Sleep Disturbance in Adults with Sickle Cell Disease: Relationships with Executive and Psychological Functioning

Abstract

Sleep disturbance is common among children with sickle cell disease (SCD) and is related to neurocognitive difficulties. However, research on sleep disturbances and related variables among adults with SCD is extremely limited. The present study examined the relationship between sleep, executive functioning, and emotional functioning among 62 adults (29 females; M age = 32 years, SD = 7.79) with SCD preparing to undergo a stem cell transplant. Participants were administered a neurocognitive evaluation that included objective and subjective measures of executive functioning, and they completed PROMIS self-report measures of anxiety, depression, and pain intensity. Results showed that about 17% of participants endorsed clinically significant sleep disruptions, while 16.1% and 8% endorsed clinically significant symptoms of anxiety and depression, respectively. Sleep disturbance in these adults was not significantly correlated with objective or subjective measures of executive functioning. Moreover, anxiety, but not depression, was a significant mediator between self-reported sleep difficulties and both objective and subjective measures of executive functioning while controlling for pain intensity. Future research on sleep interventions will be essential for ameliorating the effects of sleep disturbance on executive functioning and anxiety among adults with SCD.

Sickle cell disease (SCD) is an inherited group of red blood cell disorders resulting in a mutation of hemoglobin structure that impacts approximately 100,000 Americans according to the Centers for Disease Control and Prevention [1]. Impacted individuals are primarily of African descent. An abundance of research involving SCD has focused on children. In the 1970s, the estimated survival rate for SCD was 20 years, whereas today, the median survival age is in the mid-40s [2]. As the survival rates for SCD increase, research on adults is becoming more pertinent.

In SCD, red blood cells are shaped like a sickle which leads to minimized blood flow and oxygen depletion. This can result in very intense pain episodes (i.e., vaso-occlusive crises) and damage to the body’s internal organs [1]. Common medical complications reported in people with sickle cell disease include pulmonary hypertension, leg ulcers, stroke, ocular damage, osteonecrosis, acute chest syndrome, and cholelithiasis [3].

Among individuals with SCD, pain has been shown to be associated with sleep difficulties [4, 5]. Sleep-disordered breathing (SDB), characterized by abnormal respiratory patterns (e.g., hypopnea, apnea) or insufficient ventilation during sleep, is associated with low hemoglobin oxygen saturation and a variety of SCD-related complications [6, 7]. Obstructive sleep apnea (OSA) is the most common condition of SDB [8]. An expanding body of research investigating children with SCD has revealed higher occurrences of OSA [9] and generally poorer sleep [10-13] compared to the general population. However, far fewer studies have included adults in their investigations. Sharma and colleagues [8] found that 44% of adults with SCD who underwent overnight polysomnography had sleep-disordered breathing. Another study investigating adults with SCD found that approximately 70% report sleep disturbances [14].

In addition to sleep and other somatic issues, researchers have examined numerous aspects of socio-emotional functioning in SCD. With respect to emotional well-being, symptoms of depression and anxiety are documented in patients with sickle cell anemia at rates of up to 44% and 12.7%, respectively [15, 16]. Importantly, sleep disturbances have been shown to relate to emotional functioning in SCD. Wallen et al. [14] noted a correlation between sleep disturbance and depression (r_s = 0.51, p < .001), while Simo and Siela [17] found significant relationships between sleep, depression, and pain in adults with SCD. While no known studies have examined the role of anxiety in sleep disturbances among adults with SCD, researchers in other populations have documented its importance. For example, a review on this topic among children suggested a reciprocal relationship between sleep and anxiety [18]. Among older adults, the relationship between sleep quality and anxiety symptoms has been documented as well [19, 20], with one study finding that sleep efficiency moderated the relationship between anxiety and one aspect of executive functioning (i.e., inductive reasoning) [21]. Thus, it is plausible that anxiety plays a role in the sleep quality of individuals with SCD.

Another area that may relate to sleep disturbances in SCD is cognitive functioning. Hollocks and colleagues [7] found that decreases in oxygen saturation were associated with reduced performance on cognitive and executive functioning assessments in children with SCD. In addition, a recently published study by Bills, Katz, McNeil, and Schatz [22] examined sleep apnea and cognitive functioning (specifically processing speed, language, visual-motor skills, and early academic skills) in pediatric sickle cell disease compared to age matched peers. In contrast to previous research, there was no difference in cognitive functioning between children with SCD and OSA compared to children with SCD only. However, exploratory analyses suggested that more severe symptoms of OSA may relate to worse cognitive functioning.

