The sample included families of residents in three residential homes for older people. The inclusion criterion was having a member living in one of the residential homes for older people where the intervention was conducted. The exclusion criterion was an inability to speak or read Swedish. Four families at each unit (12 families in total) were enrolled in the intervention. None of the residents participated in the FamHcs because of their condition or illness. Two families who participated in the FamHCs declined to participate in the follow-up interviews due to health-problems and lack of time, respectively. Consequently, 10 families comprising 22 family members were included in the study. The participating family members (20 women and 2 men) were 39–84 years old (Median = 55).

Prior to the intervention, three nurses from three residential homes for older people participated together in session related to FamHC. The education was a tailored and shortened version of a regular university course in FamHC [39]. The nurses learned about the theoretical basis of family-system nursing and practiced FamHC in role-playing exercises.

The theoretical assumptions of FamHC comprise a number of core components [20]. Reciprocity, humility, and respect for the families and their situation, their beliefs, and ideas are central to the intervention. A non-hierarchical approach between the nurse and the family is also essential in building a trusting relationship. A non-hierarchical approach is characterized by reciprocity, where the nurses and each family member are acknowledged as equally valid [27]. The intervention consisted of families participating in a series of three 1-hour conversations about every other week. Two nurses led each conversation, one nurse who had participated in the education and one nurse from the research group previously trained in FamHCs. The family members and the nurses created the conversational content together with the aim of finding alternative solutions to problems described by the families. During the first conversation, all family members were invited to tell their stories and encouraged to listen to each other’s. The second conversation was intended to focus on the problems and suffering identified in the first conversation. The third conversation focused on family strengths and resources for the future [27]. Two weeks after the last conversation, a closing letter was sent to all the family members. The letter provided the nurses’ reflections over the three conversations, acknowledging the families’ suffering and highlighting their resources [40].

Semi-structured group interviews were conducted six months after the FamHCs intervention to collect qualitative data from each family [41]. Individual interviews were used in the two families that had only one participant. The interviews concerned families’ experiences of and reflections on participating in the FamHCs and followed an interview guide to ensure that questions relevant to the study aim were asked [42]. The family members were asked whether and how anything had changed in the family after their participation in the conversations and how the conversations had affected them. The interview started with an open question: “Can you tell me about your experiences of participating in the Family Health Conversations?” The interview continued with follow-up questions “Can you tell me whether or not the conversations have been helpful to you in your family, and if so, how?” “Can you tell me whether or not the conversations have contributed to a change in your family, and if so, how?” “Did you experience the conversation as positive, and if so, why / what / how?” “Did you experience the conversations as negative, and if so, why/what/how?” Probing questions, such as “Who?”, “When?” and “What do you mean?” were asked when necessary. The interviews, performed by one researchers who had not been involved in the intervention, took place in a conversation room in the particular residential home and lasted about 60 minutes. All interviews were tape-recorded and transcribed verbatim.

The interviews were analysed using qualitative content analysis [43] performed in several steps. First, all transcripts were read through several times in order to gain a sense of the whole [44]. The text was then reread and divided into meaning units, each representing a single unit of content [45]. The meaning units were then condensed and labelled with codes. Through comparing the codes for similarities and differences, two categories with seven related subcategories were abstracted. The analysis was an interprative ongoing process, in which the authors reflected until agreement was reached.

The Family Hardiness Index (FHI) [46] was used to measure the intervention’s effect on family stress resistance and adaptation resources. The instrument consists of 20 statements scored on a 4-point Likert-type scale. In this study, the version consisting of the 3 subscales Commitment (the family´s ability to work together and the confidence in handling problems), Challenge (the family´s approach and attitude towards new experiences), and Control (the sense of being in control of family life) was used. A total score was calculated, ranging from 0 to 60, in which a higher score reflected higher family hardiness. The Swedish version of the FHI used in the study indicates good internal consistency (α = 0.86) for a newly published instrument [47].

The Swedish Health-Related Quality of Life Survey (SWED-QUAL) was used to measure the effect the intervention had on the families’ health-related quality of life. The instrument consists of 61 items calculated on 12 subscales reflecting different aspects of health-related quality of life. The response options are of the Likert type and offer four or five response alternatives. The scores are transformed into a 0–100 score index, with higher scores reflecting better functioning and well-being. The instrument shows good internal consistency (α = 0.78) for all subscales [48].

Comparisons of SWED-QUAL and FHI results from baseline to 6-month follow-up post intervention were made using Cohen’s d to calculate effect size. According to Cohen [49], benchmarks for effect size are 0.2, 0.5, and 0.8, respectively, for small, medium, and large effect sizes. A value > 0.25 was considered clinically significant [cf.50]. Clinical significance may be defined as a difference large enough to have an impact on a person's health [51]. Wilcoxon signed-rank were conducted to determine statistically significances between baseline and follow-up. A p-value <0.05 was considered statistically significant.

To integrate the quantitative and qualitative results we used triangulation as a methodological metaphor as described by Erzberger and Kelle [37] and exemplified by Östlund et al. [38]. This method links theoretical assumptions to empirical findings and assigns the different data sources equal weights. The sides of the triangle represent relationships between the theoretical assumptions and the empirical findings originating from qualitative and quantitative data. Results that are found to be convergent, complementary, or divergent are reflected by different types of triangle [38].

Permission to conduct the study was given by the head of administration and the heads of the units. Written and verbal information concerning the study aims, voluntary participation, and confidentiality were given to the participants in accordance with research ethics. Consent to carry out and record the interview was sought from the families before they were included. The Regional Ethical Review Board approved the study (2011-335-31M).

