Arterial blood gas and pH analysis. Clinical approach and interpretation.

Arterial blood gas and pH analysis are performed during anesthesia or critical care medicine for (1) assessment of acid-base balance, (2) assessment of pulmonary oxygenation of arterial blood, and (3) assessment of alveolar ventilation by measurement of arterial blood PCO2. Total physiologic and alveolar dead spaces are evaluated by comparing the alveolar PCO2 with the mixed expired and mixed alveolar PCO2, respectively. This article provides a clinical approach and interpretation of arterial blood gas and pH analysis.

Importance of temperature and humidity in the measurement of pulmonary oxygen uptake per breath during anesthesia.

Traditionally, measurement of pulmonary O2 uptake uses mass balance of N2 to correct for differences between inspired and expired volume (V) due to temperature (T) and relative humidity (RH). Often during anesthesia, N2 balance cannot be invoked due to high inspired O2 fraction (FIO2) or nonsteady state conditions. Then, O2 uptake per breath (VO2,br) must use assumed or measured T and RH differences between inspirate and expirate. This numerical analysis study examines how errors in inspired RH and T can affect VO2,br. Equations were developed to simulate a baseline metabolic and ventilatory condition. A unit error in inspired RH of 0.5 (during constant inspired T of 22 degrees C) caused percent errors in VO2,br of 5.6% during FIO2 = 0.2% and 28.8% during FIO2 of unity. Per(-57.6 x FIO2-0.115) VO2,br was given by (change in RH) (R2 > 0.999). Errors in inspired T (during constant inspired RH of 0.5) had similar effects on percent error in VO2,br( =-8.75 x FIO2-0.093) x (change in T) (R2 = 0.999). Because inspired VO2 is larger at higher FIO2 and because VO2,br is the difference between inspired and expired VO2, VO2,br is most affected by the inspired V error at the largest FIO2 . When tissue O2 consumption decreases relative to minute ventilation, T and RH errors have a greater effect on VO2 br because the error in inspired V affects a smaller VO2,br. At lower barometric pressure, RH errors affect VO2,br more because water vapor V occupies a larger fraction of inspired V. In summary, because inspired RH and T can vary significantly during anesthesia, a fast-response humidity and T sensor, combined with flow and FO2 measurements, are needed to allow accurate determination of VO2,br x VO2,br should become an important measure of metabolism and patient wellness during anesthesia.

Carbon dioxide kinetics and capnography during critical care.

Greater understanding of the pathophysiology of carbon dioxide kinetics during steady and nonsteady state should improve, we believe, clinical care during intensive care treatment. Capnography and the measurement of end-tidal partial pressure of carbon dioxide (PETCO2) will gradually be augmented by relatively new measurement methodology, including the volume of carbon dioxide exhaled per breath (VCO2,br) and average alveolar expired PCO2. Future directions include the study of oxygen kinetics.

Non-steady state monitoring by respiratory gas exchange.

Traditionally, the study of CO2 and O2 kinetics in the body has been mostly confined to equilibrium conditions. However, the peri-anesthesia period and the critical care arena often involve conditions of non-steady state. The detection and explanation of CO2 kinetics during non-steady state pathophysiology have required the development of new methodologies, including the CO2 expirogram, average alveolar expired PCO2, and CO2 volume exhaled per breath. Several clinically relevant examples of non-steady state CO2 kinetics perturbations are examined, including abrupt decrease in cardiac output, application of positive end-expiratory pressure during mechanical ventilation, and occurrence of pulmonary embolism. The lesser known area of non-steady state O2 kinetics is introduced, including the measurement of pulmonary O2 uptake per breath. Future directions include the study of the respiratory quotient per breath, where the anaerobic threshold during anesthesia is identified by increasing respiratory quotient.

How does positive end-expiratory pressure decrease pulmonary CO2 elimination in anesthetized patients?

