Implementation, efficacy, costs and processes of inpatient equivalent home-treatment in German mental health care (AKtiV): protocol of a mixed-method, participatory, quasi-experimental trial

The overarching goal of the AKtiV trial is to examine implementation processes, treatment processes, clinical efficacy, costs, and subjective experiences of IEHT compared to inpatient treatment from the perspective of service users, relatives or informal caregivers, staff and other stakeholders in mental health care. To maximize the transferability of study results and to cover a broad spectrum of IEHT experience, 10 hospitals from different regions in Germany (e.g. rural, urban, east, west) are participating in this study. Combining routine data, primary data and prospective follow-up data, the study results will be compiled in a comprehensive database. Furthermore, the combination of clinical and health economic data will enable the assessment of costs and benefits from a national perspective, an important characteristic, given that there are only a few studies with health economic evidence dealing with acute outreach mental health care [6, 26‐29]. The qualitative evaluation of IEHT processes and outcomes uses a collaborative-participatory approach that aligns with current demands for more user orientation and/or the involvement of people and researchers with relevant lived experience in the process of developing interventions and their evaluation [30]. The trial’s mixed-method design corresponds with current standards of empirical social research enabling the triangulation of hypothesis-confirming quantifiable factors and hypothesis-generating qualitative aspects [31]. By parallelizing quantitative and qualitative data on the one hand and routine data on the other with primary data, data on implementation processes and data on treatment processes, different facets from different perspectives and levels of IEHT are targeted. This allows for a comprehensive, holistic assessment of this innovative treatment option.

The study population consists of service users seeking IEHT and their caregivers living in the same household, as well as staff delivering IEHT and other stakeholders in the mental health care system and politics.

After admission to IEHT, staff will conduct an individual needs assessment based on which the team will develop a treatment plan. This plan may include treatment goals, various measures such as medication, psychotherapy, training and other daily or therapeutic activities, as well as therapeutic interactions with relatives, informal caregivers, legal guardians, and other persons from the participant’s social network. Furthermore, it will consider potential triggers for future crises, list former treatments, and individual preferences. Therapeutic interventions are adapted to the user’s needs daily. The team establishes daily contacts that can either take place at home, at the hospital, or at any place the service user feels comfortable with, but six encounters per week with service users must be realized outside of the hospital. In the weekly consultation by the psychiatrist in charge, the treatment progress will be reflected upon, and further interventions will be planned, such as therapeutic sessions or a change of medication. Treatment is realized according to the available resources and standards of the study sites. Service users will be discussed extensively in regular, inter-professional IEHT team meetings at least once a week, involving medical staff and nurses with participation of at least one psychologist, social worker or member of another professional group. Discharge planning follows inpatient procedures.

The main quantitative outcomes and hypotheses of the trial are shown in Fig. 1. The hospital re-admission rate within 12 months of the index crisis that originally led to the need for immediate admission either to IEHT or inpatient treatment serves as the primary outcome. This outcome has been used in most international home treatment studies [32]. The re-admission rate is by no means a perfect quality indicator of psychiatric care [33]. Nevertheless, it is an indicator of successful acute treatment, recovery and needs met within community mental health care [34].

Clinical and research staff from six institutions from different regions in Germany collaborate in the AKtiV trial which is divided into five modules: Module A examines quantitative outcomes among service users and relatives or informal caregivers; Module B uses a qualitative and collaborative-participatory methodology to map the care providers’ experiences; Module C analyzes the implementation processes and the treatment processes of IEHT; Module D evaluates health care costs, and Module E is in charge of all biometric and statistical questions relating to the trial.

Study sites were selected ahead of the trial in order to avoid disturbances during the recruiting period. Ten IEHT teams, associated with ten different psychiatric hospitals, from both urban and rural areas, located in different regions of Germany, agreed to participate in the trial: The Immanuel Clinic Rüdersdorf, Vivantes Hospital Am Urban (Berlin), Vivantes Hospital Neukölln (Berlin), and Charité – University Medicine Berlin in the North East, and the hospitals ZfP Zwiefalten, Ravensburg, Reutlingen, Isar-Amper Hospital Munich, University Clinic Tübingen and the Hospital ZfP Reichenau, in the South West of Germany. IEHT implementation had been started in these study sites between July 2018 and August 2020. The heterogeneity of hospitals is a strength of this trial as it allows for IEHT to be analyzed in different care settings and different arrangements.

