The IFIP trial included 15 psychiatric outpatient clinics from 12 CMHCs in five Health Trusts in South-Eastern Norway, which together covered a catchment area of 1.3 million inhabitants, including both urban and rural populations (27). Each of the 15 CMHC units represented one cluster, except two units that were merged, resulting in 14 clusters (45). A detailed account of the 15 participating CMHC units (27) and their randomization into 14 clusters have been published (40). The units of analysis in the current study were relatives.

The participating CMHC units were asked to include patients with a psychotic disorder together with their closest relative in pairs during the recruitment period from March 2019 to September 2020. Patients were invited to participate in the study, and those who agreed to participate provided written informed consent to participate and for the clinician to contact a close relative for participation. Subsequently, relatives were contacted and invited to participate, and consenting relatives provided written informed consent. At the study outset, each CMHC unit had between 28 and 217 patients with psychotic disorders, with 1392 patients in total. Patients and relatives were recruited in pairs to explore and compare independent and dependent variables from both respondents. Patient results and dyadic results will be published elsewhere.

The inclusion criteria for patients were: having an established or tentative psychotic disorder (F20–29 in ICD-10) certain enough to begin treatment; being 18 years or older. The diagnosis was made by local clinicians. The exclusion criteria for patients were: being a forensic client; not being competent to consent to participation; not having any relatives; having previously completed more than five sessions of FPE in single-family groups or more than ten sessions of FPE in multi-family groups or a similarly structured FI. Inclusion criteria for relatives were: being a relative of a patient with a diagnosis as described above; being 18 years or older.

The respondents were recruited independent of the decision to offer FI or other treatments. This means that those recruited did not have to undergo any specific treatment, intervention, or support during the follow-up period, in either the intervention or control arm. Similarly, there was no obligation for patients and relatives who received FI, to participate in the study. This separation of research from treatment had an ethical purpose of avoiding favoritism towards participants with better treatment. Additionally, there was an academic rationale: to examine the effectiveness of improved practices of family involvement in clinical units on a broader group of patients and relatives, not just those who received a specific intervention (39).

The intervention period started in February 2019 with a half-day “kick-off” about the significance of family involvement in all the intervention arm units, and subsequently a one week FPE course was arranged. During the intervention period the control arm units provided treatment as usual (TAU). Research activities for the IFIP trial are seen in Supplementary File 2 .

Preliminary mapping in the participating CMHC units showed that the proportion of patients who received basic family involvement practices was 20–40%, while for FPE it was 4% (27). The purpose of the IFIP intervention was to establish, improve and maintain basic family involvement practices and FPE as routine treatment, through a set of implementation interventions, and thereby improve outcomes for the patients and their relatives. The IFIP intervention was based on two levels of family involvement applicable for patients with psychotic disorders and their relatives in Norway: Basic family involvement practices for the health- and care services in general based on legal regulations, research evidence, ethical considerations, and discussions between key stakeholders and experts (37), FPE as first-line treatment in the national guidelines for persons with psychotic disorders. The latter is based on a systematic review and quality assessment of relevant literature, and professional, experience-based, and contextual considerations made by the parties involved in the work (24). The approach for the clinical interventions was that clinicians should provide FI consisting of basic family involvement practices to all patients and their relatives, and FPE to as many as possible. The background and development of the IFIP intervention and the chosen implementation strategies, are described and discussed in detail elsewhere (39, 40).

As background variables we collected a range of sociodemographic data from the patients and the relatives (e.g. age, sex, civil status) and the patients’ level of functioning measured with the Global Assessment of Functioning (GAF) scale (52). Relatives’ outcomes were assessed through self-reported questionnaires. Outcome of interest was defined as difference from baseline to 6 and 12 months follow-up. The rationale behind the choice of outcomes and full details of all measures are described in the IFIP study protocol (39).

Power analyses were performed for the satisfaction with health service support with CWS-B (53). Seven clusters in each arm and a sample size of 112 relatives per arm was needed to detect an increase of 0.5 SD from baseline to 12 months, with 80% power and 5% two-tailed significance level in a cluster-randomized trial with intra-cluster correlation coefficient (ICC) = 0.05 (45, 57).

