COGNITIVE PERFORMANCE AS A PREDICTOR OF HEALTH CARE TRANSITION IN SICKLE CELL DISEASE

Abstract

Neurocognitive deficits in sickle cell disease (SCD) may impair adult care engagement. We investigated the relationship between neurocognitive functioning and socio-environmental factors with health care transition outcomes. Adolescents ages 15–18 years who had neurocognitive testing and completed a visit with an adult provider were included. Transition outcomes included transfer interval from pediatric to adult care and retention in adult care at 12 and 24 months. Eighty adolescents (59% male, 64% HbSS/HbSβ⁰-thalassemia) were included. Mean age at adult care transfer was 18.0 (±0.3) years and transfer interval was 2.0 (±2.3) months. Higher IQ (p=0.02; pFDR=0.05) and higher verbal comprehension (p=0.008; pFDR= 0.024) were associated with <2 and <6 month transfer intervals, respectively. Better performance on measures of attention was associated with higher adult care retention at 12 and 24 months (p=0.009; pFDR=0.05 and p=0.04; pFDR=0.12, respectively). Transfer intervals <6 months were associated with smaller households (p=0.02; pFDR=0.06) and households with fewer children (p=0.02; pFDR=0.06). Having a working parent was associated with less retention in adult care at 12 and 24 months (p=0.01; p=0.02, respectively). Lower IQ, verbal comprehension, attention difficulties, and environmental factors may negatively impact transition outcomes. Neurocognitive function should be considered in transition planning for youth with SCD.

INTRODUCTION

In the United States, >95% of children with sickle cell disease (SCD) survive into adulthood,^1–4 but experience an increase in SCD-related complications,⁵ acute care utilization,⁶ and mortality shortly after leaving pediatric care.⁷ Care continuity during health care transition may improve health outcomes through early identification and treatment of disease complications.

Among individuals with SCD, ^8,9 cognitive dysfunction is prevalent, ^10–12 including deficits in attention, processing speed, IQ, and executive functions.^13–16 Cognitive impairments may reduce the ability to assimilate concepts and organize tasks,¹⁷ which potentially diminishes transition readiness (e.g., medication management, planning of medical visits). Executive dysfunction has been linked to adaptation and transition to adult care among adolescents with chronic medical conditions,^18,14 but it has not been adequately studied in SCD.

Our group has previously shown that pediatric patients who performed better on measures of attention were more likely to establish care with an adult provider; for every 1-point decrease in age-standardized score (less attention deficit), the odds of completing a first adult visit within 6 months from leaving pediatric care increased three-fold (OR = 3.2).¹⁹ The relationship between pediatric cognitive function and integration into adult health care has not been investigated, however. To test the hypothesis that neurocognitive deficits would negatively affect health care transition outcomes, we investigated the relationship between cognitive function prior to transfer to adult care and 1) latency to engage in adult care, 2) 12 and 24-month adult care retention, and 3) utilization of adult ambulatory services. Secondarily, we investigated the association between socio-environmental factors, disease severity, and disease-modifying treatment (chronic transfusions, hydroxyurea treatment) with transition outcomes.

METHODS

Study design and participants

Participants were eligible for this study if they met the following criteria: 1) followed by the IRB-approved Sickle Cell Clinical and Intervention Program (SCCRIP), a prospective-retrospective longitudinal lifetime cohort study that collects clinical, neurocognitive, geospatial, psychosocial, and health outcomes,²⁰ 2) received care at St. Jude Children’s Research Hospital (St. Jude), 3) completed their first adult visit after leaving pediatric care between January 2013-December 2017, and 4) received neurocognitive screening between ages 15 and 18 years.