A plethora of research has investigated sleep disturbances, emotional functioning, and cognitive impairment in other populations. For example, among adults with major depressive disorder, perceived sleep quality predicted cognitive function [23]. Further, depression and sleep disturbance have been shown to mediate the pathway between pain and cognitive dysfunction in individuals with systemic lupus erythematosus (SLE) [24]. In individuals with chronic obstructive pulmonary disease, sleep disorders and comorbid depression and anxiety increased cognitive impairment by 40% [25]. In older adults, sleep complaints have been shown to influence cognitive impairment, specifically phonemic fluency, motor programming, inhibitory control, and working memory [26]. In the same study, depression also significantly influenced cognitive impairment, specifically motor programming, inhibitory control, and working memory. Similarly, depression and sleep duration have been shown to be associated with cognitive decline in individuals with chronic migraines [27]. In those with multiple sclerosis, poorer sleep was significantly associated with greater levels of objective but not subjective measures of cognitive dysfunction [28]. Thus, while the relationship between sleep, cognitive functioning, and emotional functioning in other illness populations has been explored in some depth, to our knowledge, no studies have examined these variables among adults with SCD.

Present study

Due to the scarcity of research on the relationship between sleep and cognitive and emotional functioning in adults with SCD, the current study aims to detect and explore relationships among these variables. By understanding these interrelationships, psychological interventions may be developed to target enhanced sleep quality and/or emotional functioning. Specific hypotheses include (1) higher levels of sleep difficulties will be related to higher levels of anxiety and depression; (2) self-reported sleep difficulties will predict scores on subjective and objective measures of executive function; and (3) anxiety and depression will each mediate the relationships between sleep and executive functioning variables.

Method

Participants

To be eligible for the current cross-sectional study, all patients ages 18 years and older had to be enrolled on one of four medical protocols for individuals with sickle cell disease preparing to undergo a stem cell transplant at the National Heart, Lung, and Blood Institute. Exclusion criteria for this sleep substudy included psychotic symptoms, extreme behavioral difficulties, and/or severe cognitive impairment (i.e., IQ estimated to be below 70).

Measures

The Patient-Reported Outcomes Measurement Information System The Patient-Reported Outcomes Measurement Information System (PROMIS) [29] is a self-report questionnaire that measures physical, mental, and social health in individuals living with chronic conditions. The Sleep Disturbance, Anxiety and Depression subscales and the Pain Intensity item were used in analyses. Higher T-scores on the PROMIS reflect more difficulties with the domain assessed.

Behavior Rating Inventory of Executive Functions - Adult The Behavior Rating Inventory of Executive Functions - Adult (BRIEF-A) [30] is a self-report subjective questionnaire that includes nine clinical subscales in two domains, the Behavioral Regulation Index (BRI) and the Meta-Cognition Index (MI); these indices yield the General Executive Composite (GEC) score. The MI was deemed most relevant to the current study because this domain relates to the ability to cognitively manage attention and solve problems and aligns more closely with the objective measure of executive skills described later [30]. Raw scores are converted to T-scores with a mean of 50 and SD of 10, and higher scores indicate more concerns in metacognitive executive functions.

Delis-Kaplan Executive Function System The Delis-Kaplan Executive Function System (D-KEFS) [31] is a standardized test used to assess executive skills such as planning, organization, and the ability to flexibly shift between tasks. The Number-Letter Switching task of the Trail Making test is considered the primary executive-function measure on the D-KEFS and assesses flexibility of thinking. Therefore, the Number-Letter Switching task (referred to as Trails Switching from this point forward) was used as an objective measure of executive functioning in analyses.

Procedures

The Institutional Review Board of the National Heart, Lung, and Blood Institute approved the four transplant studies that included the present sleep substudy. Informed consent was obtained from all patients for being included in the study. Participants in the medical studies traveled to the National Institutes of Health from all over the USA as well as several African countries. Neurocognitive assessments typically lasted 2 to 3 h and were conducted in a quiet room in an outpatient clinic. Participants underwent neurocognitive testing prior to their pre-transplant conditioning regimen and were asked to complete self-report measures to assess sleep, executive functioning, depression, anxiety, and pain intensity. Participant’s demographic data such as age, sex, and race were self-reported and confirmed using electronic medical records.

No patients declined to participate in this sleep substudy.

