Address for correspondence: Dr Judith de Niet, UBC, L408 - 4480 Oak Street, Vancouver, BC V6H 3V4, Canada. E-mail: email@example.com
Maintaining weight loss results in childhood obesity treatment is difficult to achieve. Self-management techniques such as self-monitoring are associated with increased weight loss and maintenance. This study analyzes whether self-monitoring of lifestyle behaviours through a short message service maintenance treatment (SMSMT) via mobile phones with personalized feedback positively effects weight, lifestyle behaviours and psychological well-being in obese children.
After 3 months of behavioural lifestyle treatment, 141 overweight and obese children (7–12 years) were randomly assigned to an intervention group receiving SMSMT for 9 months (n = 73) or to the control group (n = 68). The intervention group sent weekly self-monitoring data on exercise and eating behaviour and their mood via mobile phones. In return, they received tailored feedback messages. Primary treatment outcomes were weight, eating behaviour and psychological well-being, i.e. competence, self-esteem and quality of life. Secondary outcome was adherence to the SMSMT. Data were analyzed with mixed modelling.
SMSMT did not improve treatment outcomes. Controls gained temporarily in physical health scores (P = 0.01). SMSMT completers sent on average every 2 weeks an SMS. Children who had greater weight loss during the first 3 months of lifestyle treatment sent more SMSs (P = 0.04).
We did not find a positive effect of SMSMT on weight, eating behaviour or psychological well-being in obese children. SMSMT seems to be a feasible method of treatment delivery. Future research should study variations of SMSMT to investigate how SMSMT can be more effective.
The literature points out that not all overweight and obese children profit from lifestyle programs.
In addition, maintenance of achieved weight loss and adherence to treatment seem to be hard to achieve. However, studies on maintenance treatments are relatively scarce.
To our knowledge, this is one of the first randomized controlled studies that examined the effect of a Short Message Service Maintenance Treatment (SMSMT) on BMI-SDS, competence, self-esteem, and Quality of Life in a sample of overweight and obese children participating in a lifestyle treatment.
What this study adds
We believe this study contributes to the understanding of the effect of new technology methods in the treatment of childhood overweight and obesity.
Our findings provide insight in how to tailor and optimize maintenance care interventions by the use of text messaging.
Rates of paediatric obesity are rising worldwide . The obesity epidemic is accompanied by an increase in the number of children with type 2 diabetes mellitus, which was traditionally considered to be a disease of middle age . Although lifestyle programmes have led to weight loss and physical and psychological improvements , it seems that not all children benefit from these interventions. Moreover, long-term follow-up is difficult to achieve and interpret, partially due to high dropout rates . Maintaining treatment results and improving childhood obesity treatment outcomes requires long-term support from healthcare providers. Maintenance treatments using self-management skills such as self-monitoring might improve treatment outcomes: self-monitoring has been shown to be an important behavioural strategy in weight loss programmes resulting in greater weight loss in children as well as adults [5, 6]. New technologies, such as the short message service (SMS), are a common and popular way of communication among youngsters and might be a promising tool to enhance existing interventions.
In a previous study, we reported preliminary results of a SMS maintenance treatment (SMSMT) in a paediatric lifestyle programme showing the SMSMT to be a feasible way of communication. The two main components of the SMSMT are self-monitoring and personalized feedback. Self-monitoring is a behavioural strategy, which refers to the awareness of bodily symptoms, daily activities and cognitive processes through the self-recording of behaviour  and is an important element in self-regulation theory. Improved self-monitoring is associated with weight loss in children and their participating parents . Self-monitoring mostly consists of keeping track of the amounts of food taken, the level of exercise and counting calories. Furthermore, it has been shown that tailored feedback is more effective than generic feedback . SMS seems to be an ideal medium for self-monitoring and delivering tailored feedback in health care. It has several advantages such as accessibility any time, interactivity and rapidity, and limited costs. Technically, SMS enables frequent monitoring of behaviours and can be pre-programmed and tailored to the needs of an individual. Greater adherence to self-monitoring of health behaviour has been found in healthy children and adolescents using SMS compared to paper and pencil diaries . Despite some evidence that behavioural interventions through SMS has positive short-term behavioural outcomes in adolescents and adults , the use of SMS in health care is still in its early stages and needs to be further studied to evaluate its potential.
