Identifying heterogeneity in the Hawthorne effect on hand hygiene observation: a cohort study of overtly and covertly observed results

BACKGROUND: Observation and feedback are core strategies of hand hygiene (HH) improvement. Direct overt observation is currently the gold standard method. Observation bias, also known as the Hawthorne effect, is a major disadvantage of this method. Our aim was to examine the variation of the Hawthorne effect on HH observation in different healthcare groups and settings. METHODS: A prospective cohort study was performed in a tertiary teaching hospital during a 15-month period. Up to 38 overt observers (82% nurses) and 93 covert observers (81% medical students) participated in HH observation. The HH events observed overtly were matched for occupation, department, observation time, and location with those observed covertly. The data of matched pairs were then analysed to detect possible Hawthorne effects on different variables. RESULTS: A total of 31,522 HH opportunities were observed (4581 overtly, 26,941 covertly). There were 3047 matched pairs after 1:1 matching of overt and covert observations. The overall HH compliance was higher with overt observation than with covert observation (78% vs. 55%, p < 0.001). The Hawthorne effect was nearly three times larger in nurses (30 percentage points) than in physicians (11 percentage points) and was significantly greater in outpatient clinics (41 percentage points) than in intensive care units (11 percentage points). The magnitude of the Hawthorne effect varied among healthcare worker occupations and observation locations (p values both < 0.001) but not among departments, observation times, or HH indications. CONCLUSIONS: Heterogeneity in the Hawthorne effect may influence the interpretation of overt observations and prevent the correct identification of target populations with poor HH compliance. Therefore, directly observed HH compliance may not be an adequate performance indicator for infection control.