Recent Contributions of Air- and Biomarkers to the Control of Secondhand Smoke (SHS): A Review

Since the publication of the US Surgeon General Reports in 1996 and 2006 and the report of the California Environmental Protection Agency in 1999, many reports have appeared on the contribution of air and biomarkers to different facets of the secondhand smoke (SHS) issue, which are the targets of this review. These recent studies have allowed earlier epidemiological surveys to be biologically validated, and their plausibility demonstrated, quantified the levels of exposure to SHS before the bans in various environments, showed the deficiencies of mechanical control methods and of partial bans and the frequently correct implementation of the efficient total bans. More stringent regulation remains necessary in the public domain (workplaces, hospitality venues, transport sector, etc.) in many countries. Personal voluntary protection efforts against SHS are also needed in the private domain (homes, private cars). The effects of SHS on the cardiovascular, respiratory and neuropsychic systems, on pregnancy and fertility, on cancers and on SHS genotoxicity are confirmed through experimental human studies and through the relationship between markers and prevalence of disease or of markers of disease risk.


Introduction
The main issues of secondhand smoke (SHS) have been extensively addressed in two successive reports of the US Surgeon General [1,2] and in the report of the California Environmental Protection Agency [3].
Early publications about the health consequences of SHS (also referred to as Environmental Tobacco Smoke [ETS]) were mainly epidemiological surveys through questionnaires for general or specific populations. Their value is limited, as the respondents' answers may (voluntarily or not) be incorrect and the large numbers of factors influencing the prevalence, intensity and duration of SHS exposure cannot be correctly estimated in a questionnaire. On the other hand, measurement of air-and biomarkers of SHS exposure is more quantitative and can complement and validate the results of the questionnaire surveys.
Recent contributions of air-and biomarkers to different facets of the SHS issue are reviewed in this paper. Medline was searched from 2005 with the keywords -SHS‖ and -air‖ or -atmospheric markers‖ and -SHS and biomarkers‖. The full collection of Tobacco Control and Nicotine & Tobacco Research was examined from 2005 onward, and some references from bibliographies were also recorded. The Cochrane Review 2010 [4] was another source of information. To produce an overview compatible with the length of a review article, some publications were summarized and others had to be set aside.
Experimental studies on cells, germs and animals, transplacental foetus exposure during pregnancy and the economic consequences of smoking bans were excluded, as were the health effects of SHS estimated from risk models.

Atmospheric Markers
Among the potential atmospheric markers of SHS, carbon monoxide (CO) is usually excluded due to its poor specificity, although its non-tobacco-specific origin in closed environments is easily detected. In ambient air, its safe limit was fixed at 8.5 ppm in a European Directive [5].
Nicotine is a specific, sensitive and valid marker of atmospheric SHS that is absent in tobacco smoke-free environments. Air nicotine can be used to estimate respirable particulate matter (PM) exposure from SHS in indoor environments if smoking occurs regularly, if the system is in a near quasi-steady state and if the sampling time is longer than the characteristic times for removal processes [6].
Despite a background concentration of respirable particulate matter (PM) due to cooking and infiltration from outdoors, PM is considered as a surrogate marker of SHS, at least at relatively high levels [2]. PM < 10 µm enter the respiratory tract, but are primarily produced by mechanical processes such as construction activities, road dust resuspension and wind. PM < 2.5 µm (fine particles) originate primarily from combustion sources (such as smoking). Their background value has been estimated to be 3-5 µg/m 3 in the US and Western Europe, and the limit of air quality is estimated at 10 µg/m 3 annual mean and 25 µg/m 3 24 h mean. Those limits are those when respectively long versus short term health effects are expected to appear. For ultrafine particles (<0.1 µm), no recommendation can be provided, as the existing body of epidemiological evidence on the exposure-response relationship is insufficient [7].
Mathematical models were developed and applied to existing statistics in order to predict respirable SHS suspended particles (<3.5 µm) and nicotine in air inside the home from building smoke density, air exchange rate, exposure duration, etc., to convert serum cotinine into urine cotinine, and to estimate atmospheric SHS-PM and nicotine from cotinine in vital products. These models, called -Rosetta Stone‖ Equations by Repace, permit broad comparison of clinical and epidemiological studies using respectively atmospheric markers and biomarkers for SHS [8]. For each microgram of atmospheric nicotine, there is an estimated increase of about 10 µg in respirable SHS particle concentration [1].
Fluorescing particulate matter (FPM) and ultraviolet-absorbing particulate matter (UVPM) are also sensitive atmospheric markers [1], although not frequently used. The carcinogenic particulate polycyclic aromatic hydrocarbons (PPAH) are also used for measuring air quality [9,10]. Benzene, another carcinogenic substance of tobacco smoke, is sometimes used as an air marker of SHS (UE security limit for ambient air: 5 mcg/m 3 [5]. For volatile organic compounds and semi-volatile compounds, sorption on and desorption from indoor surfaces influence exposure.

Biomarkers
The characteristics of an ideal SHS biomarker are the following: specificity for tobacco, appropriate half-life, sensitivity and precision, sampling with non-invasive techniques, low cost, association with health effects or with an agent having health effects, quantitative relationship to prior exposure to SHS [1].
CO in exhaled air and blood HbCO are both inadequate biomarkers of exposure to SHS due to poor specificity and a short half-life. Thiocyanate is inadequate due to lack of sensitivity. Nicotine is specific, but its variations due to its 1-3 h half-life limit its value as a marker of exposure. When measured in hair (where it is incorporated into the growing shaft over time), nicotine seems to be a better marker of chronic exposure. Hair nicotine levels were better able to identify infants aged 3-27 months according to smoking in their households (no smoking, smoking only outside the home, smoke inside the home) than urine nicotine (p < 0.0001), and were correlated with the number of smokers [11]. An evident positive association between hair nicotine concentrations in non-smokers and higher numbers of cigarettes smoked per day in a household was demonstrated [12]. Of a total of 1,746 cases from six US, Canada and France databases, the cut-off value of hair nicotine distinguishing active smokers from passive smokers or subjects unexposed to smoke were 0.8 ng/mg for non-pregnant and 0.2 ng/mg for pregnant women. The cut-off value between exposed and unexposed children was 0.2 ng/mg [13].
In 2,485 women participating in the Nurse's Health Study in 1982, toenail nicotine levels differed significantly according to reported smoking status (median level for non-smokers without SHS exposure 0.10 ng/mg; with SHS exposure 0.14 ng/mg; and for active smokers 1.77 ng/mg), but with considerable overlap in nicotine levels between the reported types of smoking status [14]. Its value is thus limited.
Cotinine, the main metabolite of nicotine, is used most frequently as its half-life (±16-18 h) [15] allows levels to remain fairly constant during the day. The cut-off point of 14 ng/mL plasma cotinine Among 183 pregnant Polish women, for example, serum cotinine levels corresponding to active smoking were detected in 17% of self-reported unexposed non-smokers and levels corresponding to SHS exposure in 74% [31].
Dipstick tests (NicAlert and TobacAlert strips) were used in urine and saliva for qualitative assessment of cotinine levels. In urine samples both TobacAlert and NicAlert performed poorly at low nicotine levels, and cannot be used reliably to indicate a specific level of urine cotinine at the low levels found in SHS-exposed people [32]. The specificity of NicAlert in saliva is 95%, its sensitivity 93%, its positive predictive value 95% and its negative predictive value 93% for verifying the smoking status (>10 ng/mL) of self-identified smokers versus non-smokers with gas chromatography-nitrogen phosphorus detection as reference standard. In the readings from all five non-smokers exposed to SHS (among a total of 45 non-smokers), the NicAlert test was always negative. Thus the method did not detect all levels of SHS exposure [33].
Protein and DNA adducts are not often used for the detection of SHS exposure. Due to their half-life of ±40 days, urine NNAL and NNAL glucuronide [NNAL-O-gluc] and [NNAL-n-gluc] (metabolites of the carcinogenic tobacco specific nitrosamine NNK [4-(methylnitrosaminol)-1-(3-pyridyl)-1-butanone] can monitor long periods of exposure and are well correlated with urine cotinine levels. In a review of the literature, NNAL and NNAL-glucuronide were the most consistently elevated biomarkers in people exposed to SHS [34].
In urine samples, the levels of 1-hydroxypyrene-O-glucuronide, the major metabolite of pyrene, a common polycyclic hydrocarbon, correlated (p = 0.04) with the hours of secondhand exposure in non-smokers; poor specificity, however, does not allow them to be used to measure SHS exposure [35]. Other carcinogenic products, benzene and 1-3-butadiene, were also measured in air and in body fluids of non-smokers [36] Biomarkers can allow a valuable quantification of tobacco smoke absorption but can also establish the biological plausibility of the epidemiological observations [37].