In the current study, we evaluate the effect of a SMSMT (using mobile phones) in overweight and obese children. The SMSMT was introduced after 3 months of participation in a 1-year family-based multidisciplinary lifestyle treatment, called the ‘Big Friends Club’ (BFC). To our knowledge, this is one of the first randomized controlled studies examining the effect of SMSMT on BMI, eating behaviour, Body Mass Index-Standard Deviation Scores (BMI-SDS), eating behaviour and on self-perceived competence, self-esteem and quality of life (referred to as psychological well-being) in a sample of overweight and obese children participating in a lifestyle treatment. The primary aims of the current study were to evaluate the effect of the SMSMT on BMI-SDS, eating behaviour and psychological well-being. We hypothesize that the SMSMT positively affects BMI-SDS, eating behaviour and psychological well-being as children receive long-term support and attention as well as that self-monitoring might improve their awareness of their lifestyle behaviours. Secondary outcomes were the effect of the SMSMT on the amount of self-reported health behaviours and mood and the feasibility of the SMSMT. Losing or maintaining weight might indirectly influence their mood as well as their psychological well-being.
A multi-centre randomized controlled trial was conducted among overweight and obese children participating in the BFC programme in hospitals in the Netherlands. The BFC programme is a 1-year family-based multidisciplinary behavioural group intervention for children aged 8 to 12 years that is limited to groups of 10 children. The programme has been described in detail elsewhere [12, 13]. The treatment team includes a psychologist, dietitian, paediatrician and physiotherapist. The therapy is aimed at reducing BMI-SDS, preferably by maintaining weight during growth and adopting a healthy lifestyle. Other treatment goals include creating a positive self-image and improving skills to cope with the psychosocial consequences of obesity (e.g. bullying). The programme consists of an intake session, eight children sessions and a minimum of three parent sessions during the first 3 months. The first six children sessions are weekly and the last two sessions are bi-weekly. The three parent sessions take place at the same time of the childrens' 1st, 4th and 8th session. Additionally, follow-up group sessions for children and parents and individual appointments are organized at 6, 9, and 12 months after the start of the programme. Summarized, children receive 11 group sessions in 1 year. The programme uses different behavioural modification techniques such as goal setting, problem solving and self-regulation skills. The first 90 min of the children's group sessions address healthy eating and exercise behaviour and strategies to deal with difficulties concerning eating or physical activity. Furthermore, attention is paid to the psychosocial aspects of obesity like being picked on by peers. The last hour of each session is led by the physiotherapist. The children exercise together and these sessions are meant to improve their physical condition and create positive exercise experiences through games and sports. Parent group sessions focus on healthy diet and exercise behaviour, psychosocial aspects of obesity and the increased risk of physical and psychological morbidity. Part of the sessions is devoted to changing interaction patterns between parents and their children by teaching them how to support their child instead of controlling, give positive feedback and apply positive reinforcement.
After the first 3 months of the BFC programme, children and their parents experience a transition from a period with regular contact with the BFC team to a period with less intensive contact. Therefore, SMSMT is introduced after 3 months of BFC treatment. In this parallel trial design with balanced groups, participants were randomly assigned to the intervention group receiving SMSMT for 9 months (until the end of the BFC follow-up sessions) or to the control group receiving no SMSMT (BFC only; Fig. 1). Randomization allocation in a 1:1 ratio was applied in a randomized block design. The blocks were formed by the treatment groups. The randomization allocation was printed on paper in a sealed envelope. An equal number of SMSMT and control notes were put in the envelopes. The researcher randomized the children to the SMSMT or control group by picking an envelope from a basket. Measures were assessed at baseline, start of randomization (after 3 months of the BFC) and at 6, 9, and 12 months.
