 |
Aart N. Mudde, School of Psychology, Open University of The Netherlands, Heerlen, The Netherlands
Hein de Vries, Department of Health Promotion and Health Education, University of Maastricht, Maastricht, The Netherlands
Design/methodology/approach – Written questionnaires based on the I-Change Model, a socio-cognitive model, were left in general practices where the expert system had been trialled. Respondents intending to continue their use (intenders, n=62) were compared to those who did not (non-intenders, n=27).
Findings – Eighty-nine individuals from 55 practices responded (73 per cent). GPs were more often intenders than general practice assistants. Responses from the same practice were not significantly related to each other. Intention to continue using the expert system was determined by a more positive attitude towards the expert system, a social norm towards engaging in smoking cessation activities, and higher self-efficacy. Practice staff who had actively offered the expert system to their patients were more likely to be an intender.
Originality/value – Cognitive factors and trial involvement determined intention to further implement the expert system. Discussing barriers with practice staff could increase motivation to implement and ownership. Intenders can aid the implementation process by sharing experiences with non-intending peers.
Introduction
Advice from general practitioners (GPs) has a significant positive effect on smoking cessation rates, even if this advice is brief (Lancaster et al., 2000). GPs recognise the importance of smoking cessation activities and indicate that these activities at least partly belong in the general practice (McAvoy et al., 1999; McEwen and West, 2001). Nevertheless, GPs report barriers such as time constraints and concern about damaging the doctor-patient relationship with health education activities related to smoking cessation (Hoving et al., 2006; Yarnall et al., 2003). They prefer to discuss cessation only with patients with smoking-related problems or patients already motivated to quit smoking (Coleman et al., 2000; Coleman and Wilson, 1999; Helgason and Lund, 2002; Senore et al., 1999).
These implementation barriers suggest that an innovation with a self-help nature and a minimum of effort from the GP will be of practical use in the general practice and is therefore most likely to be implemented on a larger scale. Computer-generated tailoring is an expert system that meets these recommendations. It offers general practices an opportunity to provide their smoking patients with an individualised cessation method without a specific role for practice staff. Dijkstra et al. (1998) tested the impact of an expert system in a randomised control trial. This expert system, based on the I-Change Model, was developed by the University of Maastricht in cooperation with the Dutch Foundation on Smoking and Health (Stivoro, 2006). Results indicated that the expert system had a significant positive effect on smoking cessation rates in a sample of self-selected subjects. Smoking patients who use the expert system filled in a questionnaire containing 54 personal questions on smoking and smoking cessation. This questionnaire acted as a screening instrument. Smokers sent their filled-in questionnaire by mail to the national processing centre. There, answers from the questionnaire were entered into the expert system. The expert system consisted of a feedback library with messages related to the topics in the questionnaire. Only messages relevant for the individual smoker appeared in their feedback, a written letter consisting of three to five pages. This feedback was then sent to their home address.
In order for the expert system to reach smoking patients, it has to be implemented by GPs and their staff. As this study was partly focused on cognitive factors of intention to implement an innovation, a model in which these factors are more prominently present was used as the theoretical framework for this study: the I-Change Model (De Vries et al., 2003). This model incorporates several cognitive models, such as the Transtheoretical Model (Prochaska et al., 1997) and the Theory of Reasoned Action (Fishbein and Ajzen, 1975). The I-Change model states that the intention to change behaviour is determined by predisposing factors (behavioural, psychological, biological, social cultural factors), awareness factors (knowledge, cues to action, risk perception), information factors (message, channel, source) and motivational factors (attitude, social norm, self-efficacy). The intention to change in combination with an individual's abilities and experienced barriers influence the likelihood of accomplishing the desired behaviour change. Its predecessor, the ASE Model (De Vries et al., 1988) has been used previously in studies concerned with the implementation of health education programmes (Paulussen et al., 1995).
The I-Change Model recognises that offering an opportunity to try out the expert system (trialability) could encourage the transition from adoption (planning to start using an intervention) to implementation (embedding an intervention into daily practice). This research aimed to study factors related to this transition among Dutch general practices after a trialability opportunity.
