|Home | About | Journals | Submit | Contact Us | Français|
In the absence of comprehensive smoking bans in public places, bars and nightclubs have the highest concentrations of secondhand tobacco smoke, posing a serious health risk for workers in these venues.
To assess exposure of bar and nightclub employees to secondhand smoke, including non-smoking and smoking employees.
Between 2007 and 2009, we recruited approximately 10 venues per city and up to 5 employees per venue in 24 cities in the Americas, Eastern Europe, Asia and Africa. Air nicotine concentrations were measured for 7 days in 238 venues. To evaluate personal exposure to secondhand smoke, hair nicotine concentrations were also measured for 625 non-smoking and 311 smoking employees (N=936).
Median (interquartile range [IQR]) air nicotine concentrations were 3.5 (1.5, 8.5) µg/m3 and 0.2 (0.1, 0.7) µg/m3 in smoking and smoke-free venues, respectively. Median (IQR) hair nicotine concentrations were 6.0 (1.6, 16.0) ng/mg and 1.7 (0.5, 5.5) ng/mg in smoking and non-smoking employees, respectively. After adjustment for age, sex, education, living with a smoker, hair treatment and region, a 2-fold increase in air nicotine concentrations was associated with a 30% (95% confidence interval 23%, 38%) increase in hair nicotine concentrations in non-smoking employees and with a 10% (2%, 19%) increase in smoking employees.
Occupational exposure to secondhand smoke, assessed by air nicotine, resulted in elevated concentrations of hair nicotine among non-smoking and smoking bar and nightclub employees. The high levels of airborne nicotine found in bars and nightclubs and the contribution of this exposure to employee hair nicotine concentrations support the need for legislation measures that ensure complete protection from secondhand smoke in these venues.
Secondhand tobacco smoke exposure is a major cause of respiratory, cardiovascular and cancer morbidity and mortality around the world.1, 2 As of 2004, over 10 million disability-adjusted life years, 0.7% of total worldwide burden of disease, were lost due to secondhand smoke exposure.2 Of all public places, bars and nightclubs have the highest air levels of secondhand smoke,3–5 posing a serious health risk for employees spending long hours in their work environment.6 Epidemiological studies have shown that hospitality employees have 50 to 60% greater risk of lung cancer compared to other populations.7, 8 To protect all people including workers from exposure to secondhand smoke, Article 8 of the World Health Organization (WHO) Framework Convention on Tobacco Control (FCTC) calls for comprehensive smoke-free legislation eliminating tobacco smoking in all indoor public places and workplaces.9, 10 As of 2011, however, only 11% of the world’s population was protected by smoke-free policies that included bars, restaurants and nightclubs.11 These venues have largely been excluded from smoke-free workplace and public place legislation due to the tobacco industry’s influence on the hospitality sector to oppose smoking regulations.12, 13 In this study, we assessed secondhand smoke exposure in bar and nightclub employees from large cities around the world. To assess personal exposure to secondhand smoke, we evaluated the relationship of workplace air nicotine concentrations with hair nicotine concentrations, a biomarker of internal dose, in both smoking and non-smoking employees.
This study is part of a multi-city effort designed to assess tobacco control measures in bars, cafes/tea houses and nightclubs around the world.14 By design, the study recruited approximately 10 venues per city and up to 5 employees per venue in 24 large cities in the Americas, Eastern Europe, Asia and Africa. Study sites, located in countries with a wide range of smoking legislation in public places (Appendix 1), were selected based on previous collaborations3, 15 and the presence of a within-country study coordinator and team with experience in tobacco control research. Training of the study coordinators was conducted centrally, in person or online following a training manual, simulated interviews and question-by-question guidelines. Each study coordinator was responsible for the training of the fieldworkers.
A total of 238 venues were recruited between January 2007 (Baltimore) and September 2009 (Ulaanbaatar, Mongolia), ranging from 6 venues in Kremenchug, Ukraine to 11 venues in Baltimore, USA and Mexico City, Mexico (Appendix 1). In each city, bars, cafes/tea houses or nightclubs were recruited from 2 to 3 neighbourhoods with a high concentration of public places where people, especially young adults, spend time or gather socially. Venues were selected from popular areas, covering different socioeconomic sectors and neighbourhoods using a door-to-door sampling strategy, except in Manila where venues were randomly selected from rosters by public health inspectors. If voluntary smoke-free venues existed in the city, at least 2 of the selected venues were required to be smoke-free. If smoke-free legislation existed in the city or country, all venues were supposedly smoke-free (Appendix 1).
The minimal requirements for a venue to be in the study included owner agreement and that at least one smoking and one non-smoking employee were willing to provide a hair sample. Venue median response rate was 59% and ranged from 8% in Bishkek, Kyrgyzstan to 91% in Bangkok, Thailand (Appendix 1). In Buenos Aires, Argentina only smoke-free establishments agreed to participate. Informed consent was required for the participating venue owners/managers and employees. The goal was to recruit at least one smoking and two non-smoking employees (up to 5 employees per venue). A total of 936 employees (625 non-smokers and 311 smokers) were recruited, ranging from 20 in Buenos Aires, Argentina to 76 in St. Louis, USA. The study protocol and consent forms were approved by the Institutional Review Board of the Johns Hopkins Bloomberg School of Public Health and by a local ethics committee in each participating city. All participants provided written informed consent.
The bar owner/manager and the employees completed standardized questionnaires administered in the local language of the country by trained interviewers during work hours, but before the venues were open to the public. The employees were asked to provide information on demographics, smoking behaviour, exposure to secondhand smoke at work, home and other places, and opinions about secondhand smoke legislation. Employees were classified as non-smokers if they had not smoked a cigarette in the past 12 months and as smokers if they had smoked in the past 12 months. To confirm the smoking or non-smoking status of the employees, an extra question was asked before the hair sample collection to enquire about smoking (even a single puff) in the last 30 days. The bar owners/managers were asked to describe general characteristics of the venue, including number of employees, hours of operation, occupancy, ventilation systems and smoking policy (smoke-free or smoking allowed).
Time-weighted average air nicotine concentrations in each venue were measured for one week using passive samplers originally developed by Hammond and Leaderer.16 Samplers comprised a filter treated with sodium bisulphate, placed in 37 mm sampling cassette, and covered with a porous diffusion membrane. Two monitors were placed in each bar/nightclub with locations selected to represent areas of the venue where employees most frequently worked. A total of 9 samplers were lost or stolen during the fieldwork. At the end of the sampling period, the remaining 467 samplers were securely closed and shipped to the Exposure Assessment Laboratory of the Institute for Global Tobacco Control at the Johns Hopkins Bloomberg School of Public Health where the nicotine collected by each sampler was extracted and analyzed using gas chromatography with nitrogen-selective detection. The airborne concentration of nicotine was estimated by dividing the amount of nicotine collected by the filter (µg) per volume of air sampled (m3). The volume of air sampled is equal to the total of sampling time in minutes multiplied by the flow rate (25 mL/min).
Air nicotine concentrations from samplers placed in the same venue were comparable and concentrations are presented as the average of the two air nicotine samplers in each venue. For quality control purposes, 10% of samplers were duplicates and/or blanks. The intra-class correlation coefficient between duplicate samples was 0.94. Blanks were used to determine the blank-corrected nicotine concentrations and to calculate the nicotine limit of detection (range 0.002– 0.009 µg/m3). A total of 6 samples had air nicotine concentrations below the limit of detection. For samples below the limit of detection, a value of half the limit of detection was assigned for statistical analyses.
Hair samples from the employees were collected on the day the nicotine samplers were installed. A small hair sample (~30– 50 strands) was obtained near the hair root from the back of the scalp where there is the most uniform growth pattern between individuals. Hair samples were placed in labelled sealed plastic bags and shipped to the Exposure Assessment Laboratory of the Institute for Global Tobacco Control at the Johns Hopkins Bloomberg School of Public Health. Up to 3 cm of hair from the scalp, thoroughly cleaned to remove any nicotine in the outside of the hair, was used to evaluate secondhand smoke exposure during the most recent months. Hair nicotine was measured by gas chromatography mass spectrometry (GC/MS) following the method described by Kim et al.17, 18 Hair nicotine concentrations were calculated by dividing the amount of nicotine measured in each hair sample (ng) by the mass of hair analyzed (g). The limit of detection was 0.02 ng/mg for a 30 mg hair sample and 87 samples were below the limit of detection and were assigned a value of half the detection limit. The percent coefficient of variation ranged from 12% to 20% for higher (approximately 3 ng/mg) and low (approximately 0.5 ng/mg) hair nicotine concentrations, respectively. For quality control purposes, duplicate analysis of 10% of hair samples were analyzed by the laboratory. The intra-class correlation coefficient between duplicate samples was 0.97.