Exposure to SHS
Exposure to SHS is indirectly evaluated through studies of air markers or can be measured directly by biomarkers of intake. Both can aid in selection of preventive measures (e.g., partial or total smoking bans) and assess their efficiency and implementation.

Hospitality venues
Before the smoking bans in hospitality venues high concentrations of SHS were demonstrated mostly by large PM 2.5 levels of above the American safety standard of 35 µg/m 3 per 24 h [38]. This was the case for pubs, cafes and bars [39], particularly in pubs frequented by deprived communities versus affluent communities [40], but also and even at higher levels in discotheques and at lower levels in restaurants and restaurant-cars [39].
That ventilation fails to control SHS was demonstrated by the high levels of PM 2.5 and PPHA in pubs despite a per occupant ventilation rate that conformed to the recommendations [10].
Even in outdoor dining areas, but at only one meter of a smoker, the PM 2.5 levels, although below the safety standard, triple during the smoking period. Being under an overhead awning increases the average exposure to PM 2.5 by around 50% in comparison with the background level [41].
The same trends were observed when air nicotine (lower limit of detection 0.01 µg/mL) was detected in a cross-sectional study of hospitality venues in eight European countries: nicotine was detected in 97.5% of the samples. The highest median values were observed in discos/pubs, and the lowest in restaurants. The concentrations were higher in smoking vs. non-smoking areas (ratio 3.12). Most countries had only smoking restrictions or no regulations at all, but in Ireland and Italy, where a smoking ban had been implemented, the nicotine concentrations were much lower [42]. There was a trend towards lower concentrations in this 2008 study in comparison with the earlier 2005 data in hospitality venues [43].
In capital cities of Latin America, the highest levels of nicotine were measured in bars, but a median concentration of 0.60 µg/m 3 nicotine was also detected in the non-smoking areas of restaurants, showing that unisolated non-smoking areas are not fully effective in controlling SHS exposure [44].
PM 2.5 concentrations were compared in 32 countries from the five regions of the World Health Organization (WHO). The levels were 7.5 times higher in countries without indoor air policies compared with countries with such policies. The concentrations were also much higher in venues where smoking was noticed than in those where it was not; they are linked with the average number of cigarettes (or water-pipes) per 100 m 2 [45]. In venues where smoking was noticed, the observed levels were well above the WHO quality targets of 10 µg/m 3 annual mean and 25 µg/m 3 24h mean [7].

Hospitals
Air nicotine concentrations were quite dissimilar in the hospitals of five European cities in 2001-2002 (ranging from 0.14 to 4.0 µg/m 3 ) [43]. In Latin America, despite restrictions or bans on smoking, low levels of nicotine were detected in 95% of hospital samples, with a large range within and across hospitals (0.01-1.33 µg/m 3 ) [44].

Aircrafts, airports and train stations
In-flight air quality measured in ±250 American aircrafts before the 1989 smoking ban violated the PM 2.5 federal air quality standards in the cabins by approximately three fold for flight attendants due to major deficiencies in the ventilation systems [46]. In seven European cities in 2001-2002, despite smoking restrictions in most areas, ranges of air nicotine concentrations were 0.1-5 µg/m 3 in airports and 0.5-10 µg/m 3 in train stations [43].

Motor vehicles
In smokers' vehicles, median air nicotine concentrations were 9.6 µg/m 3 , whereas there were non-detectable concentrations in non-smokers' vehicles. The increase was 1.96 fold per cigarette smoked. The level of air nicotine was higher than in restaurants and bars [47]. In 18 motionless cars, windows closed, the average levels of PM 2.5 were >3,800 µg/m 3 ; with air conditioning, the levels decreased to 844 µg/m 3 , and holding the cigarette next to a half-open window, to 223 µg/m 3 [48].
In three stationary cars in Crete, where the outside baseline measurements were 13 µg/m 3 , with a cigarette left to burn until it extinguished itself, and without heating/air conditioning, the mean PM 2.5 was between 12,000 and 13,000 µg/m 3 ; it decreases significantly with increase in air circulation (from windows fully closed, to half open or fully open), and in vehicles with a large interior passenger volume [49]. The nicotine level in air sample, surface wipe and dust from used cars for sale was higher in 87 smokers' cars than in 20 non-smokers' cars. When smokers had imposed a car smoking ban, the air nicotine levels were significantly lower, but dust and surface contamination levels remained at similar levels. This shows the importance of the issue of -third hand smoke‖ [50].

Schools and universities
In seven European cities, median air nicotine concentration ranged from 0.07 to 0.80 µg/m 3 in schools and from 0.01 to 0.50 µg/m 3 in universities, with similar data whether or not smoking was allowed in a particular area [43]. In Latin America, despite a smoking ban in most countries, nicotine was detected in 78% of samples, but at a low median concentration [44].

Homes and residences
Since the implementation of bans of smoking in public places, homes have become a major contributor to SHS exposure. PM 2.5 was measured over a 3-day period in nine homes where smoking was allowed and in three smoke-free homes [53]. PM 2.5 levels in the primary smoking area of the smoking homes were significantly higher than in distal areas (84 vs. 63 µg/m 3 ). Smoking in only part of the home offers thus no complete protection for people anywhere inside the home. In smoke-free homes, PM 2.5 levels were 9 µg/m 3 lower than the Environmental Protection Agency (EPA) annual standard for air quality (15 µg/m 3 ).
A partial explanation of the persistence of SHS in homes where a smoker is living but not smoking inside the home could be the washout time for residual tobacco smoke from the lungs after the smoker's last puff , as 10 repeated re-entries of a smoker after the last puff increases the PM 2.5 in the room from a background of 0.56 µg/m 3 to 3.32 µg/m 3 [54].
In the Greater Boston Area, air nicotine concentrations and the air exchange rates were measured in 49 low-income multi-unit residences. There was a wide range of levels, from the limit of detection to 26,92 µg/m 3 , with a mean value of 2.20 µg/m 3 . Smoke contamination is not limited to houses with smokers, but increased levels of nicotine concentration were observed in non-smoking homes, suggesting SHS infiltration from neighboring units [55]. PM 2.5 levels measured through personal monitors worn by 407 non-smoking pregnant Polish women were higher when exposed to >10 cigarettes per day (CPD) (88.8 µg/m 3 ) than when exposed to ≤10 CPD (46.3 µg/m 3 ) or unexposed to SHS (33.9 µg/m 3 ) [56].
3.1.8. Water-pipe smoking During laboratory sessions, the levels of PM 2.5 in the air increased from 48 µg/m 3 at background to 264 µg/m 3 after water-pipe smoking by 10 individuals, while after smoking 10 cigarettes the levels increased from 44 µg/m 3 to 267 µg/m 3 . Water-pipe smoking thus contributes as much to SHS as cigarette smoking [57].

General population
A decrease in SHS exposure has been observed in the USA [58,59]. In the cross-sectional National Health and Nutrition Examination Survey (NHANES) surveys of the US civilian population aged ≥4 years, among non-smokers (serum cotinine ≤ 10 ng/mL), exposure to SHS (defined as a detectable serum cotinine level of ≥0.05 ng/mL) declined significantly from 83.9% in 1988-1994 to 46.4% in 1999-2004. The percentage of decline was lower among the younger population aged 4-19 years than for those aged ≥20 years, and smaller for the lowest income group compared with the higher income groups. This decrease can be linked with the smoking bans and the reduced prevalence of active smoking [58].
The prevalence of serum cotinine levels ≥0.05 ng/mL in the non-smoking US population decreased significantly, from 52.5% during 1999-2000 to 40.1% during 2007-2008. Nevertheless, an estimated 88 million non-smokers aged ≥3 years are still exposed. Children remain among the most exposed [59]. Among a nationally representative sample of 4,952 non-smoking employed adults who reported no home exposure to cigarette smoke, the geometric mean serum cotinine (ng/mL) ranged from 0.09 for farming, foresting and fishing occupations to 0.22 for operators, factory workers, laborers and to 0.47, the highest, among waiters for the period 1988-1994, with values decreasing between 1988 and 1994 [60].
In US, with a cut-off point of ≥0.05 ng/mL serum cotinine for adult non-smokers, the prevalence rates of exposure to SHS in the general population were 44.9% nationally (4,476 subjects in the 2003-2004 survey) and 56.7% in New York City (2004 survey), respectively, contrasting with active smoking prevalences of 29.7% and 23.3%, respectively [61].
Among African-American or Dominican pairs of women and newborns residing in underserved neighborhoods of New York City with levels of plasma cotinine <25 ng/mL, 53% of mothers and 49% of infants had levels indicative of SHS exposure (≥0.03-<25 ng/mL) [62].

