• path analysis
• economic evaluation
• bariatric surgery
• quality of life
• regression tree

The present thesis aimed at introducing and analyse QoL as additional measure for outcome assessment in the evaluation of clinical “problems” – thus adopting a multidisciplinary perspective - with practical application in the context of obesity and particularly in the evaluation of bariatric surgery procedures. Moreover the thesis aimed at presenting statistical models able to deal with huge data and to inform clinical practice.
The first two Chapters introduced the concept of quality of life and outlined the importance that its assessment – by means of instruments some of which are presented in Chapter 3 – has gained in many fields, and particularly in medicine and public health, during the last decades.
Whit these premises and the introduction of the obesity context (Chapter 4), in which, despite the flourishing of disease-specific instruments (Chapter 5), the effect of treatment on HRQoL is still unclear (Chapter 6) because HRQoL assessment in obesity is even recent and not yet standard practice treatments evaluation, the thesis outlined the potentiality of the proposed multidisciplinary approach as well as statistical methods in the evaluation of bariatric surgery.
In a population of obese subjects undergoing different bariatric procedures (GB, RYGB and SG) over six centres in two Italian regions, the present study demonstrated that 12 months after surgery the majority of patients reach desirable weight loss (as evaluated in terms of %EWL) or improvements in HRQoL. Anyway, although more than half of the subjects (56%) achieved both successful weight reduction and improvement in HRQoL, about 44% of the population examined did not reach both goals.
Results also showed that the correlation between excess weight loss after surgery and changes in HRQoL, as measured by the SF-36, but the degree of associations was absolutely weak. This confirms already available evidence thus supporting the fact that weight loss do not necessary meant also HRQoL improvement, i.e. physiological results do not necessarily met with patient perception of the effect of surgery.
Indeed, the path analysis conducted to evaluate the determinants of HRQoL changes at 12 months after surgery demonstrated that HRQoL is partly explained by %EWL but also other individual and clinical variables, particularly comorbidities, contributes directly or indirectly to the variability of HRQoL.
According to this figure it became evident the need for exploring a more comprehensive measure for outcome assessment following bariatric surgery.
Particularly, to offer a more general evaluation of outcome 1 year after surgery a cluster analysis was conducted to identifying subgroups of subjects with similar characteristics with respect to the three main goals of bariatric treatment: %EWL, comorbidities and HRQoL. Results from cluster analysis showed the presence of two subgroups having different 12-months outcome being characterized, on one side by generally good results on the three treatment goals meaning satisfactory %EWL, absence of comorbidities and good HRQoL, the opposite in the other subgroup.
The prediction tree adapted to assess the probability of pertaining to one group rather than the other, that means satisfactory vs unsatisfactory results of surgery, demonstrated that few baseline characteristics could predict 1 year outcome: PCS, the presence of anxiety and depression, BMI level and the type of surgery. Combination of these variables may help determining the probability of success of a given patient guiding clinical decision and favouring the plan of possible actions to avoid failure.
Subgroups of subjects with good/poor outcome identified in the analysis also differ in terms of associated costs with subjects showing unsatisfactory outcome revealing generally increasing level of costs.
Costs associated with bariatric surgery, despite trying to remedy to a condition – obesity - that is extremely costly for the Healthcare System but also for the society, are not negligible and may be acceptable if they are projected to induce savings. Surgery failure thus meant damages for subjects, that will be not restored from their problems, but it also implies additional costs for a pathology that is even expensive.
Properly targeting surgery and using evidence to eventually plan additional or alternative interventions for those subjects being at risk of failure is thus necessary both for patients than for the Health System and the society too. Producing evidence about partial results of surgery by focusing only on weight loss or comorbidities resolution without combining the real perception of the patient may thus be unhelpful.
Despite the effort in using a big set of variables to explore HRQoL, objective clinical outcome and costs of bariatric surgery, that is one of the major advantage of the present study together to the adoption of statistical models able to deal with big data, several limits need to be outlined.
First of all follow-up duration is quite limited and do not allow for the examinations of long-term results that actually constitutes one of the main focus of clinicians.
Moreover some clinical parameters, such as, bio-chemical data that would help the evaluation of physiological profile both at baseline and at follow-up, have not been collected being the study mainly oriented for the economic evaluation.