Neurocognitive and Emotional Factors Predict Pain-Related Healthcare Utilization in Children with Sickle Cell Disease

Abstract

Background:

Youth with sickle cell disease (SCD) experience increased rates of neurocognitive and emotional difficulties. Cross-sectional studies suggest neurocognitive and emotional functioning are associated with health outcomes in SCD. We investigated whether neurocognitive and emotional factors predicted future pain-related healthcare utilization in children with SCD.

Procedure:

112 youth with SCD between ages 7-16 years reported sociodemographics and completed measures of neurocognitive functioning and emotional well-being. The number of emergency department (ED) visits and hospitalizations for pain 1 year and 3 years after enrollment were determined by chart review.

Results:

The mean age of participants was 10.61 years (SD=2.91) with most being female (n=65; 58%). Eighty-three (74%) participants had either HbSS or HbSβ⁰ thalassemia. Regression analyses showed that attention significantly predicted ED visits and hospitalizations for pain at 1 year and 3 years after enrollment (all p values ≤.017), such that poorer attention was associated with higher healthcare utilization. Lower emotional quality of life also predicted more ED visits for pain at 3 years (b=−0.009, p=.013) and hospitalizations for pain at 3 years (b=−0.008, p=.020).

Conclusions:

Neurocognitive and emotional factors are associated with subsequent healthcare use in youth with SCD. Poor attentional control might limit implementation of strategies to distract from pain or could make disease self-management behaviors more challenging. Results also highlight the potential impact of stress on pain onset, perception, and management. Clinicians should consider neurocognitive and emotional factors when developing strategies to optimize pain-related outcomes in SCD.

Introduction

Sickle cell disease (SCD) is a hemoglobinopathy associated with hemolytic anemia, acute and chronic pain, chronic organ damage, and increased mortality.¹ Pain is the most common symptom experienced by individuals with SCD. Acute recurrent episodes of pain, known as vaso-occlusive crises (VOCs), may start as early as 4 – 6 months of age and occur throughout life. The frequency of pain increases with age, with many patients developing chronic pain.² VOCs are the primary reason for SCD-related emergency department (ED) visits and hospitalization with annual health care costs of 1.1 billion dollars.^3,4 SCD-related pain profoundly impacts quality of life and remains a significant challenge for patients, families, and treating healthcare providers.

Individuals with SCD are also at risk for mental health difficulties such as anxiety and depressive symptoms.⁵ These affective comorbidities are associated with use of maladaptive strategies for coping with pain and poorer SCD management, which have the potential to lead to greater healthcare utilization.^6,7 There is evidence that children with SCD who report more anxiety symptoms have higher admission rates for pain and longer hospital stays.⁸ In adults with SCD, depressive symptoms have been associated with higher healthcare utilization including increased ED use and hospitalizations for pain.^9,10 Emotional functioning plays an important role in both processing one’s own pain and implementing coping strategies that may influence healthcare outcomes.

SCD also contributes to direct and deleterious neurological effects that can lead to impairments in neurocognitive functioning. The most substantive neurocognitive impairments typically result from overt stroke, though deficits are also associated with the presence of silent cerebral infarcts.¹¹ Moreover, mounting evidence suggests that even patients with SCD without a history of cerebral infarct are at risk for neurocognitive deficits through varied pathways.^12–14 Robust data highlights an association between neurocognitive performance and hemoglobin and hematocrit and recent studies point to a role for other biomarkers of disease severity such as cerebral blood flow, oxygen extraction, inflammatory cytokines, and platelets.^12,15–17 Worsening neurocognitive functioning poses significant risks to patient health and quality of life due to potential for increased difficulty managing one’s disease (e.g., adhering to treatment recommendations, navigating transition to adult health care).^18–20

Neurocognitive deficits and emotional distress are common in SCD and research has shown that neurocognitive and emotional factors are associated with pain and other health outcomes in SCD.^18,21 However, the extent to which neurocognition and SCD-specific emotional functioning prospectively influence healthcare utilization is largely unknown. Therefore, we explored whether neurocognition and SCD-specific emotional distress may be associated with the frequency of pain-related hospitalization and ED visits over a three-year period. We hypothesized that lower neurocognitive functioning and greater emotional distress at baseline would be associated with increased hospitalizations and visits to the ED for SCD pain one and three years later.

