3. How can tobacco products and their ingredients be assessed?
- 3.1 What criteria for assessment should be used?
- 3.2 What methods can support assessment?
- 3.3 Are current methods adequate?
3.1 What criteria for assessment should be used?
The SCENIHR opinion states:
In human studies there are clinical criteria for dependence (e.g. diagnostic and statistical manual of mental disorders (DSM), difficulty in quitting), laboratory measures of self-administration (e.g. neurobiological measures) and smoking frequency and depth of inhalation, as well as preference studies. These criteria indicate that tobacco in humans has a high addictive potential, but they have limitations when assessing the addictiveness of individual additives in the final tobacco product. There is no widely-agreed universal standard for human studies and as a result various possible endpoints exist. An addicted individual can be considered as someone who is suffering from a specific set of chronic conditions related to a modification of the regulation of their neural networks. It is the potential to induce these modifications which should be the criteria used to define the addictive potency of a product.
In animal studies the reinforcing potency of a drug is used as a criterion for addictive potential. Self-administration studies indicate that the abuse liability of pure nicotine is weaker than the addictive potential of tobacco products in humans. At present it is not possible to evaluate whether additives increase the addictive potency of the final tobacco product. Drugs of abuse such as nicotine induce different types of behavioural and neurochemical dysregulations in animal studies but no consensus about which of those are directly related to the addiction process in humans has yet been attained among scientists.
In conclusion, the criteria for defining dependence indicate that tobacco is highly addictive in humans. Animal studies that use intravenous administration indicate that the abuse liability of pure nicotine is weaker than the addictive potential of tobacco products in humans. In contrast to additives, the combustion-product acetaldehyde has been widely investigated in animals.
3.2 What methods can support assessment?
The SCENIHR opinion states:
Several animal models are available to study particular responses that are related to nicotine addiction. Thus, predictive models are available in animals to evaluate the development of nicotine tolerance and physical dependence as well as the rewarding/reinforcing effects produced by nicotine. The animal methods currently used to evaluate nicotine addictiveness are mainly based on the evaluation of its rewarding/reinforcing properties. New complex behavioural models that resemble the main diagnosis for drug addiction in humans have been developed very recently. However, these new models can only be applied for some particular drugs and are not yet available for nicotine addiction.
Experimental models to evaluate the development of nicotine tolerance and physical dependence.
Long-term consumption of nicotine produces adaptive changes in the central nervous system leading to the development of tolerance and physical dependence that can be easily evaluated in animal models. Thus, chronic nicotine administration produces tolerance to most of its pharmacological effects (Benowitz 2008). Tolerance to several nicotine responses such as hypolocomotion, convulsive effects, hypothermia or antinociception has been widely described in animal models, whereas an absence of tolerance to the effects on cognitive processes has been currently reported in these studies (Benowitz 2008, Collins et al. 1988, Damaj and Martin 1996, Marks et al. 1986, Miner and Collins 1988).
In humans, cessation of tobacco intake precipitates both somatic and affective symptoms of withdrawal which may include severe craving for nicotine, irritability, anxiety, loss of concentration, restlessness, decreased heart rate, depressed mood, impatience, insomnia, increased appetite and weight gain (Hughes and Hatsukami 1986, Hughes 2007). In rodents, nicotine withdrawal is also characterised by the manifestation of both somatic signs and affective changes similar to those observed in humans. The somatic signs include teeth chattering, palpebral ptosis, tremors, wet dog shakes, and changes in locomotor activity and other behavioural manifestations (Malin et al. 1992). Although the development of nicotine tolerance and physical dependence is concurrent to the development of addiction, they are not aetiologically related to nicotine addiction (Volkow and Li 2005). However, the affective manifestations of nicotine withdrawal seem to play an important role in the maintenance of the nicotine addictive process. These manifestations can be evaluated in rodents by measuring several emotional symptoms such as increased anxiety, aversive effects and reward deficits (Jackson et al. 2008b, Johnson et al. 2008). The aversive manifestations of withdrawal are mainly evaluated in rodents by using the place conditioning paradigm, whereas the associated reward deficits are currently evaluated using intracranial self-stimulation techniques. Both behavioural paradigms have also been extensively used to evaluate nicotine rewarding effects and will be described in the next section.
Experimental models to evaluate nicotine rewarding effects
Drug consumption is promoted and maintained by the rewarding properties of the drug. However, it is important to underline that drug consumption is a requirement for the development of addiction, although addiction is not a necessary consequence of drug intake.
Self-administration methods are widely used to directly evaluate the reinforcing properties of a drug. The procedures are considered by most researchers to be valid and reliable models of drug consumption in humans, and to have a high predictive value. It is assumed that the neurobiological mechanisms involved in drug self-administration in animals are similar to those underlying drug-intake in humans (Sanchis-Segura and Spanagel 2006). Self-administration methods can be classified considering the route of administration and the behavioural paradigm. From a behavioural perspective, these methods can be classified as operant and non-operant procedures. Non-operant paradigms are centred on the amount of drug consumed whereas the operant procedures require a conditioned response in order to obtain the drug, and the analysis of this response provides valuable information about different behavioural aspects of drug consumption. Non-operant paradigms in animals are mainly restricted to oral self-administration and they are very useful for alcohol research considering the similarities with the route of alcohol consumption by humans. The use of the appropriate route of self-administration for each drug of abuse provides an additional source of validity to these animal models, and these non-operant paradigms are therefore not useful in evaluating nicotine rewarding effects. The use of operant models is based on the learning contingency defined as “positive reinforcement”. In these models, the drug constitutes a positive reinforcer that is delivered contingently to the completion of the schedule requirements (Sanchis-Segura and Spanagel 2006). The operant chambers are equipped with one or more manipulandi, transmitting the operant response and devices to deliver the drug (reinforcer). Usually, there is an active manipulandum that is linked to the delivery of the drug and an inactive one, which results in the delivery of the drug vehicle or lacks any programmed consequence. The programmes of reinforcement commonly used are the fixed ratio and the progressive ratio schedule and the animal species currently used for nicotine self-administration is the rat. It is suggested that fixed ratio schedules measure the pleasurable or hedonic effects of a drug (McGregor and Roberts 1995, Mendrek et al. 1998), whereas progressive ratio schedules are more related to motivation and provide a better measure of incentive salience or craving (Arnold and Roberts 1997). Under a fixed ratio schedule, the drug is delivered every time that a pre-selected number of responses are completed. For nicotine self-administration, the number of responses required to obtain the drug is generally kept low, and the most used is the fixed ratio 1 (a nicotine delivery after each response in the active manipulandum), although fixed ratio 3 and 5 schedules of reinforcement have also been used (for instance, Shram et al. 2008). Multiple studies have demonstrated that rats easily maintain an operant behaviour to self-administer nicotine under these fixed ratio experimental conditions (Maldonado and Berrendero 2009). In contrast with other drugs of abuse, when dose-response curves have been constructed for nicotine self-administration in rats, they have been relatively flat or inverted U-shaped, which may be because of the aversive effects and toxicity associated with high doses of nicotine (Corrigall and Coen 1989, Shoaib et al. 1997). In a large number of studies the dose of 0.03 mg/kg (free base) per infusion showed very robust self-administration behaviour in rats (Corrigall and Coen 1989, Donny et al. 1999, Shoaib et al. 1997).