Data analysis

Analyses were conducted using Statistical Package for the Social Sciences (SPSS) Software version 26 [32]. Data analysis consisted of calculations of descriptive statistics on demographic and cognitive variables. Pearson’s r correlations were planned between the study variables from the PROMIS, BRIEF-A, and the D-KEFS. Predictive relationships were tested using single linear regression focusing on sleep as the predictor variable. Moreover, mediation analyses were conducted to explore if the relationships between these variables were mediated by depression and/or anxiety scores. Since pain has been related to sleep and depression in SCD, pain intensity was included in the model as a covariate. All mediation analyses were performed using bootstrapping (5000 samples) in PROCESS to estimate indirect effects and bias-corrected confidence intervals (BCa CI) [33]. These indirect effects were considered to be significant when the BCa CIs did not include zero [34].

Preparatory data analyses were run to determine that there were no problems in the data set prior to the main analysis (i.e., missing data, outliers, skewness) and to ensure that the data sample met the assumptions of parametric statistical analyses. These assumptions include (a) linearity, (b) homoscedasticity, (c) independence of errors, and (d) multivariate normality. Because assumptions were violated within this sample, robust bootstrapping methods (set at 5000 samples) were conducted so that regression and mediational analyses could be run without meeting the assumptions of normally distributed data. Missing data was handled by excluding cases listwise.

Results

Descriptive statistics and preliminary analyses

As shown in Table 1, the sample consisted of 62 adults with SCD (33 males and 29 females) aged 19 to 47 years (M = 32.18 years and SD = 7.79). Racial composition of the sample was 95.2% Black/African American and 4.8% multiple race. The mean score on the PROMIS Sleep Disturbance scale was within normal limits (M=51.24, SD = 10.62), but 18% of participants reported sleep difficulties in the clinically significant range (T > 60). In addition, 16.1% of participants reported anxiety (M= 49.74, SD=9.12) and 8% reported depression (M= 46.97, SD=8.22) in the clinically significant range. Means of criterion and outcome measures used in the analyses are presented in Table 2. No demographic information was shown to predict or be associated with the criterion variable (sleep disturbance), examined through a series of Spearman’s ρ correlations and one-way independent ANOVAs.

Table 1.

Demographic Characteristics of Participants (N = 62)

Characteristic	n	%	M	SD
Gender
Male	33	53.2
Female	29	46.8
Age (years)	62		32.18	7.79
Race
Black/African American	59	95.3
Multiple Race	3	4.8
Occupational Status
Not Employed	40	64.5
Yes, Full-time	16	25.8
Yes, Part-time	1	1.6
On Medical Leave	5	8.1
Martial Status
Single	41	66.1
Married	18	29.0
Divorced	2	3.2
Co-habitating	1	1.6

Table 2.

Descriptive Statistics of the Major Study Variables

				Range
Variable	n	M	SD	Possible	Actual	Skew
PROMIS
Sleep Disturbance	62	51.24	10.62	30.5-77.6	31-78	.317
Anxiety	62	49.74	9.12	37.1-83.1	37.1-67.7	.047
Depression	62	46.97	8.22	38.2-81.3	38.2-65.8	.508
Pain Intensity	59	4.47	2.83	0-10	0-10	−.112
BRIEF-A MI	62	51.23	9.99	36-107	35-79	.709
Trails Switching	61	8.15	3.39	1-19	1-14	−.709

Note. MI = Metacognitive Index. Scores are compared to a mean of 50 (SD = 10). Higher scores reflect greater concerns of the domain assessed.

Correlations between sleep, emotional well-being, and cognitive variables

Zero-order Spearman’s ρ correlations for the PROMIS subscales, BRIEF-A MI, and Trails Switching test are included in Table 3. Scores on the PROMIS Sleep Disturbance subscale were positively related to PROMIS Anxiety (r_s = .58, p < .001) and Depression (r_s = .30, p < .05) scores. Additionally, Sleep Disturbance scores were unrelated to BRIEF-A MI and Trails Switching scores. Further, Anxiety scores were positively related to Depression scores (p < .001), the BRIEF-A MI (p < .001), and Trails Switching scores (p < .05). Depression scores were significantly associated with the BRIEF-A MI (p < .001) but not Trails Switching scores (p = .867). Interestingly, scores on the PROMIS Pain Intensity item were related to anxiety and depression scores (ps < .001) but were not significantly associated with Sleep Disturbance, the BRIEF-A MI, or Trails Switching scores (ps > .05).

Table 3.

Correlations between sleep disturbance, emotional functioning, and executive functioning.

