Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article

Degree

Doctor of Philosophy

Program

Epidemiology and Biostatistics

Supervisor

Harris, Stewart B.

2nd Supervisor

Klar, Neil S.

Co-Supervisor

Abstract

Clinical outpatient strategies to accurately predict diabetes-related iatrogenic severe hypoglycemia (SH) are lacking. To redress this gap, we conducted the first-ever prognosis investigation of guideline-defined (Level 3) SH in the United States (US) (iNPHORM).

Chapter 4 details the design and implementation of iNPHORM: a prospective 12-wave panel survey (2020–2021). N=1206 adults with type 1 or insulin- and/or secretagogue-treated type 2 diabetes mellitus (T1DM or T2DM) were recruited from a US-wide, probability-based internet panel. For one-year, we collected monthly data on SH occurrence (frequencies, detection methods, symptoms, causes, and treatments) and related factors (anthropometric, sociodemographic, clinical, environmental/situational, behavioural, and psychosocial).

iNPHORM data were analyzed in Chapter 5 to characterize and quantify Level 3 SH (N=978). Overall, 60% of events were treated outside the healthcare system; <5% required hospitalization (T1DM: 1.6%; T2DM: 4.9%, p-value=0.0014, α=0.0083). About one-third of participants experienced ≥1 event(s) over prospective follow-up (T1DM: 44.2% [95% CI: 36.8% to 51.8%]; T2DM: 30.8% [95% CI: 28.7% to 35.1%], p-value=0.0404, α=0.0007). The incidence rate was 5.01 (95% CI: 4.15 to 6.05) events per person-year (EPPY) (T1DM: 3.57 [95% CI: 2.49 to 5.11] EPPY; 5.29 [95% CI: 4.26 to 6.57] EPPY).

Chapter 6 describes the development and internal validation of the iNPHORM prognostic model. We modelled one-year recurrent Level 3 SH using Andersen-Gill Cox proportional hazards and penalized regression with multiple imputation (N=986). A range of anthropometric; sociodemographic; and clinical (diabetes-, hypoglycemia-, and general health-related) candidate variables were selected for their relevance and feasibility. The final model demonstrated strong discriminative validity and parsimony (optimism corrected c-statistic: 0.77).

The results of this dissertation promise to enhance real-world SH screening; evidence-based, risk-tailored prevention; and ultimately cost containment.

Summary for Lay Audience

Certain diabetes medications can make a person’s blood sugar drop too low. This condition, known as hypoglycemia, can occur frequently and without warning. Hypoglycemia can trigger symptoms like sweating and shakiness. In very severe cases, events can cause confusion and clumsiness, seizures, coma, and even death. Nevertheless, little is known about who is most likely to get severe hypoglycemia and how often. Such insight could help clinicians deliver better diabetes care that is not only effective but also safe.

For this dissertation, I designed and carried out the first-ever long-term research project on self-reported severe hypoglycemia in the United States, called the iNPHORM study. Over the course of one year, our team at Western University emailed monthly questionnaires to 1206 adult Americans with type 1 or type 2 diabetes mellitus at-risk of hypoglycemia. The questionnaires asked respondents about how often they experienced low blood sugar. We also collected information on various clinical and socio-demographic traits. Based on these data, we analyzed 1) the total number of severe hypoglycemia events, and 2) the factors associated with event occurrence (i.e., predictors).

Our study showed that severe hypoglycemia is alarmingly common among Americans with diabetes. After one year, about a third of participants reported at least one severe hypoglycemia event and had, on average, five events per person-year. In total, 60% of events were treated outside the healthcare system and less than 5% required hospitalization.

To identify the predictors of severe hypoglycemia, we used a statistical method called prediction modelling. Our analysis linked higher severe hypoglycemia risk to a range of different predictors including diabetes type and duration, medication type, age, sex, marital status, race, and general health. The results of iNPHORM will be used to create a tool that can predict severe hypoglycemia risk during routine medical appointments. Clinicians could use this tool to adjust treatment and care so that, in the future, severe hypoglycemia happens less often, or not at all.

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Download

Share

COinS