Background

In 2021, almost half of patients in German hospitals were 65 years and older [1]. Older age is associated with cognitive impairment, with a prevalence among elderly hospital patients ranging from 40 to 67.5% [2, 3]. This mainly includes people living with dementia, mild cognitive impairment, and with delirium [3].

Hospitalisation of people with cognitive impairment is associated with several challenges. One challenge is a frequently missing formal diagnosis of cognitive impairment [2, 3], resulting in care that is not tailored to the needs of this vulnerable patient group. An integrative review on hospital outcomes [4] found, that people with cognitive impairment encounter a hospital environment focussed on acute illness rather than their individual needs, as shown by avoidable bed moves, inadequate pain management, inappropriate catheterisation, and poor communication and relationship building with healthcare professionals. Negative experiences are common, frequently leading to feelings of fear and an increase in behavioural and psychological symptoms. In addition, people with cognitive impairment are more likely to encounter a range of related adverse events such as the incidence of delirium, urinary tract infections, pneumonia, pressure ulcers, adverse drug reactions, and falls, that may lead to prolonged hospital stays and increased mortality [4]. Healthcare professionals struggle with the discrepancy between desirable hospital care and the current reality of care for people with cognitive impairment, resulting in distress and frustration, uncertainty in dealing with changed behaviour, conflicting priorities, and the inability to meet individual needs [5].

A key to improve hospital care for people with cognitive impairment is the concept of person-centred care (PCC), which has been recommended in guidelines [6, 7] and by representative organisations [8]. Person-centredness has been applied to different target groups such as people living with dementia [9], different healthcare professionals such as nurses [10], and different care contexts such as hospitals [11]. Despite the widespread use of PCC, there is no consensus on how it is operationalised conceptually. According to a recent scoping review [12], the most common of nine identified dimensions of person-centredness are “perception of the individual as a unique person”, “sharing of responsibility between person and practitioner”, and “formation of a therapeutic relationship". However, the literature indicates a gap between the theoretical concept and practical implementation of PCC in the hospital setting [13]. Reported barriers to provide PCC include lack of knowledge, understanding and training of healthcare professionals [5, 14], ward and institutional cultures, and inappropriate physical environment [5]. To promote PCC in hospitals, change is needed at both institutional and ward levels. Here, leadership of key personnel plays a crucial role, as staff training is not sufficient to implement PCC successfully [15,16,17]. At ward level, so-called change agents with clinical expertise can support staff awareness and initiate change processes in care [17]. This role of a clinical leader reflects one core competence of an Advanced Practice Nurse (APN) [18], a specialised nurse with a master’s degree, possessing expert knowledge, complex decision-making skills, and clinical competencies [19]. APN interventions for older people have shown positive outcomes in primary and long-term care. Evidence for the acute setting shows mixed results [20, 21].

As Germany is at the beginning of establishing APN, Expanded Practice Nurses (EPN) at bachelor’s level are deployed to facilitate the transition to Advanced Practice Nursing [22, 23]. Expanded nursing roles for PCC for people with cognitive impairment in hospitals are rarely described in the literature [24, 25].

The development and implementation of an EPN role is complex as various components and their interactions need to be taken into account. To comprehensively investigate complex interventions, the guidance of the British Medical Research Council (MRC) recommends the evaluation of effects, while considering underlying processes and influencing factors as well as costs [26].

In the ENROLE-acute (Expanded Nursing ROLEs for person-centred care for people with cognitive impairment in ACUTE care) project, we are developing, implementing, and evaluating a complex PCC intervention for people with cognitive impairment, delivered by EPN on peripheral wards in one acute hospital. We hypothesise that by introducing PCC, the individual needs of people with cognitive impairment will be addressed in a timely manner and thus adverse events will be reduced or avoided, leading to a reduction in the length of hospital stay.

Objectives

This study aims to test the feasibility and clinical effectiveness of the newly developed PCC intervention. Therefore, our objectives are to assess (a) initial effects, (b) the degree and quality of implementation, mechanisms of change and contextual factors, and (c) potential cost savings of the intervention.

Methods / Design

Overall study design

The ENROLE-acute project follows the first and second stage of the MRC guidance [26] for complex interventions, with this study protocol addressing the feasibility stage. Protocol development adheres to the SPIRIT checklist [27] (see Additional file 1).

To examine initial effects of the ENROLE-acute intervention, we will conduct an exploratory non-randomised controlled clinical trial (CCT), comparing the PCC intervention and optimised care as control intervention. The CCT will be accompanied by an embedded mixed-methods process evaluation to explore how and under what circumstances outcomes are achieved. In addition, we will conduct a health economic evaluation. Figure 1 summarises the overall study design.

Fig. 1
figure 1

Participant timeline

Full size image

Setting

We will conduct the study in one university hospital in the western part of Germany. The hospital has about 1,500 beds, with inpatient care provided to 57,900 patients in 2022. According to internal hospital controlling, about 1,500 people with diagnosed cognitive impairment were treated as inpatients in the same year.

The hospital has about 12,000 employees [28]. In recent years, first EPN have been introduced. At present, there are 36 nurses working in expanded roles in wound management, oncology, psychiatry, obstetrics and gynecology, surgery, and cardiology [29, 30].

Intervention

In the first phase of the ENROLE-acute project, we comprehensively developed the complex intervention based on two systematic reviews, two surveys [31], one qualitative interview study, and two workshops with relevant stakeholders from clinical practice, research, and representative organisations of people with cognitive impairment. We will publish details of intervention development elsewhere.

The complex intervention aims to foster PCC for people with cognitive impairment through EPN acting as change agents (see Fig. 2). We defined 14 components reflecting EPN’s role in collaboration with the interprofessional team. The model includes core components that are mandatory (no. 1–12) and peripheral components that are adaptable (no. 13–14) [32]. We differentiate between patient-related and system-related components. Patient-related components are aimed at individual patients and include screening and expanded nursing assessment at patient admission (no. 1–2); planning, conducting, and evaluating PCC interventions in collaboration with registered nurses (no. 3, 4 and 9); and further EPN interventions including case conferences on discharge planning, coaching relatives, and conducting and initiating consultations (no. 5–8). System-related components include EPN interventions at the organisational level, e.g., training interventions for staff, monitoring of change processes, promoting hospital-wide collaboration, conceptual work, and nursing research (no. 10–14).

Fig. 2
figure 2

Intervention components

Full size image

Each intervention ward receives two EPN, qualified at level six of the European Qualifications Framework (bachelor’s degree or equivalent further education) [33]. EPN are integrated into the ward's work schedule and have one day each every two weeks for additional activities.

We developed different implementation strategies to prepare the field. EPN received a 200-h training regarding fundamentals of caring for people with cognitive impairment, principles of PCC, and Advanced Practice Nursing in six modules. In addition, EPN gained knowledge of established expanded nursing models in other hospitals during a 10-day internship. During the intervention period, we accompany the EPN fortnightly in two-hour coaching sessions for reflection and implementation support. We also train the interprofessional team at the beginning of the intervention phase on PCC principles and Advanced Practice Nursing during a one-hour kick-off meeting and through written information.

Team members of control wards provide optimised care with knowledge acquired through a one-time half-hour session and written information on PCC.

Evaluation of effects

Participants

For details on the inclusion criteria of all participants, please see Additional file 2.

Patient level

We defined the following inclusion criteria for potential participants: (1) age of 65 years and older, (2) sufficient knowledge of German language, (3) hospital stay of unclear duration or over 48 h, (4) ability to consent or existence of a legal representative, and (5) present cognitive impairment (e.g., diagnosis of dementia) or risk for cognitive impairment (e.g., risk for delirium) according to predefined criteria.

We will recruit potential participants on six days a week over a time period of six months. Contact persons of the respective wards will inform eligible people with cognitive impairment and, if necessary, their representatives about the study and ask for consent to be approached by the research team for obtaining informed consent.

Staff level

Members of the interprofessional team are registered nurses (European Qualifications Framework level 4), nursing assistants, and physicians of all participating wards. We exclude therapists as well as service and administrative personnel as they do not form the core of the interprofessional team on a ward. We will recruit members of the interprofessional team during kick-off meetings and via superiors.

