 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
a University of Illinois at Chicago, Chicago, Illinois
b Northwestern University/Children's Memorial Hospital, Chicago, Illinois
c Clinical Trials Surveys Corp, Baltimore, Maryland
d Baylor College of Medicine, Houston, Texas
e Columbia College of Physicians and Surgeons, New York, New York
f State University of New York Health Science Center at Brooklyn, Brooklyn, New York
g Boston Medical Center, Boston, Massachusetts
h University of Puerto Rico, San Juan, Puerto Rico
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
METHODS. Serial cognitive assessments were performed for 117 children who were infected vertically and 422 children who were exposed to but not infected with HIV, in a multicenter, natural history, longitudinal study. Repeated-measures analyses were used to evaluate the neurocognitive development of children between the ages of 3 and 7 years, as measured by the McCarthy Scales of Children’s Abilities (MSCA).
RESULTS. Children with HIV infection and class C status scored significantly lower in all domains of cognitive development, across all time points, than did those who were HIV infected without an AIDS-defining illness and those who were HIV exposed but not infected. There were no significant differences between the 2 latter groups in General Cognitive Index or specific domain scores. Rates of change in cognitive development were comparable (parallel) among all 3 groups over a period of 4 years. Factors that were associated consistently and significantly with lower mean scores were HIV status, number of times an examination had been completed previously, primary language, maternal education, and gender. No factors were related to rate of change of any mean domain score.
CONCLUSIONS. An early AIDS-defining illness increased the risk of chronic static encephalopathy during the preschool and early school age years. Children with HIV infection but no class C event performed as well as noninfected children in measures of general cognitive ability. No significantly different profiles of strengths and weaknesses for verbal, perceptual-performance, quantitative, or memory functioning were observed among children with or without HIV infection. A number of factors were found to have significant effects on the mean scores of children in all 3 groups; however, they were not related to the rate at which learning occurred.
Key Words: pediatric HIV • cognition • longitudinal
Abbreviations: WITS—Women and Infants Transmission Study • MSCA—McCarthy Scales of Children's Abilities • GCI—General Cognitive Index • HAART—highly active antiretroviral therapy
The effects of HIV infection on children's physical growth, psychological health, and neurodevelopment can range from mild to devastating.1 Delays in the mental and motor development of very young children with vertically transmitted HIV infection have been well documented.2–5 Studies found that these delays may begin as early as 4 months of age, continue into the preschool years, and manifest as either global or selective delays in neurodevelopment.
Advances in the medical treatment of children with HIV, including the development and use of combination therapies and supportive medications, not only have prolonged survival but also have promoted normal growth and development and have improved quality of life. Despite these advances, many children experience direct and indirect effects of HIV infection that affect their experience as effective learners. The direct effects may damage the functioning of the central nervous system and result in varying degrees of developmental dysfunction. Exposure to HIV among children often occurs in the context of environmental factors that pose equal or greater risks to a child's development, compared with HIV,4 including unstable or multiple caregivers, maternal drug and/or alcohol use, poverty, and low maternal education.
Few studies of HIV-related central nervous system effects focused on children with vertically acquired infection beyond the preschool years; the studies that did reported inconsistent and difficult-to-interpret results, because of small sample sizes and lack of appropriate control groups. Also, many studies were performed in the context of clinical drug trials rather than natural history studies, which may not represent a realistic view of the natural disease processes and their treatments.
In the preschool age range, findings of high rates of cognitive functioning in the "mentally retarded" range were not uncommon in early studies.6–9 However, more-recent studies of children in the preschool age range described much lower prevalence of severe impairment,10 as well as no differences between infected and noninfected children with respect to cognitive scores.11 Studies of older children vary significantly with respect to methods, which makes comparison of findings difficult. Overall, cognitive outcomes were found to be in the normal range for noninfected and infected children in the school age range, with only subtle isolated deficits among children with HIV infection.12,13
The purpose of this study was to investigate the effects of HIV infection on the cognitive development of preschool- and school-aged children over time and to determine whether these effects are related to child health status. In addition, this study sought to clarify whether there are areas of cognitive functioning (ie, verbal, perceptual-performance, quantitative skills, and/or memory) that are affected differently by the disease process. Finally, because of the concomitant environmental factors that are often associated with HIV infection and with poor developmental outcomes, the study sought to identify what effect, if any, these factors have on the cognitive functioning of this population.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
HIV-infected and noninfected, first-born, singleton children who had been born to HIV-positive women enrolled in WITS between 1990 and September 2000 and who had completed 
1 McCarthy Scales of Children's Abilities (MSCA) were included in the analysis. Of the 569 children who met all other criteria, 30 did not have 1 completed MSCA score in the database. Of the 539 children who were eligible to participate in this study, 117 were HIV positive and 422 were HIV negative.