For the first research question, the number and percentages of each type of family involvement and support services for relatives were calculated and compared between the intervention and the control arm using the chi-squared test. Some relatives had already received a certain level of services at baseline (in accordance with the inclusion criteria). This occurred because the recruitment of participants took place concurrently with the implementation of the clinical interventions ( Supplementary File 2 ). Additionally, various types of family involvement and support occurred independently of the IFIP trial (before, during, and after) in both arms due to the naturalistic study design. Because of this, the category “previous participation” was included in the analysis where this had been collected.

For the second research question, means, standard deviations, and number of observations for all scores were described. For the primary outcomes of interest, CWS-B for relatives, we estimated linear mixed effects models estimating the mean outcome in each treatment arm at 0, 6 and 12 months. The fixed part of the model consisted of intercept, treatment arm, dummy variables for time at 6 and 12 months, and interaction terms between treatment arm and dummy variables for time. The random part of the model consisted of random intercept for relative identification nested within center. If the ICC for center was negligible, the random part of the model was reduced to random intercept for relative identification. We included only relatives with at least one follow-up measurement (6 months, 12 months, or both). Missing data were handled by multiple imputation using predictive mean matching with 30 imputed data sets (58). The imputation model included age and sex in addition to the treatment arm, center, and any previous scores. For selected secondary outcomes, we applied the same model as for the primary outcome.

The results were reported as regression coefficients, standard errors, 95% confidence intervals (CI) and p-values. We further reported the estimated differences in change with time between the two arms together with 95% CI, effect sizes (Cohen’s d) and p-values. For the primary outcome we illustrated graphically the estimated outcomes at each time point with 95% CI. All tests were two-sided with a significance level of 0.05, and residual diagnostics were conducted by assessing the residuals graphically. No adjustment for multiple testing was implemented, and we present the crude p-values for the reader to interpret. In all the mixed effects model analyses, we used R version 3.4.4 (lmer-functions).

The models were rerun on a subset of the data with complete outcomes at all three time points (complete case analysis). The regression coefficients from the complete case analysis (data not shown) were in similar magnitude and the same direction as the results from the main analysis with relatives with at least one follow-up presented above.

Baseline characteristics of the participants who were lost to follow-up, and therefore excluded from the analysis, were compared to the included relatives (with at least one follow-up) with respect to age and sex, ECI positive and negative subscale and CarerQoL-VAS. For relatives in the control arm, there were more males in the subgroup with no follow-ups (N = 10, prop males 60%), compared to those with at least one follow-up (N = 86, prop males 24%), (Fisher’s exact,p = 0.03). Relatives in the intervention arm with no follow-up were more happy (N=10, CarerQol-VAS 7.80 (SD = 1.90)) than those with at least one follow-up (N = 124, CarerQoL-VAS 6.48 (SD = 1.75)), (Wilcox rank sum, W = 362, p = 0.03).

In order to measure the effectiveness of the clinical interventions directly we compared the following subgroups: 1. Relatives in the intervention arm receiving FPE the last 12 months after inclusion (N=29), 2. Relatives in the control arm not receiving FPE the last 12 months before inclusion (N=94). The linear mixed effects models for the subgroup analysis were adjusted for the patient’s duration of psychotic disorder and GAF-F and for the relative’s civil status, if they had higher education, and if they were living with the patient.

In total 261 patients (150 intervention; 111 control) and 239 relatives (139 intervention; 100 control) from the 14 randomized clusters consented for the trial. This paper presents the analysis from the relatives in 231 patient/relative pairs (135 intervention; 96 control) ( Figure 1 . Consort diagram).

Sociodemographic characteristics for patients and relatives, and the patients’ level of functioning and symptoms measured with GAF are presented in Table 1 .

Before the COVID-19 pandemic lockdown, 26% of the relatives had answered the 6-months follow-up questionnaire. The remaining data for the 6- and 12-months follow-up were collected during the pandemic.

