- Email: [email protected]
Using combined research methods for exploring diabetes patient education
Patient Education and Counseling 51 (2003) 45–52
Using combined research methods for exploring diabetes patient education Helen Coopera,*, Katie Boothb, Geoff Gillc a
Department of Primary Care, Liverpool University, Whelan Building, Brownlow Hill, Liverpool L69 3GB, UK b McMillan Practice Development Unit, Manchester University, Manchester, UK c Department of Diabetes and Endocrinology, University Hospital, Aintree, Liverpool, UK Received 25 May 2002; received in revised form 5 November 2002; accepted 2 December 2002
Abstract This paper explores a trial of an educational intervention designed for people who have Type 2 diabetes. The aim of the trial was to understand how the intervention had influenced outcomes in the context of participants’ everyday lives. A randomised-controlled wait-list trial design was used. The study was also informed by a qualitative approach which explored the meanings held by participants for informing their behaviours. Outcomes were measured using diabetes-specific questionnaires and clinical measures of blood glucose control, weight and drug treatment. Alongside these, focus group discussions were held to explore how outcome effects had transpired. Using these different methods resulted in two separate data sets which required diverse methods for analysis. This paper uses examples of compatibilities and contradictions between the data sets to look at how they were combined to produce valid results. Conclusions drawn showed that a combined methods approach was essential to expand the scope and improve the analytic power of trials of patient education. It produced illuminating results which provided guidelines for practice and suggested further areas for research. # 2003 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Patient education; Diabetes; Complex interventions; Combined research methods
1. Introduction The prevalence of Type 2 diabetes is predicted to rise over the next decade which will pose a major public health challenge in the near future [1]. Such a challenge relates to the pattern of this disease which can lead to chronic complications if poorly managed. The United Kingdom Prospective Diabetes Study clearly showed that long-term outcomes for patients with Type 2 diabetes can be improved with careful self-management aimed at controlling glucose, lipid and blood pressure levels [2]. In the United Kingdom, this has led to increasing calls for better education for people with diabetes to enhance their personal control over the dayto-day management of their diabetes [3]. The focus of such self-management education is on three interconnecting elements, namely lifestyle, drug treatments, and surveillance strategies for early detection and treatment of complications. Various reviews have been undertaken to try to identify suitable education models that meet this requirement [4–7]. *
Corresponding author. Tel.: þ44-151-794-5612; fax: þ44-151-794-5604. E-mail address: [email protected] (H. Cooper).
All have identified a gap in available models that have been rigorously evaluated over a long-term period. In addition, no research study has addressed the fact that diabetes patient education is a complex intervention, and that recognition of this is crucial if it is to become evidence-based. Complete evaluation therefore needs to follow a phased approach, each phase requiring different but linked methodologies from a variety of research traditions. These phases have been described by Campbell et al. [8] as part of a strategic Medical Research Council document and are summarised in Table 1. This paper describes how such an approach was utilised in a study of diabetes patient education.
2. Methods 2.1. Pre-clinical phase A review by the authors of patient education research for a range of chronic diseases found that interventions tested were poorly described and had failed to adhere to theoretical models [4]. Lack of methodological rigour in the research was also found making it impossible to judge the effects of
0738-3991/$ – see front matter # 2003 Elsevier Science Ireland Ltd. All rights reserved. doi:10.1016/S0738-3991(02)00265-3
46
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
Table 1 Phases required to evaluate complex interventions Phase
Structure
Pre-clinical/theoretical
Exploration of relevant theory to: direct choice of intervention and hypothesis; predict major confounders and strategic design issues Identification of components of intervention and underlying mechanisms which will influence outcomes to provide evidence of how they relate to and interact with each other Description of constant and variable components of a replicable intervention and a feasible protocol for comparing the intervention with an appropriate alternative Use of a protocol that is theoretically defensible, reproducible and adequately controlled in a study with appropriate statistical power Reliable replication of intervention and results in uncontrolled settings over the long term
I: modelling II: exploratory III: randomised-controlled trial IV: long-term implementation
patient education. Where randomised-controlled trials (RCT) had been conducted, the effect size was usually small and was only known for 6 months of follow-up. The findings from this review have been reinforced by other researchers [5–7]. 2.2. Phase I: modelling In response to these findings we constructed and trialed an intervention based on the theories associated with health education (empowerment, radical and preventive theories), adult education (humanist, cognitive and behaviourist theories), and health protective behaviours (theories of reasoned action, stages of change and personal models of illness) [9]. Research evidence suggested consistencies between them in relation to their focus on learning as a continuous process grounded in experience and which takes into account the role of emotional feelings and choice alongside cognition. This aligns patient education to an experiential learning process within which beliefs about self-efficacy and the effects of social-environmental influences are central to outcomes. It highlighted the importance of education which does not necessarily result in behavioural change, but which may affect changes across a range of variables including: knowledge; skills; understanding; attitudes and/or behaviour. These findings illustrated how the chosen theories inter-related to provide a model which could be used to direct intervention goals. The intervention comprised eight-weekly sessions, each lasting approximately two hours. Course content included exercise and relaxation training and health topics covering information about the diabetes disease process, an overview of self-management principles, exploration of feelings about having diabetes, and how to make the best use of the health service. The format of the course was largely interactive using workshops, and group work underlay its philosophy. It was delivered by nurse tutors who were trained together and provided with a teaching manual to ensure standardisation of content. Training was undertaken by a specialist team of health educators working in liaison with the first author who was employed as a diabetes specialist nurse at the time of the study. Trainee tutors attended a taught course run over 12 days with an additional time requirement for home study
and preparation. The course provided a theoretical understanding of how people change and also the practical skills of change management. It enabled tutors to increase their understanding of adult learning and group work skills and they were taught how to run an exercise and relaxation programme. Ideas for teaching a variety of health topics relating to lifestyle were also covered including healthy eating, weight management, exercise, handling stress, alcohol and smoking. Validation was based on trainer evaluation of micro-teaching sessions within the training group, and of a pre-registration course. At this time, successful validation resulted in a Royal Society of Arts certification which was later converted to a National Vocational Qualification. The primary aim of the patient’s course was to develop self-management skills with the aim of improving physiological outcomes, mediated through changes in knowledge and attitudes to the disease and its treatment, and through improved patient self-confidence. The research study aimed to measure outcomes in relation to illness beliefs (attitudes and personal models), self-care behaviours (diet, exercise and self-monitoring) and physiological outcomes (blood glucose control, weight and drug treatments). It also aimed to evaluate how the intervention had influenced these outcomes in the context of participants’ everyday lives. 2.3. Phase II: exploratory trial Recent trends in evidence-based practice have reinforced the need for RCT designed studies. However, any research design involving patient education has to take account of ethical limitations whereby a control group cannot be excluded from the intervention. These limitations were overcome by using a randomised-controlled ‘wait-list’ trial design whereby all patients were allocated to the treatment but at different times [10]. Half of the trial group were therefore randomly allocated to the treatment initially, whilst the other half acted as the short-term control group. These two groups then combined to form the long-term trial group. The longterm control group was offered the treatment at the end of the trial period. In this study, five trial interventions were delivered at staggered intervals over a period of 14 months, as shown in Table 2.
