**4. Discussion**

In this study we observed that a high attrition rate was the most relevant and limiting factor in our outpatient specialized assistance pediatric obesity clinic, with a very high drop-out rate in the early stages of intervention as well as a low mean duration of followup. We have seen how the fulfillment of behavioral counselling and the attainment of BMI reduction occurs most frequently in the first year of intervention, with a significant percentage of patients being very successful, and with metabolic improvement being attained independently from the time required to achieve weight loss and that can be sustained if follow-up is retained. We have also seen that controlled intervention does not affect growth or pubertal development, nor does it impair the attainment of the adult target height, although the timing and pace of growth are influenced by obesity and its severity. Finally, we saw that insulin resistance is related to metabolic status in patients with obesity, even before the onset of other metabolic alterations, and should be considered when defining whether a person is metabolically healthy, especially when considering that in childhood obesity the metabolic statis is a dynamic condition related to the evolution of the patient's BMI over time.

This study, similar to most preceding reports, highlights the evidence and relevance of the high attrition rate in intervention programs for childhood and adolescent obesity [8–11,28], with the number of patients in follow-up declining over time. Here, we found that 43.7% dropped out before their third visit and only 7% extended their visits up to 7 years. Around 60% of patients unilaterally decided to stop the follow-up. As might be expected, 84.1% of them had not fulfilled the therapeutic recommendations in their visits prior to withdrawal, whereas only 5.4% stated that they had difficulties to attend the programmed visits.

Several factors have been analyzed to predict and avoid a patient's dropout, including misperception of disease status, ethnicity, socio-economic or cultural determinants or unavailability to attend the visits [7,9–11,29]. Among these possibilities, misperception of the parents and children of their weight status or the conception that obesity is not a real disease is particularly important in our [30] and most western environments [31,32]. This underestimation of the pathogenicity of childhood obesity could, at least in part, explain why 11.2% of the patients in our cohort did not even perform the complementary test requested and withdrew after their first interview and clinical evaluation, even though they had been referred for specialized care by their primary care physician. This limited parental and child concern could also be involved in the high rate of patients not fulfilling the recommendations while in follow-up prior to attrition. Additionally, a large percentage of the population assumes that it is the onset of metabolic comorbidities, particularly type 2 diabetes or dyslipidemia, but not weight excess itself that confers the pathogenic potential to obesity [9]. This could influence the additional 32.5% of patients who dropped out at their second visit after getting the results from their metabolic evaluation.

This degree of acceptance of obesity is further enhanced if a positive background of familial obesity exists [31], with more severe obesity and higher prevalence of comorbidities observed in the offspring of parents with obesity at the time of consultation [33], with this being closely related to ethnic, socio-economic and cultural factors [31]. Although the parental academic level background distribution in our cohort was similar to that of the general population in our country [34], no family economic data were registered. These factors could potentially underly the significantly higher attrition rate observed in Latino patients compared to Caucasians at early stages of intervention.

Interestingly, up to one third (33.5%) of the parents asked to discontinue the visits, most after the children had improved their BMI or metabolic comorbidities. This early withdrawal from obesity intervention programs is thought to be a result of the misperception that the obesity was cured [35], which can result in weight regain and later re-consultation as observed in 8% of our patients, most of them after having regained previously lost weight.

Even assuming the high drop-out rate and the positive selection bias derived from a higher attrition rate in those patients not fulfilling therapeutical recommendations, mean follow-up duration in our cohort was similar to that reported in other long-term follow-up series [36], with a retention rate of 54.5% after 1 year; with 72% of the patients showing some BMI reduction (mean over −1 SDS) and with almost 20% being excellent responders (75% of these also in the first year). Subsequently, 31.6% and 15.4% of the cohort extended their follow-up over 2 and 3 years, respectively, resulting in a reduction in mean BMI at the end of follow- up of over 1 SDS in 62% of patients. The degree of fulfillment of the main behavioral recommendations, snacking avoidance, control of compulsive eating and scheduled physical activity, followed a parallel pace to patients' BMI evolution, significantly improving as early as the second visit and remaining stable to the end of follow-up. This reinforces the relevance of the changes attained in the early stages of the intervention to its final outcomes.

Our data show a higher retention rate and success in weight loss in this cohort compared to previous series [8,36], even though the proximity to the patient's home and the possibility of a greater frequency of contact, reported to positively influence behavioral outcome [29,37], is limited at the tertiary care level. These results could be influenced by the relevance that the parents place on being sent to specialized care by the primary care physician, the reduced mean age (10.46 years), the severity of the patient's obesity (mean BMI above +4 SDS) and the high prevalence of metabolic comorbidities in our cohort, all of which could enhance the parental perception of disease in their offspring. Consequently, this could underly the higher rate of excellent responders among prepubertal patients that can be more influenced by their parents' concerns compared to pubertal patients. Additionally, the development of the intervention in a socialized national healthcare system, with no economic factor biasing visit schedule, and the designation of the same physician for the successive visits of every patient could, among other factors, increase their loyalty and the retention rate once follow-up is settled [38].

The combined analysis of these data suggests that the first year of intervention is crucial to the final outcome and efforts should be focused on the attainment of the maximum BMI change in the first year of follow-up, as this increases the chances of sustaining the achieved weight loss, at least in the following 5 years, as seen in the excellent responder group. However, early weight loss can result in a "double edged sword"; that is, it can be the first step for sustained BMI improvement, but also can determine a misconception of definite success resulting in follow-up withdrawal and weight regain due to the return to unhealthy behaviors, which should be prevented [35]. Thus, rapid weight loss may not be the best option for every patient and an individualized strategy regarding the amount and pace of weight loss should be agreed upon in each singular case. This is supported by two observations: (1) among the excellent responders no correlation was observed between time to attain weight reduction and the magnitude of metabolic changes; (2) the excellent responders that required a longer time to achieve weight loss, thus reducing the influence of weight loss on growth, or those patients achieving a more modest weight reduction, independently of the time spent to attain it, also showed significant metabolic improvement.

Consequently, efforts must be made for the continuation of follow-up to consolidate the decrease in BMI, as its duration is an independent predictor of success [39]. Indeed, a duration of at least three years of follow-up has been proposed [17] and it is consistent with the maximum follow-up time observed for the majority of patients in our cohort. However, at the end of the intervention, only 6.6% of the patients normalized their BMI and 10.3% shifted to overweight, with 82.1% of them remaining above the threshold of obesity

(+2 BMI-SDS), which emphasizes the difficulties with reverting to this chronic condition and suggests that the coordinated assistance between specialized care (possessing resources and specialized units) and primary care (having proximity and accessibility) could enhance and prolong the benefits of obesity-oriented interventions [40].