Descriptive analyses were stratified by region (Americas, Eastern Europe, Asia, Africa) and/or employee smoking/non-smoking status. Distributions of air and hair nicotine concentrations were described using the median and interquartile range. We used crude and multivariable adjusted mixed-effect linear models with city-specific intercepts to evaluate the dose-response relationship between air nicotine and hair nicotine concentrations. Hair nicotine concentrations (independent variable) was log-transformed to improve normality. Air nicotine was modelled using different strategies to evaluate the shape of the dose-response using tertiles, log2-transformed, and the original scale. In tertile models, we computed ratios (95% confidence interval) of the geometric mean of hair nicotine concentrations comparing tertiles 2 and 3 to the lowest tertile of air nicotine. In log2-transformed models, we evaluated the ratio of the geometric mean of hair nicotine concentration with a doubling in air nicotine concentrations. In models with air nicotine concentrations in the original scale, we evaluated the ratio of the geometric mean of hair nicotine with a change in 1 µg/m3 of air nicotine concentrations. Multivariable models for the association between air and hair nicotine concentrations were adjusted for age, sex, education, living with a smoker, hair treatment and region. Models for non-smokers were further adjusted for never/former smoking status. For smokers, further adjustment for number of cigarettes smoked per day did not change estimates (not shown). Multiple imputation,19 assuming data was missing at random, was used to impute values for employees missing data on age (N=2), education (N=5), living with a smoker (N=13), hair treatment (N=16) and number of cigarettes smoked per day (N=56). To assess differences in the association of air and hair nicotine concentrations across employee characteristics, we also estimated the ratio of hair nicotine concentrations by air nicotine concentrations for subgroups defined by sex, age, region, education, living with a smoker and hair treatment. Analyses were conducted using Stata version 11.1 (Stata corporation, Texas, USA). The statistical significance level was set at α = .05. All statistical analyses were 2-sided.
Among the 238 venues included in the study, 18% were smoke-free as reported by the owner/manager (Table 1, Appendix 1). The median maximum legal occupancy ranged from 80 in venues recruited in Eastern European cities to 150 in venues recruited in American cities. The median number of employees per venue was 15 with small differences across regions. For descriptive purposes, 61%, 41% and 43% of the venues served a full menu, had dance space and offered live music, respectively; 68% sold cigarettes at the bar counter or from a vending machine and 29% reported receiving promotional items from tobacco companies.
Fifty percent of the participants were male, the mean age was 29 years old, 37% had a college-level education and 62% were bartenders or waiters (Table 2, Appendix 1). The median (interquartile range) hours worked per week was 48 (40, 60). Among non-smoking employees, 11% were former smokers. Forty-six percent (43% and 52% of non-smoking and smoking employees, respectively) reported living with a smoker. Fifty-seven percent (64% non-smokers and 44% smokers) supported smoke-free policies and 78% (87% non-smokers and 61% smokers) reported they would prefer to work in a smoke-free environment.
Median (interquartile range) air nicotine concentrations were 3.5 (1.5, 8.5) µg/m3 and 0.2 (0.1, 0.7) µg/m3 in smoking and smoke-free venues, respectively (Table 1, Appendix 2). Median (interquartile range) hair nicotine concentrations were 6.0 (1.6, 16.0) ng/mg and 1.7 (0.5, 5.5) ng/mg in smoking employees and non-smoking, respectively (Table 2, Appendix 2). Median (interquartile range) hair nicotine concentrations for all employees working in smoking and smoke-free venues were 2.7 (0.8, 9.2) ng/mg and 1.3 (0.3, 6.5) ng/mg, respectively.
After adjustment for age, sex, education, living with a smoker, hair treatment, region (and former smoking status for non-smokers), a 2-fold increase in air nicotine concentrations was associated with a 30% (95% confidence interval 23%, 38%) and a 10% (2%, 19%) increase in hair nicotine concentrations in non-smoking and smoking employees, respectively (Table 3). For non-smokers, the hair nicotine concentrations were 2.54 (95% CI: 1.91, 3.39) and 3.77 (95% CI: 2.62, 5.42) higher for tertiles 2 and 3 of air nicotine concentrations compared to tertile 1 (p-value for trend <0.001). The corresponding ratios comparing tertiles 2 and 3 to the lowest tertile were 1.55 (95% CI: 1.03, 2.34) and 1.53 (95% CI: 0.95, 2.46), respectively among smokers (p-value for trend= 0.02). For each 1 µg/m3 increase in air nicotine concentrations, hair nicotine concentrations increased 5% (95%CI: 3%, 8%) and 3% (95% CI: 1%, 6%), respectively, for non-smoking and smoking employees. After multivariable adjustment, hair nicotine concentrations were higher among men and among employees without chemical hair treatment in the past month (Table 3). Concentrations were also significantly lower in smoking employees in Eastern Europe compared to other regions.
The association between air nicotine and hair nicotine concentrations was similar across participant characteristics except for evidence of a stronger relationship among smoking employees without chemically treated hair compared to those with chemical treatment (p-value for interaction =0.05) (Figure 1).
