Primary dementia care based on the individual needs of the patient: study protocol of the cluster randomized controlled trial, DemStepCare

Some of the problems can be ascribed to general practitioners (GPs), who should often be the first to detect the onset of dementia and deal with it adequately. It has been shown that GPs are often reluctant to face the diagnosis of dementia and do not follow the existing guidelines [5‐9]. GPs do not want to “scare off” their patients and they feel that they have little chance to do anything about the disease [10]. Affective factors and experienced self-efficacy are of great importance here, not just the GPs’ previous knowledge and competence [11, 12]. As a result, dementia diagnoses are often not made [13, 14] or only when home care is already at risk. From a therapeutic perspective, however, it has been shown that early diagnosis of dementia is crucial to positively influence the course of the disease [15‐17]. A major goal in improving primary dementia care is therefore the timely and accurate diagnosis of dementia.

However, not only the diagnostic process needs to be improved. It has been shown that GPs are often overburdened or succumb to a number of misconceptions when it comes to the treatment of dementia [18, 19]. In many cases, the existing guidelines [20, 17] are not followed [21]. For example, there are often errors in the prescription of medicines [22‐25]. One difficulty here is that PwD are at higher risk of multimorbidity and polymedication due to their usually older age [26]. Dementia-specific training courses for GPs might be beneficial here [27, 28].

Other problems of primary dementia care are structural in nature. The complexity of the clinical picture necessitates multi-professional and multimodal care [29, 30]. Needs vary greatly between individuals and over the course of the disease [31], making individualized and stepped interventions necessary. Integrated, cross-sectoral approaches to effective outpatient dementia care are needed [32]. GPs often lack knowledge about existing regional services to support PwD and their caregivers [33]. Studies on case management in dementia care have not yet revealed consistent results [34], but case management seems to be a promising factor that might facilitate the needs-based [35], stepped activation of different healthcare professionals and thus contribute to improving the care of PwD [36].

Furthermore, the different healthcare professionals involved must communicate effectively with each other to ensure continuity of care. There is a lack of efficient (digital) communication structures, especially in rural areas [37]. User-friendly solutions are needed here.

Family caregivers are exposed to increased physical and psychological stress [38‐41]. This in turn can affect the caregivers’ health [41, 38]. If any additional crises occur, this can lead to hospitalization of the PwD they care for [42]. However, it has been shown that hospitalization of PwD tends to have negative consequences for their health [43] as well as for the caregivers’ emotional stability [44]. In addition, hospital stays are expensive [45] and burden the healthcare system. It is therefore an important objective to avoid unnecessary hospitalization [46], while providing effective support for caregivers.

The first part of the intervention involves training of the GPs: GPs are to identify dementias earlier and become more confident in dealing with the diagnosis. The training courses impart knowledge about the appropriate diagnosis and treatment of PwD following the guidelines of the German Society for Psychiatry and Psychotherapy, Psychosomatics and Neurology (Deutsche Gesellschaft für Psychiatrie und Psychotherapie, Psychosomatik und Nervenheilkunde, DGPPN) [17].

The so trained GPs enroll patients and if possible their caregivers in the study program. Patients with a pre-existing dementia diagnosis as well as patients who receive a new dementia diagnosis while DemStepCare is running can be included in the study. If there is a suspected dementia, the GPs can either diagnose dementia themselves or refer the patients to a specialist / memory clinic. Patients with a confirmed dementia diagnosis and if possible their caregivers are informed about the study and asked if they are willing to participate. If patients and their caregivers wish to participate, they must give their written consent. If there is a legal guardian, the legal guardian must give written consent in addition to the patient. All patients require a basic competency to consent, otherwise they cannot participate. Having given their written consent, patients are assigned to a case manager who conducts a first counselling interview, including information on regional support options (e.g. a supervised relatives’ group that is part of DemStepCare, counselling centers) and advice on care. At this first encounter, the case manager assesses the individual risk constellation of the home care situation using a dementia specific screening tool (in German: Demenzspezifisches Screening zur Versorgungssituation, DSV [47]). The DSV addresses both aspects concerning the patient and aspects concerning the caregiver. When evaluating the DSV, there are two possible results: green (stable situation) or yellow (increased risk). In the case of an increased risk, two further tests are to follow, the CMAI, which measures agitated behavior [48, 49], and the NOSGER, which covers clinically relevant behavior of psychiatric disorders of old age [50, 51]. Based on the results of these tests, it is decided whether the case can be assessed as yellow or whether it is a red case (acute crisis).

Depending on the respective category, the case manager proceeds differently: In green cases, the case manager offers information on regional support options and sets a next appointment for six months later. In yellow cases, the case manager provides more intense care contacting the patient at least once a month. In red cases, the case manager calls in the outpatient clinic that consists of specialized nursing staff, a psychiatrist, and a social worker. The outpatient clinic continues to provide care until the crisis is overcome. To this end, the outpatient clinic offers home visits of the appropriate frequency to avoid hospitalization. If it is mandatory, an inpatient stay in a geriatric psychiatric clinic is initiated.