Hospitality venues
Twenty four h urine samples were collected on working and non-working days in non-smokers working in restaurants and bars that permitted smoking. Significant increases in nicotine + nicotine-glucuronides, cotinine + cotinine N glucuronide and total NNAL were observed during the working period [63].
In New Zealand hospitality venues, the saliva cotinine concentrations increased more between the beginning and the end of the workshift among workers in premises allowing smoking among customers everywhere than in smoke-free premises, and in premises permitting smoking only in designated areas [64].
In Dakota, USA, after exposure to SHS in three heavily polluted bars, the net mean increases in urinary cotinine among eight healthy non-smoking patrons were measured before and estimated after 6 hours (from the mean of 2 h and 12 h measurements). The post-exposure values in the three bars were 4.28 ng/mL, 6.88 ng/mL and 9.55 ng/mL, respectively, above the pre-exposure background values [65].
Prior to the smoking ban, the increase in saliva cotinine was measured after 3-hour exposure among non-smokers in bars in New Zealand: the mean increase was 0.66 ng/mL, i.e., ±8-fold, higher in winter than in spring, even in venues that seemed to be smoke-free on direct observation [66].
In Hong Kong, among 104 workers in catering facilities not exposed outside the workplace, mean urinary cotinine levels ranged from 6.4 and 6.1 ng/mL when smoking was not allowed or restricted to 15.9 ng/mL with unrestricted smoking [67].
Urine concentrations of cotinine were measured in 100 non-smoking volunteers before entering a Mexican discotheque and 6 hours after the end of the exposure. In males, pre-exposure levels of cotinine (3.7 ng/mL) increased to 49.1 ng/mL post-exposure and in females from 2.2 ng/mL to 15.7 ng/mL [68].
A volunteer sample of bar (86%) and restaurant (14%) workers was recruited for a hair nicotine prevalence study. SHS exposure at work, at home, in the car or truck and in any other source of regular exposure was self-reported. Among 129 non-smokers, the hair nicotine levels were 0.63 ng/mg with no exposure, 1.18 ng/mg with one source, 1.32 ng/mg with two and 1.96 ng/mg with three or more; type of establishment, bar versus restaurant (p = 0.001) and number of sources (p = 0.007) were the best predictors of hair nicotine level. The total variance explained by the model is only 12%. The total hours of reported exposure were associated with both the number of sources (p < 0.0001) and hair nicotine level (p < 0.0001) [69].

Casino patrons
Mean increases in total urine cotinine (0.044 nmol/mg creatinine) and total NNAL (0.018 pmol/mg creatinine) were significant (p < 0.001) among non-smoker casino patrons before and after 4 hours stay in a casino where smoking was allowed [70].

Hospital inpatients
In a public hospital of San Francisco, in 2005-2006, significant exposure to SHS (possibly before hospitalization) was demonstrated through measurement of serum cotinine levels in 32% of inpatients denying smoking [71].

Aircraft
In flight attendants, before the smoking ban in aircrafts, SHS assessed through cotinine urinary dosimetry was estimated to be 14-fold that of the average person. Ventilation declined by 33-60% between 1970 and 1987, and massively failed to control SHS air pollution in aircraft cabins [46].

Home
In the Health Survey for England, nationally representative samples of children aged 4-15 years (n = 13,365) and their parents were interviewed in the home. The proportion of homes where one parent was a smoker but that were smoke-free, increased from 21% in 1996 to 37% in 2007, and where both parents smoked but were also smoke-free, from 6% to 21%. Most homes with non-smoking parents were smoke-free (95% to 98%). The geometric mean saliva cotinine across the years was 0.22 ng/mL for children with non-smoking parents living in a non-smoking home. When one parent was smoking, the mean cotinine levels were 0.37 ng/mL in non-smoking homes and 1.67 where there was smoking in the home. For the children with two smoking parents, these values were 0.71 and 2.46 ng/mL, respectively. Across the years, there was a declining trend in saliva cotinine concentration in smoke-free homes for children of smoker or non-smoker parents [72].
In Poland, one year after delivery, 46% of children are exposed to SHS as confirmed by urine cotinine measurements [73].
In Spain, in a cohort study of 987 infants, the median urinary cotinine increases 1.4 times when father smokes and 3.5 times when mother smokes. In most children considered unexposed to SHS by their mother, urine cotinine levels above 1 ng/mL are detected [74].
Hair cotinine was measured in 104 mother/child pairs. In general, child hair cotinine levels were higher than maternal levels (1.18 ng/mg versus 0.78 ng/mg, p < 0.001). Levels in children of nonsmokers were higher than maternal levels (0.77 ng/mg versus 35 ng/mg, p < 0.001) while levels were not different between smokers and their children (1.91 ng/mg versus 1.92 ng/mg, p = 0.978) [75]. Geometric means of hair nicotine of pregnant women are the lowest (0.33 ng/mg) when the spouse is not smoking (A), intermediate (0.51 ng/mg) when the spouse is a smoker not smoking in the home (B) and 0.58 ng/mg when the spouse is smoking in the home (C). Differences between A and B and A and C are significant, but not between B and C. Not smoking only in the home is thus inadequate to protect completely pregnant women from SHS [76].
Pregnant women frequently underestimate their exposure to SHS. In a study of 698 pregnant women reported as non-smokers, 305 (43.7%) claimed not to be exposed to SHS, yet 196 of these (64.3%) had plasma cotinine levels above the limit of detection [77].
Among 114 infants living in homes with parents who smoked, urine total NNAL was detectable in 67, for whom the number of cigarettes smoked per week in the home or car was significantly higher. The mean level of NNAL in the 144 infants was 0.083 pmol/mL, higher than in most other field studies of SHS, probably linked to the proximity of infants to parents who smoke [78].
In families where the pregnant women was neither smoking nor exposed to external sources of SHS, and where the father was smoking indoors, the hair nicotine concentrations of the neonate are significantly higher (p < 0.05) than in the group with non-smoking fathers or fathers smoking only outdoors. Paternal smoking inside the home thus produces maternal and fetal intake of SHS [79]. Children's urine cotinine decreased non-significantly in children of mothers receiving counseling to reduce children's secondhand exposure versus a control group of mothers with usual care [80].

Experimental conditions
Forty non-smokers were exposed for 4 hours to aged, diluted SHS generated by a smoking machine (140 µg nicotine/m 3 ). Short-term increases in saliva cotinine were similar, at approximately 12 pg/mL/min, among men and women, African-Americans and Whites, suggesting an absence of metabolic differences between the groups [26].

Pets
Urinary cotinine was detected in 15 domestic dogs from homes where residents smoked at least 20 CPD for a minimum of 24 months, and not in 15 dogs from homes without smokers. In exposed dogs, broncho-alveolar lavage detected an increase in macrophage and lymphocyte population [81].
Cats from 19 households in which smoking was reported had significantly higher concentrations of urine total nicotine (70.4 ng/mL), total cotinine (8.53 ng/mL) and total NNAL (0.0562 pmol/mL) compared with cats that lived in 42 households in which there was no smoking. (4.89 ng/mL, 0.74 ng/mL and 0.0182 pmol/mL, respectively [82]. The risks incurred by their pets can sometimes push smokers towards cessation.

Effects of Smoking Bans
Air and biomarkers measurements allow both the effectiveness of the smoking bans and the level of their effective implementation to be quantified.