Cluster level

We will purposefully select six peripheral wards with a prevalence of at least 30 people with cognitive impairment per month according to data from hospital controlling. We will exclude intensive care, intermediate care, psychiatric, palliative, and paediatric wards. Based on the recommendation of nursing management, we will identify eligible wards and recruit them through individual information meetings with ward managers.

Outcomes, data collection and management

For an overview of primary and secondary outcomes, instruments used, data collection time points, and data sources, please see Table 1.

Table 1 Overview of the evaluation of effects
Full size table

Patient level

At patient level, we will collect data at a maximum of six timepoints: t1 (at admission), t2 (at day 3), t3 (at day 7), t4 (at day 14), t5 (at discharge), and t6 (30 days after discharge). We will collect study documentation data (e.g., participant flow) paper-based and outcome data electronically using REDCap®. Trained study staff will accompany people with cognitive impairment for self-rating. Proxy-rating will be carried out by trained study staff or nurses who have been responsible for the participant for at least one shift within the past 24 h. In case of non-participation of people with cognitive impairment in the study, we will document the reason.

Data at patient level will be documented using participant codes. We will keep identification lists and paper-based documents under lock, separate from other study material and only accessible to data collectors. Electronic data will be password-protected and only accessible to the research team. We will automatically carry out a complete backup of the data every day. With the set-up of the database, we have stored definitions (e.g., minimum and maximum) for each value to be entered, which minimises the entry of invalid data. In addition, the biostatistician will perform quality and plausibility checks (data validation).

Staff level

For outcomes at staff level, we will use a paper-based questionnaire at two timepoints: t0 (before intervention phase) and t7 (after six months). We will collect the data pseudonymously and use a self-generated code to allow both sets of data of one participant to be matched over the course of the study. To increase retention, we will use incentives and email reminder to ward managers.

Only data collectors will have access to the paper-based questionnaires, which are kept under lock. We will keep potentially identifying data separate from the rest of the study material. One person will conduct electronic data entry and random samples of the record will be double-checked by another person. In case of discrepancies, the entire data entry will be double-checked.

Assignment of intervention

Allocation of wards as intervention or control ward was based on ward managers’ decision to participate in the intervention. Participants’ assignment will be determined by their wards’ assignment.

Blinding

Due to the nature of the intervention, blinding of participants, EPN providing the intervention, and researchers collecting data is not feasible. However, we will blind the biostatistician during data analyses regarding group allocation of clusters (neutral group IDs) and participating people with cognitive impairment.

Sample size

The sample size is based on a recent retrospective cohort study which investigates the length of hospital stay in older people living with dementia by matching people without dementia in Germany [63]. We assume a mean reduction in the length of hospital stay of three days (intervention vs. control). With a standard deviation of 18 vs. 16 days, an intraclass correlation coefficient of 0.05 [64], and three clusters per treatment group with 120 people with cognitive impairment each, the adjusted two-sample t-test will reach a power of 81.5% at one-sided significance level of 50% [65]. Thus, this exploratory trial will be sufficiently powered to show the hypothesized direction of effect only.

Data analysis

Statistical analysis is according to intention-to-treat (no exclusions). We will evaluate the primary outcome length of hospital stay by the stratified log-rank test (death is censored). Nota bene, the stratified log-rank test has very similar characteristics as the stratified van-Elteren rank-sum test and is available in many (statistical) software packages. We will use a propensity score approach (PSA) to balance individual characteristics between participants in intervention and control groups. In the multivariable logistic regression for PSA, we will use the independent variables age, gender, severity of cognitive impairment, Charlson Comorbidity Index, admission type, expected admission diagnosis-related or ward-related length of stay, and hospitalisations during the last 7 days [66, 67] as these variables are likely to be associated with outcome. Moreover, we will carefully monitor mortality in both arms, although we do not expect any difference. We will evaluate secondary outcome measures by (generalised) mixed models for repeated measures over time. If feasible, we will take outcome measures (particularly those reported by people with cognitive impairment) over the course of hospital stay. Thus, we may need to evaluate the outcome trajectories by random coefficient models. We will categorise and compare safety outcomes by frequency, relatedness, seriousness, and severity. Moreover, we will summarise variables by mean, standard deviation, and percentiles (0, 25, 50, 75, and 100) respectively by absolute and relative frequencies (percentage). Due to the small number of clusters, we will at least account for them in a fixed effect analysis. Though we will attempt to estimate corresponding random effects (e.g., to derive intraclass correlation), this seems prone to bias.