Eight Dutch hospitals complying with the BFC protocol during the period 2006–2009 participated in this study. Inclusion criteria were being overweight or obese, parent participation in the BFC and a sufficient knowledge of the Dutch language. Exclusion criteria for overweight and obese children participating in the BFC programme included behavioural problems defined as a score >70 on the Child Behaviour Checklist (CBCL)  and any disease causing overweight that could be treated with drugs and mental retardation. Furthermore, parents and child had to be both sufficiently fluent in the Dutch language and show sufficient motivation to actively participate in the programme. Motivation was assessed using the technique of motivational interviewing . The BFC team discussed motivation with participants at intake after they explained the intensity and time investment of the BFC programme and emphasized expected efforts. Participants were recruited to the study by the paediatrician at intake. All participants gave written informed consent before entering the study. The study protocol was approved by the Medical Ethics Committee of the Erasmus MC University Medical Centre Rotterdam and by the medical ethics committees of the participating hospitals.
The short message service maintenance treatment (SMSMT)
The methods primarily used in the SMSMT intervention are self-monitoring and personalized feedback. The SMSMT intervention is based on various behaviour change methods from social learning and cognitive theories, self-regulation theories and behavioural models [16-18]. Behavioural models use various methods, which are directed at changing behaviour directly. Self-regulation refers to a series of self-care behaviours including behavioural methods such as self-monitoring (self-observation), goal setting (self-evaluation) and reinforcement . There is evidence indicating that feedback providing personally relevant information (i.e. personalized feedback) is more effective in behaviour change than other types of feedback such as generic feedback . Personal feedback has a greater probability of addressing the person's unique need than general feedback.
Participants in the SMSMT group received a mobile phone for the period of the SMSMT intervention. The researcher explained participants how to use the mobile phone; all children were able to use the mobile phone effortlessly. The children were requested to send weekly self-monitoring data on a five-point Likert scale via SMS. They sent self-monitoring data on physical activity, healthy eating pattern and their mood. The higher the number on the Likert scale, the more favourable the behaviour in question was. The children sent three numbers in total. Data was stored on a secure web server at the Centre for Psychotherapy Research at the University Hospital Heidelberg, Germany. A software program automatically compared the incoming self-monitoring data to the data in the previous SMS. In this way, the program detected deterioration, improvement or maintenance of behaviours. Next, the program suggested a feedback message with a 160-character limitation out of a large pool of pre-formulated statements, which was tailored to the child's individual pattern of change in behaviour. The feedback was checked for plausibility by a researcher who tailored the feedback message when needed (Fig. 2). Tailoring aspects included taking into account previous SMS responses, participant's name and personal difficulties. The feedback messages were formulated according to four principles: (i) promoting social support; (ii) motivating participants; (iii) reinforcing positive changes; and (iv) suggesting behaviour modification and self-management skills that were learnt during the BFC programme. The researcher enhanced compliance by sending an SMS reminder after 1 week of non-responding. The software underlying the SMSMT has been successfully used in four other studies using SMS to enhance mental health and well-being (e.g. ).
Besides the demographic variables, we assessed all measurements at baseline at 3, 9, and 12 months.
Demographic and anthropometric variables
Information on demographic characteristics including age, gender and ethnicity were collected. Ethnicity was divided into two categories: Caucasian (both parents) and non-Caucasian (foreign ethnicity of at least one parent). Overweight and obesity were measured in terms of BMI-SDS, which are BMI-SDS corrected for age and gender. A BMI-SDS of 1.1 represents the threshold for overweight, and a BMI-SDS of 2.3 indicates obesity . The children's weight and height were measured in the hospital by a physician. The BMI-SDS was calculated using the Growth Analyser Version 3.5 .
Eating behaviour: the Dutch Eating Behaviour Questionnaire (DEBQ)
The DEBQ  was administered to assess two over-eating styles namely emotional eating (eating in response to negative emotions) and external eating (e.g. overeating in response to the smell of food) in 7–12-year-old children. In addition, the DEBQ measures restrained eating (eating less than desired with the aim to lose or maintain weight). High scores reflect a high degree of the eating behaviour in question. All subscales have good internal reliability (Cronbach's α = 0.80, 0.74 and 0.81, respectively) . We used norm scores of Dutch children from the general population with a mean age of 9.6 years and an age range of 7–12 years . In the current study, these norm scores were used to visualize the position of the participants in figures. Although contradictory results have been found for restrained eating with respect to weight loss [23, 24], high degrees of restrained , emotional  and external eating  seem to be related to higher BMI. Lifestyle interventions have been found to significantly improve eating behaviour styles [27, 28].