A comparison of demographic characteristics was completed to determine differences between otherwise eligible children with and without MSCA scores. The children without MSCA scores were more likely to be born earlier than those with MSCA scores (1993.20 vs 1994.75; P = .010). There were no significant differences between the 2 groups with regard to HIV status, gender, ethnicity, birth weight, prematurity, intrauterine hard or soft drug exposure, maternal education, maternal health status, primary language, and number of primary caregivers.
During the course of this study, a number of HIV-negative children were removed from the study randomly, because of overrepresentation. This had a significant impact on the number of HIV-exposed but noninfected children seen at the 7-year visit, compared with the number at the 3-year visit.
Procedures
The MSCA was administered to all children every 6 months, beginning when the child attained the age of 3 years. The measure was administered at each visit up to and including the child's evaluation at 7 years of age. Information on a number of potential intervening variables and covariates, including demographic factors (eg, primary caregiver, changes in living situations, primary language, maternal education, and hard drug use), medical information, child grade level, and use of special educational resources, was collected concomitantly during interviews with primary caregivers. The child also received physical and neurologic examinations and had blood drawn as part of the full study visit.
Instrument
The MSCA14 is a measure of intelligence for children 2.5 to 8.5 years of age. It contains 18 subtests of cognitive and motor ability. Weighted raw scores from the 18 subtests are grouped in various ways to yield scores that reflect ability in 5 domains, ie, verbal, perceptual-performance, quantitative, memory, and motor. The weighted raw scores from the first 3 of those domains are converted to an age-related scaled score, or index, and are then combined to yield a General Cognitive Index (GCI). The reliability and validity of the MSCA are considered adequate for children between the ages of 3 and 7 years.15
The mean for the GCI has been set at 100, with a SD of 16. Index scores for the 5 domain scales have a mean of 50, with a SD of 10. Items from the first 3 domains (verbal, perceptual-performance, and quantitative) are content oriented, with no subtests from one domain contributing to the score of another domain. The memory and motor domains are process oriented, with all subtests in the memory domain overlapping with verbal, perceptual-performance, or quantitative domains. The MSCA also has a motor subtest, but these data were not collected in the WITS study.
Definitions
A child's HIV status was defined as described previously.16 All children who were HIV positive received antiretroviral therapy according to accepted guidelines on the standard of care for pediatric HIV patients.
Children in the study who were HIV negative were assigned to the noninfected group, children who had HIV but did not experience a class C event over the course of the study were assigned to the HIV/no C group, and children who were HIV positive and experienced a class C event before the end of the study period (before age 7) were classified in the HIV/C group. Class C refers to the categorization by the Centers for Disease Control and Prevention (1994)17 of signs or symptoms of HIV infection that are considered severe and includes a number of manifestations, such as serious bacterial infections, encephalopathy, lymphomas, and Pneumocystis carinii pneumonia.
Hard drug exposure (ie, opiates, cocaine, and/or methadone) was defined as described previously,18 through urine toxicologic tests or self-report. Soft drug exposure was defined as positive self-report of alcohol, tobacco, or marijuana use during pregnancy. Data on timing and degree of exposure were collected at each trimester throughout the pregnancy but, for the purpose of this analysis, a self-report was considered to be positive if the woman answered "yes" to having used any of the listed hard or soft substances within 30 days before the study visit.
Prematurity was characterized by gestation of <37 weeks. Highly active antiretroviral therapy (HAART) represents the combined use of reverse transcriptase inhibitors with protease inhibitors for treatment of HIV infection.