Under the progressive ratio schedule, the response requirement to deliver the drug escalates according to an arithmetic progression. The common index of performance evaluated in this schedule is the break point defined as the highest number of responses that the animal accomplished to obtain a single delivery of drug. In rats, several studies have also revealed that nicotine can maintain self-administration on a progressive ratio schedule of reinforcement. The break point achieved for nicotine self-administration has been compared by the authors with other drugs of abuse. They found that it was lower than the final ratio obtained for cocaine under an identical schedule of reinforcement, higher than that reported for heroin under similar progressive ratio schedule, and slightly lower than heroin when a slowly accelerating schedule was used (Donny et al. 1999). However, comparison across studies and drugs is difficult due to procedural differences in training parameters, sequence of progressive reinforcement or degree of drug dependence (Stafford et al. 1998). Increasing doses of nicotine usually resulted in a more linear increase in the performance in the progressive ratio schedule than in the fixed ratio schedule (Donny et al. 1999). The maximum break points usually reached by the adult rats when using the progressive ratio schedule are around 50 responses to obtain a single nicotine injection (Shram et al. 2008). Interestingly, higher break point values were obtained in adolescent rats (around 95) than in adult rats (Shram et al. 2008).
Operant nicotine self-administration has been difficult to establish in mice. A recent study has reported the validation of a new reliable operant model of nicotine self-administration, extinction and relapse in mice. This model was developed in C57BL/6 mice which are particularly sensitive to the behavioural effects of nicotine (Martín-García et al. 2009). Mice were successfully trained to self-administer a dose of nicotine similar to that previously used in rats (0.03 mg/kg, free base) under a fixed ratio 1 schedule of reinforcement. An inverted U-shaped dose-response function was also obtained using mice to self-administer different doses of nicotine (Galeote et al. 2009). Similar to other drugs of abuse, the break point achieved for nicotine self-administration in mice was lower than in rats. Indeed, the maximum break point (27 responses to obtain a single nicotine injection) was reached by the mice when using the dose of nicotine of 0.042 mg/kg (free base) (Galeote et al. 2009).
Conditioned preference paradigms
In the conditioned preference paradigms, the subjective effects of the drug are repeatedly paired to a previously neutral stimulus. Through this repeated conditioning process, this stimulus acquires the ability to act as a conditioned stimulus, and the animal will prefer or avoid this conditioned stimulus depending on the rewarding or aversive effects produced by the drug. The most commonly used paradigms apply a spatial environmental stimulus as conditioned stimulus and the animal will show a conditioned place preference or a conditioned place aversion for the environment associated with the effects of the drug or its withdrawal. Although a conditioned approach/avoidance towards specific stimuli can also occur in humans as a result of drug consumption (Bardo and Bevins 2000), the place conditioning paradigms are not primarily intended to model any particular feature of human behaviour. These paradigms mainly represent an indirect assessment of the rewarding or aversive effects of a drug or its withdrawal, by measuring the response of the animal towards the conditioned stimulus. Drugs of abuse display a differential ability to produce conditioned place preference. Opioids and psychostimulants easily produce robust place preference over a wide range of experimental conditions, whereas other drugs such as ethanol, cannabinoids or nicotine produce more inconsistent results (Sanchis-Segura and Spanagel 2006). Thus, nicotine has been shown to induce in rodents conditioned place preference across a wide range of doses in some experiments, although inverted U-shaped dose-response curves have been often reported, and the magnitude of the effect is generally small and affected by environmental stimuli or previous handling history (Castañé et al. 2006, Forget et al. 2005, Grabus et al. 2006, Le Foll and Goldberg 2004). Nicotine also produced aversive effects when used at high doses in some, but not all, studies (Grabus et al. 2006, Le Foll and Goldberg 2004). These results suggest that the rewarding effects of nicotine may be weaker than other drugs of abuse in this particular experimental paradigm (LeFoll and Goldberg 2004). Interestingly, sex differences were clearly revealed in mice exposed to nicotine in the conditioned place preference paradigm. Thus, female mice responded more to the conditioned rewarding effects of nicotine compared with males (Isiegas et al. 2009).