Hospitality venues
Major improvements were observed in hospitality venues of many countries after the introduction of smoking bans. The levels of PM 2.5 were measured before and after the 2003 New York State law on smoke-free workplaces in a purposeful sample of 20 hospitality venues. After the ban, a general decrease of 84% of PM 2.5 was observed. In bars and restaurants where smoking was occurring at baseline, a decrease of 96% was observed (towards mean values below the security standard). In the restaurant portion of the two bar-restaurants initially allowing smoking in the bar but not in the restaurant section, the decrease averaged 58% in the restaurant section. In four large recreation venues (two bowling alleys, a pool hall and a bingo hall) there was a 76% decrease [83].
In Wilmington, Delaware, where the smoking ban was extended in 2002 to hospitality venues, the air PM 3. 5 and PPAH were measured in a casino, six bars and a pool hall/bar where smoking was initially allowed. The average prelaw values of PM 3.5 were 231 µg/m 3 vs. 9.5 µg/m 3 outdoors and 134 ng/m 3 PPAH vs. 21 ng/m 3 outdoors. Mean post-ban values are indistinguishable from outdoors. Ventilation or air cleaning would be unable to obtain such a decrease [9].
After the smoke-free workplace law in Boston, MA, the PM 3.5 decreased by 23-fold (to an average of 7.7 µg/m 3 ) in six pubs, and the PPAH by nearly 10 times, to 6.32 ng/m 3 . PM reached the levels of good air quality in 6/7 venues, and PPAH levels were lower than outdoors in the seven pubs [10].
The effects of the Irish smoking ban in bars, restaurants, cafes and hotels (excluding bedrooms, outdoor areas and properly designed smoking shelters) were assessed via air nicotine measurements. There was an 83% reduction in air nicotine concentration from a median 35.5 µg/m 3 to 5.95 µg/m 3 (p < 0.001). Validated exposure and the number of observed hours of exposure were thus reduced but not totally eliminated after this partial ban [84]. In 42 Irish pubs, a PM 2.5 reduction of 83% was observed after the ban. In 26 pubs, benzene concentrations were reduced by 80.2% [85].
In the English hospitality industry, indoor PM 2.5 concentrations decreased by 95% from 217 µg/m 3 at baseline to 11 µg/m 3 one month after the ban [86]. In a random selection of 41 pubs in two Scottish cities, the levels of PM 2.5 averaged 246 µg/m 3 before and 20 µg/m 3 after the ban. In most pubs, the level of PM 2.5 became comparable with the outside ambient air PM 2.5 level. The levels of benzene decreased also from 18.8 µg/m 3 preban (above the security limit of 5 µg/m 3 ) to 3.6 µg/m 3 postban. Reductions were observed in all post-ban visits, showing high compliance with the ban [87].
Full shift PM 2.5 geometric means were measured with personal PM 2.5 monitors in six bar workers in Scotland. They fell from 202 µg/m 3 pre-ban to 28 µg/m 3 post-ban, meaning (after log analysis) an average reduction of 86% [88].
In selected restaurants and bars with a serving area larger than 100 m 2 in Finnish cities, measurements were conducted before and after enforcement of the act imposing smoke-free zones in such venues. The overall air nicotine concentrations decreased in 10 of 18 establishments. The geometric mean nicotine concentration moved in restaurants from 0.7 µg/m 3 to 0.6 µg/m 3 , in bars and taverns from 10.6 µg/m 3 to 12.7 µg/m 3 , in discos and nightclubs from 15.2 µg/m 3 to 8.1 µg/m 3 . In the smoke-free sections, the concentrations were practically unchanged (2.9 µg/m 3 ). The small improvements in some sites were linked with increased ventilation, but physical barriers separating smoking from smoke-free areas are thus required to obtain a major reduction in SHS [89].
In 13 Norwegian bars and restaurants, the arithmetic mean concentration of air nicotine declined from 28.3 µg/m 3 to 0.6 µg/m 3 after the total smoking ban. Total dust declined from 262 µg/m 3 to 77 µg/m 3 [90].
In nine Swedish communities (bingo halls, casinos, bars and restaurants), exposure above the air nicotine cut-off level chosen to identify possible health risk (<0.5 µg/m 3 ) existed in 87% before and 22% after the smoking ban; the pre-ban exposure levels were highest in bingo halls and casinos [91].
In Australia, where an indoor smoking ban was promulgated, in a convenience sample of pubs and bars that had at least one indoor area with an adjacent semi-enclosed eating area, PM 2.5 concentrations were reduced indoors by 65% (p = 0.004) and outdoors by 38.8% (p = 0.037) after the ban. However, after adjustment, a 100% post-ban increase in geometric mean outdoor PM 2.5 was associated with a 36.1% increase in geometric indoor PM 2.5 , showing that the air quality in smoke-free indoor areas may be compromised by smoking in adjacent outdoor areas [92].
The indoor air quality was compared by measuring PM 2.5 before and after enforcement of a smoking ban in Italy in 40 public places (bars, pubs, restaurants and game rooms). PM 2.5 decreased from 119.3 µg/m 3 to 38.2 µg/m 3 after 3 months (p < 0.005) and to 43.3 µg/m 3 one year later. The mean outdoor measurements of PM 2.5 were 20.8 three months post-law and 27.2 one year post-law [93].

Prisons
In 22 dormitory and common areas in six North Carolina prisons, the PM 2.5 levels decreased by on average 77% after the indoor smoking ban [94]. This is particularly important, as the prevalence of tobacco use is much higher among inmates and prison staff than in the general population.

Public places, industrial service sector and office workplaces
In eight Finnish public and private workplaces, median indoor airborne nicotine concentrations decreased from 0.9 µg/m 3 before to 0.1 µg/m 3 one year after and then remained unchanged 3 years after implementation of the national smoke-free legislation [95].
In Montevideo, Uruguay, a middle income country, the median air nicotine concentrations of 100-103 indoor samples were 0.75 µg/m 3 before compared to 0.07 µg/m 3 after the legislation, but with a large overlap of the interquartile rate (IQR) between the two periods. The overall reduction was 91% after adjustment for room volume and ventilation. The greatest reduction was observed in schools (97%), followed by airports (94%), hospitals (89%), local government buildings (86%) and restaurants/bars (81%). The reason for the much lower overall air nicotine concentration in Uruguay versus other countries is not mentioned, but could be due to the selection process or the differences in prevalence of active smoking or increased ventilation in warm countries and the different patterns of active smoking [96].

Contrast between countries with and without smoking bans
The air quality differences of hospitality premises between a country with (Italy) and without (Austria) a smoking ban was assessed before and 2 years after the introduction of the Italian smoking ban in respectively 28 and 19 hospitality premises. Median airborne nicotine concentrations decreased from 8.86 µg/m 3 to 0.01 µg/m 3 in Italy (p < 0.001) and, in contrast, increased from 11.00 µg/m 3 to 15.76 µg/m 3 in Austria, a non-significant increase [97].

Effects of Smoking Bans on Biomarkers of SHS
Effectiveness and implementation of smoking bans can be measured through direct comparison of biomarkers pre-and post ban in the same areas, and can also be estimated through indirect simultaneous comparisons of biomarkers in areas with and without bans (with possible bias linked with other inter-area differences). The fact that smoke-free bans are not -applicable‖ to home and car exposure could explain the higher serum concentration in children, while higher SHS exposure also seems responsible in non-Hispanic blacks [58].
In the general population of New York, saliva cotinine levels decreased by 47.5% among non-smoking adults after implementation of the 2003 New York state ban on smoking in indoor workplaces and public places [98].

Comparisons between areas
In the US, SHS exposure, defined as serum cotinine ≥ 0.05 ng/mL, was determined from 5,866 non smoking adults categorized by their level of coverage by a smoke-free law. The odds of exposure was much lower with extensive coverage (0.10 in men and 0.19 in women) and lower with limited coverage (0.57 in men and 0.90 in women) than with no coverage, used as reference value [99].

