Can an EASYcare based dementia training programme improve diagnostic assessment and management of dementia by general practitioners and primary care nurses? The design of a randomised controlled trial

Dementia is an increasing challenge for health care and social systems in developed countries. In Europe, 6.4% of the elderly over 65 years suffer from dementia, with an increase from 0.8% in the group aged 65–69 years to 28.5% at age 90 years and older. The total number of dementia patients in Europe is expected to increase from 7 million in 2000 to over 16 million in 2050 [1]. Currently, dementia seems to be under-diagnosed [2‐4]. More than 50% of dementia patients living in the community have not been diagnosed by a GP or specialist [5, 6]. In the Netherlands, detection of dementia takes place around one year before admission to a nursing home and three years before death [7], which is rather late.

Early diagnostic evaluation of patients possibly suffering from dementia is beneficial for both patient and caregiver. Reversible causes of dementia will be identified and treated timely. Formal disclosure of dementia diagnosis allows patients and carers to make future plans and provides early access to support services [8]. These actions may prevent or decrease psychological distress in patients [9]. Early education and support for caregivers facilitates adjustment over a longer period of time; it prevents crisis situations and delays nursing home admissions [10, 11]. Non-pharmacological interventions, such as psychosocial interventions (10;11) and occupational therapy [12], have been shown to play a key role in dementia management. Pharmacological treatment may primarily be beneficial when started in the early stages of dementia. However, drugs that are currently available have only moderate effect on cognitive symptoms and ADL [13]. In the near future, disease-modifying drugs might become available since they are already subject to phase two and three trials.

Despite the acknowledged benefits of early diagnosis of dementia, both patients and doctors are reluctant to initiate cognitive assessment. Patient-related delay in early recognition of dementia is often caused by lack of insight into their condition, their view of memory loss symptoms as being normal for their age and fear of the negative consequences of dementia diagnosis [4]. GPs report that their own lack of knowledge and skills in diagnosing and treating dementia prevents them from starting diagnostic work-up in the early stages of dementia [14]. Other GP-related barriers include the absence of clear diagnostic guidelines and reliable, user-friendly screening tools, lack of time, of financial reward, of adequate resources such as access to neuropsychological consultations and neuro-imaging investigations, and lack of prescription right for cholinesterase inhibitors [15]. Thus, many GPs are sceptical about the benefits of early diagnosis, because they feel they have little to offer dementia patients and their caregivers [14, 16, 17]. Moreover, disclosure of diagnosis is considered to be difficult, because it may negatively influence patient-doctor relationships and take away patients' and caregivers' hopes [18, 19]. GPs also fear the risk of diagnostic errors. Evidence of under-recognition of dementia has also been shown in primary care nurses for similar reasons [20]. Also, collaboration between primary care nurses and GPs is not very well developed, although dementia diagnosis and management may profit from it. This lack of collaboration is caused by conflicting expectations, domain discussions and poor coordination of care [21].

To improve early detection in primary care, guidelines on dementia diagnosis and management were developed in several countries [22‐25]. A multifaceted implementation program in Denmark did not show effects on adherence to practice guidelines [26]. Research teams in the UK and USA also developed educational programmes for implementation of their national guidelines. These programmes were successful in raising the rate of early dementia diagnosis and improving professionals' knowledge [27, 28]. However, they showed only minor improvement in attitudes of health care providers regarding dementia diagnosis and regarding management of dementia in primary care [29, 30]. Implementation strategies in these programs included the use of small group sessions, a Computerized Clinical Decision Support system (CDSS) [27], and internet support [28, 29].

The educational programs in the studies discussed above primarily focused on GPs. However, dementia care management performed by collaborative interdisciplinary teams was found to be more effective in improving the quality of care than that performed either by GPs or primary care nurses [31]. Therefore, we designed a Dementia Training Program (DTP) for collaborative duos of GPs and primary care nurses, focused on teaching them how to share tasks in early diagnosis and management of dementia. Another unique component of our DTP is individual coaching, which has shown promising results in modifying behaviour of health professionals [32]. Furthermore, our Dementia Training Program is a multifaceted program [33] and consists, in addition to individual coaching, of two small-group interactive workshops [34, 35] and a Computerized Clinical Decision Support System.

The objective of this paper is to describe the design of a randomised controlled intervention study, aimed at determining the effectiveness of a multifaceted Dementia Training Program (DTP) for general practitioners and primary care nurses, based on current national guidelines. We expect the DTP to improve professional performance in dementia diagnostics and disease management and GPs' and nurses' attitudes and knowledge regarding dementia.

This study is an outcome assessor-blinded randomised controlled trial. A cluster-randomised design will be used to compare duos of GPs and nurses (Figure 1).

We plan to recruit 100 duos of GPs and practice or district nurses in the province of Gelderland, the Netherlands. We will approach all general practitioners in this province by mail and ask them to participate in the study. Participating GPs may choose to cooperate with their own practice nurse or with a district nurse. All GPs and affiliated nurses within the county are eligible for this study.