Methods

The study sample consisted of 112 youth with SCD ages 7-16 years old who received care at an urban pediatric hospital in Washington, DC. Data for the current analyses came from patients who participated in one of two IRB-approved clinical trials of a computerized working memory training program that ran from 2013 to 2017. All neurocognitive and patient-reported data were collected at the baseline visit of the study before any intervention procedures were implemented. Patients were considered eligible if they had any SCD genotype, possessed adequate English fluency, and had the physical ability to use technology for assessment and intervention procedures. Patients were excluded if they had started taking a stimulant in the past 30 days. After providing informed consent and assent, youth completed an abbreviated neurocognitive assessment and filled out standardized questionnaires to report on emotional functioning. Primary caregivers also filled out a survey to report on sociodemographic characteristics. Each family received a $20 gift card and transportation-related assistance (e.g., parking pass or public transportation fare card) to lessen the financial costs associated with participating in the study. Retrospective review of the electronic medical record was completed by trained research staff to quantify healthcare utilization.

Study data were collected and managed using REDCap electronic data capture tools.²² REDCap is a secure, web-based application designed to support data capture for research studies, providing 1) an intuitive interface for validated data entry; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for importing data from external sources.

Measures

Medical characteristics.

SCD genotype was obtained by reviewing participants’ electronic medical records and coded as either HbSS and HbS-beta⁰ thalassemia or HbSC and HbS-beta⁺ thalassemia, as is common practice given phenotypic similarities.²³ Although data are mixed on the best approach to quantifying disease severity, genotype was included in analyses to control for possible effects of disease severity on outcomes. Of note, previous studies have reported that patients with more severe SCD genotypes exhibit greater neurocognitive impairment.²⁴ The literature is less compelling regarding genotypic differences in mood-related concerns; however, there does appear to be an association between disease severity, symptom burden, and psychological distress.^25–27 Ongoing treatments at the time of study enrollment (e.g., hydroxyurea, chronic blood transfusion therapy) and hemoglobin collected as part of routine outpatient hematology care were also extracted from participants’ medical records.

Sociodemographics.

Caregivers completed a sociodemographic questionnaire to report on multiple characteristics of the patient, primary caregiver, and family. Specifically, they reported the patient’s age and gender, highest level of education attained by the primary caregiver, and household income.

Healthcare utilization.

Data were extracted from a medical chart review to characterize participants’ healthcare utilization across the 3 years following study enrollment. Documentation regarding ED visits and hospitalizations were reviewed to identify the primary presenting problem. For each participant, the number of ED visits for pain and the number of hospitalizations for pain at 1- and 3-years post-enrollment were recorded. These time periods were selected to allow for analyses examining differential short- and long-term effects (i.e., 1 year and 3 years, respectively) of observed neurocognitive and emotional functioning. Research staff conducting chart reviews completed training to standardize the approach to data collection and each participant’s chart was reviewed by two coders to identify and resolve potential discrepancies.

Neurocognitive functioning.

All participants completed a measure of general intellectual functioning: either the four-subtest version of the Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI-II) or the Wechsler Intelligence Test for Children, Fifth Edition (WISC-V), depending on which study the participant was enrolled.^28,29 Scores are standardized based on data from large normative samples with mean scores of 100 and standard deviations of 15. The WASI-II and WISC-V are well-researched and validated measures that serve as a composite indicator of intelligence and their full-scale IQ scores are highly correlated (corrected r coefficient is .87).³⁰

Participants also completed cognitive functioning tests from the Cogstate computerized battery, a standardized assessment administered on a touch-screen laptop computer shown to be reliable and valid for use with diverse pediatric populations.^31–33 For the purposes of this study, we examined Cogstate tasks measuring cognitive domains commonly affected by SCD: attention, working memory, and executive functioning.