In SCCRIP, all patients are approached consecutively in the outpatient setting. Most (99%) of the active SCD patient population at our institution has been enrolled into SCCRIP. All participants, or their legal guardians, provided informed consent. SCCRIP participants receive the same transition preparation services regardless of neurocognitive ability through the St. Jude Transition to Adult Care Program.²¹ Our transition program has been described in detail previously. ²¹ In brief, our program spans the ages of 12–25 years, prepares, tracks transition readiness, and completes transfer to adult care at age 18 years in >90% of participants within 6 months.²¹ The transition policy is presented to the patient at age 12. Prior to leaving pediatric care, the program assists in identification of an adult provider and scheduling of first visit. Once the patients transfer to adult care, they are no longer seen at the pediatric hospital, and are followed by separate partner institutions where care is co-located by pediatric and adult providers.²²

Neurocognitive Measures

SCCRIP participants receive neurocognitive screening every 4 years starting at age 4.^20,23 As these assessments were not clinical referrals, participants were not selected for prior neurologic findings, disease severity, or concern for cognitive decline. The screening battery for adolescents included measures of estimated IQ (Wechsler Abbreviated Scale of Intelligence, Second Edition [WASI-2], 4-subtest Full Scale IQ²⁴), sustained attention (Conners’ Continuous Performance Test, Second Edition [CPT-2]²⁵ or Third Edition [CPT-3]²⁶, executive function (Trail Making Test, Conditions 2 and 4; Delis-Kaplan Executive Function System [DKEFS]²⁷), and processing speed (Coding subtest from age dependent Wechsler measure ²⁸). All measures have established reliability, validity, and clinical utility and are age standardized. Higher scores on IQ and executive function measures indicate better performance (Standard Scores [SS], Mean=100, SD=15; Scaled Scores [ScS], Mean=10, SD=3); whereas, higher scores on the attention measure indicate worse performance (T-scores, Mean=50, SD=10).

Socio-environmental and Clinical Factors

Socio-environmental factors included the economic hardship index (EHI),²⁹ primary guardian education and working status, and number of persons living in the household. Higher scores indicate higher level of neighborhood hardship. EHI score and socio-environmental data were calculated at time of cognitive testing.

Clinical variables were abstracted from the medical record or SCCRIP database. Disease severity was defined as prior stroke or prior abnormal mean maximum Transcranial Doppler velocity (≥200 cm/sec). Disease-modifying exposure was defined as treatment with hydroxyurea or transfusions in childhood, therefore including the time when they received cognitive testing. Utilization of adult ambulatory services was calculated as yearly number of outpatient visits during the first two years in adult care.

Statistical Analysis

We examined two main health care transition outcomes: transition latency and adult care retention. Transition latency was defined as the interval between the last pediatric visit and the first adult visit. We examined 2 transition latencies: 1) within 6 months, based on national guidelines,³⁰ and 2) within 2 months, as an exploratory analysis, based on recommendations from the 2017 Delphi survey.³¹ Adult care retention was defined as continuous engagement in adult care (with the same or different adult SCD provider) at 12 and 24 months timepoints, as previously studied.³² We analyzed both transition outcomes as binary and continuous variables (interval from: last pediatric to first adult visit, and from first to last adult visit).

The relationships between predictors (neurocognitive performance, environment, disease severity, treatment exposure) and transition outcomes were assessed using independent sample t-tests if the data were normally distributed or the Wilcoxon rank-sum test otherwise. Categorical variables were analyzed with Fisher’s exact test. The analyses were adjusted for EHI (except for latency to adult care at 6 months due to small sample size) using generalized linear regression. Time to a transition outcome was conducted using Cox proportional regression analysis. Censoring was applied at their last follow up visit. The proportional hazard assumption was assessed with Schoenfeld residuals for each variable. Cohen’s d³³ was calculated for estimates of effect size for the neurocognitive variables. For 6-month latency to adult care, a permutation-based analysis was performed to account for small sample size.

The yearly adult routine visits rate was calculated by dividing the number of adult routine visits by the interval from first to last adult visits and analyzed as a continuous variable using a logarithm transformation. Univariate linear regression analysis was performed to test the association between covariates and adult ambulatory services. Covariates with p-values <0.1 were further assessed in multivariate analysis using backward model selection based on Akaike information criterion. For continuous outcomes, Shapiro-Wilk test was used to test for normality of the data or model residuals.

