Theresa M Marteau & John Weinman. The Sage Handbook of Health Psychology. Editor: Stephen Sutton, Andrew Baum, Marie Johnston. Sage Publishing, 2008.
Communication between patients and practitioners is a central part of health care. Effective communication is essential, for example, for practitioners to understand the nature of a patient’s problem and the patient’s perception of this. Explaining how future illness might be averted or current illness treated also requires effective communication. Failure to communicate effectively has numerous adverse effects, including patients not understanding the results of tests they have undergone (Maissi et al., submitted; McBride et al., 2002), false and failed reassurance (Lucock, Morley White & Peake, 1997; Smith, Shaw & Marteau, 1994), non-adherence (Haynes, McKibbon & Kanani, 1996), and longer lengths of stay in hospital (Johnston & Vogele, 1993). Failed communication also lies at the heart of much litigation (Petticrew, Sowden, Lister-Sharp & Wright, 2000). In addition, it is claimed that good communication skills can avoid ‘burnout’ in health professionals (Feinmann, 2002). The scale of these problems is large. For example, 20 per cent of mothers taking their children to a paediatric outpatient clinic were not informed clearly about the diagnosis and nearly 50 per cent were uncertain afterwards as to the course of their child’s illness (Korsch & Negrete, 1972). Half of those informed about an enhanced genetic vulnerability to lung cancer were they to continue to smoke failed to understand this (McBride et al., 2002). One-third of those receiving a normal cervical smear test result misinterpreted this as meaning that they had no chance at all of having cervical cancer, as opposed to a low risk of being affected (Maissi et al., submitted).
Problems in communicating between health care professionals and patients have been widely documented over several decades. Ley (1988) produced one of the first systematic attempts based on psychological principles to understand and thereby to address these problems. Ley started from the observation that communication is the aspect of the clinical encounter that engenders least satisfaction. He saw this as stemming from two problems: a failure of patients to understand the information given; and related to this, a failure to recall what had been said.
Failure to understand information was seen as having two main causes. First, information is presented in too difficult a way, using unfamiliar and technical terms or using familiar terms in unfamiliar ways. For example, in screening the term ‘positive result’ is used to denote one indicating a problem, in contrast with its favourable connotation in the vernacular. The second main cause of misunderstanding is that the information presented clashes with patients’ own representations and is interpreted within these. Thus, patients who perceive hypertension as an acute symptomatic condition are less likely to adhere to medical advice to take medication than are those whose perception of the condition coincides with the medical view (Meyer, Leventhal & Guttman, 1985).
Recall of information is affected both by understanding of the information as it is presented and by its amount and complexity (Ley, 1988). Drawing upon findings such as these, Ley developed what he termed a ‘cognitive model’ depicting the relationships between understanding, memory, satisfaction and compliance. Although disarmingly simple, this model highlights the importance of presenting information in a way that it is understood and recalled. Since this time, research has broadened to consider how the process of communication influences outcome. It has also built upon research on communication in routine consultations to consider communication in specific areas aimed at achieving particular outcomes in different patient groups.
This chapter starts by presenting a summary of some general principles of effective communication focusing on two common media: spoken and written language. The next three sections build upon this, to review the nature of communication problems and their effects in relation to three significant and contrasting areas for health psychology, focusing on (1) healthy individuals, (2) those facing surgery, and (3) those having to engage in longer term adherence to medical treatments.
Communicating Information: General Principles
People are not passive recipients of information. They actively process information, trying to make sense of it by drawing upon pre-existing representations or schemata. Such schemata function as a set of expectations that guide the encoding of new information (Hampson, Glasgow & Zeiss, 1994; Leventhal & Diefenbach, 1991). For example, perceptions of the causes of familial adenomatous polyposis, an inherited form of bowel cancer, influence how accurate the results of a genetic test are seen to be (Michie et al., 2002). When genes were seen as the sole cause, those receiving a test result showing they had not inherited the gene were reassured and did not want bowel screening. But when the cancer was seen as caused by many factors, including diet and stress, as well as genes, the genetic test was seen as less accurate at predicting disease onset, with a resultant continued desire for bowel screening, despite a test result showing that they had not inherited the gene for this dominantly inherited predisposition to bowel cancer. Such results highlight the dynamic nature of effective communication requiring eliciting current understanding of a situation and presenting information so that any new information makes sense within an existing schema. Where such schemata will not accommodate some new piece of information or do not exist, the task of the communicator is to present information in order to develop a schema within which the new information will make sense. This approach formed the basis of the first experimental study intervening to alter illness representations to improve outcomes after a heart attack (Petrie, Cameron, Ellis, Buick & Weinman, 2002). The intervention was effective at altering patient representations of a myocardial infarction and reduced the length of time taken to return to work as well as the reporting of symptoms of angina. While further research is needed to determine the most effective ways of eliciting, challenging and changing schemata, such interventions that have much promise in improving patient outcomes depend critically upon basic communication skills.
Communicating with Different Groups
There is good evidence of ethnic and social inequalities in health care and health outcomes (Cooper & Roter, 2002; Department of Health and Social Security, 1980). While much work has focused on access to health care, there is some evidence to show that health care professionals’ communication with these groups may contribute to these inequalities (Cooper & Roter, 2002). Thus, ethnic minority patients report less involvement in medical decisions, less partnership with their providers and lower levels of satisfaction with care (Cooper-Patrick et al., 1999; Saha, Komaromy, Koepsell & Bindman, 1999). Patients from lower socioeconomic groups are given less information in consultations (Pendleton & Bochner, 1980; Waitzkin, 1985). This stems from them being less verbally active in consultations (Bain, 1977), often mistaken by practitioners as indicating little interest in gaining information. Following a review of this literature, Cooper and Roter (2002) recommend that communication skills training programmes need to be more broadly based to train health care practitioners to communicate in a culturally more sensitive way, and that strategies are needed to empower patients across ethnic and social groups to participate more in their care.
Communicating effectively with those with low levels of literacy and those from minority ethnic groups requires some different considerations from those that govern communication with those who are literate and those from dominant ethnic groups. Those with low levels of literacy, estimated at about 20 per cent of the US and UK populations, derive little benefit from written communication. In inner city hospitals in the US, levels of functional illiteracy are as high as 35 per cent amongst English-speaking patients and 62 per cent in Spanish-speaking patients seeking care (Williams et al., 1995). Their lack of written language skill affects their processing of spoken language, making it less effective than those who can read and write fluently (LeVine et al., 1994). Understanding and comprehension of simple instructions regarding treatment are much improved for this group using pictographs, that is, cartoons depicting instructions. Recall of instructions regarding a range of medical problems was dramatically better when oral instructions were accompanied by pictographs (Houts et al., 1998; Houts, Witmer, Egeth, Loscalzo & Zabora, 2001). Giving patients tape recordings of their consultations may be particularly useful for this group, as would be videotapes presenting salient information about particular illnesses and their prevention or management. Health practitioners are, however, poor at identifying those with low literacy skills, with patients often keen to disguise this.
Values, customs and beliefs about illness vary across, as well as within, cultural groups. Such variations will affect how information is understood as well as responses to it (Baumann, 2003). Some of the clearest examples of differences in illness beliefs between cultures concern causal beliefs (Landrine & Klonoff, 2001). Cultures characterized as egocentric (mostly Northern European and North American) embrace the dominant medical model of disease, and with it mind-body dualism. By contrast, cultures characterized as sociocentric draw few distinctions between mind and body, religious, spiritual, medical, social and emotional processes. Such differences will affect the effectiveness of communications about illness, treatment and prevention. Thus, informing those who share the beliefs of a sociocentric culture that they can reduce their risks of heart disease by such behaviour changes as increasing levels of physical activity is likely to have little impact when the dominant view of the cause of illness is an imbalance arising out of relationship conflicts, moral transgressions and negative emotions. While it is relatively easy to predict when communications based on an egocentric culture will be ineffective for those subscribing to the values of a sociocentric culture, it is currently unknown how such communications can be made effective. Eliciting and acknowledging existing illness beliefs will be a starting point.
