Pre-Analytical and Analytical Factors Affecting the Accuracy of Blood Gas Analysis—A Systematic Review with Structured Narrative Synthesis

Background. Blood gas analysis (BGA) is a decision-critical test with low biological variation, and it is increasingly performed outside the central laboratory. We synthesised the evidence on pre-analytical and analytical factors that affect its accuracy. Methods. Six databases were searched from inception to July 2026 for studies of a defined pre-analytical or analytical factor affecting blood gas parameters. The effect measure was the within-specimen paired change from immediate analysis. No stratum held more than two comparable studies, so the synthesis follows SWiM and reports no pooled magnitude. Changes were compared with published analytical specifications and with exploratory pragmatic limits. Results. Fourteen studies were included, four with extractable change data. Among pre-analytical factors, 60 min of undisturbed room-temperature storage in plastic syringes lowered pH in both contributing studies (−0.020, −0.021) and raised pCO2 in both (+2.10, +1.43 mmHg), without a consistent pO2 change. Lactate rose in all three studies with extractable lactate data (+0.55 to +1.37 mmol/L). Within patients, chilling raised pO2 by +18.0 and +4.5 mmHg, a four-fold difference confounded with temperature, baseline pO2 and study. Delay combined with mechanical stress raised pO2 by +33.7 mmHg, and air bubbles, plastic syringe walls and haemolysis also biased pO2, with haemolysis further biasing pH and electrolytes. Among analytical factors, within-run imprecision met desirable specifications, whereas the only point-of-care versus central-laboratory comparison (pO2 bias −10.1 mmHg) could not separate method bias from a median 56 min delay. Three of 14 threshold classifications changed with the limit applied, and only the 4 °C pO2 artefact exceeded every limit. Conclusions. In this limited evidence base, delay and temperature affected oxygenation and metabolic analytes in opposite directions. When pO2 is required, prompt analysis at room temperature with air expelled appears preferable. When lactate, glucose or acid–base status is the priority and delay is unavoidable, chilling may limit metabolic drift. The analytical interchangeability of point-of-care and laboratory analysers cannot be judged from the available data.