shows that individuals can be trained to recognise
stress as an error inducer—for example, by crew
resource management training21—and continue to
improve with recurrent training.29
Measuring teamwork attitudes and behaviour
Ratings of teamwork and communication differed sub-
stantially among groups of respondents. The percep-
tion of poor teamwork by one team member, whether
actual or perceived, is enough to change the dynamics
within that team, causing that team member to
withdraw. Preliminary data from behavioural observa-
tions of teamwork by trained observers of operating
theatre teams suggest that these attitudes are represen-
tative of behaviour (particularly with respect to
teamwork between surgical and anaesthetic staff; fig
2).12 27 30 Future research should investigate teamwork
in medicine, and its relationship to error rates and
error severity. In addition to being an error-
management technique, effective teamwork and
communication also has several positive side effects,
such as fewer and shorter delays, and increases in
morale, job satisfaction, and efficiency. Behaviour was
observed in 3204 commercial flights, from before
departure to landing, and in 96 randomly selected sur-
gical procedures from patient arrival to transfer to the
recovery room. Examples of a poor rating would be
failed communication of skin incision or removal of
the aortic-cross clamp or implementation of Trendle-
burg position without notifying the surgeon.
There is a relationship between perceptions of
teamwork and status in the team. Surgeons are most
supportive of steep hierarchies in which junior staff do
not question senior staff. Surgeons also perceive team-
work and communication in the team to be of a higher
quality than the rest of the team. Similarly, in intensive
care, doctors rated teamwork with nurses higher than
did nurses with doctors. However, future research
should not focus exclusively on consultants. Indeed,
our experience in aviation tells us that poor communi-
cation does not equate to an obstinate captain but to
poor threat and error management at the team level.
Highly effective cockpit crews use one third of their
communications to discuss threats and errors in their
environment, regardless of their workload, whereas
poor performing teams spend about 5% of their time
doing the same.31
Limitations
The most important limitation of our study was the
small sample of hospitals, and these data should there-
fore be considered preliminary. As more data are
collected, the issues of hospital to hospital variation
and non-response biases can be addressed empirically.
Our research in aviation found no significant
differences between cockpit crew responders and non-
responders on demographic variables such as sex,
years experience, background (military or civilian), and
position (captain, first officer, second officer).
Survey data are limited by reliance on self
reporting, are potentially biased by non-responders
(little is known about non-response biases in
healthcare surveys such as these), and are not the
panacea for what ails a safety critical system. However,
they can be used to diagnose the strengths and
weaknesses of an organisation, to create data driven
training interventions, and to assess the effect of train-
ing. Survey data also help to tailor training interven-
tions to address local issues.
Changing the professional culture in aviation
After the introduction of jet transport in the 1950s,
accident rates due to mechanical failure dropped
steeply. As data on accidents accumulated, it became
obvious that most accidents were related to break-
downs in crew coordination, communication, and deci-
sion making. The resulting shift toward a more open
culture that accommodated questioning and recog-
nised human limitations was a gradual but steady
progression.
The change came about through the involvement
of the research community, National Aeronautics and
Space Administration, regulatory agencies, and the use
of data driven initiatives to raise awareness of the
limitations of human performance and the importance
of effective teamwork. For the first time there were
instruments to collect reliable human factors data—a
combination of individual attitudes, organisational
norms, and assessments of behaviour before and after
training interventions determined if change was
actually taking place without having to rely on
retrospective data from accident investigation. Data
collection instruments such as the cockpit manage-
ment attitudes questionnaire were used to show
changes in safety related attitudes before and after
training, and these changes mapped on to actual
behaviour in the cockpit.19
Selection and training processes were amended.
Pilots began to be selected not only for technical skills
but also their ability to coordinate activities, learn from
error, and recognise that others can contribute to
problem solving. Airlines initiated a new approach to
training and assessing pilot skills by moving away from
training the individual pilot to training the entire
crew—recognising that safety and good performance
was not just a function of the captain but of the
captain using all available resources. The aviation
approach is to deal with errors non-punitively and
proactively, and this approach defines behavioural
strategies taught in crew resource management train-
ing (currently in its fifth generation)32
as error
countermeasures that are used to avoid error
whenever possible, to trap errors when they do occur,