Information about health care is available from many sources using a variety of media. While there is increasing focus on the use of the internet and interactive media, the two most commonly used media for communicating information in health care are in person and in writing.
There has been a considerable amount of research over the past 40 years aimed at ascertaining the core communication skills needed for effective face-to-face communication. There is some evidence, for example, that the process of communication influences patient outcomes. Thus, Hall and colleagues’ overview of research in this area (Hall, Roter & Katz, 1988) showed that patient satisfaction was higher following consultations in which the health care professional engaged in more social conversation, positive verbal and non-verbal communication and partnership building. Ascertaining the precise relationships between process and different outcomes has, however, proved a difficult task, made more difficult by a paucity of reliable measures of dynamic processes as well as a lack of theoretical models to guide such measurement.
A number of systems have been developed to measure aspects of the doctor-patient relationship (Stewart et al., 1995). A meta-analysis of communication studies found that the many different elements of communication measured in the studies fell into five broad categories: information giving; question asking; partnership building; rapport building; and socioemotional talk (Hall et al., 1988). The links between these and outcomes were reviewed by Stewart (1995). Informativeness was an important factor influencing health outcomes, particularly when accompanied by emotional support, resulting in reduced psychological distress, enhanced symptom resolution and reduced pain. Physicians asking about patients’ understanding, concerns and expectations was also important in achieving these outcomes. Using cluster analysis, Roter and colleagues (1997) identified five patterns of relationship in primary care consultations: narrowly biomedical; biomedical (in transition); biopsychosocial; psychosocial; and consumerist. The first two are considered to reflect a paternalistic model, the third and fourth to reflect variations on patient-centred communication, and the final one to reflect a consumerist model.
Skills that facilitate the effective communication of information include using language that is readily understood, presenting information in a way that takes account of the patient’s beliefs, and checking understanding of any information that has been given (Ley, 1988). Video recordings of consultations show that these skills are frequently absent in routine consultations (Braddock, Fihn, Levinson, Jonsen & Pearlman, 1997; Campion, Foulkes, Neighbour & Tate, 2002). Thus, for example, recording of consultations conducted by primary care physicians in the US revealed that in just 2 per cent were direct questions asked of the patient to check understanding (Braddock et al., 1997). Similarly in an analysis of videotaped consultations selected by candidates as part of a qualifying examination for membership of the Royal College of General Practitioners, checking patient understanding was evident in just 20 per cent of the consultations (Campion et al., 2002).
Communication skills can be improved by training (Fallowfield et al., 2002; McGuire, Fairburn & Fletcher, 1986). There is, however, little evidence regarding the effective components of such training. This might be enhanced if a theoretical basis for the type of training is made explicit. For example, based on evidence regarding the acquisition of other types of skills, feedback, preferably based on videos, seems an important component (Hall, 1983). Such a technique may be particularly important in developing communication skills whose quality is difficult to judge and is, therefore, subject to such face-saving attribution biases as attributing patient confusion to the patient’s inherent incompetence but patient clarity to one’s own skills (Miller & Ross, 1975). Bandura’s (1997) social cognitive theory is one of the few theoretical models to be applied to describing and changing health care professionals’ communication behaviour (Parle, Maguire & Heaven, 1997). Using this model, Maguire and colleagues highlight, for example, the importance of addressing health professionals’ negative beliefs about the consequences of giving certain types of information to patients, as well as their self-efficacy in being able to communicate effectively. In addition to using theory to develop effective training, there is an evident need to ensure that such training is an integral part of health professional training. Such training is rarely mandatory. In a study of medical students, confidence in communicating effectively was inversely related to actual skills (Marteau et al., 1991) suggesting that health professionals’ ability to judge their own need for training would not ensure a skilled workforce. There is also evidence that those with the most communication skills are those most likely to persist with any training (Michie & Marteau, 1999). Research is needed to ascertain the attitudinal barriers to health professionals embracing more fully the centrality of effective communication skills to their practice, be it as a surgeon, physician or nurse.
Written information is sometimes used as the sole means of communication, for example when inviting people to participate in population-based screening programmes. More often it is used to supplement orally presented information, acting as a reminder or as a source of additional information, for example, about self-management of chronic conditions. Reviews of written information about a variety of health topics reveal several common problems including the use of inaccurate information, missing information and text that is too complex to be understood by the majority of the intended audience (Coulter, Entwistle & Gilbert, 1999; Ley, 1988; Slaytor & Ward, 1998). Based on research evidence as well as consensus, several checklists have been developed to guide the process of developing written information as well as its content (Coulter et al., 1999; Jadad & Gagliardi, 1998; Silberg, Lundberg & Musacchio, 1997). One of the earliest approaches to improving written information, still used today, is the use of readability formulae. These most often incorporate estimates of sentence length and vocabulary load (number of syllables per word) to generate an index of the level of education needed to understand the text (Flesch, 1948). Wright (1980) described ‘usability’ as a helpful heuristic in designing written, factual information. Usable information is designed to take account of readers’ fluency and familiarity with the subject matter as well as their cognitive abilities and existing conceptual structures. The order, layout of materials and paragraph structure are also important in contributing to the usability of written information. A checklist generated by Coulter and colleagues (1999) summarizing important steps in developing written information is shown in Table 10.1.
While such checklists are useful in describing the information that is needed, they provide little guidance on the most effective way of presenting it. Some of the evidence relating to this is included in the following sections.
Health Risk Information
Health risk information comprises, as a minimum, information about the likelihood and nature of an illness or condition that threatens an individual’s health. Such information is commonly given as part of health risk assessments or health screening. There is no one best way of presenting risk information. It will depend in part upon the aim of communicating the information. Thus, for example, if the aim is to reassure, likelihood information may be more effective at achieving this if it is presented in the form of an absolute as opposed to a relative risk (Brase, 2002).
Achieving Understanding of Information
Information about the nature of a threat is commonly assumed and thus often not presented in detail. For example, participants in population-based screening programmes are given relatively little information about the nature of the condition for which screening is being offered. Such information may be omitted based on a frequently erroneous assumption that familiarity with a term means familiarity with the condition. Thus, for example, while the great majority of people are familiar with diabetes and different types of cancers, there is a tendency to overestimate the severity of all cancers while underestimating the severity of diabetes (Farmer, Levy & Turner, 1999; Gigerenzer, 2002). Understanding the nature of a condition can be important in terms of influencing both representations of its prevention and treatment and, in turn, emotional and behavioural responses (Leventhal et al., 1997).
|Table 10.1 Checklist for patient information materials|
|1||Involve patients throughout the process|
|2||Involve a wide range of clinical experts|
|3||Be specific about the purpose of the information and the target audience|
|4||Consider the information needs of minority groups|
|5||Review the clinical research evidence and use systematic reviews wherever possible|
|6||Plan how the materials can be used within a wider programme promoting shared decision making|
|7||Consider cost and feasibility of distribution and updating when choosing media|
|8||Develop a strategy for distribution|
There have been relatively few studies documenting effective ways of increasing understanding about the nature of an illness. Morgan, Fischhoff, Bostrom and Atman (2002) describe a method of communicating about a threat, based on a mental models approach to risk communication. This is based on a premise similar to that at the root of Leventhal’s self-regulation model, namely that individuals actively process information within the confines of existing schemata. The starting point for communicating about a risk is therefore to elicit lay models using first open-ended interviews followed by confirmatory questionnaires to ascertain the prevalence of salient beliefs and misconceptions.
