Introduction
This study, a randomized controlled trial of the ‘Happiness Route,’ a positive psychology intervention, specifically focuses on a vulnerable group that is suffering from an accumulation of difficulties. Vulnerable adults are defined as individuals who are susceptible to harm [
1]. Their vulnerability is the result of the complex interaction between a lack of available resources and the challenges they have to face in their lives. Our target group has a disadvantaged socio-economic status (SES) as well as a lack of a social support network (leading to increased levels of loneliness), combined with personal limitations in the form of health problems. Each of these factors alone – a low SES, loneliness and health problems – is a risk factor for diminished well-being [
2‐
9]. In addition to the negative effects of these risk factors on the individual, these factors can also result in negative societal outcomes, such as the growing costs associated with loneliness [
10]. Low SES and loneliness can also lead to a greater risk of early mortality [
11] and increased morbidity [
12] and are associated with physical illness [
13] and mental disorders such as depression [
14]. Such an accumulation can precipitate a negative downward spiral, resulting in a serious loss of well-being and an exacerbation of symptoms of (mental) health complaints [
15,
16], consequently leading to more health care consumption and higher economic costs.
In this study, rather than treating symptoms, we examined an approach that is based on the principles of positive psychology directed towards improving the well-being of this group of vulnerable adults. The field of positive psychology has expanded traditional psychology by including and examining topics such as strengths, growth and well-being [
17]. The promotion of well-being is now widely recognized as a new goal in mental health care to complement the traditional focus on preventing and treating problems [
2,
18,
19]. Improved well-being has positive effects on health and personal functioning, resulting in health gains at both the individual and societal level [
20].
Several meta-analyses show that positive psychology interventions (PPIs) can improve emotional and psychological well-being [
21‐
23] . However, despite earlier findings that PPIs tend to yield better results in these populations [
20,
21], only a minority of intervention studies target groups with multiple health and psychosocial problems. Studies addressing clinical populations were all directed to groups with a specific problem, either a physical or mental disorder (e.g., cancer or depression), a problematic condition (e.g., loneliness), or to a specific age group (e.g., the elderly). Yet many people suffer from multiple problems and diseases making them especially vulnerable [
24]. Given this population’s high level of suffering and the resultant negative implications for society, PPIs need to be more widely used among groups with more complex vulnerabilities. Therefore, we evaluated a PPI for a group with a complex of problems spanning across their lifespan, rather than for a target group with a specific disease or condition.
One of the few happiness-based interventions that has been successfully implemented in practice in local communities in the Netherlands [
25] is called the ‘Happiness Route’ [
26]. To evaluate the intervention, this study uses a multicentre trial design and tests the intervention the way it is delivered in the everyday practice of social work, comparing it to an active control condition. One of the goals of our study, therefore, was to add valuable theoretically and practical information with regard to whether PPIs can improve well-being in people with great vulnerabilities and how the PPIs might do so.
The aims of this randomized controlled study were to examine (1) reach: the characteristics of the reached participants; (2) effectiveness: effects of the Happiness Route intervention in comparison to Customized Care regarding the (a) primary outcome measure of well-being and the (b) secondary outcome measures of resilience, purpose in life, depression, health-related quality of life, loneliness, social participation and health care costs; (3) treatment satisfaction of the participants, comparing the Happiness Route intervention to Customized Care; and (4) cost-effectiveness: to compare the cost-effectiveness of the Happiness Route with the control condition.
Discussion
No significant differences were found between the Happiness Route and Customized Care on primary and secondary outcomes. Both groups improved significantly in well-being, depression, and loneliness. The standardized effect sizes were somewhat higher for the Happiness Route than for Customized Care and the percentage of languishers decreased significantly from 32.8% at baseline to 15.8% at follow-up for the participants in the Happiness Route, but not in Customized Care. No effects were found for the other secondary outcomes: resilience, purpose in life and health-related quality of life. Participants adhered less often in the experimental than in the control condition. Regarding the level of satisfaction of participants, participants were very satisfied with the Happiness Route. They highly rated the relationship with their counsellor with a grade of 8.5. It is notable that none of the participants were negative about the intervention at 3 months and more than 90% were positive. More than half of them said that they had experienced positive effects both at 3 and 9 months. By addressing positive aspects of life such as values, strengths and passion in the intervention instead of focusing on people’s shortcomings, the happiness-based intervention was received more positively than the problem-based condition. Our cost effectiveness analysis shows that the Happiness Route may save considerable costs, but results in slightly less accrued QALYs compared to Customized Care.
