Comparison between arterial and peripheral-venous blood gases analysis in patients with dyspnoea and/or suspected acute respiratory failure

Withdrawal of blood for arterial blood gases analysis (ABGA) has formed the mainstay of treatment in patients with dyspnoea and suspected acute respiratory failure for the past 50 years [ ]. The diagnosis of acute respiratory failure is based on the results of ABGA and it is essential in the treatment of, amongst many others, various respiratory, metabolic and renal diseases, and intoxications [ ]. In critically ill patients, access to arterial blood is required for regular ABGA and continuous arterial blood pressure measurements [ ]. However, only around 10–20% of hospitalized patients who require oxygen therapy are treated in intensive care or similar units, where arterial line placement is possible [ ]. In a majority of other patients, arterial blood is obtained via radial artery puncture [ ]. Although this procedure is rarely associated with severe complications, it causes significant discomfort [ ]. We hypothesized that for patients with dyspnoea and/or suspected acute respiratory failure, and without circulatory failure, clinically relevant results can be obtained from peripheral venous blood gases analysis (PVBGA). We assumed that pCO₂ is 7.5 mmHg (1 kPa) lower in ABGA compared to PVBGA, that there is no difference in pH and HCO₃ between ABGA and PVBGA and that there is no difference in oxygen saturation between ABGA (SaO₂) and peripheral oximetry (SpO₂). Institutional ethics committee approval was obtained (No 22/19) and we performed a prospective, observational study in a medical emergency department of a university hospital. Samples for PVBGA were obtained within 5 min of ABGA in patients where ABGA withdrawal was clinically indicated by the treating physician. We included 102 patients from March to May 2019 (56 males and 46 females; mean age 70.1 ± 15.6), 64.7% of patients were admitted to hospital, 44.2% received oxygen therapy and 39.0% bronchodilator therapy. Regarding concomitant illnesses, 25.0% of patients have been previously diagnosed with COPD. 37.3% of patients had clinical signs of congestive heart failure, and 23.6% of patients have been diagnosed previously as such. Mean blood pressure on admission was 101 ± 17.1 mmHg and mean pulse 84.9 ± 18.7 bpm. Median serum lactate value in ABGA was 1.5 ±. 0.8 mmol/L, and none of the patients required support with noradrenalin, dopamine or dobutamine. APACHE II score was 11.4 ± 5.2 points. We observed statistically significant differences between ABGA and PVBGA in pCO₂ (36.3 vs. 42.9 mmHg; p = 1.3 × 10⁻¹⁰), pH (7.43 vs.7.39; p = 2 × 10⁻¹⁰) and HCO₃ (23.8 vs. 25.7 mmol/L; p = 7.3 × 10⁻⁵). There were no differences between SaO₂ and SpO₂ (95% vs. 94%; p = 0.487). The parameters above were then assessed for correlations between ABGA and PVBGA and between SaO₂ and SpO₂ (Fig. 1). The results have shown strong positive statistically significant correlations for pH (ρ=0.590), HCO₃ (ρ=0.901), pCO₂ (ρ=0.740) and SpO₂ (ρ=0.645). We also observed that the statistically significant difference in pCO₂ between ABGA and PVBGA is no longer observable if 7.5 mmHg is subtracted from venous pCO₂ (36.3 mmHg and 35.3 mmHg; p = 0.26).