We conducted a study to test whether the ongoing use of a MET system could be associated with a sustained reduction in the incidence of cardiac arrests over the four years after its introduction. We found that education alone decreased the incidence of cardiac arrests. However, the risk of cardiac arrest further decreased after the introduction of the MET and remained lower over time.
We also found an inverse association between levels of MET use and the risk of occurrence of cardiac arrests, suggesting a 'dose effect' of the MET on cardiac arrests.
Implementation of MET systems has previously been shown to reduce the incidence of cardiac arrests in hospital patients in a number of short term before-and-after studies [5
]. DeVita and co-workers [8
], however, reported that increasing the use of an existing MET was associated with a 17% reduction in the incidence of cardiac arrests in the subsequent 1.8 years to a rate of 5.4 per 1,000 admissions. In the present study, the incidence of cardiac arrests was 4.06 per 1,000 admissions before the introduction of the MET. After the introduction of the MET, the incidence of cardiac arrests was 1.90 per 1,000 admissions (53% reduction) and this decreased to 1.3 per 1,000 admissions (69% reduction) in 2004. These later results suggest a greater reduction in the incidence of cardiac arrests than previously reported following the introduction of a MET service.
Sustained system change requires strong organizational commitment to safety [9
]. System change may fail to be institutionalized because of a turnover of key employees [10
], resulting in new doctors and nurses unfamiliar with the MET system concept [8
]. Prior to the implementation of the MET service at our hospital, a 12 month period of preparation and education was undertaken (educational phase). Since the introduction of the MET service we have provided ongoing education to all existing and new staff members. This approach appears to have been successful in producing a sustained and progressive reduction in the incidence of cardiac arrests at our hospital.
In addition to the reduction in incidence of cardiac arrests demonstrated after the introduction of the MET service, we have found an inverse association between levels of MET activation and the incidence of cardiac arrest. This suggests an association between increasing use of the MET and a reduction in cardiac arrests. From this association, we estimate that for every 17 MET calls a single cardiac arrest may be prevented.
Multivariate logistic regression analysis revealed male gender and a primary rhythm of asystole or pulseless electrical activity as predictors of increased risk of death following cardiac arrest. Predictors of improved outcome following cardiac arrest included admission under cardiothoracic surgery, the spinal cord injury unit, or cardiology. This improved outcome may be secondary to increased levels of monitoring leading to earlier detection of cardiac arrests in these patients, or may suggest greater reversibility of the underlying process leading to the arrest (for example, myocardial ischemia or altered autonomic tone with spinal cord injury). Improved outcome for cardiac arrests occurring between 08:00 and 10:00 may relate to increased levels of staffing and earlier detection of events.
Our study has several strengths and limitations. It is a prospective before-and-after study that demonstrates a progressive and dose-dependent reduction in the annual incidence of cardiac arrests in a large teaching hospital in a study period including over 145,000 admissions. It is not randomized, blinded or placebo-controlled, however, and only represents the findings of a single center. A large cluster-randomized multi-centre study of the effect of the MET on cardiac arrests in 23 Australian hospitals (the MERIT study) has been recently published [13
]. The incidence of cardiac arrests in the 12 hospitals randomized to receive MET implementation was not statistically different to that in the 11 hospitals randomized to continue the existing form of cardiac arrest teams. This study involved an education and preparation period of only four months, however, and assessed the impact of the introduction of MET reviews over six months only. In addition, the average call rate (cardiac arrests and MET calls) was reported as only 8.3 calls per 1,000 admissions. This call rate is one-fifth of our current call rate. These observations suggest that time may be required before a MET system can show its full effectiveness in reducing the incidence of cardiac arrests. The benefits from other system changes, such as trauma services [14
], have also been shown to take some time to mature. These results also suggest that the 'dose' of MET reviews (calls per 1,000 patients) may have an impact on patient outcome.
The second limitation of our study relates to the inclusion of episodes of insufficient data. In analyzing the effectiveness of the education process and MET service, however, we have assumed that these events represented a true cardiac arrest. In doing this, we would have actually underestimated the effectiveness of the MET system, as such calls likely inflated the true cardiac arrest call value after implementation of the MET.
Our study does not reveal the mechanisms responsible for the reduction of the cardiac arrests. It is possible that the observed reduction in the incidence of cardiac arrest may be due to the education of staff alone, as the incidence of cardiac arrests in the educational phase decreased by 40% compared with the period before the introduction of the MET. However, the incidence of cardiac arrests after the introduction of the MET was 23% lower than during the education and preparation period, and has continued to fall with increasing use of the MET. This suggests education and awareness together with a system to promptly review unwell ward patients work synergistically in reducing the incidence of cardiac arrests.
It is possible that our favorable findings were due to a high incidence of cardiac arrests in the control period or an abnormally low seasonal incidence in the intervention period. Australian data  show an incidence ranging from 3.6 to 5.1 per 1,000 admissions. Recent data from the MERIT study, which included several smaller hospitals with patients of limited acuity and counted all day visits (no overnight stay) as admissions, showed a pre-intervention cardiac arrest rate of 2.08 cases per 1,000 admissions [13
]. Our incidence of cardiac arrests was 3.2 per 1,000 overnight hospital admissions during the control period.
We studied the MET within a single institution. Its findings might not apply to other hospitals. Institution specific heuristics and unique administrative features may have lent themselves to making the impact of the MET approach greater in our institution than in others. However, our institution has all the organizational, structural and logistic features of a typical tertiary referral hospital. The way our MET was configured might differ from the way other institutions implement such a service [13
]. Whether organizing the MET service in different ways has an impact on its efficacy remains unknown. We believe that our approach is simple and low cost. It is also possible that the decrease in cardiac arrests was secondary to some other improvements in patient care during the period that separated the control from the intervention period. There were no changes in the structure, referral pattern or activity of our hospital, however, as supported by the total number of admissions during the two study periods, which remained unchanged. Furthermore, there were no changes in 'not for cardio-pulmonary resuscitation' policy, hospital admission policy, discharge practices or surgical case mix during the study. We are not aware of any improvements or advances in medical or surgical treatment that could explain a greater than 60% reduction in cardiac arrests and a 25% reduction in overall mortality.