The aim of this study was to evaluate how the Family Talk Intervention (FTI) was implemented over time from the perspective of HSWs in their everyday clinical practice among families with a severely ill parent or child in need of palliative care.

This study is a part of a larger project that was carried out between the years 2021–2024 in which FTI is evaluated with an effectiveness-implementation hybrid design, type 2 (39). The current study used the Normalization Process Theory (NPT) as the basis for the evaluation of the implementation of FTI. NPT characterizes and explains key mechanisms that promote and inhibit the implementation, embedding, and integration of complex interventions in healthcare (28, 29).

The study has a longitudinal design based on quantitative data collected over a year from 19 HSWs who were educated and trained in using FTI. In addition to the HSWs, the study participants also included two nurses. However, for readability purposes, all participants (21) will be referred to as HSWs throughout the manuscript.

The HSWs all worked with families affected by severe illness in four different healthcare contexts in an urban area in Sweden: (1) Specialised palliative home care for adults, where the HSWs were included in multi-professional teams that offered 24 h care. Patients in these settings had diagnoses such as cancer, cardiovascular diseases, and chronic obstructive pulmonary diseases. (2) A university hospital with cancer care units for adults with different types of cancer, both emergency care and outpatient care. (3) A pediatric hospital, with both emergency care and outpatient care (all pediatric units, e.g., neurology, oncology, haematology, intensive care). (4) A children's hospice for pediatric palliative care.

FTI, also called Beardslee's family intervention, is a psychosocial family-based intervention, built on systems theory and originally developed in the context of psychiatry for families where a parent has an affective disorder (40, 41). FTI is manual-based and involves all family members. The main goals are to facilitate family communication about illness-related subjects, support parenting, increase knowledge about the illness, and emphasize the children's needs (42). For this study, the original FTI-manual was modified for the care contexts in focus (23, 43). Resilience is a key goal of FTI, since the intervention aims to help the family to identify their strengths and protective factors to help them cope with the current situation (40). FTI involves six meetings (meetings 1–6), including various family members, (with extra meetings 7–11 if needed), led by a HSW (or other healthcare professional) educated and trained in FTI (44), (Table 1). The FTI meetings are held at intervals of 1–2 weeks (45). FTI has an eclectic approach that includes psycho-educative, narrative, and dialogical ways of working (42). The psycho-educative element is focused on raising awareness about illness-related problems, such as children as next-of-kin or prognosis. FTI also includes the family's own story: individual family members share their perspectives to create a common story, the narrative element. Further, FTI has a dialogical perspective: highlighting difficult situations by emphasizing the children's voices and sharing and seeing the different perspectives of all family members.

The implementation strategy to support the integration of FTI was multifaced, including educational outreach visits, education, facilitation, clinician implementation meetings, and audit and feedback. The components of the multifaced implementation strategy are presented in detail below and further described through the ERIC taxonomy (Expert Recommendations for Implementing Change) (36), (Figure 1).

After their FTI-education, the HSWs in various clinical settings invited families with a parent or child with a severe illness to participate in FTI. The families were included based on the definitions of palliative care for adults (47), and children (48), respectively. Further inclusion criteria were that families in adult care should have at least one child/youth aged between 0 and 24 years, and that families in pediatric care should have a child between 0 and 18 years. All families needed to be able to communicate in Swedish. During the implementation period, HSWs delivered FTI to 70–80 families affected by severe illness. FTI could be delivered in physical or virtual meetings with the families.

Managers in the clinical settings included consented to their units being part of the FTI- project. According to the principles of the Declaration of Helsinki (49), HSWs were informed (verbally and in writing) about the current implementation study and were asked to provide informed consent to participate. In total, 43 HSWs with permanent employment in the four different healthcare contexts consented to participate in the current study after they had taken part in FTI-education and training during 2021–2022.

To follow the implementation of FTI over time, HSWs were asked to answer web-based questionnaires at three time points: directly after the FTI-education (time point 1, (T1), after 6 months (T2), and after 12 months (T3). The questionnaires were sent to the HSWs' professional e-mail addresses with a link to the questionnaire in the web-based program Webropol. Reminders were sent if they did not respond within three weeks. Part of the questionnaire was used in the current study, which was based on the Swedish Normalization Process Theory Measure (S-NoMAD) (Table 2). The web-based questionnaires also included questions about the implementation process, such as the organizational context, as well as background characteristics, such as age, sex, and workplace.

