A randomized trial found lower illness in the intervention arm, but 20% of participants in the intervention arm did not use assigned sanitizer while 5% in control obtained sanitizer independently. Which analysis approach best preserves randomization and estimates effectiveness?