A total of 135 subjects were invited to participate through the collaboration of various schools in the town of Cáceres (Spain). The inclusion criteria were: (i) a body mass index (BMI) equal to or greater than the 97th percentile for the age and sex of the subject; (ii) age between 8 and 11 years as defined by Spanish population curves (Hernández et al., 1988). Subjects were excluded if they were: (i) regularly practising PA, or following an exercise program or some other therapy (n=65); (ii) involved in any weight control program (n=18); (iii) were taking any medication (n=8); (iv) had any type of dysfunction limiting their physical activity (n=2); other reasons (n=9). The final sample comprised 27 Caucasian sedentary children (10.4±1.0 years). They were divided into two groups: the G1 group who followed a multi-sports exercise program (n=11: 8 boys and 3 girls), and the G2 group who followed a combination of two programs–the above exercise program plus a low calorie diet (n=16: 10 boys and 6 girls). Table 1 presents the characteristics of the G1 and G2 groups. All the children's parents completed a prior informed consent form. The study was approved by the Bioethics and Biosecurity Committee of the Universidad of Extremadura (Spain) and respected the principles of the Declaration of Helsinki.

Nutrition was assessed with a self-reported 3-day food record (2 weekdays and 1 weekend day in succession, i.e., Thursday, Friday, Saturday) filled in by the parents. The weight of the food was estimated from the parents’ records. A computerized database [deleted for blinding purposes] was used to calculate the daily intake [deleted reference for blinding purposes], and the program logged the mean of the three days (kcal/day).

Daily PA was measured before the intervention, during the follow-up, and at the end of detraining, using a validated uniaxial accelerometer (Caltrac), and covering a 3-day period (Thursday, Friday, and Saturday) except during bathing and swimming. All participants were instructed to record the amount of time spent cycling or swimming during the evaluation period. At the beginning and the end of the day, the subjects recorded the number of “motion counts” of the accelerometer, following previously published protocols (Sallis, Buono, Roby, Carlson, & Nelson, 1990). The final Caltrac score was recorded, as also was the average of the three days (motion counts per day).

The accelerometer used (Hemokinetics, Madison, WI, USA) was programmed to function as a physical activity monitor (Sallis et al., 1990). It contains a piezoelectric bender element which assesses the intensity of movement in the vertical plane. Its validity has been demonstrated as a method for estimating energy expenditure in children (Maliszewski, Freedson, Ebbeling, Crussemeyer, & Kastango, 1991), and it has been used in other studies (Raudsepp & Päll, 2006; Sallis et al., 1990). Although it does not record such activities as rowing or swimming, no activities of these kinds were used in the subjects’ daily physical activity for the duration of the study.

Pubertal stage was evaluated by a trained pædiatrician according to pubic hair development using the Tanner classification criteria (Tanner, Whitehouse, & Takaishi, 1966). The kinanthropometric measurements followed the ISAK protocol (Norton et al., 1996): body height, body weight, body fat mass, and fat free mass (bio-impedance). Standard equipment was used: a scale-mounted stadiometer (Seca, Berlin, Germany), a weight scale (Seca, Berlin, Germany), and a bio-impedance analyser (Bodystat 1500, Bodystat Ltd, Douglas, Isle of Man, UK). BMI was calculated as weight divided by height squared (kg/m2), and the status ponderal (BMI z-score) were determined [deleted reference for blinding purposes].

The participants’ motivation to engage in exercise was measured using the 19-item Behavioural Regulation in Exercise Questionnaire-2 (BREQ-2; Markland & Tobin, 2004) in its form validated for the Spanish context (Moreno, Gimeno, & Camacho, 2007). The BREQ-2 consists of 19 items divided into 5 subscales (1=not true for me, 5=very true for me), one measuring amotivation (four items; e.g., “I think exercising is a waste of time’), and four measuring forms of regulation of exercise behaviour: external (four items; e.g., “I am physically active because I feel under pressure from my friends/family to exercise”), introjected (three items; e.g., “I feel ashamed when I miss an exercise session”), identified (four items; e.g., “I am physically active because it's important to me to exercise regularly”), and intrinsic (four items; e.g., “I am physically active because I enjoy my exercise sessions”). The test-retest Cronbach's α reliability coefficients for this Spanish version of the BREQ-2 subscales range from .81 to .89 (Moreno et al., 2007), and the questionnaire has also been specifically validated for use in obese pædiatric populations (Verloigne et al., 2011).

The present intervention it is a quasi-experimental study (Hartley, 2012; Ramos-Álvarez, Moreno-Fernández, Valdés-Conroy, & Catena, 2008). All the variables satisfied the tests of homoskedasticity (Levene variance homogeneity test) and normality (Kolmogorov-Smirnov test) of their distributions. The basic descriptive statistics (means and standard deviations) were calculated. A repeated-measure ANOVA was used to compare the interaction between the different groups (G1 and G2 group) and the different evaluations (baseline, 3rd-year, and detraining), and another ANOVA was applied to compare the differences between the three evaluations. The Tukey post hoc test was used to compare means. Cohen's categories were used for the magnitudes of the effect size (ES): small if 0 ≤ |d| ≤ .2; medium if .2<|d| ≤ .5; and large if |d|>.5 (Cohen, 1988). Obese prevalence was compared using Pearson's χ2. The level of significance for all statistical tests was set at p ≤ .05. All calculations were performed using SPSS (version 16.0).

There were reductions in amotivation from the baseline in both groups: in the 3rd year in the G1 (ES=-1.52) and the G2 group (ES=-3.15), and in detraining in the G1 (ES=-1.96) and in the G2 group (ES=-3.10). These results contrast with a recent study which found no change over time in amotivation after a 10-month multi-component exercise and diet program based on SDT (Verloigne et al., 2011). This might indicate that long-term intervention is required to generate changes in this motivational parameter. Similarly, it has been shown that residential treatment programs might have put pressure on the children to become physically active (Markland & Ingledew, 2007), which could have led to there being no decrease in amotivation in the medium term. Thus, one study concludes that, although obese adolescents may perceive losing weight as an important benefit of being physically active (“feeling or looking better”, and “improving health and physical condition”), this might not encourage continued participation in physical activity programs (Deforche et al., 2006). Probably, as the results of the present study suggest, more time is needed for obese children to perceive exercise and balanced diet as habits of a healthy lifestyle (Reinehr, Kleber, Lass, & Toschke, 2010).

Similarly, changes were observed in intrinsic regulation in both groups: in the 3rd year of the intervention in the G1 (ES=2.78) and in the G2 group (ES=2.04), and in detraining in the G1 (ES=2.70) and in the G2 group (ES=3.00). Unlike the case of the previous parameter (amotivation) however, the present study confirms the findings of the aforementioned multi-component program of exercise and diet in the medium term (Verloigne et al., 2011). It seems that this combined procedure is effective at positively changing obese subjects’ intrinsic motivation in the long term. Similar results were observed with interventions of physical exercise alone. According to SDT, long-term participation is likely to be sustained by intrinsic, but not by extrinsic, motives (Deci & Ryan, 1985). This is reflected in the maintenance of this intrinsic motivation in the detraining period, with an increase in physical activity of the subjects compared to their baseline (Table 2). These findings could be explained by a lower perceived barrier to physical activity (Deforche et al., 2011). After several years of multi-sport exercise, the subjects have a more positive attitude, possibly practising some physical activity at which they feel competent (Craig, Goldberg, & Dietz, 1996). This highlights the importance of activities that are pleasurable and generate satisfaction when they are carried out (Hwang & Kim, 2011; Silva et al., 2010; Verloigne et al., 2011).