Frail elderly people, suspected of suffering from cognitive problems are the target group to be diagnosed and treated according to Dutch dementia guidelines. [22, 25]. From this target group additional data on satisfaction and informal carer burden will be obtained. Informed consent will be obtained from patients and from their legal representatives. The Local Medical Ethical Committee, Commissie Mensgebonden Onderzoek Regio Arnhem – Nijmegen, concluded that this study did not include experiments with patients and therefore did not need to be tested for approval.

Randomisation will take place after measurement of baseline data. Duos of GPs and nurses will be randomly allocated to one of the two conditions: 1. Dementia Training Program (DTP) and 2. no training at all; control group. (Figure 1). Randomization will be concealed; a person who is not responsible for recruiting subjects and has no knowledge of the study conduct will perform it on a computer.

Cluster-randomisation will be performed in order to avoid contamination by the effects of possible exchange of information within a cluster. A cluster was defined as all GPs working in the same practice or as all GPs working together with the same nurse.

In order to assure an equal distribution of baseline characteristics, duos will be randomised with adaptive weights regarding cluster size (one vs. more than one GP), age, sex, high or low percentage of elderly patients in practice (< or > 15%), practice location (rural or urban area) and nurse affiliation (district or practice).

Adherence to guidelines will be assessed by two independent researchers. In order to exclude the possibility of detection bias, these researchers will be blinded to the outcome of randomization.

DTP consists of two workshops and an individual coaching programme, including case-based consultation either by phone or by e-mail. Participants have access to an internet forum for discussion with colleagues, additional literature and individual training on dementia diagnosis and management. A Computerised Clinical Decision Support System on dementia diagnostics and management will be available to support GPs in decision making in daily practice. The content of the DTP and the CDSS is based on the recently published, evidence-based Dementia Guideline for Primary Care [36]. The DTP is graphically presented in Figure 2[37].

Methods used in the DTP to stimulate collaboration between GPs and nurses are the following:

Duos in the control group will not receive the training during the trial. They will, however, be given the opportunity to attend to the DTP after the trial. This means that duos in the control group will enter the DTP with a 9 months' delay.

Primary outcome measures are the numbers of cognitive assessments and dementia diagnoses over a period of nine months, which will be measured at baseline (T0) and nine months later, at the end of the study (T1). Secondary outcomes are GPs' and nurses' attitude, confidence and knowledge regarding dementia diagnosis and management, which will be measured at baseline (T0) and nine months later, at the end of the study (T1). Other secondary outcomes are the rate of adherence to national guidelines for dementia diagnosis and management, patient satisfaction, informal carer burden and satisfaction, number of emergency calls, visits and consultations. Baseline demographic characteristics collected from GPs and nurses are: age, sex, practice experience, practice size, percentage of elderly people > 65 years and the availability of chronic disease management programs in practice. Baseline demographic characteristics collected from patients and their informal caregivers are: age, sex and co-morbidity (Table 1).

Data on adherence to national guidelines, the number of emergency situations and the number of cognitive assessments and dementia diagnoses will be retrieved from GPs' Electronic Medical Records (EMD), from interviews with GPs and nurses and from self registration forms [41]. Self-made questionnaires will provide information on baseline characteristics, on data on GPs' and nurses' attitudes, competencies and knowledge regarding dementia diagnosis and management and on patient and informal carer satisfaction. Burden of informal carers will be assessed using the Short Sense of Competence Questionnaire (SSCQ) [42]. Data on duos' performance during the 'coaching phase' will be included in the analysis.

We expected a change in the primary outcome measure of the incidence of dementia diagnoses and cognitive assessments from 50% to 65% of the total population of dementia. In previous studies on implementation of several other guidelines in general practice the average change was 10% [43]. However, these implementation studies focused on diseases, to which GPs already showed a high adherence to the guidelines. Because of the low adherence to dementia guidelines we saw more room for improvement and therefore we expected a change of 15% to be realistic. Cluster randomisation was taken into account when we calculated the sample size needed. In this calculation, the ratio between the number of duos in the control group and the number of duos in the intervention group was 1:2. Clusters are expected to include 5 patients on average. Intra class correlation (ICC) is expected to be 0.05 or lower. For a power of 0.80, and two-sided testing at 0.05, a total of 91 general practitioners is required.

In addition to a confirmatory analysis using the MANOVA-method, we will also perform an exploratory analysis. The aim of the exploratory analysis is to gain more insight into possible causal relations between the rate of success of the intervention components and the outcomes. Possible mediator variables are change in knowledge, attitudes and collaboration rate. We will design a path-model, specifying hypothesized relations between predictors, mediators, moderators, and effects, and we will test its goodness of fit with the data. In this way we hope to be able to specify the most plausible interdependence pattern between the variables used in the present study.