Self-perception: the Self-perception Profile for Children (SPPC)
The Dutch version of the SPPC  aged 8–12 years was used and the questionnaire measures self-perceived competence on five domains: scholastic competence, social acceptance, athletic competence, physical appearance and behavioural conduct. In addition, the SPPC assesses global self-worth. High scores reflect greater perceived competence or global self-worth. This study examined the three subscales: athletic competence, physical appearance and global self-worth, all having moderate to good internal reliability for boys and girls (α = 0.70, 0.82, and 0.74, respectively) . We used separate norm scores of Dutch schoolchildren for boys and girls (mean age = 10) because of significant differences in scores between boys and girls. The SPPC was only not measured at 9 months. Obesity seem to negatively impact self-perceived competences such as athletic and physical competence and self-esteem in obese adolescents [31, 32], especially in girls [31, 33]. Lifestyle treatment significantly improves athletic and physical competence and global self-esteem .
Health-Related Quality of Life (HRQoL): the Child Health Questionnaire – PF50 (CHQ-PF50)
The Dutch validated version  of the CHQ-PF50  was used to assess parent-rated health-related QoL in children. In the current study, we used ‘physical’ and ‘psychosocial’ CHQ summary scores , both having good internal reliability (α = 0.93, and 0.87, respectively) . The highest summary scores (100) reflect the best possible health state, and the lowest scores (0) the poorest health state. We used norm scores of a population of Dutch schoolchildren aged 5–13 years . Obese youth rate their general and physical health lower than their normal-weight peers and report impaired social functioning and psychosocial health [37, 38]. It has been shown that lifestyle treatment leads to improved subjective health scores in obese children .
The following three questions were answered weekly by the children in the SMSMT group via text messages and answered on a five-point Likert scale (Fig. 2): (i) how many hours did you engage in physical activity during the last week? e.g. sporting, cycling, walking or playing outside; (ii) how many days did you adhere to a healthy eating pattern last week?; and (iii) how often did you feel sad or unhappy last week?
In order to detect a medium-sized difference between children in the SMSMT group and controls, respecting an alpha level of 0.05 and a power of 0.80, 64 cases were needed in each treatment group. To test differences in the anthropometric and demographic variables between the intervention group and the control group Student's t-tests were used for continuous variables and Fisher's exact tests were used for categorical variables. Paired Student's t-tests were applied to test differences on BMI-SDS, DEBQ, SPPC and the CHQ-PF50 between 0 and 3 months of BFC treatment.
Longitudinal analyses of SMSMT
Longitudinal analyses over the period 3 to 12 months were performed with mixed modelling using SAS 9.2 . This technique allows the use of incomplete cases if missing at random  and facilitates within subject analyses. Nine saturated models were postulated with the dependent variables BMI-SDS, emotional eating, external eating, restrained eating, athletic competence, physical appearance, global self-worth, physical health and psychosocial health. SMSMT, age, gender, linear time and quadratic time and interactions were entered into the models as main effects. The covariance structure was determined using the restricted maximum likelihood estimation method . In a backward procedure (interaction), effects with a P value>0.05 were removed, respecting that interaction effects must be nested under their respective main effects. In this way, final parsimonious models were reached that did not differ significantly from the saturated models.
Effect sizes were calculated by dividing the difference between follow-up estimate and estimate at the start of randomization by the estimated standard deviations from the models (Equation 1), pooled at start of randomization and follow-up.
Differences at various time points between the SMSMT and the control group were estimated by formulating contrasts in the mixed models.
Self-reported lifestyle behaviours and adherence to the SMSMT
The general linear model was applied to test the difference in mean self-reported scores on physical activity, healthy eating and mood over time for the SMSMT group. To test the differences in scores on adherence to self-monitoring of physical activity, a healthy eating pattern, and mood, anova post-hoc analyses were applied within the three SMSMT subgroups: SMSMT completers, SMSMT dropouts and BFC treatment dropouts. SMSMT completers were those children who completed the 9 months SMSMT intervention and therefore also the BFC. SMSMT dropouts were those who only dropped out the SMSMT but continued with the BFC. BFC treatment dropouts dropped out of the BFC treatment and therefore were not allowed to continue with the SMSMT. Pearson correlations were calculated to analyze the relationship between changes in BMI-SDS and the number of submitted SMSs in the first 3 months of the SMSMT.
