Investigating Novel Predictions of Hypoglycemia Occurrence in Real-world Models
Investigating Novel Predictions of Hypoglycemia Occurrence in Real-world Models
Hypoglycemia is the most common diabetes-related adverse event. However, it is often under-reported to healthcare providers by patients and simultaneously, not often asked about by healthcare providers. As a result, little is known about how often hypoglycemia occurs and consequently, which individuals with diabetes will experience such events. The aims of this study are to determine the real- world occurrence of hypoglycemia and develop/validate real-world risk prediction models for hypoglycemia. These risk prediction models will generate a risk score that indicates an individual's risk for hypoglycemia given their socio-demographic, clinical, and/or behaviour-related characteristics. They can be used to promote clinician awareness around patients' hypoglycemia risks, guide point- of-care and patient decision-making with regard to treatment changes, inform the development and conduct of population-based interventions, and lead to tailored, cost-effective management strategies.
The overarching purpose of the proposed investigation is to develop and validate three real-world risk prediction models for: 1) severe hypoglycemia, 2) non-severe daytime hypoglycemia, and 3) non-severe nighttime hypoglycemia, that are applicable to the general population with diabetes (Type 1 and Type 2). These prediction models, which will produce risk scores, will be generated using long-term, prospective data on the frequency and multidimensional risk factors of real-world hypoglycemia. Self-reported hypoglycemia data - a pragmatic and significant patient-important outcome in the clinical management of diabetes - will collected in a non-clinical setting as they are crucial to determining the true distributional burden of events and impactful avenues for prevention, especially given the known epidemiological challenges of existent data collection strategies (e.g., via RCT- or registry-based designs). The use of real-world data will also enhance the generalizability and thus, clinical value of hypoglycemia risk prediction models.
The study will employ an ambidirectional (one-year retrospective and one-year prospective) observational cohort design such that multiple exposures (i.e., risk factors) will be collected and evaluated in relation to the occurrence of an outcome (hypoglycemia events). Participants will be enrolled into a prospective, observational cohort referred to as the 'Diabetes iNPHORM Community'. Data will be collected through online questionnaires administered at baseline (to collect retrospective data) and each month of the one-year prospective period. A pilot test will be conducted prior to the enrollment of participants into the Diabetes iNPHORM Community. The purpose of this pilot test is to test the usability of the online question platform, flow and format of the questionnaires, and the readability of the questions.
Participants will be recruited into the pilot test and the observational cohort of the study from a pre-existing online panel representative of the general public that has been developed and managed by Ipsos Interactive Services (IIS), a global leader in survey conduct. All individuals in the pre-existing online panel provided profile information and consented to be approached by IIS and its subsidiary partners to complete surveys. For this study, individuals approached to participate in the pilot tests will not subsequently be invited to participate in the observational cohort.
Inclusion Criteria:
Exclusion Criteria: