributed to excess alcohol consumption. Disruptive passenger behavior can range from noncompliance with safety instructions to the physical assault of airline staff.

Although there are indications that such incidents may have declined in recent years,3 they were described as having reached “epidemic proportions” at the 2000 International Aircraft Safety Symposium.4 The International Air Transport Association reported a fivefold increase in air rage incidents between 1994 and 1997.5 In that same time period, a voluntary government reporting system in the United States recorded an 800% increase in reports of unruly passenger behavior.6

Although it has been acknowledged that too little is known about air rage,7 two separate surveys of airline employees cited alcohol as its major cause.8,9 In the United Kingdom, alcohol was identified as a contributory factor in 45% of the incidents reported to the government between 1999 and 2003.3 Similarly, when US officials analyzed a year's worth of voluntary reports on passenger misconduct, “overindulgence in alcoholic beverages” emerged as “the most prevalent contributory factor…by far.”4 An analysis of newspaper accounts of air rage cases in Canada cited “drunkenness/being cut off from alcohol” as their most common precipitating event.10 Anglin and colleagues have characterized our understanding of this problem as being analogous to “where the drunk driving problem was 40 years ago.”11

Disruptive passenger behavior is just one of several public health issues that may be related to airline passengers' use of alcohol. Many people drive home from airports after deplaning. Almost half of all rental car revenue in the United States is generated at airports,12 yet we have no data on the proportion of those leaving airports who are intoxicated. In one civil lawsuit that was brought against an airline for overserving a passenger who went on to drive while intoxicated, the plaintiff's private investigator was reportedly served 12 drinks while sitting in the first class cabin for a 90-minute flight.13 In New Mexico, authorities halted alcohol sales on two airlines after a recently deplaned passenger caused a crash that killed a family of five.14 His blood alcohol level was four times the state's legal limit, and witnesses reported that he had been served alcohol in-flight despite appearing intoxicated.

Another potential concern is a passenger's ability to evacuate a commercial aircraft in the event of an emergency. In one report of a runway collision that resulted in a fire, 76% of the passengers in a Boeing 737 were able to escape to safety. Among the fatalities was a first class passenger with a blood alcohol level of 0.24%, who had not even unfastened his seat belt.2 This issue is particularly relevant for people sitting in the overwing exit row, who assume responsibility for operating that hatch when they agree to sit in those seats. This task has been documented as challenging from both physical and cognitive perspectives.15,16

We were also interested in whether airline passengers who drink alcohol are less likely to wear their seat belts, as has been documented for automobile occupants.17–19 Unbelted airline passengers are at increased risk for turbulence injuries, which can be fatal.20 That aspect of our study has been published elsewhere.21 In brief, we found that airline passengers who intended to consume alcohol on their upcoming flight were twice as likely to report that they “rarely or never” wore their seat belts when the seat belt sign was not illuminated.

A final reason to increase our understanding of airline passengers' alcohol use is that air travelers are often advised to forgo alcohol consumption in hopes of reducing their risk of developing deep vein thrombosis.22

Methods

Since the aircraft cabin environment has not been studied in relation to alcohol consumption, we felt it was important to collect qualitative data before beginning our quantitative study. To this end, we held three focus groups with airline passengers. Their input, and our review of the general alcohol literature, resulted in the list of candidate independent variables displayed in Table 1. We developed a survey instrument to measure these factors, using standardized items when they were available.23

Table 1.  Factors hypothesized as being associated with airline passengers' alcohol consumption
Passenger characteristicsSituational characteristics
  1. *These variables were added or refined as a result of our focus groups.

DemographicsDay of week
Fear of flyingTime of day
Water/hydration concerns*Flight duration
Mood*Purpose of travel
Alcohol outcome expectanciesWork responsibilities*
Perceived normsDesire to sleep*
Past alcohol consumptionCultural context
Alcohol dependencySpecial occasion*
Nicotine dependencyHeading home or away
 Travel companions
 Site of consumption
 Class of service
 Exit row seating
 Aisle vs other seating*
 Driving after deplaning

Our survey instrument and study procedures were approved by the human subjects committee of the Uniformed Services University of the Health Sciences. Our questionnaire was distributed to passengers who were waiting to board 24 domestic flights at a major international airport located in the South Atlantic United States. Flights were selected to cover a range of destinations, lengths, and dates/times of departure, but those departing before noon were excluded from the study. Our data were collected from November 2005 through March 2006.

Interviewers were trained to approach all adult passengers who appeared to be waiting for the study flight. Passengers were eligible if they confirmed that they were ticketed for the study flight, were at least 21 years of age, could speak and read English or Spanish, and were not airline employees. Data collection began approximately 2 hours before the flight's scheduled departure and continued until the flight's last passengers had boarded. Our introduction informed passengers of the study's voluntary and anonymous nature. They were also informed of the study's sponsor, focus (ie, “airline passenger behavior”), length (“less than 10 min”) and the fact that they would receive a gourmet coffee coupon worth $3 if they agreed to participate.

