The Discursive Construction of Risk and Trust in Patient Information Leafl ets

There is wide recognition that the communication of risk in Patient Information Leafl ets (PILs) – the instructions that accompany medications in Europe – problematises the reception of these texts. There is at the same time growing understanding of the mediating role of trust in risk communication. This paper aims to analyse how risk is discursively constructed in PILs, and to identify and analyse discourses that are associated with trust-generation. The corpus (nine PILs chosen from the British online PIL bank, www.medicines.org.uk) is analysed using Foucauldian (1972) discourse analysis: specifi cally, this involves identifying the functions of the statements that constitute the discourses. A discourse analysis of the corpus of PILs reveals that the discourse of risk revolves around statements of the potential harm that may be caused by taking the medication, whilst trust is constructed through three discourses: the discourses that relate to competence and care, in accordance with the trust theories of Poortinga/Pidgeon (2003) and Earle (2010), and a third discourse, corporate accountability, which functions to construct an ethical (trustworthy) identity for the company. This paper contributes to PIL literature in the following ways: it introduces a methodology that has not been used before in relation to these texts, namely, Foucauldian discourse analysis; it helps to identify the presence of trust-generating discourses in PILs; and analysing the discourses of risk and trust at statement-level facilitates a better understanding of how these discourses function in texts that are generally not well-received by the patients for whom they are intended.