The proportion of patient/relative pairs in the intervention arm who had received FPE from the CMHC units at baseline was 17.2% (23/134), compared with 3% (3/96) in the control arm, and 37.9% (50/132) and 7% (6/86) respectively at 12 months follow-up. This increase in the intervention arm was significantly higher than the increase in the control arm, (p < 0.05). After the pandemic outbreak, only six new FPE groups in the intervention arm and two in the control arm were started.

Table 2 shows the proportion of relatives who had received various types of family involvement and support services. The proportion of relatives who received at least one CMHC consultation “without the patient” had a larger increase from “previous” at baseline to the last 24 months at 12 months follow up in the intervention arm (from 27.6% to 64.2%) compared to the control arm (from 28.1% to 45.7%), (p < 0.05). At 24 months follow-up, the proportion of CMHC consultations “without the patient” and consultations “with patient” both were significantly higher in the intervention arm (64.2% and 72%) than in the control arm (45.7% and 53%), (p-values < 0.05).

Figure 2 shows that the proportion of relatives who had participated in at least one CMHC consultation in the intervention arm was highest at 6 months, before it was reduced at 12 months. The level of “other services” decreased in the intervention arm and increased in the control arm from baseline to 6 months.

In the intervention arm 79.8% (95/119) of the relatives had participated in a CMHC consultation during the one-year follow up, compared to 62.0% (49/79) in the control arm, (p = .007). The proportion of relatives who had participated in at least one CMHC consultation, at all time points (baseline, 6 and 12 months) did not differ significantly between the intervention arm (28.7%, 35/122) and the control arm (25.6%, 22/86).

Information and/or involvement of relatives in the patients’ crisis plan increased from 38.8% (52/134) at baseline to 55.3% (73/132) at 12 month in the intervention arm, and from 37.5% (36/96) to 38.8% (33/85) in the control arm.

The relatives in the intervention arm received an increased amount of FI (both FPE and BFIS), including involvement in the patients’ crisis plan, and experienced decreased patient dependency compared to the control arm. There was also a tendency towards increased satisfaction with health service support. Altogether, the IFIP intervention, despite being substantially affected by the pandemic, influenced the clinicians’ family involvement practices sufficiently to have a beneficial effect on relatives’ outcomes and received FI.

The proportion of patient/relative pairs receiving FPE increased considerably in the intervention compared to the control arm, consistent with the findings in our fidelity outcomes study (40). Our findings demonstrate that systematically offering FPE in clinical practice increases the proportion of patients and relatives receiving family involvement in line with recommended treatment for psychotic disorders. This proportion is comparable to the OPUS-study and the TIPS Project of FPE to patients with first-episode psychosis and their relatives (59, 60).

A significantly higher proportion of the relatives in the intervention compared to the control arm had at least one CMHC consultation during the follow-up (80% vs 62%). Most likely, the increase in CMHC consultations in the intervention arm is largely linked to providing BFIS, based on the changes between the arms from baseline to 6 months, which corresponds to the period when the clinical interventions were well underway. This proportion is similar to the study from the OnTrackNY Early Psychosis Services: patients were consistently encouraged to involve their relatives in the treatment, and 73% to 84% of relatives at each measurement point had at least one family consultation (29). The increased proportion of relatives involved in crisis planning as part of the FI, may have contributed to promote the patients’ mental health and reduce the relatives’ stress related to relapse (61).

The COVID-19 pandemic significantly disrupted the delivery of health services. In the intervention units, BFIS and FPE was put on hold, reduced, modified (e.g. digital meetings rather than in-person), terminated earlier than planned, or not initiated at all. Due to the pandemic it was difficult to ensure that the clinical interventions were adherent to the original design in content and quantity, and subsequently to improve FI as intended. The low number of new FPE groups reported after the pandemic outbreak, was also documented in the fidelity study (40). The pandemic led to a sharp decline in CMHC consultations at 12 months in both arms, but more so in the control units. The IFIP intervention focused on better routines, more family-friendly culture, and increased competence and understanding of the significance of family involvement among clinicians (62). At the time of the pandemic outbreak, FI was a more integrated part of the intervention units’ clinical practice and organization, which made them better equipped than the control units to maintain contact with the relatives.

