A new paradigm for human resuscitation research using intelligent devices.

OBJECTIVES: To develop new methods for studying correlations between the performance and outcome of resuscitation efforts in real-world clinical settings using data recorded by automatic devices, such as automatic external defibrillators (AEDs), and to explore effects of shock timing and chest compression depth in the field.
METHODS: In 695 records of AED use in the pre-hospital setting, continuous compression data were recorded using AEDs capable of measuring sternal motion during compressions, together with timing of delivered shocks and the electrocardiogram. In patients who received at least one shock, putative return of spontaneous circulation (P-ROSC) was defined as a regular, narrow complex electrical rhythm > 40 beats/min with no evidence of chest compressions at the end of the recorded data stream. Transient return of spontaneous circulation (t-ROSC) was defined as the presence of a post-shock organized rhythm > 40 beats/min within 60s, and sustained > or = 30 s. 2x2 contingency tables were constructed to examine the association between these outcomes and dichotomized time of shock delivery or chest compression depth, using the Mood median test for statistical significance.
RESULTS: The probability of P-ROSC for first shocks delivered < 50 s (the median time) after the start of resuscitation was 23%, versus 11% for first shocks > 50 s (p=0.028, one tailed). Similarly, the probability of t-ROSC for shorter times to shock was 29%, compared to the 15% for delayed first shocks (p=0.016). For shocks occurring > 3 min after initiation of rescue attempts, the probability of t-ROSC with pre-shock average compression depth > 5 cm was more than double that with compression depth < 5 cm (17.7% vs. 8.3%, p=0.028). For shocks > 5 min, the effect of deeper compressions increased (23.4% versus 8.2%, p=0.008).
CONCLUSIONS: Much can be learned from analysis of performance data automatically recorded by modern resuscitation devices. Use of the Mood median test of association proved to be sensitive, valid, distribution independent, noise-resistant and also resistant to biases introduced by the inclusion of hopeless cases. Efforts to shorten the time to delivery of the first shock and to encourage deeper chest compressions after the first shock are likely to improve resuscitation success. Such refinements can be effective even after an unknown period of preceding downtime.