The effects of cognitive behavioral therapy for insomnia in people with type 2 diabetes mellitus, pilot RCT part II: diabetes health outcomes

A systematic review and meta-analysis of epidemiological studies found that people with type 2 diabetes (T2D) and sleep disturbances were more likely at risk of poor glycemic control as measured by glycated hemoglobin (HbA1c) and suboptimal diabetes self-care behavior (DSCB) [1, 2]. A recent study found that people with T2D and insomnia symptoms had worse scores in several health domains related to DSCB compared to people with T2D without insomnia symptoms [3]. Additionally, increases in insomnia severity was associated with suboptimal DSCB among people with T2D [3]. It is possible that sleep disturbances lead to fatigue and physical inactivity, which then contributes to poor diabetes health outcomes [4].

Previous studies have shown that insomnia symptoms are common among people with T2D [5, 6], and insomnia itself may independently increase mortality rates [7, 8]. In addition, the increased mortality risk in people with insomnia might be due to inflammation, which is also associated with cardiovascular diseases [9]. The mechanisms underlying the relationship between T2D and insomnia symptoms are not well understood [10, 11], yet there is a need to identify an effective treatment for insomnia symptoms to improve T2D health outcomes.

Among therapeutic options for insomnia, the American Academy of Sleep Medicine recommends the employment of Cognitive Behavioral Therapy for Insomnia (CBT-I) as first-line therapy [12], since CBT-I can be superior when compared to sleep medications [13]. These sleep medications can possibly lead to negative side effects [14] or metabolic alterations [15, 16]. Further, a recent study supported the need to investigate the effects of CBT-I on people with T2D due to the harmful side effects of pharmacological treatments and the limited evidence of effectiveness [17].

CBT-I is a potentially efficacious intervention for people with T2D as it may address an altered metabolism. Generally, the components of CBT-I including sleep restriction therapy, stimulus control therapy, cognitive therapy, and relaxation therapy strengthen the association between bed and sleep connection, readjust the homeostatic mechanisms and the circadian rhythm, and reduce anxiety and rumination about sleep [18]. CBT-I modules may interrupt the physiological mechanisms such as hypothalamic pituitary-adrenal (HPA) axis activation [10, 11], which may be one link between insomnia symptoms and T2D. For example, it has been shown that an association between sleep homeostasis and glucose regulation could be adjusted using sleep restrictions and stimulus control therapies [19]. Improvement of the circadian cycle misalignment has intimate correlations with glucose metabolism for people with T2D [20]; however, this has not been proven yet in CBT-I studies. In addition, ancillary modules in CBT-I, such as relaxation techniques and sleep hygiene, could play a role in reducing stress and nocturia episodes (i.e., the number of bathroom visits per night) [21‐23].

The objective of this study was to explore the effects of 6 sessions of CBT-I on HbA1c, DSCB, and fatigue. We hypothesized that the participants in the CBT-I group would have greater improvements in HbA1c, DSCB, and fatigue compared to the participants in the health education (HE) group. The effects of CBT-I on sleep outcomes were reported separately (Part I of the intervention trial: Using Cognitive Behavioral Therapy for Insomnia in People with T2D, Pilot RCT Part I: Sleep Outcomes and Concomitant Symptoms) [24]. We anticipated improvements in sleep and concomitant outcomes which will positively impact people with T2D and health outcomes because of the relationship between insomnia symptoms and poor diabetes-related health outcomes. Both intervention studies were designed to serve as pilots for a subsequent larger scale intervention trial.

The consort of this intervention trial shows a total of 28 participants enrolled in the study and 26 participants completed the study (Fig. 1). There were no baseline differences between groups in demographics including age, sex, ethnicity, and education (p > .05) [24]. In addition, there were no significant between-group differences in the baseline assessments of HbA1c, RBG, DCP composite score, or FSS (Table 2).

There were significant between-group post-intervention differences in HbA1c (d = 0.41, p = .01), DCP composite score (d = 1.01, p = .01), and FSS (d = 1.07, p = .009; Fig. 2; Table 2). There were significant within-group differences for the CBT-I group in HbA1c (p = .02), DCP composite score (p = .03), and FSS (p = .002), which are also shown in Table 2. However, there were no within-group differences in HbA1c, DCP composite score, or FSS for the HE group.

We noted declines in the blood glucose levels before bedtime and after first awakening in the morning for the CBT-I group using the R software package (Fig. 3). The linear regression analysis showed significant association between the number of days in the CBT-I intervention with a blood glucose level before bedtime (B = -0.56, p = .009) and after first awakening in the morning (B = -0.57, p = .007) (Fig. 3).

To the best of our knowledge, this was the first RCT to examine the effects of CBT-I on diabetes outcomes and daytime functioning in people with T2D and insomnia symptoms. This study suggested CBT-I was effective in improving HbA1c, DSCB, and FSS for people with T2D and insomnia symptoms. Glucose blood levels, both before bedtime and after first awakening in the morning, also decreased over the course of the CBT-I intervention.

Diabetes outcomes improved following CBT-I, with a clinically meaningful difference in HbA1c. Clinical improvement in HbA1c may have been due to reductions in insomnia severity or psychological symptoms, which might foster an improved DSCB. After the CBT-I intervention, as shown in Fig. 2, there was a 0.05% absolute percentage reduction in HbA1c that suggests a clinically significant change based on the American Diabetes Association statistics [41]. It is recommended that people with T2D maintain HbA1c levels below 6% to reduce their risk of developing microvascular complications, although HbA1c between 6.5 and 7.9% is often considered acceptable by physicians [42]. Interestingly, the HE intervention provided to the control group, which included sleep hygiene, diet, and physical activity, was not sufficient to improve HbA1c. However, the baseline data of HbA1c for people in the HE group were at the borderline of optimum glycemic control that has been recommended by the American Diabetes Association. Thus, future research is needed to consider HbA1c in the power calculation and randomization process. As shown in the initial part of this intervention trial, improving insomnia symptoms following CBT-I may produce reductions in depression and anxiety symptoms that are often associated with daily hyperglycemia [24]. Previous studies have shown the negative influence of the combination of insomnia and depression on an individual’s glucose metabolism [43], which could be adjusted using CBT-I [44].

