The aim of the study was to examine the effectiveness of a health promotion intervention targeting PA and healthy eating in individuals with MD living in sheltered housing. The study period consisted of a 10-week intervention period followed by a 6-month follow up period.
From baseline to the end of the intervention period (at ten weeks), significant differences in changes in body weight, BMI, WC, and fat mass between the intervention and control group were observed. In the intervention group, a decrease in these outcomes was found, while they increased in the control group. For WC and fat mass, this intervention effect was independent of confounding variables. Within the intervention group, a larger decrease in weight, BMI, WC, and fat mass was found in those who completed the intervention than those who did not. From baseline to the end of the study, the decreases in the intervention group in the primary outcomes disappeared, except for the outcome of fat mass. From baseline to the end of the intervention period, a significant difference between the intervention and control group was observed for the pedometer-determined steps/day. In the intervention group, the mean number of daily steps increased, while it decreased in the control group. No other significant differences between the intervention and control group in changes in the secondary outcomes were found.
The baseline characteristics demonstrated the unhealthy lifestyle behaviour of the study population. For example, a mean BMI of 30 kg/m2 was found in our study population, compared with a mean BMI of 25.3 for the general population in Belgium . Smoking prevalence in the study population also greatly exceeded that of the general population in Belgium. Amongst the study population, about 60% were daily smokers, while in the general population it is about 21% .
It is well established that individuals with MD are at a greater risk of being overweight or obese than the general population [2, 3]. Important reasons for this high prevalence consist of the use of SGAs , lower PA levels [10, 11] and unhealthy eating habits [15, 17]. It is therefore promising that growing attention is being given to the importance of health promotion interventions targeting PA and healthy eating in this population. Several guidelines already emphasized the importance of PA and healthy eating [44–46] and the relevance of health promotion in mental health care is also acknowledged by the European Psychiatric Association .
The results of the study demonstrate that relatively small but significant reductions in body weight, BMI, WC, and fat mass are possible following a 10-week health promotion intervention targeting PA and healthy eating. Previous research has shown that weight loss through lifestyle interventions in individuals with MD is possible. The results of these studies must, however, interpreted cautiously due to their methodological limitations such as small sample sizes or the absence of a control group [30, 48]. The results of our study are more promising in those participants who completed the intervention. It is also important to note that 68% of the participants in the intervention group were taking a SGA. It is well established that these drugs are associated with weight gain in individuals with MD [8, 9].
The decreases in weight, BMI, WC, and fat mass in the intervention group disappeared in the period from the end of the intervention period to the end of the study. So, although emphasis on health promotion targeting PA and healthy eating during a certain period of time is beneficial, it probably needs to be continued. Therefore, the integration of health promotion activities, alongside other treatment aspects such as psychological and medication treatment, into the daily care of individuals with MD should be considered. Lifestyle interventions are essential in lowering the risks and morbidity associated with obesity and should be integrated into the daily care of individuals with MD [49, 50].
All mental health nurses involved in the study received the same training by one (the same) member of the research team. They received detailed instructions concerning the delivery of the group-based sessions as well as how to support the participants individually. The aim was to obtain as much consistency as possible concerning the information provided. We are, however, aware that the views and attitudes of individual nurses towards PA and healthy eating may to some extent have influenced the way in which they communicated with the participants.
The design of evaluation studies of public health interventions, like health promotion programmes, poses several problems and they require multiple, flexible, and community-driven strategies . Randomization at the individual level may cause contamination bias if individuals in the control group receive some aspects of the intervention by being in proximity to individuals in the intervention group [52, 53]. To avoid the risk of contamination bias arising from the fact that participants in the intervention and control group could be living together in the same SHO, it was decided to use a cluster design with the SHO as the unit of clustering. It was assumed that if SHOs were assigned to a non-preferred study arm, they could be disappointed and their interest in participating in the study could be reduced as a result [23, 54]. A substantial risk of non-participation on the SHO-level was also assumed based on the results of previous qualitative research which has identified lack of time due to the high workload in the daily care of patients with MD as a common barrier for mental health nurses to engage in health promotion programmes [55, 56]. For these reasons, a preference design appeared to be appropriate. As far as is known to the authors, this is the first trial examining the effectiveness of a health promotion intervention using a cluster preference RCT design.
Besides the significance of the results it is also important to consider their clinical relevance. According to the UK Department of Health , reductions in body weight of 5% or more are considered to greatly reduce the risks of physical health problems. At the end of the intervention period (at ten weeks), only 5.5% of the participants in the intervention group reached this target. Among those who completed the intervention, however, the figure jumps to 10.7% with a further 13% losing between four and five per cent of their body weight. To our opinion, integrating PA and healthy eating into the daily care of individuals with MD has the potential to increase the number of them losing at least 5% of their body weight.
