The present paper describes the reach, dose (delivered and received) and fidelity of the PRO-FIT intervention, a combination of a web-based computer-tailored lifestyle advice (PRO-FIT*advice) and (face-to-face and telephone) counselling guided by MI. The results indicate that a representative proportion of the intended study sample agreed to participate of whom only half logged on at the PRO-FIT*advice website and completed at least one of the advice modules. Almost all participants received face-to-face counselling, however with low MI fidelity, and the majority of the planned number of telephone booster calls was delivered.
Despite its representativeness, only 34% of the people with FH invited to participate in the PRO-FIT project took part in the study. This low participation rate, as well as the StOEH screening rate, has implications for the generalizability of the results, as the sample was self-selective. Participants are likely to be more motivated to change lifestyle behaviour and our study showed significanty higher LDL-C levels in non-participants compared to participants. This is disappointing, since people with elevated LDL-C levels are most in need for a lifestyle intervention. In addition, because of the low participation rate, a decreased (cost-) effectiveness is expected on a population level [43, 44]. By conducting measurements and providing counseling sessions at the participant’s home, we already tried to minimize the main burden and time investments of the participants. However, in future comparable trials, other proactive strategies to recruit high-risk participants are suggested, such as the incorporation of healthcare professionals (e.g. medical specialists or StOEH genetic field workers) during the recruitment phase, and the provision of incentives for participation.
Despite the high dose of the PRO-FIT*advice accounts delivered, the extent to which participants actively engaged in using the website as intended was disappointing. The power of web-based interventions is that they can be delivered at almost any time and anywhere, as suites the individual participant . However, suboptimal exposure to web-based interventions has already been pointed out as a major concern in such health promotion studies . Apparently, dose received is a less controllable process element as compared to dose delivered, which is under the control of the implementers. Robroek et al also evaluated the use of an internet-delivered behaviour change program for construction workers and found 43% of them visiting the website . PRO-FIT*advice was based on the Dutch GezondLevenCheck, a quite comparable web-based tool which contains 5 (instead of 6) advice modules and is freely available to the general public and online registration before entering the advice modules is required. Comparable to PRO-FIT*advice, multiple visits to the GezondLevenCheck were possible and recommended, but not mandatory. Brouwer et al. reported a registration rate of 29% and found 91% of the registered users actually finishing at least one module . This confirms that, despite the potential of PRO-FIT*advice (or web-based interventions in general) to be delivered at a high dose, achieving an acceptable dose received remains challenging and less controllable. The length of the screening questionnaires of the advice modules could have inhibited participants from completing an advice module, particularly since they overlapped with the questionnaires for evaluative purposes. In future studies on computer-tailoring, the burden of filling in (screening) questionnaires should be brought to a minimum in order to keep participants motivated, e.g. by creating a joint questionnaire, for both evaluative and tailoring purposes. Thereby, it is known that incorporating iterative feedback and interactive website components are positively associated with exposure to web-based interventions . The combination of PRO-FIT*advice and personal counselling could be more successful if counsellor support is also available at an interactive communication board/forum, whereon participants also can communicate with each other. Still, the consequences of the low dose received of PRO-FIT*advice remain to be questioned, as the complete PRO-FIT intervention also incorporated face-to-face and telephone booster calls. In other words, to what extent were the gaps with regard to (un)completed advice modules and (lack of) formulated action plans, filled in by the content of the face-to-face counselling sessions?
Regarding face-to-face counselling, the dose delivered again appeared to be high, since almost all participants were visited by their personal coach. However, none of the analysed face-to-face counselling sessions met the MITI thresholds. Other studies on MI counselling have also reported below-threshold scores [48–51]. The association between MI fidelity and efficacy could not be tested in this study, but previous studies showed that a better MI performance is associated with larger intervention effects [21, 52]. It has often been reported that skills required for effective MI may take longer to develop than the 3-day MI workshop in our project [53, 54]. Probably, the provided MI workshop was not sufficient and more thorough monitoring and supervision of counselling skills during the intervention should have been built in. Beyond meeting MI thresholds, the face-to-face counselling sessions were part of the complete PRO-FIT intervention, and also included the discussion of the given advice at PRO-FIT*advice, and/or the (re)making of action plans. Thus, despite being a useful supplement to PRO-FIT*advice, this could have worked at the expense of fidelity to MI. Strict separation between the intervention components was impossible and undesirable.
The significant difference between the two coaches in MI fidelity, is noteworthy. By providing a 3-day workshop and an intervention protocol to both coaches, we attempted to achieve comparable delivery of MI throughout the sessions. Nevertheless, despite all effort, differences in background, demographics and other personal characteristics (e.g. counselling style) were unavoidable, and undoubtedly must have affected counselling performance. The analysed sessions showed that the coach with a more extended and diverse counselling history performed poorer than the coach with a more limited (though lifestyle counselling-) background. Literature has also shown that it has advantages to train more inexperienced coaches, e.g. students . Overall, we should keep in mind that in a real-life setting, differences in the above-mentioned inter-coach characteristics are indispensable.
The secondary aim of this paper was to investigate whether the dose of: A) PRO-FIT*advice, B) face-to-face counselling, C) telephone booster calls, and D) the complete intervention-package, was associated with change in lifestyle behaviour and LDL-C levels. The delivery of the complete intervention-package as intended led to non-significant improvements in LDL-C and lifestyle behaviours. More particular, associations between the completion of the separate advice modules of PRO-FIT*advice and change in LDL-C and related lifestyle behaviours were positive, but non-significant. Other studies also showed weak or absent dose–response relationships regarding web-based lifestyle interventions [56, 57]. Further, generally negative associations were found between the number of telephone booster calls and LDL-C and lifestyle behaviours, but these associations were also not statistically significant. Even if these negative associations are valid, this does not necessarily mean that the telephone booster sessions might have inhibited behavioural improvements. It may be that with fewer sessions performed, more improvements regarding lifestyle behaviours may already have been made and no further session were necessary, given that the participants were encouraged to plan the telephone sessions themselves according to their need for additional counselling.
This process evaluation has limitations. At first, the sample in this process evaluation (n = 181) might be too small to draw firm conclusions, since sample size calculations in the PRO-FIT project were based on the power to statistically detect an intervention effect . Further, associations of process indicators with demographic (e.g. age), psychosocial (e.g. motivation) and behavioural (e.g. physical activity level) correlates, that could further clarify for whom the intervention works best, were not included in this process evaluation. Also, not all recommended process elements were incorporated in this process evaluation, e.g. maintenance. In general, to produce lasting effects, interventions will need to address successful intervention components/strategies that lead to sustained behavioural change. We cannot draw conclusions on the longer-term effects of the PRO-FIT intervention and the association with intervention dose. Further, the assessment of MI fidelity was limited to 20 counselling sessions, which was sufficient for determining MI quality, but made it unable to explore its association with efficacy.
Strengths of the present process evaluation include that a thorough, theory-based approach was conducted incorporating the most important process indicators. Data were mostly collected from objective sources, such as website data/coach logs. By linking these indicators to efficacy, we meet the call for more insight in the association between the process of delivery of intervention components and efficacy, contributing to a more transparent evaluation of a public health intervention and being able to indicate facilitators and barriers in translating such an intervention into practice.