Intracranial self-stimulation paradigms
Intracranial electric self-stimulation procedures were essential in the discovery of the brain reward circuits (Olds and Milner 1954) and are now widely used to study the effects of drugs of abuse in the activity of the reward circuits (Sanchis-Segura and Spanagel 2006). In this paradigm, animals are trained to maintain an operant behaviour in order to obtain an electric pulse through an electrode that has been previously implanted in a reward-related brain site, most frequently the lateral hypothalamic area. The threshold of the minimal current needed to promote intracranial electric self-stimulation is estimated. A drug that stimulates the reward circuit will decrease this threshold, which would be related to its rewarding properties, whereas a drug having aversive effects will enhance the minimal current required to maintain the self-stimulation (Markou and Koob 1993). Nicotine as well as other drugs of abuse such as psychostimulants, opioids or ethanol, reduces the threshold to promote intracranial electric self-stimulation in some reward brain areas (Huston-Lyons and Kornetsy 1992, Kornetsky and Bain 1992, Wise 1996). Therefore, this behavioural paradigm clearly demonstrates the capability of nicotine to activate the brain reward circuits.
Experimental models to evaluate nicotine addiction
The behavioural models available to evaluate drug rewarding effects have been very useful in clarifying the neurobiological basis of drug taking. However, addiction is not just the taking of drugs, but represents a relapsing disorder characterised by compulsive drug use maintained despite adverse consequences for the user (APA 1994). Behavioural models that resemble the main diagnosis criteria for addiction are difficult to validate in animals. Recently, two independent research groups have validated behavioural models of compulsive drug seeking in rodents that resemble addictive behaviour in humans (Belin et al. 2008, Deroche-Gamonet et al. 2004, Vanderschuren and Everitt 2004). In these models the authors have evaluated the difficulties in stopping drug use by measuring the persistence of drug seeking during a period of signalled non-availability. The extremely high motivation of the addicts to take the drug has been evaluated by using a progressive ratio schedule where the number of operant responses to obtain a single drug injection was increased progressively within the same session. The maximal amount of work that the animal performs before cessation of responding (referred to as the break point) is considered a reliable index of the motivation for the drug. These new animal models of addiction report a break point over 500 to obtain a single cocaine injection in “addict rats” (Deroche-Gamonet et al. 2004). In these new animal models of addiction, the continued use of the drug despite its harmful consequences has been resembled by the persistence of the animal’s responding for the drug when drug delivery was associated with a punishment.
However, these models validated for cocaine consumption are still not available for other drugs, such as nicotine. Indeed, nicotine self-administration has not been reported to be maintained when drug delivery was associated with a punishment. In addition, only moderate break point values were obtained when a progressive schedule of reinforcement was used for nicotine self-administration. Thus, the maximum break points usually reached to obtain nicotine, i.e. around 50 responses in adult rats (see for instance, Shram et al. 2008) and around 95 in adolescent rats (Shram et al. 2008), are far away from the break point values (over 500) reached to obtain cocaine by the “addicted rats” (Deroche-Gamonet et al. 2004). In contrast, recent advances using animal models of relapse have shown that nicotine seeking after extinction of the operant behaviour can be triggered in rats and mice by nicotine-associated (conditioned) cues (Caggiula et al. 2002, Liu et al. 2007, Martín-García et al. 2009), stressors (Bilkei-Gorzo et al. 2008, Buczek et al. 1999) (e.g. mild footshocks) and re-exposure to the previously experienced drug (Chiamulera et al. 1996, Dravolina et al. 2007, Shaham et al. 1997), which are the same events that trigger nicotine craving and relapse in humans. Nicotine-paired cues have a critical role in sustaining nicotine self-administration after prolonged periods of abstinence and in maintaining smoking behaviour in humans. Indeed, a critical role of the environmental stimuli previously associated with drug consumption has been attributed when explaining the high rate of nicotine relapse (Caggiula et al. 2001, Caggiula et al. 2002, Liu et al. 2007). In agreement, the exposure to the associated cues was the most effective stimulus reinstating nicotine-seeking in mice, whereas stress exposure reinstated nicotine-seeking behaviour in half of the mice, and a priming injection of nicotine only reinstates seeking behaviour in a low percentage of mice (Martín-García et al. 2009). The neurobiological mechanisms involved in the processes underlying relapse to nicotine seeking are poorly understood. Further studies will be required to clarify the mechanisms involved in nicotine relapse using these animal models now available.
