Meta-Analysis of Adherence Promotion Interventions in Pediatric Asthma

Background Youth with asthma commonly have suboptimal adherence to inhaled corticosteroids (ICS). It is critical to systematically evaluate the effectiveness of ICS adherence promotion interventions and discern which techniques are most effective. Objective This study aims to (1) quantify the extent to which interventions improve ICS adherence in pediatric asthma, (2) explore differences in effect size estimates based on intervention and study characteristics, and (3) characterize the risk of bias across interventions. Methods We conducted literature searches across five databases. Included studies quantitatively measured ICS adherence as an intervention outcome among youth (<18 years old) diagnosed with asthma and were published after 1997. We analyzed aggregate effect sizes and moderator variables using random-effects models and characterized risk of bias using the Cochrane Collaboration tool. Results Thirty-three unique studies met inclusion criteria. At post-intervention, the aggregate effect size for pediatric ICS adherence promotion interventions was small but significant (n = 33, g = 0.39, 95% confidence interval [CI] = 0.24-0.54); however, the aggregate effect size at follow-up was not statistically significant (n = 6, g = 0.38, 95% CI = -0.08 to 0.83). Method of adherence measurement and intervention format were significant moderators. Most interventions had a high risk of performance bias and an unclear risk of bias in one or more domains. Conclusions ICS adherence promotion interventions are effective among youth with asthma. Additional longitudinal research is needed to quantify a more precise measure of intervention effectiveness over time, and moderators of intervention effectiveness should be reassessed as the literature base expands.

Automated summary

This study aims to quantify the effectiveness of interventions to improve ICS adherence in pediatric asthma, explore differences in effect size estimates based on intervention and study characteristics, and characterize bias risk across interventions. Results showed that most interventions were effective, with some significant moderators, but had a high risk of bias. Additional longitudinal research is needed to assess intervention effectiveness over time.