Measure	1	2	3	4	5	6
1. Sleep Disturbance	—	.58**	.30*	.25	.24	.24
2. Anxiety	.58**	—	.67**	.45**	.40**	.25*
3. Depression	.30*	.67**	—	.49**	.45**	−.03
4. Pain Intensity	.25	.45**	.49**	—	.23	−.12
5. BRIEF-A MI	.24	.40**	.45**	.23	—	−.11
6. Trails Switching	.24	.25*	−.03	−.12	−.11	—

Note. Spearman two-tailed correlations are reported.

^*p < .05

^**p < .001.

MI = Metacognitive Index.

Regression and mediation testing

Sleep disturbance-anxiety-BRIEF-A metacognitive index

Anxiety, b = .17, 95% CI [0.0082, 0.3908], significantly mediated the relationship between sleep disturbance and metacognitive scores. Sleep disturbance scores did not exert a significant total effect on metacognitive scores, b = .21, R² = .10, p > .05, 95% BCa CI [− 0.0301, 0.4539]. The direct effect of sleep disturbance on metacognitive scores also was insignificant in the mediation model, b = 0.05, p > .05, 95% BCa CI [− 0.2232, 0.3243]. See Fig. 1 for all regression results for variables with anxiety as the mediator.

Figure 1.

Anxiety as the indirect effect between sleep and measures of executive functioning. Pain is included in the model as a covariate.

Sleep disturbance-anxiety-D-KEFS trails switching

Anxiety, b = .16, 95% CI [0.0219, 0.2976], significantly mediated the relationship between sleep disturbance and executive skills (Trails Switching). The direct effect of sleep disturbance on executive skills was insignificant in the mediation model, b = 0.03, p > .05, 95% BCa CI [− 0.0627, 0.1238].

Sleep disturbance-depression-BRIEF-A metacognitive index

Depression, b = .06, 95% CI [− 0.0290, 0.2276], did not significantly mediate the relationship between sleep disturbance and metacognitive (BRIEF-A MI) scores. In addition, sleep disturbance did not exert a significant total effect, b = .21, R² = .10, p > .05, 95% BCa CI [− 0.0301, 0.4539] nor direct effect, b = .15, p > .05, 95% BCa CI [− 0.0290, 0.2276] on metacognition. See Fig. 2 for all regression results for variables with depression as the mediator.

Figure 2.

Depression as the indirect effect between sleep and measures of executive functioning. Pain is included in the model as a covariate.

Sleep disturbance-depression-DKEFS trails switching

Depression, b = .08, 95% CI [− 0.0062, 0.1641], did not significantly mediate the relationship between sleep disturbance and executive skills. The total effect of sleep on executive functioning was nearing significance, b = .08, R² = .09, p = .0557, 95% BCa CI [− 0.0020, 0.1654]. Additionally, the direct effect was insignificant in the model, b = .08, p > .05, 95% BCa CI [− 0.0062, 0.1641].

Discussion

People with sickle cell disease often have complex medical concerns as well as sleep disturbances, which can affect quality of life. This study aimed to explore whether sleep difficulties are associated with executive functioning and to determine the role of emotional functioning in these relationships among adults with SCD. Our hypotheses were partially supported. Our primary findings indicated that anxiety, but not depression, acted as a significant mediator between reported sleep difficulties and both objective and subjective measures of executive functioning. Without anxiety or depression in the model, sleep difficulties did not exert significant influence on executive functioning measures, contrary to our hypotheses. The relationship between sleep and executive skills must be considered in the context of anxiety among adults with SCD.

The fact that depression did not exhibit the same indirect effect as anxiety is surprising given that measures of anxiety and depression are often moderately to strongly correlated, as seen in our data sample (r_s = .67, p < .001). Sleep did not exert a direct effect on depression in our models and results of the broader mediation models were nonsignificant suggesting that depression does not play a role in understanding sleep’s effects on executive skills in SCD. The only significant finding in the Sleep Disturbance-Depression-BRIEF-A Metacognitive Index model, presented in Fig. 2, is that depression scores appear to predict subjective reports of executive dysfunction, b = .52, p = .005. Future studies further investigating this relationship are warranted. Sleep disturbance and neurocognitive functioning have been studied in other populations including MDD, systemic lupus erythematosus, chronic obstructive pulmonary disease, older adults, chronic migraines, and multiple sclerosis. In addition, depression but not anxiety is widely investigated with respect to these constructs. However, the interactions and specific pathways are still not well understood. The current study presents a model for a better understanding of sleep and executive functioning by emphasizing anxiety as a crucial component in these statistical pathways. In contrast, depression did not significantly contribute to the relationship between sleep disturbance and execution functioning as reported in previous research.

