Clustering and variable selection in the presence of mixed variable types and missing data

We consider the problem of model-based clustering in the presence of many correlated, mixed continuous, and discrete variables, some of which may have missing values. Discrete variables are treated with a latent continuous variable approach, and the Dirichlet process is used to construct a mixture model with an unknown number of components. Variable selection is also performed to identify the variables that are most influential for determining cluster membership. The work is motivated by the need to cluster patients thought to potentially have autism spectrum disorder on the basis of many cognitive and/or behavioral test scores. There are a modest number of patients (486) in the data set along with many (55) test score variables (many of which are discrete valued and/or missing). The goal of the work is to (1) cluster these patients into similar groups to help identify those with similar clinical presentation and (2) identify a sparse subset of tests that inform the clusters in order to eliminate unnecessary testing. The proposed approach compares very favorably with other methods via simulation of problems of this type. The results of the autism spectrum disorder analysis suggested 3 clusters to be most likely, while only 4 test scores had high (>0.5) posterior probability of being informative. This will result in much more efficient and informative testing. The need to cluster observations on the basis of many correlated, continuous/discrete variables with missing values is a common problem in the health sciences as well as in many other disciplines.

Evaluation of the ability of neurobiological, neurodevelopmental and socio-economic variables to predict cognitive outcome in premature infants

BACKGROUND

When developed in the 1990s, the Neurobiologic Risk Score (NBRS) and Neurodevelopmental Risk Exam (NRE) correlated well with developmental outcomes in premature infants. Given recent advances in neonatology, we assessed their present ability to predict cognitive outcome, alone and combined with socio-economic factors.

METHODS

One hundred and twenty-nine neonates <32 weeks gestational age were assessed at 6, 12 and/or 24 months corrected age with the Cognitive Adaptive Test/Clinical Linguistic and Auditory Milestone Scale (CAT/CLAMS). Indices of socio-economic status included maternal education and marital status.

RESULTS

At 24 months corrected age (n= 67), the NBRS (r=-0.5), maternal education (r= 0.46) and marital status (r= 0.37) correlated with the CAT/CLAMS. These correlations increased when NBRS and maternal education were combined (r= 0.63) and when specific NBRS components (intraventricular haemorrhage, periventricular leukomalacia, seizures) and maternal education were combined (r= 0.66).

CONCLUSIONS

In the contemporary neonatal intensive care unit, measures used to predict cognitive outcome should incorporate both neurobiological risk factors and socio-economic variables.

Relationship between omega3 long-chain polyunsaturated fatty acid status during early infancy and neurodevelopmental status at 1 year of age

OBJECTIVE

To determine the influence of alpha-linolenic acid (ALA; 18 : 3omega3) intake and, hence, the influence of plasma and/or erythrocyte phospholipid content of docosahexaenoic acid (DHA; 22 : 6omega3) during early infancy on neurodevelopmental outcome of term infants.

METHODS

The Bayley Scales of Infant Development (second edition), the Clinical Adaptive Test/Clinical Linguistic and Auditory Milestone Scale (CAT/CLAMS) and the Gross Motor Scale of the Revised Gesell Developmental Inventory were administered at a mean age of 12.26 +/- 0.94 months to 44 normal term infants enrolled in a study evaluating the effects of infant formulas differing only in ALA content (0.4, 1.0, 1.7 and 3.2% of total fatty acids).

RESULTS

As reported previously [Jensen et al., Lipids 13 (1996) 107; J. Pediatr. 131 (1997) 200], the group fed the formula with the lowest ALA content had the lowest mean plasma and erythrocyte phospholipid DHA contents at 4 months of age. This group also had the lowest mean score on every neurodevelopmental measure. The difference in mean gross motor developmental quotient of this group versus the group fed the formula with 1.0% ALA but not of the other groups was statistically significant (P < 0.05). Across the groups, motor indices correlated positively with each other and with the plasma phospholipid DHA content at 4 months of age (P=0.02-0.03). The CLAMS developmental quotient correlated with the erythrocyte phospholipid content of 20 : 5omega3 (P < 0.01) but not with DHA.

CONCLUSIONS

These statistically significant correlations suggest that the omega3 fatty acid status during early infancy may be important with respect to neurodevelopmental status at 1 year of age and highlight the need for further studies of this possibility.

Internal consistency, temporal stability, and reproducibility of individual index scores of the Test of Variables of Attention in children with attention-deficit/hyperactivity disorder

Psychometric properties of the Test of Variables of Attention (TOVA) were examined in a cohort of children (n=63) strictly diagnosed with attention-deficit/hyperactivity disorder (AD/HD). Internal consistency was assessed via correlational analyses to determine the degree of agreement among various test portions. The temporal stability of errors of omission, errors of commission, response time, and response time variability was evaluated using test-retest reliability. Reproducibility of individual scores for the same indices was assessed using the Bland-Altman procedure. Select TOVA index scores exhibited high internal consistency in this cohort. Although the temporal stability of group scores (test-retest reliability) was satisfactory, individual test scores were less reproducible. Temporal stability and individual test-retest score agreement were greater for response time and response time variability than for errors of omission and errors of commission.