The Cogstate attention task (“Identification”) presented examinees with a playing card that was either red or black. The program asked participants to consider the question, “Is the card red?”, and press one button if the answer was “yes” and a different button if the answer was “no” as quickly as possible. Performance was assessed in terms of reaction time and resulted in age-normed standard scores where higher scores reflected better attention. The Cogstate working memory task (“One Back”) presented examinees with a series of individual playing cards and asked, “Is the previous card the same?” Participants responded by pressing a button to indicate a response of “yes” or “no”. Performance was assessed based on reaction time. The Cogstate executive functioning task (“Groton Maze Learning Test” (GMLT)), presented examinees with a 10 x 10 grid of gray squares and asked them to follow set of rules to find a hidden pathway to the end of the maze (e.g., only tap squares next to the last correct square; the path cannot move diagonally or back onto a previous part of the path). Participants tapped the gray squares to try to uncover the path and were shown a green checkmark or red “X” for correct or incorrect selections. Performance was evaluated based on the total number of errors made in trying to learn the pathway across five consecutive trials. The normative mean across Cogstate tests is 100 with a standard deviation of 10, where higher scores indicate better functioning.

Parent-rated Executive Functioning.

A primary caregiver completed the parent proxy-report form of the Behavior Rating Inventory of Executive Functioning (BRIEF).³⁴ The BRIEF is a standardized questionnaire that assesses children’s everyday executive functioning. The form consists of 86 items, which map onto eight clinical scales: Inhibit, Shift, Emotional Control, Initiate, Working Memory, Plan/Organize, Organization of Materials, and Monitor. These eight clinical scales make up a Behavioral Regulation Index and a Metacognition Index, which load onto an overall score known as the Global Executive Composite (GEC). The GEC score is representative of a child’s overall executive function.

Emotional functioning.

Select subscales from the PedsQL^™ Sickle Cell Disease Module patient self-report and parent proxy-report were used to evaluate emotional functioning.³⁵ Specifically, scores from the “Worry 1” (i.e., anxiety about common SCD complications) and “Emotions” (i.e., anger about having SCD and pain) domains were used for the current analyses given their focus on emotional reactions to having SCD, pain, and other disease-related complications. Participants were asked to report how much of a problem various issues had been for them (e.g., “I worry that others will not know what to do if I have pain”): Never, Almost Never, Sometimes, Often, or Almost Always. Similarly, primary caregivers reported their observation of how much of a problem the same issues had been of concern for their child. Ratings are transformed onto a scale ranging from 0 to 100 where higher scores indicate better quality of life. Raw scores are not converted to standardized scores; for reference, the initial validation study reported mean scores of 63.5 (SD = 26.2) for child-rated Worry I and 62.0 (SD = 33.1) for child-rated Emotions.³⁵ The same study reported mean scores of 60.2 (SD = 31.7) for parent-rated Worry I and 64.7 (SD = 32.7) for parent-rated Emotions.

Data Analysis

Linear regression models were used to explore the relationship between each of the dependent variables (ED visits for pain at 1 and 3 years after enrollment and hospitalizations for pain at 1 and 3 years after enrollment) and the independent variables, which included neurocognitive predictors (IQ, Cogstate attention, Cogstate working memory, Cogstate executive functioning, BRIEF GEC), emotional variables (child- and parent-reported PedsQL^™ Worry I domain, child and parent-reported Emotions domain), and control variables (age, patient gender, SCD genotype, parent education, and family income). All dependent variables were skewed; therefore, a log-transformation was used: log(y + 1) (with plus 1 to avoid log(0)). We applied the Akaike information criterion (AIC)³⁶ to do stepwise variable selections to select important independent variables; i.e., start from the full model (including all covariates in the model), and stepwise to obtain a final model. Estimated coefficients, standard errors, associated p-values, and R² are reported for the final model for each dependent variable. These calculations were implemented in R.³⁷