Because of its higher clinical severity, we repeated all analyses for the HbSS/HbSβ⁰-thalassemia sickle genotypes subgroup, except for transition latency at 6 months due to small sample size. All statistical tests were two-sided, and p<0.05 were considered statistically significant. False discovery rate (FDR) adjusted p-values (p_FDR) were calculated due to multiple comparisons.

RESULTS

Participants’ Characteristics

Between January 2013 and December 2017, 178 patients initiated adult care (Supplemental Figure 1). Of these, 80 completed a neurocognitive screening assessment in adolescence. They were similar in demographics and transition outcomes to the 98 youth with SCD who transferred to adult care within the same period but did not complete neurocognitive testing (Table 1, Supplemental Table 1).

Table 1.

Sample Demographics.

		Completed Neurocognitive Screening N=80	No Neurocognitive Screening N=98	p-value	pFDR
Sex, n (%)
	Male	47 (59)	52 (53)	0.45	0.64
	Female	33 (41)	46 (47)
Age at Cognitive Testing (years)				NA	NA
	Mean (SD)	16.5 (0.6)	NA
	Median (Range)	16.5 (15.0–18.0)	NA
Age at Last Pediatric Visit (years)				0.34	0.57
	Mean (SD)	18.3 (0.3)	18.2 (0.3)
	Median (Range)	18.2 (17.3–18.9)	18.1 (17.0–19.0)
Sickle Genotype, n (%)				0.82	0.82
	HbSS/HbSβ0-thalassemia	51 (64)	59 (60)
	HbSC/HbSβ+thalassemia	28 (35)	38 (39)
	Other	1(1)	1(1)
Disease-modifying therapies, n (%)		80	98
	Hydroxyurea	55 (69)	48 (49)	0.01	0.1
	Chronic transfusion	12 (15)	23 (23)	0.19	0.57
Disease severity, n (%)		80	98
	History of overt stroke in childhood	1(1)	7 (7)	0.08	0.40
	History of abnormal TCD in childhood	8 (14)*	8 (15)*	0.79	0.82
Economic Hardship Index	N	80	96	0.33	0.57
	Median (Range)	42.5 (22.2–70.5)	41.0 (23.1–64.3)
Adult Care Retention, n (%)
	12-Months	50 (82)	54 (90)	0.30	0.57
	24-Months	31 (72)	36 (68)	0.82	0.82

Two-sample t- or Wilcoxon rank sum test to compare continuous variables and Fisher’s exact test to compare categorical variables between two groups. p<0.05 indicates significance (two-sided). False discovery rate (FDR) adjusted p-values >0.05 for all tests. NA = not available.

^*Data only available for 59 (73.8%) and 52 (53.1%) of patients who completed and did not complete neurocognitive screening, respectively.

Of the 80 adolescents included, the mean (±1SD) age at neurocognitive screening was 16.5 (±0.6) years. The majority were sickle genotype HbSS/HbSβ⁰ thalassemia (64%), and all patients identified as African American/Non-Hispanic (Table 1). Fifteen percent of patients received chronic transfusions and 69% were prescribed hydroxyurea. Further, 14% were identified as having severe disease, indicated by history of prior stroke or abnormal TCD (Table 1). A subset of these 80 patients were included in our preliminary work that investigated the relationship between attention and adult care transfer. ¹⁹

Healthcare Transition Outcomes

Of the 80 patients, 64 (80%) and 76 (95%) transferred to adult care within 2 and 6 months, respectively. The mean (±1SD) age at time of transfer was 18 (±0.3) years. There were no significant differences in age at neurocognitive testing and demographics among individuals who transferred to adult care <2 vs. ≥2 months or <6 vs. ≥6 months. Of the 80 participants, 61 and 43 had sufficient follow-up time to investigate retention in adult care at 12 and 24 months, respectively (Supplemental Figure 1). Fifty (82%) remained in adult care 12 months, whereas 31 (72%) remained in adult care 24 months after engaging in adult care.