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
47
Table 2 Trial protocol Time (months)
Group A (n ¼ 30)
Group B (n ¼ 23)
Group C (n ¼ 36)
0
Baseline data Interviews Questionnaires Clinical measures # INTERVENTION Week 8 Focus groups
Baseline data Interviews Questionnaires Clinical measures # Short-term Control Period
Baseline data Interviews Questionnaires Clinical measures # Long term Control Period
6
Clinical measures Questionnaires
Pre-course Clinical measures Questionnaires # INTERVENTION Week 8 Focus groups
12
Clinical measures Questionnaires Focus groups
Clinical measures Questionnaires
18
Using a controlled type design ruled out threats to valid inference. It maximised the number of people taking part in the trial, and permitted collection of data to measure cause and effect constructs over short- and long-term periods. It also addressed the ethical and psychological issues involved in recruitment and allocation of patients to treatment within a trial. In this way, it ruled out threats to internal validity associated with inequities between the way participants are treated [11]. However, this method precluded the collection of data on process indicators. Denzin and Lincoln [12] describe the RCT as separating the variables of interest from their local everyday milieus, entering them into a controlled research environment, and then trying to fit the results back into the original context. This did not account for the experiences of participants who were dealing with their disease as a ‘lived experience’. Exploration of how they connected their learning experiences to their individual worlds was considered to be essential for understanding the inter-relationships between the quantitative findings and the social factors that shaped these outcomes. To meet these needs a qualitative approach was required. In qualitative research there are a range of theoretical frameworks which provide guidelines for the research design. Livta and Eyles [13] argue that unlike quantitative research, where the RCT is taken as the gold standard, there is no contest in qualitative research as to which theory describes the social world better. It is rather a matter of which is most appropriate for the task in hand. For the study of patient education, ‘symbolic interactionism’ is the
Pre-course Clinical measures Questionnaires # INTERVENTION
Clinical measures Questionnaires Focus groups
preferred theoretical perspective as its roots lie in behavioural psychology. Blumer [14] describes three principles of interactionism: the first is that people will act towards things based upon the meaning that they have attached to them. The second is that meanings are attached to things as a result of social interactions with other people. The third is that meanings are altered and negotiated through an interpretative process used by individuals in dealing with things that they encounter. This perspective emphasises both the individualistic nature of the diabetes experience and the social nature of living with the disease. Mead [15] described how these meanings are not regarded as being intrinsic but arise through the process of social interaction, such as that encouraged in educational group sessions. The use of this perspective therefore recognised three important criteria for this trial. Firstly, that it treats patients as active participants in their care. Secondly, that it recognises the interaction between the body and the mind and acknowledges internal forces (conversations with self) as well as external forces (conversations with others) as having impact on how diabetes is perceived and how it is responded to. Finally, it acknowledges that perceptions of an illness are influenced by socialenvironmental factors. These three factors correlated with the theoretical framework identified for the study. 2.4. Phase III: trial protocol Using a dual methods approach meant that we needed to explore how the two sets of results could fit together.
48
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
Both approaches were given equal emphasis in this study but differences between them needed to be understood from the analytical and interpretative perspectives. Bryman [16] describes how such differences are often marshalled as evidence of the essential incompatibility of the two approaches. This leads into the debate about whether you can actually combine the two methods. This debate centres on differences in relation to paradigms and their associated methods. Paradigms of inquiry are world views that signal distinctive ideas about the nature of knowledge which translates into issues of scientific methodology [17]. As the roots of quantitative and qualitative research lie at either end of a spectrum of research methods, the different approaches need to remain distinct. This trial therefore essentially involved two separate studies that were linked by their evidential and theoretical base to form a scientific process. In this study, the following instruments were used to research outcomes: Structured interview tool to collect quantitative data on the characteristics of those taking part in the trial. Diabetes-specific/validated questionnaires measuring attitudes to diabetes and its treatment [18], personal models of diabetes [19], and behaviour including exercise, diet and self monitoring [20]. Physiological measures of blood glucose control (HbA1c), weight (body mass index) and changes in drug treatment. Focus group discussions using a semi-structured interview tool to explore group perceptions of the educational process and its effects in ‘real life’. Use of multiple outcome measures meant that different dimensions of patient education could be captured. It therefore expanded the scope of the study in two ways: firstly it allowed for triangulation, to ensure corroboration of data; and secondly it allowed for complementarity, to clarify, explain and more fully elaborate the results of the analyses. The aim of using this approach was to improve the quality of the research and confirm the conclusions induced from the data. In this way we hoped to understand how the educational intervention had influenced outcomes in the context of participants’ everyday lives. Creating such an account, however, relied upon exploration of both compatibilities and contradictions between the data sets. This required strong theoretical underpinnings to allow sense to be made of the combined data. This point has been clearly demonstrated in the work of a number of researchers which have been reviewed by Brannen [21] and Carey [22], although none could be located in the field of diabetes research. 2.5. Analysis of the data Analysis of both data sets followed rigorous methods. The quantitative data was processed using the statistical package for the social sciences (SPSS). Parametric data was compared using independent sample t-tests and non-parametric
data using the Mann–Whitney U-test. Regression analysis was used to compare changes in HbA1c levels, allowing for differences in participants’ baseline levels. Categorical data was compared using w2, and McNemar’s test was used to measure categorical changes within groups. Results were considered significant with the conventional P 0:05, although the ‘P’ values were considered to be an indicator of differences rather than a conclusive finding. How the quantitative and qualitative data fitted together was considered to be more indicative a test of the hypotheses. The qualitative data was examined and the meanings and relevant implications for the research questions explored. All the interviews were tape recorded (with informants’ permission) and transcribed verbatim. The constant comparative method was used to develop categories and meanings. These were then tested and modified through addition cycles of data collection, analysis and systematic search for negative cases [23]. In addition, a second researcher checked the written interpretations for confirmatory purposes.
3. Results Various publications have reported on the overall findings of this trial [4,9,24]. This paper looks at the compatibilities and contradictions between the data sets with a focus on the main aims of the study which were to measure quantitative outcomes and to explore how the intervention had influenced these outcomes in the context of participants’ everyday lives. Table 3 provides details of the characteristics of those who took part in the trial. Overall, participants in both the shortand long-term trial and control groups were comparable in relation to their demographic, medical, and social characteristics. 3.1. Clinical outcomes Significant differences for blood glucose (HbA1c) levels were found at 6 months only (P ¼ 0:005 (6 months), P ¼ 0:84 (12 months)). This reflected differences in the rates of deterioration in HbA1c, with the trial group showing less deterioration than the control group. Baseline levels of HbA1c were found to predict outcome at 12 months (P ¼ 0:01) so that those entering the trial with higher levels were significantly more likely to improve their blood glucose levels longitudinally. This demonstrated the success of the intervention at raising perceptions about glycaemic risk factors in those most at risk and this was confirmed during the focus group discussions. There was also significant correlation between attendance rates to the course and reductions in HbA1c levels at 12 months (P ¼ 0:05). This finding highlighted the importance of attending all eight of the educational sessions for maximum clinical benefit. The fact that effects on HbA1c did not persist longitudinally was mirrored by a lack of significant effect on weight (P ¼ 0:84, P ¼ 0:11), but wide variations were noted
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52 Table 3 Demographic, medical and social characteristics of trial participants (n ¼ 89) Participant characteristics Mean age (year) Male:female (%) Mean time since diagnosis (year) Employed (%) Unemployed (%) Retired (%) Mean HbA1c (%) Desirable weight (BMI) (%) Current smokers Alcohol intake Lives alone (%) Perception of effect of diabetes on lifestyle (%) Treatment Diet only (%) Diet and tablets (%)
58.6 (range 35–73) 56:44 5.7 (range 1–30) 35 35 30 7.5 (range 4.6–11) 7 21 11% above safe limits 17 57
28 72
Concurrent complications Hypertension (%) Hyperlipidaemia (%) Ischaemic heart disease (%) Retinopathy (%) Neuropathy (%)
25 26 18 11 12
Management of diabetes Shared care (hospital/GP) (%) GP care only (%)
82 18
Sources of information on diabetes Leaflets (%) Practice nurse and GP (%) Diabetes specialist nurse (%)
21 20 17
Table 4 Overview of changes in HbA1c and weight using individual patient data Blood glucose (HbA1c)
Short-term Short-term Long-term Long-term trial control trial control (6 months) (6 months) (12 months) (12 months)
Improved (%) Deteriorated (%) No change (%) Range (1)
53 17 43 43 74 51 3 9 6 3.8 to 2.6 0.2 to 3.2 5.4 to 5.6 57 53 29 40 14 6 5.6 to 1.1 4.6 to 6.4
Minus ranges denote improvement.