The S-NOMAD (Normalization Measure Development) instrument is based on The Normalization Process Theory (NPT), translated and adjusted to Swedish conditions (28, 29, 50). The concept of normalization is central to NPT and the theory defines and clarifies factors that influence the implementation of new interventions into healthcare settings by focusing on how individual healthcare professionals and teams work to make new interventions become embedded and integrated into routine work processes (29, 51). The NPT is used within complex interventions to assess the implementation process from the perspective of the involved individuals. It is based on four constructs: Coherence (the sense-making work done to understand the new practice), Cognitive Participation (the relational work done to engage people in the new practice), Collective Action (the operational work done to enact the new practice), and Reflexive Monitoring (the appraisal work done do to assess the new practice (29, 52), (Table 2).

S-NoMAD consists of 23 questions. The first three are general questions about the intervention that can be answered on a 10-point numeric rating scale, ranging from “not at all” (0) to “completely” (10). These questions are usually regarded as single questions (1). Further, the instrument includes 20 detailed questions about the implementation of the intervention divided over the four constructs described above (Table 2). These questions are answered on a 5-point Likert scale ranging from “agree completely” (1) to “do not agree at all” (5), there is also the option “not relevant” (6). The constructs could be summed and used as subscales, but the questions could also be considered as single questions. All the questions were worded in a positive way except for question 13 (FTI disrupts working relationships), which was negatively worded. This question was consequently reversed and recoded before analysis. In the present study, Cronbach's alpha's values for the four constructs were 0.46 for coherence, 0.79 for cognitive participation, 0.43 for collective action, and 0.75 for reflexive monitoring. As S-NoMAD is relatively new, and hardly used in the Swedish healthcare context (50), it is used in its original form with the result both from the constructs and single questions.

Descriptive statistics were used to describe the sample's characteristics and study variables. Non-ordered categorical data were presented as frequencies and percentages, ordered categories as median (Mdn) and percentiles (P₂₅–P₇₅), and continuous data as means and standard deviations (SD). Since the study was based on repeated measures, only HSWs with complete data from all three time points were included. To minimize the risk of a biased sample, comparisons of the baseline characteristics were made between the HSWs who completed all three assessments and those who did not, using independent sample t-test and chi-square test.

Friedman's test was used to evaluate HSWs perceptions of the implementation of FTI (i.e., S-NoMAD). The tests were conducted both on the single questions and the constructs. Wilcoxon signed-rank test with Bonferroni corrected p-values were used as post-hoc test.

Due to the small sample size, and thereby an increased risk of type II errors, both Bonferroni corrected, and non-corrected results are presented. Kendall's W was used as an effect size measure: 0.1– < 0.3 small effect, 0.3 – < 0.5 moderate effect, and ≥0.5 large effect. The level of statistical significance was set at 5% (p < 0.05). All analyses were conducted in the SPSS (53).

In total, 21 HSWs completed S-NoMAD at all three time points (see Table 3). Their mean age was 48 years (min-max: 27–65 years) and all but one were women. They had worked as a HSW for an average of 10 years (min-max: 2–25 years). No significant differences were found between HSWs who completed all measurements (n = 21) and those who did not (n = 19). Reasons for dropping out of the study were mainly due to HSWs ending their employment; other reasons are illustrated in Figure 2. There were also several changes in management at some of the clinics during the research project.

The results of Friedman's test regarding the three general questions in S-NoMAD demonstrated that for the HSWs, the FTI intervention became more familiar and normalized over time (Table 4). A significant change was found for the first general question in S-NoMAD where HSWs reported more familiarity with using FTI over time (p = 0.013). The uncorrected post-hoc tests demonstrated that the change was significant both between T1 and T3 and between T2 and T3; however, the significant difference between T2 and T3 was lost in the Bonferroni correction. A significant positive change over time was also seen regarding the HSWs reports about FTI as a normal part of their work (p = 0.025). Both corrected and uncorrected post-hoc tests showed that a significant change occurred between T2 and T3, where HSWs median score increased from 4 to 6 on a scale between 0 and 10. No significant change was observed for the question of whether HSWs believed that FTI will become a normal part of their work. However, HSWs rated the question highly from T1–T3 as the median score was stable and high, being 7 at all three time points.

