Effects of type 1 diabetes on sleep duration and timing-a comparison within the general population

Effects of type 1 diabetes on sleep duration and timing – a comparison within the general population Nat´ alia Marhefkov´ a a,b , Martin Sl´ adek c , Jan Klus´ aˇ cek d , Michaela R¨ oschov´ a d , Michal Kahle e , Robert Roland f , Michal Pazderník f , Peter Wohlfahrt f , Peter Novodvorský a , Martin Haluzík a , Alena Sumov´ a c,* , Michal Dubský a,b,** a Diabetes Centre, Institute for Clinical and Experimental Medicine, Prague, Czech Republic b First Faculty of Medicine, Charles University, Prague, Czech Republic c Laboratory of Biological Rhythms, Institute of Physiology, The Czech Academy of Sciences, Prague, Czech Republic d Institute of Sociology, The Czech Academy of Sciences, Prague, Czech Republic e Department of Statistics, Institute for Clinical and Experimental Medicine, Prague, Czech Republic f Cardiac Centre, Institute for Clinical and Experimental Medicine, Prague, Czech Republic ARTICLE INFO Keywords: Circadian rhythmicity Chronotype Sleep Social jetlag Type 1 diabetes mellitus Glycemic control Glycemia risk index Chronic diabetic complications ABSTRACT Study objectives: The role of circadian misalignment has been implicated in metabolic health, not excluding glycemic control, but its role in type 1 diabetes mellitus (T1DM) has not been fully understood. The aim of the study was to investigate the relationship between the degree of circadian misalignment and the efficacy of glycemic control in T1DM patients. Methods: This cross-sectional study included 879 patients within the general Czech population who have been diagnosed with T1DM for a period of at least 10 years (mean HbA1C 52.9 ±0.4 mmol/mol). The patients were aged 18–86 years and comprised 45 % females. All participants filled out the Munich Chronotype Questionnaire (MCTQ) to assess their individual chronotype and calculate social jetlag (SJL). The individual chronotypes were also assessed using self-reporting best alertness time (BAmid). Sleep quality was self-assessed on a 4-point scale from 1 (best) to 4 (worst). T1DM management was evaluated through continuous glucose monitoring data (CGM) including the glycemia risk index (GRI). For the non-diabetic control group, we utilized data obtained from wave 5 of a nationally representative longitudinal Czech Household Panel Survey (CHPS), which involved 1757 subjects without diabetes aged 18–86 years. Data were analyzed through cross-sectional univariate and multivariate analyses. Results: T1DM patients had later average chronotype and significantly higher SJL compared to controls. Across all adjusted models, higher SJL was consistently associated with T1DM. Within the T1DM cohort, increased SJL was positively associated with higher GRI (r =0.12, P =0.0029), while chronotype itself showed no independent association with glycemic control. Overall, metabolic markers showed better outcomes in T1DM patients than in control non-diabetic population. Conclusions: Social jetlag was robustly correlated with both T1DM and impaired glycemic management. These findings call for systematic examination of such characteristics in patients with T1DM and may open the door to taking circadian aspects into account when striving for optimal management of T1DM. 1. Brief summary Data regarding the effects of poorly adjusted circadian rhythms is insufficient to support the idea that they influence glycemic control. Many studies have addressed this phenomenon in relation to T2DM, yet there is still little known about its association with glycemic control in T1DM. The aim of this study was to demonstrate the importance of examining sleep quality in patients with T1DM. We aim to incorporate circadian aspects of overall health into clinical practice for better * Corresponding author. Praha 4, Vídeˇ nsk´ a 1083, 142 20, Prague, Czech Republic. ** Corresponding author. Diabetes Centre, Institute for Clinical and Experimental Medicine, Praha 4, Vídeˇ nsk´ a 1958, 140 21 Prague, Czech Republic. E-mail addresses: [email protected] (A. Sumov´ a), [email protected] (M. Dubský). Contents lists available at ScienceDirect Sleep Medicine journal homepage: www.elsevier.com/locate/sleep https://doi.org/10.1016/j.sleep.2025.106868 Received 28 July 2025; Received in revised form 22 September 2025; Accepted 15 October 2025 Sleep Medicine 136 (2025) 106868 Available online 15 October 2025 1389-9457/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). management of T1DM. 2. Introduction The circadian system controls the timing of a variety of physiological processes in our body and synchronizes them with the time of day as well as with each other [1]. In real life, synchrony effect is maintained by ensuring adequate exposure to natural outdoor light, increased daytime activity, and restricting meals to daylight hours while avoiding artificial light at night and aligning sleep with the nighttime period. These factors are necessary to keep the circadian system synchronized and to avoid any disturbance in circadian timekeeping. Numerous studies have linked disruption of intrinsic circadian rhythms to an increased risk of various diseases, with cardiovascular and metabolic diseases