Treatment of Obesity in Young People—a Systematic Review and Meta-analysis

An extensive literature search of the MEDLINE and EMBASE databases from inception (last search May 2016) was conducted. Search terms were included to detect studies investigating lifestyle modifications, pharmacological therapies, endoscopic treatments and bariatric surgery in the young population with obesity (Supplementary Appendix 1).

Studies were selected according to the following pre-defined eligibility criteria. Type of participant: We define young people as those aged 21 years or younger. Thus, included study participants were aged 21 years or younger with a BMI of 25 kg/m² or higher. Studies with pregnant females, neonates and patients with obesity-related genetic syndromes (e.g. Prader Willi syndrome) were excluded. Type of intervention: Four interventions were included: lifestyle modifications, pharmacological therapies, endoscopic treatments and bariatric surgery. There was no restriction on who delivered the interventions, e.g. specialist doctors or teachers. The interventions could be community-, school- or clinic-based. Type of comparison: The highest level of available evidence was used in each group. Lifestyle and drug studies were all randomised control trials and had controls which had either no intervention, placebos or minimal interventions such as a single information session. Only a single randomised control trial (RCT) was retrieved assessing surgical management, so non-randomised studies were included for this sub-analysis. Type of outcome measure: Studies included one or more of the following outcomes: weight (kg), BMI (kg/m²), BMI standard deviation score (BMI-SDS or BMI-Z score), waist circumference (cm), fat mass (kg) or waist to hip ratio. These could either be reported as a primary or secondary outcome of the study.

Two independent reviewers (SS and NA) assessed the search strategy results. Study titles were examined for potential relevance and the abstract was then reviewed. Subsequently, the full text was retrieved to ensure eligibility. Review articles and studies with insufficient data for meta-analysis were excluded. No limits were applied for language and foreign papers. This involved the translation of one Portuguese study [9]. The bibliographies of relevant articles were inspected for further eligible studies. In the case of any uncertainty regarding study inclusion, another investigator (NP) was consulted to assess eligibility.

Eligible studies were sub-divided into four groups: lifestyle modifications, pharmacological therapies, endoscopic treatments and bariatric surgery. For the included studies, the following variables were obtained: study title, publication year, first author, type of intervention, number of participants, metric of measure for weight loss, baseline control and interventional group weight metrics, post-intervention control and interventional group weight metrics. Due to a sparsity of evidence using other weight metrics, BMI was utilised for the quantitative analysis.

Methodological quality of RCTs was assessed using the Cochrane Collaboration Tool (Supplementary Appendix 2) [10]. This included assessment of random sequence generation, allocation concealment, blinding of participants and assessors, incomplete outcome data, selective reporting and other sources of bias [10].

Any missing standard deviations were derived from other metrics provided, using standard formula found in the Cochrane handbook. The chi-squared test was completed to establish the degree of heterogeneity [11]. Higgins et al. suggest that 25, 50 and 75% indicate low, medium and high heterogeneity respectively [11]. As I² values were over 75%, the restricted maximum likelihood random-effects method was used to generate pooled mean BMI differences [11]. Publication bias was assessed using funnel plots (Supplementary Appendix 3). Statistical analyses were performed using the Metafor package in R (version 3.2.4).

This study investigated the comparative efficacy of BMI-reducing treatment options for obesity in the young population. It illustrates that as a single intervention, bariatric surgery is the most effective treatment for BMI reduction in the short and long terms. Furthermore, LSG and RYGB caused a statically significant reduction in BMI compared to LAGB at 12 months.

Although this systematic review and meta-analysis is not designed to evaluate the relative risks between different treatment modalities for obesity in young patients, it is important that the risks that accompany surgery are well established. This study suggests that bariatric surgery has a good safety profile in the paediatric population. However, there are existing concerns over the long-term consequences of bariatric surgery on pre-pubertal children with severe obesity, including factors such as growth. Recently, Alqahtani et al. showed that, contrary to expectations, growth was improved after LSG, with approximately 1 extra millimetre of height gain per month compared with obese children who did not undergo surgery [41]. However, of the 2155 patients included that underwent surgical intervention, one death relating to the surgery was reported 48 months following LAGB [35]. The 19-year-old patient underwent emergency endoscopic evaluation 4 years post-gastric banding due to melena and haematemesis. Gastric haemorrhage with band erosion was found. During gastric band removal, acute haemorrhage occurred and the patient subsequently died of uncontrolled bleeding [35]. This illustrates that although rare, severe post-operative complications are possible. Furthermore, evidence regarding the long-term complications associated with bariatric surgery on young patients is still sparse [43]. Therefore, it is important that the risks that accompany surgery are not underestimated and are well established in each individual case [44].

Moreover, evidence regarding the long-term efficacy of bariatric surgery in young patients is limited. However, recently, Olbers et al. conducted a prospective study analysing 5-year outcomes of adolescent patients after Roux-en-Y gastric bypass. It demonstrated a BMI reduction of 13.5 kg/m² compared to a control group undergoing lifestyle intervention that experienced an increase in BMI of 3.3 kg/m² [45]. Additionally, Inge et al. analysed 3-year outcomes of adolescent patients after Roux-en-Y gastric bypass and sleeve gastrectomy. The study found that patient BMI decreased by 15 and 13 kg/m², respectively [46]. Nevertheless, despite emerging studies that demonstrate the longer-term efficacy of bariatric surgery in adolescent patients, it is important to highlight that long-term data in this cohort is still limited and further studies are required.

Assessment of lifestyle interventions revealed that a combined approach of behaviour, exercise and dietary treatments resulted in a statistically significant greater BMI reduction compared to a solitary lifestyle intervention. Moreover, both lifestyle modifications and pharmacological therapies have similar impacts on BMI change in the short and medium terms.