Introduction
Patient Information Leafl ets (henceforth PILs) are legally required to accompany medicine available for purchase in Europe: they inform about side effects and give instructions on how to take the medication appropriately, information which patients say they want to receive (Kenny et al. 1998, Berry et al. 1997).Somewhat paradoxically, however, research shows that PILs are not well-regarded by the patients for whom they are written: despite attempts to improve PILs, including ongoing research and new EU legislation for compulsory user-testing in 2004, PILs are considered 'dysfunctional (…) due to very complex production and reception processes' (Askehave/Zethsen 2008: 171).Patients have criticised PILs for many reasons including a register that is too complex (Askehave/Zethsen 2000), poor presentation and layout (Berry 2006, Bernardini et al. 2001, Koo et al. 2003), and lack of user-friendliness (Askehave/Zethsen 2000).As a measure of how patients 'do not value the written medicines information they receive' (Raynor/Dickinson 2009: 701), many opt not to read them (Raynor/Britten 2001, Horwitz et al. 2009).The fact that some patients choose not to read PILs is problematic as the information provided is important for the health and safety not only of the patient taking the medication, but also for others, including foetuses (in the case of pregnant women), infants (in the case of breastfeeding mothers), and indeed the general public (where, for example, medication may affect one's capacity to drive safely).
PILs have been studied in two main ways: through reception studies, and with focus on the texts themselves.Within the latter stream, Askehave/Zethsen (2000), for example, when analysing Danish PILs, found that the complex word-choice and convoluted syntax of these texts made them diffi cult to understand for the lay receiver; Clerehan/Buchbinder (2006) devised a means of evaluating Australian PILs using the linguistic approaches of Systemic Functional Linguistics (SFL) -it was found in a follow-up reception study (Hirsch et al. 2009) that patients felt that PILs that were revised using the evaluative framework had been improved; Connor et al. (2008) compared American and Spanish PILs and found evidence of differences that were ascribable to the effects of two different sets of cultures and institutions.
The present paper contributes to this text-oriented stream of research into PILs by focusing on how risk and trust are expressed in these texts.Previous studies of risk in PILs have tended to focus on their reception, examining the effects of the communication of risk in PILs on the patient.It has been shown, for example, that communicating effectively about risk in these leafl ets is diffi cult due to the burden of negative information that risk messages must bear (Berry 2004: 5).Amery (1999: 122) points out that an unbalanced focus on side effects in PILs can lead to 'relatively negative information overload' and result in texts which patients describe as 'negative' (MHRA 2005: 18).A study by Knapp et al. (2010: 270) similarly found evidence that patients tend to over-estimate the risks associated with side effects in PILs.Trust, on the other hand, is a more neglected area of PIL research; it has been investigated more in terms of patients' mistrust of the sender, the pharmaceutical industry, in reception-oriented studies (see, for example, Consumers' Association 2000).
The regulatory agencies, the bodies that evaluate PILs, have explicitly called for further investigation into how the communication of risk in PILs could be improved.Thomas Lonngren, the former head of EMA (the European Medicines Agency, the body responsible for the regulation of PILs at European level), underlined the need to improve how risk is communicated in PILs, as it has the effect of 'scaring people' (Smith 2008) with the unintended result that patients do not take medication appropriately.Similarly, the Medicines and Health Products Regulatory Agency (MHRA 2005), the British government body responsible for regulating PILs, pinpointed the communication of risk as one of the areas of greatest weakness in these texts, calling for further research in this area.This paper can be considered a response to the problems associated with the communication of risk in PILs identifi ed by patients and researchers alike, and calls by the regulatory agencies for further research in this area.