In anesthetized, mechanically ventilated patients, 10 cm H2O positive end-expiratory pressure (PEEP10) immediately decreased the CO2 volume exhaled per breath (V(CO2,br)) by 96%, as exhaled tidal volume (VT) decreased to expand functional residual capacity during the first 8 breaths after PEEP10 began. Then, the sustained decrease in V(CO2,br) for over 10 min was due to the 19% decrease in cardiac output (QT, decreased CO2 delivery from tissues to lung) and to the decrease in alveolar ventilation (VA). In turn, decreased VA resulted from decreased VT (loss of inspired volume into the compressible volume of the ventilating circuit) and possibly from increased physiological dead space, due to the potential for new high alveolar ventilation-to-perfusion (VA/Q) lung regions. V(CO2,br) increased and recovered to baseline by 20 min of PEEP10 ventilation because QT increased to augment the CO2 delivery to the lung and alveolar P(CO2) increased (increased mixed venous P(CO2) and tissue CO2 retention) to increase V(CO2,br) while alveolar VT remained depressed. End-tidal P(CO2) (PET(CO2) progressively increased during PEEP10 and did not detect the decrease in V(CO2,br) during PEEP10 ventilation because PET(CO2) does not account for exhaled volume.

Can capnography detect bronchial flap-valve expiratory obstruction?

OBJECTIVE: We have previously shown in a mechanical lung model [1] that bronchial flap-valve expiratory obstruction results in sequential lung expiration, best detected by prolonged and low magnitude tracheal expired flow (V) from the obstructed lung. However, the normal expiratory resistance of clinical ventilation circuits might also generate prolonged, low value exhaled V, that could be confused with bronchial flap-valve obstruction. We reasoned that bronchial flap-valve obstruction would also cause sequential CO2 unloading from each lung and result in a biphasic tracheal capnogram. METHODS: To test this hypothesis, we ventilated (VT, 650 ml; f, 10 br/min) a dual mechanical test lung, with each side connected to a separate alcohol-burning chamber. An airway adapter-monitor system measured airway V, P, PCO2, and FO2. The circumference of the diaphragm in a respiratory one-way valve was trimmed to generate unidirectional resistance to expiratory V. Measurement sequences were repeated after this flap-valve was interposed in the left "main-stem bronchus." RESULTS AND DISCUSSION: During moderate or severe left bronchial flap-valve obstruction, left bronchial V was delayed so that the left lung anatomical dead space (devoid of CO2) mixed with normal right exhalate to depress the expiratory upstroke or early plateau of the tracheal capnogram. During severe obstruction, decreased perfusion of the left lung caused lower alveolar PCO2. Then, prolonged low V from the left bronchus also resulted in depression of the end of the tracheal alveolar plateau. In general, the low magnitude of bronchial V from the obstructed lung limited its effect on the tracheal capnogram and the best marker of sequential lung emptying during bronchial flap-valve obstruction may be late exhaled V without reduction in total tidal volume.

Carbon dioxide spirogram (but not capnogram) detects leaking inspiratory valve in a circle circuit.

UNLABELLED: Expiratory valve incompetence in the circle circuit is diagnosed by using capnography (PCO2 versus time) when significant CO2 is present throughout inspiration. However, inspiratory valve incompetence will allow CO2-containing expirate to reverse flow into the inspiratory limb. CO2 rebreathing occurs early during the next inspiration, generating a short extension of the alveolar plateau and decreased inspiratory downslope of the capnogram, which may be indistinguishable from normal. We hypothesized that CO2 spirography (PCO2 versus volume) would correctly measure inspired CO2 volume (VCO2) during inspiratory valve leak. Accordingly, a metabolic chamber (alcohol combustion) was connected to a lung simulator, which was mechanically ventilated through a standard anesthesia circle circuit. By multiplying and integrating airway flow and PCO2, overall, expired, and inspired VCO2 (VCO2,br = VCO2,E - VCO2,I) were measured. When the inspiratory valve was compromised (by placing a wire between the valve seat and diaphragm), VCO2,I increased from 2.7 +/- 1.7 to 5.7 +/- 0.2 mL (P < 0.05), as measured by using CO2 spirography. In contrast, the capnogram demonstrated only an imperceptible lengthening of the alveolar plateau and did not measure VCO2,I. To maintain effective alveolar ventilation and CO2 elimination, increased VCO2,I requires a larger tidal volume, which could result in pulmonary barotrauma, decreased cardiac output, and increased intracranial pressure. IMPLICATIONS: Circle circuit inspiratory valve leak will allow CO2-containing expirate to reverse flow into the inspiratory limb, with subsequent rebreathing during the next inspiration. This CO2 rebreathing causes imperceptible lengthening of the alveolar plateau of the capnogram and is detected only by using the CO2 spirogram (PCO2 versus volume).

