Bariatric surgery is one of the most effective interventions to achieve long-term weight loss in patients with severe obesity, but frequently leads to serious side effects(1–3). One of them is postprandial hypoglycemia (PBH), previously referred to as late-dumping, affecting in standardized test settings about 50% of patients with a gastric bypass(4–6) although it is frequently undiagnosed(7). This may lead to a reduced quality of life(8), secondary weight gain(9, 10) and may even be associated with an increased mortality(11–13). Serious hypoglycemia requiring assistance of third parties and/or hospitalization for hypoglycemia affects around 1%(8, 14). Although some patients are able to effectively control hypoglycemic episodes with dietary modification, there is no established treatment for severely affected patients with frequent postprandial hypoglycemia(5).
A recent exploratory trial with 12 patients after gastric bypass indicated that both the SGLT2-inhibitor empagliflozin and the IL-1 receptor antagonist anakinra may reduce postprandial insulin release and prevents hypoglycemia(15). These findings call for further evaluation in a larger randomized clinical trial.
We aimed to develop an evidence-based study design for such a trial that tests these treatment options and can provide useful patient-relevant evidence on their merits. However, selecting a primary outcome that appropriately reflects the needs of patients turned out to be a major challenge. Ideally, such a trial would explore patient-relevant outcomes but also improve the understanding of the underlying metabolic effects of the tested treatments that would mediate potential patient-centered benefits (such as glucose levels).
Clear frameworks such as those established for assessment of anti-diabetic treatments that may help to utilize hypoglycemia as trial outcome in non-diabetic persons are lacking. Extrapolation of such frameworks to nondiabetic patients may be inappropriate and would require strong assumptions as changing glucose levels in diabetic and nondiabetic persons may correspond to different symptoms and health effects.
For postprandial hypoglycemia, no core outcome set (COS), as “an agreed standard set of outcomes that should be measured and reported, as a minimum, in all clinical trials in specific areas of health or healthcare”(16) has been published. In 2016, the BARIAtric and metabolic surgery Clinical Trials (BARIACT) project defined COS for patients undergoing bariatric surgery based on patient and healthcare provider perspectives. However, postprandial hypoglycemia was not among the list of candidate outcomes includeed in the delphi processes(17).
We systematically searched in September 2019 for clinical trials specifically including patients with symptomatic postprandial hypoglycemia and found only small trials using biomarkers (e. g. time of glucose levels below a specific threshold)(18, 19) or clinical scoring systems (e. g. Edinburgh Hypoglycemia Scale) mostly within standardized mixed meal test-settings(20–22) which do not reflect real-life circumstances and provide no patient-relevant information. Only one trial that we identified reported effects on quality of life as a secondary outcome(23).
The measurement of hypoglycemic episodes by continuous glucose measurement systems appears to accurately depict real-life glucose level variation and may give an objective measurement of the metabolic effects of an intervention aiming to alleviate symptomatic metabolic disorders(24). However, it is unclear how relevant it is for patients with symptomatic hypoglycemic episodes to measure merely metabolic changes, since some hypoglycemic episodes may not be recognized (7, 25, 26) and have varying symptomatic presentation, especially when occurring frequently or during the night(7). Measuring the frequency of hypoglycemic episodes alone without consideration of their severity and patient-reported symptoms seem inadequate to reflect patients’ needs. Furthermore, using the classical Whipple's triad (i.e. appearance of typical symptoms, low plasma glucose, and relief of symptoms following glucose administration) would be unfeasible in a real-life evaluation. We, therefore, sought to identify the outcome parameter that is most relevant for patients and their subjective wellbeing.