Sibutramine appeared to be the most effective drug at achieving BMI loss in the short term. It facilitates weight loss by acting as a serotonin reuptake inhibitor causing activation of anorexigenic pathways to prevent cognitive sensations of hunger [47].

Metformin caused similar BMI changes over short- and medium-term administration periods. However, there is no clear benefit on BMI loss after long-term metformin administration compared with short- and medium-term. The mechanism by which metformin results in weight loss is unclear, but it is likely to be multifactorial [48]. Its effect on weight may be mediated by increasing glucagon-like peptide 1, reducing dipeptidyl peptidase-4 activity, modifying the gut microbiome, increasing lipid oxidation and promoting central satiety [48].

Orlistat inhibits gastric and pancreatic lipase and decreased absorption of triglycerides by approximately 30% [49]. It is the only weight-reducing drug approved for paediatric patients with obesity in USA [49]. However, it is imperative that more studies investigate the weight-reducing effects of orlistat in young adults to conclusively determine whether it should be preferred to other available treatment options.

Endoscopic treatments appear to cause statistically significant weight loss in the short term. However, given the low number of studies included, it is vital that more studies are conducted to further assess its impact on weight loss. The mechanism behind the IGB includes the inflation of a balloon in the stomach of patients to fill its capacity, promoting satiety and decreasing calorie intake [50]. The Endobarrier leads to weight loss by creating a 60-cm-length barrier along the duodenal wall to reduce absorption of nutrients [51].

To our knowledge, this is the first comprehensive systematic review and meta-analysis to directly compare the efficacy of weight-reducing treatment options for obesity in the young population. Lifestyle modification and pharmacological therapy meta-analyses contain a large number of RCTs. The surgical meta-analysis contains five times the number of overall patients and double the number of LSG studies compared to the most recent topical paper [52].

This study includes sibutramine despite its withdrawal from the USA and UK market following the SCOUT study [53]. It was illustrated that sibutramine carries a relative risk of 1.16 for causing cardiac-related morbidity [54]. However, sibutramine can be found in small amounts within some diet pills and herbal remedies, which can be purchased online [55]. Therefore, it is important to know its possible impact on young people as well as using it as a historical benchmark to compare against current licenced medications.

Surprisingly, dietary interventions were rarely evaluated as a sole component of treatment in comparison with a minimal-intervention or no-treatment control group [56]. Therefore, many trials did not meet the inclusion criteria for statistical analysis in this study. Nevertheless, the available evidence demonstrates that dietary interventions do have the potential to cause a statistically significant decrease in BMI. Partsalaki et al. compared the efficacy of a ketogenic and a hypocaloric diet in obese children without the use of a control group [57]. At 6 months, both groups had a statistically significant decline in mean BMI of 3.7 and 3.3 kg/m², respectively. This illustrates a higher BMI reduction compared to RCTs included in this study. There is a clear need for well-designed RCTs that assess the long-term efficacy of alternative dietary interventions to allow conclusive determinations regarding its efficacy.

There is a lack of information from RCTs on sibutramine and its impact on BMI change after 6 months. A prospective observational study demonstrated that a large dropout rate was seen due to a slow rate of weight reduction after 6 months compared to the first 6 months of administration [47]. It was concluded that sibutramine has a limited effect on weight loss in the long term [47].

The surgical meta-analysis included non-randomised trials (Fig. 4) making it difficult to directly compare the efficacy of surgery to other interventions. The studies do not take into account the placebo effect and allow bias from potential confounding factors [58]. However, the lack of RCTs reflects the ethical concerns of randomising a vulnerable population, in this case children and adolescents with obesity [52]. Surgery includes a higher risk of complications and a greater cost, making it difficult to carry out RCTs in this area [59].

Another limitation includes the use of BMI to measure weight change in a paediatric population [59]. BMI can be inaccurate, as the calculation does not distinguish between lean and fat mass [59]. Moreover, variation in age, sex and maturation results in a large range of normal BMI scores in the paediatric population [59]. BMI-Z score would be a better measure of weight change as it considers age and gender [59]. However, studies most commonly reported BMI, and therefore, it was the used metric for this analysis.

The results from this meta-analysis apply to the general paediatric population of overweight and obese individuals below the age of 21 with a BMI of 25 kg/m² or more. In the included studies, mean age ranged from 4.8 to 19 years. As a whole, the patients undergoing bariatric surgery were older than those carrying out lifestyle modifications. This may have influenced findings. However, the limited number of studies prevented investigations through further stratification of results according to age groups.

Finally, a lack of studies with large sample sizes is a problem when evaluating the efficacy of orlistat and endoscopic treatment. It is important that more studies are carried out to further assess their impact on weight loss.

McGovern et al. (2006) concluded that a combined lifestyle intervention is more effective than a single lifestyle intervention approach on BMI change and that pharmacological therapy (for example, sibutramine) can be beneficial in the short term [60]. Our study confirms this. We also extend this inference to note that overall lifestyle and pharmacological interventions have very similar impacts on BMI reduction, while metformin demonstrates similar effects on BMI reduction regardless of the time interval for which it is administered. Our results are also consistent with another review that found that metformin did not demonstrate any clear benefit on weight loss after long-term administration [61].

Our surgical meta-analysis contained solely non-randomised trials; however, its results mirrored the only published RCT in this area [62]. This RCT of 50 patients illustrated that the surgical arm reduced mean BMI by 12.7 kg/m2 (95%CI − 11.3 to − 14.2), compared to a reduction of 1.3 kg/m² (95%CI − 0.4 to − 2.9) in the lifestyle modification control arm [62].