Paper -Published On SAGE: When Hotel Guests Complain About Tobacco, Electronic Cigarettes, and Cannabis: Lessons for Implementing Smoking Bans

The study population of interest was hotels listed by TripAdvisor in the US, the most popular travel-related website featuring user-generated content with little evidence for false content.^17–20 To equally represent states across the range of smoking prevalence in the US, we selected two states in each of the five smoking prevalence categories designated by the Centers for Disease Control and Prevention.²¹ The prevalence categories and selected states were: 9% to < 12.24%, California and Utah; 12.24% to < 15.48%, Texas and New York; 15.48% to < 18.72%, Oregon and Nebraska; 18.72% to < 21.96%, Michigan and Alabama; and 21.96% to 25.2%, Arkansas and Kentucky. Within each state, we then selected the largest city by population: Los Angeles, CA, Salt Lake City, UT, Houston, TX, New York City, NY, Portland, OR, Omaha, NE, Detroit, MI, Birmingham, AL, Little Rock City, AR, and Louisville, KY. For each of the ten cities, a list of all hotels represented on TripAdvisor.com was generated and organized by traveler ranking. We then randomly selected from each city’s list 50 hotels with at least 50 reviews for a total sample size of N = 500 hotels. Although all hotels had ≥50 review ratings, six hotels did not have written comments and were excluded from analyses of TEC complaints. In addition, room cost per night was unavailable for 16 hotels, and no star ratings were given for seven hotels, resulting in a total sample size of 477 hotels with complete data sets.

Hotel websites and TripAdvisor comments were reviewed from November 2018 to February 2019 using the web scraper ParseHub.²⁰ Altogether, hotel guests left 332,277 comments.

Descriptive analyses were conducted to report major characteristics of the hotels and their TripAdvisor reviews, identify influential data points, and scrutinize variable distributions. We report the mean numbers of adjusted complaints per 100 reviews, their 95% confidence intervals, as well Tukey’s five-number summary (Minimum, 1^st Quartile, Median, 3^rd Quartile, Maximum).²⁸ Hypotheses about predictors of adjusted TEC complaints were tested based on a multivariable negative binomial regression model with hotel location, hotel star rating, hotel brand, hotel capacity, number of hotel rooms, hotel room cost, and hotel smoking policy as covariates. The negative binomial regression models initially included an adjustment for the clustering of hotels within cities, but this was found to be unnecessary due to an intraclass correlation of 0. Categorical variables were dummy-coded. All analyses were conducted using Stata version 17.²⁹ The Type I error rate was set at an alpha = .05.

Table 1 shows the locations and major characteristics of hotels. According to their own websites accessed between November 2018 and February 2019, 63% of hotels were 100% smokefree; according to TripAdvisor, 96% were smokefree. Of the 465 hotels identified by TripAdvisor as smokefree, 298 hotels (64%) provided information on their website confirming the smokefree designation, 36 (8%) indicated that nonsmoking guest rooms are available, and 131 (28%) hotels provided no information about their smoking policy. Hotel capacity ranged from 11 to 1705 (Mdn = 115) guest rooms, and the cost for a standard room ranged from $41 to $650 (Mdn = $95).

Table 2 shows the median total number of reviews was 374 (Interquartile Range – IQR: 164 to 818). This translates to a median of 43 reviews per year (IQR:19 to 102) or a median of 34 reviews per 100 guest rooms per year (IQR: 19 to 71). TripAdvisor reviews covered periods ranging from .6 to 16.7 years before our assessment.

The analysis of guest reviews for TEC-related complaints (see Table 2) showed an average of 9.6 complaints per hotel (95% Confidence Interval: [6.9, 10.1]). Per 100 guest reviews, the mean number of TEC-related complaints was 2.3 (95% CI: [2.2;2.9]). Closer inspection showed the distribution of complaints was heavily skewed, with a median of .9 and an IQR of .2 to 3.4 complaints per 100 reviewed. The largest proportion of TEC-related complaints concerned THS residue (M = 2.08; 95% CI: [1.78;2.38]), and only a small fraction was attributable to SHS (M = .23; 95% CI: [.14;0.32]). Table S2 in the supplemental material gives examples of SHS and THS-related reviews. None of the guests registered a complaint specifically about electronic cigarettes. Cannabis-related complaints were reported at an average of .24 per 100 reviews (95% CI: .19;0.30). For comparison, Table 2 also shows dirtiness and noise complaints based on a subsample of N = 99 and N = 149 hotels, respectively. For every TEC-related complaint, hotels received, on average, four noise and four dirtiness complaints. TEC-related complaints were weakly and positively correlated with noise (Spearman rho = .271, P < .01) and dirtiness complaints (Spearman rho = .273, P < .01).

