In the current study we evaluated the effectiveness of an innovative, theory-based lifestyle intervention carried out in a primary health care setting. To our knowledge, we are the first to report on the effects of a lifestyle intervention to reduce the estimated risk of developing T2DM and CVD mortality. The cognitive behavior program was provided in the participants’ own general practice, by practice nurses instead of researchers in the study. At the same time, we were able to monitor the intervention carefully by providing standardized training for the nurse practitioners and feedback on the counseling sessions by means of tape recordings. In contrast to the procedure in former lifestyle interventions [4, 5], the participants in our study were encouraged to find solutions instead of being told how to change their behavior, and they were also taught how to implement these solutions into their life. However, our findings show that the lifestyle intervention was not more effective than the provision of health brochures. Implementing lifestyle interventions in everyday practice poses challenging issues that require further investigation. A special focus in this regard may be the controllability of the dose of intervention in real-life settings, as this will probably be lower as compared to highly controlled settings. In addition, the role of participants’ social and physical environment may be of greater importance .
Earlier research in controlled settings has demonstrated that, separately, MI and PST are more effective than attention alone [35, 36], and there is evidence to support the efficacy of MI in a number of programs promoting change in lifestyle behavior . Although it has been convincingly demonstrated that T2DM can be delayed or prevented in high risk individuals, it is still a considerable challenge to provide evidence-based lifestyle programs for high risk populations in ‘real life’ settings. Few earlier randomized trials evaluated the effectiveness of diabetes prevention. These studies were, in line with our findings, not able to reproduce the very positive outcomes of previous efficacy trials [13, 14]. Most diabetes prevention studies that were carried out in primary care had insufficient power, used single-group designs, and/or had high rates of attrition, and should therefore be interpreted with caution [37, 38]. In the Hoorn Prevention Study we had sufficient power, randomization was performed at individual level, and relatively few participants were lost to follow-up. Other strengths include an appropriate design, choice of staff, staff training, adequate sample size and the choice for simple outcome measures. In addition, the participants did not receive financial incentives, and were recruited from the general population. Whereas the latter two arguments may be considered as strengths of the present study, they may also be associated with the lack of effectiveness of the intervention.
The findings of our process-evaluation indicate that the recruitment strategy was adequate and resulted in a reasonably high reach of the target population . Practice nurses were competent and confident to provide MI and PST, and participant satisfaction was high. Nevertheless, the amount of attended sessions was low, and almost no effects on determinants of behavioral change were seen . The rather low attendance rate may have contributed to the absence of an intervention effect, since a median number of 2 counseling sessions were attended. On the other hand, after per protocol analyses (which only included participants who had attended at least 4 counseling sessions) no change in the results were seen. Six or less face-to-face sessions as we provided may not have been enough to induce a sustainable lifestyle behavioral change, given that previous efficacious lifestyle interventions provided at least 12 sessions [40–42]. However, since the participants in our effectiveness study were not even motivated enough to attend 6 sessions, it is unlikely that they would be willing to attend more. It must also be mentioned that participants in our study were, on average, younger, and had a lower absolute risk of developing T2DM than those enrolled in previous effectiveness studies on lifestyle interventions. With regard to the external validity we like to point out that the study sample was not culturally diverse. Although our population (predominantly from a Western European culture) was representative for the study region in the Netherlands, this may affect the generalizability of findings.
Efforts were made to reduce barriers for participation to a minimum, as non-respondents are a potential threat to the external validity of the results . We chose to approach potential participants via correspondence at multiple moments, as described by Dillman et al. . Other efforts to reduce non-participation included the choice to provide the intervention in general practices (which are near to the homes of the participants), as well as to minimize the burden of the measurements by using short questionnaires, and to refrain from using unpleasant measurement methods such as 2-hour oral glucose tolerance tests.
After 6 and 12 months the lifestyle intervention was not more effective than the provision of health brochures in improving estimated risk scores for T2DM and CVD or lifestyle behavior in an at-risk population. Hence we conclude that the provision of this primary prevention approach was not effective in a Dutch ‘real life’ primary care setting.