Pre-and post comparisons
In a pre-post follow-up study, saliva cotinine specimens were available among 24 non-smoking workers in restaurants, bars and bowling facilities followed up until 12 months after the New York smoke-free law. Saliva cotinine concentration decreased from 3.6 ng/mL (95% confidence interval [CI] 2.6-4.7) to 0.8 ng/mL (95% CI 0.4-1.2), suggesting compliance with the law [100].
The effects of the Lexington smoke-free law covering all buildings open to the public were investigated respectively before and 3 months after the law was implemented in a convenience sample of 105 workers (including 40 active smokers), of whom 92 worked in restaurants and 13 in bars. The geometric mean of hair nicotine among bar-workers was 4.51 ng/mg before and 1.06 ng/mg 3 months post-law, and in the restaurant cohort 1.54 ng/mg before and 1.20 ng/mg after. After adjustment for CPD, the decline from pre-to post-law was significant for bar workers (p = 0.0004), but marginal among restaurant workers (p = 0.07). Subject retention was a problem, as only 71 of the 105 initial participants were followed up at 3 months [101].
In Minnesota communities, among 24 non-smokers working in hospitality venues (12 bars, 6 restaurants, 5 bowling alleys and one unknown workplace) and living in non-smoking households, the median percentage decrease after the ban was very significant in urine as well for total cotinine (83.3%) as for total NNAL (76.6%). Levels of total NNAL/mg creatinine were significantly correlated with the number of hours worked in the smoking section (r = 0.43; p < 0.05) [102].
In England, among 75 non-smoking employees of the hospitality industry, saliva cotinine was reduced from 3.6 ng/mL one month before the smoke-free legislation to 0.9 ng/mL one month after its implementation (p < 0.001) [86].
The same trend was observed in Norway among 25 non-snuffing non-smoker employees in bars and restaurants from a geometric mean urine cotinine concentration of 9.5 µg/g creatinine preban to 1.4 µg/g creatinine post-ban (p < 0.01). However, there was also a reduction among 29 non-snuffing smokers, from 1,444 µg/g creatinine to 688 µg/g creatinine (p < 0.05), suggesting a reduction in active smoking after the intervention [90].
The impact of the Irish smoking ban on the saliva cotinine concentration in a cohort of 35 workers from 15 city hotels translated into a 69% decrease from a median value of 1.6 ng/mL to 0.5 ng/mL (p < 0.0005), showing the persistence of limited exposure. Self-reported exposure to SHS nevertheless decreased from a median of 30 hours a week to zero [84].
In Italy, the concentration of urine cotinine among non-smoker workers in 40 hospitality venues decreased from 17.8 ng/mL to 5.5 ng/mL 3 months after (p < 0.0001) and to 3.7 ng/mL 12 months after (p < 0.0001) the smoking ban, together with a contemporaneous reduction in indoor fine and ultrafine particles [93].

Inter-area comparisons
The effects of the New York clean indoor legislation appear also in a pre-postlaw comparison of non-smoking hospitality workers compared with American Indian casino workers exempt from the law. The median urine cotinine decreased from 4.93 ng/mL prelaw to 0.30 post-law (p < 0.01) among the former, but remained at a higher level in the latter (8.40 vs. 6.49). Urine cotinine levels were significantly related with self-reported numbers of exposures and number of days exposed [103].
In Oregon, the influence of exposure was assessed among CO-validated non-smokers employed in 52 bars or restaurants where smoking was allowed or in 32 where smoking was prohibited by local ordinances. After a workshift of 4 hours, among workers exposed to workplace SHS, the odds of having a detectable urine level of total NNAL were almost six times higher than for protected workers (aOR = 5.66; p = 0.005) . The same was true for detectable levels of urine nicotine (aOR = 109.01; p < 0.001) and cotinine (aOR = 95.21; p < 0.001). The post workshift mean level of urine total NNAL, nicotine and cotinine was 0.02 pmol/mL, 1.39 ng/mL and 1.4 ng/mL for protected workers vs. 0.04 pmol/mL, 44.36 ng/mL and 20.20 ng/mL for exposed workers. Each hour of self-reported exposure to SHS was associated with an increase of 6% in total NNAL, of 33% in total nicotine and of 12% in total cotinine [104].
Following the smoke-free Spanish law of 2006, in bars and restaurants >100 m 2 , the proprietors were still allowed to provide a separate, independently ventilated smoking area comprising less than 30% of the total floor area. For venues with a floor area <100 m 2 , the owner could choose between a totally non smoke-free or a totally smoke-free venue (in fact only 10-20% of those venues were smoke-free). In Portugal and Andorra, no smoking ban was in effect at the time of the study. In a convenience sample in Spain, Portugal and Andorra, hospitality venues were recruited based on their size (< or >100 m 2 ) and area: these venues were pubs, bars, restaurants and discos; 117 non-smoking workers completed the study in Spain and 20 in Portugal and Andorra. Non-smokers were defined as self-reported never or former smokers, with a saliva cotinine concentration <20 ng/mL, a high cut-off level justified by the potentially high exposure to SHS in hospitality venues. Before the Spanish law, the median stimulated saliva cotinine concentrations among non-smokers were 2.0 ng/mL in Spain and 1.2 ng/mL in Portugal and Andorra (p < 0.01). In Spain, in venues where smoking was totally banned, the median saliva cotinine decreased from 1.6 ng/mL before the ban to 0.5 ng/mL after the ban (p < 0.01), a 56.6% reduction; in 22 venues where smoking was permitted in designated areas, from 1.8 ng/mL to 1.1 ng/mL (p = 0.0618), a non-significant 31.9% reduction; in 63 venues where smoking was permitted throughout the premises, saliva cotinine remained stable from 2.5 ng/mL to 2.6 ng/mL (p = 0.475). In Portugal and Andorra, the median saliva cotinine remained unchanged before (1.2 ng/mL) and after (1.2 ng/mL) the Spanish ban. Self-reported duration of SHS exposure decreased by 100% in Spanish venues where smoking was totally banned (p < 0.01), whereas a borderline significant decrease (from 8 hours median exposure per day to 1 hour per day; p = 0.055) was observed in venues where smoking was partially permitted. Median duration of exposure increased significantly in venues with no smoking restrictions (from 8.0 to 10.0; p < 0.01).
In Portugal and Andorra, duration of exposure to SHS in hospitality venues remained unchanged, at 8 hours. SHS exposure outside the workplace decreased in Spain regardless of the type of postban regulation (between −16% and −54.6%) but increased in Portugal and Andorra (+54.2%) [105].

Pre-and post comparisons
In 73 non smoking barmen working in Dublin pubs, there was a 79% reduction in exhaled breath CO and a 81% reduction in saliva cotinine after a total workplace smoking ban, together with a 90% decrease in total exposure at work, and a 42% decrease outside of work [85].
The same type of comparison was conducted in Canada between Toronto (79 eligible non-smoking bar workers), where a smoke-free bylaw was enacted and Windsor (49 eligible non-smoking bar workers), with no bylaw change. Urinary cotinine levels decreased from 24.2 ng/mL to 7.8 ng/mL in Toronto immediately after the ban, with the reduction sustained 1, 2 and 9 months post-ban, while there was no change among Windsor bar workers. The compliance with the ban was both high and persistent [106].
Prior to the introduction of the Scottish smoke-free legislation and 2 months and 1 year after its implementation, saliva samples were collected in their places of work in 191 bar workers. In non-smokers, the saliva cotinine level decreased from 2.94 ng/mL to 0.41 ng/mL at 1 year (a reduction of 89% in non-smokers' cotinine levels). Among smokers, a reduction of 12% was recorded during the same period, probably reflecting both reduction of exposure to SHS at work and a decrease in active smoking [88].
The same trend was observed for benzene and 1-3 butadiene levels, for which the average dose after a standard exposure of 3 hours in Irish pubs diminished by 91 and 95%, respectively, after the ban [36].

Inter-area comparisons
In the Republic of Ireland, among 158 non-smoker bar staff, a decrease in saliva nicotine concentration was observed before and after legislation for smoke-free workplaces (from 29.0 nmol/L to 5.1 nmol/L), while in Northern Ireland, without a ban, over the same period the decrease was much more limited (25.3 nmol/L to 20.4 nmol/L) and not significant, with overlapping of the confidence intervals. After adjustment, the decline was twice as great in the Republic than in Northern Ireland [107].

Bar patrons
In Dakota, USA, in three heavily polluted bars, the net mean increases in urine cotinine among eight healthy non-smoking volunteers were measured before and estimated after a 6-hour exposure (from the mean of 2 h and 12 h measurements). The post-exposure values in the three bars were 4.28 ng/mL, 6.88 ng/mL and 9.55 ng/mL above the pre-exposure background values [65] Saliva cotinine levels were measured among non-smoking volunteers before and after a 3-hour visit to 30 randomly selected bars in New Zealand. Before the comprehensive smoke-free legislation, there was an average increase in saliva cotinine of 0.66 ng/mL. After the law change, the mean saliva cotinine increase was much lower, but persisted at 0.08 ng/mL, despite quasi total compliance with the law (only 1 lit cigarette observed in 30 visits) [108].