Both risk and trust have been researched from a psychological perspective: trust has been defi ned as a psychological phenomenon, and risk has been defi ned as having psychological impact.The psychologising of risk is evident in PIL research where the emotive effects associated with the communication of risk in PILs are highlighted.In psychological models of risk reception, risk information is seen as input to the brain -especially playing a role in triggering neurological reactions in the part of the brain called the amygdala, also called the 'fear module' (Öhman/Mineka 2001: 483), in a stimulus-response system where risk information is the 'stimulus' and the receiver 'responds' physiologically.Functional magnetic resonance imagery experiments conducted by Isenberg et al. (1999) demonstrated that the amygdala is activated by 'linguistic threat' -in other words, risk 'input' can consist of words that are read by the receiver, which is a relevant fi nding for any risk communication that depends on text.In psychological models of risk reception, risk information is portrayed as having the potential to overwhelm or engulf 'normal' cognitive processes, revealing the limits of human rationality.
As risk 'may frighten and frustrate the public' (Slovic 2000: 184-185), risk messages are often accompanied by messages designed to engender trust.Trust has been considered a psychological state (Rousseau et al. 1998: 395); Giddens (1990: 92-100) defi nes it as an emotional state of ontological security.Trust can be considered a social lubricant (Luhmann 1988), reducing complexity and uncertainty, hence the aptness of its service to the mediation of risk communication where, again using the psychological perspective, complexity and uncertainty abound (Kasperson et al. 1992).Moreover, the nature of the relationship between risk and trust has been theorised from a psychological perspective.Eiser et al. (2002) postulated two models: the causal model in which the trust one has, or invests, (in a product, institution, company etc.) determines one's perceptions of risks and the acceptability of those risks; and the associationist model in which the acceptability of perceived risks determines trust.A study by Poortinga/Pidgeon (2005), which focused on the British public's perceptions of genetically modifi ed food, provides evidence that supports the associationist model.
Risk has not only been theorised from a psychological perspective, however.Writers such as Lupton (1999Lupton ( , 1993)), Turner/Tennant (2009) and McKay (2006), working within a social constructionist paradigm, posit that risks are discursively constructed.In this conceptualisation, risks are 'viewed as assemblages of meanings, logics and beliefs cohering around material phenomena, giving the phenomena form and substance' (Lupton 1999: 30).The description of risk as forms of meaning that constitute phenomena is reminiscent of Foucault's (1972: 49) defi nition of a discourse as 'practices that systematically form the objects of which they speak'.As risk within a social constructionist paradigm can be considered to be discursively constructed, it follows that risk is amenable to discourse analysis.
When it comes to trust, more 'translation' of framework (from psychological to discursive) is required than in the case of risk.This has something to do with the different ontic natures of risk and trust: risk is considered something that is external to individuals, whilst trust tends to be viewed as a state of mind.How then can trust be analysed discursively?The approach that is taken here is to examine how trust is generated.Trust has been conceptualised as the by-product of 'solidarity' and 'accuracy' (Earle 2010: 570).'Solidarity' relates to the relationship between sender and receiver where the truster is assured that the trustees have his/her best interests at heart; 'accuracy' relates to the trustees appearing knowledgeable in their fi eld of expertise.Poortinga/ Pidgeon (2003: 962) describe these facets of trust generation in the similar, related terms of 'competence' and 'care', where one is more likely to trust, say, an institution if it appears to be expert and knowledgeable in its fi eld, and if it is perceived to be showing genuine interest in the wellbeing of at-risk individuals or groups.Characterising trust in terms of these two parameters (competence/accuracy and care/solidarity) facilitates the necessary translation from a psychological framework to a discursive one, as discourses that are associated with competence and care may be identifi ed.Hence, the discourse analysis undertaken here relates to identifying discourses associated with trust-generation rather than trust itself.Of course, the fact that discourses that are associated with the generation of trust may be present in PILs does not imply that individual patients will 'trust' PILs: other methods, such as interviews, would be required to investigate that issue.What discourse analysis offers is a means of identifying whether risk and trust-generating discourses are present and what their salient features are.
The question that this paper pursues is: How are risk and trust discursively constructed in PILs?It engages in discourse analysis from the social constructionist point of view that language does not simply refl ect reality but creates it.The purpose is to understand the functioning of these texts better, given that the 'reality' that they create is often rejected by the patients for whom they are intended.

