Impact of Dietary Carbohydrate Content on Physical Activity-Associated Hypoglycemia in Individuals with Type 1 Diabetes Mellitus

Abstract

Type 1 diabetes mellitus (T1DM) is characterized by absolute insulin deficiency. Despite major therapeutic advances, glycemic control often remains suboptimal, placing millions at risk for complications. Glycemia surrounding physical activity (PA) is especially challenging; hypoglycemia— arising from PA-related increased glucose disposal—can have detrimental consequences. This is frequently managed with supplemental glucose ingestion, reducing some of the health benefits of exercise and often resulting in hyperglycemia. The impact of dietary carbohydrate content in this setting is poorly understood. The common need for supplemental glucose raises concern that a very low carbohydrate diet (VLCD, or ketogenic) may increase hypoglycemia risk. Conversely, metabolic adaptations on this diet may be protective, due to the availability of a different fuel source—fatty acids. To clarify the impact of dietary carbohydrate on PA-associated hypoglycemia, we performed a scoping review of the literature and conducted a 12-week randomized pilot study to test the hypothesis that a VLCD would reduce PA-associated hypoglycemia. For the scoping review, we systematically searched the PubMed, Embase, Web of Science, and CINAHL databases for peer-reviewed studies that report primary data addressing T1DM, dietary carbohydrate intake, PA, and hypoglycemia or glycemic variability surrounding PA. Of 65 unique records found, only 3 met inclusion/exclusion criteria. A supplemental Google Scholar search with screening of the first 50 results found no records meeting inclusion/exclusion criteria. The 3 studies meeting criteria exhibited methodologic heterogeneity, included small numbers of participants, and had variation in reporting of outcomes. No studies definitively addressed the research question. In the pilot randomized study, participants received food deliveries following a standard (n=4) or VLCD (n=3) diet. PA and glycemia were monitored using accelerometers and continuous glucose monitors. Using a mixed effects Poisson regression, no statistically significant effect of diet, physical activity as approximated by step count, or their interaction on frequency of hypoglycemic episodes (primary endpoint) was found (incidence rate ratios and 95%CI respectively: 0.95 [0.25,1.24], p=0.95; 1.02 [0.83,1.24], p=0.87; 0.95 [0.67,1.34], p=0.76). Analyses of secondary endpoints showed statistically significant reductions of frequency and amount of supplemental carbohydrate intake and % coefficient of variation (%CV) in the VLCD group. However, increasing step count was associated with increases in supplemental glucose intake and %CV on the VLCD relative to the standard carbohydrate diet. Together, these investigations show a notable gap in published evidence and can inform the design of definitive studies on this important topic. Our scoping review was limited by the paucity of identified literature; a broader scoping review could be conducted to further query the existing literature. Limitations of our pilot study include the small sample size limiting statistical power to address the research question conclusively. For future clinical trials on this topic, we suggest a crossover design given the large observed inter-individual differences in frequency of hypoglycemic episodes. There remains equipoise regarding the effects of dietary carbohydrate content on PA-associated hypoglycemia in individuals with T1DM, warranting further research.