Predictors of Transition Latency

Overall mean (SD) group performance on the estimated IQ measure was within the low average range of WASI-2 4-subtest Full Scale IQ 83.9 (±10.9) points.²⁴ As a group, participants demonstrated below average range scores on the verbal comprehension 85.2 (±11.5) and perceptual reasoning 84.9 (±12.1) indexes.²⁴

Participants who completed a first adult visit <2 months from the last pediatric visit had better mean performance on estimated IQ in comparison to those who completed their first adult visit ≥2 months [85.03 (±11.4), 79.4 (±6.9), p=0.02, p_FDR=0.05; Table 2). After adjusting for EHI and sickle genotype, estimated IQ remained associated with completion of first adult visit <2 months (estimate=0.07, standard error/se=0.034, p=0.042). Among participants who initiated adult care <6 months from their last pediatric visit, estimated mean verbal comprehension performance was higher than those who established adult care ≥6 months from completing pediatric care [85.5 (±11.8), 79 (±2), p_perm=0.008, p_FDR=0.024, respectively]. When the same comparisons were repeated in the HbSS/HbSβ⁰-thalassemia subgroup, similar results were seen (Supplemental Table 2). No other differences in cognitive measures were observed relative to transfer latency.

Table 2.

Comparison of Neurocognitive Performance for participants with SCD according to transfer latency and care retention.