The findings revealed two categories, Improved family functioning and Enhanced emotional well-being, reflecting the responses of the families after their participation in the FamHCs.

The results of the families’ assessments showed tendencies of improvement concerning the subscale negative affect at follow-up. The result also showed a tendency that pain were scored lower and sleep problems were scored higher at follow-up compared to baseline, however, not to a significant extent. When calculating for effect size, the result showed a positive clinical significant effect in commitment, control, and total family hardiness at follow-up. Emotional well-being, negative affect, and sleep problems also proved to be showed a clinical significant effect, while pain showed a medium worsening effect at follow-up compared to baseline (Table 1). These results indicate that the intervention improved the families control over their situation and their mental health but led to more pain among the participant.

The qualitative and quantitative results were then integrated with the theoretical assumption. This integration, using the methodological metaphor of the triangle, of the empirical results and theoretical assumptions is presented in this section and illustrated in Fig. (1). In this study, we have interpreted the results from the quantitative and qualitative findings to be convergent, i.e. pointing to the same conclusion as the outcomes and responses demonstrating improvements in family functioning and health. When deductively testing the theoretical assumption and asking about families responses after participating in FamHC, the theoretical assumption were aligned with empirical results.

Fig. (1). Triangulation diagram of the logical relationships between the theoretical assumption, the quantitative data from SWED-QUAL and FHI, and qualitative findings from the interviews.

The theoretical assumption suggests that FamHCs will promote healing and sustain family health. This is supported by the quantitative findings. An improvement of hardiness was seen in the FHI total score and the subscale measuring commitment (families’ internal strengths and ability to work together) from baseline to 6-month follow-up. The result from SWED-QUAL showed increased emotional well-being in family members and decreased negative affect, negative affect include feelings of anxiety, sadness, nervousness, and tension, and these were decreased after participating in the FamHCs.

The qualitative results also support the theoretical assumption. The participants described improved cooperation, communication, and relationships within the family. They also said their families had become better able to share responsibilities. On a more personal level, participants described feeling stronger after the conversations and better able to process their feelings about having a family member living in a residential home. Anxiety and frustration were also relieved, which made it easier for them to deal with their situation. They felt their consciences were eased, and that led to their improved emotional well-being and created a change in their beliefs about their own insufficiency and guilt.

The part of the theoretical assumption that suggests that FamHC creates a context of change and supports the creation of new beliefs, new meaning, and new opportunities to manage problems was also supported by both the qualitative findings and the quantitative data. The families described how they had changed their ways of looking upon the situation and had begun to think in new ways. They had changed how they communicated with each other and had learned about how other family members viewed the situation. Furthermore, they saw problems from a different perspective, which made it possible to think differently about how to deal with their situation and continue with their lives. The quantitative data showed a clinical significant effect on the families’ abilities to take control of their family lives rather than being shaped by outside circumstances or events.

In this study, we used mixed methods research design, based on integrating qualitative and quantitative data with the intervention’s theoretical assumptions. The methodological metaphor of triangulation helped to describe the logical relations between the qualitative and quantitative findings on the empirical level and the explicitly stated assumptions on the theoretical level. A mixed methods research is suitable to investigate effects in complex interventions since findings from one method strengthen from the other method and a combination of the two methods help explain the findings and neither method alone will answer the study object [62].

The sample size was 22 family members, which can be assumed as a sufficient number for the qualitative analysis since the interviews were rich in content and gave a good variation in their responses. However, this is a small sample when considering the quantitative analysis and we therefore used a non-parametric test since there can be questions concerning the normality of the data. As statistical significance is just a statement about likelihood of findings being due to chance and the p value on its own provides no information about the meaning of the results to clinical practice or the overall importance [63], in the study design, a measure to establish clinical significance were chosen. Thus, Cohen’s d >0.25 was considered as a benchmark for the intervention is clinically significant, as proposed by Wolf [50]. We consistently used validated instruments in the quantitative part of the study. To strengthen the trustworthiness of the qualitative data, the person performing the interviews with the families had not participated in the conversations with the families. To strengthen the credibility of the qualitative data all authors discussed each step together in the analysis process until a consensus was reached. All authors reflected independently upon the findings to ensure that nothing were overlooked.

When including families none of the family members living at the residential homes were included due to illness, fatigue, frailty and communication difficulties caused by dementia, which could have caused confusion to the residents if participated. Moreover, there was a predominance of female family members which may have affected the results. Even though the sample size was small, the choice of using mix-methods design integrating qualitative and quantitative findings was seen as a strength in this study. We achieved a more comprehensive in-depth information about the responses to and effects of the FamHC on the family members. Even so, although we believe that our results are transferable, we recommend that generalizations to other populations should be made cautiously.

From the integration of the empirical results and our theoretical assumptions it can be concluded that participating in FamHCs facilitates a context for change and supports the creation of new beliefs, new meanings, and new opportunities in relation to the problems described by families with a member living in a residential home for older people. The family members’ abilities to manage the transition of moving their sick member from home to assisted living improved with the intervention. This study shows the importance of nurses caring not only for the resident but also for the whole family. Therefore, FamHCs should be offered as a part of standard care to support family health. For nurses it is essential to expand the approach from patient-centred care to an approach in which the family is viewed as a unit of care, as in FSN. Family Systems Nursing interventions such as FamHC should be considered as a natural responsibility for nurses to support family well-being. Consequently, opportunities for education in training for and conducting this kind of family intervention should be considered, as there is a further need for knowledge transfer in this area from theory to practice. Even if this study adds to the evidence base of FSN interventions in general and FamHCs in particular more empirical research is still needed to strengthen the research evidence in this specific context.