Relatives in the intervention group tended to be more satisfied with support from the health services, mostly in terms of information and advice, and with support from staff. The positive tendencies of support are consistent with implementation studies of FPE in regular clinical practice (43, 63) and with qualitative data published from our study (64). In the latter study clinicians in the intervention arm described that through FPE alliance sessions, they could offer adequate information, guidance and support, helping the relatives to cope with the situation and to support the patient. The clinicians also experienced that family involvement contributed to shared understanding, acknowledgment and mutual dialogue between relatives, patients and clinicians (64). The significance of FI lies in the potential to alleviate the relatives’ sense of isolation in shouldering responsibility for the patient’s situation. Additionally, these interventions play a crucial role in ensuring that the relatives’ efforts are acknowledged by staff and address any negative experiences they may have had with mental health services, as described in previous studies (35, 36, 65).

The level of relatives’ experienced patient dependency was significantly lower in the intervention arm. Dependency is understood as the state in which a person requires the help of others in order to perform daily activities (66). The level of dependency of the care recipient has been shown to be associated with the severity of the illness, and for being the most important predictor of relatives’ burden (29). In FI, the patient’s personal goals for recovery serve as guidance for support and problem solving, through an equal partnership between the patient, relatives and clinician. A strengthened collaboration may have relieved the relatives’ experience of being alone with the responsibility and burden of the patient’s decreased functioning and thereby resulted in a reduced perception of patient dependency. This adds to earlier studies showing strong correlations between family burden and professional support (67, 68).

Both contextual factors and factors related to the complex (and pioneering) study design, may have weakened the difference in exposure to FI before and after baseline in the intervention arm, as well as the difference in exposure between the intervention and control arm during the intervention period. In sum, it is likely that this sub-study underreports the effectiveness of the IFIP intervention on relatives’ outcomes such as the degree of satisfaction with health service support.

The response rates at the follow-ups were notably high ( Figure 1 ), especially considering the life situations of the included relatives. Another strength lies in the results from the subgroup analyses, consistently reinforcing the main findings concerning relatives’ positive outcomes. Moreover, the naturalistic design contributed to knowledge that is relevant to ordinary clinical practice, which is highly needed in mental health services (73). An important implementation strategy was the “whole ward approach”, to increase the probability that all patients with psychotic disorders and their relatives were offered some kind of systematic FI. Both BFIS and FPE were provided by ordinary clinical staff, as one of several strategies to enhance sustainment of high-quality FI after the project period. Finally, the IFIP trial is a unique large-scale project to implement basic family involvement and FPE, with fidelity measurements (27, 40), qualitative data (62, 64, 74, 75), and health register data which complements the quantitative data from relatives presented in this paper and strengthens our understanding of FI.

Family involvement is consistently recommended in treatment guidelines worldwide for individuals with a psychotic disorder, which should entail that relatives receive continuous information, guidance, and support from mental health services (30). Therefore, it is crucial that all CMHC units ensure routines, knowledge, and competence among clinicians to systematically offer FI tailored to the needs of patients and their families. The IFIP-trial, with this sub-study on relatives’ outcomes and the fidelity outcome study on the cluster-level, has demonstrated the possibility of enhancing FI on a large scale with positive outcomes for relatives. The forthcoming patient outcome study (in process) aligns with the results in the current study. Thus, it is possible and feasible for CMHCs to implement and integrate FI for patients with psychotic disorders and their families in an evidence-based, safe, and effective manner. Our approach to implement FI seems to reduce relatives’ needs for other support and more ad hoc services. A systematic implementation of FI may overall be the most sustainable and cost-efficient approach. Future research and publications will examine the costs and sustainability associated with implementing the IFIP intervention in the mental health services.