There are many different ways in which information about the likelihood of a condition can be given. These include the use of verbal descriptors (e.g., high or low risk), numerical expressions of likelihood expressed in language (e.g., absolute and relative risks, expressed as percentages, single event probabilities), or graphically. Such information can be framed positively (e.g., the chances of not developing a disease) or negatively (e.g., the chances of developing a disease). There is much debate about the most effective expressions and formats to use. To date, there have been too few comparable studies in clinical contexts for many patterns to emerge (Brun, 1993; Edwards, Elwyn, Covey, Matthews & Pill, 2001; Lipkus & Hollands, 1999). The one dimension of likelihood information where there has been sufficient research for a consensus to begin to emerge concerns the presentation of numerical expressions. Quantitative information is best understood when it is presented using simple frequencies (e.g., 1 in every 100 people) as opposed to single event probabilities (e.g., 0.01) or relative frequencies (percentages, e.g., 1 per cent) (Brase, 2002; Gigerenzer, 2002). In addition, people are better able to compare frequencies when they are presented using the same denominator. Thus, for example, pregnant women were more likely to correctly understand which of two risk estimates was larger when the information was presented as 2.6 versus 8.9 per 1,000 than when it was presented as 1 in 384 versus 1 in 112 (Grimes & Snively, 1999).
Understanding the likelihood of an outcome is, however, different from perceiving personal vulnerability. Thus, for example, smokers may perceive smoking as hazardous but perceive their own risks as lower than those of the average smoker (Weinstein, 1998). In a review of approaches to communicating health risk information, Rothman and Kiviniemi (1999) observe that informing people that particular behaviours can increase the probability of adverse health outcomes generally has little effect on perceptions of personal vulnerability. The authors attribute this in part to a misunderstanding of numerical information and in part to a minimizing or discounting of unfavourable health information. Attempts to overcome these motivational processes in order to motivate behaviour change have generally been unsuccessful (Weinstein & Klein, 1995; Weinstein, Sandman & Roberts, 1991). The relatively weak association between risk perception and behaviour (van der Pligt, 1998) suggests that altering risk perceptions may not be the most effective way of changing behaviour. This then raises the question of what people need to understand about a health threat for its communication to be considered effective. It may be sufficient for individuals to appreciate that they have some vulnerability to a health threat in order for them to process further relevant information about a threat. The amount of vulnerability experienced may be less important than the fact that some vulnerability is experienced. If risk perceptions reflect fear minimization processes, with little effect on behaviour (Wiebe & Korbel, 2003), it perhaps should not be considered a failure of risk communication that an individual at higher than average risk perceives that they are vulnerable but that their level of vulnerability is lower than average.
Avoiding or Reducing Anxiety
Informing individuals of an increased chance of developing an illness is associated in the short term with raised general levels of anxiety (Shaw, Abrams & Marteau, 1999). After 1 month such anxiety is rarely evident. Anxiety is often an adaptive, initial response to health risk information (Cameron, 2003). It can enhance attention to health threats, elaborate representations and motivate behaviour to reduce threats. It is important, however, that the anxiety and fear generated by health risk information is manageable. A number of cognitive processes have been described which have the effect of minimizing the threat and resultant fear following the provision of health risk information. These include downplaying the seriousness of a problem (Croyle, Sun & Hart, 1997), downplaying the accuracy of the test (Ditto & Lopez, 1992) and unrealistic optimism (Weinstein, 1999). While these defensive processes reduce distress there is too little evidence to judge their impact upon risk reducing behaviour (Wiebe & Korbel, 2003). But while anxiety, particularly in the first few weeks after learning of a health risk, may signal the operation of adaptive processes, it can have adverse effects, particularly when it is excessively high and unabated by time, resulting in anxiety-infused information processing, and a focus upon short-term alleviation of anxiety.
Research effort in this area has focused upon two problems: first, high levels of short-term anxiety, particularly in those informed of increased risks of a cancer; and second, continued concern in those informed that the results of investigations are negative, indicating no problem.
Effective ways of preventing raised anxiety in those informed of an increased risk of disease include presenting information to prevent overestimation of the likelihood of having a significant health problem, a significant predictor of anxiety (Maissi et al., submitted). So, for example, women informed of an abnormal cervical smear test result are significantly less anxious when also informed that such a result is only rarely associated with cancer (Wilkinson, Jones & McBride, 1990). Such effects might be enhanced further by presenting the absolute chances of developing cervical cancer given such a test result. Given the association between perceived seriousness and vulnerability (Hendrickx, Vlek & Oppewal, 1989), interventions aimed at reducing the perceived severity of abnormal cervical smear test results might reduce perceived vulnerability and in turn levels of distress.
Many of those undergoing diagnostic tests, particularly following screening, will have no abnormalities detected. Despite this, some will experience continued concern. While simple reassurance can be very effective in reducing concern (Lucock et al., 1997; Watson, Hall, Langford & Marteau, 2002), it is less so for those with high health anxiety. In a prospective study of patients undergoing gastroscopy which revealed no serious illness, Lucock and colleagues (1997) showed that while oral reassurance from the physician that there was ‘nothing seriously wrong’ was effective in reducing worry for 24 hours, for those with high health anxiety the worry resurged after this point and was maintained 1 and 12 months later. These results may be explained by findings from a series of experimental analogue studies which suggest that informing someone concerned about their health that they are ‘fine’ can actually serve to increase their concern (Cioffi, 1994). For information to reassure in the face of uncertainty it needs to be presented in the same frame as the one the patient is using, that is to feature positive information. In one study, students were informed that they were either at low, moderate or high risk of developing a cancer caused by exposure to asbestos. Half of each group then received information about the number of healthy cells they had or the number of mutant cells. Those informed they were at moderate risk (maximum uncertainty) and who received information about the number of healthy cells in their body had least confidence in their health of all six experimental groups. Thus, someone concerned about their health is more likely to be made more concerned by informing them of the chances that they are well as opposed to the chances that they are ill.
In addition to the framing of the information designed to reassure, any explanation is more likely to be accepted and to provide reassurance if it fits with the individual’s existing illness schema. This requires eliciting patients’ illness schemata alongside their concerns in order that information can be provided that makes sense within that context. So, for example, if someone is presenting with headaches fearing a brain tumour, for which tests reveal no evidence, a reassuring explanation is likely to be one that elicits and maybe alters the existing schema of brain tumours and headaches to allow the one to exist without the other.
The majority of health risk assessments and many diagnostic test procedures reveal no problems. Such results mean that there is a low chance, but rarely do they mean no chance, of developing or having the tested condition. Erroneously believing that a negative or normal test result means no residual risk may be associated with reinforcement of unhealthy lifestyles (Tymstra & Bieleman, 1987), delay in seeking help when symptoms of disease arise, as well as delay in health professionals responding to such symptoms (Petticrew et al., 2000), poorer adjustment (Hall, Bobrow & Marteau, 2000), and litigation (Petticrew et al., 2000). The challenge is how to present the meaning of a normal test result to avoid false reassurance, that is the belief that low risk means no risk, without inducing anxiety about the residual risk.
Until recently false reassurance was not widely considered a problem. The focus in health risk communication had been firmly upon avoiding disabling levels of anxiety largely in response to the results of a study reporting the adverse effects upon work attendance of detecting hypertension in a group of steel workers (Haynes, Taylor & Sackett, 1979). The perceived need to avoid high levels of anxiety has probably unwittingly fed into a desire to present health risk assessments in an overly reassuring way. Thus information on test sensitivity (i.e., the proportion of cases a test can detect and, by implication, the proportion it cannot) is most often not given. For example, in an analysis of 58 leaflets describing breast cancer screening, information on the proportion of breast cancers detected by screening was mentioned in just 15 of them (Slaytor & Ward, 1998). In a study of HIV testing, five of 19 counsellors studied incorrectly informed an individual undergoing testing that it was impossible to get a false negative test result (Gigerenzer, Hoffrage & Ebert, 1998).