In an attempt to confirm that reported intentions to consume alcohol were well correlated with actual in-flight behavior, we provided 20% of passengers with a $10 gift check and invited them to call a toll-free number after deplaning. Those who called were asked for their coupon number (ie, so that their responses could be linked to their preflight survey), whether they had consumed any alcohol on the study flight, and (if so) the type and amount. We also purchased alcohol sales data—by flight—from the airline.

We used chi-square and t-tests to explore associations between independent variables and reported intentions to consume alcohol in-flight. Factors with bivariate associations of p < 0.10 were entered into a logistic regression model. Three of our four measures of usual/past alcohol consumption habits were excluded due to collinearity, and we reduced our model several times to produce a final model that included only factors independently associated with drinking intentions at the p < 0.05 level. A Hosmer-Lemeshow test of our fitted model was also conducted. To account for the possible effects of clustering (ie, unmeasured common traits that might be shared by passengers on the same flights), we used the Huber-White robust variance estimator.24 Potentially dangerous patterns of alcohol consumption will also be described.

Results

Sample

Of the 2,013 passengers to whom we offered survey participation, 80% ( n = 1,615) agreed. Usable surveys were collected from 96% of those volunteers, resulting in a final sample size of 1,548. Participants are described in Table 2.

Table 2.  Characteristics of airline passengers in survey sample
Characteristic% ( n)
Gender ( n = 1,484) 
 Male55% (817)
 Female45% (667)
Age ( n = 1,457) 
 Mean44 years SD = 14.16
Race ( n = 1,548) 
 White78% (1,201)
 Asian9% (139)
 African-American4% (55)
 Other10% (153)
Hispanic ethnicity ( n = 1,484) 
 No95% (1,391)
 Yes5% (74)
Education ( n = 1,471) 
 High school or less7% (100)
 Some college13% (194)
 College degree28% (410)
 Graduate work52% (767)
Household income ( n = 1,387) 
 < $50,000 per year15% (207)
 $50,000–$99,99929% (401)
 ≥ $100,00056% (779)

Factors Associated with Alcohol Consumption

Fourteen percent ( n = 212) of respondents responded that they intended to consume alcoholic beverages on the flight they were waiting to board. Male passengers ( p = 0.07), those who were White ( p = 0.01), non-Hispanics (p = 0.03), and respondents who reported higher household incomes ( p < 0.01), were more likely to answer this question in the affirmative. Participants were more likely to report drinking intentions if they were not “jittery right now” ( p < 0.01) and did not agree that “airlines charge a lot for alcoholic beverages” ( p < 0.01). The outcome expectancies that were positively associated with reported alcohol intentions included believing that alcohol is: relaxing ( p < 0.01), increases their enjoyment of flying ( p < 0.01), makes them more friendly to the person sitting next to them ( p < 0.01), does not increase their aggressiveness ( p < 0.01), does not have more of an effect if consumed during a flight (p < 0.01), and does not make jet lag worse ( p < 0.01). Our social norm items were also significantly associated with drinking intentions [ie, reporting that the respondent would not feel embarrassed if other passengers thought they were drunk ( p < 0.01), and thinking that passengers “should feel free” to have a couple of drinks when they are flying ( p < 0.01)]. Most of our questions relating to usual (ie, nonaviation) drinking habits were associated with preflight drinking intentions: more frequent alcohol use ( p < 0.01), consuming more drinks on a “typical” day of drinking ( p < 0.01), frequency of drinking “five or more drinks on one occasion” ( p < 0.01), and having consumed a higher number of drinks in a 1 day period in the last 12 months ( p < 0.01). Finally, passengers who believed that being sober did not offer a survival advantage in the event of an emergency landing were more likely to report that they intended to drink on the flight they were waiting to board ( p < 0.01).

Situational characteristics that were associated with a significantly higher likelihood of respondents reporting that they were going to drink on board the plane included: evening flights ( p < 0.01), longer flights ( p < 0.01), sitting in First or Business class ( p < 0.01), having a nonaisle seat ( p = 0.07), having no work responsibilities on the day of the flight ( p = 0.06), traveling with friends ( p = 0.02), flying under conditions that are associated with alcohol use in the United States (eg, to a football playoff, to Las Vegas, on New Year's Eve) ( p = 0.02), and having already consumed alcohol on the day of the flight ( p < 0.01).