The ratings of the construct coherence in S-NoMAD were stable with a median score of 2 (agree) for all three time points, indicating that the HSWs attitude to the differentiation, communal specification, individual specification, and internalization of FTI was positive, with no significant changes. The median score across the questions in the construct (N4–N7) were 1 (agree completely) or 2 (agree) and did no change between the three time points.

The questions in the construct (N4–N7) ranged between 1 and 2 (agree completely- agree) for all questions at all three time points.

A similar positive attitude was shown for cognitive participation (initiation, enrolment, legitimation, and activation) with median scores of 2 (agree) for all three time points.

The HSWs ratings of the construct collective action (interactional workability, relational integration, skill-set workability, and contextual integration) of FTI had ratings or median 2 (agree) at all three time points. However, in the individual questions (N12–N18) lower scores were given regarding their ability to integrate FTI into their daily work (N12), but also to the resources provided (N17), sufficient training (N16), and support from their management (N18). The median scores for these questions ranged between 2 and 3 (agree-neither agree nor disagree). Over the three time points, lower scores were given regarding provided resources (N17) and support from their managers (N18). A tendency for significant changes was seen over time for question N16, sufficient training (p = 0.052), indicating a greater variation in the ratings.

HSW ratings for the construct reflexive monitoring (systematization, communal appraisal, individual appraisal, reconfiguration) were very stable with median scores of 2 (agree) for all time points. Similarly, median scores for all questions (N19–N23) at all time points were also 2.

The importance of implementing evidence-based ways of working into HSWs practice has been raised earlier (62, 63). This study has a unique strength as it involves the implementation process of a family-based psychosocial intervention delivered by HSWs among families living with severe illness, as earlier intervention studies have mainly focused on intervention effectiveness (14, 15). The current intervention, FTI, has been delivered in real-life conditions to vulnerable populations. The study includes a wide representation of HSWs with different working experience from various healthcare settings. In addition, since data were collected longitudinally over 16 months, more all-encompassing results can be presented from the implementation process than if the design had been cross-sectional. The used sample represented various working units, both adult and children's care, implying that the results from the study might be applicable to similar psychosocial interventions in several healthcare contexts.

Limitations of the study include its small sample size and the risk of type-II errors. The original inclusion of 43 HSWs was reduced to 21 in the final sample, mainly due to the HSWs changing workplace. A high level of job resignations was observed over time, contributing to the high attrition. This reflects the difficulties that arise when conducting implementation studies within clinical practice, where this kind of dropout can occur. The challenges of staff turnover during the implementation of new intervention has been raised before (64). Nevertheless, no differences regarding background characteristics were found between the initially included HSWs and the sample used for this study. As the study population is homogenous, e.g., similar educational background, a small sample size may be adequate (65). In longitudinal studies, sample sizes commonly decrease over time, with attrition being more common with long time intervals between the time points (65, 66). This reflects the difficulties that arise when conducting implementation studies within clinical practice, where this kind of dropout can occur. In longitudinal studies, sample sizes commonly decrease over time, with attrition being more common with long time intervals between the time points.

Implementation fidelity refers to what degree an intervention is delivered as it was intended (67), in this case to what degree the HSWs delivered FTI according to the intervention description. Within the current FTI-study, discussions were held between the external facilitators and HSWs regularly regarding how FTI should be delivered (56). Fidelity was also a central topic in the supervision sessions with the key-facilitator where adherence to the manual was discussed. Even if several strategies for fidelity were conducted in this study, we cannot present evidence that complete fidelity was the case. However, flexibility is often necessary to fit the specific context, and it does not mean that the fidelity of core components of the innovation is threatened (68).

The Cronbach´s alpha values for the constructs in S-NoMAD ranged from 0.43 to 0.79, indicating a varied internal consistency. This differs from the Elf, et al. (50) Swedish study (50) using S-NoMAD in health care among health care professionals (not HSWs), which showed higher alpha values. However, since the constructs include few questions, this could influence the test and result in lower alpha values, without necessarily meaning that the internal consistency is low (69). This sample is also relatively small, which might also be one reason for low alpha values on two constructs in S-NoMAD (Coherence and Collective action). The Collective action construct included a question that was negatively worded (FTI disrupts working relationship) unlike the other questions which could also incorporate lower internal consistency (70).