The few studies conducted to date attempting to avoid or reduce false reassurance show that using the term ‘normal’ to describe test results encourages false reassurance with as many as 50 per cent believing their results indicate no residual risk (Marteau, Senior & Sasieni, 2001). A range of verbal and numerical expressions to depict residual risk is effective at reducing rates of false reassurance in the short term (Marteau et al., 2000, 2001). While rates of false reassurance can be reduced, as many as 30 per cent in one study continued to be falsely reassured (Marteau et al., 2001), this being more marked in those with lower levels of education who were more likely to be falsely reassured. Further attempts to avoid false reassurance need to focus on this group in particular. There is also a need to understand the emotional and cognitive processes that affect longer-term recall. In one of the few studies assessing recall of the meaning of low risk test results over time, we found that while recall of residual risk was high immediately after testing, with over 90 per cent of participants correctly understanding that their negative test result meant that they had a very low (but not no) risk of being a carrier for cystic fibrosis, 3 years later only 50 per cent correctly recalled this, with over 43 per cent erroneously believing that they had no risk at all (Axworthy, Brock, Bobrow & Marteau, 1996). Such an erosion may reflect the operation of simplifying heuristics (low risk eventually becoming seen as no risk) or the operation of defensive biases, protecting individuals from perceiving themselves at continuing, albeit low levels of risk.
False reassurance is a concept implicit in risk homeostasis or risk compensation theory (Wilde, Robertson & Pless, 2002). This theory is based on two core assumptions. The first assumption is that people are utility maximiz-ers, in that they will behave in the way that maximizes overall benefits. The second assumption is that people have an implicit level of risk that they consider acceptable. By acting to reduce a risk, for example by wearing a seat belt, this decreases their risk below the level they consider acceptable, resulting in behaviour that then increases their risk, for example by driving faster, back to their target level. While the evidence to support the theory and in particular the second premise are contested (McKenna, 2002; Robertson, 2002), it may have some merit as a general explanatory framework in accounting for the failure of many health promoting interventions to reduce risk. Given the well documented use of fear control processes (Croyle et al., 1997), it is plausible that acting to reduce a risk may result in an overestimation of the risk reduction achieved and an underestimation of the risks entailed in unchanged risk enhancing behaviours. Thus, increasing levels of physical activity may result in an overestimation of the risk so reduced, and an underestimation of the risk induced by continuing to smoke or eating a high fat diet.
Risk Information Fails to Result in Behaviour Change
Presenting risk information, including information based on biomarkers, infrequently leads to behaviour change (McClure, 2002). For example, informing people of a genetic susceptibility to lung cancer does not result in smoking cessation (Audrain et al., 1997; McBride et al., 2002). Dietary change does not follow the provision of information about raised cholesterol (Strychar et al., 1998), nor is activity increased when individuals are informed of their cardiovascular unfitness (Godin, Desharnais, Jobin & Cook, 1987). Such findings reflect in part the limited impact of risk information upon risk perception (Avis, Smith & McKinlay, 1989; Kreuter & Strecher, 1995; Rothman & Kiviniemi, 1999), as well as the limited role that perceived risk plays in motivating behaviour change (van der Pligt, 1998).
The focus of presenting risk information has been upon communicating the likelihood of disease largely to the neglect of other key pieces of information (Rothman & Kiviniemi, 1999; Weinstein, 1999). Recent systematic review evidence shows the importance in motivation to change behaviour in the face of a health threat of not only perceiving the threat as likely, but also perceiving the threat as serious, the recommended behaviour as effective in reducing the risk, and the ability to perform the behaviour as existent (Floyd, Prentice-Dunn & Rogers, 2000; Milne, Sheeran & Orbell, 2000; Witte & Allen, 2000). Much of this research has been conducted using non-clinical samples, often in descriptive studies. The challenge now is to develop effective ways of communicating about these salient dimensions of a risk to motivate behaviour change in clinical populations.
While intention to perform a behaviour is a reliable predictor of actual behaviour, it accounts for between 20 and 30 per cent of the variance (Sheeran, 2002; Sutton, 1998). Many models of health behaviour draw a distinction between intention to engage in a behaviour and the enactment of that intention (Schwarzer, 1992; Sheeran, 2002). It appears that thinking about future behaviour in relation to specific environmental cues of when, where and how to enact a particular behaviour increases the frequency of that behaviour. So, for example, students were more likely to have a tetanus inoculation when the fear message was accompanied by details of where on campus the inoculation could be obtained (Leventhal, Singer & Jones, 1965). More recently action plans have been studied under a new name of implementation intentions (Gollwitzer, 1999). These take the form of individuals writing when, where and how they will enact a particular behaviour. This includes taking vitamin tablets, undergoing exercise, and attending for screening (Sheeran, 2002). While this approach shows much promise, its effectiveness in non-student samples awaits further studies.
Anticipating the negative emotion of regret can also result in people being more likely to engage in behaviour to reduce risks. Thus, those asked to focus on how they would feel were they to experience a related adverse outcome having not engaged in a recommended preventive action are more likely to perform the action (Parker, Stradling & Manstead, 1996; Richard, van der Pligt & De Vries, 1995). Anticipated regret has an effect by strengthening the association between intentions and behaviour, thereby reducing the widely acknowledged gap between intentions and behaviour (Abraham & Sheeran, 2003).
The Stress of Surgery
The Nature of the Problem
There are many situations and contexts in health care where routine communication or communication-based interventions can have important effects on patient wellbeing. This is particularly true for patients coping with medical procedures, which can be painful or distressing. One of these, surgery, has also been found to produce significant adverse emotional effects (Johnston & Wallace, 1990; Kiecolt-Glaser, Page, Marucka, MacCallum & Glaser, 1998), the extent of which will depend on its severity and duration, as well as on psychological factors such as the patient’s expectations and coping style and the quality of health care communication.
Recovery from surgery is both multidimensional and variable (Johnston, 1984). There are many indices of recovery, including mood, pain, wound healing and length of hospital stay. Studies that have examined patterns of post-operative recovery provide clear evidence that, whatever recovery index is used, there is considerable variation between patients who have undergone the same surgical procedure. Even for discrete procedures such as minimal access (keyhole) surgery, some patients show rapid return to full function whereas others may take a significantly longer time (McGinn et al., 1995; Schlumpf, Klotz, Wehrli & Herzog, 1994). Moreover, for some quite widely used surgical procedures, some patients may fail to obtain therapeutic benefit. For example, following surgery for a lumbar spinal stenosis, approximately one-third of patients fail to benefit substantially (Turner, Ersek, Hernon & Deyo, 1992). In contrast, there is recent evidence that for another widely used surgical procedure, knee arthroscopy, patients who have undergone ‘placebo surgery’ show a very similar pattern of post-operative recovery to those patients who have received the full surgical intervention (Mosely et al., 2002).
The determinants of this variability are multifactorial and depend on a range of physical, contextual and psychosocial factors (e.g., Iverson, Daltroy, Fossel & Katz, 1998). One factor, which has been investigated quite extensively, is the pre-operative mood of the patient, particularly their level of state anxiety, which, in turn, may reflect their level of preparedness for the surgery. A large number of studies have investigated the relation between pre-operative mood and post-operative recovery, and the majority have demonstrated a link between the greater anxiety or distress and a range of recovery indicators (see Munafo & Stevenson, 2001, for a review). Early researchers (Janis, 1958) proposed a curvilinear relation between pre-operative distress and recovery but subsequent work indicates that this is more likely to be linear (e.g., Johnston & Carpenter, 1980). Thus patients with higher levels of pre-operative anxiety or distress report more post-operative pain, negative mood and poorer recovery (Munafo & Stevenson, 2001). There is also experimental evidence which has shown that stress slows the speed of wound healing and has implicated the role of the immune system in this process (Kiecolt-Glaser et al., 1998).