Results

Sociodemographic characteristics of the study sample are reported in Table I. The median number of ED visits for pain over a period of 1 year from baseline was 1 visit (IQR = 2) and 3 visits (IQR = 6) at 3 years after enrollment. The median number of hospital admissions for pain over a period of 1 year from baseline was 0 admissions (IQR = 1) and 1 admission (IQR = 4) at 3 years after enrollment. Healthcare utilization variables did not differ by gender or SCD genotype and were not significantly correlated with age or hemoglobin. Descriptive statistics for cognitive and emotional functioning independent variables are presented in Table II. Age was negatively correlated with IQ (r = −.23, p = .02) and working memory performance (i.e., Cogstate One Back task; r = −.20, p = .04) and positively correlated with parent-rated executive functioning difficulties (r = .22, p = .02). Cognitive measures did not differ by gender, SCD genotype, or hemoglobin. Furthermore, emotional functioning variables were not associated with age, gender, SCD genotype, or hemoglobin.

Table I.

Sociodemographic characteristics of the sample

	Mean or n (%)	Standard Deviation
Age	10.61	2.91
Sex (Female)	65 (58%)
Genotype (HbSS or HbSβ0 thalassemia)	83 (74%)
Prescribed hydroxyurea	57 (51%)
Receiving chronic blood transfusion therapy	29 (26%)
Hemoglobin (g/dL)	9.51	1.57
Prescribed stimulant medication for attention problems	3 (3%)
Primary caregiver highest education attained
Less than high school diploma	11 (10%)
High school diploma	24 (21%)
Some college	28 (25%)
College degree	28 (25%)
Master’s or doctorate-level degree	19 (17%)
Declined to answer	2 (2%)
Household income
≤ $49,999	58 (52%)
$50,000 – $99,999	21 (19%)
$100,000 - $149,999	15 (13%)
≤ $100,000	11 (10%)
Declined to answer	7 (6%)

Table II.

Descriptive statistics for cognitive and emotional functioning variables

	Mean	Standard Deviation
WASI-II/WISC-V
Estimated IQ	91.03	13.34
Cogstate
Identification (attention)	94.17	10.86
One Back Speed (working memory)	91.36	11.47
Groton Maze Learning Test (executive functioning)	89.14	13.57
BRIEF
Global Executive Composite	56.58	11.35
PedsQL™ Child Self-Report
Worry I	60.40	29.62
Emotions	62.39	35.46
PedsQL™ Parent Proxy-Report
Worry I	70.14	27.22
Emotions	64.91	29.93

Note. The population mean for the WASI-II and WISC-V estimated IQ is 100 with a standard deviation of 15. The population mean for Cogstate tasks is 100 with a standard deviation of 10. The BRIEF Global Executive Composite population mean is 50 with a standard deviation of 10. PedsQL^™ scores range from 0 to 100 where higher scores indicate better QOL (i.e., fewer problems with worry or emotions).

Findings from a series of linear regression analyses are presented in Tables III and IV, with each table displaying results of models with different dependent variables: predicting ED visits for pain at 1 year and at 3 years (Table III) and hospitalizations for pain at 1 year and at 3 years (Table IV). In the final model obtained using the AIC criterion, the Cogstate attention task (“Identification”) significantly predicted the number of ED visits for pain at 1 year post-enrollment (p = .001). Specifically, better attentional performance predicted fewer ED visits for pain over 1 year. The Cogstate attention task also significantly predicted ED visits for pain at 3 years (p = .004), with better attention associated with fewer ED visits for pain over 3 years. In addition, higher scores on the parent proxy-report PedsQL^™ Emotions subscale (i.e., fewer SCD-related emotional problems) significantly predicted fewer ED visits for pain over 3 years (p = .013).

Table III.

Estimated coefficients and standard errors with associated p-values based on the final model selected by the AIC criteria predicting ED visits for pain.