being the most common [2]. In type 2 diabetes mellitus (T2DM), maintaining the correct phase relationship between the sleep-wake cycle and metabolism appears to be of crucial importance [1]; however, knowledge of what effect this relationship has on type 1 diabetes mellitus (T1DM) control and its complications is lacking. Individuals differ in their preference for a distinct timing of sleep and activity over 24 h, which compares the distinct sleeping patterns between workdays and work-free days; this preference is referred to as chronotype. A person’s chronotype is largely regulated by the circadian clock, which is age-dependent and influenced by both genetic variations and environmental factors [3]. An individual’s chronotype can be directly assessed by measuring the phase of the circadian clock (e.g., by dim light melatonin onset analysis), through actigraphy or by administrating validated questionnaires, such as the Munich Chronotype Questionnaire (MCTQ), which takes into account the distinct sleeping patterns between workdays and work-free days during the week [4]. The MCTQ is a useful tool for assessing objective chronotypes based on self-reported time of sleep or wakefulness as well as sleep latency and inertia. The questionnaire expresses the chronotype as a midpoint of sleep on work-free days (Mid-Sleep on Free Days or MSF). The marker reports on the phase of the central clock in the suprachiasmatic nuclei of the hypothalamus. Notably, it has been shown that the chorotype is also reflected in the actual phase of clocks in other cells, so that internal timing shifts throughout the body according to preferred schedules [5]. Daily schedules may force subjects to rearrange their waking time according to work and school hours in discordance with their internal time. Such a situation leads to so-called social jetlag (SJL) [3] in analogy to jetlag caused by traveling across time zones. Social jetlag can be defined as a chronic misalignment between sleep time on workdays and free days, and it is remarkably common in industrialized societies [6]. Approximately 70 % of the studying/working population experiences at least 1 h of social jetlag, while nearly half experience 2 h or more [7]. SJL is calculated as the difference in midpoint sleep between workdays (MSW) and free days (MSF), with both demonstrated at local time (SJL =|MSF −MSW|) [7]. The extent of SJL varies widely across the population. Recent research demonstrated that SJL negatively affects glycemic control in people with T1DM [8–10]. Social jetlag is also associated with adverse endocrine, behavioral and cardiovascular risk profiles in otherwise healthy participants and increases the risk of metabolic diseases and mental disorders [11]. In people with diabetes, glycemic control is assessed by parameters obtained from continuous glucose monitoring (CGM) systems [9]. These parameters include glycemic variability (GV), which refers to fluctuations of interstitial glucose levels over a given time interval. Further CGM metrics include time in range (TIR: percentage of time when glucose levels are between 3.9 and 10 mmol/L), time below range (TBR: percentage of time when glucose levels are <3.9 mmol/L) and time above range (TAR: percentage of time when glucose levels are >10 mmol/l). Novel CGM-derived indices, such as the glycemia risk index (GRI), have recently been introduced. The GRI is a CGM metric used to characterize the overall quality of CGM data, which consider the proportion of time a patient displays in very low/low and high/very high interstitial glucose concentrations [12]. The GRI has been shown to provide precise estimation of the risk of hypoglycemia and hyperglycemia, with a single number ranging from 0 (minimal risk) to 100 (maximal risk) [10]; a lower number indicates more favorable glycemic control, while a higher number indicates less favorable glycemic control. Our first aim was to characterize the chronotype, and to identify SJL and selected metabolic markers prevalent in a cohort of patients with T1DM. Secondly, we compared people with T1DM with those in a group of non-diabetic individuals. Finally, we evaluated the association between these parameters and glycemic control quality in T1DM patients. 3. Material and methods 3.1. Study participants and data collection (diabetes group) Our experimental dataset consisted of 879 patients with T1DM aged 18–86 years (mean age 45.3 ±14.3). The T1DM patients were prospectively recruited from the outpatient clinic of Diabetes Center (Institute of Clinical and Experimental Medicine, Prague, Czech Republic). Baseline characteristics of the T1DM experimental cohort, including patient history, data on the presence of diabetic complications and CGM metrics, is shown in Table 1. Specific insulin therapy regimens are shown in Table 2. A substantial number of our patients (43 %) utilized automated insulin delivery (AID). We also recorded the type of CGM device used by participants (Dexcom [42 %], FreeStyle Libre [18 %], Medtronic Guardian [20 %], Glunovo [<1 %], shown in Table 3]). Continuous glucose monitoring metrics included TIR, TBR and TAR. The mean HbA1c was 53 mmol/mol and TIR was 73 %. The glycemia risk index was calculated as