All statistical testing took place at a 0.05 level of significance (two-tailed).
A total of 144 children (mean age 9.9 years; standard deviation (SD) = 1.3; range 7–12 years) and their parents participated in the current study. Three children dropped out in the first 3 months of the BFC programme (Fig. 1). In total, 141 children were randomly assigned to either the intervention group (n = 73) or the control group (n = 68). The two groups did not differ in terms of baseline, clinical, and demographic characteristics (Table 1). However, the two groups did differ at baseline with respect to CHQ psychosocial health scores: the SMSMT group had higher scores at baseline (P = 0.04). The total group had a mean BMI-SDS of 2.6 (SD = 0.4). At baseline, 74% were obese compared to 56% after 1 year. At start of randomization, the control group had higher scores than the intervention group with respect to CHQ physical health (P = 0.03; Table 2). Only 10 children in the intervention group dropped out of the BFC treatment (14%) in the period between 3 and 12 months compared to 21 children in the control group (31%). In the intervention group, 12 children discontinued the SMSMT but continued with the BFC treatment between 3 and 12 months (Fig. 1).
Table 1. Baseline characteristics of the total study population, the control and the intervention group
aValues are mean (SD) [range], range, n or number (%) of participants.
bContinuous data were tested by t-test for independent samples and categorical (binary) data by Fisher's Exact test. Tested at a 0.05 level of significance (two-tailed).
BMI-SDS, body mass index – standard deviation scores; CHQ, Child Health Questionnaire; DEBQ, Dutch Eating Behaviour Questionnaire; SD, standard deviation; SMSMT; short message service maintenance treatment; SPPC, Self-Perception Profile for Children.
9.9 (1.3) [7–12]
0.95 (139) 0.35
2.59 (0.45) [1.12–3.60]
1.25 (136) 0.21
DEBQ emotional eating
1.23 (0.31) [1–2.7]
0.97 (134) 0.34
DEBQ external eating
1.92 (0.51) [1–3]
1.09 (134) 0.28
DEBQ restrained eating
2.00 (0.47) [1–3]
0.36 (134) 0.72
SPPC athletic competence
15.5 (4.0) [6–24]
0.47 (134) 0.64
SPCC physical appearance
14.4 (4.0) [6–24]
0.13 (135) 0.90
SPCC global self-worth
17.8 (4.0) [6–24]
0.69 (135) 0.49
50.0 (9.2) [1–63]
0.64 (127) 0.53
49.0 (7.4) [27–67]
Table 2. Characteristics of the total study population, the control, and the intervention group at randomization (3 months)
bContinuous data were tested by t-test for independent samples.
BMI-SDS, body mass index – standard deviation scores; CHQ, Child Health Questionnaire; DEBQ, Dutch Eating Behaviour Questionnaire; SD, standard deviation; SMSMT, short message service maintenance treatment; SPPC, Self-Perception Profile for Children.
2.43 (0.53) [0.64–3.63]
0.85 (128) 0.40
DEBQ emotional eating
1.18 (0.28) [1–2.29]
0.81 (135) 0.42
DEBQ external eating
1.97 (0.47) [1–3]
0.56 (135) 0.58
DEBQ restrained eating
2.05 (0.43) [1.14–3]
0.35 (135) 0.73
SPPC athletic competence
16.3 (4.2) [7–24]
1.19 (129) 0.24
SPCC physical appearance
15.6 (4.6) [6–24]
0.25 (129) 0.81
SPCC global self-worth
18.6 (4.6) [6–24]
0.34 (129) 0.74
52.2 (7.3) [6–62]
51.1 (7.9) [23–65]
1.67 (128) 0.10
Effect of the BFC treatment
With respect to the effect of the first 3 months of the BFC programme, we found significant improvements for the total group on BMI-SDS (P < 0.001), athletic competence (P = 0.012), physical appearance (P = 0.004), global self-worth (P = 0.035), physical health (P = 0.001), and psychosocial health (P = 0.017). BMI-SDS characteristics of children at 3, 6, 9 and 12 months are shown in Table 3.