Homes
Since the smoking bans in many countries, the home has become a major source of exposure to SHS. In Scottish adult non-smokers from non-smoking households, geometric mean saliva cotinine concentrations fell by 49% (from 0.35 ng/mL to 0.18 ng/mL, p < 0.001) within 1 year after implementation of the smoke-free legislation in enclosed public places and workplaces. A smaller (16%) fall in cotinine concentration in non-smokers from smoking households was not statistically significant. The reduction in self-reported exposure to SHS was clear in public places, but not in homes and cars [109].

Adolescents and children
In a national cross-sectional survey of the US population, serum cotinine levels were available for 11,486 non-smoking children and adolescents (3-19 years) from 117 survey locations with extensive versus no coverage by a smoke-free law. Among those without home SHS exposure, in a county with extensive coverage of a smoke-free air law, the adjusted prevalence of detectable cotinine was 0.61 times lower and the geometric mean 0.57 times lower compared with children and adolescents from counties without a smoke-free air law. In children and adolescents with home SHS exposure, after adjustment for covariables, the differences between counties with or without smoke-free air laws was no longer observed [110].
In a nationally representative cross-sectional survey of ±2,500 children in Scotland, the geometric mean saliva cotinine concentrations in non-smoking children fell from 0.36 ng/mL to 0.22 ng/mL after the introduction of the smoke-free legislation (a 39% reduction). The extent of the fall was statistically significant only among pupils living in households in which neither parent figure smoked (from 0.14 ng/mL to 0.07 ng/mL) and among pupils in households in which only the father figure smoked (from 0.57 ng/mL to 0.32 ng/mL) [111].
In Australia, feedback to parents of asthmatic children of the results of the urinary cotinine to creatinine level of their children did not significantly change the prevalence of smoking bans at home or in the car, nor the daily consumption of cigarettes or consumption in front of the child [112].

Health Effects of Exposure to SHS
The health effects of exposure to SHS concern the general population, hospitality workers (for whom exemptions from the smoking bans are frequently considered), non hospitality workers and finally children (as smoking bans do not concern private housing). The types of effects are disease symptoms, disease prevalence and issues, hospitalization rates for SHS exposure related diseases and biological data possibly related to exposure-related diseases.
As the reader's interest is frequently disease-oriented, the main exposure-related diseases, i.e., cardiovascular, respiratory diseases in adults and children, fertility and pregnancy, sudden infant death syndrome, cancers, neuropsychiatric diseases and genotoxicity, will be considered successively.

Cardiovascular Diseases
In 20.8% of a nationally representative sample of 13,443 English and Scottish adults, high levels of saliva cotinine (0.71-14.99 ng/mL) were observed in non-smokers. After a mean follow-up of 8 years, these were associated with all-cause deaths (age adjusted hazard ratio [HR] 1.25 [95% CI 1.02-1.53)] and cardiovascular deaths (adj HR 1.21 [95% CI 0.85-1. 73)]. High SHS was also associated with elevated C-reactive protein (CRP), which contributes to explaining the association between SHS and deaths due to cardiovascular disease (CVD) [113].
Admission serum cotinine concentrations were measured in consecutive patients admitted to nine Scottish hospitals over 23 months. Among 1,261 never smokers, within 30 days, 50 (4%) had died and 35 (3%) had had a non-fatal myocardial infarction. All-cause death increased from 2.1% in those with cotinine ≤0.10 ng/mL to 7.5% in those with cotinine >0.9 ng/mL (χ 2 test for trend p < 0.001). This difference persisted after adjustment for confounders (adj OR for cotinine >0.9 ng/mL = 4.80 [95% CI 1. 96-11.83]; p = 0.003). The same trend was observed for death, cardiovascular death, or myocardial infarction. SHS exposure is thus associated with worse early prognosis after acute coronary syndrome [114].
From the 10-month period preceding the passage of the Scottish smoking ban (2006) to the same period the next year, among all patients admitted with acute coronary syndrome in nine hospitals, there was a decrease in the geometric mean concentrations of serum cotinine among never smokers from 0.68 ng/mL to 0.56 ng/mL (p < 0.001), and a 21% reduction in the number of admissions for acute coronary syndrome, together with a self-reported decrease in the weekly duration of exposure to SHS. Reductions in admissions for acute coronary syndrome were also observed among former smokers (19%) and smokers (14%), while the number of deaths among patients with acute coronary syndrome who were not admitted to the hospital decreased by 6%. These reductions contrast with the smaller overall reduction in hospital admissions of 4% in England (where such legislation did not exist) during the same period, and with the 3% mean annual decrease in admissions in Scotland during the decade preceding the study [115].
In Rome, Italy, acute coronary events were compared between the pre-and post-legislation period, taking into account several time-related potential confounders, including PM 10 air pollution. The PM 10 concentration decreased, but not significantly, between the year before and after the ban. The age-standardized rate ratios of acute coronary events (defined as hospitalization records of acute coronary events + other out of hospital deaths due to ischemic heart disease from the mortality register) were 0.89 (95% CI 0.85-0.93) post-smoking ban versus pre-smoking in the 35-64 years age group and 0.92 (95% CI 0.88-0.97) in the 65-74 years age group, but 1.02 (95% CI 0.98-1.07) in the 75-84 years age group. After simultaneous adjustment for both time trends and hospitalization rates, the results become borderline in the youngest age group (RR 0.94; 95% CI 0.89-1.01) but remain significant in the 65-74 olds (RR 0.90; 95% CI 0.84-0.96). It remains uncertain whether these limited effects (11.2% reduction in the population aged 35-64 years and 7.9% in those aged 65-74 years) are effectively linked with the smoking ban or with the decrease in active smoking suggested by the reduction in cigarette sales in 2005, or with other, perhaps not controlled factors. The level of implementation of the law was not discussed in this study [116].
In US, the cross-sectional relation between SHS exposure and recognized biomarkers of heart disease risk was investigated among 5,599 never smoking adults (serum cotinine levels ≤15 ng/mL) from the third (1988-1991; 1991-1994) national representative examination survey. Serum cotinine (threshold of detection 0.05 ng/mL) was not detectable in 18% of the subjects; the other subjects were distributed into low (0.05-0.215 ng/mL) and high cotinine levels (0.215-15 ng/mL). Levels of fibrinogen and homocystein were significantly higher in people with detectable serum cotinine versus those without (9-10 mg/dL, p = 0.03 and 0.8 µmol/L, p < 0.001) higher respectively, but were not statistically different between low and high cotinine levels. White blood cell (WBC) counts decreased only for high cotinine levels. CRP did not increase in relation to cotinine. This lack of association between cotinine and CRP could be attributed to the low sensitivity laboratory method used for CRP in this survey. Among 4,990 current active smokers, an adjusted mean increase of 29.2 mg/dL of fibrinogen and 1.8 µmol/L of homocystein was observed in comparison with never smokers without detectable cotinine. The apparent effects of SHS on fibrinogen and homocystein, markers of cardiovascular risk, in non-smokers were approximately one third to one half those observed in active smokers, while cotinine levels of passive smokers were only ≈ 0.1% of those of active smokers. These disproportionate associations fit the epidemiological evidence of a similar disproportionate association of SHS exposure and active smoking with coronary heart disease risk. SHS is thus an important avoidable cause of cardiovascular disease [117] In the US NHANES (1999)(2000)(2001)(2002), in non-smoking children and adolescents aged 6-18 years, multiple regression analysis indicated that a change in serum cotinine level of 0.5 ng/mL was associated with a 0.96 mg/dL change in CRP (95% CI 0.93-1.00) even after adjustment. There is thus a significant association between serum cotinine and elevated serum CRP [118].