Bymixer provides on-line calibration of measurement of CO2 volume exhaled per breath.

The measurement of CO2 volume exhaled per breath (VCO2.br) can be determined during anesthesia by the multiplication and integration of tidal flow (V) and PCO2. During side-stream capnometry, PCO2 must be advanced in time by transport delay (TD), the time to suction gas through the sampling tube. During ventilation, TD can vary due to sample line connection internal volume or flow rate changes. To determine correct TD and measure accurate VCO2.br during actual ventilation. TD can be iteratively adjusted (TDADJ) until VCO2-br/tidal volume equals PCO2 measured in a mixed expired gas collection (PECO2) (J Appl. Physiol. 72:2029-2035, 1992). However. PECO2 is difficult to measure during anesthesia because CO2 is absorbed in the circle circuit. Accordingly, we implemented a bypass flow-mixing chamber device (bymixer) that was interposed in the expiration limb of the circle circuit and accurately measured PECO2 over a wide range of conditions of ventilation of a test lung-metabolic chamber (regression slope = 1.01: R2 = 0.99). The bymixer response (time constant) varied from 18.1 +/- 0.03 sec (12.5 l/min ventilation) to 66.7 +/- 0.9 sec (2.5 l/min). Bymixer PECO2 was used to correctly determine TDADJ (without interrupting respiration) to enable accurate measurement of VCO2.br over widely changing expiratory flow patterns.

Protective effect of stroma-free methemoglobin during cyanide poisoning in dogs.

BACKGROUND: During fire exposure, cyanide toxicity can block aerobic metabolism. Oxygen and sodium thiosulfate are accepted therapy. However, nitrite-induced methemoglobinemia, which avidly binds cyanide, decreases oxygen-carrying capacity that is already reduced by the presence of carboxyhemoglobin (inhalation of carbon monoxide in smoke). This study tested whether exogenous stroma-free methemoglobin (SFmetHb) can prevent depression of hemodynamics and metabolism during canine cyanide poisoning. METHODS: In 10 dogs (weighing 18.8 +/- 3.5 kg) anesthetized with chloralose-urethane and mechanically ventilated with air, baseline hemodynamic and metabolic measurements were made. Then, 137 +/- 31 ml of 12 g% SFmetHb was infused into five dogs (SFmetHb group). Finally, the SFmetHb group and the control group (n = 5, no SFmetHb) received an intravenous potassium cyanide infusion (0.072 mg.kg-1.min-1) for 20 min. Oxygen consumption (VO2) was measured with a Datex Deltatrac (Datex Instruments, Helsinki, Finland) metabolic monitor and cardiac output (QT) was measured by pulmonary artery thermodilution. RESULTS: From baseline to cyanide infusion in the control group, QT decreased significantly (p < 0.05) from 2.9 +/- 0.8 to 1.5 +/- 0.4 l/min, mixed venous PCO2 (PvCO2) tended to decrease from 35 +/- 4 to 23 +/- 2 mmHg, PvO2 increased from 43 +/- 4 to 62 +/- 8 mmHg, VO2 decreased from 93 +/- 8 to 64 +/- 19 ml/min, and lactate increased from 2.3 +/- 0.5 to 7.1 +/- 0.7 mM. In the SFmetHb group, cyanide infusion did not significantly change these variables. From baseline to infused cyanide, the increases in blood cyanide (4.8 +/- 1.0 to 452 +/- 97 microM) and plasma thiocyanate cyanide (18 +/- 5 to 65 +/- 22 microM) in the SFmetHb group were significantly greater than those increases in the control group. SFmetHb itself caused no physiologic changes, except small decreases in heart rate and PvO2. Peak SFmetHb reached 7.7 +/- 1.0% of total hemoglobin. CONCLUSIONS: Prophylactic intravenous SFmetHb preserved cardiovascular and metabolic function in dogs exposed to significant intravenous cyanide. Blood concentrations of cyanide, and its metabolite, thiocyanate, revealed that SFmetHb trapped significant cyanide in blood before tissue penetration.

How does experimental pulmonary embolism decrease CO2 elimination?