30%) had their drug treatment increased relative to baseline, but differences were not significant (w2, P ¼ 0:16). During the focus groups, participants spoke about using their new found competencies to self-monitor their drug treatment and increase their participation in the medical consultation. They found, however, that health professionals did not always embrace their self-perceived needs, and talked about the difficulties they had in getting them to respond in what they considered to be an appropriate fashion. They recognised a lack of specialist knowledge in primary care which would have permitted the on-going support that many participants felt they now needed. This was particularly apparent for those requiring a change from oral therapy to insulin treatment. There were 19 participants (21%) who were on combined maximum therapy at the start of the trial. Of these 19, only four were changed to insulin therapy over the trial period despite these people recognising their need for such a change. As a progressive disease, Type 2 diabetes requires continual systematic monitoring of treatment. The lack of an appropriate response by health professionals highlighted the interaction of factors that can affect the clinical outcomes associated with educational trials. 3.2. Psychological status
around the means for both these variables. Analysis using individual patient data found that the trial groups had greater numbers of people improving their HbA1c levels than the control groups. For reductions in weight, this position was more pronounced at 12 months, reinforcing the need for weight loss to be measured longitudinally (Table 4). Changes in drug treatment at 12 months showed that proportionally more patients in the trial group (46% versus
Weight (BMI) Improved (%) 61 Deteriorated (%) 36 No change (%) 4 Range (1) 3.9 to 5.0
49
19 71 10 3.8 to 2.6 43 43 13 4.2 to 1.9
Significant differences between the groups for changes in attitudes were seen at both 6 months (P ¼ 0:04) and 12 months (P ¼ 0:01), whilst changes in perceptions of selfcare treatment effectiveness were seen at 6 months only (P ¼ 0:02, P ¼ 0:23). There were no significant differences for changes in perceptions of seriousness of diabetes or personal control. For both groups, diabetes was perceived to be ‘fairly’ serious, pre-and post-trial, illustrating that they all showed some concern about the gravity of their disease and its threat to their future health. The mean and median values for self-care treatment effectiveness were higher than the values for seriousness for all participants. This showed that beliefs about the value of treatment were stronger than beliefs about the seriousness of diabetes. However, only the trial groups showed an increase in their beliefs about the value of self-care treatment, as shown by an increase in median scores at 6 and 12 months. All participants felt they had a moderate level of control over their diabetes. This perception did not change over the trial period. However, the data showed that at 6 months, 71% of the trial group and 67% of the control group continued to feel, or had moved toward feeling more in control of their diabetes. These proportions had changed to 74 and 52% respectively at 12 months. The control group’s perceptions had therefore deteriorated, whilst the trial group had improved their perceptions of personal control. This trend was reflected by changes in the P values over time: 6 months P ¼ 0:72; 12 months, P ¼ 0:06. During the focus groups participants spoke about a heightened awareness of the seriousness of their type of
50
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
diabetes and acceptance of its permanence. This was reflected by a change in their perception of Type 2 diabetes being a ‘mild’ disease. These findings contradicted those from the questionnaire. However, such inconsistency was explained by reference to the theories of personal models of illness [25]. During the focus groups it was found that participants’ perceptual changes were balanced against a reduction in feelings of helplessness, less worry about the disease, and feelings of having been provided with the ‘tools’ to control any threats. They had therefore altered their personal models of diabetes and with it, their coping patterns. Such (re)appraisal resulted in participants feeling that diabetes had become an integral part of them and not something to be dealt with outside their normal living experiences. Feeling that the disease was less of a threat did not therefore mean that they viewed it as being less serious. The outcome of this learning process varied, depending upon environmental and psycho-social constraints, which appeared to both positively and negatively affect health outcomes. This was illustrated by the quantitative data which found significant differences in perceptions of treatment effectiveness at 6 months only. The fact that such differences were not shown later, after participants had had a chance to test out their new found skills and knowledge, indicated that their beliefs had been affected by experiences in the ‘real world’. This concurred with the clinical outcomes. 3.3. Self-management behaviours There were no significant differences between the groups for changes in exercise and diet scores. The trial groups, however, showed larger increases in their mean scores than the control groups, but wide variations around the means were noted for all the groups. The number of participants self-monitoring increased over the two time periods for both the trial and control groups, although differences were only significant at 12 months (P ¼ 0:21, P ¼ 0:002). Pre-trial, 67% of participants in the long-term trial group and 47% in the long-term control group were self-monitoring. At 12 months, this proportion had changed to 92 and 63%, respectively. Changes for the trial group related to highly significant increases in the numbers blood glucose testing (McNemar’s test, P < 0:001), but not for the control group (McNemar’s test, P ¼ 0:13). During the focus groups participants described how living with diabetes often meant that compromise was required to be able to conform to the treatment regime. They discussed how they viewed what happened in ‘real life’ to be of great importance to their health outcomes and they wanted this to be acknowledged by health practitioners. They spoke of intentions to change, as well as behaviour change. This highlighted how the quantitative questionnaire data had focused upon single outcomes, i.e. changes in exercise and diet scores, rather than on the process of changing these
variables. This meant that some were contemplating, some preparing, some taking action, some maintaining change, or some relapsing [26]. This qualitative data tied in with the wide variations around the quantitative mean. The intervention therefore appeared to act as a catalyst, moving participants into a spiral of behavioural change, where support strategies were essential to help participants cope. Treatment relapse was discussed by some participants in relation to loss of support at home or from the group which reinforced the importance of having such systems in place. When participants were asked about their self-monitoring practices at 6 months, some participants described encountering discrepancies between what their doctors advised and what they now perceived to be important. At 12 months this situation had changed and participants described how the intervention had empowered them to negotiate for the equipment they required, i.e. they had developed greater self-efficacy. This linked into changes in their perceptions of personal control. Participants spoke of self-monitoring allowing them to ‘see’ and thus learn how their metabolism deals with such things as variations in meals and activities. This had motivated adherence to the behaviour.
4. Discussion and conclusions 4.1. Discussion This study aimed to explore the outcomes of patient education by measuring effects upon participants’ self-care behaviours and their illness beliefs. Alongside this it investigated how the intervention had influenced these outcomes in the context of participants’ everyday lives. The main limitation of the trial related to the number of people who agreed to take part in the study. Recruitment demonstrated that patients differed in their willingness to be educated, as reflected by the fact that only 40% of people asked to take part agreed to do so. How typical the trial participants were in terms of the overall population of patients with Type 2 diabetes could best be understood by studying those who refused to take part in the educational intervention. Apart from characteristics relating to age, ethnicity and sex ratio (for which there were no differences) this data was not collected and is a limitation to generalising the results of this study. It also implies the need for further research into the reasons why people with Type 2 diabetes choose to engage (or not) in educational initiatives. For those who did choose to take part, it can only be presumed that they were orientated toward wanting a different kind of management for their diabetes. This has implications for the results of the study because it suggests that some or all of the participants were ready to make changes. However, by using a randomised-controlled trial design, which achieves comparability by equating the average unit within each of the groups, this limitation was minimised [11].