There are at least three possible reasons why an interaction effect was not found. First, power was too low, as The power size calculated beforehand was not reached. Second, the differences between the experimental intervention and the control condition in their approach might have been smaller than expected. For example, some counsellors registered participants in the control condition for a social activity, such as activities in a senior club, or found the participant social support, such as via a buddy project. Furthermore, counsellors looked at the whole picture of care in the Customized Care condition, whereas in everyday practice, different care professionals often approach specific problems from more isolated areas of expertise. Third, the adherence-rate was much lower in the Happiness Route compared to the control condition. This could be explained by the fact that the Happiness Route was more intensive and asked more of the participants than the passive control condition. The Happiness Route required active participation in an activity, often after years of isolation and inactivity on the part of the participant. We have to realize that this could be an immense, perhaps frightening step for many people of this group. When they did persevere, they were very satisfied with the Happiness Route and the effects it had on them. This result corresponds to findings that engaging in a leisure activity that one is intrinsically interested in, such as sports or social activities, is linked to improved well-being [
57].
Although there were no differences between the conditions, we found effects for emotional and social well-being, depression, and loneliness over time, which might have been due to regression toward the mean [
58], as baseline scores were extremely low. On the other hand, scores for the MHC-SF have proven to be remarkably stable over a period of 9 months in a large panel, representative for the Dutch population [
59]. Also, it has been shown to be very difficult to improve loneliness with interventions [
60,
61]. Furthermore, only three measures improved over time, whereas resilience, purpose in life and health-related quality also started extremely low, but did not improve significantly over time. These are all strong arguments against the regression toward the mean explanation.
A meta-analysis on PPIs has shown that - even under more controlled conditions and often with non-active control conditions - effect sizes were small, with a standardized mean difference of .34 for subjective well-being, .20 for psychological well-being and .23 for depression [
22]. Except for psychological well-being, these findings are in line with the effect sizes that we found.
An explanation for higher satisfaction could be that people generally perceive well-being interventions as less stigmatizing than formal mental health services and, consequently, accept them more easily [
61]. Higher satisfaction indicates that a positive psychology approach could be a viable alternative, especially for structurally isolated, vulnerable people who demonstrate care-avoidant behaviour and tend to keep health care that targets their problems at a safe distance [
8]. This behaviour is in line with findings from a qualitative study amongst community-dwelling, lonely older people. For the majority of the group, primary care and community-based services addressing their loneliness were seen as neither desirable nor helpful. Yet they considered group-based activities with a shared interest preferable to other forms of support [
62]. Being able to offer a more attractive, acceptable and non-stigmatizing approach could help to care for groups that are normally difficult to reach within the formal (mental) health care system [
61]. Concerning future research, the intervention could be studied with other target groups at risk for a languishing condition that might be difficult to reach with formal mental health care, for example refugees.
While the cost effectiveness analysis showed that costs were saved in the Happiness Route, willingness to accept thresholds to interpret the results of disinvestment studies have not yet been defined in the Netherlands. However, the willingness to pay for an additional QALY lies between €10,000 and €90,000 and is dependent on the disease burden of the population under consideration. Using these threshold values, the estimated savings of €161,953 per QALY would suggest that the Happiness Route is an acceptable intervention. It should be noted, however, that a disparity exists between the willingness to pay and the willingness to accept; people are generally willing to pay a lower amount of money to acquire one QALY compared with the amount they would minimally want to receive before they are willing to forego 1 QALY [
62].
Although recruitment proved to be difficult, a very vulnerable group was reached, which makes our target group especially interesting. The group we included was even more vulnerable than expected based on the inclusion criteria. On average, participants were severely lonely, with a high comorbidity of diseases, serious depressive symptoms and low levels of well-being as compared to the Dutch population. In fact, participants scored almost twice as high as the group in the Dutch population with the highest average loneliness scores [
28]. Comparable to patients with moderate depressive symptoms [
57], the participants’ health-related quality of life was seriously impaired, with a score that was almost half as low as a representative sample of the Dutch population [
63]. Well-being had a low total mean score of 1.99 (SD 0.91) on the MHC-SF, which is more than a standard deviation below the mean of 2.98 (SD 0.85) for the normal Dutch population [
31]. While only 5% of the Dutch population languish, 31% of the participants were languishing at the start of the intervention [
64]. The yearly mean health costs they produced at baseline were around €8000. Indeed, their high level of suffering and care consumption show that they are an important group to pay attention to and study.
Limitations and recommendations
Our study had a couple of limitations that have to be kept in mind when interpreting the results. First of all, the study was underpowered. The study had under-recruited and could not reach the sample size that was calculated in the power analysis. Additional to the problems with inclusion, there was a high number of people who did not complete the intervention and who did not complete all questionnaires, both leading to a serious problem with power. Therefore, it is possible that not finding a difference between the experimental and control condition was due to insufficient power, rather than equivalence of the interventions. Accordingly, a possible reason that we did not find an interaction effect between time and group could be that the study was underpowered.