Methods
Respondents and recruitment
The general practices eligible for inclusion in this study had been participating in an effectiveness study concerning the expert system described above. For this effectiveness study, 512 general practices were asked for permission to distribute the expert system questionnaires. The questionnaires with return envelopes were displayed in a separate folder stand in the waiting room. Pens with the project logo to fill in the questionnaires were provided, and two A4-size posters were displayed to draw attention. All practices were visited monthly to replenish project material when necessary. A contact person within the practice monitored the questionnaires taken from the stand. Practice assistants were preferred over GPs due to logistical reasons (better availability and closer ties to the waiting room). Neither GP nor practice assistant had to carry out any additional tasks. The effectiveness study served as a trialability opportunity to become familiar with the expert system. A total of 75 general practices agreed to participate in this study.
After eight months, the effectiveness study ended and two process-evaluation questionnaires were left behind in each practice. One general practitioner and one practice assistant from each general practice were invited to fill in the questionnaire. Responses were anonymous, but general practice of origin was recorded. After two weeks, a written reminder was sent by mail to all non-responders.
Questionnaire
The three-page questionnaire consisted of 29 questions concerning demographics, motivational determinants on both smoking cessation activities in general and the expert system specifically, previous smoking cessation education activities, the degree of implementation of the expert system during the effectiveness study and future action plans concerning the use of the expert system.
Seven questions measured demographic variables of the respondent: occupation (general practitioner or practice assistant), practice type (solo – one GP with assistance; duo practice – two GPs with assistance; group practice – more than two GPs with assistance), practice location (rural, semi-urban or urban), smoking status (smoker, non-smoker or ex-smoker) and the percentage of registered patients who smoke (respondents could also tick a box if this percentage was not known).
The main cognitive variables (attitude, social influence/social norm and self-efficacy) were scored on a five-point scale (from −2 for “disagree” to 2 for “agree”), as were previous cessation education activities, the degree of implementation during the expert system trial and future implementation plans.
Two items assessed attitudes concerning smoking cessation education in general (“I think that smoking cessation education is important”, “I think that smoking cessation education has a place in the general practice”, α=0.55), whereas three items assessed attitudes towards the expert system (“I think the expert system is an effective smoking cessation method”, “the expert system is a good addition to the smoking cessation materials we already had in use”, “I think the expert system is very usable in the general practice by general practitioners and practice assistants alike”, α=0.82).
Social norm concerning smoking cessation education in general was assessed by four items (“My colleagues inside/outside the practice think that smoking cessation education is important”, “My colleagues inside/outside the practice think that smoking cessation education has a place in general practice”, α=0.84).
Self-efficacy concerning smoking cessation activities in general was assessed with two items (“I do not find it difficult to address smoking and smoking cessation”, “I have influence on the smoking behaviour of my patients“, α=0.18), whereas self-efficacy concerning using the expert system was assessed with four items (“I will succeed to use the expert system on smoking patients who are asking for a cessation advice”, “I will succeed to use the expert system on smoking patients who are not asking for a cessation advice”, “I will succeed in supplementing the expert system questionnaires in the practice so they can always be offered”, “I will succeed in offering the expert system even when there are other smoking cessation methods available”, α=0.81).
The questionnaire assessed the degree of implementation of the expert system trial with six items (e.g. “During the project we kept two posters in the waiting room at all times”, “All employees were informed about the project”, α=0.55).
Intention to keep using the expert system in the future was assessed by six items (“I plan to order the expert system”, “I plan to discuss further use of the expert system within our practice”, “I plan to keep offering the expert system in our waiting room”, “I plan to actively offer the expert system to our smoking clients”, “I plan to stimulate colleagues within our practice and outside to use the expert system”, α=0.85).
Analysis
Data was analysed using SPPS 11.0. Respondents were assigned to one of two groups based on their intention to use the expert system in the future: intenders and non-intenders. This was determined by computing a mean score of the six items concerning intention to keep using the expert system in the future; respondents with a score above zero were considered intenders, respondents with a score of zero or below were considered non-intenders.