To test how large pulmonary embolism changes non-steady state CO2 kinetics, the right pulmonary artery (RPA) was occluded in 5 anesthetized, ventilated, thoracotomized dogs. By 1 min after RPA occlusion, CO2 volume exhaled per breath (VCO2,br) decreased from 9.3 +/- 2.8 to 7.0 +/- 2.6 ml and end-tidal PCO2 (PETCO2) decreased from 28.7 +/- 4.2 to 21.8 +/- 3.3 Torr. During the ensuing 70 min, VCO2,br increased back to baseline but PETCO2 was still 13% less than baseline. Both PaCO2 (41.5 +/- 1.7 to 55.1 +/- 8.1 Torr) and PvCO2 (48.2 +/- 1.9 to 62.8 +/- 6.5 Torr) steadily increased and approached equilibrium by 45 min of RPA occlusion. Cardiac output did not significantly change. In summary, RPA occlusion immediately decreased VCO2,br by 25%, due mostly to increased alveolar VD (VDalv). Then, VCO2,br recovered back to baseline as CO2 accumulated in tissues and lung. In contrast, elevated VDalv caused persistent decreased PETCO2, which did not detect recovery of VCO2,br nor increase in PaCO2 during RPA occlusion.

Carbon dioxide elimination measures resolution of experimental pulmonary embolus in dogs.

Patients with severe pulmonary embolism can suffer progressive hypercapnia refractory to supramaximal mechanical ventilation, and may require open-thoracic or transvenous emergency embolectomy in addition to anticoagulation and/or thrombolysis. The functional recovery of gas exchange would be signaled by an increase in pulmonary CO2 elimination and decrease in CO2 retention; such data could guide the course of operative embolectomy. Accordingly, we studied five chloralose-urethane anesthetized, mechanically ventilated dogs with open thoraces in which the right pulmonary arteries (RPAs) were reversibly occluded with cloth snares. After waiting for steady state, we abruptly released the snare to restore RPA perfusion and experimentally simulate resolution of pulmonary embolism. For 70 min we serially measure the CO2 volume exhaled per breath (VCO2,br), arterial, mixed venous, and end-tidal PCO2 (PACO2, PVCO2, PETCO2), cardiac output (QT), and the alveolar dead space fraction (VDalv/VTalv = [PaCO2 - PETCO2/PaCO2). RPA reperfusion caused VCO2,br to significantly and abruptly increase from 8.9 +/- 2.7 to 11.6 +/- 3.6 mL; 70 min later VCO2,br had returned to baseline. PaCO2 and PVCO2 steadily decreased during 70 min of RPA reperfusion. PETCO2 increased from 25 +/- 5 to 33 +/- 5 mm Hg immediately after RPA reperfusion, as VDalv/VTalv decreased from 54% +/- 10% to 32% +/- 12%, but PETCO2 was still significantly greater than baseline at 70 min of RPA reperfusion. QT did not significantly change. We conclude that intraoperative measurement of VCO2,br should immediately detect and follow the resolution of CO2 retention in the lung and peripheral tissues after RPA reperfusion. PETCO2 could not detect the decrease of VCO2,br back to baseline because PETCO2 does not measure exhaled volume or the PCO2 waveform.

Measurement of blood CO2 concentration with a conventional PCO2 analyzer.

OBJECTIVES: CO2 content can be determined from the Pco2 in an acidified (forces all CO2 into solution) and diluted blood sample. However, Pco2 concentrations measured in conventional blood gas analyzers are only correct for samples with a significant buffer capacity (such as whole blood), so that mixing with the Pco2 in the rinse solution and tubing walls does not significantly change the sample Pco2. This study describes a calibration method and validation data for the Radiometer Medical ABL2 CO2 electrode system to accurately measure unbuffered blood samples used in the determination of blood CO2 content (or other aqueous fluids). DESIGN: Prospective, criterion standard. SETTING: Laboratory. MEASUREMENTS AND MAIN RESULTS: Blood samples (0.4 mL) were acidified and diluted with 0.2 M lactic acid. After measuring Pco2, CO2 content was calculated using the CO2 solubility coefficient and the dilution factor of 20. CO2 content was determined in a series of sodium carbonate (Na2CO3) solutions spanning the physiologic range of CO2 content. Regression of the measured vs. the actual CO2 content data generated a straight line with a slope of 0.796 and y-intercept of 12.5 (r2 = .99; n = 48). These coefficients were successfully used to correct CO2 content determined in blood samples into which graduated amounts of sodium carbonate were added. CONCLUSIONS: This calibration procedure allows accurate measurement of Pco2 in aqueous samples using the Radiometer ABL2 electrode system, and should be applicable to other blood gas analyzers. Necessary syringes and chemicals are readily available, the method is fast and simple, and the sample volume is small. In the practice of critical care medicine, accurate Pco2 measurement in aqueous acidified and diluted blood provides direct determination of blood CO2 content (useful in calculations of modified Fick cardiac output or tissue CO2 production). Determinations of absolute CO2 content in blood requiring complex methodology are not necessary. In addition, accurate measurement of aqueous gastric Pco2 can help determine gastric pH, which is an important marker of tissue perfusion.

Measurement of pulmonary CO2 elimination must exclude inspired CO2 measured at the capnometer sampling site.

OBJECTIVE: The pulmonary elimination of the volume of CO2 per breath (VCO2/br, integration of product of airway flow (V) and PCO2 over a single breath) is a sensitive monitor of cardio-pulmonary function and tissue metabolism. Negligible inspired PCO2 results when the capnometry sampling site (SS) is positioned at the entry of the inspiratory limb to the airway circuit. In this study, we test the hypothesis that moving SS lungward will result in significant inspired CO2 (VCO2[I]), that needs to be excluded from VCO2/br. METHODS: We ventilated a mechanical lung simulator with tidal volume (VT) of 800 mL at 10 breaths/min. CO2 production, generated by burning butane in a separate chamber, was delivered to the lung. Airway V and PCO2 were measured (Capnomac Ultima, Datex), digitized (100 Hz for 60 s), and stored by microcomputer. Then, computer algorithms corrected for phase differences between V and PCO2 and calculated expired and inspired VCO2 (VCO2[E] and VCO2[I]) for each breath, whose difference equalled overall VCO2/br. The lung and Y-adapter (where the inspiratory and expiratory limbs of the circuit joined) were connected by the SS and a connecting tube in varying order. RESULTS: During ventilation of the lung model (VT = 800 ml) with SS adjacent to the inspiratory limb, VCO2[E] was 16.8 +/- 0.4 ml and VCO2[I] was 1.1 +/- 0.1 ml, resulting in overall VCO2/br (VCO2[E] - VCO2[I]) of 15.7 +/- 0.4 ml. If VCO2[I] was ignored in the determination of VCO2/br, then the %error that VCO2[E] overestimated VCO2/br was 7.2 +/- 0.3%. This %error significantly increased (p < 0.05, Student's t-test) when VT was decreased to 500 mL (%error = 12.4 +/- 0.8%) or when SS was moved to the lungward side of a 60 mL connecting tube (VCO2[I] = 2.8 +/- 0.2, %error = 18.2 +/- 1.6) or a 140 mL tube (VCO2[I] = 5.9 +/- 0.3 mL, %error = 37.5 +/- 3.3). CONCLUSIONS: When the SS was moved lungward from the inspiratory limb, instrumental dead space (VDINSTR) increased and, at end-expiration, contained exhaled CO2 from the previous breath. During the next inspiration, this CO2 was rebreathed relative to SS (i.e. VCO2[I]), and contributed to VCO2[E]. Thus, VCO2[E] overestimated VCO2/br (%error) by the amount of rebreathing, which was exacerbated by larger VDINSTR (increased VCO2[I]) or smaller VT (increased VCO2[I]-to-VCO2/br ratio).

How does positive end-expiratory pressure decrease CO2 elimination from the lung?