described above, which resulted in a mean GRI score of the experimental group was 30.4 for the experimental group. Participants filled out the Czech language version of the MCTQ. The chronotype was evaluated via the MCTQ for 594 patients from the T1DM cohort. In addition to clinical and metabolic parameters, we also explored the socio-demographic background of our subjects. Their employment Table 1 Characteristics of the diabetes (T1DM) group. Baseline demographic, clinical, and metabolic data of the 879 adults with T1DM included in the study. Continuous glucose monitoring (CGM) metrics are defined as follows: Time in Range (TIR): Percentage of time glucose is within 3.9–10.0 mmol/l; Time Above Range (TAR): Percentage of time glucose is above target range (Level 1 TAR: 10.0–13.9 mmol/L; Level 2 TAR: >13.9 mmol/L); Time Below Range (TBR): Percentage of time glucose is below target range (Level 1 TBR: 3.0–3.9 mmol/l; Level 2 TBR: <3.0 mmol/L); and Glycemia Risk Index (GRI): Composite score (0–100) summarizing risks of both hypoglycemia and hyperglycemia based on CGM data (0 =lowest risk; 100 =highest risk). Parameter Average value Female 395/879 (45 %) Age 45.32 ±14.34 (years) BMI 26.54 ±4.42 (kg/m 2 ) Smoking 141/735 (19 %) Retinopathy 168/879 (38 %) Neuropathy 334/879 (19 %) Dyslipidemia 450/879 (51 %) Triglycerides 0.89 mmol/l Cholesterol 4.48 ±0.87 mmol/l HDL 1.50 ±0.42 mmol/l LDL 2.50 ±0.75 mmol/l Cortisol 447.85 ±182.01 nmol/l Glycemia 7.99 ±3.41 mmol/l HbA1C 52.96 ±0.42 mmol/mol (DCCT 7.0 %) TIR 73 % TAR (Level 1) 19 % TAR (Level 2) 3 % TBR (Level 1) 2 % TBR (Level 2) 0 % GRI 30.40 N. Marhefkov´ a et al. Sleep Medicine 136 (2025) 106868 2 status (employee, unemployed/retired/student, worker, or unknown) is presented in Table 4. These variables were analyzed descriptively in order to characterize the study cohort (see Table 5). 3.2. Study participants and data collection (control group) The control group dataset was obtained from the Czech Household Panel Survey (CHPS), which is a nationally representative longitudinal survey of Czech non-institutionalized individuals living in private households (conducted by the Institute of Sociology, Czech Academy of Sciences; CERGE-EI; Faculty of Social Studies, Masaryk University). The households were selected by a two-stage stratified probability sampling design; questionnaires were administered through face-to-face and penand-paper-personal-interviews. Fieldwork was organized by MEDIAN and STEM/MARK agencies in six subsequent waves (from 2015 to 2020). For this study, only data from wave 5 performed between July and November 2019 are used. A subsample of 1957 subjects aged 18–86 years from more than 5000 original participants interviewed in the previous wave was used; two hundred of them self-reported a diabetes diagnosis (type was not reported) and were excluded from the dataset. The participants filled out the Czech language version of the MCTQ and provided blood samples. The resulting sample age was 42.8 ±22.7 years. The specific chronotype was successfully evaluated for 1136 control subjects [6]. 3.3. Combined dataset To resolve the imbalance between T1DM and control datasets, we calculated propensity score based on three major covariates (age, sex and BMI), followed by Inverse Probability Weighting (IPW). The balance of covariates before and after weighting is visualized in Fig. S1. Subjects that reported shift work during the previous three months (n =276) were excluded from the analysis; for selected multivariate models, these subjects were included together with additional binary covariant “shiftwork”. Up to 521 additional missing values (mostly missing information on cortisol levels and the inability to calculate objective chronotype – see the methodology in Circadian and sleep variables section) were excluded from the analysis depending on the number of parameters used in the models of increasing complexity. 3.4. Blood samples (control group) Participants were asked to visit a regional medical facility linked to a certified diagnostic company and provide a morning blood sample while in a fasted state (n =1757); the time of sampling was 8.32 ±1.13 h (AM, mean ±standard deviation). Plasma levels of high-density and lowdensity lipoprotein cholesterol (HDL, LDL), total cholesterol (CHL), triglycerides (TAG), glucose (GLC), and cortisol (COR) were measured using standard clinical methods (levels of CHL, LDL, HDL, TAG, GLC are expressed in mmol/l; COR in nmol/l). To minimize batch effects introduced by slightly different sensitivity, accuracy or precision of analytical instruments from five participating diagnostic companies with eight individual laboratories, biomarker levels were median-centered by subtracting a small correction, which was calculated as the difference between the median of measurements in a given lab and the median of all measurements. 