Data Analysis
Univariate summaries comparing characteristics of mothers and children across the 3 HIV cohorts were created by using the SAS UNIVARIATE and FREQ procedures (SAS version 8.2; SAS Institute, Cary, NC). Regression modeling of the longitudinal, neurodevelopmental, mean MSCA profiles assumed normally distributed errors and estimated parameters with the generalized estimating equations method.19 The SAS GENMOD procedure was used to perform these analyses.
Multivariate backward-elimination models were constructed that retained covariates by using an 
= .10 significance-level inclusion criterion. Mean profiles were approximately linear and were assumed to be linear in the regression models. Meaningful profile summaries were intercepts and slopes that measured the overall level and rate of change in mental functioning, respectively. Models were constructed both for individual MSCA scales and for all MSCA scales simultaneously. The model analyzing the mean scales simultaneously allowed formal comparisons of differences in HIV cohort performances in different domain scaled scores (verbal, perceptual-performance, quantitative skills, and memory). Differences in rates of change and overall mean levels among the cohorts were compared across different scales. This comparison allowed us to investigate whether some mental processes were more vulnerable to the effects of HIV than others. Significance was assessed by using a conservative significance level of = .01 to control for multiple comparisons.
The longitudinal models used in this study adjusted for a number of variables that were potentially affected by either the outcome variables or the relationship between the outcome variable and an independent variable. The variables that were included in the models were practice effects associated with repeated test administration (number of previous tests taken), social demographic variables (age at study visit, gender, maternal education, race/ethnicity, total household income, primary language, number of changes in caregiver, and number of changes in residences), and maternal factors during pregnancy and delivery (maternal exposure to hard and soft drugs during pregnancy, prematurity of birth, and child's birth weight). Other factors that were adjusted for included calendar year of enrollment in the study (to control for cohort effects), maternal disease stage (class B events), and child's disease stage (class A or B events). Although regression analyses assumed a linear model in age at visit, graphical summaries of the profiles are displayed that allowed the adjusted means to vary with the factor levels of age at visit.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
No differences were found among the 3 groups with respect to maternal factors of age, education, income, and health. Similarly, no differences were found among the 3 groups in child factors of in utero hard and soft drug exposure, gestational age, ethnicity, site of enrollment, premature birth, primary language, and gender.
The children in the HIV/C group were more likely to have an earlier year of birth and a greater number of changes in caregivers. Children without HIV were less likely to experience exposure to hard drugs during pregnancy than were children in the HIV/no C group. Significant differences were also found among the 3 groups in the kinds of antiretroviral treatment to which children were exposed in utero (P = .0001).
All children in the HIV/C group experienced a class C event before or during the study period, with a mean age at class C diagnosis of 1.92 years. Forty percent of these children had experienced a class C event by 12 months of age, 64% by 24 months, 79% by 3 years, 88% by 4 years, and 100% by 6 years. Within the HIV/no C group, 
18% of the children had class N (no symptoms) as their highest classification, 29% had a classification of class A (mild symptoms), and 54% had class B (moderate symptoms) as their most severe classification.
Differences in Group Means and Rates of Learning Over Time
The analyses of data for GCI and each test domain revealed strikingly similar patterns (Figs 1–5). In all instances, the rates of change with age were homogeneous across HIV status groups (P = .95). The tests for homogeneous intercepts across HIV status groups were significant for the GCI (P 
.0001) and for all index scales (P .0002), and in all instances the HIV/C group had a significantly lower mean profile than the other 2 groups. Results indicated that, for the GCI, the HIV/C group scored 15.9 points lower than the noninfected group and 14.9 points lower than the HIV/no C group. For all indices, the HIV/C group scored 5.6 to 7.7 points lower than the HIV/no C group and 5.6 to 8.2 points lower than the noninfected group. Comparisons involving the HIV/no C group and the noninfected group revealed no differences between those 2 groups. Comparisons of differences in means across HIV status groups at specific visits were not made because the graphical summaries suggested an approximately parallel pattern, with differences in profiles across HIV status groups appearing as shifts in the levels of the profiles. These differences seemed to be uniform across the visits, rather than accentuated at specific visits; therefore, the adjusted profiles were assumed to be approximately parallel in the analyses.
|
|
|
|
|
Different Functioning in Specific Areas of Cognition
A simultaneous analysis of the domains was performed to determine whether one or several of the domains were more vulnerable to HIV than others. The rate of change over time did not vary according to domain for any of the 3 groups (P = .47). However, when domain scores were compared, the difference between the noninfected group and the HIV/C group was slightly larger in the memory domain than in the quantitative domain (P = .0069).
