Clinical utility of ADHD symptom thresholds to assess normalization of executive function with lisdexamfetamine dimesylate treatment in adults.

OBJECTIVES

This analysis assessed the relationship of various cutoff scores of the ADHD Rating Scale IV (ADHD-RS-IV) to levels of improvement in ADHD-related executive function (EF), measured by the Brown ADD Scale for Adults (BADDS), which may provide a measure of clinically meaningful EF improvement after ADHD treatment.

METHODS

Post hoc analysis of a 4-week, open-label, dose-optimization phase in a double-blind, placebo-controlled study of lisdexamfetamine dimesylate (LDX) in adults with ADHD. The BADDS for Adults, a validated, normed, self-report measure of EF in ADHD, provides a qualitative measure to rate treatment progress. The ADHD-RS-IV assesses current symptom status based on DSM-IV criteria. Postbaseline ADHD-RS-IV scores were categorized according to four cutoff criteria of symptom remission: (1) ADHD-RS-IV total score ≤ 18; (2) ADHD-RS-IV total score ≤ 10; (3) no ADHD-RS-IV item scored >1; and (4) ADHD-RS-IV total score ≤ 18 and ≤ 2 items per subscale with a score of 2. Sensitivity and specificity of criteria for identifying participants with optimal BADDS scores were assessed using receiver operating characteristics (ROC). Safety evaluation included treatment-emergent adverse events (TEAEs).

RESULTS

At endpoint, 85/127 participants had optimal BADDS scores. Linear ANOVA indicated limited overlap between BADDS and ADHD-RS-IV scores (r (2) = 0.20; P < 0.0001). Specificity was similar for criteria 1-4 (0.46, 0.39, 0.39, and 0.42), as were ROC (0.699, 0.776, 0.732, and 0.668). Sensitivity was high for criteria 2 and 3 (0.96, 0.92), lower for criteria 1 and 4 (0.72, 0.75). TEAEs were consistent with those of stimulants.

CONCLUSION

Criteria 2 and 3 had satisfactorily high sensitivity, but no criteria had adequate specificity. AUC comparison indicated that criteria 2 and 3 ADHD-RS-IV thresholds may be more accurate assessments of EF normalization as measured by the BADDS. The open-label design, small sample size, and selection criteria limit the applicability of these results to a larger treatment population.