In support of hypotheses, higher sleep disturbance scores are related to higher levels of self-reported anxiety (r_s = .58, p < .001) and depression (r_s = .30, p < .05). With respect to the prevalence of sleep disturbance and emotional well-being in our sample, some differences emerge when comparing our results to previous research. For example, 17.7% of respondents reported clinically significant sleep disruptions, which is much lower than another study that reported sleep disturbances in adults with SCD at around 71% [14]. Additionally, 16.1% of our sample reported clinically significant anxiety, which is much higher than would be suggested by the 6.5% of anxiety disorders noted in a large-scale study by Levenson and colleagues [35]. These differences may be attributed to methodology and definitions of the constructs of interest, since our finding examined clinically significant scores on a self-report scale and not formal diagnoses. By contrast, only 8% of patients in our sample scored in the clinically significant range on the depression measure, which is lower than other estimates of depression in SCD ranging from 18 to 44% as documented by rating scales [14, 16, 35-37]. Our sample appears to report symptoms of depression at a similar prevalence rate (8.1%) to healthy adults in the USA [38]. The patients participating in the present study were all newly enrolled on protocols involving stem cell transplants, which holds a potential cure for SCD. Thus, they may have felt more hopeful about the future compared to others with SCD, resulting in lower endorsement of depressive symptoms. Although the means of both anxiety (M = 49.74) and depression (M = 46.97) fell within normal limits, more anxiety scores fell within the clinical range than depression (16.1% and 8%, respectively) suggesting that anxiety may have been more relevant for the patients in the present study than depression as they prepared to undergo extensive transplant protocols that were potentially curative, yet experimental.

In addition, results also showed that scores on subjective and objective measures of executive functioning were not significantly correlated. This finding replicates previous research that did not find associations with subjective and objective measures of executive functioning in medical populations [39, 40]. Therefore, both objective tests and subjective rating scales of executive functioning should be included in future studies, as they may be measuring separate constructs and cannot be substituted for each other.

Finally, pain intensity and sleep were not significantly correlated in SCD. This finding is especially surprising since previous research has found predictive relationships between pain and objective sleep patterns in sickle cell disease [41, 42]. However, similarly to the present finding, several other studies found that sleep-disordered breathing was not related to pain in sickle cell patients [43-45]. The reported research on the sleep-pain relationship in sickle cell disease continues to be inconsistent. Researchers should investigate these variables with larger samples and use consistent measures across studies.

Limitations and future directions

This study should be considered in the context of several limitations. First, the sample size is relatively small and may have prevented the identification of other significant relationships. Moreover, patients were all preparing for a stem cell transplant; substantial morbidity is required for eligibility on the medical protocols from which our sample was drawn, and thus, results may not be generalizable to the overall population of individuals with SCD. Our patients also may exhibit characteristics (e.g., hopefulness, anxiety) that are different from other SCD populations as they prepare to undergo a potentially curative but experimental medical procedure. Additionally, sleep was examined through self-report measures rather than through more objective observational methods (e.g., actigraphy or polysomnography) for measuring sleep duration, sleep efficiency, and sleep latency. Future studies should utilize objective measures of sleep and determine whether those indices relate to cognitive functioning in ways that are consistent with the current findings.

Another important direction for future research is sleep interventions for individuals with SCD, since they have not been a focus in the literature to date. Research with healthy adolescents and adults, as well as adults with chronic pain and cancer, suggests that sleep can be significantly improved through psychoeducation [46], mindfulness-based cognitive-behavioral interventions [47], and mindfulness-based stress reduction (MBSR) [48-50]. Studies examining the efficacy of such interventions among individuals with SCD are sorely needed.

In conclusion, extant literature suggests that pain, sleep, depression symptoms, and cognitive difficulties often are reported by patients with SCD. Anxiety often has not been included in analyses investigating the relationships among these variables. In the present study, patients with SCD reported fewer sleep difficulties, increased anxiety, and decreased depression compared to previous studies examining SCD, although inconsistent measurement tools across studies make comparisons tenuous. Efforts to standardize outcome measures across studies is an important step for the SCD research community. Self-reported sleep difficulties did not predict executive dysfunction (in both objective and subjective measures). However, anxiety significantly mediated the influence of sleep difficulties on executive functioning while depression did not exhibit the same effect. These findings demonstrate the significance of anxiety in understanding sleep and cognitive functioning among individuals with SCD. Future investigations on psychological treatment for sleep and anxiety may lead to improved executive functioning in SCD.

Data availability

The data included in the study is not publicly available according to the NIH Publishing Agreement.