Corpus
The corpus under investigation consists of the nine PILs that went online on www.medicines.org.uk on the 22 nd March 2010.The reason for choosing this site is that it provides up-to-date PILs that have been validated either by the MHRA or EMA (in other words, a fair cross-section of PILs that the British public would be using, as PILs in Britain are validated either nationally by the MHRA or at European level by EMA).A random approach is more characteristic of quantitative research, but a 'random purposeful' selection has its place in qualitative research (Patton 2002: 240-241) where a snapshot of a very large corpus is required; the random approach also helps to add 'credibility' (Patton 2002: 241) in an explorative study such as this where a 'window' on the genre is required.The corpus consists of the following texts (see Table 1).

Foucauldian Discourse Analysis
Foucault 's (1972) approach to discourse as outlined in The Archaeology of Knowledge forms the basis of the analytic approach used here.His intention was to provide 'a method of analysis' (1972: 16) that was not about examining structure or subjectivity (structuralism and phenomenology are research paradigms which he rejects) but rather the discourses that constitute our understanding of the world, which mediate our access to the world.As individuals, we are 'subject' to these discourses as they tend to be taken-for-granted, defi ning what can be done and said (and not), as well as determining the limits of knowledge itself.Foucauldian analysis involves examining text closely, though Foucault insists that one must not confl ate discourse analysis with linguistic analysis (1972: 48).Language for Foucault represents the 'exteriority' of discourse (1972: 140).When language is examined in Foucauldian discourse analysis, it is in order that discourses may be scrutinised through the substance of language.Key concepts in Foucauldian analysis are discursive formations (Foucault's word for 'discourse' as a structured body of knowledge), and the statement (énoncé) which is the smallest indivisible building block or 'atom' of a discursive formation (Foucault 1972: 80).The statement is roughly equivalent to a sentence, but it is discursive rather than linguistic in identity, pointing backwards towards the rule of formation that generated it, rather than to objects in the world, as a sentence would.Moreover, as Foucault (1972: 49) considers discourses to be 'practices', a statement should be analysed in relation to its function -the discursive work it performs.In analysing the function of statements, Foucault (1972: 27) usefully provides the following question: 'how is it that one particular statement appeared rather than another?'This involves questioning what is 'special' about the function of the statement, and, as discourses in Foucault's discourse theory are linked to institutions (Foucault 1972: 41-42), such a functional approach to the statement also provokes the question of whose interests a statement serves.
This emphasis on the statement's function has an impact on the presentation of the analysis.The end-point of Foucauldian discourse analysis is a better understanding of discourses; it is not concerned with the linguistic expression of the statements in the discursive formations, or the frequency of the occurrence of similar statements in texts.In depicting the discourses of risk and trust generation, it is thus not necessary to include in the analysis all of the statements that relate to the same function, as this would lead to redundancy in the analysis.To illustrate this point, the statements: 'Like all medicines, Neurontin can cause side effects, although not everybody gets them' in the PIL for Neurontin, and 'Most people can use this medicine without any problems but sometimes you may notice some side effects' taken from the PIL for Boots Nic-Assist Fruit Fresh Gum, would be considered from a Foucauldian perspective to have the same function, namely, 'To inform about the possibility of side effects with the medication'.Thus, one example only of the statements that refl ect this function is included in the analysis (see Statement Nr. 3 in the Analysis section).Foucauldian discourse analysis is, as I have argued, a means of uncovering and describing the discourses that underlie and motivate texts, rather than having the language of the texts as the focus of analysis.
The methodology of Foucauldian (1972) discourse analysis presents the analyst with two main advantages.First of all, it examines and describes discursive formations on the basis of the individual statements in texts.The method is simple and elegant, in that a picture of discursive formations emerges on the basis of statements that have been spoken or are written down that relate to a particular fi eld (such as 'scientifi c expertise').Unlike Critical Discourse Analysis (CDA), Foucauldian discourse analysis does not consider it necessary to bring the optic of social theories or linguistic tools to bear on the analysis (see Fairclough 2003: 2-3): for Foucault, the elements of discourse are present in the statements; a functional analysis of statements allows the discourse analyst to 'describe' (Foucault 1972: 38) discursive formations.Besides its descriptive powers, Foucauldian discourse analysis provides a second advantage: an identifi cation of the functions of statements in a discursive formation facilitates the introduction of normative and evaluative stances.This is because after one has analysed the functions, one can then consider whether the rules of formation that generate the statements are fair and reasonable, or need to be challenged.Foucault (1972: 25-26) expresses this as follows: 'we must show that they [pre-existing forms of continuity] do not come about by themselves, but are always the result of a construction the rules of which must be known, and the justifi cations of which must be scrutinized'.The purpose of this paper is to map out the discourse of risk and the discourses associated with trust-generation -this relates to the fi rst advantage of Foucauldian discourse analysis (its descriptive/ analytical powers) outlined above.The results of this paper could, however, be taken and interfaced with communication theory or the fi ndings of best practice in PILs research (such as results from reception studies) for a more critical perspective.
A wide diversity of discourse methodologies characterises the study of health communication, giving rise to what Sarangi (2010: 413) describes as 'analytical eclecticism' in the fi eld.Arribas-Ayllon et al. (2011, forthcoming), for example, advocate the application of 'rhetorical discourse analysis' when investigating the negotiation of genetic knowledge in health settings.Måseide (2003) marries a Foucauldian approach (the idea of 'order of discourse') with Goffman's concepts of 'framing' and 'footing' when examining moral discourses in medical collaborative settings.Roberts/Sarangi (2005) propose a multi-method approach called 'theme-oriented discourse analysis' to examine how language is used strategically to construct professional practice.Taylor et al. (2011, forthcoming) link their narrative approach to the 'discursive turn' in the study of medicine and healthcare.As already noted, the aim of this paper is to chart the discursive landscape of PILs, specifi cally examining how risk and trust are discursively constructed in these texts.For this task, Foucault's descriptive approaches and his focus on discourses as they appear in texts make Foucauldian discourse analysis seem an appropriate and obvious choice.