Passenger characteristics that were measured but found to be nonsignificant included age, gender, education, being “a nervous flyer,” taking care not to get dehydrated when flying, having concerns about the quality of the drinking water served on planes, being in a “bad mood,” being nicotine-dependent,25 being alcohol-dependent, and knowing the fine for interfering with a flight attendant. Nonsignificant contextual factors included weekend (vs nonweekend), trip purpose, intending to sleep, celebrating something, leaving versus coming home, being seated in an emergency exit row, having a nonaisle seat, and planning to drive home from the airport.

The factors that emerged as being significantly associated with intended alcohol consumption after we performed the exploratory regression analyses described under Methods are displayed in Table 3. Our final regression model was estimated using data from 1,444 surveys, because 7% of study participants were missing responses to key variables.

Table 3.  Fitted regression model predicting intended alcohol consumption by airline passengers
VariableOdds ratio95% Confidence intervalp Value
Seating class  <0.01
Coach/Economy1.00  
Business/First class5.473.29–9.09<0.01
Don't know4.351.78–10.62<0.01
Alcohol is expensive  <0.01
Agree1.00  
No opinion0.310.19–0.51<0.01
Disagree1.170.72–1.890.52
Enjoy flying more when drinking  <0.01
Agree1.00  
No opinion0.460.30–0.71<0.01
Disagree0.280.17–0.45<0.01
Jetlag worse with alcohol  0.05
Agree1.00  
No opinion1.530.97–2.390.07
Disagree1.921.12–3.280.02
It's okay to drink on planes  0.07
Agree1.00  
No opinion0.860.51–1.440.56
Disagree0.420.20– 0.870.02
How often drink  <0.01
Never1.00  
Monthly or less2.580.53–12.510.24
2–4 times month9.462.14–41.73<0.01
5 or more per month11.362.58–49.93<0.01
4 or more per week26.735.63–126.82<0.01
Flight length  <0.01
<3 hours1.00  
3–4 hours2.561.2–5.440.02
>4 hours2.701.79–4.08<0.01
Traveling companions  <0.01
No one they know1.00  
People from work0.660.35–1.240.20
Relative(s)1.510.96–2.360.07
Friend(s)2.501.41–4.42<0.01
Other1.350.61–2.990.47
Drank today   
Yes1.00  
No0.380.25–58<0.01

We found that endorsing social norms that condoned in-flight drinking, holding positive outcome expectancies (or not endorsing negative statements about alcohol), traveling with friends, sitting (or hoping to be seated) in First or Business Class, longer flights, drinking more alcohol in other contexts, and having consumed alcohol earlier in the day were associated with reported intentions to drink on the flight that respondents were waiting to board. This same set of variables remained significant after our standard errors were adjusted for cluster sampling.

Dangerous Patterns of Alcohol Use?

Of those who said they intended to drink, the average level of consumption reported was 1.6 drinks ( SD = 1.04). When passenger levels of intended consumption were divided by flight length, 95% of respondents reported that they would consume less than one drink per hour.

It is worth noting, however, that 14% of respondents had already consumed alcohol on the day of their flight. According to their surveys, 78% had had one drink, and half had consumed their last drink within an hour of completing the survey. Almost half of these passengers reported having had their drinks in the airport; a third had had their drinks on a plane.

The alcohol use of certain subgroups of airline passengers was of particular interest because they pose a potential threat to other people. These included passengers sitting in an overwing (“emergency”) exit row, people with a history of aggressive behavior after using alcohol, and those who planned to drive a motor vehicle from their destination airport. Passengers who reported that they had gotten into a “physical fight” while drinking or right after drinking in the last 3 years were not significantly more likely to report that they would consume alcohol on their upcoming flight. Passengers among that group who did intend to drink, however, said they would order 2.6 (SD = 2.4) drinks on average, versus 1.6 (SD = 0.91) for other passengers. Three of those who reported a history of fighting after drinking ( n = 49) said that they intended to drink more than a drink per hour.

Agreeing that, “Drinking tends to increase my aggressiveness,” was negatively associated with intending to consume alcohol on the upcoming flight. Two passengers who endorsed that statement, however, intended to drink more than one drink per hour of flight.

Passengers sitting in overwing exit rows were not significantly less likely to drink than their fellow passengers. The amount of alcohol they said they would consume, however (1.2 drinks on average, SD = 0.44), would not be expected to render them impaired.

Passengers who expected to drive away from the airport upon reaching their final destination did not differ significantly from nondriving respondents, with regard to whether they intended to drink in-flight, or the amount they said they would consume (1.6 drinks, SD = 1). Three of these respondents reported that they would consume more than a drink per hour of flight.

We collected height and weight data to estimate26 crude blood alcohol levels for our respondents. These calculations suggest that three respondents would achieve a blood alcohol level that exceeded 0.08 (ie, the legal limit for operating a motor vehicle in the United States) over the course of their flight (factoring in drinks that passengers reported consuming just before boarding). One of these passengers said that he planned to drive home from his destination airport.