Kolmogorov-Smirnov tests showed that all interval variables to be included in the analyses were not normally divided, indicating the need for non-parametric testing. Therefore, attrition analyses were conducted by means of in order Mann-Whitney U tests and χ2 tests to determine a selection bias between those who had responded and those who had not. Furthermore, differences between non-intenders and intenders were determined to reveal relations between future action plans concerning the expert system and demographic variables, main cognitive variables and the degree of implementation during the trial. To find out whether a GP's intention to further implement the expert system was related to the intention of the assistant from the same practice, a χ2 test was conducted. In order to minimise the number of separate tests on such a limited sample, mean scores were calculated within concepts, based on the reliability scores shown in Table I. Mean scores with a reliability score larger than 0.7 were used (as suggested by Nunnally, 1978). In case mean scores had a lower reliability score, the individual item scores were entered into the analysis. Items and concepts were then introduced into a backward logistic regression analysis in three blocks with intention to keep using the expert system in the future as dependent variable; the first block contained four demographic variables (job function within the general practice, practice type, location of the practice and smoking status of the respondent), the second block contained seven items and mean scores concerning the main cognitive variables (attitudes concerning smoking cessation education in general and towards the expert system, social norm concerning smoking cessation education in general and self-efficacy concerning smoking cessation activities in general and using the expert system) and the third block contained six items concerning the degree of implementation of the expert system during the trial. Variables were excluded on basis of the change in the model's likelihood ratio.
Results
Sample
A total of 95 respondents (63 per cent) returned the questionnaire, representing 73 per cent of participating practices. Six questionnaires contained too many missing values (>10 per cent) and were excluded from further analysis. Attrition analyses showed no significant difference concerning function within the practice (general practitioner or practice assistant) or practice type (solo, duo or group practice) between respondents and those who did not respond.
In the final sample, 48 respondents functioned as assistants (53.9 per cent) and 41 as general practitioners (46.1 per cent). A total of 46.1 per cent (n=41) worked in a solo practice, 37.1 per cent (n=33) in a duo practice and 16.9 per cent (n=15) in a group practice. Respondents represented 27 solo practices (50.0 per cent), 19 duo practices (35.2 per cent) and eight group practices (14.8 per cent). From 33 practices, both GP and assistant responded (59 per cent of the 52 participating practices). Most respondents (n=70, 78.7 per cent) did not know the percentage of smokers among patients registered with their general practice, but from those respondents (n=19) who did give an answer, the mean percentage of registered smokers was 14.5 per cent. Most respondents worked in general practices in an urban (n=35, 39.3 per cent) or semi-urban (n=33, 37.1 per cent) environment, 21 respondents (23.6 per cent) worked in a general practice situated in a rural area. A majority of the respondents had never smoked (n=56, 62.9 per cent), ten respondents were currently a smoker (11.2 per cent) and 23 respondents were ex-smokers (25.8 per cent). Many respondents indicated that they intended to further implement the expert system in their general practice (n=62, 69.7 per cent); 80.5 per cent of the GPs and 60.9 per cent of the practice assistants could be classified as intenders.
Differences between intenders and non-intenders
The group of intenders included significantly more GPs than the non-intender group (χ2=4.22, p<0.05). GPs' intentions were not significantly related with those of assistants in the same practices (χ2=0.75, p=0.39). Table I shows the differences between intenders and non-intenders concerning attitudes, social norm, self-efficacy and degree of implementation. Intenders considered smoking cessation activities in the general practice more important and were more likely to think that smoking cessation activities had a place in the general practice. They also held a more positive attitude towards using the expert system. Intenders also reported a more positive social norm concerning the system and were more confident about using the expert system.
Table II shows the results of the logistic regression analysis. Logistic regression analysis showed that practice assistants were less likely to implement the expert system in their practice than GPs, when correcting for general practice characteristics. After including attitude, social pressure and self-efficacy, the influence of occupation became even more apparent. A higher self-efficacy score concerning using the expert system and experiencing a positive social norm also contributed to the decision to further implement the expert system. Also, respondents were more prone to further implement the study if they had actively offered the expert system to smokers during the trial. The model explained 52 per cent of the variance.