Analysis
What follows are the fi ndings from a Foucauldian discourse analysis of the risk and trust generating discourses in the corpus of PILs.Foucauldian discourse analysis is utilized to unpack these discourses; it employs the concept of the 'statement' to reveal the functional components of these discourses.

Quote Source text
Statement Function

1.
Boots Nic-Assist Fruit Fresh Gum Read all of this leafl et carefully because it contains important information for you.
To inform patients to take the product with care, follow procedures.

Cozaar
It is important to tell your doctor before taking Cozaar: if you have had a history of angioedema….
To inform patients of the need to collaborate with their doctor to ensure the safety of the product.

3.
Neurontin Like all medicines, Neurontin can cause side effects, although not everybody gets them.
To warn about side effects and patients' need to be prepared to take appropriate action.

4.
Enalapril Maleate Breastfeeding new-born babies (fi rst few weeks after birth), and especially premature babies, is not recommended whilst taking enalapril tablets To remind patients that their responsibility for the safe administration of the medication extends to the health and safety of others.

Doribax
This medicine has been prescribed for you.Do not pass it on to others.
To remind patients of the dangers of sharing the medication.

6.
Cozaar Ask your pharmacist how to dispose of medicines no longer required.
To warn of the risks to the environment.
Table 2. Foucauldian analysis of the risk discourse The risk discourse revolves around the need for care due to the potential harm that the medication can cause.A Foucauldian discourse analysis of the statements in the risk discourse reveals that the patient is constructed as needing to take responsibility to ensure the safety of the product, as patients are urged to exercise caution with the medication (1), to act responsibly (2), as the medication can have dangerous side effects (3).The person most affected by the potential risks of the medication is the patient him/herself, but the medication may also be toxic for others, such as infants through breastmilk (4); others through the inappropriate sharing of medication (5); nature through inappropriate disposal of the medication (6).What emerges is a view of risk as polluting (to self, others, environment) that is reminiscent of Mary Douglas's (2002 [1996]) identifi cation of danger with impurity.And whilst Foucauldian discourse analysis is more concerned with isolating the main functions of the statements in the discourses rather than with indicating the frequency of certain statements, it should also be noted that the discourse of risk dominates the communication in PILs as the risk discourse appears in four of the six sections of a standard PIL, namely, Before you take X, How to take X, Possible Side Effects, How to Store X; the two sections not directly concerned with risk are What X is and what it is used for, and Further Information.

Trust Discourses
As noted earlier, risk messages tend to rely on trust, where the truster invests trust in the trustee: it can be considered a kind of relinquishing of the truster's autonomy to the trustee, even a leap of blind faith (Kierkegaard 1985(Kierkegaard [1843]]: 85).What follows is a discursive analysis of a range of discourses that refl ect Earle's (2010) and Poortinga/Pidgeon's (2003) conceptualisation of trust as relating to two qualities which the trustee should be seen by the truster as possessing: competence (accuracy) and care (solidarity).The inclusion of these trust discourses in PILs appears strategiceven 'rhetorical' (Foucault 1972: 59) -to make the prevailing risk messages more palatable.The main trust discourses that were identifi ed in PILs were the following: science/expertise (which relates to competence, and to some extent care); empathy (which relates to care); and corporate accountability (this relates to to the discursive construction of the ethical credentials of the pharmaceutical company).

Science/Expertise Discourse
The science/expertise discourse relates to the notion of competence, which, as noted earlier, has been recognised as being important in instilling trust.Findings from the analysis are presented in the following table (Table 3): Quote Source text Statement Function

7.
Reyataz These medicines control Human Immunodefi ciency Virus (HIV) by stopping a protein that the HIV needs for its multiplication.
To indicate the pharmaceutical company's scientifi c knowledge of the patient's disease.

Cozaar
The side effects of medicines are classifi ed as follows: very common: affects more than 1 user in 10… To indicate the pharmaceutical company's knowledge of the side-effects of the medication.

9.
Cozaar Cozaar has been studied in children.To underline that the pharmaceutical company has conducted the necessary clinical trials.

Enalapril Maleate
Enalapril tablets may prevent heart failure from getting worse in some patients who have symptoms.
To inform about the benefi ts of the medication.
Table 3. Foucauldian analysis of the science/expertise discourse In the present corpus, the science/expertise discourse is built on the basis of four statements that relate to biomedical and pharmacological knowledge (7), knowledge of the statistical likelihood of side-effects and their nature (8), evidence of clinical trials (9), as well as accounting for the benefi ts that patients can expect from the product (10).The science/expertise discourse contributes to the generation of trust as it projects the image of a professional company capable of producing a scientifi cally sound and effi cacious product -in other words, a trusted company that can produce a trustworthy product.By including statement (10) which revolves around the promise or hope of a treatment/cure of the patient's illness, the science/expertise discourse reminds the patient why they might want to trust the medication (take it despite possible side-effects), that the pharmaceutical company and the patient share a common goal of the patient's improved wellbeing and health: in other words, that the pharmaceutical company operates with the patient's interests at heart.This reveals one of the strategic functions of the statements of benefi t -to offset the negatives of the risk discourse.This statement relating to the benefi ts of taking the medication can furthermore be considered to bring together the trust-engendering elements of expertise and care, as it constructs scientifi c expertise as improving the condition of the patient, which in turn suggests the company's concern for the wellbeing of the patient.However, as noted earlier, the risk discourse dominates the discursive terrain of PILs.Statements associated with the scientifi c benefi ts of taking the medication appear in only one section of the PIL -entitled 'What X is and what it is used for', whereas the risk discourse appears in four.

The Empathy Discourse
The empathy discourse focuses on the emotional needs of the patient and is used communicatively to establish an empathetic relationship with the patient as a means of engendering trust.This relates to the notion of the pharmaceutical company as provider of care (see Table 4):

11.
Boots Nic-Assist Fresh Fruit Gum Because smoking is an addiction you may fi nd it diffi cult to give up.
To reveal the pharmaceutical company's understanding that it is diffi cult for the patient and refl ect their intention to encourage the patient.