Figure 1 shows the rank-ordered distribution of TEC-related complaints per 100 reviews showing that 17.6% had no TEC-related complaints, and 54.3% of hotels had ≤1 TEC complaint per 100 reviews. In contrast, about 1 in 4 hotels (24.5%) had three or more TEC-related complaints, with a maximum of 26.5 TEC-related complaints per 100 reviews.

Negative binomial regression models (see Table 3) showed that hotel smoking status as reported on the hotel website, guest review rate, hotel star rating, room cost, hotel brand, and hotel location were independently and significantly associated with TEC complaints (overall model fit statistics χ²(24) = 288.38, P < .001). In contrast, hotel smoking policy as reported by TripAdvisor (P = .605), hotel capacity (P = .767), review interval (P = .691), and state’s smoking prevalence (P = .142) were not associated with TEC complaints (Table S4 in online supplement for additional model information). Tables S5 to S8 in the online supplement provide the estimated marginal means and 95% confidence intervals based on the model presented in Table 3.

Controlling for all other variables, lower-star hotels (≤2 star ratings) had significantly more TEC complaints (Mean = 3.4) than higher-star hotels (2.5 to 4 star ratings; Means: 1.3-2.1; χ²(1) = 15.58, P < .001). Hotels with 2.5 to 5 star ratings did not differ from each other. Figure 1 illustrates this association showing that hotels with ≥2.5 stars (blue) tended to have fewer TEC complaints than hotels with ≤2 stars. Hotels with higher-price standard rooms had fewer complaints. For a $20 increase in room cost per night, TEC complaints were reduced by 13% (z = −3.56, P < .001). Hotels with more guest reviews per 100 rooms had fewer TEC complaints. A 10-point increase in the number of guest reviews per 100 rooms was associated with an 11% reduction in TEC complaints independent of other variables(z = 6.65, P < .001). Compared to nonsmoking hotels (Mean = 1.91), those offering designated smoking rooms (Mean = 2.57) and those providing no information about their status (Mean = 2.44) had 35% (P = .046) and 27% (P = .039) higher TEC complaints, respectively. Among the hotel brands, Marriott and Hilton had the overall lowest mean TEC complaints (.94 and .98, respectively). This is in contrast to Wyndham (3.36), Choice (3.14), other brands (Mean = 2.34), and independent hotels (Mean = 2.25). That is, Wyndham and Choice hotels had more than three times as many TEC complaints per 100 reviews than Marriott and Hilton. Among the ten different locations, Salt Lake City, UT, and Omaha, NE, had the lowest mean TEC complaints, with 1.64 and 1.71, respectively. In contrast, Los Angeles, CA, Detroit, MI, and Birmingham, AL, had the highest means with 3.11 (P = .01), 2.84 (P = .007), and 2.54 (.028), respectively, controlling for other variables.

Based on the existing scientific evidence on SHS and THS in multiunit buildings and findings about guest complaints from the present study, restricting smoking to particular guest rooms or areas within the hotel is an ineffective policy and fails to protect guests staying in nonsmoking rooms of the same building from SHS and THS exposure.^5,9 Fifteen years after the first hotel chains adopted 100% smokefree policies, it is misleading for hotels to suggest nonsmoking rooms offer a smokefree environment when in the same building smoking is allowed in other rooms. When deciding which hotel to stay in, hotel guests rely on accurate information provided on a hotel website to make reservations for nonsmoking rooms expecting protection from SHS and THS exposure. Our findings suggest that many hotels claim to offer a hotel stay free of SHS and THS when in fact, their policies cannot assure this level of protection to their guests. To remove this information asymmetry, hotels offering dedicated smoking rooms should present themselves as smoker hotels and caution their guests that they cannot protect nonsmokers from SHS and THS exposure in nonsmoking rooms. Not only will this allow guests to make informed decisions, but it will also reward hotels that do provide a 100% smokefree-building hotel.