Tobacco addiction is maintained by nicotine, and tobacco products that do not deliver nicotine do not sustain addiction. However, it is important to distinguish between the stages of tobacco use, from early experimentation and initiation (prior to the development of dependence), through to regular use (and possible dependence) and possibly eventual cessation. Therefore, nicotine and additives may play different roles, or may differ in their relative importance during experimentation and initiation compared with the progression to regular use. In addition, the role of additives will differ according to whether the tobacco is delivered as a smoked or smokeless product.
Smoking and inhalation into the lungs, in particular, is a highly efficient form of nicotine administration, as the drug enters the circulation rapidly through the lungs and moves into the brain within seconds. This also allows precise dose titration, so a smoker may obtain the desired effects (Benowitz 2008). Therefore, additives and design characteristics which require the inhalation of tobacco smoke will be associated with increased dependence potential, and this will be particularly true when inhalation into the lungs (as opposed to the oral cavity only) is encouraged. In addition, various tobacco additives and flavourings can modulate the impact of nicotine, including via administration and inhalation behaviour. The impact of these additives on the attractiveness and palatability of tobacco products, in particular in naive users, may influence initiation of use and progression to regular use, before dependence is established.
Tobacco dependence is operationalised in multiple ways, but all definitions share core features of tolerance and withdrawal symptoms in relation to tobacco use. Most studies use either the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria for tobacco dependence, or a proxy measure such as the Fagerström Test for Nicotine Dependence (FTND), or the number of cigarettes smoked per day. However, the number of cigarettes smoked per day is often a poor measure of dependence, given the substantial inter-individual variability in the amount of nicotine extracted from a cigarette. The majority of the variance in scores on the FTND is accounted for by the first item (“How soon after you wake do you have your first cigarette?”), and it is likely that many dependent cigarette users can be identified by how soon after waking they smoke their first cigarette. This is most likely due to the short half-life of nicotine, which means that after a period of sleep most tobacco users have very low levels of circulating nicotine, resulting in withdrawal symptoms which are rectified by tobacco use.
Human behavioural studies require either subjective or objective measures of the effects of tobacco, and this allows a comparison of these effects between tobacco products which do and do not contain specific additives. Subjective measures include self-report measures of mood and craving, which may be as simple as single visual analogue scale measures of liking (e.g. “How much do you like the taste of this cigarette”), or include validated questionnaire measures (e.g. the Positive and Negative Affect Schedule). The latter refers to a range of laboratory assessments, including actual smoking behaviour through smoking topography measurement, which allows the detailed measurement of number of puffs taken per cigarette, depth of inhalation, inter-puff interval, and so on. This may also include self-administration or cigarette choice paradigms (e.g. presenting participants with two cigarettes, only one of which contains an additive of interest, to determine which is preferentially smoked), which are more closely comparable with paradigms used in animal studies. These measures are generally impossible or impractical to collect in survey studies, although the rates of use of different tobacco products, containing different additives, may allow their attractiveness to certain sub-groups (e.g. defined by age or ethnicity) to be inferred. A further complication is the possibility that what constitutes an attractive or palatable product may be culturally or ethnically specific.
Methods to assess attractiveness
According to the World Health Organisation (WHO), the terms “attractiveness” or “consumer appeal” refer to factors such as taste, smell and other sensory attributes, ease of use, flexibility of the dosing system, cost, reputation or image, assumed risks and benefits, and other characteristics of a product designed to stimulate use (WHO 2007b).