The “Health promotion on well-balanced eating and healthy physical activity” program  served as the basis for that used in our study. The use of this programme appeared reasonable to us as its target population consists of the general population in the Flanders region (Belgium). Some adjustments were made to it to better meet the needs and interests of the population of individuals with MD included in our study. At this point, it is important to emphasize that, as far as is known to the authors, the efficacy of this general population programme has not been tested, resulting in no information on possible effect size. For this reason, the sample size calculation for our study was performed using a between group change of 3.5 kg found in a systematic review we performed . The mean between group change of 0.57 kg found in our study was yet to a large extent lower as the change of 3.5 kg. To our opinion, two possible explanations for the deviation between the weight loss of 3.5 kg found in our review and the amount of weight loss of 0.57 kg found in our study exist. First, the larger amount of weight loss found in the review may be explained by the fact that all but two of the fourteen studies reviewed consisted of an intervention duration in excess of the 10-week intervention period of our study (range: 2 – 12 months). Second, the studies included in the review consisted of a psycho-educational and/or behavioural intervention. In seven of them, this was combined with supervised exercise. Additionally, in some studies, the intervention also included restricted energy intake/ energy expenditure. This may be a second explanation for the larger amount of weight loss observed in the review. The health promotion programme assessed in our study did not include individualized energy restriction or energy expenditure goals such as low fat or low calorie diets. Based on the results of previous qualitative research , it was assumed that such an intervention would be too demanding for both the patients and the mental health nurses. Further research is required to examine the long-term effects of such an intervention (for example: providing the intervention twice a year, intervention with a longer duration).
The sample size calculation, identified that 371 individuals in each group were needed. This sample size was not reached as only 201 individuals in the intervention group and 83 in the control group agreed to participate. This number represents about only 20 per cent of the individuals living in these settings in the Flanders region, which places a limitation on the generalizability of the findings to the wider population of individuals living in sheltered housing. Compared with the most recent available data from the Federal Public Service of Health  on the sheltered housing population in the Flanders region, our study population had a higher proportion of women (39 vs. 30%), mean age was slightly lower (46.3±12.3 vs. 50.0±13.0 years), and it was more frequently diagnosed with mood disorders (25 vs. 16%). Data on the duration of stay in sheltered housing (4 vs. 4.2 years) and the proportion of individuals with schizophrenia (38 vs. 39%), substance misuse (16 vs. 18%), and personality disorders (14 vs. 12%) were comparable .
The study sample was characterized by high drop out rates. At the end of the study, 40% of the participants in both the intervention and control group were lost to follow-up. The main reason for dropping out was no further interest or motivation to participate. This is congruent with the results of previous research on barriers to individuals with MD engaging in health promotion activities which report lack of motivation and energy as a consequence of the MD and side effects of psychotropic drug use like sedation [60–62].
Another element of concern was the high number of individuals who did not fill out the various questionnaires at the end of the intervention period. This was related to the considerable drop-out rates from the study as a result of the lack of further interest and/or motivation to participate. For example, only 56% in the intervention group and 61% in the control group filled out the SF36 Health Survey questionnaire at the second measurement at ten weeks. Only 46 and 48 per cent of the participants in the intervention and control group respectively registered the number of daily steps during the second registration period (at ten weeks). For this reason, the promising results of the increase in steps/day from baseline to the end of the intervention period must be interpreted cautiously.
We are aware that omitting mixed model analysis is another limitation of the current study. The decision to omit mixed model analysis was based on the fact that a limited group-level variance compared to the total variance was present. For all primary outcomes, an ICC below 5% was found, indicative of a low level of variance at the level of the SHOs . Moreover, the number of participating SHOs and the number of individuals per SHO was below the minimum number of groups and individuals recommended for mixed model analysis . We nevertheless performed unadjusted mixed model analysis to examine whether the SHO clustering had an impact on the intervention effect for the primary outcomes from baseline to the end of the intervention at ten weeks. The significant differences between intervention and control group in changes in body weight (p=0.111) and BMI (p=0.109) disappeared, while they remained significant for WC (p=0.006) and fat mass (p=0.013) (data not shown). No ICC related to health promotion programmes with a cluster randomized controlled design in individuals with MD was found in the literature. To account for the degree of correlation within the several clusters in our study, the calculated sample size was multiplied with a design factor. As for the ICC, no design factor for the type of intervention assessed in our study was found in the literature. So, an assumption was made based on a design effect of 1.5 used in previous studies [32, 63].
The target population of our study comprised individuals with a wide variety of psychiatric diagnoses such as schizophrenia, mood disorders, and personality disorders. From a methodological point of view it may have been more suitable to focus only on individuals with a specific diagnosis, which would probably have lead to different results. However, it has already been well established that overweight and obesity affects individuals with MD irrespective of their specific psychiatric diagnosis [1, 64, 65]. In this sense, health promotion targeting PA and healthy eating appears to be important and desirable for individuals with MD independent of their diagnosis. In any case, sheltered housing is aimed at individuals with a wide variety of mental health problems. For these reasons, it was decided to include individuals irrespective of their diagnosis in the study population. Further research is nevertheless required to examine the effects of this kind of intervention in individuals with a specific psychiatric diagnosis.