Our findings show, however, that declaring a 100% smokefree building policy or a state-wide hotel smoking ban itself does not automatically protect nonsmokers from exposure to SHS and THS. The substantial differences between hotel brands in TEC-related guest complaints independent of their smoking policies demonstrate that the implementation of such policies matters. Implementation starts with disclosing on their website and other public relations material that a hotel follows and enforces strict 100% smokefree building policies. Effective implementation also requires hotel staff training, signage, and reminders at check-in, hallways, balconies, conference rooms, staircases, elevators, and guest rooms. In addition, implementation includes consistent enforcement of policies and practices. Finally, implementation requires listening and responding to guest feedback, such as online reviews to identify and remedy discrepancies between how a hotel presents and how guests experience a hotel’s smoking policy.

With 80% of TEC complaints attributable to THS residue left behind by previous guests, hotels must pay attention to two separate issues. Hotels have to be deliberate and persistent in implementing smoking bans to avoid the new accumulation of THS. In addition, hotels have to address existing THS reservoirs left behind from years of permissive smoking policies. THS is highly persistent and will not disappear by itself, so a single THS-polluted room will cause repeated complaints. Even though guests may complain about the offensive odor, the underlying THS reservoir will not be reduced by using fragrances or other chemical approaches to cover up or remove the odor (eg, ozonation). Instead., hotels must identify, clean, and remove the reservoirs where THS pollutants are stored, such as carpets, beds, mattresses, furniture, and wallboard, to prevent future odor complaints caused by re-emission from THS reservoirs.

Hotel guest complaints about tobacco, electronic cigarettes, and cannabis should be interpreted with caution. The number of TEC-related complaints we recorded is likely to be an undercount of the actual number of TEC-related incidences experienced by hotel guests because only a fraction of such incidences is likely to be reported, and only a fraction of those are provided in writing as guest reviews on TripAdvisor. It is also possible that guests may have misinterpreted odors as coming from tobacco, cannabis, and e-cigarette products or failed to attribute unpleasant odors to tobacco and cannabis products. Irrespective of a potential undercount, the number of TEC complaints per 100 reviews provides a valuable metric describing the relative frequency of TEC complaints compared to the total number of guest reviews and sheds light on differences between hotels and the factors affecting such complaints. While our study relies on a random sample of hotels in the largest cities of 10 states with 50 or more guest reviews, our findings cannot be generalized to hotels in smaller cities, rural areas, or hotels generating fewer than 50 guest reviews. The distinct differences in TEC complaints between locations suggest location-specific factors that affect TEC complaints, but these factors remain unknown and should be further explored. This study could not examine changes in TEC complaints over time as hotels may have transitioned to smokefree policies. Hotel and TripAdvisor smoking status used in this study was based on hotel policies from November 2018 to February 2019. The official hotel policies at the time guests made reservations could not be confirmed except for hotels that had publicly declared smokefree policies since 2006 or 2011. However, this does not alter our conclusions as all TEC complaints came from guests reporting reservations for smokefree rooms or in smokefree hotels, that is, hotels offering smokefree accommodations at the time of a hotel stay.

From a consumer’s perspective, hotels share many characteristics with other indoor environments for which comprehensive smoking bans already exist. Like restaurants, bars, airplanes, and public transportation, hotels have a high occupancy turnover and are used by a broad cross-section of the population. While at first it seemed impossible, 100% smokefree policies for these indoor environments are now widely implemented and enjoy high compliance. A uniform, comprehensive building-wide smoking ban for hotels would remove ambiguity and facilitate the implementation of consistent policies across different hotel brands to protect the vast majority of guests seeking a sanctuary from secondhand and thirdhand smoke pollution caused by tobacco, cannabis, and electronic cigarette use. Until such uniform bans are in place, individual hotels should give up the illusion that designated smoking rooms can protect nonsmokers and fully commit to implementing and enforcing a 100% smokefree-building policy.