	ED visits for pain at 1 year
	Estimate	Std. Error	t value	Pr(>|t|)
(Intercept)	3.064*	0.856	3.581	0.001
Attention	−0.025*	0.007	−3.543	0.001
Executive functioning	0.009	0.006	1.542	0.127
Parent-reported executive functioning	−0.013	0.007	−1.898	0.061
Parent-reported emotional QOL	−0.004	0.003	−1.454	0.150
Caregiver Education	0.118	0.071	1.661	0.101
Household Income	−0.073	0.043	−1.698	0.093
	ED visits for pain at 3 years
	Estimate	Std. Error	t value	Pr(>|t|)
(Intercept)	4.368*	1.138	3.837	0.000
Attention	−0.030*	0.010	−3.003	0.004
Executive functioning	0.015	0.008	2.007	0.048
Parent-reported executive functioning	−0.017	0.009	−1.819	0.072
Parent-reported emotional QOL	−0.009*	0.003	−2.543	0.013

Note. R² = .206 for the model predicting ED visits at 1 year. R² = .174 for the model predicting ED visits at 3 years. Attention is measured using the Cogstate Identification task, executive functioning is measured using the Cogstate Groton Maze Learning Test, parent-reported executive functioning is measured using the BRIEF Global Executive Composite, and SCD emotional QOL is measured using the PedsQL^™ Emotions Parent Proxy-Report subdomain.

Table IV.

Estimated coefficients and standard errors with associated p-values based on the final model selected by the AIC criteria predicting hospitalizations for pain.

	Hospitalizations for pain at 1 year
	Estimate	Std. Error	t value	Pr(>|t|)
(Intercept)	2.263*	0.691	3.275	0.002
Attention	−0.015*	0.006	−2.678	0.009
Executive functioning	−0.010	0.006	−1.686	0.096
Parent-reported emotional QOL	−0.003	0.002	−1.559	0.123
Caregiver Education	0.133*	0.060	2.229	0.028
Household Income	−0.065	0.036	−1.799	0.076
	Hospitalizations for pain at 3 years
	Estimate	Std. Error	t value	Pr(>|t|)
(Intercept)	2.904*	1.124	2.583	0.012
Attention	−0.023*	0.010	−2.432	0.017
Executive functioning	0.013	0.007	1.743	0.085
Parent-reported executive functioning	−0.012	0.009	−1.411	0.162
Child-reported emotional QOL	−0.004	0.003	−1.384	0.170
Parent-reported emotional QOL	−0.008*	0.003	−2.364	0.020
Caregiver Education	0.132	0.077	1.723	0.089

Note. R² = .177 for the model predicting hospitalizations at 1 year. R² = .214 for the model predicting hospitalizations at 3 years. Attention is measured using the Cogstate Identification task, executive functioning is measured using the Cogstate Groton Maze Learning Test, parent-reported executive functioning is measured using the BRIEF Global Executive Composite, and SCD emotional QOL is measured using the PedsQL^™ Emotions Parent Proxy-Report subdomain.

In the regression model with number of hospitalizations for pain at 1 year post-enrollment as the dependent variable, the Cogstate attention task significantly predicted hospitalizations for pain (p = .009), with better attention associated with fewer hospitalization for pain. Number of hospitalizations for pain at 3 years was significantly predicted by performance on the Cogstate attention task (p = .017) and parent proxy-report PedsQL^™ Emotions subscale scores (p = .020). Better performance on the Cogstate attention task and higher scores on the PedsQL^™ Emotions subscale (i.e., fewer SCD-related emotional problems) were associated with fewer hospitalizations for pain 3 years after enrollment.