Furthermore, recruitment turned out to be very difficult. First, the ‘invisibility’ of the target group of lonely people could have hindered finding candidates. Being socially isolated and thus having no contact with the outside world makes these people very difficult to detect. As Machielse states, structurally socially isolated people are invisible to society [
16]. Second, the fact that the intervention was part of a study with a control condition could have played a role in the limited amount of applications. Intermediaries sometimes wanted to ‘protect’ their clients from the possible stress a research project might expose them to (e.g. filling in long questionnaires), as well as from the chance that they could be randomized to the control condition. Future studies working with intermediaries should be aware of the concerns they might have regarding research and be prepared to adequately and proactively respond to those concerns.
Only 12% of the participants that have been applied for participation had to be excluded, indicating that the intermediaries knew quite well who was qualified for the study. However, the extreme scores of the participants could be a sign that the intermediaries had a picture of the target group that was too extreme. For example, candidates were allowed to take part with a loneliness score of 3, while the mean score at baseline of the included participants was around 9 (with 11 as the highest possible score). These high loneliness scores indicate that the intermediaries seemed to have sought out people who were extremely lonely, and as such, had a distortedpicture of how vulnerable the target group had to be – a possible disadvantage when using sampling by referral. This misperception on the part of the intermediaries could have impeded recruitment, as 28% of the Dutch population is moderately lonely and only 4% is severely or extremely lonely [
28]. For future studies, we recommend emphasizing that moderate levels of loneliness also indicate that someone could be applied for the study. Although the real-life setting held limitations for recruitment, it also meant that the external validity of this study was good, as it took place in the field and the recruitment as well as the intervention were conducted as usually practiced.
It is interesting that, while there were no significant differences in effect between groups when measured quantitatively, participants from the Happiness Route, compared to their counterparts in Customized Care, were significantly more likely to describe experiencing positive effects when responding to the open questions, both directly after the intervention and 6 months after home visits had stopped. This disparity could mean that the questionnaire was less sensitive to change than the participants’ own perceptions. Nevertheless, this possible limitation demonstrates the added value of using open questions to validated questionnaires.
Moreover, we did not compare the amount of contact that was given between the two conditions. It is likely that participants of the Happiness Route received more visits than people in the control condition, and this was not taken into account in the analysis. The decision to not level the two conditions concerning the number of sessions was due to the fact that one important aspect of the Happiness Route is the program’s personalized fit to the individual’s needs. While one participant might need only two sessions, another might need twice as much time to discover their passion. We gave the counsellors the freedom to fit the number of sessions (within certain limits) to each individual participant.
Due to the small sample size, we could not conduct any moderator or mediator analysis, which would have been relevant to differentiate for whom the intervention works best. Therefore, we could also not take into account that in the control condition, all counsellors were professionals, while in the Happiness Route, counsellors were both professionals and volunteers. The reason for this difference was that we wanted to stay as close as possible to the traditional health care in the control condition, which meant using only professionals in that condition. We do not expect that there were large differences between the counselling provided by professionals and volunteers. The Happiness Route was a new intervention for both professionals and volunteers and they were trained in the exact same manner, thus both started at the same level. Furthermore, the volunteers often had many years of experience in the health care sector or even worked in the sector, but volunteered for this project.
A final challenge of this study was the difficulty in clearly differentiating the working ‘ingredients’ of the intervention. A future study could use different conditions, e.g., comparing the original intervention with a condition that only offers the rapportbuilding part, a condition where participants receive the money without counsellor support, and a condition where participants receive support, but no money. Furthermore, an analysis of treatment integrity, for example, based on audio-visual recordings, would allow for more in-depth insights into which parts of the intervention were actually used as intended.
Acknowledgements
We want to thank our team from Arcon, project leader Aad Francissen (AF) and administrative head Eddy Wezenberg, who initiated the research project and made substantial contributions to the conception, design. Special thanks to Aad who helped during the implementation. We are deeply thankful for the hard work and enthusiasm of the dedicated project leaders Marja Scheper, Chantal Bloemen and former project leader Lizzy Meijerink, Adri de Raaf, Johan Strik, Anoeska van Gorkum, Janet Dekker, Jan Bouwman, Carla Eefting and the founder of the Happiness Route, Gerard Noordkamp. We also want to thank all counsellors of ‘Zorg op Maat’ in both the Happiness Route and the control condition. We are thankful for the medical expertise that Irina Pologos so kindly provided us and for her help with categorizing the diseases. We are most grateful to our participants, without whom this study would not have been possible. We also want to thank our student assistant Pauline van den Hazel (PH), who proved to be invaluable, especially with the data collection. Finally, we are grateful for the language editing done by Catherine Lombard.
A version of this article is part of the first author’s PhD thesis, ‘Direction: Happiness. Improving well-being of vulnerable groups’.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.