736 Assessment of blinding of hand hygiene observers in randomised controlled trials of hand hygiene interventions

Sunday, March 21, 2010: 10:45 AM
Regency VI-VII (Hyatt Regency Atlanta)
Chris Fuller, RGN, MSc , University College London, London, England
Joanne Savage, BSc, MRes , University College London, London, England
Sarah Besser, MSc , Kings College London, London, England
Sheldon Stone, BSc, MD, FRCP , UCL Royal Free Campus, London, England
Ellen Fragaszy, MSc , University College London, London, England
Barry D. Cookson, FRCPath, FRCP , Health Protection Agency, London, England
John McAteer, MSc , University College London, London, England
Susan Michie, PhD , University College London, London, England
Julian Gardiner, MA, MSc, PhD , London School of Hygiene and Tropical Medicine, London, England
Background: Trials evaluating interventions to improve healthcare workers' hand hygiene compliance use directly observed compliance as a primary outcome measure. Observers should be blinded to the intervention,that is, have no knowledge of the allocation of wards or units to the intervention, and the effectiveness of blinding assessed, to prevent systematic bias with compliance over-estimated in intervention wards and underestimated in non-intervention wards. This issue has not been addressed in the literature.

Objective: The study aimed to design a method of assessing blinding, use it to determine the efectiveness of blinding in a current randomised controlled trial, and thus inform future hand hygiene trial methodology.

Methods: A method of assessing blinding was designed and used to assess blinding in a current randomised controlled trial, the Feedback Intervention Trial (NRR website N0256159318) . The study hand hygiene observer, trained another researcher, blinded to the intervention, in use of a robustly standardised, valid, reliable and sensitive measure of hand hygiene compliance, the Hand Hygiene Observation Tool (HHOT) (www.idrn.org/nosec.php) . 1030 simultaneous observations were carried out over 20 hours, on 7 intervention and 6 non-intervention wards in the trial. Between observer differences were compared for both types of ward using Cohen's kappa (individual hand hygiene behaviours) and the Mann-Whitney U (overall compliance).

Results: Raw agreement between observers for individual hand hygiene behaviours was excellent in both intervention wards (91.5%, kappa = 0.886[95% CI 0.792, 0.98) and non-intervention wards (92.4%, kappa = 0.894 [95% CI 0.845,0.987). There was no statistically significant difference between observers for overall compliance.

Conclusions: This study describes a robust and pragmatic method for assessing the adequacy of blinding in hand hygiene intervention trials. It demonstrates that blinding of the study hand hygiene observer to ward allocation was effective in a current trial, and that assessments of hand hygiene compliance were not biased. All hand hygiene studies should blind observers to the intervention and assess the effectiveness of blinding.