This pragmatic lifestyle intervention study aims to reduce the risk of obesity among a selected risk group of preschool children. It is based on following components: intensified diet and physical activity counseling during pregnancy; advicing mothers to favour breastfeeding and mother and father to help their child to adopt healthy food preferences; advicing parents to encourage their children to be physically active, minimizing sedentary activity time, and also reminding parents about the role of an appropriate amount of sleep in weight development. The intervention is still ongoing and the final feasibility and effectiveness of this trial will be assessed at offspring age of six years, but we will report also earlier preliminary results.
Pragmatic trials are considered to have greater external validity than do explanatory studies, but instead the internal validity is considered to be lower . Our study was designed to be integrated in routine health care practice and to maximize the applicability of results to usual care setting. Practitioners who perform the intervention may vary in the way they deliver the planned counseling and the compliance of both practitioners and families may be variable. This results in weaker internal validity. For the internal validity randomization is considered the best way to select the trial participants also in pragmatic trials . However, conducting a randomized controlled trial is not always feasible, and the randomization process may reduce willingness to participate in the trial especially in lifestyle interventions, where problems in recruiting enough participants are usual. Case control study design is considered the second best design in intervention studies when randomization is not feasible, especially when the study groups are matched for the characteristics that may affect the result . Raaijmakers M et al. assessed randomized vs. non-randomized study with group matching design in their intervention and found that non-randomized appropriate matching resulted in 34% of their simulated trials in a more equally balanced distribution of their key characteristics compared with randomization, thus suggesting acceptability of non-randomized trials when the key characteristics are equally balanced in the groups . It has also been stated that even retrospective recruitment could be used in primary care settings when non-acute conditions are studied . In retrospective recruitment randomization is not possible and may make participant flow irregular, as it does in our study, too. The advantages of retrospective recruitment are that it shortens the recruitment time and reduces practical staff workload. Disadvantages of retrospective recruitment include weaker control of bias or other unknown factors influencing the effectiveness of the trial.
There are several methodological limitations in our study. We did not use randomizing and recruited the control group retrospectively in order to get a larger sample size for our trial planned to be performed in only one city and one risk group. The rationale behind this group selection was that it made the recruitment period shorter, the health centre staff’s workload lower in recruiting, and we were able to recruit larger study groups in this small city willing to participate this long-term intervention study. Randomization is not possible when the control group is retrospectively recruited, but because motivation to participate trials is a problem, especially when lifestyle intervention is involved, the randomization could have further reduced families’ willingness to participate in the study. However, we estimated that the groups would be comparable being families living in the same city and recruited from the same population. We also assumed that the one-year retrospective control group would not induce bias in the results since there were no major changes in municipal health care practices during that period. The retrospective control group could however cause bias, because the same PHNs who perform the intervention take care of the usual counselling practice of the control group in child health care clinics, but not in maternity clinics.
The strength of this study is that is designed to be a pragmatic trial integrated in health care practice thus having good prospect to be a sustainable part of municipal health care. The intervention costs are low since the existing clinical staff are the intervention practitioner. The follow-up time will be up to six years of offspring’s age thus providing fairly long-term follow-up. Our intervention targeted several lifestyle factors that are known to affect the child’s weight gain. Multifaceted intervention programmes are thought to be suitable for pragmatic trials and most effective in preventing overweight, since obesity is a result of many lifestyle factors in addition to genetic susceptibility. Moreover our intervention already starts during pregnancy and is planned to continue until the offspring is five years of age. This longer duration gives the intervention better chances of resulting in healthy weight gain. We are targeting mothers known to be at risk of having overweight or obese offspring. The positive intervention effect is more probable than in a trial targeting a population without specifically sought risk characteristics. We also offered the parents a chance to monitor their own weight, BP, WC and metabolic markers, which we thought would motivate the parents to continue in the study, and also to function as a public health promotive act by helping to find parents at risk of cardiovascular diseases.