In post-hoc analyses aimed at further explicating the relationship between attention and pain-related healthcare utilization, we conducted a Kruskal-Wallis test to examine whether healthcare utilization varied by level of attentional ability. Participants were categorized as having Very Low attention functioning (standard scores < 80; n = 14), Below Average attention functioning (80 – 89; n = 17), or Average to Above Average attention functioning (90 or higher; n = 79). Median healthcare utilization statistics by attentional classification are presented in Figures 1 and 2. Results showed that there was a significant difference in ED visits and hospitalizations for pain 1 year after enrollment, such that participants in lower ranges of attentional functioning demonstrated higher healthcare utilization, χ²(2) = 10.30, p = .006 and χ²(2) = 12.30, p = .002, respectively. Dunn’s pairwise tests revealed differences in ED visits for pain across 1 year between the Very Low and Average to Above Average attention groups (p = .007) and between the Below Average and Average to Above attention groups (p = .030). Pairwise tests also identified differences in hospitalizations for pain across 1 year between the Very Low and Average to Above Average attention groups (p < .001). There were no significant differences between attentional functioning classification and ED visits or hospitalizations for pain 3 years after enrollment, though was a trend toward significance for differences in hospitalizations over 3 years, χ²(2) = 5.38, p = .068.

Figure 1.

Differences in Median Number of Emergency Department Visits for Pain by Attention Functioning

Figure 2.

Differences in Median Number of Hospitalizations for Pain by Attention Functioning

Discussion

This study examined the role of neurocognitive functioning and disease-specific emotional well-being as predictors of healthcare utilization for pain in children with SCD. Results showed that lower neurocognitive functioning, specifically in attentional abilities, predicted higher pain-related ED visits and hospitalizations 1 and 3 years later. Parent ratings of their child has having lower emotional quality of life (e.g., frequently feeling angry about having SCD) predicted higher ED visits and hospitalizations for pain 3 years later. These results demonstrate that children’s neurocognitive and emotional functioning play an important role in pain outcomes and should be considered when developing comprehensive pain management plans for children with SCD.

Attentional abilities emerged as a reliable predictor of pain-related healthcare utilization and could signal that poor attention makes it difficult to implement non-pharmacologic strategies that are frequently recommended to distract from pain (e.g., shifting attention away from pain and towards an alternate stimulus). Optimal performance on tasks measuring attention involves strong attentional control, or the ability to sustain and intentionally direct focus towards a desired stimulus while inhibiting a response to an undesired stimulus. Prior studies have shown that intentionally attending to alternate stimuli during experimentally induced pain is associated with reduced activity in neural regions responsible for pain processing and lower pain intensity.^38,39 Therefore, our findings that better attentional abilities are associated with fewer ED visits and hospitalizations for pain could reflect skillful implementation of distraction strategies that lead to reduced experiences of pain.

Another possible explanation is that children with poorer attention have greater difficulty managing behavioral aspects of their treatment, though this was not directly assessed. The current study included both children and adolescents so there was likely variability in the degree to which patients assumed responsibility for disease self-management tasks. However, there is evidence that neurocognitive factors influence medication adherence particularly for adolescents,^40,41 with individuals with poorer attention having difficulties with forgetfulness, distractibility, and disorganization that potentially interfere with medication-taking and other aspects of disease self-management. As a result, patients may be less likely to engage in preventive or health-promoting behaviors (e.g., active avoidance of pain triggers) or to experience therapeutic benefits of prescribed medications that are not taken consistently. Given the known heritability of attentional abilities,⁴² it is also possible that the parents of children with attention problems have their own cognitive challenges that interfere with execution of disease management tasks. These are important considerations, given that a significant proportion of youth with SCD will experience neurocognitive sequelae. Current results suggest the negative effects of attention problems are not limited to academic domains, but also include pain-related healthcare utilization.

Emotional functioning—specifically, feeling angry about having SCD and pain—also predicted the number of future ED visits and hospitalizations for pain, highlighting the impact of psychological factors on pain perception, onset, and management. This connection between mood and pain outcomes is well-documented in the broader pediatric pain literature,⁴³ and ample evidence supports a link between emotional functioning and pain in SCD.^6,44 However, studies have not previously examined the influence of emotions on pain-related healthcare utilization in pediatric SCD prospectively.