Table 3. Characteristics of the total study population, the control and the intervention group with respect to BMI-SDSa
aValues are mean (SD) [range], range, n or number (%) of participants.
bContinuous data were tested by t-test for independent samples and categorical (binary) data by Fisher's exact test. Tested at a 0.05 level of significance (two-tailed).
SMSMT, short message service maintenance treatment; SD, standard deviation; BMI-SDS, body mass index – standard deviation scores.
Obesity at 3 months
Obesity at 6 months
Obesity at 9 months
Obesity at 12 months
Longitudinal analyses of SMSMT: the effect of SMSMT on BMI, eating behaviour and psychological well-being
Longitudinal analyses using mixed modelling showed that the two groups did not change in different ways between 3 and 12 months after the start of the BFC with respect to BMI-SDS. Table 4 shows the estimated weights and significance levels of the longitudinal mixed models, Table 5 shows the estimates and effect sizes at the follow-ups, and the course in time is depicted in Fig. 3. No difference in change in BMI-SDS was found after 12 months for the SMSMT group. The level of the normal population and one SD was added in Fig. 3 to illustrate the position of the participants compared to the general population. The scores of the dependent variables emotional eating and external eating decreased significantly over time: these effects are less than small and small, respectively . All children had reductions in external eating scores. Physical health increases somewhat in the control group, while it decreases a little for the SMSMT group: this difference was significant at 3 and 6 months (both P = 0.01), but not at 9 months (P = 0.09). No significant differences between the control and the intervention group were found for emotional and restrained eating, athletic competence, physical appearance, global self-worth, and psychosocial health. Minimal gains were found for athletic competence (P at 3 months = 0.01, 6 months 0.02 and 9 months 0.11), physical appearance (P respectively = 0.01, 0.02 and 0.25) and global self-worth (P respectively = 0.01, 0.03 and 0.37); however, those gains were non-significant at 9 months of SMSMT.
Table 4. Longitudinal analyses of the effect of SMSMT on BMI-SDS, eating behaviour, and psychological well-being: final MIXED models (start randomization – end BFC: 3–12 months)a
Clarification of the results in this table: each row in this table represents a regression equation. These show estimations of the scores for all time points. For the control group the estimation of external eating scores at a certain moment after the start of randomization is: 1.94 − 0.09 × months + 0.007 × months squared. For example, at months 3 (randomization), the estimation is: 1.94 − 0.09 × 0 + 0.007 × 0 × 0 = 1.94 and after 9 months, the estimation is 1.94 − 0.09 × 9 + 0.007 × 9 × 9 = 1.94 − 0.81 + 0.567 = 1.697. For the SMSMT group, the equation is (1.94 + 0.044) – (0.09 + 0.048) × months + (0.007 − 0.005) × months squared = 1.984 − 0.042 × months + 0.002 + months squared, at randomization, the estimation is 1.984 − 0.042 × 0 + 0.002 × 0 × 0 = 1.984, and after 9 months the estimation is 1.984 − 0.042 × 9 + 0.002 × 9 × 9 = 1.984 − 0.378 + 0.162 = 1.786.
aTested at a 0.05 level of significance (two-tailed).Data were analyzed with mixed modelling that provides insight in significant changes of outcomes in course of time.
BMI-SDS, body mass index – standard deviation scores; CHQ, Child Health Questionnaire; DEBQ, Dutch Eating Behaviour Questionnaire; est., estimated regression; SMSMT, short message service maintenance treatment; SPPC, Self-Perception Profile for Children.
Besides these SMSMT interaction effects, the results showed older children having lower scores on the subscales physical appearance and global self-worth, although this gain was somewhat moderated in the long run.
