EoP was successful in recruiting its target population and compliance was high. EoP had a small positive effect on PA during leisure time (short-term and long-term) as well as on PA during household activity (long-term), but no effect was found on the total amount of PA. The effect of EoP on total PA, health status and healthcare use was not significant.
Although the target group was reached, the effect of EoP on PA, health status and healthcare use was very small. This is not in accordance with many other ERS programs. However, there is an important difference with other ERS programs in that our target group consisted of physical inactive women from a multi-ethnic population living in deprived neighborhoods. It is known that these women more frequently face complex socioeconomic problems like poverty, unemployment, social isolation and mental illness. Therefore, the limited effects that we found in our study could possibly be explained by these problems, as incorporation of PA in daily life may not have taken priority.
Moreover, compared to other ERS programs, the training frequency of only once a week was low. We are aware of the fact that the frequency of just one single training session a week is probably too low to expect changes in aerobic fitness, weight loss or waist circumference [24
], although results of previous studies on the association between intensity of the intervention and its effects are mixed [9
]. However, our expectation was, based on social cognitive models [25
], that EoP would induce an increase in total PA and that this increase together with the single weekly training session would cause an improvement in health status. Unfortunately, this was not the case. This is probably not only related to the socio-economic problems the target group faces, but also to the fact that 60% of these women were obese. Morgan et al. [8
] concluded in their review on evidence for exercise referral schemes, that these programs can improve PA levels at short-term in those who are overweight but not in those who are obese. A reason for that might be that the EoP program, like most other EoP programs, focused just on changing PA levels and not on healthy food patterns.
The EoP dropout rate was only 14%. Compared to other literature this is a very low number [12
], for example, Gidlow et al. [26
] reported in their review on attendance of ERS in the UK an adherence level of approximately 20%. EoP was developed to meet the environmental, economic and cultural needs of migrant women living in deprived neighborhoods in the Netherlands, as described by Schmidt et al. [16
]. Hence, women were referred by their GP, training sessions were held in their neighborhood in a supportive environment under the supervision of a female coach and financial incentive was available. These factors may be part of the success of the program in reaching its target group and in the adherence to the program. The low dropout rate may, however, also be an explanation for not finding substantial effects of EoP. Perhaps in other studies only the motivated participants who managed to incorporate a healthier lifestyle finished the program, which then led to positive effects. For motivation of the participants seems to be the key factor for the effectiveness this kind of programs [9
The recruitment of the intervention group was a challenge. Multiple factors which have been reported in the literature, such as the GP’s collaboration and the cooperation of a difficult target population were of influence [28
]. In particular in view of the insufficient recruitment of the participants by the GPs, we decided to leave our initial protocol and to make use of the natural patient flow of the EoP program for our intervention group. Due to the different selection processes, we observed differences in the health status of the intervention and the control group. More specifically, the health of the EoP group was worse than that of the control group: they had a higher prevalence of obesity, had a poorer mental well-being, a lower subjective health and a poorer fitness level. One of the explanations for this finding could be the selective referral of women by their GP (confounding by indication), perhaps only the women with obvious health problems were referred. Another explanation could be that women with severe health complaints who were referred to the control group refused to participate in the study. We do not expect these differences between the intervention and control group to have influenced the results, however, as these differences have been controlled for in the statistical analyses.
Another possible limitation might be related to the measurement of the primary and secondary outcomes. This includes the use of self-reported data on physical activity, and the use of a single item measurement of self-perceived health. Participants might have over- or underreported their actual level of physical activity, and the reliability of a single item measurement for self-perceived health might be limited. We do not expect these biases to be substantial, however, given that we found the results to be consistent over a broad range of outcome measures, measured with different types of instruments.