The Scandinavian Solutions for Wellness study - a two-arm observational study on the effectiveness of lifestyle intervention on subjective well-being and weight among persons with psychiatric disorders

Weight gain and consequent obesity is a serious and growing health problem, associated with an increased risk of a number of somatic illnesses, e.g., type II diabetes and cardiovascular disorders and with a reduction in quality of life [1‐3]. Persons with schizophrenia are even more prone to weight gain and obesity than the general population; possible reasons for this include change in caloric intake and physical inactivity [4]. In addition to this, weight gain is an adverse effect associated with many antipsychotics, antidepressants and mood stabilizers [4‐8].

Several studies have investigated whether behavioral interventions can be effective in curbing the weight gain induced by psychotropic medication [9‐24]. The intervention programs evaluated in these studies ranged from comprehensive group programs with motivational counseling, physical exercises and teaching healthy eating or with cognitive therapy focusing on cognition in relation to food and body weight to educational mailings about the importance of diet and exercise. The designs of the studies were diverse, encompassing single-arm observational studies, studies with matched controls and randomized controlled trials. In the recent years, a few reviews of studies of behavioural weight management in persons with serious mental illness have been made [25‐28]. These reviews concurrently conclude that it is possible to manage weight gain in persons with serious mental illness by means of behavioral interventions.

Only a few of the studies mentioned above took the next step and included evaluation of the impact of behavioural weight management on the quality of life (QoL) or subjective well-being of the participants. In one randomized trial, a one-on-one nutrition education program was found to increase QoL and reduce weight gain [18]. Another randomized trial evaluating an individual program focusing on diet and exercise found that the persons attending the program reduced the weight more than the controls, but no difference was found between the groups for QoL [10]. A third randomized trial evaluated a group program comprising diet and physical activity [24]. This program was also found to reduce weight, but QoL was not improved. A single-arm study found decreased weight and improved QoL after a group program comprising teaching healthy lifestyles as well as physical exercise [9].

The Solutions for Wellness (SfW) educational program in Scandinavia (Denmark, Norway and Sweden) is a well-established 12-week group program for persons who suffer from psychiatric disorders, use psychotropic medications and have weight problems. The program focuses on working towards a healthier lifestyle, with two main elements: healthier diet and increased physical activity. The primary objective of the observational study presented here was to investigate the impact of the Scandinavian SfW educational program on the development of subjective well-being and QoL as measured by changes in the Subjective Well-Being under Neuroleptics (SWN) scale. The population studied was suffering from chronic psychiatric illness and treated by any psychotropic medication. Secondary objectives included to assess if potential changes in subjective well-being and QoL during the program were maintained at a follow-up 3 months after the program ended, to describe the association between participants' characteristics at baseline and SWN changes during the program and 3 months later, and finally to describe development in weight, BMI and waist circumference after completion of the program and 3 months later, as well as potential associations with baseline variables.

The SWN total score response rates for the period between baseline and 3 months were not found to be significantly different for participants who participated in the SfW program and for participants in the control group, neither in the primary analysis nor after controlling for differences in baseline variables. However, for one of the SWN subscales, emotional regulation, OR for response was almost 2 for SfW participants compared with controls in the sensitivity analysis. The content of the emotional regulation subscale is hope for the future, intensity of emotions, interest in what is happening and confidence. Participation in regular group activities for the SfW group participants may have had a positive impact on these aspects. For the other four SWN subscales SfW participation was not associated with a higher chance for improvement of the score.

The only clinical factor associated consistently with improvement in SWN scores was the SWN score at baseline (lower subjective well-being giving a higher probability of improvement). Greater age of the person and shorter duration of illness both were associated with improved subjective well-being on two dimensions each at 3 months: age on physical functioning and social integration and duration of illness on self-control and physical functioning. Diagnosis and symptom severity at baseline were not found to play any role here.

The choice of the short-form of the SWN scale in this study was governed by the history of the scale as it was developed and validated in a clinical trial measuring the subjective well-being of persons with schizophrenia [30], and also because this scale was already part of the SfW program in Scandinavia and the feedback from the SfW instructors about the ability of the participants to fill in the scale was positive.