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
The strength of the study lay in its design and the fact that there was considerable overlap between the quantitative and qualitative data sets. This meant that there was less probability of drawing erroneous conclusions and served to validate the findings from each method. Where inconsistencies existed, explanation was sought from the theories used to underpin the trial. From a purely quantitative perspective the findings show that patient education appears to have short-term clinical effects, long-term psychological effects and no effects on lifestyle behaviours. However, when looked at from a combined data perspective the conclusions drawn are quite different and clearly demonstrate the value of using a combined methods approach. Firstly, whilst there were no significant changes in diet and exercise behaviour participants showed that they had entered a behaviour change cycle. The focus group discussions highlighted the lack of an effective quantitative tool to measure these processes. From this perspective behaviours, rather than just physiological outcomes, need to be rewarded if changes are to be encouraged and maintained. This finding has implications for clinical practice and future research. Secondly, the data on self-monitoring revealed participants’ perceptions of lack of support for this practice. Coster et al. [27] conducted a meta-analysis of six RCTs which had compared blood testing, urine testing and no testing in Type 2 diabetes. There were no significant differences in HbA1c outcomes for the three methods. Such findings may well have influenced policy decisions on providing blood testing equipment for such patients. However, Coster et al. also found only anecdotal evidence concerning patients’ perceptions of monitoring, that patient training was not addressed systematically, and that no trial required subjects to modify their drug therapy in accordance with self-monitoring results. Whilst the intervention provided a standardised educational programme, participants were not provided with information about drug self modification in accordance with self-monitoring results. Rather, they were directed to seek advice from health professionals where participants reported specialist knowledge to be inconsistent. They also felt that health professionals were not always willing to value their expertise and spoke angrily about the tendency to focus on the physical aspects of their condition rather then on their emotional needs. This appeared to have acted as a barrier to effective learning in the past and has implications for professional training and future research studies. Thirdly, the findings are consistent with the psychological theories chosen to underpin the trial and illustrate how they can be integrated to form a framework for guiding intervention to enhance patient self-management. This model argues for the need to expand the traditional approach to patient education so that it allows for patients’ self-perceived needs. This highlights the active role that patients need to play so that they become partners in their health care process and calls into question the relationships between health professionals and patients. Empowerment, described by the United Kingdom government in terms of ‘patient expertise’
51
[28], involves relationships with others because it is ‘‘. . . nurtured by the effects of collaborative efforts’’ [29]. Our study has verified this statement and has highlighted the training needs of health professionals at both under- and post-graduate levels.
5. Practice implications By using a combined methods approach to research patient education this trial has explicated both its effectiveness and its shortcomings. It has provided a vehicle for bridging the gap between scientific evidence and real life practice. It has demonstrated that questionnaires, clinical tests and focus groups provide access to different types of knowledge and that they are all equally valid. What was found to be important, however, was how the different types of data ‘fit’ together. For this to be achieved strong theoretical underpinnings were required, including an understanding of the theoretical philosophies of research. This trial has shown that the use of such an approach led to a reinforcement of theory, suggested new research directions and provided guidance for clinical practice and professional education. An RCT alone would have restricted applicability of our findings because it does not address the human elements of living with a disease such as diabetes. It has therefore strengthened our conviction that a combined methods approach is required for researching complex interventions such as patient education.
Acknowledgements The authors would like to thank Diabetes United Kingdom who funded this research study. References [1] Amos AT, McCarty DJ, Zimmet P. The rising global burden of diabetes and its complications: estimated predictions to the year 2010. Diabet Med 1997;14(Suppl 5):S1–S85. [2] Turner R, Cull C, Holman RR. United Kingdom prospective diabetes study 17: a 9-year update of randomised controlled trials on the effect of improved metabolic control on complications in non-insulin controlled diabetes. Ann Int Med 1996;124:136–45. [3] Audit Commission. Testing times: a review of diabetes services in England and Wales. The Audit Commission for local authorities and the National Health Service in England and Wales, London; 2000. [4] Cooper H, Booth K, Fear S, Gill G. Lessons from chronic disease patient education. Patient Educ Couns 2001;44:107–17. [5] Griffin S, Kinmouth AL, Skinner C, Kelly J. Educational and psychosocial interventions for adults with diabetes article (online), March 1999. Summary of report available on: http://www.diabetes.org.uk. [6] Roter DL, Hall JA, Merisca R, Nordstrom B, Cretin D, Svarstad B. Effectiveness of interventions to improve patient compliance: a meta-analysis. Med Care 1998;36:1138–61. [7] Skinner TC, Cradock S. Empowerment: what about the evidence? Pract Diab Int 2000;17:91–5.
52
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
[8] Campbell M, Fitzpatrick R, Haines A, Kinmonth AL, et al. Framework for design and evaluation of complex interventions to improve health. Br Med J 2000;321:694–6. [9] Cooper H, Booth K, Fear S, Gill G. Patients’ perspectives on diabetes health care education. Health Educ Res 2003;18:101–16. [10] Bradley C. Designing medical and educational intervention studies. Diabetes Care 1993;16:509–18. [11] Cook DC, Campbell DT. Quasi-experimentation: design and analysis issues for field settings. Boston MA, USA: Houghton Miffllin; 1979. [12] Denzin NK, Lincoln YS. Handbook of qualitative research. Beverley Hills, CA: Sage; 1994. [13] Livta A, Eyles J. Coming out: exposing social theory in medical geography. Health Place 1995;1:5–14. [14] Blumer H. Symbolic interactionism: perspective and method. Englewood Cliffs: Prentice-Hall; 1969. [15] Mead GH. Mind, self and society. Chicago: University of Chicago Press; 1962. [16] Bryman A. Quantity and quality in social research. London: Routledge; 1988. [17] Sandelowski M. Combining qualitative and quantitative sampling, data collection, and analysis techniques in mixed-methods studies. Res Nurs Health 2000;23:246–55. [18] Welch G, Beeney L, Dunn S, Smith RBW. The development of the diabetes integration scale: a psychometric study of the ATT 39. Multiva Exp Clin R 1994;11:75–88. [19] Glasgow RE, Toobert DJ, Hampson S, Noell JW. A brief office based intervention to facilitate diabetes self-management. Health Educ Res 1995;10:467–78.
[20] Toobert DJ, Glasgow RE. Assessing diabetes self-management: the summary of diabetes self-care activities questionnaire. In: Bradley C, editor. Handbook of psychology and diabetes, vol. 15. Switzerland: Harwood; 1994. p. 351–75. [21] Brannen J. Mixing methods: qualitative and quantitative research. Aldershot: Avebury; 1992. [22] Carey JW. Linking qualitative and quantitative methods: integrating cultural factors into public health. Qual Health Res 1993;3:298–318. [23] Denzin NK. The art and politics of interpretation. In: Denzin NK, Lincoln YS, editors. Handbook of qualitative research. Beverley Hills, CA: Sage; 1994. [24] Cooper H, Booth K, Gill G. Diabetes education: the patient’s perspective. J Diabetes Nurse 2002;6:91–5. [25] Leventhal H, Nerenz DR, Steele DJ. Illness representations and coping with health threats. In: Baum A, Taylor SE, Singer JE, editors. Handbook of psychology and health. Hillside NJ: Erlbaum; 1984. [26] Prochaska JO, DiClemente CC, Norcross JC. In search of how people change: applications to addictive behaviours. Am Psychol 1992;47: 1102–14. [27] Coster S, Gulliford MC, Seed PT, Powrie JK, Swaminathan R. Monitoring blood glucose control in diabetes mellitus: a systematic review. Health Technol Assess 2000;4. [28] Department of Health. The expert patient: a new approach to chronic disease management for the 21st century. London: HMSO (Books); 2001. [29] Keiffer C. Citizen empowerment: a developmental perspective. Prev Hum Serv 1984;3:9–36.