Validating Our Outcome of Interest

Our attempts to verify whether reported intentions to consume alcohol in-flight were well correlated with actual in-flight consumption met with mixed results. Of the 281 passengers who received gift checks requesting that they participate in a brief telephone survey after deplaning, only 27% ( n = 77) complied. A comparison of their preflight intentions to drink and their postflight reports, however, showed 91% agreement, which is significantly higher than would have been expected by chance ( κ = 0.314,z = 2.73,p < .01).

More objectively, we were able to purchase data on the number of drinks sold to coach passengers for 18 of our 24 study flights. These figures correlated with the number of intended drink purchases reported by study participants ( r = 0.59,p < 0.01). We also found significant correlations within two of three categories of alcoholic beverages. When we calculated the correlation between drinks sold per economy passenger and drinks reported per economy survey respondent, our correlation was 0.83 ( p < 0.01). We were encouraged by these results, because our study volunteers represented only a subset of the passengers who actually flew and were eligible to purchase alcohol on-board.

Finally, the dose response curve observed between how much alcohol respondents said they usually drink, and whether they intended to drink on their flight, also suggested that passengers were being candid.

Discussion

Some of the factors that have been associated with alcohol consumption in nonaviation contexts (e.g. prior history of alcohol consumption, product pricing) were also found to operate among this sample of airline passengers. Other significant findings may appear unique to this setting (eg, perceived association with jet lag), but most are consistent with previously recognized patterns of association (ie, positive alcohol expectations, social norms).

Some variables that usually show association with alcohol intake—such as gender—dropped out of our model. Although it is possible that this pattern is the result of previously identified phenomena (eg, gender convergence operating among this upper socioeconomic status sample), it may also derive from the aviation setting. Previous studies have shown that drinking context can have important effects on alcohol intake,27 and that subgroups of alcohol users prefer to drink in different venues.28,29

The fact that alcohol dependence was not associated with increased reporting of intention to drink in-flight may be attributable to our measurement tool. CAGE (cut down, annoyed, guilty and eye opener) has proven to be more valid when applied to clinical (vs. general) populations and is not recommended as a screening test for “heavy or hazardous drinking.”30,31 Fear of flying and nicotine addiction were also nonsignificant, which is interesting because there has been speculation that airline passengers might self-medicate with alcohol in response to such stressors.

Although airline passengers' alcohol use has not been studied in the past, our findings share at least two results with the nascent air rage literature. A study of major Canadian carriers reported that most events occur on flights that exceed 3 hours in length.1 In both Canada and the United Kingdom, it has been reported that passengers who were traveling with others were more likely to become involved in disruptive incidents.1,32

Limitations

This study has a number of limitations. It relies upon self-reports of intentions to consume alcohol. The alcohol literature contains numerous reports which demonstrate that intentions to behave are well associated with subsequent behavior. Such associations are particularly likely when the actions under study follow reported intentions closely in time,33 when they are measured in the same referent context,34 and when the behavior is under the respondent's control.

Our data were collected at one airport in the United States, from one airline's passengers. We did not include morning flights in this study nor did we collect data during the spring or summer. These constraints call the generalizability of our findings into question. When our sample is compared to the flying public nationwide, they are similar with regard to travel purpose and age distribution but different with regard to traveling companions, household income, and education.35 For these reasons, it is advisable to focus on the associations identified in this study rather than our prevalence estimates.

Those associations suggest that many of the major covariates of alcohol use on commercial aircraft are modifiable (eg, alcohol expectancies and the amount of time that passengers have access to beverage service in flight). Future studies should attempt to measure alcohol consumption directly, expand to multiple airports and airlines, and collect data year-round.

Conclusions

If the results of future work support the findings of this study, we might conclude that problematic drinking in the aviation context is relatively rare. This is consistent with studies of air rage, one of which estimated that 1 in every 138,308 airline passengers is involved in a passenger misconduct incident.1 A second investigation concluded that one serious on-board incident occurs for every 2 million passengers carried.3

In the United States alone, however, 650 million people fly each year and that number is expected to increase. Even small proportions of such a large population should not be ignored.

Acknowledgments

This study was funded by the National Institute on Alcohol Abuse and Alcoholism Grant No: AA013428-01A2. The authors thank Professor Susan Baker and Dr Norman Giesbrecht for reviewing this article and providing methodologic guidance to our study overall. Suman Sharma oversaw our data collection and database creation. Dorothy Becher supported our data analysis and results reporting. This article represents the views of its authors, who are not speaking for any government agency.

Declaration of Interests

The authors state they have no conflicts of interest to declare.

Ancillary