The ways in which pre-operative anxiety influences post-operative recovery are not clearly understood. Some studies (e.g., Kain, Sevarino, Alexander, Pincus & Mayes, 2000) have used path analysis of data collected at multiple time points and have proposed that anxiety has a critical role in the chain of events influencing post-operative pain levels and responses. In contrast, other researchers maintain that the apparent influence of anxiety on recovery is explained by its association with pre-operative function and fatigue. Hence Salmon, Hall and Peerbhoy (2001) propose that while emotional response to surgery partly predicts post-operative recovery, the critical component of emotion is fatigue rather than anxiety.
Patients’ emotional responses prior to surgery relate to concerns about the procedure and the outcome (Weinman & Johnston, 1988). Patients have been found to experience emotional reactions, such as fear of surgery and anaesthesia, as well as lack of information about medical details, the roles of different health care professionals and discharge from hospital (Breemhar, van den Borne & Mullen, 1996). To a large part these problems are caused by inadequate information provision which, in turn, may be due to a range of organizational barriers. One large recent study of over 3,000 surgical patients has sought to investigate the difficulties patients have in obtaining information and the ways in which this can affect the patients’ evaluation of their experience in hospital (Krupat, Fancey & Cleary, 2000). This study identified four distinct information factors, namely surgical information, recovery information, sensory information and general information, each of which was related to patients’ evaluations. They also showed that individual patient factors, such as desire for information and perceived control, were important in moderating and mediating the value and impact of information in determining patients’ responses to impending surgery.
Some of these individual difference factors will be considered in more detail later, but the general point to be made here is that lack of patient preparation for surgery and the consequent effects on emotional reactions provide one explanation for the variability in patient recovery. Another potentially influential pre-operative determinant of recovery, which can be influenced by the level of preparation or information provided, is the patient’s expectation regarding the surgery. There is a small recent literature on the role of patients’ expectations in post-operative recovery and this provides clear evidence that these can influence such factors as pain, functional recovery and satisfaction (Iverson et al., 1998; McCarthy, Lyons, Weinman, Talbot & Purnell, 2003).
This brief overview of pre-operative psychosocial influences on post-operative recovery points to the potential value of different types of communication in preparing patients for surgery. The next section describes a range of approaches to pre-operative psychological preparation and examines the evidence for their efficacy.
Psychological Interventions for Stressful Medical Procedures
Since studies have shown a relation between patients’ psychological state and their recovery, it has been recognized that there could be considerable gains from providing a psychological intervention designed to reduce or minimize the psychological impact of a medical procedure. There is a range of interventions that have been used to prepare patients for surgery or other stressful procedures in the hospital setting. In broad terms they can help by providing the patient with information to reduce the uncertainty of the event, or with specific behavioural or cognitive skills to help with some of the discomfort or pain. The main approaches are as follows.
This is probably the most widely used approach and consists of providing information about the various procedures that will take place before and after the operation. It therefore involves the patient being provided with information about what will happen to them at different stages pre- and post-operatively Sometimes this information is also accompanied by an explanation of the purpose of each of the procedures that are described. Thus it provides the patient with a realistic set of expectations about the events that will occur and, in doing so, can reduce the uncertainty of the whole process.
This describes what patients are likely to feel, particularly during the immediate post-operative pain period. The function of this information is to provide matter-of-fact or benign interpretations of the sensations so that the patient can recognize them as part of the expected postoperative process. Thus the patient who can recognize post-operative pain as an expected sensation caused by the incision and reflecting the healing process will be far less likely to be distressed than someone who has not been prepared for the pain and who may think of it as a problem or a complication of the surgery.
Contrada, Leventhal and Anderson (1994) have discussed the benefits of sensory and procedural information from the perspective of the self-regulatory model. They propose that the sensory preparation serves to provide a script which describes internal sensations, and that the procedural preparation provides a script of the objective external events involved in surgery. They maintain that it is the availability of the script which reduces uncertainty and worry for the patient. More specifically, sensory information should be particularly helpful since it focuses on potentially threatening sensations (e.g., pain, discomfort) with the aim of ensuring that these are processed as non-threatening or less threatening. However, evidence on the efficacy of both types of preparatory information indicates that procedural information is at least, if not more, effective in producing favourable post-operative outcomes (Johnston & Vogele, 1993: see below).
These typically describe different behaviours that the patient can use to help before, during and after surgery. These include instructions about ways to cough and move in bed that will reduce the likelihood of pain associated with these movements. Other behavioural instructions such as deep breathing and ambulation exercises may also reduce the incidence of pain or complications as well as facilitating recovery.
This is based on the use of filmed models who can be seen undergoing the same procedure as the patient. Following Bandura’s (1986) social learning theory, modelling or the observation of others completing a difficult or stressful task can serve to increase the individual’s sense of self-efficacy for managing the same task. Two main types of model have been investigated: mastery models, who are shown dealing with the task with ease and ability; and coping models, who are shown as having some anticipated concerns but who nevertheless are able to overcome these and cope with the procedure. The coping type of model has been found to be a more effective preparation for children undergoing surgery. For example in a study of children about to be inoculated, Vernon (1974) has compared a group of children who saw a preparatory film which was ‘realistic’ (the child in the film is seen to experience short-lived, moderate pain and emotion) with a group who saw an ‘unrealistic’ film (the child shows no pain or emotional expression) and a group who saw no preparatory film. The realistically prepared group were found to experience least pain when receiving their injections. These methods have been more widely used with children than with adults, particularly since it may be more difficult to provide children with sensory or procedural information or behavioural instructions in a helpful way.
These can involve a number of different techniques, including deep breathing, progressive muscle relaxation or, less frequently, hypnosis. Relaxation can be used both to provide a general preparation involving anxiety reduction and as a specific skill which can be used post-operatively for coping with pain and discomfort at times of increased stress or tension.
Cognitive Coping Procedures
These focus on patients’ concerns and fears about the surgery and assist in dealing with them in one of two ways. First, they may make use of coping strategies that the patient has used successfully in the past for coping with stressors, enabling the patient to rehearse and apply these in the surgical context (Langer, Janis & Wolfer, 1975). The second cognitive approach involves coping with negative thoughts by distracting attention from them and by focusing on positive aspects of the surgery and repeating positive self-statements (Ridgeway & Mathews, 1982).
The efficacy of these interventions has been evaluated by examining their effects on a range of post-surgical outcomes, including anxiety, pain and use of pain medication, length of stay in hospital and various indicators of recovery. All the interventions have been found to be successful in improving at least one outcome and the majority have a positive impact on many of the outcomes. The different interventions have been examined and compared systematically in a meta-analysis by Johnston and Vogele (1993). Across the range of outcomes, the largest recovery effects were obtained for pain, negative affect and physiological indices of recovery but there was considerable variation in the magnitude of these effects. Smaller but more consistent advantages of psychological preparation were found on pain medication and length of hospital stay. The interventions that had the most widespread effects on all the outcomes were found to be procedural information provision and behavioural instructions. Relaxation was also found to have beneficial effects on the various outcomes. Whereas Mathews and Ridgeway (1984) had indicated that cognitive coping interventions were most likely to have the greatest efficacy, the meta-analysis results show that their effects appear to be restricted to specific outcomes. Thus cognitive interventions have been shown to have positive effects on negative affect, pain and use of pain medication, and clinical recovery, but do not appear to result in shorter lengths of stay or in improved physiological indices or behavioural recovery. Surprisingly, in view of the importance attached to patient evaluations of health care, only a few studies have examined the effects of these interventions on patient satisfaction but these show quite positive results, indicating that patients view them as acceptable and helpful.
There is now considerable evidence to indicate that different types of psychological preparation can not only reduce the anxiety, stress and pain involved in many medical procedures but also generate considerable related benefits (e.g., less analgesia, better recovery, faster discharge etc.). Although each of the different approaches has been described separately here, they can easily be used in conjunction and often are. What is encouraging from the research and reviews of psychological preparation for surgery is that they show that it is possible to intervene effectively using relatively uncomplicated procedures. Moreover, there is now sufficient information about their efficacy to be confident in recommending that they should be included as routine components of standard medical and nursing care for all patients undergoing surgery.
