Flexible, integrated, and person-centered psychiatric care through global treatment budgets: results of the multiperspective study PsychCare

For the quantitative study, 3594 patients were screened, of whom 2824 (78.6%) fulfilled the inclusion criteria. From these, 1495 (52.9%) consented to participate (FIT: 51.2%, TAU: 54.6%). With a return rate of 81.0% in FIT and 76.9% in TAU, 1183 patients provided assessment data at M‑I (n = 1150 after reconciliation of discharge diagnoses and exclusion of incomplete questionnaires; FIT n = 595, TAU n = 555; see Fig. 2).

Bild vergrößern

The sociodemographic and clinical characteristics of participants were comparable regarding age, sex, living circumstances, diagnoses, symptom severity, and duration of illness in FIT and TAU (see Supplement S2).

A significantly higher treatment satisfaction at M‑I (Patient Satisfaction Questionnaire [ZUF-8]: 26.30 ± 4.36 vs. 24.90 ± 4.70; p < 0.001) was found for FIT patients. The difference was statistically significant; however, it did not reach the study-specific pre-defined difference of two points assumed to indicate clinical relevance. At M‑II, while scores were numerically still higher in FIT, there was only a trend toward a difference (26.20 ± 4.34 vs. 25.40 ± 4.41, p = 0.095). At M‑I, HRQoL was also significantly higher in FIT patients (here reaching the pre-defined difference assumed to indicate clinical relevance), with again no statistical difference at M‑II [14].

At M‑I, patients in the FIT group showed a significantly higher recovery (RAS-R: 134.0 ± 35.8 vs. 119.0 ± 54.3; p < 0.001). This was also true for all five subscale scores. Again, at M‑II, there was no longer any difference (141.0 ± 36.9 vs. 141.0 ± 34.2, p = 0.864; Table 1).

The reader is referred to Supplement S3 for the results of the regression models. Higher treatment satisfaction and recovery in FIT at M‑I were confirmed. Patients in day care and outpatients were more satisfied compared to inpatients. Outpatients showed higher recovery compared to inpatients. Patients with affective disorders seemed to be less satisfied with care and to have lower recovery scores compared to patients with a substance use disorder.

About half of the participants reported a high satisfaction with clinical decision-making (46.6% FIT vs. 50.3% TAU, M‑I), while only 6.0% and 8.2%, respectively, reported low satisfaction. The implementation of shared decision-making was high: 85.0% (FIT) and 80.2% (TAU) reported having made the decision together with their clinician or having taken actively part in the process at M‑I.

Caregivers of 113 FIT and 58 TAU patients were assessed at M‑I regarding their burden. In the three domains evaluated and at both time points, caregivers of FIT patients were numerically, but not statistically, more satisfied (see Supplement S4).

The direct medical costs at M‑I were significantly lower in FIT compared to TAU (15,667.1 €, SD = 17,475.1 € vs. 18,331 €, SD 17,722.4 €, p < 0.01). Patients in the FIT group were significantly less often in inpatient care (59.9% vs. 76.6%; p < 0.01), and the duration of inpatient care was significantly shorter (28.5 days, SD = 40.4 vs. 41.9 days, SD = 44.1; p < 0.01). There was no group difference in non-medical and indirect costs. Societal costs at M‑I were 21,200 € for FIT and 23,310 € for TAU (p = 0.058). At M‑II, societal costs were also lower in FIT than in TAU (19,336 € vs. 26,654 €; p = 0.093). The statistical significance level was not reached, potentially because of the sample size, particularly at M‑II, and the high variance, as reflected by the high standard deviations (sdFIT M‑I = 19,070 €; sdTAU M‑I = 19,991 €; sdFIT M‑II = 14,992 €; sdTAU M‑II = 17,159 €). With costs being numerically lower and ZUF‑8 and HRQoL scores significantly higher at M‑I, the results suggest the cost-effectiveness of FIT at M‑I. At M‑II direct medical costs were again significantly lower in the FIT group (5807.9 €, SD = 10,262.7 € vs. 9961.6 €, SD = 15,453.3 €; p = 0.005), again with a significantly shorter duration of inpatient treatment. No group differences were found with regard to societal costs, and cost-effectiveness could not be appraised due to the low number of individuals participating at that time point.

Good internal consistency was found for the NEPT (Cronbach’s α = 0.89). There was an increasing trend (p = 0.03) in the overall value across three independent groups based on the degree of FIT contract design, and the NEPT construct demonstrated a comparable validity to the ZUF‑8 construct. A significantly higher degree of implementation of the specific care components was observed for FIT compared to TAU (reported separately, [24]). They consistently exhibited higher averages reaching the significance level except for continuity of treatment team, multiprofessional collaboration, and sovereign steering of therapeutic measures (Tab. 2). In the mixed-method convergence design, clear associations emerged between hospital types and the degree of implementation of care components. In a two-dimensional representation, FIT with 100% FIT-contract design and/or long contract durations was distinctly separated from TAU and FIT that did not meet either criterion.

Providers reported a de-economization of the clinical decision processes. The cooperation and shared aims of hospitals and SHI were seen to be central to the implementation. Participant observation revealed central challenges for realizing the transfer between the various service components: the spatial proximity/distance of the components, the realization of innovative alliances between these, and the information flow. Staff members perceived the mobility and flexibility needed as freedom and potential but also as an additional burden in terms of time.