Using combined research methods for exploring diabetes patient education Helen Coopera,*, Katie Boothb, Geoff Gillc a
Department of Primary Care, Liverpool University, Whelan Building, Brownlow Hill, Liverpool L69 3GB, UK b McMillan Practice Development Unit, Manchester University, Manchester, UK c Department of Diabetes and Endocrinology, University Hospital, Aintree, Liverpool, UK Received 25 May 2002; received in revised form 5 November 2002; accepted 2 December 2002
Abstract This paper explores a trial of an educational intervention designed for people who have Type 2 diabetes. The aim of the trial was to understand how the intervention had influenced outcomes in the context of participants’ everyday lives. A randomised-controlled wait-list trial design was used. The study was also informed by a qualitative approach which explored the meanings held by participants for informing their behaviours. Outcomes were measured using diabetes-specific questionnaires and clinical measures of blood glucose control, weight and drug treatment. Alongside these, focus group discussions were held to explore how outcome effects had transpired. Using these different methods resulted in two separate data sets which required diverse methods for analysis. This paper uses examples of compatibilities and contradictions between the data sets to look at how they were combined to produce valid results. Conclusions drawn showed that a combined methods approach was essential to expand the scope and improve the analytic power of trials of patient education. It produced illuminating results which provided guidelines for practice and suggested further areas for research. # 2003 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Patient education; Diabetes; Complex interventions; Combined research methods
1. Introduction The prevalence of Type 2 diabetes is predicted to rise over the next decade which will pose a major public health challenge in the near future [1]. Such a challenge relates to the pattern of this disease which can lead to chronic complications if poorly managed. The United Kingdom Prospective Diabetes Study clearly showed that long-term outcomes for patients with Type 2 diabetes can be improved with careful self-management aimed at controlling glucose, lipid and blood pressure levels [2]. In the United Kingdom, this has led to increasing calls for better education for people with diabetes to enhance their personal control over the dayto-day management of their diabetes [3]. The focus of such self-management education is on three interconnecting elements, namely lifestyle, drug treatments, and surveillance strategies for early detection and treatment of complications. Various reviews have been undertaken to try to identify suitable education models that meet this requirement [4–7]. *
Corresponding author. Tel.: þ44-151-794-5612; fax: þ44-151-794-5604. E-mail address: [email protected] (H. Cooper).
All have identified a gap in available models that have been rigorously evaluated over a long-term period. In addition, no research study has addressed the fact that diabetes patient education is a complex intervention, and that recognition of this is crucial if it is to become evidence-based. Complete evaluation therefore needs to follow a phased approach, each phase requiring different but linked methodologies from a variety of research traditions. These phases have been described by Campbell et al. [8] as part of a strategic Medical Research Council document and are summarised in Table 1. This paper describes how such an approach was utilised in a study of diabetes patient education.
2. Methods 2.1. Pre-clinical phase A review by the authors of patient education research for a range of chronic diseases found that interventions tested were poorly described and had failed to adhere to theoretical models [4]. Lack of methodological rigour in the research was also found making it impossible to judge the effects of
0738-3991/$ – see front matter # 2003 Elsevier Science Ireland Ltd. All rights reserved. doi:10.1016/S0738-3991(02)00265-3
46
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
Table 1 Phases required to evaluate complex interventions Phase
Structure
Pre-clinical/theoretical
Exploration of relevant theory to: direct choice of intervention and hypothesis; predict major confounders and strategic design issues Identification of components of intervention and underlying mechanisms which will influence outcomes to provide evidence of how they relate to and interact with each other Description of constant and variable components of a replicable intervention and a feasible protocol for comparing the intervention with an appropriate alternative Use of a protocol that is theoretically defensible, reproducible and adequately controlled in a study with appropriate statistical power Reliable replication of intervention and results in uncontrolled settings over the long term
I: modelling II: exploratory III: randomised-controlled trial IV: long-term implementation
patient education. Where randomised-controlled trials (RCT) had been conducted, the effect size was usually small and was only known for 6 months of follow-up. The findings from this review have been reinforced by other researchers [5–7]. 2.2. Phase I: modelling In response to these findings we constructed and trialed an intervention based on the theories associated with health education (empowerment, radical and preventive theories), adult education (humanist, cognitive and behaviourist theories), and health protective behaviours (theories of reasoned action, stages of change and personal models of illness) [9]. Research evidence suggested consistencies between them in relation to their focus on learning as a continuous process grounded in experience and which takes into account the role of emotional feelings and choice alongside cognition. This aligns patient education to an experiential learning process within which beliefs about self-efficacy and the effects of social-environmental influences are central to outcomes. It highlighted the importance of education which does not necessarily result in behavioural change, but which may affect changes across a range of variables including: knowledge; skills; understanding; attitudes and/or behaviour. These findings illustrated how the chosen theories inter-related to provide a model which could be used to direct intervention goals. The intervention comprised eight-weekly sessions, each lasting approximately two hours. Course content included exercise and relaxation training and health topics covering information about the diabetes disease process, an overview of self-management principles, exploration of feelings about having diabetes, and how to make the best use of the health service. The format of the course was largely interactive using workshops, and group work underlay its philosophy. It was delivered by nurse tutors who were trained together and provided with a teaching manual to ensure standardisation of content. Training was undertaken by a specialist team of health educators working in liaison with the first author who was employed as a diabetes specialist nurse at the time of the study. Trainee tutors attended a taught course run over 12 days with an additional time requirement for home study
and preparation. The course provided a theoretical understanding of how people change and also the practical skills of change management. It enabled tutors to increase their understanding of adult learning and group work skills and they were taught how to run an exercise and relaxation programme. Ideas for teaching a variety of health topics relating to lifestyle were also covered including healthy eating, weight management, exercise, handling stress, alcohol and smoking. Validation was based on trainer evaluation of micro-teaching sessions within the training group, and of a pre-registration course. At this time, successful validation resulted in a Royal Society of Arts certification which was later converted to a National Vocational Qualification. The primary aim of the patient’s course was to develop self-management skills with the aim of improving physiological outcomes, mediated through changes in knowledge and attitudes to the disease and its treatment, and through improved patient self-confidence. The research study aimed to measure outcomes in relation to illness beliefs (attitudes and personal models), self-care behaviours (diet, exercise and self-monitoring) and physiological outcomes (blood glucose control, weight and drug treatments). It also aimed to evaluate how the intervention had influenced these outcomes in the context of participants’ everyday lives. 2.3. Phase II: exploratory trial Recent trends in evidence-based practice have reinforced the need for RCT designed studies. However, any research design involving patient education has to take account of ethical limitations whereby a control group cannot be excluded from the intervention. These limitations were overcome by using a randomised-controlled ‘wait-list’ trial design whereby all patients were allocated to the treatment but at different times [10]. Half of the trial group were therefore randomly allocated to the treatment initially, whilst the other half acted as the short-term control group. These two groups then combined to form the long-term trial group. The longterm control group was offered the treatment at the end of the trial period. In this study, five trial interventions were delivered at staggered intervals over a period of 14 months, as shown in Table 2.