Discussion
GPs had a more positive intention towards implementing the expert system than practice assistants; eight GPs versus 19 practice assistants were not inclined to further implement the expert system. Lock et al. (2000) also found that GPs involved in an alcohol screening programme had more positive attitudes towards the programme than the primary health care receptionists involved (Lock et al., 2000). This could be due to the fact that contact persons in most participating practices were practice assistants. The delegation of this task to the practice assistant might be involuntary, thus influencing the practice assistants' attitude towards the expert system. Also, as the recruitment for participating practices in the trial was focused on GPs, practice assistants may have felt overlooked. In combination with a perceived, possible involuntary, increase in workload this might have caused a less positive attitude towards the expert system and therefore less motivation to continue using the expert system after the trial in the group of practice assistants, compared to the group of GPs.
A more positive social norm towards smoking cessation activities in general practice and stronger self-efficacy expectations to use the expert system appear to positively influence the intention to use the expert system in the future as well. These findings are in line with the outline of the I-Change Model, which states that attitude, social influence and self-efficacy have an influence on the intention to change or maintain certain behaviour, in this case, the continuation of using the expert system. The high percentage concerning explained variance of the model supports this hypothesis. Practices that actively pointed out the possibility of using the expert system to their smoking clients were more likely to intend to keep using the expert system. It might be that a somewhat active role of practice staff increases involvement with the expert system.
This study is subject to some limitations. First, although the sample described in this paper is representative for the population of 75 general practices that were involved in a trial concerning the expert system, all practices had already agreed to adopt the expert system for the period of the trial and practice staff might therefore have viewed the expert system more positive than the average Dutch general practice would have. However, it would be more likely that the differences between intenders and non-intenders would increase or multiply rather than the other way around. The characteristics of non-intenders we found in this study are also likely to be applicable for general practices that did not even want to adopt the expert system for as long as the trial period; all the more reason to take notice of these characteristics. Second, it is possible that staff members were influenced by each other. In 33 general practices, opinions of GPs as well as assistant could be assessed. However, in those practices, no significant correspondence between responses of GPs and assistants were found. Third, this study focused on the intention to further implement the expert system, rather than the action of implementation itself. Even though the theoretical model we used for this study assumes that behaviour follows from the intention to perform this behaviour, it is also recognised that barriers can prevent intention to develop into action. However, the expert system was set up to diminish at least practical barriers, such as time restraints and GP's reluctance of addressing all smoking patients as much as possible.
Conclusion
This study has identified a number of determinants important in the decision whether or not to implement a smoking cessation expert system by general practice staff.
It seems that a somewhat active role of general practice staff positively influences the intention to continue using the expert system in the future. Therefore, bypassing the practice staff entirely does not seem prudent, especially since research has shown that the GP's health advice increases smoking cessation rates. However, these investments in time and effort might have their limits. Further research could determine investments by general practice staff needed to create some degree of ownership, but does not exceed the extent of investment staff is willing to make. Also, the specific role of the GP in the effectiveness of general practice-based smoking cessation advice should be studied, in order to determine whether the delegation of tasks to other staff members within the general practice might influence the positive effect of an advice concerning smoking cessation through general practice (Lancaster et al., 2000).
General practice staff willing to continue using the expert system can be characterised as having a more positive attitude towards the expert system, as well as a higher confidence in being able to use the expert system in the future. When we assume that the expert system is an innovation worth using by GPs, it is desirable for the expert system to be implemented by the entire population of Dutch GPs. Rogers indicates that social influence through members who have already implemented an innovation is important for the continuation of the process by the entire population (Rogers, 1995). This is in line with the finding in our study that intenders perceive a social norm more towards engaging in smoking cessation activities than non-intenders.
Overburdening staff could mean that the expert system will be rejected before it is even seriously considered for implementation. If discussing these and other barriers during the development or adjustment of smoking cessation methods with the persons who will actually be responsible for implementing these methods, motivation to work with the method can be increased and the base for ownership can be created.
Although there is a difference between intention to perform certain behaviour and the actual behaviour, if intenders continue to implement the expert system, they are in the position to motivate and stimulate other general practices to start using the expert system. If the trialability of the expert system is increased, a more positive attitude towards the expert system can be established, as well as an increased confidence to eventually implement the expert system into daily practice. By inviting their peers who have not yet implemented the expert system for a demonstration or by sharing their experiences with the expert system, implementers might be able to increase non-implementers' interest and self-efficacy expectations and provide a forum to address misconceptions.
 |
 |
| References |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Ciska Hoving can be contacted at: c.hoving@gvo.unimaas.nl |