A large proportion of patients whose SHI was a partner in PsychCare agreed to the linkage of their self-reported and SHI data (89% FIT, 82% TAU). Relatively high absolute concordance on resource utilization was found across all settings, with higher values for inpatient and day care compared to outpatient services (see [7]).

Compared to TAU, FIT was cross-sectionally superior with regard to treatment satisfaction and HRQoL [14], recovery, and caregiver satisfaction. The degree of implementation of specific care components for a flexible, patient-centered treatment was higher in FIT. Direct medical costs for FIT were significantly lower and FIT care seemed to be cost-efficient. Providers reported de-economization of the clinical decision processes. The cooperation and shared aims of hospitals and SHI were reported to be central to implementation. There was a higher demand for mobility and flexibility for the staff to realize the transfer between service components. Linkage of self-reported and SHI data was feasible, and a high proportion of patients agreed to this.

EVA64 found a significant reduction in the length of inpatient stay for patients initially treated in the hospital (no treatment in that hospital within the last 2 years) by about 5 days in FIT compared to TAU with a shift to day care and in some hospitals to outpatient treatment. The duration of sick leave, however, was not different in the groups [2, 12]. The odds for treatment continuity increased by 1.4 in FIT, indicating that the chance was higher for FIT patients to be seen in the psychiatric outpatient department and/or by a resident psychologist or psychiatrist within 30 days of discharge from hospital [13].

The findings of superiority of FIT at M‑I regarding treatment satisfaction, HRQoL, recovery, and satisfaction of caregivers, as well as the implementation of flexible and integrated care components, confirmed the study hypotheses. The FIT and TAU hospitals were matched regarding regional as well as structural indicators, and the patient groups were comparable in major sociodemographic and clinical variables. At the outset of PsychCare, FIT hospitals had started FIT ≥ 2 years earlier and/or had applied a precursor FIT-like model treatment already (with one hospital having started a precursor model in 2003). Thus, M‑I differences in PsychCare are most likely attributable to ongoing model care performance. Assessing changes in outcomes over time is challenging, as care models are complex and component structures vary. Models of FIT function as learning systems, starting with an introductory phase and followed by process reorganization. Our study shows greater implementation of model components alongside the share of SHI funds included in the individual model contract. This suggests that key changes, i.e., in the attitude of staff (e.g., focusing more strongly on patient- and care-giver needs, working very close in cross-setting and cross-professional way, 24/7 availability) and structures (e.g., providing more day-care and flexible outpatient capacity, cross-setting group treatment) are more likely achieved when the majority of patients receive model care, rather than limiting it to those insured by individual SHI funds [23]. The government commission on a modern and needs-based hospital care in Germany [4] recommended that all SHI funds are to contract and evaluate model care programs that already exist in regions with ≥ 25% of the population. This would lead to a consistent regional regularization of psychiatric care and be an incentive to transfer the model to regional standard care.

Recovery and high HRQoL are key treatment aims: FIT patients showed better overall and component-specific recovery and reported higher HRQoL, supporting the prioritization of FIT-like care models in future mental health financing. Regression analyses showed that day-care patients and outpatients had higher satisfaction and recovery than inpatients. As FIT aims to increase flexibility based on patient needs, the shift toward day-care and outpatient treatment aligns with these findings.

The involvement of patients in treatment decision-making was high with at least 80% in both FIT and TAU groups, which is in line with guideline recommendations and shows that care outside model treatment provides components of flexible and integrated care as well; this is supported by the results of the comparison of component implementation during the process evaluation.

Group differences were mainly found only at M‑I, suggesting context variables that were difficult to capture. PsychCare coincided with the onset of the COVID-19 pandemic. The M‑I assessment was finished before the pandemic; however, 66% of M‑II assessments were conducted after the pandemic’s onset. Exploratory analyses suggested that FIT patients assessed after this time point reported a greater HRQoL decline relative to TAU than those assessed before the onset. Pandemic-related measures affected all aspects of healthcare, especially with day-care units often forced to close temporarily. As FIT hospitals shifted toward day and outpatient care, they had reduced inpatient capacity, potentially making them more vulnerable. Moreover, complex, interactive interventions may have been harder to adapt to crisis conditions or lost their distinctive character compared to routine care.

PsychCare was conducted mainly in non-academic hospitals, where research infrastructure is often limited. Limited resources because of burdening bureaucracy were reported to be a major reason for not taking part as a study center. Higher TAU rejection rates raise concerns about bias toward more motivated or progressive TAU sites, potentially inflating control group outcomes. Recruitment success varied by setting, with challenges especially in outpatient departments due to limited patient availability for additional study visits. Recruitment challenges are well-known in research within routine care and depend greatly on local structures, access, and the motivation of on-site recruiters. For the results of healthcare research to be representative, structural resources have to be supplemented in academic and non-academic settings. Participation in the M‑II measurement fell short of expectations, highlighting the challenge of conducting longitudinal studies in individuals with severe mental illness. With respect to the participation of caregivers, the two-stage inclusion process (patient recruitment, then caregiver recruitment) and the lack of close caregivers were reasons for the participation rate of only 19% and 10%, respectively.

Due to limited SHI participation, claims data linkage was only possible for a subset of participants, thus limiting the full potential of Module F. Nevertheless, high consent rates indicate that such linkage studies are feasible in psychiatric care.

Despite the aforementioned limitations and in line with the statements of our advisory board, we are confident that the data presented in this article should be considered in future healthcare planning.