Retired subjects participating in a one- year low intensity energy balance programme decreased their waist circumference, body weight, BMI, blood pressure and most other body composition indices and improved their physical activity and dietary behaviour. Although the changes were more consistent and more pronounced among subjects of the intervention group, the between-group-differences were small and mostly not statistically significant. Additional analyses among low educated men indicate that the programme may be effective in men with a low educational level: for waist circumference and fat intake the between-group-differences were significantly different. After the follow-up period the between-group-differences more or less remained the same, though the magnitude of the differences diminished.
We hypothesised that the intervention group would maintain their waist circumference and body weight in the two years following transition to retirement as opposed to the control group. Within the control group, the waist circumference would on average increase by 0.5 cm per year. Remarkably, in men, both groups reduced their body weight and weight circumference. And although the difference in change was -0.48, we could not demonstrate a significant effect. Possibly our study lacked statistical power. The sample size was calculated based on observational data on change in waist circumference in a middle-aged population. Such data were not available for the specific group of retirees we studied. Apparently, the variance that was used for the power calculation was too low. Further, despite the randomisation, the control group had on average higher, though non significant, scores for the outcome measurements at baseline, which may have caused regression to the mean. However, we included baseline values in the models and thus have allowed for these apparent differences.
The lack of effect may also be due to our recruitment strategy resulting in a relatively healthy and health conscious group of subjects. The study participants were selected from pre-retirement workshops, often attended by higher socio-economic groups, who in general are more motivated to change physical activity and diet, which might have reduced the added value of the prevention programme. Earlier studies have described that individuals willing to participate in health promoting intervention studies are already interested in diet and physical activity and are health conscious [30
Further, study participation itself may have led to increased awareness and motivation to change physical activity and/or diet in the control group and intervention group (Hawthorne effect), which also reduced the added value of the programme. The influence of the researchers or others involved in measurements is supposed to be very low, since information associated with the content of the intervention was not discussed at the physical examinations or during other contacts.
And last but not least, transition to occupational retirement per se may have induced the changes in (lifestyle) behaviours. The study by Nooyens et al
, showed that in transition to retirement subjects decrease work-related activity and increase household activities as well as doing odd jobs [9
]. Maybe the impact of retirement itself was so great resulting in either a ceiling effect or a small added value of the intervention programme.
A limitation of the study is the small number of females. Although the percentage of women that participated is representative for the percentage of women that worked in this age group, the number was too small to draw conclusions on the effectiveness of the intervention. And although the PASE questionnaire was originally designed for older adults (65+ years) it was chosen because it distinguishes activities (household -and leisure activities) that are relevant for retired people. Moreover, the recall period was only one week which also enabled us to pick up changes over a short period of time.
The programme of our study was developed according to the Intervention Mapping Protocol [15
]. This systematic process comprises a series of five steps for the development of health promotion programmes based on theory, empirical evidence, and additional research [7
]. Since information on determinants of weight gain for the specific group of retirees or middle-aged adults was lacking we used available information from general adult populations. It subsequently appeared to be a good choice, since Nooyens et al
showed that determinants of weight gain among older populations do not really differ from determinants in adult populations [9
Our programme aimed to induce relatively small and possibly sustainable changes in physical activity and diet to prevent weight gain [7
Moreover, the programme was developed in a way that it could be implemented nationwide, thus it was of low intensity, easily accessible and home-based. As a result, participants could voluntarily use the modules of the programme in accordance with their personal preferences. As a consequence commitment and adherence of the target group may have been too weak to result in a behaviour change.
Although the use of personal computers and internet in the middle-aged has increased rapidly in recent years [32
], it is unknown to what extent this is a suitable mode to deliver health messages to this age group, as can be concluded from the data on utilisation. Only 47% of the participants in the intervention group reported to have utilised both modules 2 and 3 (CD-ROMS with computer-tailored feedback), while the use of the Internet modules (study website and Weight Co@ch) was even lower: only 16% utilised Weight Co@ch. This indicates that the exposure to the full programme was rather low. Still, computer-tailored interventions have the potential to provide individualised behaviour change information to many individuals at low costs. This approach has been shown to be more effective than general nutrition information, especially for reduction of dietary fat intake [33
], although effect sizes are mostly small and apply only to the short and medium term (follow-up up to six months) [34
]. Clearly, the delivery of computer tailored interventions in real-life settings needs more research [35
The results of this study can by used by the Netherlands Heart Foundation and others to further improve the intervention modules. At present it is not clear if or how the results of this study will lead to further development or implementation of this intervention.