All the interventions outlined above are relatively brief and can be readily incorporated into routine clinical care, provided that the health care professionals involved are appropriately trained to deliver them. However, there is also evidence that pre-operative patients can gain useful information from fellow patients and that this can also influence post-operative recovery. This evidence comes from an elegant series of studies by Kulik and colleagues (Kulik & Mahler, 1987; Kulik, Mahler & Moore, 1996), who have examined the effects on pre-operative patients of having a roommate who has either undergone or is about to undergo the same operation or a different procedure. Their findings showed the importance of social contact in the recovery of patients following surgery. In their early study, they compared the effects of sharing a room either with a patient who was also about to undergo cardiac surgery or with one who had already had the same operation. The results showed clear beneficial effects of sharing a room with someone who was recovering from surgery. The patients who had post-surgical roommates were less anxious prior to surgery, engaged in more post-surgical physical activity and were discharged sooner (Kulik & Mahler, 1987). In a more recent variation on this study, they investigated two further variables, namely having a roommate or being on one’s own, and having a roommate who was about to undergo or had gone through the same or different type of operation. In addition to replicating their earlier findings respective to the advantage of sharing with a post-surgical patient, they also found that it was advantageous to share with a post-surgical patient who had undergone the same type of surgery and that those who were in rooms on their own had the slowest recovery (Kulik et al., 1996).
From a communications perspective, these results are important for a number of reasons. They serve as a reminder that information from health care professionals is not the only form of communication that can influence the adjustment to and recovery from stressful medical procedures. Important preparatory and other useful information may be communicated verbally and non-verbally from one patient to another, and this can have considerable impact. There is clearly scope for structuring ward environments in order to maximize positive influences of fellow patients on the response to and recovery from surgery.
Pre-Operative Preparation: Is There a Need for More Individually Tailored Approaches?
One of the problems with many of the intervention methods outlined and evaluated above is that they often come as a fixed package, regardless of the level of the pre-operative patient’s informational needs or coping style. Surgery is a potential stressor for the patient, and it is known from the broader stress research literature that people cope in a range of ways, which are partly dependent on dispositional factors and partly on their appraisal of the stressor and their own coping capability (Folkman & Lazarus, 1980). One psychological dimension that pervades this research is linked to approach-avoidance, and has been investigated in a number of ways in the study of coping with surgery. Miller and Mangan (1983) have grouped individuals as ‘monitors’ or ‘blunters’, according to the extent to which they cope with a stressor in a more vigilant (monitoring) or avoidant (blunting) way, whereas others have used a more general coping style classification (e.g., Krohne, Slangen & Kleemann, 1996; Lazarus & Folkman, 1984). Broadly these studies show that those patients who typically use more avoidant approaches to coping find it most helpful to be given relatively little preparation information, and may react negatively to the provision of more detailed information. Conversely the more vigilant or monitoring copers find more detailed information more helpful in preparing for and recovering from surgery.
Recent overviews of tailored approaches to informational interventions provide further support but raise the practical question of how such tailoring can be incorporated into the routine clinical setting. One obvious approach is to adopt computer-based methods in which the level and type of preparatory information can be determined by the individual patient. These approaches can combine text, images and video sequences to provide different sorts of preparatory information. There are good recent examples of the efficacy of video information for preparing patients for a range of surgical procedures, including cardiac surgery (Roth-Isigkeit et al., 2002) and surgery for osteoarthritis of the knee (Ayral, Gicquere, Duhalde, Boucheny & Dougados, 2002).
Adherence to Treatment
The Nature of the Problem
The failure of many patients to follow recommended treatment or advice is now regarded as a major problem in health care. This is usually referred to as noncompliance or non-adherence, although the latter term tends to be used more widely now. Adherence has been defined as ‘the extent to which the patient’s behaviour coincides with the clinical prescription’ (Sackett & Haynes, 1976). Alternatively, non-adherence has been defined as ‘the point below which the desired preventative or desired therapeutic result is unlikely to be achieved’ (Gordis, 1976). Problems of adherence have been observed across a wide range of medical recommendations, including pill taking, returning for follow-up appointments, and following behaviour change advice for primary, secondary or tertiary prevention.
Low rates of adherence to recommended treatment are found in most chronic diseases including asthma (Hand, 1998), diabetes (Strychar et al., 1998), heart disease (Rich, Gray, Beckham, Wittenberg & Luther, 1996), cancer (Lilleyman & Lennard, 1996), HIV (Paterson et al., 2000) and kidney disease (Cleary, Matzke, Alexander & Joy, 1995), as well as in psychological treatments such as relaxation training for anxiety-related disorders (Taylor, Agras, Schneider & Allen, 1983). The incidence of non-adherence varies greatly, converging at 30–50 per cent in chronic illness (Haynes et al., 1979; Meichenbaum & Turk, 1987). In the area of primary prevention, it has been found that many participants drop out of lifestyle change programmes designed to improve diet or reduce health risk behaviours (Brownell & Cohen, 1995). Even patients who have experienced major health problems, such as heart attacks, may show low levels of uptake of rehabilitation programmes as well as considerable variation in the adoption of recommended lifestyle change (Petrie, Weinman, Sharpe & Buckley, 1996).
Non-adherence may be categorized in many ways but one useful distinction is between those that are active and those that are passive. Active non-adherence arises when the patient decides not to take the treatment as instructed. From a self-regulatory perspective, the level of treatment adherence may reflect a strategic coping response that is entirely consistent with the patient’s view of their problem (Leventhal & Cameron, 1987). Thus, if patients believe that their problem will not last for long, they will be less likely to adhere to their medication over a long period of time than those who perceive their condition as more long-lasting (Meyer et al., 1985). Passive non-adherence may be unintentional when it is caused by barriers such as forgetting, and inability to follow treatment instructions because of a lack of understanding or because of physical problems such as poor eyesight or impaired manual dexterity. Similarly, if the quality of communication is poor and patients receive information that is difficult to understand or recall, as has been outlined above, then this reduces the likelihood that treatment will be adhered to (Ley, 1988).
One of the major problems with all adherence research is the difficulty in obtaining accurate measures. Many different measures have been used to provide both direct and indirect markers of adherence, including tracers of metabolized medications, pill counts, electronically monitored pill bottle tops or inhalers, self-reports and the effects on health or illness outcomes (see Myers & Midence, 1998, for a review). These methods vary in their accuracy and, since none can provide an exact measure, many of the better studies have included multiple measures to provide triangulation and greater reliability. Despite these inherent problems, the evidence of the prevalence of non-adherence is widespread and, if anything, the problems of measurement probably mean that the empirical data represent an underestimation of the true extent of the problem.
The Effects of Low Adherence
Even though there are continuing problems in defining, assessing and explaining variations in adherence behaviours, there are clear reasons why low adherence is a major problem in health care. First, and most obvious, patients who are non-adherent are much less likely to gain benefit from their treatment or advice. The relation between treatment adherence and outcome is a complex one that depends on a number of factors, as has been shown in a recent meta-analysis (DiMatteo, Giordani, Lepper & Croghan, 2002). This examined and integrated the findings from 63 studies that varied considerably in sample size, patient group and study design. The overall relation between adherence and outcome was found to be positive with an effect size of 0.21, and with the great majority of studies (n = 51) showing a significant positive effect. The results indicate that greater adherence reduces the risk of a null or poor treatment outcome by 26 per cent, and that a good treatment outcome is three times more likely to be found in those who have been highly adherent. Although these effects were found to be moderated by the type of disease and treatment, it is notable that the overall effect size is broadly similar to that found in a similar review of patients receiving treatment for coronary heart disease (McDermott, Schmitt & Wallner, 1997) and to that found in a recent meta-analysis of disease management interventions in chronic illness (Weingarten et al., 2002).