Self-reported lifestyle behaviours and adherence to the SMSMT
The following analyses have been done in the intervention group only. We divided the SMSMT group into SMSMT completers (n = 51), SMSMT dropouts (n = 12), and BFC treatment dropouts (n = 10), sending in the first 3 months of the SMSMT on average 0.80 SMSs (SD = 0.19, range = 0.4–1.1), 0.55 (SD = 0.24, range = 0.2–1.0) and 0.78 SMSs (SD = 0.28, range = 0.4–1.2) per week, respectively. SMSMT dropouts sent significantly less SMSs in 1 week compared to treatment dropouts and SMSMT completers (P = 0.02 and P < 0.001, respectively). The SMSMT completers have sent 0.55 and 0.47 SMSs after 6 and 9 months of the SMSMT, respectively. Pearson correlations showed that completers with the greatest BMI-SDS change during the first 3 months of the BFC programme sent the most text messages per week in the first 3 months of SMSMT (r = 0.29, P = 0.041) as well as in the total SMSMT period (r = 0.36, P = 0.012).
For self-reported days per week of adherence to a healthy eating pattern, children reported to adhere less to a healthy eating pattern over time (test within subjects contrast F = 4.575, d.f. = 1, P = 0.037): in the first 3 months of the SMSMT, they adhered 3.78 (SD = 0.52) days to a healthy eating pattern, in months 3–6 of the SMSMT 3.74 (SD = 0.55) days, and in the last 3 months 3.60 (SD = 0.78) days. We found no significant effects with respect to physical activity or mood. In the subgroup analyses, we found a significant result for the subgroup SMSMT completers only: the SMSMT completers had a greater number of physical active days per week over time (P = 0.030, 95% CI = 0.04–0.81) and they adhered more often to a healthy eating pattern (P = 0.015, 95% CI = 0.09–0.78) than SMSMT dropouts.
In recent years, a variety of interventions for treating paediatric overweight have been developed [3, 44]. To our knowledge, this is one of the first randomized controlled studies examining the effect of a SMSMT on BMI-SDS, eating behaviour, and psychological well-being in a sample of overweight and obese children participating in a lifestyle treatment. We found that children improve their BMI-SDS, athletic competence, physical appearance, global self-worth, physical health, and psychosocial health in the first 3 months of the BFC programme. With respect to the SMSMT, we indicated that an additional text messaging intervention is a feasible method for the self-monitoring of health behaviours and providing tailored feedback to children aged 7–12 years. In the first 3 months of the SMSMT, SMSMT completers sent 80% of the requested messages, and 55% and almost half of the requested messages after 6 and 9 months, respectively. We found that children who received SMSMT were less likely to withdraw from the family-based behavioural lifestyle BFC treatment than children who did not receive the SMSMT; however, this study did not show that the use of the SMSMT positively influences BMI-SDS, eating behaviour, and psychological well-being.
In line with our findings, Newton et al. (2009), studying adolescents with type 1 diabetes who received motivational text messages for 3 months reminding them to wear a pedometer and be active, found no improvement in BMI and quality of Life (QoL) . We found that children in the present study improved their BMI-SDS and psychological well-being during the first 3 months of the BFC treatment. Consequently, significant changes by SMSMT during the maintenance treatment might have been more difficult to achieve. Although we found no significant differences in BMI-SDS between the two groups over time, both the control and the intervention group significantly reduced their BMI-SDS at 12 months compared to their baseline BMI-SDS (t(60) = 2.66, P < 0.001 and t(66) = 5.25, P < 0.001, respectively). Furthermore, it might be that the current self-monitoring procedure was not intensive enough to improve measures such as QoL and self-esteem. Future research should examine more frequent self-monitoring and whether self-monitoring of various other variables should be included, e.g. sedentary behaviours, beverage intake, and weight. Sedentary behaviours (e.g. television watching) and the intake of sugar-sweetened beverages seem to be related to higher BMI [46, 47]. Epstein et al. showed that both stimulus control and reinforcing decreased sedentary behaviours in obese children .
Children in the two groups also changed in different ways over time with respect to physical health. Self-reported physical health of the children in the SMSMT group decreased, whereas the control group initially showed an improvement in physical health, although this increase was later moderated. Because the difference in scores existed at the start of randomization, the difference in main effects can probably be attributed to randomization bias and not to the effect of the SMSMT. None of the children in the two groups achieved levels of physical health within the range of the norm group. Also, the difference between the groups with respect to physical health scores might be explained by selective dropout as the control group had a significantly higher dropout rate. An earlier study on the BFC treatment indicated that dropouts were less successful in BMI-SDS reduction during treatment . We speculate that the children that would have dropped out without the SMSMT had poorer treatment outcomes. If so, retaining these children in the BFC programme by SMSMT may have moderated the mean treatment effect of the SMSMT.