Respiratory Symptoms and Diseases in Adults
Experimental exposure to one hour SHS (set at bar/restaurant levels) by 16 voluntary non-smoking adults is followed by a mean decrease in FEV 1 (forced expiratory volume in one second) from 4.3 L to 3.8 L and of FEV 1 /FVC (forced vital capacity) from 0.9 to 0.8 (p < 0.05) immediately after exposure, but by a return to baseline values after 3 hours. Serum cotinine increased from 8.9 ng/mL before exposure to 35.5 ng/mL 3 hours after exposure. This lag between spirometric and cotinine data is due to the duration of metabolism from nicotine to cotinine. Other inflammatory cytokines are also increased after 3 hours in contrast with the return to the initial spirometric values at that moment [119].
The impact of smoking policy on respiratory health was assessed among food and beverage servers residing in the Vancouver (Canada) area. Acute and chronic respiratory symptoms were the most prevalent where smoking was allowed, less prevalent where smoking was partially restricted and still less prevalent where it was restricted in 100% of the workplaces. The FEV 1 /FVC ratio was lower (77.8%) in facilities where smoking was permitted than where it was prohibited (81.1%) (p = 0.04). In a parallel fashion with this clinical evolution, in a subset of 88 workers, the geometric mean of hair cotinine was 1.4 ng/mg for those working in facilities where smoking was prohibited, 4.6 ng/mg with partial restriction and 5.4 ng/mg with no restriction (p = 0.01). In the small subset of workers who underwent spirometric and cotinine measurements, there was no relationship between lung spirometry and hair nicotine levels [120].
In public places in Scotland, concurrently with a decrease in serum cotinine levels in 77 non-smoking bar workers, the prevalence of respiratory and sensory symptoms decreased from 79.2% before the smoke-free policy to 53.2% one month after the ban (change 82%; p < 0.001) and to 46.8% 2 months after. FEV 1 increased from 96.6% predicted to 104.8% predicted after 1 month and to 101.7% after 2 months (change 5.1%; p = 0.002). Among the asthmatic bar workers, exhaled nitric oxide, a marker of inflammation, decreased from 34.3 parts per billion (ppb) to 27.4 ppb one month after the ban (an 0.8-fold change; p = 0.04). Smoke-free legislation is thus associated with early improvements in symptoms, minimal increases in spirometric values and decrease of inflammation [121].
Full respiratory function tests were performed on 73 barmen before and after the Irish smoking ban. Among non-smokers, even if significant, the postban increase in flow value measurements remained very limited, and the diffusion capacity for carbon monoxide (DL CO ), as expected, remained unchanged, while among current smokers, practically all parameters were not significantly different between pre-and post-ban periods. The decrease in prevalence of cough and phlegm was significant and relevant in non-smokers, but not in current smokers. Symptoms and spirometric values evolve concurrently with saliva cotinine measurements [85].
In a longitudinal study of a cohort of 77 non-smoking adults with chronic obstructive pulmonary disease (COPD) in the US, higher levels of SHS exposure, as measured by urine cotinine, were marginally associated with worse COPD severity (mean score increment 4.7 points [95% CI −0.1-9.4; p = 0.054]). SHS exposure thus seems to aggravate the symptoms of COPD, even after cessation of active smoking [122].
In Scotland, among 148 patients with severe COPD (among them 39 current smokers), the symptom burden was worse in households with increased PM 2.5 , especially among current smokers [123].
In Sweden, before the ban, 87% of non-smoking hospitality workers were exposed to air nicotine levels of ≥0.5 µg/m 3 , and 22% after the ban, a relative risk of 0.25 (95% CI 0.15-0.41).There was a reduction of ±50% in respiratory and sensory symptoms, but no notable change in lung function [91].
Among 72 current biologically verified non-smokers in a cohort study of COPD, the increased levels of urine NNAL-to-creatinine ratio (a marker of long-term SHS) were associated with greater COPD severity (mean score increase 1.7 points, p = 0.0003) for each interquartile increment of NNAL/creatinine ratio, after adjustment for demographic data and past smoking. They were also related to more severe dyspnoea and more restricted activity (p ≤ 0.05). NNAL, frequently used as a marker of lung carcinogenesis, can thus also be a good long-term biomarker of SHS, showing greater impact on the severity of COPD than cotinine, a short-term biomarker [124].
In Spain [105], 12 months after the smoking ban, concurrently with a decrease in saliva cotinine and hours of SHS exposure, the prevalence of cough and phlegm considered together decreased from 40.6% to 15.6% (p < 0.05) in totally non-smoking venues. The prevalence of any respiratory symptoms also decreased post-ban in completely smoke-free venues (from 56.2 to 38.1; p = 0.012), but not where smoking was allowed in some (p = 0.625) or all (p = 0.774) of the premises. In Portugal and Andorra, with no smoking ban, the prevalence of any respiratory symptoms was not significantly different between the two periods (p = 0.07). This study shows that in Spanish venues where smoking was allowed (at least in theory) in physically separated areas, workers still complained of respiratory symptoms. Even in venues where smoking was totally banned, despite a major decrease, the median saliva cotinine concentration (0.50 ng/mL) was still above the acceptable levels of risk (i.e., 0.14 ng/mL [125]).

Low Respiratory Illnesses in Infancy and Early Childhood
In the 2006 US Surgeon General Report, among the large number of papers allowing a causal relationship to be inferred between SHS exposure after birth and lower respiratory illness in infants and children, middle ear disease in children, respiratory symptoms among children of school age, ever having asthma among children of school age and a lower level of lung function during childhood, most concern self-report exposures and only a minority refer to air or biomarkers of exposure [2].
Recent publications show a relationship between biomarkers, asthma and co-morbidities among young asthma patients. In a low-socioeconomic UK community based cross-sectional study, salivary cotinine levels were significantly increased in children with doctor-diagnosed asthma compared to those without (p = 0.002); cotinine validated exposure levels were also associated with doctor validated asthma (Adj OR 1.8; CI 1.4-2.5) [126].
Among 222 children in whom asthma was diagnosed by the primary care giver, SHS exposure in and outside the home was validated by cotinine in serum and hair, while the level of PM 5 was measured at home. Despite lower reported exposure to SHS at home (14.9 vs. 18.7 CPD), with lower levels of PM 5 in houses, the levels of serum and hair cotinine were significantly higher among African-American children than among white children (1.41 ng/mL vs. 0.97 and 0.25 vs. 0.07 ng/mg, respectively), a difference only explained partially by housing volume [127]. The presence of a strict household smoking ban (confirmed by child cotinine assays) vastly reduced SHS exposure in 91 children with asthma matched with 91 healthy children [128].
Among 200 asthmatic children enrolled in an asthma program, log salivary cotinine level was independently associated with externalizing (p = 0.04), headstrong (p = 0.04) and antisocial behavior (p = 0.04) subscales of a 28-item behavior problem index [130]. Among 220 asthmatic children enrolled in an asthma trial, child behavior problems (externalizing, i.e., hyperactivity and aggression, and internalizing, i.e., depression and behavior symptoms) increased with increased SHS exposure measured as serum cotinine, but the increase was significant only in boys [131].

Fertility and Pregnancy
For issues concerning female fertility or fecundability, spontaneous abortion, preterm delivery, congenital malformations, cognitive and behavioral development of the infant, height and growth, childhood cancers, leukemias, and lymphomas, the evidence was considered inadequate or only suggestive of a causal relationship with exposure to SHS in the 2006 Surgeon General's Report [2], but only a minority of studies relied on cotinine measures, and the latter with discordant results.
In Sweden, a population-based case-control study was conducted between 1996 and 1998 comparing exposure to SHS between 463 women with spontaneous abortion and 864 pregnant women as controls. The Adj OR of spontaneous abortion among SHS-exposed women (0.1-15 ng/mL plasma cotinine) was 1.67 and that of the active smokers 2.11 in comparison with unexposed non-smokers [132].
Among 441 non-smoking pregnant women, the OR of discontinuation of any breastfeeding after the infant's first 6 months was 2.42 in women whose blood cotinine level was above the 75th percentile of cotinine distribution (>0.15 ng/mL) compared with mothers who had lower cotinine levels, with the corresponding OR of discontinuation of full breastfeeding at 1.71 [133].
The risk of preterm delivery (<37 weeks) was higher in a group of non-smoking women whose hair nicotine after delivery was >4.00 µg/g (Adj OR 6.12) in comparison with those with the lowest hair nicotine concentration (<0.75 µg/g) [135].

Sudden Infant Death Syndrome
In the 2006 Report of the Surgeon General, the evidence was considered sufficient to confer a causal relationship between postnatal exposure to SHS and sudden infant death syndrome [2].