3.5. Circadian and sleep variables MSFsc (mid sleep phase on free days, sleep debt corrected, expressed in hours centered on midnight) was calculated for subjects (both from experimental and control groups) who reported in the MCTQ that they did not require an alarm clock, as they are able to arrange their sleep time on free days in a regular manner or to wake up at a consistent time without an alarm clock during work days. Ordinal variable “schedule” was encoded for early, medium and late MSFsc (as 1 for MSFsc <2.4, 2 for MSFsc ≤3.9, 3 for MSFsc >3.9, the cutoffs are based on frequency distribution in our representative dataset of Czech population), while subjects that reported no free days (i.e. everyday alarm use) or irregular sleeping schedules were encoded as 4. Age and sex-corrected normalized chronotypes (MSFsasc, a hypothetical chronotype at the age of 30) were calculated by fitting third degree polynomial curves to MSFsc data for men and women aged 18+and normalizing to age =30 13 . In addition to assessing the sleep phase, the subjective chronotype was determined through self-assessment of the interval of their best cognitive alertness. This interval was utilized for the calculation of BAmid, which is defined as the "best alertness midpoint." The interpretation of this measurement is expressed in hours [13]. The best alertness midpoint (BAmid) is a proxy variable designed to emulate the Morningness-Eveningness Questionnarie (MEQ) output, obviating the need to ask an additional 19 questions (questionnaire is available at Morningness-Eveningness Questionnaire (MEQ) | QxMD.) It has previously been established that time spent outdoors (LE week ) is a lifestyle factor that may significantly influence chronotype, as measured by sleep phase [14]. The extent of weekly outdoor light exposure was evaluated on the basis of time spent outside during work and free hours, which was observed during the 3 months preceding the study. This evaluation was conducted with consideration to the respondents’ answers to the questionnaire [13]. Social jetlag defines the absolute value of the difference between the mid-sleep phase on free days and workdays. Using the MCTQ, SJL was quantified as a shift between the activity-sleep phase on workdays and Table 2 Distribution of insulin therapy regimens within the study cohort. Data are presented as absolute numbers (n) and percentages (%) of the total sample (N = 879). Insulin Therapy n % CSII (Insulin Pump) 65 7.4 Automated Insulin Delivery (AID) 378 43.0 Insulin Pen (IIT) 274 31.2 Other/NA 74 8.4 PANCREAS TRANSPLANT 88 10.0 Table 3 Distribution of continuous glucose monitoring (CGM) devices used in the study cohort and percent of use of the CGM. Data are presented as absolute numbers (n) and percentages (%) of the total sample. CGM Device n % Dexcom 370 42.0 Freestyle Libre 159 18.0 Medtronic 172 20.0 Glunovo 1 0.1 No Sensor 177 20.0 THE USE OF CGM N % 0–24 % 28/704 4.0 25–49 % 58/704 8.2 50–74 % 145/704 20.6 75–100 % 473/704 67.2 N/A 175/879 20 Table 4 Description of the occupational structure of the study cohort, grouped into employed, unemployed/retired/student, workers, and unknown. Data are presented as absolute numbers (n) and percentages (%) of the total sample (N = 879). Employment Status n % Employee 380 43 Unemployed/Retired/Student 271 31 Worker 24 3 Unknown 204 23 N. Marhefkov´ a et al. Sleep Medicine 136 (2025) 106868 3 free days [3]. Sleep quality was self-reported using a 4-point scale from 1 (best) to 4 (worst). 3.6. Cross-sectional univariate analyses and visualizations Data were analyzed using Python pandas and visualized by matplotlib and seaborn. Cross-sectional analyses were performed by MannWhitney test (Fig. 1D–O) or 2-way ANOVA (Fig. 1B–C) to compare groups, with Pearson’s or Spearman’s correlation to analyze linear relationships. Spearman’s correlation index (rho) between all variables was inspected using the Hinton diagram. Relationships between selected variables were additionally visualized by polynomial curves fitted using seaborn’s x_estimator function to calculate means and standard deviations in a set amount of bins (for respective numbers of bins and polynomial degree, see figure legends). Histograms were fitted with a kernel density estimate curve using seaborn. 3.7. Cross-sectional multivariate analysis To identify variables associated with diabetes, we constructed weighted binomial generalized linear models (with robust standard errors) of increasing complexity (Table 2 and Supplemental Table S1–S2 in Supplemental Materials) depending on the set of covariates included within the model, with the presence or absence of T1DM as the outcome. Generalized least squares regression model was used to identify variables associated with continuous variable GRI (outcome) available in the T1DM dataset. Python statsmodels package was used for regression. All models were checked for multicollinearity, and weighted least squares linear models were additionally checked for normality of residuals and heteroscedasticity using Breusch-Pagan test. 3.8. Ethical consideration The study followed the principles of the Declaration of Helsinki and was approved by Ethics Committee of the Institute for Clinical and Experimental Medicine, Prague. Written informed consent was obtained from each participant who provided blood samples prior to enrolment in the study after explanation of the study procedures. 