Exposure to secondhand tobacco smoke in the workplace, assessed by air nicotine, resulted in elevated concentrations of hair nicotine among both non-smoking and smoking employees. Our study extends previous findings on elevated hair nicotine concentrations in non-smoking hospitality employees20–24 using a multi-city approach. In addition, we found that exposure to secondhand smoke is a relevant occupational hazard for both non-smoking and smoking employees. Comprehensive smoke-free legislation is needed to protect hospitality employees from involuntary exposure to tobacco smoke at work.
The negative health impact of exposure to secondhand smoke well established.25 Levels of air nicotine and indoor fine particulate matter have been shown to be much greater in bars, restaurants and nightclubs compared to other public places.3, 5, 26, 27 The health effects of exposure to secondhand smoke among this occupational group have also been investigated. In 1993, a review of six studies that examined the risk of lung cancer among bar and restaurant employees controlling for active smoking concluded that there is approximately 50% (range: 10% to 90%) increased lung cancer risk among these employees compared to the general population.7 The increased lung cancer risk was attributed to their higher exposure to secondhand smoke. In the United Kingdom, among 617 lung cancer, ischemic heart disease or stroke deaths attributable to secondhand smoke in 2003, 54 deaths were among long-term employees of the hospitality industry and almost half of these deaths were among pub, bar and nightclub employees despite that smaller size of this sector of the workforce.28 In New Zealand, a study of 435 bar employees found that among those exposed to secondhand smoke at work, 53% reported lung or throat irritation and 73% wanted to restrict smoking in bars.29 Non-smokers exposed to secondhand smoke at work have more illness-related absenteeism than non-smokers without work exposure.30 Increased hair nicotine concentration among non-smoking bar and restaurant workers has also been associated with greater number of behavioural symptoms of nicotine dependence.31 Finally, reducing secondhand smoke exposure in bars and restaurants has been associated with decreased hair nicotine21 and salivary32–35 and urine36 cotinine concentrations in employees and with decreased respiratory symptoms in studies from Norway37, Sweden38 and the United States (Kentucky and California).39, 40
Hair nicotine is a reliable and valid biomarker of secondhand smoke exposure with each centimetre of hair reflecting about 1 month of cumulative tobacco smoke exposure.41–43 In our study, air nicotine was a major determinant of hair nicotine concentrations both in non-smokers and smokers, although the association was stronger among non-smokers. Other determinants of hair nicotine included sex, with higher concentrations among men, chemical hair treatments including colouring, bleaching and perming in the past month, and region, with lower concentrations in participants from Eastern Europe. The lower concentrations of hair nicotine among participants with chemical hair treatment are consistent with previous findings,44, 45 including studies among hospitality employees from New Zealand21 and among women participating in a large multi-city study evaluating secondhand smoke exposure in women and children around the world.15 Lower hair nicotine concentrations in women from Eastern Europe versus other regions were also found in that large study.15 Employees from Eastern Europe were more likely to be female and to use more hair treatment but lower hair nicotine concentrations among Eastern Europe employees could be also be related to differences in hair nicotine uptake by hair colour or other characteristics, or by differences in nicotine metabolism. Hair colour and nicotine metabolism could also play a role in explaining the stronger association between air nicotine and hair nicotine among non-smoking employees from Africa (Figure 1).
Strengths of the study include the objective measures of secondhand smoke exposure, the multi-city study design and the standardized protocol. A few limitations should be taken into account. First, in most cities, venues and employees were selected by convenience sampling; therefore results may not be representative of secondhand smoke exposure in a particular city. The participating venues, however, were located in areas of the cities with a high concentration of places where people gather socially. Moreover, the goal of the study was not to estimate the prevalence of secondhand smoke exposure, but to evaluate the contribution of secondhand smoke exposure in the workplace to hair nicotine, a biomarker of internal dose. Although hair nicotine concentrations reflect cumulative exposure to tobacco smoke in the past months, air nicotine concentrations were only collected for 7 days. Potential differences in levels of exposure to secondhand smoke during those days could have resulted in underestimation of the relationship between air and hair nicotine concentrations. Also, while we observed lower concentrations of hair nicotine among participants with chemical hair treatment, we are unable to differentiate between different types of chemical treatment. Finally, the response rate was low in some countries, although sensitivity analyses excluding cities with low-response rates yielded similar results (data not shown).
