Sample size. To achieve the main objective of this study (i.e. to assess weight reduction in patients with DM2 on semaglutide vs patients on other GLP-1 agonists), it is expected that 360 patients will be included during the study period. It has been estimated that approximately 36 patients per centre (10 centres) will be enrolled over a 6-month period. This estimate is based on previous data on the number of monthly approvals at participating centres. Previous studies11–15 have assessed weight loss in patients with DM2 treated with semaglutide and other GLP-1 agonists, and have described mean weight losses of 3.6 to 4.9 kg in patients on semaglutide vs 0.86 to 2.96 kg in patients on other GLP-1 agonists. Based on these data, and assuming a P-value of 5% as a cutoff for statistical significance, a mean weight loss of 2.5 kg in the GLP-1 agonist group, a mean weight loss of 4.2 kg in the semaglutide group, and a pooled deviation of 3.0 kg, the inclusion of 360 patients will achieve a statistical power of more than 90% to detect differences using the Student t-test for independent samples. On the other hand, assuming a possible loss to follow-up of up to 20%, the sample could be reduced by up to 72 patients. In this situation, the statistical power to detect differences in weight loss between the study groups would still be more than 90%. The SPSS 3.0 (Chicago, USA) statistical package was used to estimate statistical power based on the sample size.

Qualitative variables will be expressed as frequency and percentage and quantitative variables will be expressed as mean and standard deviation or as median and interquartile range if they do not conform to normal distribution. Normal distributions will be determined using the Kolmogorov-Smirnov test.

The overall objective will be assessed according to weight loss in each group. Other variables associated with weight loss will be analysed using the Student t-test for quantitative variables and the chi-square test for qualitative variables. Multivariate linear regression analyses will be used to adjust for weight loss in relation to the variables shown to be statistically significantly associated in the univariate analysis. Similarly, the specific objectives will be investigated using univariate and multivariate analyses. The coding variables will be categorical, the Student t-test or ANOVA will be applied to continuous variables, and the chi-square test will be applied to categorical variables. Multivariate logistic regression will be conducted using significant variables.

Several clinical trials have been conducted on GLP-1 agonists. All these trials have demonstrated their effectiveness in reducing HbA1c in patients on monotherapy or on combinations with other oral antidiabetic drugs and/or insulin. Dulaglutide was associated with a weight reduction of –0.35 kg to –2.90 kg8. In the case of exenatide, 3% of patients experienced at least one period of rapid weight loss (greater than 1.5 kg/wk)8. In the case of liraglutide, weight reduction was more significant when the baseline BMI was higher8. Lixisenatide was associated with a weight reduction of 1.76 kg to 2.96 kg8. Finally, semaglutide was associated with a weight losses of at least 5% and at least 10% in more patients than in those on the active comparators dulaglutide and exenatide, respectively (SUSTAIN 7 and SUSTAIN 3)12–13. In the SUSTAIN 69,10 trial, weight reduction was –3.6 kg to –4.9 kg. Semaglutide was approved by the FDA for the treatment of obesity in diabetic and non-diabetic patients following the publication of the results of the STEP1 study, which found a weight reduction of –14.9% and a change in baseline weight of –15.3 kg10. Subsequently, the STEP 8 trial (weekly subcutaneous semaglutide vs liraglutide) found a weight reduction of –15.8% with semaglutide vs –6.4% with liraglutide11.

After assessing these data, it can be concluded that all GLP-1 agonists reduce body weight to varying extents: however, the clinical trials that led to their approval have some limitations that could affect the external validity of the results and their extrapolation to real clinical practice. Each of these trials had different study populations, followed different methodological designs, and differed in relation to the primary endpoint. With the exception of the STEP 1 study10, most of these trials were designed to assess HbA1c reduction and cardiovascular events as endpoints, but not to assess weight reductions as endpoints. In contrast to the SEVERAL study, no other study has compared the effectiveness of all GLP-1 agonists.

These weight losses have been accompanied by specific adverse events, the most common being gastrointestinal adverse events, such as nausea, vomiting, diarrhoea, dyspepsia, and constipation8. The most recent study on this issue, STEP 811, found that treatment discontinuation with semaglutide was 13.5% and 27.6% with liraglutide. Gastrointestinal events were reported in 84.1% of patients on semaglutide and 82.7% of patients on liraglutide. Most events were mild to moderate, did not lead to permanent discontinuation, were mainly of short duration, and occurred during dose escalation. These results may indicate a lack of adherence in real life: that is, outside the setting and control to which patients included in clinical trials are exposed. The foregoing suggestion is one of the main motivations for this study (i.e. the assessment of the safety of these drugs in real life).

The SEVERAL study has been designed to collect data on all possible adverse events during the dose escalation and follow-up periods. Data on quality of life and physical activity will also be collected and correlated with potential therapeutic failure and clinical effectiveness.

The present study has several limitations that must be taken into account when interpreting the data. The first limitation is the lack of randomisation, which means that comparative conclusions cannot be drawn: thus, the study merely provides guidance on the current situation, prescription rates, results, and tolerability. If marked differences between classes were to emerge from the present study, a randomised controlled trial could be considered.

Secondly, there is a potentially relevant bias arising from possible loss to follow-up, such that patients with poorer drug tolerability may drop out of the study, thus distorting the results. To avoid this bias, possible loss to follow-up was limited to < 20% to calculate sample size. All patients in our Autonomous Community have electronic records, and so the electronic prescription can be checked for medication collection, which will be of great help in validating adherence (as is done in daily clinical practice).

It is worth mentioning that the present working group, which is exclusively made up of professionals from pharmacy units, will minimise inclusion losses as they represent the largest nucleus of dispensing control. Thus, possible bias can be avoided had this study been initiated, for example, by cardiology or endocrinology units, where it would have been more difficult to ensure the correct inclusion of all patients who are prescribed GLP-1 agonists in our Autonomous Community.

In conclusion, this study will attempt to provide information on the effectiveness of GLP-1 agonists on weight reduction and on changes in quality of life, physical activity, and safety in patients with DM2 starting on treatment with these drugs in real life. This study attempts to compare different GLP-1 agonists in terms of effectiveness and safety in order to make better choices in the prescription of these drugs in patients with cardiovascular disease.