12.
Fluoxetine Other medicines or psychotherapy can also help to treat these conditions.
To show their awareness that patients are different and may benefi t from different types of treatment

Fluoxetine
You may fi nd it useful to tell a friend or relative about your condition, and read this leafl et with them.
To remind patients of the benefi t of availing of their social network.

Enalapril Maleate
Please do not be worried, most people taking this medicine will not experience any problems.
To indicate the pharmaceutical company's awareness of patients' concerns about risks and remind them not to be unduly worried.

Table 4. Foucauldian analysis of the empathy discourse
The empathy discourse is expressed along two main trajectories in the present corpus.The fi rst has to do with the psychological effects of the condition in question.Both nicotine-addiction (Boots Nic-Assist Fresh Fruit Gum) (11) and depression (Fluoxetine) (12-13) have a psychological profi le -one must be motivated to overcome nicotine addiction, and there is a need for emotional support in relation to depression.However, the empathy discourse is not present in all the PILs in the sample: in the PILs for chronic illnesses (heart disease and HIV infection) represented in the corpus, it may be considered conspicuously absent, as research shows that chronic illness has a major emotional impact on patients and relatives (see Rolland 1984).
The second aspect of the empathy discourse is refl ected in the statement that relates to concerns about taking the medication (14).An explicit recognition of patients' anxieties about the possible effects of the medicine is represented in only two of the nine PILs in the corpus.This particular statement is interesting as it constructs and communicates the company's awareness that damage may be caused (in terms of undue anxiety) by the risk discourse that dominates the communication in PILs.

The Corporate Accountability Discourse
Discourse analysis reveals the presence of a third discourse (besides competence and care) that can be considered to generate trust, namely the discourse of corporate accountability, which serves to construct the ethical credentials of corporations.Despite the fact that PILs are highly regulated, standardised texts, some corporate self-consciousness is present, where a company's desire to create a positive impression of corporate accountability and reliability is evident.The discourse of corporate accountability has a Janus-faced quality, positioning itself both in relation to the trust needs of the patients (here, 'consumers') and instrumental in the construction of its own positive self-image.Christensen (1997: 197) argues that marketing is a form of 'self-referential communication practices through which the organization recognizes and confi rms its own images, values and assumptions; in short, its own culture.'We can therefore see in the corporate accountability discourse of PILs the construction of a looking-glass image of self that speaks both to the pharmaceutical companies themselves as well as to their audience.Analysis of the statements of the discourse of corporate accountability is presented in To represent the contractual relationship between the patient as consumer and the pharmaceutical company.

Boots Nic-Assist Fruit Fresh Gum
To request a copy of this leafl et in Braille, large print or audio please call, free of charge: 0800 198 5000 (UK only).
To indicate the pharmaceutical company's awareness that their customers may have special needs.

17.
Ranitidine Medicines should not be disposed of via wastewater or household waste.
To underline the company's 'green' credentials.

18.
Cozaar This medicinal product is authorised in the Member States of the EEA under the following names: ...
To indicate authorisation to sell the product.

Doribax
For any information about this medicinal product, please contact the local representative of the Marketing Authorization Holder: België/ Belgique/Belgium… To indicate the company's will to be transparent, encouraging questions and informationseeking.
Table 5. Foucauldian analysis of the corporate accountability discourse Of the nine PILs in this corpus, four employ a company icon such as in (15) -a sort of badge of identity of the pharmaceutical company, underlining the source of the medication and suggestive of some corporate pride.It is relevant to note here that for Foucault (1972: 84-86), statements were not necessarily lexical items: they could take other semiotic forms too.A commitment to accessibility for blind or partially-sighted patients is underlined in ( 16), which emphasises the company's ethical credentials.Another example of this strategy is the exhortation to dispose of the medicine responsibly to protect the environment (17) which eight out of the nine PILs in the corpus included.The fact that regulation is mentioned (18) emphasises that the product meets legal standards; transparency is underlined in (19).The perceived trustworthiness of the sender is an important element in risk judgements.So besides the discourses associated with competence and care (identifi ed by Poortinga/Pidgeon 2003; Earle 2010), another discourse used to generate trust is present in the corpus, namely, the 'corporate accountability' discourse, whose function is to construct discursively the ethical credentials of the company.