Throughout the course of DemStepCare, this risk stratification is regularly repeated (at least at t0, t1, and t2, and invididually if needed) and appropriate action is taken to ensure that each patient receives the care he or she needs. Figure 2 shows an overview of the described procedure.

In addition, clinical pharmacists regularly conduct structured medication reviews with the aim of improving the quality and safety of the drug therapy. The attending GPs are alerted to inappropriate medication and drug-related problems (DRP).

Communication between the different parties (i.e. GPs, case managers, outpatient clinic, pharmacy) takes place via a specifically adapted electronic case record. Thus, timely and effective communication is facilitated.

While participating in DemStepCare, patients continue to be cared for by their GPs, who can repeatedly take part in dementia-specific training sessions as the project progresses.

GPs assigned to the control group 1 receive a shorter form of training. A case manager is employed and risk stratification is conducted, too, but the result has no influence on the further treatment of the patients, i.e. it only serves evaluative purposes. Neither the GPs nor the patients get to know the result of the stratification. There is a follow-up meeting with the case manager after nine months.

As there are small elements of intervention in control condition 1, control condition 1 cannot be considered as being merely usual care. However, the training of the GPs is intended as an incentive and the visits of the case managers are necessary to gather data for the evaluation. We therefore decided to use this slightly expanded usual care condition as control condition 1. Control group 2, the historical control condition, is inherently usual care.

Training of the GPs in intervention as well as in control condition 1 is carried out by experienced healthcare professionals. The case managers also receive training resulting in a certification as case manager according to the German Society for Care and Case Management (in German: Deutsche Gesellschaft für Care und Case Management, DGCC).

The three primary endpoints are assessed via questionnaires. An overview is shown in Table 1.

To measure the patients’ quality of life, we use the Quality of Life in Alzheimer’s Disease (QOL-AD), a validated scale developed specifically for PwD [60]. The QOL-AD encompasses 13 areas of life that need to be evaluated (e.g. family, living conditions, energy). There is a version for patients and a proxy version. In DemStepCare, we use both versions, with the proxy version being filled in by the caregivers. Both scores are integrated into a total score. There is one baseline measurement (t0) and one measurement at t1.

To measure the caregiver burden, we use the short form of the Berlin Inventory of Caregivers’ Burden with Dementia Patients (in German: Berliner Inventar zur Angehörigenbelastung – Demenz; BIZA-D) which was specifically designed and validated to measure the burden resulting from caring for a person with dementia [61]. The BIZA-D is a comprehensive instrument that, among other things, queries various care tasks and associated burdens, the social support experienced, and acceptance of the situation. Caregivers fill in the BIZA-D at baseline (t0) and at t1.

To measure the days spent in hospital, we use a self-developed item at t0 and t1. We ask for days spent in hospital in the last nine months. We use one version for patients and one proxy version for caregivers. In addition, we will receive individual level data from the participating health insurance companies for the inpatient treatment days. Since only a part of the participants will presumably be insured with these health insurance companies, we will use this data for plausibility checks (comparison self-report vs. claim data). The main data basis for the primary endpoint will be provided by the questionnaires.

Within the multi-professional concept, clinical pharmacists conduct structured medication reviews at t0, t1 and t2. In the intervention group, additional medication analyses are performed quarterly and in case of crisis.

The quality of the prescribed medication is primarily measured by the Medication Appropriateness Index (MAI), an implicit prescriptive quality measure with good inter-rater and intra-rater reliability [62, 63]. The index provides ten criteria relevant for appropriate prescription, formulated as questions regarding indication, efficacy, dosage, correct dosage regimens, practicable directions for use, drug-drug interactions, drug-disease interactions, duplicate prescriptions, treatment duration and cost [62].

For a standardized evaluation of the criteria by the evaluators, the MAI-Score provides general instructions for the use of the index, precise definitions for each criterion and specific instructions and examples for answering the ten questions [62].

The evaluation of the criteria results in a weighted point score that represents the adequacy of each prescribed drug [63]. Only nine of the ten criteria of the MAI are used in DemStepCare, the question of economic efficiency of the prescription has been excluded.

In addition to the MAI, the following aspects are also addressed within the medication review to assess and evaluate safety and quality of medication:

The required information is retrieved from the electronic case record. Available clinical data (diagnoses, laboratory values) are included in the analysis. The implementation rate of pharmaceutical recommendations in the intervention group is determined based on the update of patient records.

The regular assessment of risks in home care is a central component of the DemStepCare intervention, since the further course of treatment in the intervention group is based on the results of this assessment. The DSV is used for this purpose [47]. Its data concerning stability of care will also be integrated as a secondary endpoint in the evaluation.

Besides the primary endpoint “days spent in hospital” (intervention group vs. control group 1), two additional secondary endpoints are related to hospital stays: Firstly, a comparison of days spent in hospital is also made between the intervention group and control group 2. For this purpose, the health insurance companies involved in the project provide data for control group 2. Secondly, the number of hospital stays is also analyzed (without considering the length of each stay), both between the intervention group and control group 1 and between the intervention group and control group 2. As with the hospital days, data is available either via questionnaires (intervention group and control group 1) or from aggregated data from the health insurance companies (control group 2) as well as for a small percentage of individual data from the health insurance companies (intervention group and control group 1).