Overall, attractiveness is likely to be influenced by a subtle array and interaction of any number of these factors, although at certain times individual factors may take precedence (e.g. price, particularly during a recession). In addition, certain factors might be essential for enduring attractiveness (e.g. the presence and ease of delivery of nicotine).
The factors influencing attractiveness can be broadly divided into: extrinsic factors (e.g. marketing, packaging, pricing); and intrinsic factors (e.g. taste, smell, sensory attributes, and pharmacological factors). Additives play a role mainly in the intrinsic factor category, but marketing and packaging can also reflect the presence of additives in a way to attract and maintain customers (e.g. by signalling that the tobacco product contains menthol). Given the subtle interactions between different factors however, identifying and measuring the influence of individual addictives on attractiveness of products is difficult. Separating the role of additives in enhancing addictiveness, from their role in enhancing other attractive attributes of a tobacco product is also complex.
There are two main ways of examining the influence of additives on the attractiveness of a product. Firstly, one can assess individual tobacco products, and compare their attractiveness on a number of scales/dimensions, against other tobacco products. By then examining what is known about the additive content of these products, indirect inferences can be made as to the role of individual additives in the overall attractiveness of the product, although there are important limitations to studies of this kind. Secondly, one can examine the influence of individual additives on attractiveness of a tobacco product, along a number of scales, by experimentally adjusting tobacco products to include or exclude individual additives and testing responses to them. In addition, the quantity of the additive can be varied to assess dose response and whether there is a threshold below which any impact is not observed. However, in practice this may be difficult to achieve by research groups outside of the tobacco industry, who are likely to lack the resources to manipulate additive content in this way.
Tobacco industry documents show that the tobacco companies frequently tested human smokers on their reaction to different cigarettes using focus groups, market testing, human smoking behaviour studies or consumer panels. For example, one study carried out by British American Tobacco in 1980 exposed a panel of smokers, trained to be objective in their evaluation of cigarettes, to different conditions wherein brand identification was either masked or visible, in order to understand how brand identification and imagery affected subjective evaluation of cigarettes (Ferris 1980). The difficulties with this type of research are that ethical restrictions will usually preclude human testing of different tobacco products, particularly among non-users or children. In addition, there are technical constraints on the ability to manipulate the presence or absence of specific additives in tobacco products. While the tobacco industry may be able to achieve this, such manipulations may be beyond the resources of independent academic research groups.
Both the main methods have advantages and disadvantages and should be seen as complementary. Ideally, a variety of methods and tests would be utilised and assessors would be looking for overall consistency in the findings, in order to conclude that an additive affected attractiveness.
Measuring attractiveness of different brands
Actual brand choices
Assessing actual brand use gives an overall indicator of attractiveness of products which reflects all the factors listed at the outset of this section covering both extrinsic and intrinsic variables, of which additive content is only one factor. A major difficulty of this approach will therefore be separating the influence of these factors. The largest influence is likely to be the marketing budget. For example, the popularity of Marlboro worldwide is likely due to the substantial funding spent on its advertising and promotion. A further complication with interpretation of brand preference data over time is that the tobacco industry has been expanding the number of variants of existing brands; since 1998 brand families have increased by more than 50%. For example, Benson & Hedges increased the number of brands from four in 1998 to 12 by 2008 (ASH 2010).
Brand choices can be examined cross-sectionally in populations (nationally and globally) but longitudinal data enable trends in brand preferences to be examined over time and in relation to changing product make up (content and design) as well as tobacco control policies and other factors. Brand preferences should be examined in subpopulations such as by gender, age, and sociodemographic factors, which might reflect targeting by tobacco companies. Brand preferences in younger age groups (e.g. 11-16 year olds) are especially important to identify as these can enable an assessment of attractiveness and appeal to children. In particular, it is important to assess which brands are used initially by children, followed by those that they progress onto over time. Products that attract children to smoking have been referred to in the literature as “starter products”. This refers to two main types of products: confectionary products which are made and packaged to look like cigarettes, thereby enabling children to imitate smoking (e.g. candy cigarettes, not discussed further here), and tobacco products which are made to look like confectionary (e.g. candy-flavoured cigarettes), thought particularly to appeal to children and ethnic minorities (Connolly 2004).