Discussion and Conclusion
In this paper, I have used Foucauldian discourse analysis to investigate the discursive construction of risk and trust in PILs.The aim of the analysis was to reveal how these discourses worked by examining the function of the statements that constitute these discourses.
This paper has contributed to research into PILs in a number of ways.Methodologically, it brings a new optic, Foucauldian discourse analysis, to bear on PILs, and as such may be considered a response to Payne (2002: 177) who called for greater use of discursive approaches to written health materials, particularly 'the deconstruction of text used in leafl ets using methods such as discourse analysis'.The descriptive advantages and evaluative potential of the Foucauldian approach to discourse have been described in this paper.
The study has also contributed to the fi eld empirically by conducting an examination of the statements within the discourses of risk and trust.Whilst there has been an understanding that risk problematised communication in PILs, no study prior to this has focused on examining what risk in these texts might be considered to 'consist of'.The discourse analysis conducted here anatomises the risk discourse in a corpus of PILs, at the same time as it identifi es trust as being constructed through discourses that relate to competence and care, in accordance with the trust theories of Poortinga/Pidgeon (2003) and Earle (2010), and reveals a third discourse, corporate accountability, which also contributes to the generation of trust.Indeed, one of the advantages of discourse analysis is that it has helped to uncover the less obvious discourses related to trust generation that are present in these texts and may have some impact on reception.This is important as there has been little investigation of trust as an aspect of PILs.There is, moreover, a growing interest in trust as a discursive phenomenon (see Candlin 2010): studying the discursive construction of trust in a variety of different settings and in written texts should contribute to a better understanding of this phenomenon.
Thirdly, this paper makes a theoretical contribution in its attempt to theorise how trust may be 'discursifi ed'.It is useful that the various discourses of trust-generation are understood better given the importance of trust in society (we recall Bachmann/Zaheer's (2006) characterisation of the 'trust society ' and Löfstedt's (2008) description of our world as 'post-trust').There is no doubt scope for the theorisation of trust to be developed further, hand-in-hand with more analysis, as well as different objects of analysis.
This paper has its limitations: it is mainly explorative, introducing a new methodology to a well-recognised problem (the poor communication of risk in PILs), investigating at the same time the affordances of Foucauldian (1972) discourse analysis.It will be recalled that Foucauldian discourse analysis can be used for descriptive and evaluative purposes.This paper undertook the descriptive analysis of discourse only, as it was outside the scope of this paper to introduce the second layer of Foucauldian discourse analysis, namely, evaluation.This one could do by critically examining the fi ndings of the descriptive analysis (conducted here) in relation to a health communication model, or research into best practice in document-writing of this kind.A follow-up reception study would no doubt also be valuable, where the results of the discourse analysis could, for example, provide questions or topics for a focal group study.In any case, before the discourses of risk and trust may be evaluated, they need to be anatomised and described.It is hoped that the present paper can engender debate and interest by showing the value of bringing Foucault's (1972) descriptive approach to discourse analysis to bear on a genre that has been recognised by so many different parties as being problematic.
This paper is positioned at the intersection between discourse and communication: its fundamental presupposition is that discourse analysis can contribute to communication research as it allows one to go 'backstage' on the basic presuppositions of discourses that may problematise communication which otherwise might seem beyond scrutiny.Not all communication problems are discourse problems, of course.Discourse analysts wishing to contribute to a fi eld such as communication need fi rst to establish that the problem they wish to investigate is amenable to discourse analysis.
To conclude, this paper offers a number of insights into what can be achieved using the Foucauldian approach of employing statements to analyse discourse.What emerges from the analysis conducted here is a better understanding of how the discourses of risk and trust-generation (science, care and corporate responsibility) operate in the corpus of PILs in question at the microlevel of statements.Describing the constituent components of discourses also provides a means of challenging them as it makes underlying assumptions amenable to scrutiny and evaluation.In the case of texts that are considered not to meet the expectations or needs of those for whom they are intended, such analytical and evaluative perspectives are very valuable, as, by understanding better how these texts work discursively, it may be possible to derive arguments and strategies for their improvement.

Table 1 .
Overview of corpus