The well-being of caregivers plays a major role in DemStepCare. Therefore, some secondary endpoints also refer to the caregivers. Quality of life is measured with the WHOQOL-Bref, a comprehensive instrument of the WHO, which covers various aspects of this concept [67]. We also use the Perceived Stress Scale (PSS-10), which asks for the perceived stress in the last month [68, 69]. We assess the caregivers’ resilience by using the Brief Resilience Scale (BRS) [70]. Additionally, comorbidities are queried. For this purpose, the KoMo-Score is used [71], to which typical comorbidities known from the care of PwD have been added. Caregivers fill in questionnaires with these scales at baseline (t0) and at t1.

The case managers assess the perceived stress of the caregivers at each interaction (including t0, t1, and t2 in the intervention group) on a 10-point scale.

In addition, the case managers note whether the caregivers use any support services. This aspect is also covered by the caregivers’ self-report in the questionnaires (at t0 and t1).

The secondary endpoints relating to patients and caregivers are shown in Table 2.

All participating GPs take part in a panel survey. Here, attitudes, self-efficacy, experienced competence, commitment, and perceived challenges with regard to the diagnosis and treatment of dementia are investigated. We use the term “dementia sensitivity” for this complex, heterogeneous bundle of attitudes and experienced abilities. The questionnaire that we developed especially for DemStepCare is based on our previous experience in this field [10, 58, 59]. In the case of the GPs, there are three measurements: one at baseline (t0_GP), i.e. before randomization and the first training session of the GPs, one 12 months after completion of the first training (t1_GP), and one at the end of the intervention phase (t2_GP).

The aim is to confirm whether the intervention group performs better than control group 1 with regard to the three primary endpoints. We assume that there are low to medium effects (effect sizes Cohen’s d = 0.30). Following the usual conventions, we require a power of 0.80 and two-sided testing. For the alpha level, a Bonferroni adjustment (5%/3 = 1.67%) is made due to the multiple testing of three primary endpoints. The number of cases required under these assumptions was calculated with the software G*Power.

Due to the clustering by GP practice, we need to calculate the design effect. A value of 0.01 is assumed for the intra-cluster correlation [72]. The number of cases per cluster (patients per practice) is assumed to be N = 15. The design effect is therefore DE = 1 + 0.01x (15–1) = 1.14.

It is estimated that 13 to 15 patients per practice will be included, so with 60 participating practices per group there will be approximately N = 800 patients per group. Our previous experience with similar study conditions suggests that dropout occurs in about one third of the cases. Assuming a dropout rate of one third, 800*2/3 = 533 persons remain in each patient sample. For the sample of caregivers, we assume a comparable dropout rate, but we must also take into account that only in about two-thirds of all cases is a caregiver available and willing to participate. In the sample of caregivers, 800*4/9 = 355 cases are therefore to be expected. Comparing these numbers with the required number of N = 268, there is a buffer in case of unexpected problems during recruitment.

We will reach GPs by public advertising, sending out flyers, and advertising at events, as well as face-to-face communication by our project partners.

Patients who are potentially eligible to participate in DemStepCare are contacted by their GPs. If possible, their caregivers are also addressed. Those who agree to join the study are assigned to either the intervention or control group 1, based on the group membership of their respective GP.

To evaluate the three primary endpoints, two-sided comparisons between intervention and control group 1 are performed to test for differences with multiple test level alpha = 5% using Bonferroni-Holm adjustment (lowest p-value < 1.67%, if applicable second lowest p-value < 2.5%, if applicable highest p-value < 5%). Due to the typically skewed distribution, the evaluation of the number of days spent in hospital is performed using negative binomial regression. The quality of life of the patients and caregiver burden are investigated using generalized linear regression models with an adjustment for the respective baseline measurements. The multi-level structure of the data is considered in all regression models by means of random effects at the level of the GPs practices. In addition, it is taken into account that some of the participants were affected by a COVID-19-related treatment break. In all three analyses, the dementia sensitivity of the GPs, which is relevant for randomization, is adjusted.

In addition, exploratory analyses are carried out using structural equation models to investigate the secondary hypotheses on how the intervention works.

For health economic evaluation, the type, frequency and duration of hospitalization are first used to determine the costs of hospitalization using standardized evaluation rates [73]. Then, the intervention-related cost reduction (in terms of hospital admissions compared to control group 1 and 2) is determined. This in turn is contrasted with the costs of the intervention. Finally, the net costs of the intervention (intervention costs minus costs of avoided hospital admissions) are compared with the intervention effects in terms of quality of life of the patients and reduction of the caregivers’ burden. To show the inference for the individual cost-effectiveness relations, a 95% confidence interval is calculated using Fieller’s Theorem (REF [74]).

In addition, the precise costs are evaluated for the subgroup for which individual data of the participating health insurance companies is available.

The interviews conducted in the process evaluation will be recorded and transcribed with digital audio recorders. The evaluation of the qualitative data is based on the qualitative content analysis procedure according to Mayring [75].