Marked variability in adherence to hand hygiene: a 5-unit observational study in Tuscany.

BACKGROUND: International authorities recommend that the hand hygiene of health care workers be improved to prevent health care-associated infection. In 2005, Tuscany, a region in central Italy, initiated a campaign to improve hand hygiene that focused on raising awareness and educating health care workers. We assessed hand hygiene rates approximately 3 years after the campaign was initiated in 5 units of 2 hospitals in Florence, Italy, the capital of Tuscany. We also were curious whether variability would exist in the hand hygiene rates despite the close proximity of the units.

METHODS: We conducted a 3-month observational study in 2008 to assess hand hygiene adherence of doctors and nurses. Four of the units (ophthalmology, cardiology, geriatrics, and infectious diseases) were within one hospital, and the fifth unit (an emergency department) was in another hospital located less than 1 km away. External observers were used to assess the hand hygiene adherence of doctors and nurses before patient contact.

RESULTS: A total of 665 doctor-patient observations and 1147 nurse-patient observations were made. Doctors used some type of hand hygiene before touching the patient in 28% of their patient interactions (soap and water in 16% and alcohol-based handrub in 12%). Nurses used some type of hand hygiene in 34% of their interactions (soap and water in 27% and alcohol-based handrub in 7%). Hand hygiene adherence varied substantially across the units, from a low of 6% to a high of 66% for doctors and from 19% to 56% for nurses. The correlation between nurse adherence and doctor adherence was 0.90.

CONCLUSION: The overall rates of hand hygiene adherence observed were similar to those found when Tuscany initiated a hand hygiene campaign 3 years earlier. Focusing on overall rates may be misleading, however, because substantial variability existed between units. Furthermore, these rates come only from the "first moment" (before touching the patient) and can only be compared with rates from studies using the same approach.