Evaluation of processes

The process evaluation is based on a convergent parallel mixed-methods approach [68]. Theoretical foundations are the MRC guidance for process evaluations of complex interventions [69] and Grant’s framework on process evaluation for cluster-randomised trials [70]. For the design and conduct of the study, we developed a logic model that illustrates expected causal pathways of the ENROLE-acute intervention and potential moderating effects of relevant contextual factors (the logic model will be published as part of the intervention development). We used the logic model’s domains (implementation, mechanisms of change, and context) to organise all process variables in a predefined analysis-plan. Our purpose is that by evaluating processes before, during, and after the intervention period, observed effects can be interpreted in the light of how the complex intervention actually works. For an overview of domains, questions, data collection, participants, and time points, please see Table 2.

Table 2 Overview of the evaluation of processes
Full size table

Participants

We will ask members of the three hospital wards in the ENROLE-acute intervention group to participate in the process evaluation. Target groups of the quantitative and qualitative parts are all EPN, members of the interprofessional team, and ward managers. The qualitative part additionally includes people with cognitive impairment and their relatives as well as members of the hospital’s nursing management and nursing development unit. For the eligibility criteria for each target group, please see Additional file 2.

When selecting participants, we will combine different sampling strategies. For all quantitative parts and the qualitative parts on EPN level, we strive to recruit all members of the relevant target groups. For the remaining explorations, we will form subsamples through purposeful sampling. A predefined sampling plan ensures that different forms of cognitive impairment respectively health professions as well as all three clusters are represented in the sample. The targeted sample size will be guided by reaching data saturation.

For recruitment, we will invite members of the target groups both orally and in writing to participate in the study.

Data collection and management

We will collect both qualitative and quantitative process data on cluster and individual level at five time points: tp0 (before training, internship, and kick-off event), tp1 (after training, internship, and kick-off event), tp2 (start of the intervention period), tp3 (after transfer or discharge of people with cognitive impairment), and tp4 (6 months after start of the intervention period).

The qualitative part of the process evaluation comprises individual (EPN, ward managers, members of the nursing management and the nursing development unit) and dyad interviews (people with cognitive impairment and their relatives) as well as moderated and non-moderated focus groups (EPN, members of the interprofessional team). We will collect data face-to-face, by phone or by web-based video conferencing. Nursing researchers trained in qualitative methodology will guide the data collections. We will audiotape and transcribe interviews and group discussions verbatim.

To collect quantitative data, we will use paper-based questionnaires (EPN, members of the interprofessional team, ward managers) and reflection forms (EPN, members of the interprofessional team). A self-generated identification code will allow us to link participants’ data over time. To increase retention, we will use incentives and email reminder. After data entry, we will perform quality and plausibility checks.

For collecting quantitative and qualitative data, we refer to pre-existing and approved instruments. Only if this is not possible, we will use self-developed instruments. We conduct pretests of all instruments to be used.

In addition, we will gather different process documents and materials (e. g., structured notes on recruitment and retention, protocols and attendance lists, teaching and learning materials, and EPN logbooks).

We will store data both pseudonymously (EPN and ward managers) and anonymously (people with cognitive impairment and their relatives, interprofessional team members, and members of the nursing management and the nursing development unit), while documenting the time point and cluster assignment on data sheets. Data protection follows current data protection laws. We will keep paper-based documents under lock. Electronic data will be password-protected. We will store potentially identifying data separate from the rest of the study material. Data will only be accessible to the research team.