					Latency in Adult Care
			ALL	< 2 months	≥2 months				< 6 months	≥6 months
			N=80	N=64	N=16				N=76	N=4
Domain	Measure	Score	M (SD)	M (SD)	M (SD)	p †	pFDR *	d	M (SD)	M (SD)	pperm †	pFDR *	d
Estimated IQ	WASI-II (SS)	Full Scale IQ	83.87(10.84)	85.03(11.42)	79.44(6.86)	0.02	0.05	0.53	84.05(11.08)	80.5(3.51)	0.14	0.21	0.33
		Verbal Comprehension	85.19(11.54)	86.22(12.24)	80.91(6.86)	0.06	0.10	0.47	85.54(11.76)	79(2)	0.008	0.024*	0.58
		Perceptual Reasoning	84.94(12.11)	85.75(12.74)	81.71(8.79)	0.18	0.18	0.34	85.06(12.4)	83(6.06)	0.68	0.68	0.17
Attention	CPT-II (T)	Omissions	52.26(11.25)	50.5(8.64)	61.78(18.48)	0.06	0.2	1.10	52.23(11.53)	52.6(7.61)	0.58	0.93	0.03
		Commissions	52.6(10.44)	51.78(10.4)	57.06(10.29)	0.09	0.2	0.52	52.57(10.22)	52.99(16.04)	0.92	0.93	0.04
		Hit Reaction Time	44.64(10.8)	43.48(9.91)	50.93(13.96)	0.22	0.22	0.72	44.4(10)	48.03(22.33)	0.77	0.93	0.34
		Variability	54.57(11.78)	53.56(11.3)	60.02(13.8)	0.22	0.22	0.57	55.01(11.92)	48.34(9.03)	0.37	0.93	0.58
		Detectability	51.86(9.73)	51.22(9.97)	55.33(8.03)	0.17	0.22	0.43	51.88(9.71)	51.56(12.21)	0.93	0.93	0.04
	WISC-IV/ WAIS-IV (SS)	Digit Span	8.21(2.1)	8.28(2.22)	7.93(1.58)	0.64	0.78	0.17	8.24(2.12)	7.75(1.89)	0.64	0.64	0.23
Executive Function	DKEFS (ScS)	Color Naming	7.95(2.79)	7.82(2.78)	8.4(2.95)	0.59	0.59	0.21	7.92(2.74)	8.33(4.16)	0.85	0.85	0.15
		Trail Making (Number)	7.14(3.42)	7.36(3.33)	6.08(3.82)	0.29	0.55	0.38	7.08(3.42)	8.25(3.86)	0.56	0.75	0.35
		Trail Making (Number-Letter)	5.36(3.65)	5.52(3.57)	4.58(4.06)	0.38	0.55	0.26	5.24(3.62)	7.25(4.19)	0.31	0.75	0.56
		Verbal Fluency (Category)	8.38(2.89)	8.22(2.76)	9(3.42)	0.44	0.55	0.27	8.25(2.79)	11.33(4.16)	0.34	0.75	1.09
		Verbal Fluency (Letter)	8.38(3.03)	8.59(3.02)	7.6(3.02)	0.27	0.55	0.33	8.32(2.98)	9.5(4.04)	0.60	0.75	0.39
Processing Speed	WISC-IV/ WAIS-IV (SS)	Coding	7.53(2.33)	7.57(2.28)	7.38(2.58)	0.78	0.78	0.08	7.45(2.31)	9(2.45)	0.29	0.58	0.68
				Retention in Adult Care
			ALL		< 12 months	≥12 months				ALL	< 24 months	≥24 months
			N=61		N=11	N=50				N=43	N=12	N=31
Domain	Measure	Score	M (SD)		M (SD)	M (SD)	p †	pFDR*	d	M (SD)	M (SD)	M (SD)	p †	pFDR*	d
Estimated IQ	WASI-II (SS)	Full Scale IQ	83.81(11.63)		79.6(9.8)	84.69(11.88)	0.17	0.26	0.45	81.32(10.45)	80.45(9.72)	81.63(10.85)	0.74	0.78	0.11
		Verbal Comprehension	84.79(12.76)		79.89(10.93)	86.31(13.07)	0.16	0.26	0.52	81.58(9.35)	80.9(10.79)	82.07(8.58)	0.78	0.78	0.13
		Perceptual Reasoning	83.86(12.67)		81(9.43)	84.48(13.28)	0.37	0.37	0.28	81.12(11.02)	82.1(9.55)	80.71(11.75)	0.72	0.78	0.13
Attention	CPT-II/III (T)	Omissions	50.93(10.23)		50.47(5.27)	51.02(11.02)	0.52	0.68	0.06	52.14(11.53)	49.48(5.48)	53.17(13.16)	0.74	0.74	0.33
		Commissions	52.4(9.78)		61.42(6.85)	50.59(9.33)	0.0089	0.05	1.2	53.90(8.95)	59.31(8.38)	51.80(8.46)	0.04	0.12	0.93
		Hit Reaction Time	43.14(9.88)		35.3(12.07)	44.71(8.79)	0.051	0.12	1	41.94(10.41)	35.02(11.04)	44.64(9.09)	0.029	0.12	1.00
		Variability	53.89(11.82)		56.24(10.55)	53.42(12.17)	0.57	0.68	0.24	55.04(11.88)	53.39(12.22)	55.68(12.05)	0.68	0.74	0.20
		Detectability	52.1(7.8)		57.44(6.28)	51.03(7.71)	0.058	0.12	0.88	52.94(7.28)	55.88(7.06)	51.8(7.23)	0.22	0.44	0.59
	WISC-IV/ WAIS-IV (SS)	Digit Span	8.25(2)		8(2.14)	8.31(1.99)	0.67	0.67	0.16	8.15(1.86)	8.17(2.12)	8.14(1.78)	0.95	0.95	0.01
Executive Function	DKEFS (ScS)	Color Naming	8.32(2.04)		7.4(2.97)	8.52(1.81)	0.46	0.58	0.57	8.24(2.07)	8.00(3.03)	8.32(1.77)	0.95	0.95	0.16
		Trail Making (Number)	7.56(3.27)		6.78(3.31)	7.72(3.28)	0.45	0.58	0.29	7.51(3.3)	6.8(3.12)	7.80(3.39)	0.41	0.63	0.31
		Trail Making (Number-Letter)	5.71(3.67)		4.44(3.81)	5.98(3.63)	0.22	0.55	0.43	5.63(3.57)	4.8(3.77)	5.96(3.51)	0.37	0.63	0.33
		Verbal Fluency (Category)	8.6(3.16)		7.12(3)	8.87(3.14)	0.16	0.55	0.57	8.32(3.31)	7.67(3.24)	8.54(3.36)	0.50	0.63	0.27
		Verbal Fluency (Letter)	8.78(3.19)		8.56(2.4)	8.82(3.35)	0.78	0.78	0.08	8.50(3.19)	9.10(2.85)	8.29(3.32)	0.47	0.63	0.26
Processing Speed	WISC-IV/ WAIS-IV (SS)	Coding	7.93(2.25)		8.45(3.11)	7.82(2.03)	0.53	0.67	0.29	8.24(2.27)	8.75(3.14)	8.03(1.85)	0.31	0.62	0.32