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
47
Table 2 Trial protocol Time (months)
Group A (n ¼ 30)
Group B (n ¼ 23)
Group C (n ¼ 36)
0
Baseline data Interviews Questionnaires Clinical measures # INTERVENTION Week 8 Focus groups
Baseline data Interviews Questionnaires Clinical measures # Short-term Control Period
Baseline data Interviews Questionnaires Clinical measures # Long term Control Period
6
Clinical measures Questionnaires
Pre-course Clinical measures Questionnaires # INTERVENTION Week 8 Focus groups
12
Clinical measures Questionnaires Focus groups
Clinical measures Questionnaires
18
Using a controlled type design ruled out threats to valid inference. It maximised the number of people taking part in the trial, and permitted collection of data to measure cause and effect constructs over short- and long-term periods. It also addressed the ethical and psychological issues involved in recruitment and allocation of patients to treatment within a trial. In this way, it ruled out threats to internal validity associated with inequities between the way participants are treated [11]. However, this method precluded the collection of data on process indicators. Denzin and Lincoln [12] describe the RCT as separating the variables of interest from their local everyday milieus, entering them into a controlled research environment, and then trying to fit the results back into the original context. This did not account for the experiences of participants who were dealing with their disease as a ‘lived experience’. Exploration of how they connected their learning experiences to their individual worlds was considered to be essential for understanding the inter-relationships between the quantitative findings and the social factors that shaped these outcomes. To meet these needs a qualitative approach was required. In qualitative research there are a range of theoretical frameworks which provide guidelines for the research design. Livta and Eyles [13] argue that unlike quantitative research, where the RCT is taken as the gold standard, there is no contest in qualitative research as to which theory describes the social world better. It is rather a matter of which is most appropriate for the task in hand. For the study of patient education, ‘symbolic interactionism’ is the
Pre-course Clinical measures Questionnaires # INTERVENTION
Clinical measures Questionnaires Focus groups
preferred theoretical perspective as its roots lie in behavioural psychology. Blumer [14] describes three principles of interactionism: the first is that people will act towards things based upon the meaning that they have attached to them. The second is that meanings are attached to things as a result of social interactions with other people. The third is that meanings are altered and negotiated through an interpretative process used by individuals in dealing with things that they encounter. This perspective emphasises both the individualistic nature of the diabetes experience and the social nature of living with the disease. Mead [15] described how these meanings are not regarded as being intrinsic but arise through the process of social interaction, such as that encouraged in educational group sessions. The use of this perspective therefore recognised three important criteria for this trial. Firstly, that it treats patients as active participants in their care. Secondly, that it recognises the interaction between the body and the mind and acknowledges internal forces (conversations with self) as well as external forces (conversations with others) as having impact on how diabetes is perceived and how it is responded to. Finally, it acknowledges that perceptions of an illness are influenced by socialenvironmental factors. These three factors correlated with the theoretical framework identified for the study. 2.4. Phase III: trial protocol Using a dual methods approach meant that we needed to explore how the two sets of results could fit together.
48
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
Both approaches were given equal emphasis in this study but differences between them needed to be understood from the analytical and interpretative perspectives. Bryman [16] describes how such differences are often marshalled as evidence of the essential incompatibility of the two approaches. This leads into the debate about whether you can actually combine the two methods. This debate centres on differences in relation to paradigms and their associated methods. Paradigms of inquiry are world views that signal distinctive ideas about the nature of knowledge which translates into issues of scientific methodology [17]. As the roots of quantitative and qualitative research lie at either end of a spectrum of research methods, the different approaches need to remain distinct. This trial therefore essentially involved two separate studies that were linked by their evidential and theoretical base to form a scientific process. In this study, the following instruments were used to research outcomes: Structured interview tool to collect quantitative data on the characteristics of those taking part in the trial. Diabetes-specific/validated questionnaires measuring attitudes to diabetes and its treatment [18], personal models of diabetes [19], and behaviour including exercise, diet and self monitoring [20]. Physiological measures of blood glucose control (HbA1c), weight (body mass index) and changes in drug treatment. Focus group discussions using a semi-structured interview tool to explore group perceptions of the educational process and its effects in ‘real life’. Use of multiple outcome measures meant that different dimensions of patient education could be captured. It therefore expanded the scope of the study in two ways: firstly it allowed for triangulation, to ensure corroboration of data; and secondly it allowed for complementarity, to clarify, explain and more fully elaborate the results of the analyses. The aim of using this approach was to improve the quality of the research and confirm the conclusions induced from the data. In this way we hoped to understand how the educational intervention had influenced outcomes in the context of participants’ everyday lives. Creating such an account, however, relied upon exploration of both compatibilities and contradictions between the data sets. This required strong theoretical underpinnings to allow sense to be made of the combined data. This point has been clearly demonstrated in the work of a number of researchers which have been reviewed by Brannen [21] and Carey [22], although none could be located in the field of diabetes research. 2.5. Analysis of the data Analysis of both data sets followed rigorous methods. The quantitative data was processed using the statistical package for the social sciences (SPSS). Parametric data was compared using independent sample t-tests and non-parametric
data using the Mann–Whitney U-test. Regression analysis was used to compare changes in HbA1c levels, allowing for differences in participants’ baseline levels. Categorical data was compared using w2, and McNemar’s test was used to measure categorical changes within groups. Results were considered significant with the conventional P 0:05, although the ‘P’ values were considered to be an indicator of differences rather than a conclusive finding. How the quantitative and qualitative data fitted together was considered to be more indicative a test of the hypotheses. The qualitative data was examined and the meanings and relevant implications for the research questions explored. All the interviews were tape recorded (with informants’ permission) and transcribed verbatim. The constant comparative method was used to develop categories and meanings. These were then tested and modified through addition cycles of data collection, analysis and systematic search for negative cases [23]. In addition, a second researcher checked the written interpretations for confirmatory purposes.
3. Results Various publications have reported on the overall findings of this trial [4,9,24]. This paper looks at the compatibilities and contradictions between the data sets with a focus on the main aims of the study which were to measure quantitative outcomes and to explore how the intervention had influenced these outcomes in the context of participants’ everyday lives. Table 3 provides details of the characteristics of those who took part in the trial. Overall, participants in both the shortand long-term trial and control groups were comparable in relation to their demographic, medical, and social characteristics. 3.1. Clinical outcomes Significant differences for blood glucose (HbA1c) levels were found at 6 months only (P ¼ 0:005 (6 months), P ¼ 0:84 (12 months)). This reflected differences in the rates of deterioration in HbA1c, with the trial group showing less deterioration than the control group. Baseline levels of HbA1c were found to predict outcome at 12 months (P ¼ 0:01) so that those entering the trial with higher levels were significantly more likely to improve their blood glucose levels longitudinally. This demonstrated the success of the intervention at raising perceptions about glycaemic risk factors in those most at risk and this was confirmed during the focus group discussions. There was also significant correlation between attendance rates to the course and reductions in HbA1c levels at 12 months (P ¼ 0:05). This finding highlighted the importance of attending all eight of the educational sessions for maximum clinical benefit. The fact that effects on HbA1c did not persist longitudinally was mirrored by a lack of significant effect on weight (P ¼ 0:84, P ¼ 0:11), but wide variations were noted
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52 Table 3 Demographic, medical and social characteristics of trial participants (n ¼ 89) Participant characteristics Mean age (year) Male:female (%) Mean time since diagnosis (year) Employed (%) Unemployed (%) Retired (%) Mean HbA1c (%) Desirable weight (BMI) (%) Current smokers Alcohol intake Lives alone (%) Perception of effect of diabetes on lifestyle (%) Treatment Diet only (%) Diet and tablets (%)
58.6 (range 35–73) 56:44 5.7 (range 1–30) 35 35 30 7.5 (range 4.6–11) 7 21 11% above safe limits 17 57
28 72
Concurrent complications Hypertension (%) Hyperlipidaemia (%) Ischaemic heart disease (%) Retinopathy (%) Neuropathy (%)
25 26 18 11 12
Management of diabetes Shared care (hospital/GP) (%) GP care only (%)
82 18
Sources of information on diabetes Leaflets (%) Practice nurse and GP (%) Diabetes specialist nurse (%)
21 20 17
Table 4 Overview of changes in HbA1c and weight using individual patient data Blood glucose (HbA1c)
Short-term Short-term Long-term Long-term trial control trial control (6 months) (6 months) (12 months) (12 months)
Improved (%) Deteriorated (%) No change (%) Range (1)
53 17 43 43 74 51 3 9 6 3.8 to 2.6 0.2 to 3.2 5.4 to 5.6 57 53 29 40 14 6 5.6 to 1.1 4.6 to 6.4
Minus ranges denote improvement.
