Low partial pressure of oxygen causes significant and unrecognized under-recovery of glucose on blood gas analyzers.

Background: Blood gas analyzers employing glucose-oxidase biosensors under-recover glucose when pO is low. The manufacturer of the GEM®Premier™ series of analyzers introduced an algorithm to detect specimens at risk of low pO interference. We investigated the reliability of this algorithm.

Methods: Whole blood specimens were tested by GEM®Premier™ 4000 (GEM 4000) and 5000 (GEM 5000). Specimens with an incalculable ("incalc") error code for glucose result or that had a glucose ≥ 20 mmol/L were retested on a second analyzer of the same type within 5 min over the course of 30 months in 5 hospitals in Calgary, Alberta. Discordant retests were defined as either: 1) paired numeric results with a difference >10 %, or 2) an "incalc" code that yielded a numeric result upon retesting. Glucose recovery in relation to pO level was assessed by comparing specimens experimentally depleted of pO between GEM 5000 and a laboratory analyzer (Siemens Vista®).

Results: Of 1,776 glucose tests repeated on the GEM 5000 or 1,544 on GEM 4000, 10% were discordant. GEM 5000 produced twice as many discordant numeric retests versus the GEM 4000 [5.9% (98/1,651) vs 2.7% (38/1,391)]. The majority of "incalc" error codes repeated with a numeric glucose result on both GEM analyzers [(79.7% (122/153) vs 75.2% (94/125)]. Among specimens experimentally depleted of pO the GEM 5000 under-recovered glucose by up to 30% compared to the Siemens Vista and were not flagged by an "incalc" code.

Conclusions: The algorithm in the GEM®Premier series of analyzers that flags specimens at risk for glucose under-recovery due to low pO does not reliably detect specimens at risk for glucose under-recovery.