Data analysis

For analysing the process data, we will use a stepwise approach. First, we will analyse qualitative and quantitative data separately. Subsequently, we will merge data during the interpretation of the results.

For analysing the qualitative data (interviews and focus groups), we will conduct a thematic framework analysis [71] with a deductive-inductive approach based on the logic model. Furthermore, we will use a qualitative content analysis [72] applying a deductive-inductive approach to review the structured process documents and materials as well as open-ended questions in the questionnaires. We will conduct all qualitative analyses in the software MAXQDA [73]. To promote credibility, two persons will analyse data and discuss findings within the research team.

We will analyse quantitative data (questionnaires and reflection forms) by means of descriptive statistics (frequency, distributional location, and spread) using the software IBM SPSS Statistics [74].

We will combine qualitative and quantitative findings regarding key process variables and outcomes incorporated in our logic model.

Economic evaluation

We will perform a cost-consequence analysis from the hospital perspective [75].

Data collection and management

We will use claims data from the hospital to determine the potential cost savings of the intervention. These will be available at the hospital controlling department. This data includes Diagnosis Related Groups (DRG) charges, length of hospital stay, cost weight, gender, and primary and secondary International Statistical Classification of Diseases and Related Health Problems (ICD) diagnoses. The data will be delivered pseudonymously to the evaluators via a secure data exchange platform. In addition, we will collect training costs consisting of salaries of the nursing staff, the costs of the lecturers as well as material costs for premises, catering and travel costs.

Data analysis

In the cost-consequence analysis, we will compare the intervention group with the control group in terms of total inpatient costs, length of hospital stay, and cost weight. We will add the intervention costs (e.g., training costs) to the DRG charges of the intervention group. We will analyse differences (possibly after log-transformation) between the mean total costs using a t-test. In addition, we will perform additional analyses according to gender, age groups, and cost weight using t-test, Mann–Whitney-U-test, and other statistical tests. The analysis will be performed with IBM SPSS [74] and Microsoft Access.

Dissemination

We will publish main study results in international, peer reviewed journals. In addition, we will present our results at relevant scientific conferences and meetings. We will report the results based on this study protocol as well as the recommendations of current reporting guidelines. Main study information (e.g., intervention material and results) will be freely available at our project website (https://www.enrole-acute.uni-koeln.de/).

We will share authorship between persons involved in the study following the current guidelines of the International Committee of Medical Journal Editors. Persons not directly involved in the study will not be granted authorship.

Discussion

Complex interventions are considered most promising in promoting PCC [76]. The ENROLE-acute intervention builds on implementing an expanded nursing role as a core element to enhance PCC for people with cognitive impairment in acute care. Intervention development and testing are theory-based and follow recent guidance [26, 69]. The planned exploratory CCT will provide insights into the intervention’s initial effects, with the primary outcome focusing on patient length of hospital stay. Since there is a lack of high-quality studies to evaluate PCC interventions for people with cognitive impairment using expanded nursing roles in acute care, our study will provide important evidence. With the accompanying mixed-methods process evaluation, we will additionally examine the intervention’s implementation, mechanisms of change, and contextual factors. Analysing these underlying processes will lead to more transparency in trial results and a comprehensive understanding of how and under what circumstances the intervention will actually work. The additional economic analysis will increase our knowledge regarding the costs of the intervention.

However, some limitations of the study must be considered. Due to practical reasons, randomisation of the clusters will not be feasible. This may lead to differences between the groups at baseline resulting in a high risk of selection bias. However, we will consider these differences in the statistical analyses. Due to the nature of the intervention, blinding of participants and researchers during the study period will also not be possible. The knowledge of the group allocation might affect both the participants’ answers and data collection by researchers. At least the statistician will be blinded during data analysis. Finally, staff turnover during the six-month intervention period might result in differences between staff participating in the data collection at baseline and follow-up.

Despite the limitations, the study results will support the future development and implementation of new as well as the optimisation of existing PCC interventions based on expanded nursing roles. Our implications can particularly inform evidence-based decision making at the management and policy level. Additionally, our results will provide valuable information for planning and conducting large-scale trials as claimed in the third and fourth stage of the MRC guidance [26].