ACCURACY, ALARM LIMITS AND RISE TIMES OF 12 OXYGEN ANALYZERS

The Comite Europeen de Normalisation recently proposed a new standard for 'the particular requirements of oxygen monitors for medical use'. The feasibility of this proposed standard was tested in respect of (I) accuracy of alarm activation (2) accuracy of oxygen display value dur ing both continuous and cyclical gas flows (3) rise time during rapid changes in oxygen concentration in the following 12 analysers: Datex C apnomac II and Servomex 570A (paramagnetic); Bruel & Kjaer 1304 (magne toacoustic); Criticare Poet II, Multinex, Drager Oxydig, Drager PM 803 0, Megamed 046A (part of the Megamed 700 ventilator), Ohmeda 5120, Spa celabs Multigas, Teledyne TED 200 (galvanic); Kontron OM 810 (polarogr aphic). All the analysers tested displayed an oxygen reading, which wa s within +/-3 vol% of the actual oxygen concentrations of the test gas es (15, 21, 40, 60 and 100 vol%). A cyclical pressure of between - 1.5 to + 8kPa did not affect the measured oxygen concentration as display ed by the Bruel & Kjaer 1304, Datex Capnomac II and Servomex 570A anal ysers. The remainder, however, showed, depending on their measuring pr inciple, a display error of between -1 and +6 vol%. After exposure to high pressure all the oximeters functioned normally. Some of the teste d devices showed more than 2% of deviation between their alarm activat ion and the preset alarm limits. Only the Kontron OM 810, the Megamed 046A and the Spacelabs Multigas monitors satisfied the requirements at all the rested oxygen concentrations. The time required by the oxygen analyser to display the rise from 29 to 92 vol% after a sudden change of concentration from 21 to 100 uol% O-2 is defined as 'rise time' an d must not, according to the Comite Europeen de Normalisation standard proposal, exceed the manufacturers' specification by more than a fact or of 1.15. The Bruel & Kjaer and Poet II monitors did not comply with this requirement, although their rise times were among the shortest. We conclude that the recommended standards concerning occuracy are gen erally met during continuous gas flow. Some of the monitors tested fai led to meet the required standard during cyclical flows and in the acc uracy of their alarm response. These small deviations are,however, not likely to be clinically relevant.