In addition to the predicted adherence-outcome relation, there is also evidence now that higher adherence can produce positive effects on outcome by way of indirect factors, such as the patient’s expectations. A revealing study by Horwitz et al. (1990) compared the treatment responses of adherent and non-adherent patients to medication for myocardial infarction. This involved a group of patients in a large randomized controlled trial, in which half were randomized to the active treatment (propanalol) and half to the placebo treatment. Unlike the majority of randomized controlled trials, which either ignore or screen out non-adherence, this study made a special effort to assess the adherence of all the participants, who were then categorized as adherent or non-adherent. The comparative advantages found for the adherers on a range of outcomes, including mortality, were very similar for those on the active treatment and the placebo. These findings, which have also been shown in studies with other treatments, indicate that good adherence to any treatment, even when it is a placebo, results in a better outcome, even when factors such as illness severity, comorbidity, and psychosocial differences are taken into account. Thus, for patients, a further disadvantage of low treatment adherence, over and above the obvious lack of any clinical therapeutic effect, is that they are not able to benefit from the considerable non-specific (i.e., placebo) effects which are found with any treatment (Richardson, 1997).
The Determinants of Non-Adherence
From the accumulated research evidence on factors associated with non-adherence, it now seems untenable to accept the idea of a ‘non-compliant type’ in terms of fixed individual or social characteristics. An early systematic review of 185 studies (Sackett & Haynes, 1976) revealed no clear relationship between race, gender, educational experience, intelligence, marital status, occupational status, income and ethnic or cultural background and adherence behaviours. Moreover, there is little evidence that adherence behaviours can be explained in terms of personality characteristics (Becker, 1979; Bosley, Fosbury & Cochrane, 1995; McKim, Stones & Kozma, 1990). Also the idea that stable sociodemographic or dispositional characteristics are the sole determinants of adherence is discredited by evidence that an individual’s levels of adherence may vary over time and between different aspects of the treatment regimen (Cleary et al., 1995; Hilbrands, Hoitsma & Koene, 1995). This limitation also applies to the search for disease and treatment characteristics as antecedents of adherence since there are wide variations in adherence between and within patients with the same disease and treatment (e.g., Cleary et al., 1995; Lilleyman & Lennard, 1996).
One simple explanation for non-adherence is that patients lack basic knowledge about their medication (Cartwright, 1994; Eagleton, Walker & Barber, 1993), but again this does not provide a complete explanation. In a systematic review of the adherence literature Haynes (1976) concluded that, although 12 studies had demonstrated a positive association between knowledge and adherence, there were more that had failed to demonstrate a link. Studies conducted since that time generally indicate that associations between knowledge and adherence are at best small and inconsistent (Eagleton et al., 1993) and, as will be seen below, information-based interventions do not consistently result in increased levels of adherence (Haynes, Sackett, Taylor, Gibson & Johnson, 1978; Roter, Hall et al., 1998). Patient satisfaction may act as a mediator between information provision, recall and adherence, as would be predicted by Ley’s model which was outlined earlier. In a national UK survey of patients’ satisfaction with medicines information, over 70 per cent of respondents wanted more information than they were given (Gibbs & George, 1990). Dissatisfaction with the amount of information and explanation provided may also act as a barrier to adherence by making the patient less motivated towards treatment (Hall et al., 1988).
Adherence research has stopped trying to identify stable trait factors which characterize the non-adherent patient and now places a greater emphasis on understanding those patient cognitions which could explain why they decide to take some treatments and not others (Horne, 1998). The application of social cognition models in research indicates that medication non-adherence may arise from a rational decision on the part of the patient and identifies some of the cognitions that are salient to these decisions, particularly for the more active or intentional types of non-adherence. Although there is some variation in the specific type of beliefs that are associated with adherence across studies, the findings show that certain cognitive variables included in the health belief model (HBM: Janz & Becker, 1984) and the theory of planned behaviour (TPB: Ajzen, 1988) appear to predict adherence in certain situations. For example, higher rates of adherence have been found in patients who believe that failure to take the treatment could result in adverse consequences and that they are personally susceptible to these effects (Cummings, Becker, Kirscht & Levin, 1981; Kelly, Mannon & Scott, 1987; Nelson, Stason, Neutra, Soloman & McArdle, 1978). Additionally, adherence decisions may be influenced by a cost-benefit analysis in which the benefits of treatment are weighted against the perceived barriers (Brownlee-Duffeck et al., 1987; Cummings et al., 1981; Nelson et al., 1978). Other studies based on the TPB have shown that the perceived views of significant others such as family, friends and doctors (normative beliefs) may also influence adherence (Cochrane & Gitlin, 1988; Reid & Christensen, 1988). Several studies have demonstrated the value of interventions based on the HBM in facilitating health-related behaviours, such as attending for medical check-ups (Haefner & Kirscht, 1970), or using emergency care facilities in an acute asthma attack (Jones, Jones & Katz, 1987).
A more dynamic cognitive approach which has been used to explain non-adherence is Leventhal’s self-regulatory model (Leventhal, Brisette & Leventhal, 2003), which acknowledges the importance of symptom perception in influencing illness representations which, in turn, direct coping responses, including adherence behaviour. Early evidence for this approach is provided by findings from a study of patients with diabetes who used perceived symptoms to indicate their blood glucose levels and to guide self-treatment (Gonder-Frederick & Cox, 1991). However, patients’ beliefs about their symptoms, and estimations of their own blood glucose levels, were often erroneous and resulted in poor diabetic control. Further evidence of the role of illness representations in adherence has been found in patients with hypertension (Meyer et al., 1985). Patients who believed that their hypertension was an acute condition were less likely to continue taking anti-hypertensive medication than those who believed it to be a chronic condition. This study also showed that patients’ representations of their illness often conflicted with the medical view and provided an insight into the effects of mismatch between the patients’ representations and those of their doctor. In a group of 50 patients who had continued in treatment, 80 per cent agreed with the statement that ‘people cannot tell when their blood pressure is up’. However, 92 per cent believed that they could tell when their own blood pressure was raised by monitoring symptoms such as tiredness, headache and stress. Patients who believed their anti-hypertensive medication improved symptoms were more likely to adhere, which is consistent with more recent research on patients’ beliefs about the necessity of taking their medication (Horne, 2003: see below).
Illness perceptions have been linked with a range of adherence-related behaviours. These include various self-management behaviours, such as dietary control and blood glucose testing in diabetes, attending rehabilitation and the adoption of various lifestyle changes following myocardial infarction. In studies of non-insulin dependent diabetic patients, Hampson and colleagues have shown that personal models of diabetes are related to dietary self-management and to exercise adherence but not to the more medical aspects of control, such as blood glucose testing and taking medication (Hampson, 1997). Similarly, prospective studies of patients following first-time myocardial infarction (MI) have found that specific illness perceptions are predictive of different post-MI behaviours. Attendance at rehabilitation, which is prescribed for all patients, was predicted by the strength of their belief in the cure/control of their MI (Cooper, Lloyd, Weinman & Jackson, 1999; Petrie et al., 1996) whereas return to work depended more on the extent to which the patient saw their MI as having less serious consequences (Petrie et al., 1996). However, a number of studies have failed to show direct relations between illness perceptions and adherence levels, and have forced researchers to search for more specific cognitive predictors, focusing on beliefs about treatment. There is a small body of earlier work that has examined people’s beliefs about medicines and the ways in which these could influence adherence (Britten, 1994; Conrad, 1985; Donovan & Blake, 1992; Morgan & Watkins, 1988). The negative beliefs about medicines identified in these studies appear to be common across several illness and cultural groups and include worries about the possible harmful effects of medicines and about long-term dependence on them.