4. Results We first visualized the Spearman’s regression indexes of all analyzed variables (Fig. 1A). As expected, T1DM was positively associated with glycemia; however, interestingly, it was also positively associated with circadian variables chronotype (MSFsc) and social jetlag (SJL). The average chronotype of the T1DM patients (3.47 ±1.14 h) was significantly later than that of the control subjects (3.12 ±0.98 h) (Figs. 1B and 2-way ANOVA P <0.0001, Fig. 1D, Mann-Whitney test, P < 0.0001). The relationship was more complex for SJL (Figs. 1C and 2-way ANOVA P =0.15)), but the average SJL was on 0.26 h increased in T1DM patients than in controls (Fig. 1E, Mann-Whitney test, P < 0.0001). Both the normalized chronotype MSFsasc (Fig. 1F, P <0.0001) and subjective chronotype BAmid (Fig. 1G, P <0.0001) were similarly later for the T1DM group of adults. There was no significant difference in the weekly sleep duration between patients and controls (Fig. 1H, P = 0.34). Interestingly, the T1DM group spent significantly less time outdoors than did the control group (Fig. 1I, P <0.0001). We also identified significant differences in levels of BMI (Fig. 1J, P <0.0006), glycaemia (Fig. 1K, P <0.0001), cortisol (Fig. 1L, P <0.0001), LDL (Fig. 1M, P < 0.0001), HDL (Fig. 1N, P <0.0001) and triglycerides (Fig. 1O, P < 0.0001) between T1DM and control groups. To explore further, we constructed weighted binomial generalized linear models of increasing complexity (Table 2 +Table S1–2in Supplements) depending on the set of covariates included within the model. Both the later chronotype (0.2301 ±0.036, p <0.0001) and increased social jetlag (0.1609 ±0.043, p <0.0001) were significantly positively associated with the presence of diabetes (Table 2). Excluding the subjects (n =147) that experienced shift work in the past 3 months before the questionnaire did not meaningfully change the results (MSFsc 0.1985 ±0.036, p <0.0001, SJL 0.1875 ±0.045, p <0.0001); the more complex models do not include them. Including subjects that reported irregular sleeping or no free days (n =554) and using ordinal variable “schedule” instead of continuous variable MSFsc as a chronotype proxy did not meaningfully change the results (schedule 0.2459 ±0.032, p < 0.0001, SJL 0.1533 ±0.034, p <0.0001); the more complex models use the continuous variable MSFsc. Including sleep and light exposure covariates in the more complex model (Table S1) decreased the association of chronotype (0.0827 ±0.041, p =0.043) and social jetlag (0.1752 ±0.054, p =0.001) with the presence of diabetes, while including blood biomarkers (Table S2) further diminished (SJL 0.2964 ±0.106, p =0.005) or completely removed (MSFsc p =0.179) their effect. Interestingly, worse self-reported sleep quality (0.5511 ±0.094, p <0.0001) remained significantly positively associated and longer time spent outdoors (−0.3525 ±0.034, p <0.0001) remained significantly negatively associated with the presence of diabetes even in the most complete model (Table S2 in Supplements). Based on these results, we hypothesized that T1DM could be better controlled in patients with an earlier chronotype or lower SJL. We focused on GRI as a potentially more accurate metric for the evaluation of glycemic management. However, since we only obtained GRI for the T1DM dataset, controls were not included in this model. Interestingly, while MSFsc showed no association with GRI (Fig. 2A, P =0.2563), we observed significantly higher GRI in patients with increased social jetlag (Pearson’s r =0.12, P =0.0029, 95 % CI [0.044, 0.1918], Fig. 2). To Table 5 Weighted binomial generalized linear model (GLM) with heteroskedasticity-consistent robust Standard Errors, showing association between the presence of type 1 diabetes (dependent variable), chronotype and social jetlag; ****P <0.0001. Multicollinearity was tested by variance inflation factor with VIF <1.4 for all covariates. Dep. Variable: Diabetes I (yes 1, no 0) No. Observations: 1717 Model: GLM Df Residuals: 3445 Model family: Binomial Df Model: 7 Method: IRLS Covariance Type: HC3 Pseudo R-squ.: 0.1531 Deviance: 4496 Pearson chi-squ.: 0.0035 No. iterations: 4 Var Coef. Std.Err. z P>|z| [0.025 0.975] constant −0.3286 0.304 −1.08 0.28 −0.925 0.268 ****Chronotype (MSFsc) 0.2301 0.036 6.471 <0.0001 0.16 0.3 ****Social jetlag (SJL) 0.1609 0.043 3.775 <0.0001 0.077 0.244 Age −0.0019 0.003 −0.737 0.461 −0.007 0.003 Sex −0.0354 0.072 −0.493 0.622 −0.176 0.105 BMI 0.0055 0.008 0.722 0.47 −0.009 0.02 ****Smoker −0.214 0.025 −8.402 <0.0001 −0.264 −0.164 ****Shiftwork −1.8193 0.183 −9.959 <0.0001 −2.177 −1.461 N. Marhefkov´ a et al. Sleep Medicine 136 (2025) 106868 4 verify, we constructed generalized least squares regression model with GRI as the outcome variable (Table S3). Increased social jetlag (0.0475 ±0.02, p =0.016) remained weakly positively associated with larger glycaemia risk even when covariates such as age, sex, BMI and HbA1c levels were included in the model. Our data suggest that T1DM control may be worse in subjects with misaligned circadian system. 