Notes: SS: standard score (M=100, SD=15); ScS: scaled score (M=10, SD=3); T: T-score (M=50, SD=10); Z: Z-score (M=0, SD=1), WASI: Wechsler Abbreviated Scale of Intelligence, CPT: Conners’ Continuous Performance Test, WSIC: Wechsler Intelligence Scales for Children, WAIS: the Wechsler Adult Intelligence Scale, DKEFS: Delis-Kaplan Executive Function System

^†Two-sample t-tests or Wilcoxon rank sum tests to compare two groups; p<0.05 indicates significance; d denotes Cohen’s d effect size.

^*FDR adjusted p (p_FDR)<0.05 was displayed. p_perm was calculated based on the permutation test due to small number.

Individuals living in smaller households and in households with fewer children were more likely to transfer to adult care <6 months from a last visit to pediatric care (p_perm=0.02; 0.02, respectively; p_FDR=0.06 for both, supplemental Table 3). There was no relationship between disease severity, exposure to disease-modifying therapies including chronic transfusion or hydroxyurea, primary guardian education, or primary guardian work status and latency to adult care.

Predictors of Adult Care Retention

Group means were all within age expectations, however, patients who remained in adult care longer demonstrated less impulsivity as measured by the CPT Commissions Index (i.e., test responses were given when no letter targets were presented), at both 12 months [50.6 (±9.3) versus 61.42 (±6.9), p=0.01, p_FDR=0.05, Figure 1a] and 24 months [51.8 (±8.5) versus 59.3 (±8.4), p=0.04, p_FDR=0.12; Table 2, Figure 1b]. Individuals who remained with the adult provider at 24 months demonstrated a slower reaction time (i.e., average speed of correct responses), compared to the group that did not remain in adult care by 24 months (CPT Hit Reaction Time [44.6 (±9.1) versus 35.0 (±11.0), p=0.03, p_FDR=0.12, Table 2].

Figure 1.

Neurocognitive Performance and Retention in Adult Care. Better performance on the CPT-2 test was predictive of retention in adult care greater than 12 months (panel a) and 24 months (panel b). Green indicates favorable outcome and grey indicates an unfavorable outcome. *Lower scores indicate better performance.

In multivariate logistic regression analysis, when controlling for EHI, CPT Hit Reaction Time was no longer associated with retention in adult care, but CPT Commissions remained significant (estimate=−0.14, se=0.072, p=0.045). When controlling for EHI, CPT Omissions (i.e., failure to respond when letter targets were presented), became significant (estimate=−0.19, se=0.091, p=0.037) at 12 months, but not at 24 months.

Within the HbSS/HbSβ⁰-thalassemia subgroup, the same results were seen with the addition of significant associations in CPT Commissions and CPT Detectability (i.e., discriminability between letter targets and non-targets) at 12 (p<0.0001, p_FDR=0.0003; p=0.006, p_FDR=0.02) and 24 months (p<0.001, p_FDR=0.002; p=0.02, p_FDR=0.045) and loss of significance with CPT Hit Reaction Time (Supplemental Table 2). No other differences in cognitive measures were observed relative to adult care retention.

The Cox proportional regression analysis, after adjusting for EHI, demonstrated for every unit T-score increase in CPT Commissions, the probability of not remaining in adult care increased by 15% (hazard ratio=1.15, 95% CI: 1.02–1.31, p=0.025, p_FDR=0.075) at 12 months and by 10% (hazard ratio=1.10, 95% CI: 1.0–1.22, p=0.058, p_FDR=0.17) at 24 months. Neither exposure to disease-modifying therapies including chronic transfusion or hydroxyurea, or disease severity were associated with long-term engagement in adult care.