For the SfW group the mean change from baseline to Month 3 in the SWN total score was +3.3 (95% CI: 1.9; 4.7) and from 3 months to 6 months the mean change was +1.4 (95% CI: 0; 2.7). These changes on SWN score are much smaller than what was reported in another study [34]. That double-blind randomized clinical trial enrolled 114 persons with schizophrenia who changed from other antipsychotics to olanzapine or clozapine. Of the 99 persons who had a SWN score post-baseline, mean improvements in SWN score at 26 weeks (+SD) were found to be 11.3 ± 20.7 points for olanzapine and 8.2 ± 15.8 points for clozapine.

Reasons for observing these differences in change in SWN score may be multiple. Part of the explanation of the much smaller change in SWN score in this study may be that only persons in psychiatric day care or outpatient clinics and inpatients in non-acute wards were included, i.e., relatively clinically stable patients. Only a total of 118 persons of the sample (32%) were markedly ill or more severely affected as measured on the CGI-S scale, compared with the clinical trial where the majority (84% to 93%) was at least markedly ill. Probably, the focus on stable persons in this study makes the evaluation of the impact of participation in SfW on subjective well-being more valid as the effect of other interventions as well as clinical fluctuations over time are smaller for this group than for more acute patients. Another aspect is that the SWN scale, as mentioned, was developed for persons with schizophrenia, and in this study, unlike the trial mentioned above, 170 persons (46%) had another psychiatric disorder than schizophrenia, which means that this scale may not have captured the improvement of all almost half of the persons optimally. Still, the change in QoL as measured with 15D was modest too and not significantly different between the two groups either.

In several important respects, the SfW group differed from the control group, as it had a significantly higher baseline weight and BMI and lower QoL (15D) and subjective well-being (SWN). Furthermore, the SfW group was more severely ill (CGI-S) and contained almost twice as many persons with "other" diagnoses compared to the control group (39.2% vs. 23.7%). The fact that antidepressants were about twice as common in the SfW group (39.6% vs. 20.3%) indicates that depressed persons were substantially more common in the SfW group. Persons with depression may have more pronounced difficulties finding the motivation necessary for any change in lifestyle, both regarding diet changes or physical exercise habits, which has possibly contributed to the modest change in SWN score in the SfW group. Most of the above mentioned differences between the two groups were controlled for in the logistic regression analyses (weight, BMI, SWN and CGI-S at baseline), but as diagnoses were not collected for the persons who did not have either schizophrenia or bipolar disorder (i.e. they were categorized as "other") it was not possible to adjust fully for the differences in diagnoses between SfW participants and controls. The reason for the way diagnoses were registered in the study (schizophrenia/bipolar disorder/other) is that the SfW program was started with persons with schizophrenia in mind, and it was expected that almost all participants would have schizophrenia (which was also the reason for choosing the SWN scale as an outcome measure). However, this study has shown that the program is now used for a much more diverse population. The differences in diagnoses between the two groups is a limitation of this study (due to its non-randomized design), and likewise, there may also be other variables that influenced the outcome, which were not equally distributed between the SfW participants and the controls, and which were not measured, and which could therefore not be controlled for in the statistical analysis. Examples of such variables could be: previous weight gain, eating disorders and treatment changes during or after the program. These differences between the two groups, both the ones that are known now and the ones that may exist, but were not investigated as we wanted to keep the study simple, illustrate the main limitation of using a non-randomized design, which is the difficulty of assessing the validity of differences found, or not found, between the groups in terms of outcome of the interventions.

As described, the control group consisted of persons eligible for entering the SfW program but treated at hospital departments and outpatient clinics which did not run the SfW program. This design was chosen in order to ensure a clear separation between the SfW group and the control group in order to be able to investigate the relationship between participation in SfW and any observed change. However, whereas the treatment of the psychiatric disorders must be assumed to be comparable in centers running the SfW program and centers not running the SfW program, it was out of our control if the centers collecting participants for the control group had other programs or activities with a similar effect on the well-being as the SfW program. In theory, a better design could be to run a cluster-randomized clinical trial, randomizing departments and clinics to program participation or not, but at the time when it was decided to run this study the SfW program was already established in many settings in Scandinavia.

