The risks, potential harm and costs of adverse events for patients at intensive care units (ICU) are larger than in other hospital departments , making improvement of patient safety in ICUs all the more important. The ICU is particularly vulnerable in terms of patient safety threats related to ineffective teamwork or failure to follow protocol [2, 3]. Results of a Dutch record review study revealed that 9.4% of all patients admitted to the ICU experienced one or more adverse events, which is far above the average of a hospital (5.7%). Of these adverse events 50% were considered highly preventable. Similarly, Vincent and colleagues  reported 1.7 adverse events per patient per day in a medical-surgical ICU.
It has been demonstrated that unsafe care more often originates from problems with non-technical skills than from a lack of technical expertise [6, 7]. Non-technical skills are 'the cognitive, social and personal resource skills that complement technical skills and contribute to safe and efficient task performance' . For example to be explicit in coordinating tasks or to share information. Studies have shown that a lack of non-technical skills led to poor teamwork resulting in critical incidents in the ICU [9, 10].
Both national and international health authorities have advocated crew resource management (CRM) as a method to improve non-technical skills, especially in emergency departments, surgery and intensive care [11–13]. CRM, which has its roots in high-risk industries such as aviation , was developed in the early 1980's as a response to the finding that unsafe flight conditions were frequently the result of failures in pilots' non-technical skills rather than a lack in technical knowledge . It has been shown that CRM can effectively improve safety in a variety of professional domains, such as nuclear power and offshore oil production [16, 17]. It is plausible that the general principles of CRM can be used in the ICU as well, because the ICU shares characteristics with workplaces where CRM has been proven to be effective, such as high-stake outcomes, complex actions, and high time pressure .
CRM is based on the premise that human error is avoidable, but can never be eradicated. It is typically directed at creating awareness of human factors and performance limiters , and at teaching behaviours to neutralize these threats, for instance through leadership and speaking up . Furthermore, it forces participants to assess and think about personal and peer behaviour. Concepts that are introduced during the training include inquiry, seeking relevant task related information, advocacy, communicating proposed actions, conflict resolution and decision making [19–21]. Thus, CRM is directed at increasing awareness of human limitations and changing team behaviour and communication in order to improve the management unsafe situations and, as a result, reduce adverse events.
Although several studies have been carried out to evaluate the effectiveness of CRM in health care [22, 23], none were conducted at an ICU. Rabøl and colleagues  conducted a systematic review about the reported effects of CRM training in health care. They found that the first reaction to the training was very positive and that attitudes changed in favour of the CRM principles. For instance, France and colleagues  reported that trainees indicated that CRM has the potential to increase patient safety and quality of care. However, going one step further, looking at behavioural change and the impact of the training on the organisational level (e.g. reduced number of bed days per patient), the results are less straight forward. For example, McCulloch and colleagues  found an increase of the use of non-technical skills for nurses, but not for anaesthetists and surgeons.
There are several reasons for these inconsistent findings, which stipulate the necessity for the present research. It can be argued that a lack of consistent findings on a behavioural and organizational level is due to weak study designs. Most of the studies evaluate the effect of CRM within six months , which is a relatively short period for an innovation to be completely adopted and to become part of the daily routine . In addition, most evaluations rely on a pre- and post training comparison, but do not include a control group . In some studies the trained and non-trained participants were not separated in data gathering. Furthermore, observations have hardly ever been used to measure behavioural change, despite the high validity of this method as it measures behaviours when they actually occur.
Therefore the aim of the study is to determine the effectiveness and cost-effectiveness of CRM on attitude, behaviour and organization one year after the training. In order to reach this goal, we will use a multi-method approach using questionnaires, direct observations, and interviews. Trained ICUs will be compared with matched control units using pre- and post measurements. The purpose of the present article is to describe the study protocol and the underlying choices of this evaluation study of CRM in the ICU in detail.