Finally, having a working primary guardian was associated with shorter retention in adult care, 303 and 377 fewer days at 12 months (se=114, p=0.01) and 24 months (se=157, p=0.02), respectively. These associations remained after adjusting for EHI (estimate=−290, se=119, p=0.02 for 12 months, and estimate=−364, se=165, p=0.03 for 24 months, respectively).

Predictors of Adult Routine Visits

Attendance of adult ambulatory visits were evaluated for 67 (51 HbSS/HbSβ⁰-thalassemia) of the 80 participants whose data were available. Neurocognitive performance was not associated with utilization of ambulatory services. Having a primary working caregiver was associated with attending fewer outpatient visits (estimate=−0.39, se=0.15, p=0.014, p_FDR=0.06). In multivariate linear regression analysis, not having a primary working caregiver (estimate=−0.47, se=0.15, p= 0.0029) and treatment with hydroxyurea (estimate=0.44, se=0.15, p=0.006) were independently associated with increased attendance to ambulatory visits. Receiving chronic transfusions during pediatric care and disease-severity were not associated with adult ambulatory visit attendance.

Among the HbSS/HbSβ⁰-thalassemia subgroup, univariate linear regression analyses showed that not having a primary working caregiver (estimate=−0.51, se=0.16, p=0.003, p_FDR=0.01), higher estimated perceptual reasoning (WASI-2 Perceptual Reasoning subtest; estimate=0.02, se=0.007, p=0.04, p_FDR=0.053), higher FSIQ (WASI-2 FSIQ; estimate=0.2, se=0.007, p=0.02, p_FDR=0.053), and faster processing speed (Wechsler Coding subtest; estimate=0.08, se=0.03, p=0.01, p_FDR=0.03) were associated with better attendance to adult routine visits. Exposure to disease-modifying therapies including chronic transfusion or hydroxyurea were not associated with adult routine visits. After adjusting for primary working caregiver in multivariate analyses, perceptual reasoning, FSIQ, and processing speed remained significantly associated with better attendance to adult routine visits (p≤0.038) and estimated executive function (DKEFS Color Naming subtest; estimate=0.1, se=0.04, p=0.04) became significant.

DISCUSSION

Children with SCD survive into adulthood therefore securing care continuity is necessary to ensure adequate care delivery.¹ Disengagement in health care during transfer from pediatric to adult care and during the initial adult care years may lead to increased disease complications, placing patients at higher risk of morbidity and mortality. In our single-center analysis, lower verbal comprehension, and attention deficits were associated with longer delay in initiating adult care and decreased adult retention.

Cognitive dysfunction is prevalent among individuals with SCD, worsens with aging, and its sequalae include increased school retention and absenteeism, grade remediation, and adult unemployment.^34–37 Current guidelines from the American Society of Hematology recommend that neurocognitive screening be performed during childhood to identify possible supportive interventions.³⁸ Transition planning is an essential part of preparation for health care transition. Our work underlies the importance of including neurocognitive screening in transition planning and programming among youth with SCD, as cognitive deficits may impact early and long-term engagement in adult care.

Our previous work identified deficits in attention as predictors of not completing transfer to adult care.¹⁹ In this study, we extended our prior work and demonstrated that, among patients who successfully completed the transfer to adult care, global intelligence, specifically verbal comprehension, predicts earlier engagement in adult care and attention deficits decreased retention in adult care. Deficits in attention have been documented in individuals with SCD,³⁶ and it is possible that attention deficits may impede the ability to effectively monitor and keep upcoming appointments. Further, within our exploratory analysis, we found that better perceptual reasoning and faster processing speed was associated with increased patient attendance to adult outpatient visits, domains important to organizational skills. Therefore, tools to support schedule organization (e.g., mobile health applications that send appointment reminders) should be investigated.

Further, poorer cognitive performance may impair health literacy, the ability to understand disease management and the overall health care transition process. The Verbal Comprehension Index is comprised of vocabulary and verbal association questions that require abstract thinking and reasoning. As health literacy is vital to navigate the health care system,³⁹ and is known to be low among adolescents with SCD,⁴⁰ it is possible that interventions that target health literacy for adolescents that demonstrate low cognitive performance may improve transition outcomes for those individuals, however the relationship between cognition and health literacy it not yet established in SCD.