Alternative designs would have been to recruit persons for the SfW group and for the control group in the same departments and clinics, with or without randomization to the two groups, so that the participants in the two groups would be exposed to the same treatment of psychiatric disorders and the same "other activities", but this was found suboptimal as the care of the participants in the control group would be influenced by the fact that the program was run by the department, as the change in knowledge and attitudes induced among the caregivers through working with the program would have an impact on all patients. Still, having decided to recruit SfW patients and control patients in different departments, it could have been possible to explore and describe the differences between SfW centers and the control centers in terms of what "other activities" they were running in order to try to evaluate the possible impact of these, and using matched controls instead of controls who "only" were eligible for participation in the program could have reduced the differences between the groups and improved the validity of the results.

The country where the study participants were enrolled was significantly associated with the probability of response for both weight and waist circumference at both time points: participants in Denmark had better chances to stabilize their weight at Month 3, and having participated in the study in Denmark or Sweden was associated with a better outcome for weight at Month 6 and for waist circumference at both time points. The results presented in this publication are mean results achieved across centers with different ways of implementing the SfW program and working with lifestyle intervention in general. The differences regarding weight and waist circumference control between the three countries suggest that different approaches were developed at a country level. The way the program was implemented was not registered as part of the study, but differences between centers were noticed: some centers focused on diet, others on physical activity, some taught mainly theory, whereas others used a more active and practical approach with for example outdoor activities. The SfW as an educational program is focusing on knowledge and attitudes. But it is not increased knowledge that changes the weight, but improved lifestyle habits. A more practical approach, perhaps leading to a larger degree of change of behavior among the participants, might have generated a larger mean weight loss. The differences in results between centers with different approaches have not been analyzed in this piece of work, but would be an interesting next step for a research project.

A number of other studies evaluated the impact on weight of behavioural programs more or less similar to the Scandinavian SfW program, and the results from these studies are generally in line with the findings from the study reported here with regards to development during program participation [9‐28]. Some of these studies also followed the participants after program cessation to evaluate the persistence of the effect on weight [9, 18, 20, 21], and these studies found, like the one reported here, that the effect of the program seemed to last in the period from the end of the program and up to the end of follow-up a few months later. It is interesting to see how the results from this study and many other studies match in this respect, and it is important to acknowledge that even though the weight loss during SfW program participation was not large, 0.5 kg, the persons were referred to the SfW program because of gaining weight or having weight problems, and while participating in the program 38% of the persons lost weight and 36% stayed stable. These, findings, all pointing in the same direction, contrast with the results on subjective well-being and QoL. Only a few studies included these outcome measures, and among three randomized controlled trials only one found a positive impact of the intervention on QoL [10, 18, 24]. However, all these studies had small sample sizes, and low statistical power may have contributed to the lack of finding of positive changes in QoL in the other two studies.

The SfW group was found to improve significantly from baseline to Month 3 regarding the severity of the psychiatric disorder as measured with CGI-S score, whereas the control group did not change. When interpreting this finding it is relevant to take in consideration that the study was an observational, open-label study, and that bias may play a role for this difference between the groups. Some centers made CGI-S ratings at clinical conferences, where staff not working with SfW was also present. This is a good practice in order to reduce the bias introduced when investigators rate their own patients and through that the quality of their own work. This practice was, however, not implemented in all centers.

The statistical methods used illustrate the exploratory nature of the study. No clinically meaningful difference has been defined for the primary outcome measure, the SWN scale, and it was decided to use a difference in the proportion of participants with a SWN improvement between the two groups of 70% - 50% or larger for the sample size calculation. In order to reach a power of 80%, with a two-sided type I error of 0.05, 300 persons were enrolled in the treatment group and 55 persons in the control group. Uneven treatment groups were chosen as the widespread implementation of the SfW program would have made recruitment of more SfW-naïve departments and clinics difficult.

In conclusion, the Solutions for Wellness program was not shown to have a positive impact on the subjective well-being of the participants relative to the control group, as measured with the SWN scale, but Solution for Wellness program participation was associated with maintaining weight and waist circumference and the program may allow health care professionals to help persons under psychotropic treatment to control their weight.