Comprehensive sources of data on brand preferences at country level broken down by socio-demographics are not readily accessible. As an example, we have selected data from the UK which suggest that brand preferences of children and adults can be quite similar. The top five brands in 2009 were identified as: Lambert & Butler King Size, Mayfair King Size, Marlboro King Size Gold, Benson & Hedges King Size Gold and Richmond King Size (Hegarty 2010). Comparable data are not available for youth from 2009 but in 2006, the most popular brands with 11-16 year olds were: Mayfair (58%), Lambert & Butler (56%), Richmond (45%), Benson & Hedges (28%) and Sovereign (23%) (Amos and Hastings 2009). Four of the brands were common to both adults and youth, and for each age group there was a dominance of economy brands. Trends over time indicate increasing popularity of economy over premium brands suggesting price may be playing a key role in current brand choices. As indicated in section 3.13.2, there may be a trend in the UK for preferring brands marketed as containing no additives, but this observation needs confirmation. Careful monitoring of brand preferences over time will be important for future research, as will disclosure by the tobacco industry of detailed product content information for all brands on the market.
Perceived brand preferences
By showing different brands to consumers, assessments can be made about how attractive the products are perceived to be. For non-tobacco users, responses will largely reflect extrinsic factors such as the packaging, but will also reflect their knowledge of experiences of others with the products. For users, such assessments also reflect knowledge and experience of using the products in addition. The role of additives therefore will need to be assessed and inferred alongside these other factors, assuming that differences in additives between the different brands are known. As stated above, this research involves examining the look of a pack, and its design and packaging.
Packages can be digitally altered experimentally to test the responses of the presence or absence of attributes (e.g. whether listing an additive such as menthol alters how people respond to the product). However, studies have shown that colours of packs quickly become associated with certain attributes; for example, one study in New Zealand found that green colouring indicated the presence of menthol (Peace et al. 2007). In these types of studies, different population groups should be compared to test if some products are more appealing than others. For example, one experimental study indicated that some adolescents had more favourable impressions of tobacco brands that featured cherry flavouring in the packaging (Manning et al. 2009). This type of research has now been carried out in a variety of settings (e.g. internet, supermarket, and mall intercept studies) and using a variety of qualitative and quantitative research techniques (Hammond et al. 2009a, Hammond and Parkinson 2009b, Manning et al. 2009). The products have been assessed along several attributes including their perceived attractiveness, harmfulness, ease of initiation or cessation. Standardised designs, methodologies and questions therefore exist which can be utilised to facilitate comparative analysis.
Sensory attributes to users and others
Consumer perceptions of sensory attributes such as taste or palatability, smoke irritation and odour, can also be useful for indicating differences in brands. Although there is likely to be some impact of packaging and design on expectations of sensory effects, this area of testing will be more focused on attributes of the content and emissions of the product itself. This research can be done in two main ways: Through surveys of smokers in which questions cover reasons for selecting the brands they smoke and the role of sensory attributes.
Experimentally, using panels of test subjects trying products and expressing preferences using, for example, visual analogue scales (see section 3.10). However, whilst perceived responses to these attributes are important, it is also useful to see how sensory differences translate into topography measurements and the presence of biomarkers, such as cotinine (see below).
These factors could be attractive to a smoker as they make it less troublesome for others in their presence, who are then less likely to complain about their smoking. The sensory attributes to be measured here would include smoke irritation, smoke odour, and visibility of sidestream and mainstream smoke. These assessments can be made as described above, but of non-smokers who live, work or are in the presence of smokers.