Role of Environment in Cognition
Environmental variables that were associated consistently and significantly with all intercepts (GCI and domain mean scores) were HIV status, number of times an examination had been completed previously, primary language, maternal education, and gender. Children in the HIV/C group demonstrated consistently lower scores than did either of the other 2 groups. An increasing number of test administrations predicted increasingly higher scores. Children whose primary language was not English scored lower than primarily English-speaking children. Each year of maternal education over the 11th grade predicted higher mean scores. Lastly, girls scored higher than boys in all domains. Nonwhite ethnicity was a significant predictor of lower mean scores in areas of verbal skills, perceptual-performance, and quantitative abilities.
Additional analyses were completed to determine whether individual environmental variables affected the rate of learning within a particular domain over time. None of the environmental variables had a different effect on the rate of change for any mean domain score, regardless of HIV status.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Interestingly, although the children with an AIDS-defining illness were more cognitively delayed than others, the rates at which overall cognitive development occurred were not significantly different among the 3 groups of children. The pattern of cognitive development for children with an AIDS-defining illness suggests a static encephalopathy6 in which delays are global and significant, with a slower rate of development than expected for age, but are relatively static over time.
This study differs from much of the previous research related to the cognitive development of preschool- and school-aged children with HIV, in that it differentiates between children who are infected but have remained relatively healthy and those who have experienced a serious HIV-related illness. For example, early studies described high rates of moderate to severe cognitive impairment among preschool-aged and young school-aged children with HIV, with a minority of HIV-positive children scoring in the average range.9,20 Although the current study differs with respect to comparison groups, the rates of severe impairment were much lower than previously observed. It is possible that the higher rates of severe impairment found in earlier studies reflect the lack of treatment availability and thus higher rates of AIDS-defining illnesses.
The present study confirms and extends findings from previous research with both a younger cohort and an older cohort. Nozyce et al21 found that HIV-infected children <24 months of age performed as well as noninfected children in early measures of development unless the child had experienced an AIDS-defining illness. Significant differences were found in both mental and motor domains of the Bayley Scales for children with an AIDS-defining illness. Findings reported by Tardieu et al22 suggested that other health markers, such as circulating CD4+ lymphocyte counts during the first 1 year of life, were predictive of cognitive functioning among vertically infected school-aged children.
The second aim of this study was to investigate whether there are specific areas of cognitive functioning that are more vulnerable to the effects of HIV than others. Results proved to be consistent with findings for the general measure of intelligence on the MSCA (GCI). Children who had experienced an AIDS-defining illness early in life scored significantly lower on the verbal, perceptual-performance, quantitative abilities, and memory subtests than did either infected children with no AIDS-defining illness or noninfected children. There were no differences in functioning between the HIV-positive group with no class C event and the noninfected group. Also, similar to the GCI findings, children with an AIDS-defining illness scored 1 SD below the normative population mean in all domains, with the exception of the perceptual-performance mean, which was somewhat higher. Mean scores for the other 2 groups were all within the average range, as measured with the MSCA.
Although mean scores were consistently lower for the HIV/C group of children in all individual domains, the rates of learning were not significantly different among the 3 groups for any of the domains. Again, this profile suggests a static encephalopathy among the infected children who experienced an AIDS-defining illness early in life.
Previous studies9,11 examining specific areas of cognitive functioning found no differences in specific domain scores between HIV-infected children and noninfected but exposed children, although both groups scored in the low average (standard score: 80–89) or borderline (standard score: 70–79) ranges of intelligence. Again, small group sizes, particularly in control groups, made these findings difficult to interpret.