More systematic recent research by Horne and colleagues (Horne, 2003; Horne, Weinman & Hankins, 1999) indicates that four ‘core themes’ or factors underlie commonly held beliefs about medicines. Factor analysis of a pool of belief statements revealed two broad factors describing people’s beliefs about their prescribed medicines: their perceived necessity for maintaining health (specific necessity) and concerns based on beliefs about the potential for dependence or harmful long-term effects and that medication taking is disruptive (specific concerns). Two factors were also found to describe people’s beliefs about medicines in general. The first relates to the intrinsic properties of medicines and the extent to which they are harmful, addictive substances (general harm), and the second comprises views about whether medicines are overused by doctors (general overuse).
People’s views about the specific medication regimen prescribed for them were more strongly related to adherence reports than were more general views about medicines as a whole. Moreover, an interplay was found between concerns and necessity beliefs, which suggests that people engage in a risk-benefit analysis and consequently attempt to moderate the perceived potential for harm by taking less. Patients with stronger concerns based on beliefs about the potential for long-term effects and dependence reported lower adherence rates, whilst those with stronger beliefs in the necessity of their medication reported greater adherence to medication regimen (Horne, 2003; Horne et al., 1999). This research also showed that, although illness perceptions can explain variance in adherence, they may play a more important role in determining treatment beliefs. For example, in a study of factors influencing adherence to prevention medication, stronger beliefs about the necessity of taking this treatment were found in patients who perceived their asthma as a more chronic condition with more negative consequences (Horne & Weinman, 2002).
Another consistent determinant of adherence behaviour is the affective state of the patient. Negative affect, particularly depression, has been shown in a meta-analysis to be associated with lower levels of adherence to a wide range of treatments (DiMatteo, Lepper & Croghan, 2000). There are a number of ways in which depression could exert an influence on adherence behaviours. The negative cognitions associated with depressive states may well generalize to patients’ beliefs about their illness and treatment. Hence, in those with a physical illness, and who are also depressed, illness may be perceived as less controllable and with more negative consequences. Similarly, medications may also be perceived as having more negative side effects or as less potentially efficacious, as part of the generalized hopelessness experienced in depression (DiMatteo et al., 2000).
Interventions to Improve Adherence
The evidence that low rates of adherence are so widespread in health care at all levels from primary prevention to treatment for chronic conditions has provided the impetus for the development of a wide range of interventions. Many of the earlier attempts were predominantly information-based and sought to improve adherence by increasing patient knowledge and understanding. Although these achieved some small positive effects, they were generally of limited impact since behaviour change is very unlikely to be activated by information provision. Hence a range of other methods has been developed, and these have been described and evaluated in a number of reviews and meta-analyses (e.g., Haynes et al., 1996; Kok, van den Borne & Dolan Mullen, 1997; Macharia, Leon, Rowe, Stephenson & Haynes, 1992; Mullen, Green & Persinger, 1985; Roter, Hall et al., 1998). These studies have varied considerably in the nature of the intervention, the patient group and the study design. Roter, Hall et al. (1998) have grouped these interventions into four broad categories as follows.
These interventions are primarily concerned with the provision of information to increase the knowledge of the patients about their treatment and/or their illness (e.g., Barth, Campbell, Allen, Jupp & Chisholm, 1991; Gonzalez-Fernandez, Rivera, Torres, Quiles & Jackson, 1990). The information in these interventions has been provided in a number of differing formats (e.g., spoken, written, audiovisual) and has been presented in a range of ways (e.g., one-to-one, groups, by mail or telephone etc.).
These interventions are characterized by their use of basic behavioural principles (e.g., Grady, Goodenow & Borkin, 1988; Tucker, 1989). Thus they have been designed to work by targeting, shaping or reinforcing adherence behaviours to either ensure their adoption or increase their frequency. A range of approaches has been tried including the use of goal setting, financial rewards, contracts, skill building and behavioural modelling. As with the educational approaches described above, the behavioural interventions have been presented in a number of ways and contexts.
The affectively based methods comprise a range of approaches which attempt to influence by way of appeals to the patient’s emotions or by involving their social networks, particularly by using their social supports (Jay, DuRant, Shoffitt, Linder & Litt, 1984; Kellaway & McCrae, 1979). They include the use of various counselling approaches as well as the involvement of other family members to support the patient in their attempts to change and maintain their behaviour. They have been rarely used in isolation and are more commonly part of a broader intervention strategy.
Since it is now recognized that the behaviour of the health care professional in the consultation can have considerable effects on the subsequent affective, behavioural and health outcomes for the patient (Stewart et al., 1995), it is not surprising that approaches have been developed to change clinicians’ behaviour in order to improve patient adherence. These typically involve specific training interventions for doctors, nurses or pharmacists to increase adherence through improved communication, usually with clearer information and behavioural instructions (e.g., Berger et al., 1990). Thus they rely on changing clinicians’ behaviour in the first place, using a range of training methods and often with cues or reminders to activate the learned skills at appropriate times.
In many adherence intervention studies these approaches are used in combination to achieve maximum effect, but there have been a reasonably large number of studies that have used either educational or behavioural methods on their own (see Roter, Hall et al., 1998). One other important factor, which needs to be considered in categorizing adherence intervention strategies, is the extent to which they are based on a theory or model of adherence behaviour as well as a theory of behaviour change. For the most part, the interventions have not paid much or any attention to the determinants of non-adherence, and comparatively few take a theoretically based approach to changing adherence behaviour.
These interventions have been evaluated in terms of their effects on a wide range of adherence indicators and related outcomes with mixed success. The Roter, Hall et al. (1998) meta-analysis reveals a mixed pattern of efficacy of the various categories, but the broad picture is moderately optimistic since most types of intervention produce at least weak effects on adherence. No single type of intervention was shown to be clearly stronger than any other type but there is consistent evidence that those which used a range of methods are more effective than those which have a single focus. The overall finding was that the most powerful interventions included a blend of educational, behavioural and affective approaches, a conclusion similar to that found in an earlier review of patient education programmes in chronic illness (Mullen et al., 1985). From a health education perspective it is acknowledged not only that people vary in their preferences for different types of input but also that having a variety of approaches is a good way of increasing and maintaining learner interest (Green, Kreuter, Deeds & Partridge, 1980).
The meta-analysis also showed that evaluating the efficacy of any intervention depended very much on the type of outcome used. Interventions generally had their strongest effects on indirect measures of medication use, such as pill counts or refill records. Also, those studies that used refill records showed bigger effects than those relying on pill count measures, and that intervention effects on appointment making were consistently greater than the effects on appointment keeping. Both of these findings are taken to indicate the levels of commitment reflected in these different types of adherence indicator. Thus, while adherence interventions may have quite strong effects on patients’ intentions, a range of other factors will ultimately determine the extent to which intentions are translated into behaviours.
Implications for Communication
As with the approaches described earlier in this chapter, there is increasing evidence that, in order to improve patient adherence, communication needs to be tailored to make it personally salient to the individual patient. Hence effective adherence interventions will not only need to be multidimensional in their focus but also need to take account of patients’ pre-existing understanding and beliefs. One example of an intervention that has done this successfully was reported by Petrie et al. (2002), and was described earlier in this chapter.
The concept of effective communication has now broadened from an earlier focus upon patient satisfaction to one that incorporates engaging patients in their care. Such a broadening reflects a change in the political climate towards a consumerist role for patients. It also reflects the behavioural science evidence presented in this chapter, that to realize the benefits of predictive and therapeutic medicine requires communicating with patients from diverse social and cultural backgrounds to achieve an understanding commensurate with the problem being faced. Achieving such engagement requires specific communication skills to elicit and to work with patient representations as well as a broader appreciation of the patient’s social and cultural context. The challenge now is for health psychologists to generate evidence to increase our understanding of how representations of patients from diverse social and cultural groups are most effectively elicited and used in clinical consultations. This will form the basis for evidence-based training of health care professionals.