The purpose of this study is to assess the effectiveness of a best practices nursing protocol in intensive care units on the occurrence of critical incidents compared to the common practices.

To develop the best practices nursing protocol in intensive care units, three steps were necessary:

1. An incidence study of occurrence of critical incidents linked to nursing was conducted during a month in nine intensive care units to identify high risk clinical situations
2. A qualitative study (individual interviews and focus groups) with nurses was led to elicit nurse's risk anticipatory strategies during nursing
3. Based on the results of the two previous step, a working group proposed a best nursing practices protocol to limit critical incidents

Methodology to assess the effectiveness of the protocol of best nursing practices:

The statistical unit is the nursing act for one patient meeting the inclusion criteria (a patient is nursed two times or more per day) Nine intensive care units are included in the study Intensive care units includes 127 beds, their average occupancy rate is approximately 80 % which means about 330 nursing a day and about 9900 nursing a month.

An occurrence of critical incidents (primary outcome measure) of 25% is expected (approximately 2475 critical incidents per month). To show a reduction of 50% of critical incidents in the interventional group, 6 month inclusion will be necessary, whatever the value of the inflation coefficient.

The patient will be under observation during 60 min if a critical incident occurs. And only during nursing if no critical incident occurs.

Data will be collected at bedside.

Patients will be included for the entire duration of the hospitalization in intensive care unit, if inclusion criteria are still met.

Data collection:

• Characteristics of the intensive care unit
• Characteristics of the patient
• Characteristics of the nurse
• Characteristics of the medical equipment of the patient
• Physiological parameters before, during and after nursing (see primary outcome measures)

The best nursing practices protocol will be consider as efficient if a reduction of 50% of critical incident can be established.

Statistical analysis will be in the intent-to-treat.

Hospitals as part of qualitative policy request a declaration of each critical incident. This circus of declaration will not be changed.

Inclusion Criteria:

• Patient in intensive care units
• Patient who has received complete information on the organization of the research and who has not objected to his participation and the exploitation of his data

Exclusion Criteria:

• Moribund patients
• Nursing made in the chair or in the shower stretcher
• Patient able to wash and dress alone (or just with the assistance of a nurse)
• Person under legal protection
• Person deprived of liberty by a judicial or administrative decision
• Person being the object of psychiatric care by virtue of articles L. 3212-1 and L. 3213-1 of the french Code of Public Health

Patients in intensive care units