This multi-city study confirms that secondhand smoke exposure remains an important occupational hazard for non-smoking and smoking employees in bars, cafes and nightclubs in the absence of comprehensive smoke-free legislations. At the time of the study only 18% of venues were smoke-free and only one participating city (Montevideo, Uruguay) had comprehensive smoke-free legislation which prohibited smoking in all public places including bars and nightclubs. Since the completion of the study comprehensive policies have been introduced in Guatemala; Mexico City, Mexico; Baltimore, USA; Poland and Shanghai, China (Appendix 1). The WHO’s recent report on the Global Tobacco Epidemic found that since 2008 the number of people protected by comprehensive smoke-free laws has increased more than 385 million, representing a 6% increase of the world population that is protected.11 Article 8 of the WHO Framework Convention for Tobacco Control (FCTC) mandates participating nations to implement policies to prevent exposure to tobacco smoke in indoor workplaces, public transport and indoor public places and workplaces including restaurants, bars and nightclubs.9, 10 Many of these countries are now approaching their sixth year of implementation and are thus required to adopt comprehensive legislation to protect individuals from exposure to secondhand smoke in all indoor workplaces and indoor public places. With the exceptions of Argentina, Armenia, Guyana, Indonesia and the United States, all countries in our study had ratified the FCTC at the time of the study.
In conclusion, the high levels of airborne nicotine found in bars and nightclubs and the contribution of this exposure to employee hair nicotine concentrations support the need for legislation that regulates smoking in these environments and provides complete protection from secondhand smoke for all employees. This is an opportune moment for countries to honour their commitments under the FCTC and expand the number of people protected from secondhand smoke worldwide. In countries with comprehensive legislation, efforts are needed to ensure complete enforcement. By eliminating secondhand tobacco smoke in socializing and hospitality venues, smoke-free legislations can reduce the burden of disease related to secondhand smoke exposure.
The FAMRI Bar Study Investigators include Wifred Agbenyikey (Health Research Unit, Accra, Ghana), Mira B Aghi and Mitali Chakrabarty (Taleem Research Foundation, Guajarat, India), Marta Angueira (Union Antitabaquica Argentina, Buenos Aires, Argentina), George Bakhturidze (Tobacco Control Alliance in Georgia, Tbilisi, Georgia), Carmen Barco (CEDRO, Lima, Peru), Joaquin Barnoya (Cardiovascular Unit of Guatemala and Washington University in St. Louis, USA), Marcia Bassier-Paltoo (Ministry of Health, Georgetown, Guyana), Jamil H. Chowdhury (RTM International, Dhaka, Bangladesh), Rita Damayanti (Center for Health Research University, Indonesia), Beatriz Goja (Facultad de Medicina, Montevideo, Uruguay), Nipapun Kungskulniti and Punyarat Lapvongwatana (Mahidol University, Bangkok, Thailand), Vladimir Levshin (Russian Cancer Research Center, Moscow, Russia), Chimedsuren Ochir (School of Public Health, Ulaanbaatar, Mongolia), Oluwakemi Odukoya (Lagos University Teaching Hospital, Lagos, Nigeria), Zheng Pin-Pin (Fudan University, Shanghai, China), Krzysztof Przewozniak (Health Promotion Foundation, Warsaw, Poland), Luz Myriam Reynales-Shigematsu and Tonatiuh Barrientos-GutiÈrrez (Instituto Nacional Salud Publica, Mexico City, Mexico), Rachel Rowena Garcia and Lia Losonczy (Center for Health Development, Manila, Philippines), Arayik Sargsyan (American University of Armenia, Yerevan, Armenia), Dao Thanh Huyen (Center for Community Health Strategy, Hanoi, Vietnam), Denis Vinnikov (Public Association “Lung Health”, Bishkek, Kyrgyzstan) and Kyryll Zhyvotovskyy (European Choice, Kremenchug, Ukraine).
Funding: This project was supported by a Clinical Investigator Award from the Flight Attendant Medical Research Institute (FAMRI). Miranda R Jones and Ana Navas-Acien were also supported by the US National Cancer Institute (R03CA153959). Miranda R Jones was supported by the Cardiovascular Epidemiology Institutional Training from the National Heart, Lung and Blood Institute (T32HL007024).
Contributorship statement: ANA, JMS and PNB had the idea for the study. ANA, HW and SS directed the fieldwork. The FAMRI Bar Study Investigators directed the fieldwork within their country. ANA, EAT and MRJ prepared the study database and planned statistical analyses. PB is responsible for the air nicotine and hair laboratory analysis, quality control and assurance and interpretation of air and hair nicotine data. MRJ and ANA analysed the data and drafted the manuscript. All authors participated in the interpretation of the results and contributed to the writing of the manuscript.
Competing interests: None.
Data Sharing Statement: Data from this study are available for reanalysis and for analysis of additional research questions through contact with the study authors.