Cancers
In an European case-control study in non-smokers (190 controls and 149 cases, mostly all types of cancers), 4 ABP-Hb adducts, markers of carcinogenicity measured 7 years before the disease (median), were somewhat higher in cases than in controls [136]. In a large nested case-control study within the European prospective investigation into cancer and nutrition (EPIC), in which smoking status was supported by cotinine measurements, daily exposure to SHS for many hours during childhood was associated with lung cancer in adulthood (HR 3.63; 95% CI 1. 19-11.11) [137].
In addition to these papers, studies showing elevated levels of PPAH-gluc in the air of venues where smoking was allowed [9] or NNAL or NNAL-gluc, benzene and 1-3 butadiene in body fluids of non-smokers [34,36,63,70,78,104], all markers of carcinogenicity, support the biological plausibility of the influence of SHS on the development of cancers, earlier demonstrated in epidemiological studies.

Mental Health
In a representative sample of 5,560 biologically validated (saliva cotinine < 15 µg/L) non-smoking adults without a history of mental illness drawn from the 1998 and 2003 Scottish Health Surveys, higher saliva cotinine levels (>0.70 µg/L) were associated with higher odds of psychological distress (OR 1.49; 95% CI 1.13-1.97) in comparison with participants with cotinine levels below the limit of detection (≤0.05 µg/L). In the prospective analysis, the risk of psychiatric hospital admission was related to high SHS exposures (multivariable adjusted HR 2.84; 95% CI 1.07-7.59) [138].
The trend of cognitive impairment (assessed by the lowest 10% of scores in a battery of neuropsychological tests) in a cross-sectional analysis of a national population based survey of 4,809 validated non-smokers in England, was positively related with increasing concentrations of saliva cotinine (p for trend 0.02) after adjustment for other risk factors for cognitive impairment. The pattern of association was the same for never smokers and former smokers [139].

Genotoxicity of SHS
The issue of the genotoxicity of SHS was addressed in a recent review of human studies. Biomarker studies carried out among SHS-exposed non-smoking adults show the presence of metabolites of carcinogens in urine, of DNA and/or protein adducts in peripheral blood or other non-malignant tissues, and the presence of micronuclei in bone marrow or peripheral blood. There is thus abundant evidence of the genotoxicity of SHS [140].

Limitations
The following limitations can be identified in the mentioned studies: (1) The biological differentiation between active non daily smokers and passive smokers remains critical, due to overlapping levels of biomarkers, and particularly short-term biomarkers. The -validated‖ definition of smokers, sometimes based on different cutoff levels of plasma cotinine, is a limitation for the validity of comparisons.
(2) Some of the studies can be biased through the use of -convenience samples‖ instead of random representative sampling. Participants in convenience samples may be more conscious of the SHS problems and more prone to implement the regulations [92,105]. The statistical power in some studies was limited [100]. (3) The results of some studies were weakened by the low level of participation and/or by attrition during the process of inclusion or during follow-up [85,88,100,101]. (4) The multiple sources of SHS (also outside the workplace and home) are sometimes unrecognized and can have a confounding effect that can impair the exact quantification of differences in markers between different areas or periods [69]. The prevalence of active smoking can also influence the evolution of SHS exposure [90]. Nicotine and cotinine biomarkers can also be influenced by Nicotine Replacement Therapy.
(5) The level of compliance with the regulations is not always physically controlled and can play a role on the quantified values observed in pre-and post-ban studies or in contemporaneous comparisons between different sites.

Conclusions
Earlier and principally recent SHS publications on air and biomarkers objectively: (1) can validate the smoking status obtained via self-report (active daily smoker versus non-smoker) [1-3], measure the percentages of inexact self-reporting in different circumstances and show the possibility of understatement of SHS exposures in some self-reports [84] (e.g., pregnancy) [77].
(2) confirm the toxin intake difference between non-smokers not exposed (<0.05 ng/mL serum cotinine) and non-smokers exposed to SHS (<3 ng/mL serum cotinine) and thus the biological plausibility of the earlier epidemiological observations, a essential element for a causal link between SHS and disease [16]; the cut-off between passive and active smokers sometimes shifts towards larger values in periods or environments with higher SHS levels [15,105]. (3) quantify the level of exposure to air smoke toxicants through air markers and their intake through biomarkers in various locations and conditions and show its increase with the duration of exposure. Both are very different in various areas and both are the foundation for regulations and bans [42,45,104]. (4) show the possibility of "third hand smoke" due to sorption and desorption of nicotine from indoor surfaces, etc., while a limited contribution to SHS can result from retention of smoke in the lungs and its expiration after the last puffs [54] and also from infiltration from neighboring units [55] or from outdoor air [92]. (5) show the deficiencies of mechanical measures in controlling SHS (e.g., ventilation, non-physically separated areas for smokers and non-smokers) [10,44,46,83]. (6) objectify the control of SHS obtained by the implementation of total bans of smoking in public places, and the resulting reduction in prevalence of exposure in many Western countries [9,10,58,83,84,86,88,95,97,98,103,107,108], without a corresponding increase in SHS exposure outside the workplace [109]. (7) show in some countries the good level of implementation of the bans [87,88,90,100,106,108], which was unexpectedly satisfactory, even in some Mediterranean countries [93], but as the implementation sometimes decreases with time, also the necessity of its persistent control. (8) demonstrate that partial bans reduce SHS exposure much less than total bans and that stringent legislation is needed (e.g., bans of smoking rooms, of smoking in terraces outside cafes or restaurants whether covered or not, in airports, in discotheques, etc.) [41,43,64,68,92,99,105,110]. (9) show that in the private domain (home), currently the main location of exposure to SHS, even if it is decreasing [72] (but not everywhere [109]), individual voluntary measures of protection are needed and complete control is obtained only with a total non-smoking policy at home or, but with lower efficacy, by allowing smoking exclusively outside the home, in order to protect children, who are particularly sensitive to SHS [53][54][55]75,79,111]. In cars, levels of nicotine and PM 2.5 are high and not fully controlled by ventilation. They remain elevated even in secondhand cars sold by former smoking owners [47][48][49][50]. (10) demonstrate some health effects of SHS. All cause-and cardiovascular deaths are linked to cotinine levels [113,114]. The prognosis of coronary heart disease is negatively influenced by the levels of cotinine due to SHS exposure [115], evolving together with the levels of fibrinogen, homocystein and CRP (all correlated with the coronary risk) [113,118]. The increases in those markers of cardiovascular risk are much larger than the increase in cotinine. This can explain why the difference in coronary risk between passive and active smokers is much lower than what could be expected from their differences in toxin intake [117]. SHS exposure is thus an important avoidable cause of cardiovascular disease.
Negative short-term effects of experimental exposure to SHS [119] and of chronic SHS exposure [120] on respiratory functions are observed, but the minimal positive influence of a smoking ban on spirometric values is controversial, while a decrease in cough and phlegm is more evident [91,105,121] at least in non-smokers [85]. In established COPD, the severity of symptoms is increased by SHS exposure [122][123][124].
Spontaneous abortions are more frequent among SHS-exposed non-smoking pregnant women than among unexposed non-smokers [132]. The prevalence of fetal death and preterm delivery are linked with biomarkers of exposure [134,135]. The early discontinuation of breastfeeding is more frequent when blood cotinine levels are in the range of SHS exposure [133]. The link between sudden infant death syndrome and exposure to SHS was already demonstrated by earlier studies [2].
There is limited biologically validated evidence of the influence of SHS on cancer prevalence [136], due to the long delay between exposure and disease, but the observed increase in biomarkers of carcinogenicity [34,36,63,70,78,104] demonstrate the biological plausibility of a link between SHS and cancers, previously observed in epidemiological studies.
Higher odds of psychological distress [138] and of cognitive impairment [139] are also associated with increased cotinine levels. Genotoxicity of SHS is shown by the presence of carcinogen metabolites and of DNA adducts in blood and other tissues and by the presence of micronuclei in bone marrow or peripheral blood of non-smokers exposed to SHS [140].
In spite of the above mentioned limitations, given the consistency of the recent results (as opposed to some contradictions in the earlier periods), there is overwhelming evidence of local variations in exposure to SHS, of reductions in exposure after recent smoking bans, and of the effects of SHS on health parameters.