3.3 Are current methods adequate?
The SCENIHR opinion states:
Many different methods are used in humans, but there is a lack of consistency between these methods. Human studies have many limitations in design (e.g. the use of conditioned cues and the need to work with smokers). Furthermore, ethical issues may arise when testing substances in humans.
There is currently no animal model to assess the addictive potency of the final tobacco product; however, pure nicotine has been studied extensively.
The methods currently used in animals to evaluate the addictiveness of any drug of abuse, including nicotine, are mainly based on the evaluation of the re-inforcing properties of the drug. These experimental animal models are mainly based on self-administration protocols in rodents, usually rats. The model with the highest predictive validity is the operant self-administration paradigm. A response which is easy to evaluate is the break point. This is defined as the highest number of responses that the animal completes in order to obtain a single delivery of a drug. A higher break point represents a direct measure of the motivation of the animal to obtain the drug and is often taken to imply an increase in the addictive potency of the drug. Other models have also been used, such as the intracranial self-stimulation and the conditioned place preference paradigms. New complex behavioural models that resemble the main diagnosis for drug addiction in humans have been developed very recently, although these new models can only be applied for some particular drugs and experimental conditions at the present moment. The methods have additional limitations as in animal studies pure nicotine is injected intravenously and shows only a weak addictive potential whereas in humans tobacco is used differently (e.g. inhalation, oral consumption). The operant self-administration paradigm has been widely accepted as a reliable animal model with high predictive value for the abuse liability of a drug and therefore, possibly also for its addictive potential in humans. However, a consensus between scientists has not yet been attained on whether this method, which is appropriate to define the abuse liability, would also be the most suitable method to define the addictive potential of a drug.
In conclusion, there are many methods for assessing the addictive potency of a substance in humans, but they have limitations in design and ethical issues may arise. Animal studies using self-administration protocols evaluate the reinforcing properties after intravenous injection of the drugs but there is no consensus concerning the most suitable method for defining the addictive potential. The current methods can thus not be considered adequate.
Conclusions on methods to assess attractiveness
Attractiveness depends on multiple factors that combine to stimulate use. These include extrinsic factors such as marketing, packaging and price, and intrinsic factors such as taste and smell. It is very difficult to identify the role of individual additives in enhancing addictiveness or enhancing other attractive attributes of tobacco products. The attractiveness of a product may be assessed by the direct comparison of different products by surveys, experimental measures or human testing. Another way to examine the attractiveness of individual additives is to test a certain tobacco product by introducing the additive in different doses. When additives are thought to act in synergy, they may be tested together. In practice, however, overall attractiveness is assessed by comparison of brand choice in subpopulations according to gender, age and sociodemographic factors. By showing different brands to consumers, assessments can be made about their perceived attractiveness. Sensory attributes such as taste, irritation etc. may be tested by surveys of users or experimentally on panels of test subjects.
Animal models do not currently exist to allow the assessment of attractiveness. There are two main ways of examining the influence of additives on the attractiveness of a product which have largely been conducted by tobacco industry.
The first is to assess individual tobacco products and compare their attractiveness against other tobacco products on a number of scales/dimensions. By then examining what is known about the additive content of these products, judgements can be made as to the role of individual additives in the overall attractiveness of the product. This can be done using a variety of research methods, such as panel studies and surveys, experimental measures and human testing.
The second is to examine the influence of individual additives or combination of additives on attractiveness of a tobacco product, along a number of scales, by experimentally adjusting tobacco products to include or exclude individual additives and testing responses to them. In addition, the quantity of the additive can be varied to assess dose response and whether there is a threshold below which any impact is not observed.
The difficulties with this type of research include ethical considerations that will usually preclude human testing of different tobacco products, particularly among non-users or children.
In conclusion, it is only possible to assess attractiveness in humans, and this may be done by comparison of different products used or by adjusting tobacco products experimentally. However, such studies in human subjects are difficult to carry out due to ethical considerations and the current methods are thus not considered adequate for a reliable quantification of attractiveness in humans.