Assessing Medication Adherence in Autism Spectrum Disorders

Background: More than half of children with autism spectrum disorders (ASD) are prescribed psychotropic medication for treating problem behaviors. To date, only one study has assessed any treatment adherence in ASD (Moore 2011), even though adherence is an important predictor of treatment outcome. Adherence research in other chronic conditions suggests that medication adherence is poor among young children, particularly those with chronic or mental health conditions.

Objectives: This study will describe multiple measures of medication adherence and identify predictors of poor adherence to medications among Medicaid-eligible children with ASD.

Methods: Medicaid-eligible children who were identified with an ASD by the South Carolina Autism and Developmental Disabilities Monitoring Network (SCADDM) across 5 study years (2000, 2002, 2004, 2006, and 2008) were included. All confidentiality procedures were followed and appropriate regulatory approvals were granted. Data linkages were made using unique identifiers common to both datasets; protected health information was removed following this linkage, resulting in a completely de-identified database. 

Child characteristic variables included age, race, gender, intellectual disability, co-occurring conditions, ASD diagnostic history, and indicators of behavioral and emotional problems (e.g., the presence of associated features such as tantrums, self-injurious behavior, hyperactivity, etc and DSM-IV diagnostic criteria). Medicaid data included for each child the individual eligibility status, county of residence, amount paid per claim, dispense date, drug name and therapeutic class, dosage, quantity, days supply, and prescribing provider type. The primary outcome measure was adherence, measured by the validated (Karve 2008) Medication Possession Ratio (MPR) and defined as the number of days supply in the index period divided by the number of days in the study period. Additional outcome measures included the proportion of days covered (PDC), and the refill compliance rate (RCR).  To quantify the complexity of medication regimens and the impact on adherence, a modified version of the Medication Regimen Complexity Index (MRCI) (George 2004) was used. This index includes the total number of medications, dosing schedule, dosing form (e.g., pill, liquid, etc), and special instructions (e.g., “take on an empty stomach”). Categorical and continuous variable differences were assessed using chi-square or t tests respectively. 

Results: Medication adherence is a complex phenomenon that appears to be best represented via multiple methodologies. Patient-related factors (e.g., knowledge of the condition as evidenced by a documented formal diagnosis), condition-related factors (e.g., associated features, co-occurring mental health disorders), and medication regimen-related factors (e.g., medication regimen complexity index) are among the most robust predictors of adherence.

Conclusions: Results provide a more complete and precise estimate of medication adherence in ASD by combining population-based data and Medicaid. Appreciating factors associated with poor adherence could lead to targeted interventions aimed at improving medication-taking habits, and ultimately improving treatment outcomes.