Type 2 diabetes (T2D) patients often complain about sleep difficulties (1). For instance, about 77% regularly report insomnia symptoms, such as difficulty falling and staying asleep (2). According to the International Classification of Sleep Disorders criteria (3), having trouble falling asleep at night and waking up in the middle of the night constitute key insomnia symptoms. In addition to insomnia, T2D patients often report habitual short (usually defined as <7 hours per day) or long sleep duration (often described as >9 hours per day) (4).

Several factors may account for sleep difficulties among T2D patients, including but not limited to comorbidities such as overweight and hypertension (1). As suggested by previous studies, the type of oral antidiabetic agent used to prevent hyperglycemia may also account for sleep difficulties among T2D patients (5–7). For example, by using one-night sleep polysomnography, a study among 387 outpatients with T2D found that those on metformin slept on average about 36 minutes longer and exhibited a 6.4% higher sleep efficiency compared to patients on non-metformin (mainly sulfonylureas) (5). Metformin, a biguanide, suppresses hepatic glucose production and intestinal glucose absorption and promotes β-cell functions and insulin sensitivity (8–10). Sulfonylureas lower blood glucose by stimulating insulin release from pancreatic β-cells (11, 12).

Besides the mechanism of action, oral antidiabetics can also differ in the type, frequency, and severity of side effects (13–19). For instance, hypoglycemia represents the most common side effect of sulphonylureas (17). Diarrhea, nausea, vomiting, abdominal bloating, abdominal cramping or pain, flatulence, and anorexia, occur in approximately 10% to 30% of metformin recipients (18, 19). Side effects caused by oral antidiabetic therapy may lead to sleep difficulties, primarily if occurring during sleep.

In the present study, we wanted to examine the association of oral antidiabetics with sleep among T2D patients from the UK Biobank. Specifically, we used cross-sectional data from 11 806 T2D patients to determine whether the ability to initiate or maintain sleep and habitual sleep duration would vary by the type of oral antidiabetic agent used.

Baseline characteristics are summarized in Table 2 . In the full cohort (n=11 806), ~35% of the patients reported that they usually struggle with falling or staying asleep. As shown in the fully-adjusted logistic regression ( Figure 1 ), patients using non-metformin agents (n=815) had a 1.24-fold higher odds ratio of reporting regular difficulty falling and staying asleep at night than patients without antidiabetic medication use (n=5 366; P=0.007). Additional multivariable analysis revealed that patients using non-metformin had higher odds of having trouble falling and staying asleep when using metformin patients as the reference group (n=5 625) (OR[95%-CI], 1.244 [1.064,1.455]; P=0.006). Non-metformin patients also reported longer sleep duration than untreated and metformin patients (P=0.004 for the main effect of treatment group; derived from the fully-adjusted generalized linear model; see Figure 2 for group comparisons). Comparisons between metformin patients and those without antidiabetic medication use did not reveal significant group differences ( Figures 1 , 2 ).

There is evidence that oral antidiabetics may impact sleep; however, with mixed results. One study reported longer and more efficient sleep among T2D patients on metformin than those using non-metformin agents (5). In contrast, two other studies found that metformin may be associated with poor sleep quality and frequent nightmares (6, 7). In the present study, including 11 806 patients with T2D, we found that those treated with non-metformin were more likely to report difficulty falling and staying asleep at night compared to those without antidiabetic medication use or those on metformin monotherapy. More than four out of five patients in the non-metformin group were on sulfonylurea monotherapy. Hypoglycemia, the most common side effect of sulphonylureas (17), can induce counterregulatory stress hormone responses (28). If occurring at bedtime or during sleep, these hypoglycemia-induced endocrine stress responses may cause hyperarousal and, therefore, impair the ability of the patient to fall and stay asleep. However, due to the lack of hypoglycemia screening in the UK Biobank, this explanation remains speculative. We also found that patients on non-metformin slept about 8 to 10 minutes longer than those on metformin or without antidiabetic treatment. Whether the small sleep extension was due to accompanying sleep initiation and maintenance difficulties among non-metformin patients is unclear.

Several limitations apply to the present study. It remains unclear whether factors such as antidiabetic dosing and timing of medication administration may modulate the association of oral antidiabetics with sleep. Furthermore, the UK Biobank investigation did not survey adverse side effects of oral antidiabetics, which could explain differences in sleep outcomes between metformin and non-metformin users. In T2D patients with low kidney function, physicians often prescribe sulfonylureas (29). Chronic kidney disease has been linked to poor sleep quality (30). Importantly, only three of 11 806 patients had chronic kidney disease at the baseline investigation in the present study. Thus, it is unlikely that the observed sleep difficulties among patients treated with non-metformin are attributable to chronic kidney disease. Furthermore, in the present study, only a few patients reported the use of sleep-promoting sedatives. However, we could not control our analysis for the possible use of sleep-promoting over-the-counter-remedies (e.g., melatonin). The present study was further limited by the small sample size of patients treated with non-metformin drugs. Thus, future studies with larger sample sizes, including novel types of non-metformin oral antidiabetics than those investigated herein (e.g., dipeptidyl-peptidase 4 inhibitors, glucagon-like peptide-1 receptor agonists), are needed to advance our understanding of how oral antidiabetics impact sleep among patients with T2D. Considering that most T2D patients utilize multiple therapies to manage their glycemic control in the long term, it may also be worth investigating possible interactions of antidiabetic drugs on sleep. Finally, sleep duration was derived from self-reports, which might be subject to potential recall bias. Thus, our results must be replicated in studies using longitudinal measures of sleep duration (e.g., sleep-tracking wearables) and validated questionnaires for insomnia (e.g., insomnia severity index).

Despite these limitations, the primary strength of our study is that our results were based on one of the largest cohorts worldwide. Furthermore, findings were robust to adjustment for important confounders, such as snoring, age, BMI, blood pressure, and HbA_1c.

Recurrent problems with falling and staying asleep have been associated with impaired glycemic control among patients with T2D (31–33). To mitigate these possible adverse consequences, T2D patients on non-metformin should regularly participate in sleep screenings. As shown herein, they more often suffer from sleep initiation and maintenance difficulties.

The datasets presented in this article are not readily available because data may be obtained from a third party and are not publicly available. Data were derived from the UK Biobank investigation. Thus, data may be obtained from the UK Biobank upon request. Requests to access the datasets should be directed to https://www.ukbiobank.ac.uk/.

The studies involving human participants were reviewed and approved by National Health Service National Research Ethics Service (ref. 11/NW/0382). The patients/participants provided their written informed consent to participate in this study.

CB and PX designed the study. PX and JW performed analysis. CB and PX drafted the manuscript. All authors interpreted the results and critically revised the manuscript for intellectual content. PX takes responsibility for the accuracy of the data analysis. All authors approved the final version of this manuscript to be published.

The authors’ work is funded by the Novo Nordisk Foundation (CB; NNF19OC0056777), Swedish Brain Research Foundation (CB; FO2020-0044), Åke Wiberg Foundation (XiaoT; M18-0169, M19-0266), Fredrik and Ingrid Thuring Foundation (XiaoT; 2019-00488), and the Swedish Society for Medical Research (XiaoT). The study sponsor/funder was not involved in the design of the study; the collection, analysis, and interpretation of data; writing the report; and did not impose any restrictions regarding the publication of the report.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.