30%) had their drug treatment increased relative to baseline, but differences were not significant (w2, P ¼ 0:16). During the focus groups, participants spoke about using their new found competencies to self-monitor their drug treatment and increase their participation in the medical consultation. They found, however, that health professionals did not always embrace their self-perceived needs, and talked about the difficulties they had in getting them to respond in what they considered to be an appropriate fashion. They recognised a lack of specialist knowledge in primary care which would have permitted the on-going support that many participants felt they now needed. This was particularly apparent for those requiring a change from oral therapy to insulin treatment. There were 19 participants (21%) who were on combined maximum therapy at the start of the trial. Of these 19, only four were changed to insulin therapy over the trial period despite these people recognising their need for such a change. As a progressive disease, Type 2 diabetes requires continual systematic monitoring of treatment. The lack of an appropriate response by health professionals highlighted the interaction of factors that can affect the clinical outcomes associated with educational trials. 3.2. Psychological status
around the means for both these variables. Analysis using individual patient data found that the trial groups had greater numbers of people improving their HbA1c levels than the control groups. For reductions in weight, this position was more pronounced at 12 months, reinforcing the need for weight loss to be measured longitudinally (Table 4). Changes in drug treatment at 12 months showed that proportionally more patients in the trial group (46% versus
Weight (BMI) Improved (%) 61 Deteriorated (%) 36 No change (%) 4 Range (1) 3.9 to 5.0
49
19 71 10 3.8 to 2.6 43 43 13 4.2 to 1.9
Significant differences between the groups for changes in attitudes were seen at both 6 months (P ¼ 0:04) and 12 months (P ¼ 0:01), whilst changes in perceptions of selfcare treatment effectiveness were seen at 6 months only (P ¼ 0:02, P ¼ 0:23). There were no significant differences for changes in perceptions of seriousness of diabetes or personal control. For both groups, diabetes was perceived to be ‘fairly’ serious, pre-and post-trial, illustrating that they all showed some concern about the gravity of their disease and its threat to their future health. The mean and median values for self-care treatment effectiveness were higher than the values for seriousness for all participants. This showed that beliefs about the value of treatment were stronger than beliefs about the seriousness of diabetes. However, only the trial groups showed an increase in their beliefs about the value of self-care treatment, as shown by an increase in median scores at 6 and 12 months. All participants felt they had a moderate level of control over their diabetes. This perception did not change over the trial period. However, the data showed that at 6 months, 71% of the trial group and 67% of the control group continued to feel, or had moved toward feeling more in control of their diabetes. These proportions had changed to 74 and 52% respectively at 12 months. The control group’s perceptions had therefore deteriorated, whilst the trial group had improved their perceptions of personal control. This trend was reflected by changes in the P values over time: 6 months P ¼ 0:72; 12 months, P ¼ 0:06. During the focus groups participants spoke about a heightened awareness of the seriousness of their type of
50
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
diabetes and acceptance of its permanence. This was reflected by a change in their perception of Type 2 diabetes being a ‘mild’ disease. These findings contradicted those from the questionnaire. However, such inconsistency was explained by reference to the theories of personal models of illness [25]. During the focus groups it was found that participants’ perceptual changes were balanced against a reduction in feelings of helplessness, less worry about the disease, and feelings of having been provided with the ‘tools’ to control any threats. They had therefore altered their personal models of diabetes and with it, their coping patterns. Such (re)appraisal resulted in participants feeling that diabetes had become an integral part of them and not something to be dealt with outside their normal living experiences. Feeling that the disease was less of a threat did not therefore mean that they viewed it as being less serious. The outcome of this learning process varied, depending upon environmental and psycho-social constraints, which appeared to both positively and negatively affect health outcomes. This was illustrated by the quantitative data which found significant differences in perceptions of treatment effectiveness at 6 months only. The fact that such differences were not shown later, after participants had had a chance to test out their new found skills and knowledge, indicated that their beliefs had been affected by experiences in the ‘real world’. This concurred with the clinical outcomes. 3.3. Self-management behaviours There were no significant differences between the groups for changes in exercise and diet scores. The trial groups, however, showed larger increases in their mean scores than the control groups, but wide variations around the means were noted for all the groups. The number of participants self-monitoring increased over the two time periods for both the trial and control groups, although differences were only significant at 12 months (P ¼ 0:21, P ¼ 0:002). Pre-trial, 67% of participants in the long-term trial group and 47% in the long-term control group were self-monitoring. At 12 months, this proportion had changed to 92 and 63%, respectively. Changes for the trial group related to highly significant increases in the numbers blood glucose testing (McNemar’s test, P < 0:001), but not for the control group (McNemar’s test, P ¼ 0:13). During the focus groups participants described how living with diabetes often meant that compromise was required to be able to conform to the treatment regime. They discussed how they viewed what happened in ‘real life’ to be of great importance to their health outcomes and they wanted this to be acknowledged by health practitioners. They spoke of intentions to change, as well as behaviour change. This highlighted how the quantitative questionnaire data had focused upon single outcomes, i.e. changes in exercise and diet scores, rather than on the process of changing these
variables. This meant that some were contemplating, some preparing, some taking action, some maintaining change, or some relapsing [26]. This qualitative data tied in with the wide variations around the quantitative mean. The intervention therefore appeared to act as a catalyst, moving participants into a spiral of behavioural change, where support strategies were essential to help participants cope. Treatment relapse was discussed by some participants in relation to loss of support at home or from the group which reinforced the importance of having such systems in place. When participants were asked about their self-monitoring practices at 6 months, some participants described encountering discrepancies between what their doctors advised and what they now perceived to be important. At 12 months this situation had changed and participants described how the intervention had empowered them to negotiate for the equipment they required, i.e. they had developed greater self-efficacy. This linked into changes in their perceptions of personal control. Participants spoke of self-monitoring allowing them to ‘see’ and thus learn how their metabolism deals with such things as variations in meals and activities. This had motivated adherence to the behaviour.
4. Discussion and conclusions 4.1. Discussion This study aimed to explore the outcomes of patient education by measuring effects upon participants’ self-care behaviours and their illness beliefs. Alongside this it investigated how the intervention had influenced these outcomes in the context of participants’ everyday lives. The main limitation of the trial related to the number of people who agreed to take part in the study. Recruitment demonstrated that patients differed in their willingness to be educated, as reflected by the fact that only 40% of people asked to take part agreed to do so. How typical the trial participants were in terms of the overall population of patients with Type 2 diabetes could best be understood by studying those who refused to take part in the educational intervention. Apart from characteristics relating to age, ethnicity and sex ratio (for which there were no differences) this data was not collected and is a limitation to generalising the results of this study. It also implies the need for further research into the reasons why people with Type 2 diabetes choose to engage (or not) in educational initiatives. For those who did choose to take part, it can only be presumed that they were orientated toward wanting a different kind of management for their diabetes. This has implications for the results of the study because it suggests that some or all of the participants were ready to make changes. However, by using a randomised-controlled trial design, which achieves comparability by equating the average unit within each of the groups, this limitation was minimised [11].
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
The strength of the study lay in its design and the fact that there was considerable overlap between the quantitative and qualitative data sets. This meant that there was less probability of drawing erroneous conclusions and served to validate the findings from each method. Where inconsistencies existed, explanation was sought from the theories used to underpin the trial. From a purely quantitative perspective the findings show that patient education appears to have short-term clinical effects, long-term psychological effects and no effects on lifestyle behaviours. However, when looked at from a combined data perspective the conclusions drawn are quite different and clearly demonstrate the value of using a combined methods approach. Firstly, whilst there were no significant changes in diet and exercise behaviour participants showed that they had entered a behaviour change cycle. The focus group discussions highlighted the lack of an effective quantitative tool to measure these processes. From this perspective behaviours, rather than just physiological outcomes, need to be rewarded if changes are to be encouraged and maintained. This finding has implications for clinical practice and future research. Secondly, the data on self-monitoring revealed participants’ perceptions of lack of support for this practice. Coster et al. [27] conducted a meta-analysis of six RCTs which had compared blood testing, urine testing and no testing in Type 2 diabetes. There were no significant differences in HbA1c outcomes for the three methods. Such findings may well have influenced policy decisions on providing blood testing equipment for such patients. However, Coster et al. also found only anecdotal evidence concerning patients’ perceptions of monitoring, that patient training was not addressed systematically, and that no trial required subjects to modify their drug therapy in accordance with self-monitoring results. Whilst the intervention provided a standardised educational programme, participants were not provided with information about drug self modification in accordance with self-monitoring results. Rather, they were directed to seek advice from health professionals where participants reported specialist knowledge to be inconsistent. They also felt that health professionals were not always willing to value their expertise and spoke angrily about the tendency to focus on the physical aspects of their condition rather then on their emotional needs. This appeared to have acted as a barrier to effective learning in the past and has implications for professional training and future research studies. Thirdly, the findings are consistent with the psychological theories chosen to underpin the trial and illustrate how they can be integrated to form a framework for guiding intervention to enhance patient self-management. This model argues for the need to expand the traditional approach to patient education so that it allows for patients’ self-perceived needs. This highlights the active role that patients need to play so that they become partners in their health care process and calls into question the relationships between health professionals and patients. Empowerment, described by the United Kingdom government in terms of ‘patient expertise’
51
[28], involves relationships with others because it is ‘‘. . . nurtured by the effects of collaborative efforts’’ [29]. Our study has verified this statement and has highlighted the training needs of health professionals at both under- and post-graduate levels.
5. Practice implications By using a combined methods approach to research patient education this trial has explicated both its effectiveness and its shortcomings. It has provided a vehicle for bridging the gap between scientific evidence and real life practice. It has demonstrated that questionnaires, clinical tests and focus groups provide access to different types of knowledge and that they are all equally valid. What was found to be important, however, was how the different types of data ‘fit’ together. For this to be achieved strong theoretical underpinnings were required, including an understanding of the theoretical philosophies of research. This trial has shown that the use of such an approach led to a reinforcement of theory, suggested new research directions and provided guidance for clinical practice and professional education. An RCT alone would have restricted applicability of our findings because it does not address the human elements of living with a disease such as diabetes. It has therefore strengthened our conviction that a combined methods approach is required for researching complex interventions such as patient education.
Acknowledgements The authors would like to thank Diabetes United Kingdom who funded this research study. References [1] Amos AT, McCarty DJ, Zimmet P. The rising global burden of diabetes and its complications: estimated predictions to the year 2010. Diabet Med 1997;14(Suppl 5):S1–S85. [2] Turner R, Cull C, Holman RR. United Kingdom prospective diabetes study 17: a 9-year update of randomised controlled trials on the effect of improved metabolic control on complications in non-insulin controlled diabetes. Ann Int Med 1996;124:136–45. [3] Audit Commission. Testing times: a review of diabetes services in England and Wales. The Audit Commission for local authorities and the National Health Service in England and Wales, London; 2000. [4] Cooper H, Booth K, Fear S, Gill G. Lessons from chronic disease patient education. Patient Educ Couns 2001;44:107–17. [5] Griffin S, Kinmouth AL, Skinner C, Kelly J. Educational and psychosocial interventions for adults with diabetes article (online), March 1999. Summary of report available on: http://www.diabetes.org.uk. [6] Roter DL, Hall JA, Merisca R, Nordstrom B, Cretin D, Svarstad B. Effectiveness of interventions to improve patient compliance: a meta-analysis. Med Care 1998;36:1138–61. [7] Skinner TC, Cradock S. Empowerment: what about the evidence? Pract Diab Int 2000;17:91–5.
52
H. Cooper et al. / Patient Education and Counseling 51 (2003) 45–52
[8] Campbell M, Fitzpatrick R, Haines A, Kinmonth AL, et al. Framework for design and evaluation of complex interventions to improve health. Br Med J 2000;321:694–6. [9] Cooper H, Booth K, Fear S, Gill G. Patients’ perspectives on diabetes health care education. Health Educ Res 2003;18:101–16. [10] Bradley C. Designing medical and educational intervention studies. Diabetes Care 1993;16:509–18. [11] Cook DC, Campbell DT. Quasi-experimentation: design and analysis issues for field settings. Boston MA, USA: Houghton Miffllin; 1979. [12] Denzin NK, Lincoln YS. Handbook of qualitative research. Beverley Hills, CA: Sage; 1994. [13] Livta A, Eyles J. Coming out: exposing social theory in medical geography. Health Place 1995;1:5–14. [14] Blumer H. Symbolic interactionism: perspective and method. Englewood Cliffs: Prentice-Hall; 1969. [15] Mead GH. Mind, self and society. Chicago: University of Chicago Press; 1962. [16] Bryman A. Quantity and quality in social research. London: Routledge; 1988. [17] Sandelowski M. Combining qualitative and quantitative sampling, data collection, and analysis techniques in mixed-methods studies. Res Nurs Health 2000;23:246–55. [18] Welch G, Beeney L, Dunn S, Smith RBW. The development of the diabetes integration scale: a psychometric study of the ATT 39. Multiva Exp Clin R 1994;11:75–88. [19] Glasgow RE, Toobert DJ, Hampson S, Noell JW. A brief office based intervention to facilitate diabetes self-management. Health Educ Res 1995;10:467–78.
[20] Toobert DJ, Glasgow RE. Assessing diabetes self-management: the summary of diabetes self-care activities questionnaire. In: Bradley C, editor. Handbook of psychology and diabetes, vol. 15. Switzerland: Harwood; 1994. p. 351–75. [21] Brannen J. Mixing methods: qualitative and quantitative research. Aldershot: Avebury; 1992. [22] Carey JW. Linking qualitative and quantitative methods: integrating cultural factors into public health. Qual Health Res 1993;3:298–318. [23] Denzin NK. The art and politics of interpretation. In: Denzin NK, Lincoln YS, editors. Handbook of qualitative research. Beverley Hills, CA: Sage; 1994. [24] Cooper H, Booth K, Gill G. Diabetes education: the patient’s perspective. J Diabetes Nurse 2002;6:91–5. [25] Leventhal H, Nerenz DR, Steele DJ. Illness representations and coping with health threats. In: Baum A, Taylor SE, Singer JE, editors. Handbook of psychology and health. Hillside NJ: Erlbaum; 1984. [26] Prochaska JO, DiClemente CC, Norcross JC. In search of how people change: applications to addictive behaviours. Am Psychol 1992;47: 1102–14. [27] Coster S, Gulliford MC, Seed PT, Powrie JK, Swaminathan R. Monitoring blood glucose control in diabetes mellitus: a systematic review. Health Technol Assess 2000;4. [28] Department of Health. The expert patient: a new approach to chronic disease management for the 21st century. London: HMSO (Books); 2001. [29] Keiffer C. Citizen empowerment: a developmental perspective. Prev Hum Serv 1984;3:9–36.