5. Discussion Adults with T1DM demonstrated later chronotype and significantly increased SJL than non-diabetic controls. In weighted binominal models, both the later chronotype and increased social jetlag were significantly positively associated with the presence of diabetes. In a recently published study [15], the average chronotype in a cohort of Fig. 1. (A) Hinton diagram of correlations between circadian and physiological parameters. Size of the square is proportional to Spearman’s rho. (B) Chronotype and social jetlag (C) were calculated for controls (blue) and diabetic patients (red) and expressed in relation to their age (2-way ANOVA, ****P <0.0001). For visualization purposes, data were separated into 8 bins with mean ±SD error bars and fitted with a third order polynomial curve. (D) Chronotype and (E) social jetlag in all adults, 18–45 years old and those older than 45 years were compared between controls and diabetic patients. (F) Age and sex normalized chronotype (MSF sasc ) and (G) subjective chronotype (BAmid; query about the start and end of the most physically and mentally active interval) compared between controls and diabetic patients. (H) Sleep duration, (I) weekly time spent outdoors, (J) body mass index, (K) glycemia, (L) cortisol, (M) low and (N) high density lipoproteins, and (O) triglycerides level in plasma were compared between controls and diabetic patients (Mann-Whitney test, ***P <0.001, ****P <0.0001). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) N. Marhefkov´ a et al. Sleep Medicine 136 (2025) 106868 5 T1DM patients was characterized as intermediate. The discrepancy between this result and the observations from our current study may be related to several factors. Firstly, chronotyping is dependent on the sample size. The aforementioned recently published study included 95 T1DM patients, whereas our cohort consisted of 594 T1DM patients. Secondly, results may vary depending on the method used to assess chronotype. While the MEQ was used in the previously published study [15], we used the MCTQ, which allowed us to assess not only the chronotype but also the extent of SJL as a proxy for circadian misalignment [16]. Furthermore, the MCTQ was found to correlate slightly better than the MEQ when validated by an objective marker (dim light melatonin onset) [4]. A later chronotype was typically associated with shorter time spent outdoors [17], which was reflected in the behavior patterns of our T1DM group. In population-wide studies, a later chronotype has been associated with increased SJL [16]and the degree of the SJL exhibit age-dependence [18]. According to our results, SJL remained significantly associated with T1DM across all levels of adjustment, whereas the association with chronotype attenuated after inclusion of covariates. Recent objective data analysis has confirmed a consistent negative association between age and social jetlag, showing that for every one-year increase of age, SJL decreases by approximately 0.64 min [19]. This represents a gradual but meaningful reduction in circadian misalignment, as individuals grow older. Our findings align with the previous evidence suggesting that circadian misalignment may be more pronounced in diabetes. We also analyzed several continuous glucose monitoring (CGM) metrics including time in range (TIR), time below range (TBR; levels 1 and 2), and time above range (TAR; levels 1 and 2). None of these values revealed any significant associations with sleep or chronotype parameters. By contrast, glycemia risk index (GRI) scores, calculated from the most recent 14 days of CGM monitoring, were positively correlated with SJL, suggesting that circadian misalignment contributes to greater glycemic variability and risk exposure. Our patients utilized various CGM devices; however, the GRI metric is designed to be robust across many platforms. In a recent study published in 2024 [19], the implications of the GRI at different HbA1c levels as well as other conventional CGM metrics have been explored [20]. The findings demonstrated a linear correlation between GRI scores and both HbA1c and TIR, thus emphasizing the multifaceted implications of the GRI across distinct HbA1c levels in T1DM [19]. Specifically, the GRI indicated a connection to hypoglycemic risk with levels of HbA1c <53 mmol/mol and a connection to hyperglycemic risk with levels of HbA1c ≥53 mmol/mol. These data underscore the significance of individualized treatment approaches and posit the clinical relevance of the GRI in optimizing glycemic management strategies for individuals with T1DM. Furthermore, an examination of Czech Registry data (CENDA) from 2013 to 2020 reveals a downward trend in HbA1c levels among children diagnosed with T1DM. The mean HbA1c level exhibited a decline of 12 mmol/mol in 2020, a finding that attained statistical significance (p < 0.001) [21]. Concurrently, a decline in the prevalence of acute diabetic complications was observed. The principal factors associated with reduced HbA1c levels were treatment with modern technology (e.g. insulin pumps and CGM devices), male sex and care provided at a large diabetes center [21]. Diabetes and its management can affect both the physiological and behavioral components of sleep architecture. For example, patients with obstructive sleep apnea (OSA) are more likely to have poorer glycemic control [21]. Poor sleep quality was reported by 67.4 % of patients. Individuals with evening chronotypes scored worse in sleep quality (p = 0.05) and reported a lower sense of well-being (p =0.03) in comparison to other chronotypes [15]. The general recommendation for T1DM patients is to sleep for at least 6.5 h since shorter sleep times are associated with worse HbA1c values, as mentioned in a study from 2016 [22]. A subsequent cross-sectional study of 115 patients with T1DM confirmed these findings by demonstrating that patients with social jetlag ≥1 h exhibited significantly higher adjusted values of HbA1c than did those with social jetlag <1 h (8.7 % vs. 8.0 %, P =0.029). After adjusting for confounding variables, including age, sex, diabetes duration, insulin dose, insulin regimen, and body mass index, the association between social jetlag ≥1 h and HbA1c remained statistically significant (β = 0.253, P =0.026) [10]. Interestingly, we observed that the T1DM patients in our study spent less time outdoors per week than did the control group, which may partially explain the differences in sleep and circadian variables, as this factor was previously associated with a later chronotype. Additionally, both groups differed in body mass index (BMI), which is known to influence circadian homeostasis and could present a potential confounding factor in the observed associations [23]. Employment status and work schedules are known to shape sleep timing and circadian rhythms, and our data confirm a heterogeneous occupational structure within the T1DM cohort. Taking it together, our findings suggest that circadian misalignment, rather than chronotype itself, may play a role in glycemic management in T1DM. These results add to the limited but growing body of literature emphasizing the importance of circadian health in the context of diabetes care. 6. Limitations This study has several important limitations. First, its cross-sectional design precludes any causal inference, and the observed associations Fig. 2. In T1DM group, increased glycemia risk index (GRI) is associated with increased social jetlag (SJL). (A) Pearson’s linear regression shows the relationship between GRI and chronotype or (B) social jetlag. N. Marhefkov´ a et al. Sleep Medicine 136 (2025) 106868 6 cannot establish directionality. Second, the circadian clock is strongly influenced by environmental factors such as light exposure, work schedules, and lifestyle habits, which may have acted as potential confounders and were not fully controlled for. Third, the reliance on selfreported questionnaires for chronotype and sleep quality introduces a risk of recall bias and misclassification. Fourth, our non-diabetic control group was not assessed for diabetes-specific CGM variables (e.g., GRI), limiting direct comparability. Fifth, differences in BMI between the T1DM and control groups may have influenced circadian outcomes and should be acknowledged as a confounding factor. Finally, heterogeneity in CGM device type and employment status may have influenced sleep and circadian characteristics, and although these factors were described in detail, they were not fully controlled for in statistical models. 7. Conclusion Circadian misalignment, reflected by higher social jetlag, was associated with worse glycemic risk profiles in T1DM, while other CGMderived parameters (TIR, TBR, TAR) did not show such associations. Our data highlight the potential clinical importance of assessing circadian health in clinical practice. Systematic integration of circadian screening into routine diabetes care could support more personalized therapeutic strategies. CRediT authorship contribution statement Nat´ alia Marhefkov´ a: Writing – original draft, Investigation, Data curation. Martin Sl´ adek: Supervision, Methodology, Formal analysis, Data curation. Jan Klus´ aˇ cek: Data curation. Michaela R¨ oschov´ a: Data curation. Michal Kahle: Software, Methodology. Robert Roland: Data curation. Michal Pazderník: Supervision, Formal analysis, Conceptualization. Peter Wohlfahrt: Data curation. Peter Novodvorský: Visualization, Supervision, Formal analysis. Martin Haluzík: Resources, Project administration, Funding acquisition. Alena Sumov´ a: Supervision, Resources, Methodology, Investigation. Michal Dubský: Supervision, Funding acquisition, Conceptualization. Disclosure statement N. Marhefkov´ a: None; M. Sl´ adek: None; J. Klus´ aˇ cek: None; M. R¨ oschov´ a: None; M. Kahle: None; R. Roland: None; M. Pazderník: None; P. Wohlfahrt: None.; P. Novodvorský: None; M.Haluzik: None; A. Sumov´ a: None; M. Dubský: None. All authors have seen and approved the manuscript. Work for this study was performed at the Institute of Clinical and Experimental Medicine, Prague, Czech Republic. This study was funded by the project National Institute for Research of Metabolic and Cardiovascular Diseases (Programme EXCELES, Project No. LX22NPO5104) - Funded by the European Union - Next Generation EU. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments Funded by the project National Institute for Research of Metabolic and Cardiovascular Diseases (Programme EXCELES, Project No. LX22NPO5104) - Funded by the European Union - Next Generation EU. List of abbreviations AID Automated insulin delivery BAmid Best Alertness time CGM Continuous Glucose Monitoring GRI Glycemia Risk Index MEQ Morningness-Eveningness Questionnaire MCTQ Munich Chronotype questionnaire MSF Midpoint Sleep on Free days MSW Midpoint Sleep on Workdays MSFsc Mid sleep phase on free days, sleep debt corrected MSFasc Mid sleep phase on free days, age and sex corrected LEweek Light exposure/Time spent outdoors during week SJL Social jetlag T1DM Type 1 diabetes mellitus T2DM Type 2 diabetes mellitus TIR Time In Range TBR Time Below Range TAR Time Above Range Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.sleep.2025.106868. Data availability Raw data and technical information is freely available in the Czech Social Science Data Archive ˇ CSDA [. Hamplov´ a, Dana; R¨ oschov´ a, Michaela; Kudrn´ aˇ c, Aleˇ s; Jan, Klus´ aˇ cek, 2019, "ˇ Cesk´ e panelov´ e ˇ setˇ rení dom´ acností, 5. vlna (2019) – hlavní soubor", https://doi. org/10.14473/CSDA/PUG41W, CSDA, V2.]. The public dataset does not contain biomarker information. The ˇ CSDA research infrastructure project is supported by the Ministry of Education, Youth and Sports within the framework of grant LM2018135. References [1] Marhefkova N, Sladek M, Sumova A, Dubsky M. Circadian dysfunction and cardiometabolic disorders in humans. Front Endocrinol (Lausanne) 2024;15:1328139. [2] Sumova A, Sladek M. Circadian disruption as a risk factor for development of cardiovascular and metabolic disorders - from animal models to human population. Physiol Res 2024;73(Suppl 1):S321–34. [3] Wittmann M, Dinich J, Merrow M, Roenneberg T. Social jetlag: misalignment of biological and social time. Chronobiol Int 2006;23(1–2):497–509. [4] Kantermann T, Sung H, Burgess HJ. Comparing the morningness-eveningness questionnaire and munich ChronoType questionnaire to the dim light melatonin onset. J Biol Rhythm 2015;30(5):449–53. [5] Novakova M, Sladek M, Sumova A. Human chronotype is determined in bodily cells under real-life conditions. Chronobiol Int 2013;30(4):607–17. [6] Sladek M, Klusacek J, Hamplova D, Sumova A. Population-representative study reveals cardiovascular and metabolic disease biomarkers associated with misaligned sleep schedules. Sleep 2023;46(6). [7] Caliandro R, Streng AA, van Kerkhof LWM, van der Horst GTJ, Chaves I. Social jetlag and related risks for human health: a timely review. Nutrients 2021;13(12). [8] Larcher S, Gauchez AS, Lablanche S, Pepin JL, Benhamou PY, Borel AL. Impact of sleep behavior on glycemic control in type 1 diabetes: the role of social jetlag. Eur J Endocrinol 2016;175(5):411–9. [9] Manov AE, Chauhan S, Dhillon G, et al. The effectiveness of continuous glucose monitoring devices in managing uncontrolled diabetes mellitus: a retrospective study. Cureus 2023;15(7):e42545. [10] Rusu A, Ciobanu D, Bala C, Cerghizan A, Roman G. Social jetlag, sleep-related parameters, and glycemic control in adults with type 1 diabetes: results of a crosssectional study. J Diabetes 2019;11(5):394–401. [11] Rutters F, Lemmens SG, Adam TC, et al. Is social jetlag associated with an adverse endocrine, behavioral, and cardiovascular risk profile? J Biol Rhythm 2014;29(5): 377–83. [12] Klonoff DC, Wang J, Rodbard D, et al. A Glycemia Risk Index (GRI) of hypoglycemia and hyperglycemia for continuous glucose monitoring validated by clinician ratings. J Diabetes Sci Technol 2023;17(5):1226–42. [13] Sladek M, Kudrnacova Roschova M, Adamkova V, Hamplova D, Sumova A. Chronotype assessment via a large scale socio-demographic survey favours yearlong standard time over daylight saving time in central Europe. Sci Rep 2020; 10(1):1419. [14] Roenneberg T, Keller LK, Fischer D, Matera JL, Vetter C, Winnebeck EC. Human activity and rest in situ. Methods Enzymol 2015;552:257–83. [15] Valiensi SM, Folgueira AL, Diez JJ, et al. Is being a lark healthier for patients with type 1 diabetes mellitus? Sleep Sci 2023;16(1):75–83. N. Marhefkov´ a et al. Sleep Medicine 136 (2025) 106868 7 [16] Roenneberg T, Pilz LK, Zerbini G, Winnebeck EC. Chronotype and social jetlag: a (Self-) critical review. Biology (Basel) 2019;8(3). [17] Porcheret K, Wald L, Fritschi L, et al. Chronotype and environmental light exposure in a student population. Chronobiol Int 2018;35(10):1365–74. [18] Foster RG, Peirson SN, Wulff K, Winnebeck E, Vetter C, Roenneberg T. Sleep and circadian rhythm disruption in social jetlag and mental illness. Prog Mol Biol Transl Sci 2013;119:325–46. [19] Gottlieb E, Gupta S, Gahan L, Raymann RJ, Watson NF. Social jetlag decreases across the lifespan: a prospective big data analysis of objective sleep metrics. J Sleep Res 2025:e14433. [20] He B, Fan L, Deng C, et al. Implications of glycemic risk index across different levels of glycated hemoglobin (HbA1c) in type 1 diabetes. Chin Med J (Engl) 2024; 137(4):481–3. [21] Sumnik Z, Venhacova J, Skvor J, et al. Five years of improving diabetes control in Czech children after the establishment of the population-based childhood diabetes register CENDA. Pediatr Diabetes 2020;21(1):77–87. [22] Grimaldi D, Beccuti G, Touma C, Van Cauter E, Mokhlesi B. Association of obstructive sleep apnea in rapid eye movement sleep with reduced glycemic control in type 2 diabetes: therapeutic implications. Diabetes Care 2014;37(2): 355–63. [23] Farabi SS. Type 1 diabetes and sleep. Diabetes Spectr 2016;29(1):10–3. N. Marhefkov´ a et al. Sleep Medicine 136 (2025) 106868 8