Besides the effects of improving insomnia symptoms on HbA1c, improvements in DSCB could also explain the clinical changes in HbA1c. Our study tracked glucose levels for participants in the CBT-I group, and there was a statistically significant decrease over the course of the intervention. This is entirely observational, however, as we did not monitor glucose before bedtime and after first awakening in the morning in the HE group. Regardless, self-monitoring of blood glucose should be done as a part of DSCB when trying to minimize problems related to hyperglycemia. It has been suggested that self-monitoring of blood glucose significantly reduces HbA1c levels for people with poorly controlled T2D [45].

There are few physiologic mechanisms that might explain improvements in HbA1c. First, the negative effects of sleep disturbances on metabolism might cause decreased brain glucose utilization, which could lead to hyperglycemia [46]. Reducing sleep disturbances via CBT-I might regulate glucose utilization, which could improve HbA1c. Second, previous studies have suggested a U-shaped relationship between sleep duration with HbA1c levels, where excessively short or long sleep durations have been noted to be associated with higher HbA1c levels. Sleep restriction therapy might lead to improved HbA1c levels by maintaining sleep durations within an optimal range of 7–8 h. Third, sleep disturbances are associated with appetite hormone dysregulations [46], and these dysregulations could be adjusted through sleep hygiene and stimulus control therapy. Sleep hygiene and stimulus control might help the participants in scheduling meals and acquiring a better understanding of their bodily needs regarding food consumption. Fourth, abnormal HPA axis activation might be normalized as a result of improving insomnia symptoms. This normalization could reduce cortisol secretion during sleep, which has been linked to reduced morning glucose levels. Finally, people with poor sleep and T2D at risk of impaired decision-making [47], which might be improved following CBT-I. Effective decision making may assist people with T2D in understanding domains related to diabetes such as food choices, control problems, diabetes distress, and medication adherence.

A population-based RCT of 109 older adults with insomnia showed the risk of 8 biomarkers of various health conditions following mood and daytime functioning promotions as a part of cognitive behavioral therapy [48]. The authors suggested that improving sleep quality was associated with reducing the risk of cardiovascular, metabolic, and inflammatory diseases [48]. As mentioned previously, the possible changes in the physiological systems, might be associated with an increased allostatic load, which was related to diabetes, hypertension, metabolic syndrome, and cardiovascular disease [48]. This suggests sleep promotion programs may help in re-establishing healthy patterns of various biomarkers for various health conditions due to the association between the circadian clock regulation and the rhythms of body organs or hormonal secretory patterns that alter numerous biological processes such as metabolism and blood glucose regulation [46]. The current study adds to the literature in regard to the benefits of CBT-I on an important indicator of diabetes control, HbA1c, in people with T2D. Part I of this intervention trial approved the effects of CBT-I on various sleep parameters and psychological symptoms which might indirectly show a positive impact on diabetes health outcomes. However, future study is needed to investigate the effect of CBT-I on a wide range of metabolism parameters for people with T2D.

Although the HE group received the same amount of face-to-face attention, we did not find any significant improvement in their diabetes and daytime functioning outcomes. This reiterated the importance of considering CBT-I as a treatment in diabetes clinics for people with T2D who suffer from insomnia symptoms. Contrary to the positive results of CBT-I on insomnia symptoms, glycemic control, and fatigue in people with T2D, there were no significant improvements in RBG for either group. Several factors might explain these results such as the specificity of interventions, the short-term intervention, or the methodological factors. CBT-I is designed to change detrimental beliefs about sleep behavior, which could demonstrate a secondary effect on RBG over time. However, the data from this study suggests blood glucose levels could be decreased at night before bedtime and in the mornings after first awakening.

This study has identified the effects of CBT-I on diabetes health outcomes in people with T2D; however, some limitations need to be considered for future research. First, including other highly sensitive glucose metabolism measures on larger sample sizes, such as a homeostatic model assessment and an oral glucose tolerance test, may generalize the other effects of CBT-I on diabetes parameters. Second, diabetes management includes interdisciplinary approaches, such as diet, physical activity, and medication adherence, to ensure optimal HbA1c. Future work needs to track daily changes in these activities to better explain the impact of CBT-I on HbA1c. Third, comprehensive functional assessments, including variables such as cognition, motivation, and activities of daily living, may help to efficiently identify other results following CBT-I. Fourth, we were not able to determine if the improvement in HbA1c was mediated by a reduced severity of insomnia or psychological symptoms. A longitudinal design for a larger sample size with T2D is needed to better understand the factors that promote good glycemic control following CBT-I. Finally, although this study demonstrated clinical improvements in HbA1c and DSCB after participants underwent six sessions of CBT-I, future research is needed to measure the sustainability of these improvements for at least 3 months.

This study focuses on clinical information about the effectiveness of CBT-I on diabetes health outcomes. CBT-I showed a clinically meaningful effect on HbA1c and significant improvements in optimal DSCB and fatigue in people with T2D and insomnia symptoms. There is still a need to understand the underlying mechanism of these enhancements, and future research is needed to investigate the long-term effect of CBT-I on diabetes blood parameters and to understand the underlying mechanisms of these improvements.