The Science Journal of the American Association for Respiratory Care

1995 OPEN FORUM Abstracts


Thomas D. East.Ph.D., Fidel Silva, MD, Jeff Anderson, MS Pulmonary Division, LDS Hospital, 8th Ave and C St, Salt Lake City, UT 84143

Introduction: The purpose of this research was to develop the automated testing tools and techniques to thoroughly evaluate the performance of the Siemens Servo 300 ventilator under a variety of different conditions. The specific aims were: 1. Construct an accurate single compartment adult, pediatric and neonatal test lung that was interfaced to a computer for data collection. 2. Develop a set of tests and failure conditions which would adequately challenge the ventilator 3. Test the ventilator under a comprehensive set of conditions.

Methods: The adult and pediatric test lung used a Michigan Instruments TTL test lung that was modified to measure volume with 1% accuracy using a linear variable displacement transducer (LVDT) to record bellows displacement. The pressure transducers at the mouth (proximal) and in the test lung (distal) were Sensym (± 1 psi) solid state pressure transducers. FiO2 was measured using a Ventronics Oxygen Fuel Cell Electrode. Data was collected using a National Instruments analog/digital converter in a Macintosh computer. The neonatal test lung was constructed using three fixed volume plexiglas cylinders that were packed with copper wool to provide isothermal behavior. The cylinders were designed to provide 0.47, 0.88 and 1.20 ml/cm H_2O compliance. Three different neonatal endotracheal tubes (2.0, 2.5 and 3.0 mm ID) were connected to the test cylinder. A spring loaded pneumatically driven syringe was used to trigger the neonatal test lungs. The cylinders were calibrated and volume accuracy wasə%.

There are three different patient sizes and 11 different mode and triggering combinations on the Servo 300 (33 different permutations). Each permutation was tested at three different levels of ventilatory support; low, medium and high (99 tests). 27 different conditions were developed that simulated normal conditions as well as leaks, electrical and pneumatic failures, and other common clinical conditions known to cause ventilator failure (i.e. bovie use). Each condition was tested with all 99 tests unless the ventilator was non-functional. 60 seconds of data was collected for each test, converted to STPD and stored in a database for analysis.

Results: 709 tests were made under 27 different conditions. 58 comparisons were made per test (41,122 test results). The ventilator performed very well, even with the toughest of challenges. Over all test conditions the average accuracy of all the different variables was 0.87% and the precision was 28%. This represents a global worst case performance as it includes many different failure modes in which the ventilator could not deliver the set values. With normal conditions the average accuracy was 1.19% and the precision was 14.6%. The control of the inspired tidal volume was exceptionally accurate with an accuracy of -0.37% over all conditions, patient sizes, ventilatory support conditions and modes. FiO2 control was accurate (0.06% FiO2) and precise (6 % FiO2). The pressure control level was 1.4±3 cm H_2O below the setting. Pressure support level was 0.4±4 cm H_2O below the setting. PEEP was less than setting (difference -0.56±1.6 cm H_2O); however, in adults it was larger (-2±2 cm H_2O). Mean airway pressure measured by the Servo 300 was low (-0.9±1.37 cm H_2O, -12±19%). Errors in pressure were smaller with decreasing patient size. Ventilatory support level had little impact on errors. Bovie use near the ventilator did interfere with monitoring and the digital interface.

Conclusions: The Servo 300 performed very well even when faced with common clinical situations that have been known to produced faults in mechanical ventilators. The small difference between the pressure settings and measured values are likely due to knob calibration errors and are not clinically significant. It is difficult to compare results to other ventilators as there are no published data for performance over a wide range of ventialtory support and failure conditions.

Acknowledgments: Siemens Life Support Systems, Solna, Sweden.


You are here: » Past OPEN FORUM Abstracts » 1995 Abstracts » SIEMENS SERVO 300 PERFORMANCE ANALYSIS