The aim of evidence-based medicine is to integrate current best evidence from research into clinical policy and practice. However, this does not necessarily result in different treatment of individual patients. Difficulties in developing evidence-based clinical policies, organizational barriers and ineffectual educational programmes are identified as barriers to successful application of research evidence to health care [3
]. Existing hierarchical structures within and between the different professional groups obstruct routine decision-making processes and integration of evidence-based medicine [4
]. In intensive care, clinical practice is still influenced by a combination of theory, experience and evidence [5
Because of the various problems encountered in the management of a patient in the ICU setting, communication between member of the MDT is of particular importance. Good evidence has demonstrated the positive impact of a MDT approach on caring for ventilator-dependent patients [10
Ventilatory protocols that rely on multidisciplinary ICU expertise are becoming more frequently recommended in the ICU setting [6
]. Because nurses and RCPs also spend more time at the patient's bedside, their input regarding readiness for extubation is invaluable. Hence, we strongly believe that the integration of this evidence into the design of an extubation ICU protocol should take opinions and experiences of the MDT into consideration. Previous studies have shown that protocol-based approaches may shorten the duration of MV in comparison with weaning without a standardized approach [16
], but no approach has been established as superior over any other. Our intention was not to show that standardized extubation is a better approach to liberating patients from MV, but that it is possible to design and implement a MDT-driven extubation protocol that has a positive impact on the patient and the MDT, in a safe manner.
The unique feature of the present study is the description of the process undertaken to implement a MDT-driven extuba-tion protocol and what we felt was the best application of evidence-based medicine in our ICU.
Previous studies have identified both unfamiliarity of physicians with a RCP-driven protocol for ventilator weaning and the lack of consistent assignment of the RCPs to the same ward as two important barriers to successful implementation of a weaning protocol [8
]. During the design of our protocol, these barriers were taken into consideration in order to decrease the risk of protocol noncompliance. First, our ICU has its own dedicated RCP and nursing staff. Second, educational sessions were held, in which didactic teaching reinforced confidence of staff working with the protocol. Finally, the protocol was introduced to housestaff at the start of their ICU rotation.
The MDT agreed to ensure safety of the implemented protocol. In order to ensure safety and efficacy, we compared durations of MV and ICU stay, as well as rate of reintubation with those from previous literature and a cohort group; we found no significant differences between the protocol group and the control group in this regard. All of our findings are comparable to those from previously published studies (Table ).
Because a large number of patients from our protocol group were still in the process of weaning at the time that the first SBT was performed, and perhaps would not have been challenged had the protocol not been in place, we feel that in our ICU this protocol probably reduced ventilation time. However, this was not a primary outcome, given that it has already been proven. No serious adverse effects occurred during the SBTs. Finally, the MDT felt that the implementation of the protocol allowed patients to be extubated earlier for two reasons: most patients were still in the process of weaning, because historically we waited until they were on a PS of 5 cmH2O before extubating; and extubation no longer had to wait until the end of bedside morning rounds.
In order to draw a meaningful conclusion from this intervention, the following weaknesses need to be considered. First, the present study had a small number of subjects (n = 47), so any statistical result would be under-powered. Second, the use of historical control individuals limits our ability to draw strong conclusions. Despite these limitations, we felt it was necessary to ensure there was no obvious harm in implementing our protocol by comparing it with previous literature as well as a control group.