In line with our preliminary results , we found that completers with the greatest BMI-SDS change during the first 3 months of the BFC programme sent the greatest amount of text messages per week during the SMSMT. Whereas a successful experience might motivate children and improve their adherence to the SMSMT, a failure experience during the intensive treatment might be demoralizing and result in a higher chance of giving up.
In the SMSMT group, no significant improvement over time was found for self-reported mood and adherence to physical activity, but a significant deterioration over time was found for self-reported adherence to a healthy eating pattern. It is difficult to interpret the size of this deterioration because we had no reference group to test the difference in improvement/deterioration on self-reported adherence to health behaviours. Given the small size of deterioration, we believe that this finding is not clinically relevant. However, no decrease in physical activity and no deterioration over time for mood were found, implying that these patterns were stabilized over time.
Furthermore, 10 children dropped out the BFC and therefore out of the SMSMT, while 21 children in the control group not receiving the SMSMT dropped out of the BFC programme. The children who dropped out the SMSMT but stayed in the BFC might have not needed the SMSMT to complete the programme or the SMSMT might not have been attractive enough to keep them engaged. Future research should study the predictors of dropout from the SMSMT.
One of the drawbacks of this study is that adherence to each specific health behaviour was measured with only one item using a five-point Likert scale. Future studies should evaluate adapted versions of the SMSMT and their effects by measuring physical activity with accelerometers, pedometers or questionnaires, and by measuring dietary intake with validated questionnaires. Another limitation of the study is that the number of analyses  is rather large relative to the number of included subjects (140). Also, we only measured eating behaviour, competence and self-esteem, and QoL. We are aware that other psychological aspects such as self-efficacy might have been improved by the SMSMT. One study in adolescents with diabetes mellitus found that self-efficacy was improved in the intervention group receiving text messages compared to the control group receiving no text messages . Although a strength of the SMSMT was the simplicity of the programme and a minimal time investment for both the health care provider and the participants, the SMSMT might have been too limited to improve psychological well-being. However, failing to find a significant effect does not mean that the programme is not effective in other formats. Other strategies such as problem solving, a stronger emphasis on coping with difficult situations, and educational messages in a positive and stimulating way might be considered for future research. Also, some evidence for enhanced effectiveness of behaviour interventions was found when messages and feedback by SMS, mail or phone were received more frequently . In the current study, the participants received feedback only once a week.
The use of SMS in the treatment of paediatric overweight has several advantages and seems to be particularly useful as a maintenance treatment to reduce dropout rates. In a previous study examining predictors of dropout from the BFC, we found dropouts to be more likely non-Caucasian and less active, have higher baseline BMI-SDS and not live in families with an authoritative structure . Especially, children with those characteristics might profit more from the SMSMT. Future research should study these associations. The use of SMS in health care is still in its early stages and needs to be further explored. Adapted versions of the SMSMT should be developed and their effect on BMI-SDS and psychological well-being should be further studied, together with their cost-effectiveness.
Conflict of interest statement
The authors declare no conflict of interest.
The authors thank the Sint Franciscus Hospital Rotterdam, Maasstad Hospital Rotterdam, Atrium Medical Centre Heerlen, Gelderse Roos and Gelderse Vallei Hospital Ede, GGZ Helmond and Elkerliek Hospital Helmond, Scheper Hospital Emmen, Medical Centre Leeuwarden, Hospital Diakonessenhuis Zeist for providing data, and the Centre for Psychotherapy Research, University of Heidelberg for their (technical) support and for sharing their expertise on the short message service maintenance treatment.
Financial disclosure: Support for this study was received from Vodafone, the Netherlands. Grants were received from the Erasmus University Medical Centre Rotterdam – MRACE (Medical Research Advice Committee) grant no. 2006-26 and Innovation Fund Insurances (Innovatiefonds Verzekeringen) grant no. 06–334.