A randomized, double-blind, placebo-controlled trial of docosahexaenoic acid supplementation in children with attention-deficit/hyperactivity disorder

OBJECTIVE

To determine whether docosahexaenoic acid (DHA) supplementation for 4 months decreases the symptoms of attention-deficit/hyperactivity disorder (ADHD).

STUDY DESIGN

Sixty-three 6- to-12-year-old children with ADHD, all receiving effective maintenance therapy with stimulant medication, were assigned randomly, in a double-blind fashion, to receive DHA supplementation (345 mg/d) or placebo for 4 months. Outcome variables included plasma phospholipid fatty acid patterns, scores on laboratory measures of inattention and impulsivity (Test of Variables of Attention, Children's Color Trails test) while not taking stimulant medication, and scores on parental behavioral rating scales (Child Behavior Checklist, Conners' Rating Scale). Differences between groups after 4 months of DHA supplementation or placebo administration were determined by analysis of variance, controlling for age, baseline value of each outcome variable, ethnicity, and ADHD subtype.

RESULTS

Plasma phospholipid DHA content of the DHA-supplemented group was 2.6-fold higher at the end of the study than that of the placebo group (4.85 +/- 1.35 vs 1.86 +/- 0.87 mol % of total fatty acids; P <.001). Despite this, there was no statistically significant improvement in any objective or subjective measure of ADHD symptoms.

CONCLUSION

A 4-month period of DHA supplementation (345 mg/d) does not decrease symptoms of ADHD.

Stimulant medications decrease energy expenditure and physical activity in children with attention-deficit/hyperactivity disorder

OBJECTIVE

To determine the effect of stimulant medications used to treat children with attention-deficit/hyperactivity disorder (AD/HD) on energy expenditure, fuel utilization, and physical activity.

STUDY DESIGN

Energy expenditure and physical activity were measured, respectively, by room respiration calorimetry and microwave motion detectors in 31 children with AD/HD (26 boys and 5 girls; ages 6 to 12 years) both while they were receiving their prescribed stimulant medication and after the medication had been discontinued for at least 24 hours. Fuel utilization was calculated from calorimetry data.

RESULTS

Total and awake energy expenditure including energy expended while doing schoolwork, riding a stationary bicycle, resting, and watching a movie were from 4% to 8% lower when the children were receiving their prescribed stimulant medication. Total and awake activity were also lower while they were receiving medication (16% to 22%) and accounted for the lower rates of energy expenditure. Sleeping metabolic rate, basal metabolic rate, and fuel utilization were unaffected by medication.

CONCLUSIONS

Stimulant medications decrease physical activity, and hence, decrease the activity component of total daily energy expenditure in children with AD/HD.

Biochemical features of a patient with Zellweger-like syndrome with normal PTS-1 and PTS-2 peroxisomal protein import systems: a new peroxisomal disease

The peroxisomal disorders represent a group of inherited metabolic disorders that derive from defects of peroxisomal biogenesis and/or from dysfunction of single or multiple peroxisomal enzymes. We described earlier an 8 1/2 year-old with a history of progressive developmental delay, micronodular cirrhosis, and elevated very long chain fatty acids in plasma and skin fibroblasts. These findings were felt to be compatible with both neonatal adrenoleukodystrophy (nALD) and Zellweger syndrome (ZS). This patient is now 21 years old and his clinical course, inconsistent with either nALD or ZS, led us to examine his peroxisomal status in light of a possible new peroxisomal disease. The normal levels of bile acid precursors found in this patient suggest that peroxisomal beta-oxidation is functional. The activities of dihydroxyacetone phosphate acyltransferase and oxidation of lignoceric acid and phytanic acid were 14, 17, and 15% of the control, respectively. This partial activity for oxidation and the normal levels of bile acid precursors suggests that this patient has peroxisomes containing beta-oxidation enzymes. Western blot analysis of subcellular organelles showed that beta-oxidation enzyme proteins are present at normal levels in catalase-negative peroxisomes of density equivalent to normal peroxisomes. The presence of acyl-CoA oxidase and 3-ketoacyl-CoA thiolase in catalase-negative peroxisomes suggests that both peroxisomal targeting signal-1 (PTS-1), and peroxisomal targeting signal-2 (PTS-2)-mediated protein transport processes into peroxisomes are normal in this patient. These findings of catalase-negative peroxisomes of normal density and normal PTS-1 and PTS-2 import machinery with partial peroxisomal functions clearly demonstrate that this patient differs from those with known disorders of peroxisomal biogenesis.