Socioeconomic and environmental factors may play a role in health care transition outcomes in SCD, but their relationship is complex, and causation is not yet defined. Individuals who lived in households with greater number of people (total and children) had increased latency to adult care. In crowded households, competing demands may limit access to transportation and other primary needs, that take precedent over attending medical appointments. It has been shown that greater parental-educational is associated with better cognitive performance in SCD.⁴¹ However, due to sample size limitations we were unable to perform a mutli-variable relationship between cognition, parental educational, and latency to adult care. We did find that having a working primary guardian was associated with lower retention in adult care. Although counterintuitive, it is plausible that a working parent is less likely to be available to assist with transportation to appointments or help with schedule organization. It is also plausible that patients’ employment obligations may be a barrier to compliance with adult visits. Although, we were not able to test this hypothesis in our work, further examination of the socio-environment and how it influences health care transition is warranted.

The mechanism of progressive cognitive dysfunction in children with SCD is unclear.³⁷ Recurrent micro-infarction and chronic hypoxic damage likely contribute to the development of cognitive deficits.⁴² Our group and others have shown that treatment with hydroxyurea and elevated fetal hemoglobin were linked to improved cognitive functioning.^23,43,44 Although we did not establish a relationship between favorable transition outcomes and disease-modifying therapy in our analysis, hydroxyurea may mitigate cognitive deficit in adolescents with SCD. Our study design was not appropriate to address this question; however, prioritization of disease-modifying therapies that could preserve cognitive function and their possible relationship to transition outcomes should be studied.

Interventions targeting attention, pharmacological and behavioral modification, and verbal comprehension, e.g. reading programs, should be considered during health care transition planning in SCD. Further, ensuring that patients with SCD and cognitive deficits have access to school accommodations (Section 504 and Individualized Educational Plans) may help mitigate academic losses. Finally, further investigation of disease-modifying therapies and central nervous system stimulants (e.g., methylphenidate) to potentially mitigate the negative effects of cognitive deficits, is warranted.

LIMITATIONS

Our work had limitations. First, this was as single-center study, thus replication of our findings in other SCD populations is important. Secondly, our sample size was small and thus limited the strength of association between variables investigated at subsequent time points. Thirdly, not all participants had brain imaging studies, therefore the number of participants with imaging was too small to perform a meaningful analysis. In our center, imaging is only performed if clinically indicated, therefore, correlation with cognitive functioning and brain imaging studies was not possible, as it would be limited to a very small non-representative high-risk group. Lastly, it may be possible that the relationship between poor transition outcome and attention deficits is confounded by other cognitive or environmental factors not examined in this study. Numerous multidimensional environmental factors, like family dynamics, beyond standard unidimensional of socio-economic status, such as parental education^41,45, may also impact behavior and cognitive performance, but were beyond the scope of this analysis.^45,46

Unfortunately, not all individuals who completed pediatric care received neurocognitive screening. Although not statistically significant, amongst the patients who did not undergo neurocognitive screening, there was a greater number of participants who experienced overt stroke as children. If they had received cognitive evaluations, given the higher prevalence of cognitive deficits among stroke patients, this may have potentially increased the power of our analysis. Noncompliance with pediatric visits often resulted in missing the opportunity for neurocognitive screening, especially if screening was not offered on the same day as a regular clinic visit, however no differences in demographics or transition outcomes were found in relation to completion of neurocognitive screening. Finally, no data were available concerning patients who were lost to follow up, and further attempts at identifying them in our health system will be undertaken in the future.

CONCLUSION

Our work identified an association between cognitive ability (IQ, verbal comprehension, and attention) and transition outcomes, namely delay in establishing initial adult care and retention in adult care. Our work highlights the potential utility of embedding neurocognitive screening in transition programs to inform the development of individualized transition plans for youth with SCD and continued investigation and development of disease-modifying therapies that preserve cognitive function.