Power calculations for the primary outcome were based on international RCTs included in the Cochrane Review [7] as well as on an analysis of routine data on IEHT from the participating study sites. In a first step, weighted mean re-admission rates by Hoult et al. [42], Fenton et al. [43], and Johnson et al. [44] as presented in the Cochrane Review were used. This led to an expected ratio of 70% re-admissions in the CG versus 45% in the IG within 12 months after admission, or rather a quotient of re-admission rates between IG and CG of approximately 1/3 (exact 0.358). However, data from nine participating study sites from overall N = 37,007 service users receiving either IEHT or inpatient treatment within 12 months showed re-admission rates among inpatient treated service users of 52.1%. This is about 20% lower than the internationally reported numbers. Analysis in a small pilot study (n = 86) site showed an even smaller difference between the re-admission rate of IEHT (37.2%) and inpatient treatment (46.5%) in a 12-month-follow-up-assessment [45], i.e. a reduction quotient of about 1/5. Following this data, expecting a quotient of readmission rates between IEHT and TAU of about 1/3 as internationally observed seemed to be too optimistic. Thus, for the trial presented, we used our own TAU value of 52.1% and combined all data sources in order to receive a realistic IEHT rate by a weighted mean of the reduction factors applying a 10% higher weight for the international value than for the small pilot study factor. As a result, an IEHT rate reduction by a quotient of about 0.288 was obtained. Thus, an admission rate of 37.4% after IEHT is expected and should be detected. A power calculation using nQuery Advisor 7.0. with an alpha of 5% and a power of 80% revealed in a two-sided chi square test a total number of 360 service users (n_IG = 180, n_CG = 180) that must be included into the trial to show the effects named above. Assuming a non-response rate of 10% as shown in two pilot studies [46, 47], approximately 400 service users (n_IG = 200, n_CG = 200) have to be canvassed in order to achieve the optimum number of participants for the trial.

The recruitment period of the AKtiV study lasts 12 months. The number of IEHT cases in the preceding 12 months regarding the preparation of this manuscript ranged between 44 and 209 with a mean of approximately 120 IEHT cases per study site. Available treatment units at participating study sites currently range from five to 21 resulting in 130 IEHT units overall (mean = 13 IEHT units per study site). The mean stay of IEHT service users during the index period in the preceding 12 months was approximately 32 days. Beyond this background, the inclusion of 180 IEHT users into the trial within 12 months across all ten study sites is ensured. Recruitment of the same number of participants for CG is unproblematic in all the study sites since the amount of inpatient treatment units outnumbers the amount of IEHT in all hospitals by far.

All assessments are conducted face to face either in the hospital, at the service user’s home or in his or her social surroundings. The sequence of assessments in each study cohort is outlined in Table 1. The following incentives for study participation are provided: service users receive 50 € after completing the 12-month-follow-up assessment; relatives or informal care givers of service users receive 20 € after completing the baseline assessment.

Module C1 includes a) institutional and structural data from the study departments, b) routine data about service provision, c) additional data about IEHT implementation, and d) quantitative and qualitative primary data collected from IEHT team members about their work satisfaction. To explore treatment processes, personal routine data according to § 301 SGB V will be used. This includes performance data involving admission, working and discharge diagnoses, information about therapy times, pretreatment, and transfer from IEHT to inpatient setting or vice versa. Routine data will be complemented by additional personal user data about the psycho-social background, treatment course, and illness chronicity. Based on qualitative surveys with the IEHT team members, comprehensive information about the implementation process will be collected. In addition, the employee’s job satisfaction will be assessed using quantitative measures.

A comparison between the type of services provided by different professional groups and the amount of time dedicated to service users, therapeutic groups as well as organization and administration, IEHT team structure and data about service users enables an exploration of the impact of the IEHT team processes on treatment output and outcome.

Furthermore, Module C analyses quantitative data assessed by Module A regarding satisfaction with the provided services as well as aspects of shared decision making from the perspective of users and their relatives or informal caregivers. In addition, these analyses will be related to quantitative data regarding employees’ satisfaction and data from qualitative focus groups that will be carried out in teams providing IEHT. The latter make it possible to consider primarily non-quantifiable data on the relevant effects of team building, cohesion and the thematic foci of IEHT, which in turn determine the way in which IEHT services are provided and structured.

To address these research questions, service users’ characteristics (such as age, gender, diagnoses, symptom severity, psycho-social functioning, social network, marital status, socioeconomic status, and previous treatment episodes), treatment processes and outcomes regarding the two admission pathways will be examined. A central goal of Module C2 is to gain knowledge about how to improve indication of IEHT and as well as process and outcome quality.

We will use both single- and multi-center analyses. The multi-center study will include user-specific admission data, data about the treatment process as well as outcome data across all recruiting sites. The single-center analysis will include more detailed quantitative data (such as patients’ aggression etc. at direct IEHT admission vs. inpatient treatment admission) as well as qualitative data, which will not be available in all study centers. It will be conducted at the Munich study site, currently holding the largest German IEHT unit with 20 treatment places.