The third aim of this study was to determine the factors that, either alone or in combination with HIV, mediated an effect on cognitive outcomes. A number of covariates were associated consistently and strongly with cognitive outcomes for all groups. As expected, HIV status was a significant predictor of outcomes in all domains, including the GCI. Other environmental variables that were identified independently and consistently as important across all domains and GCI, with control for other variables, were primary language, maternal education, and gender. Ethnicity was also predictive of outcomes in the areas of verbal skills, perceptual-performance, and quantitative abilities.
When possible, the MSCA was administered by testers whose primary language was the same as the child's, but often it was necessary to use a translator during administration. It is accepted that, when a measure is translated and adapted for use in different cultures, reliability and validity may be weakened, resulting in different rates of outcomes.23 Because Hispanic children scored significantly lower in several of the domains in which ethnicity was also a significant predictor, it may be that the results were confounded by test language differences.
The findings regarding the influence of other demographic variables concur with a number of previous studies documenting the effects of low socioeconomic status and ethnic minority status on standardized psychological test scores24,25 and the MSCA in particular.26 Gender has also been documented to be a factor predicting lower scores on some standardized psychological tests, but this has not been found specifically with the MSCA.23 In addition, a study examining behavior problems in the same cohort found that the demographic variables (ie, ethnic minority status, gender, and maternal education) that were found to be important in the present study were more predictive of high rates of behavior problems than was HIV disease itself.27
The positive effect of repeated testing on mean scores was consistent across all domains, including the GCI. This learning effect increased as the number of examinations taken increased, with a ceiling effect by the sixth or seventh examination. Previous studies documented the variance in the magnitude of this effect as a function of type of measure, test-retest interval, age, and overall competency level of the participant.28 In a study involving repeated testing with the Wechsler Intelligence Scale for Children-Revised and the Wechsler Adult Intelligence Scale-Revised among children with hemophilia,29 practice effects were evident for some domains after 2 years of annual assessments, whereas other domain scores remained unaffected by repeated exposure for up to 4 years. Additional study is warranted to determine how this information contributes to the understanding of practice effects with this particular measure and, more generally, for young children who are participating in clinical trials and longitudinal investigations.
The results of this investigation suggest strongly that prevention of early AIDS-defining illness among young children with HIV infection is crucial for preventing poor neurodevelopmental outcomes. With early, appropriate, antiretroviral therapy, not only do the odds for survival increase but also children may avoid significant developmental/cognitive impairment.
It is not yet clear what effect this pattern of chronic static encephalopathy may have on academic achievement or quality of life for older children. Families and school systems should be familiar with this potential risk and should be prepared to offer children with HIV an individualized educational plan that includes collaboration with medical health professionals and provision of comprehensive services. Because the present study suggests that cognitive impairments for a subset of young children tend to be global in nature, rather than in a specific area, some children may need additional educational and psychosocial resources in the classroom and regular reevaluation of their progress and learning needs.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Much remains to be learned about the effect of HIV on the central nervous system past the preschool and early school age years. As children with HIV continue to survive and thrive into their middle school and adolescent years, it is critical for future research to determine whether children with early AIDS-defining illnesses continue this chronic profile of static encephalopathy.
There are many stressors that are experienced by children affected by and infected with HIV in the United States, including familial chronic illness, loss, family disruption, poverty, parental drug use, and social stigmatization. It is important for future studies to continue to investigate the long-term consequences of HIV, alone and in combination with these environmental stressors. With improved understanding of the risks inherent in HIV-related illness and effective medical, educational, and psychosocial interventions, children and their families may be assured improved health and quality of life in the future.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Address correspondence to Renee Smith, PhD, Department of Pediatrics/WITS, 840 S Wood St, M/C 856, Chicago, IL 60612. E-mail: resmith{at}uic.edu
The authors have indicated they have no financial relationships relevant to this article to disclose.
Dr Brouwers' current address is National Institutes of Health, National Institute of Maternal Health, Bethesda, MD 20614.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. C. Lindsey, K. M. Malee, P. Brouwers, M. D. Hughes, and for the PACTG 219C Study Team Neurodevelopmental Functioning in HIV-Infected Infants and Young Children Before and After the Introduction of Protease Inhibitor-Based Highly Active Antiretroviral Therapy Pediatrics, March 1, 2007; 119(3): e681 - e693. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2006 American Academy of Pediatrics. All rights reserved. |
||
|
||