Although there is some overlap in the mechanisms through which mood and attention modulate pain, mood is known to exert a particularly salient influence on perceptions of pain unpleasantness via distinct neural systems.^39,45 It is also possible that persistent feelings of frustration and anger about SCD lead to heightened stress, which triggers marked vascular responses in individuals with SCD with the potential to contribute to the onset of VOCs.⁴⁶ Prior research also highlights a relationship between mood and use of cognitive-behavioral coping strategies that has implications for pain management. For example, research has shown that that anxiety and depression are associated with maladaptive coping strategies such as denial, avoidance, and wishful thinking, which in turn lead to heightened experiences of pain, engagement in behaviors that increase risk for pain, and pain interference.^47,48

Findings from the present study must be interpreted with caution and in consideration of several limitations. For instance, although the original studies from which these data were extracted were not limited to patients with neurocognitive deficits, it is possible that patients and families were more motivated to enroll due to potential benefits of participating in the neurocognitive training intervention studies. Therefore, it is possible that results reflect relationships observed in a subset of children with SCD and are not generalizable to all patients; though, it should be noted that participants exhibited a wide range of neurocognitive performance, with many scoring in the Average range or higher on assessments. Although no participants experienced an overt stroke during the study period, it is possible that new silent infarcts occurred of which we were unaware as we did not conduct brain MRI scans. However, the risk for cerebral infarcts in SCD is high early in childhood^49–51 and given that the mean age of participants was 11 years and participants known to be at risk for stroke were receiving chronic blood transfusion therapy, we estimate that there was a low risk for new infarcts.

Participants may have experienced effects of the cognitive training intervention or a host of other possible treatments across the study period. However, only about half of the study sample received the intervention and half of those completed a substantive portion of the training.^52,53 Moreover, immediate effects of the cognitive intervention were modest and although data on long-term maintenance of treatments effects are lacking in SCD, effects have been shown to wane in other populations.⁵⁴ We also did not restrict participation in any clinically indicated treatments; therefore, it is possible that participants were exposed to community-based interventions for cognition or mood for which we are unable to account. Future studies could attempt to overcome this limitation by leveraging observational natural history cohorts and carefully monitoring exposure to treatments outside of clinic.

Another limitation is that we assessed emotional functioning at a single point in time and it is possible that responses to questionnaires were influenced by acute situational factors. We posit that child and parent ratings of emotional functioning represent generally stable measures as the PedsQL SCD Module has demonstrated good reliability and asks specifically about SCD-specific worries and emotions, which are likely to have been formed over many years.³⁵ However, we are not able to confirm these hypotheses and future studies should evaluate the stability of SCD-specific emotional functioning and temporal relationships with health outcomes. Also related to measurement limitations, we relied on two short subscales from the PedsQL SCD Module. These measures were selected due to their focus on SCD-specific emotional functioning and have been shown to be significantly correlated with the PedsQL^™ Generic Cores Scales Psychosocial Health Summary Score.³⁵ However, given that they only contain 2-5 items, it is possible that they are not fully capturing the emotional experiences of youth with SCD and future studies should incorporate more comprehensive assessments.

We are also unable to confirm directionality of the findings. Although regression models showed that neurocognitive and emotional factors predicted future ED use and hospitalizations for pain after controlling for SCD genotype, it is possible that children with severe disease phenotypes who are at higher risk for pain may also be at increased risk for neurocognitive and mood-related comorbidities. Nonetheless, our findings underscore the association between neurocognitive and SCD-specific emotional functioning with pain burden. Future longitudinal cohort studies are needed to confirm directionality and mediators of these associations as well as the impact of treating psychological and neurocognitive comorbidities on long-term pain outcomes. We recommend that assessment of attention and SCD-specific emotional distress be included as integral components of comprehensive SCD and pain management plans, which may help improve outcomes.

Abbreviations:

AIC

Akaike information criterion

BRIEF

Behavior Rating Inventory of Executive Functioning

ED

Emergency department

GEC

Global Executive Composite

GMLT

Groton Maze Learning Test

IQR

Interquartile range

SCD

Sickle cell disease

SD

Standard deviation

VOCs

Vaso-occlusive crises

WASI-II

Wechsler Abbreviated Scale of Intelligence, Second Edition

WISC-V

Wechsler Intelligence Test for Children, Fifth Edition