Following the net benefit approach, a cost-utility analysis (CUA) based on primary data will be conducted [58‐60]. The aim of this analysis is an estimation of the maximum willingness to pay (MWTP) needed for gaining one quality-adjusted life year (QALY) through IEHT compared with TAU. The CUA from a societal perspective is based on a complete analysis of direct and indirect health costs as well as measuring subjective quality of life with a preference based method [61, 62]. Registering all health care costs is only possible by directly asking service users. This is because mental health care services in Germany are funded by various different bodies according to the Social Code Book and other laws [60, 63]. Therefore, direct and indirect medical and mental health care costs will be assessed with the German Version of the Client Socio-Demographic Service Receipt Inventory (CSSRI-D) [35, 64]. Health care related costs will be calculated by multiplying units of care with relevant, published unit costs values [60, 65, 66].

Quality adjusted life years (QALYs) will be estimated on the basis of health states estimated by means of the EQ. 5D-5L [67, 68] and utility values from the German population provided by Ludwig et al. [69].

Based on the “state of the art” discussion regarding the choice of threshold values for the MWTP, threshold values of 25,000, 50,000, 75,000, and 100,000 € will be taken as the basis for net benefit regression models [70]. However, thresholds can always be differentiated further.

Before starting the recruitment process, logistic regression analyses of the previous year’s data from all participating study sites were conducted to calculate PS functions. Data included those of service users of IEHT as well as inpatients fulfilling the trial’s inclusion criteria over a period of 12 months. “Participation in IEHT” (yes/no) was chosen as a dependent dichotomous variable, along with the independent variables age, sex, primary diagnosis (FX), and number of psychiatric hospitalizations for the last 2 years in the corresponding study site. These analyses provided every study site with a specific function for calculating their PS given their respective regression variables. By means of this function, each participant included into the IG will be matched with the most comparable service user treated in inpatient units of the participating study centers.

Intention-to-treat (ITT) analysis will be conducted to avoid selective reporting and biases in the analysis due to possible non-random attrition of participants and missing values [71]. Additionally, multiple imputation will be conducted for missing values.

Regression variables used for matching as well as propensity scores will be tested for homogeneity in an exploratory manner. Descriptive statistics will be performed for all baseline values and outcomes at follow-up. Differences between intervention and control group will be analyzed based on the endpoint or working hypothesis categories “primary”, “secondary”, and “tertiary” following the null hypothesis “No difference between intervention and control group.” Among service users, these categories applied as followed: P = hospital re-admission rate; S1 = combined re-admission rate (inpatient + partly inpatient + IEHT), S2 = inpatient hospital days; S3 = continuity of care/withdrawal from treatment, S4 = health related quality of life, S5 = psycho-social functioning, S6 = job integration, S7 = recovery, S8 = satisfaction with care, S9 = shared decision making; impact of T1 = diagnosis groups, T2 = region, T3 = employment status, T4 = chronicity of disorder, T5 = additional therapies. Assessments among relatives or informal caregivers will be analyzed by Module C and belong to secondary endpoints: S10 = burden, S11 = satisfaction with care. Thereby, P1, S1, S2, S4, S5, S6, and S7 refer to the time span of 12 months (main analysis) and 6 months (supporting analysis) after baseline assessment. S3, S8, S9, S10 and S11 refer to the index episode.

All statistical analyses will be conducted with SAS. Prior to the end of data assessment, a statistical analysis plan (SAP) will be created determining statistical methods, e.g. for transformations, imputations and, analyses of quantitative data to be processed especially for Module A and partly Module C. A synthesis of results from the different Modules will be described procedurally in the SAP. The main features of the plan are as follows:

Results will be interpreted regarding reference values and similar study populations.

One limitation of the “AKtiV” study is the lack of direct randomization of study participants. Randomization is not possible as IEHT already is standard care and a preference-based alternative to psychiatric inpatient treatment. Furthermore, the randomization of service users in the CG may lead to conflicts if they have important reasons to prefer IEHT. This could e.g. be the case if a user is a single parent care giver or is caring for an elderly person living in the same household. On the other hand, some service users may not or cannot be treated at home even though they fulfill the inclusion criteria. Moreover, the fact that the number of IEHT units per clinic ranges from 5 to 21 makes randomization challenging. In addition, samples from cohort studies appear to be more representative for clinical routine samples than samples derived from RCTs because they are not biased. Against this background, it was decided to carry out a quasi-experimental cohort study with PS matching, rather than a randomized controlled trial. Another limitation is the relatively short follow-up period. Since funding guidelines only allow for a study period of a maximum of 3 years, the final follow-up-assessment cannot be longer than 12 months. Even though a longer observation period would be desirable in order to analyze the long-term effects and sustainability of IEHT, the current trial duration is sufficient for an initial methodologically sound evaluation of this new home treatment model.