Six chloralose-urethane anesthetized dogs (23 +/- 2 kg) underwent median thoracotomy (open pleural spaces) and constant mechanical ventilation with O2. We conducted measurements at baseline and during 25 min of ventilation with 3.3 cmH2O positive end-expiratory pressure (PEEP3) or 10.7 cmH2O PEEP (PEEP 11), including breath-by-breath values in the first 2 min after PEEP began. PEEP 11 immediately decreased pulmonary CO2 elimination per breath (VCO2,br, digital integration and multiplication of exhaled flow and FCO2) from 8.4 +/- 2.0 to 4.5 +/- 1.6 ml (P < 0.05) by significantly decreasing alveolar ventilation (VA) (29% increase in anatomical dead space (VDana) and generation of high VA/Q regions) and by decreasing alveolar PCO2 (PACO2) from 42.5 +/- 3.5 to 35.9 +/- 3.5 Torr (decreased CO2 transfer to the lung as electromagnetic aortic cardiac output (QT) decreased by 51%). The immediate dilution of alveolar gas and PACO2 by fresh gas as PEEP increased functional residual capacity by 1152 +/- 216 ml was offset by simultaneous decreased expiratory volume and, hence, CO2 accumulation. Compared to baseline, the 17% reduction in VCO2,br was sustained at 25 min after addition of PEEP 11 because VA remained depressed. Then, VCO2,br could only be restored to baseline if PACO2 sufficiently increased. However, CO2 transport was still in unsteady state at 25 min of PEEP. Peripheral tissue retention of CO2 and the significant increase in mixed venous PCO2 (PVCO2, 62.4 +/- 6.2 Torr) were not enough to normalize CO2 transfer to the lung and to sufficiently increase PACO2, especially during the continued depression in QT that occurred at higher PEEP. The sustained decrease in VCO2,br during PEEP was not mirrored by changes in end-tidal PCO2 (PETCO2).

Comparison of end-tidal PCO2 and average alveolar expired PCO2 during positive end-expiratory pressure.

The measurement of average alveolar expired PCO2 (PAECO2) weights each PCO2 value on the alveolar plateau of the CO2 expirogram by the simultaneous change in exhaled volume. PAECO2 can be determined from a modified analysis of the Fowler anatomic dead space (VDANAT). In contrast, end-tidal PCO2 (PETCO2) only measures PCO2 in the last small volume of exhalate. In conditions such as mechanical ventilation with positive end-expiratory pressure (PEEP), where the alveolar plateau can have a significant positive slope, we questioned how much PETCO2 could overestimate PAECO2. Accordingly, in six anesthetized ventilated dogs, we digitally measured and processed tidal PCO2 and flow to determine VDANAT. We determined PETCO2 and PAECO before and after the application of 7.6 cm H2O PEEP. Alveolar dead space to tidal volume fraction (VD/VT) was determined by [arterial PCO2- alveolar PCO2]/arterial PCO2, where alveolar PCO2 was determined by either PETCO2 or PAECO2. During baseline ventilation, PETCO2 was 3.4 mm Hg (approximately 11%) greater than PAECO2. Because PEEP significantly increased the slope of the alveolar plateau from 28 to 74 mm Hg/L, the difference between PETCO2 and PAECO2 significantly increased to 6.6 mm Hg (approximately 20% difference). The concurrent increase in VDANAT during PEEP decreased alveolar tidal volume and tended to limit the overestimation of PETCO2 compared to PAECO2. When alveolar PCO2 was estimated by PETCO2, alveolar VD/VT was 18%, compared to an alveolar VD/VT of 26% when alveolar PCO2 was estimated by PAECO2. This difference was significantly magnified during PEEP ventilation. The overestimation of PAECO2 by PETCO2 can result in a falsely high assessment of overall alveolar PCO2. Moreover, the use of PETCO2 to estimate alveolar PCO2 in the determination of the alveolar dead space fraction can result in falsely low and even negative values of alveolar dead space.

Effect of oxygen and sodium thiosulfate during combined carbon monoxide and cyanide poisoning.

In a canine model of combined carbon monoxide (CO) and cyanide (CN) poisoning, cardiac output (QT) and oxygen consumption (Vo2) decreased but recovered to baseline values by 15 min after toxic exposure; elevated blood CN and lactic acidosis persisted for at least another 10 min. Given the rapid spontaneous recovery after cessation of toxic exposure, we questioned the efficacy of usual treatment with oxygen (O2) and sodium thiosulfate (Na2S2O3) for CN poisoning. Accordingly, in seven dogs (26 +/- 3 kg, chloralose and urethane anesthesia), we sequentially administered CO by closed circuit inhalation (231 +/- 42 ml) and potassium CN by intravenous infusion (0.072 mg.kg-1.min-1 for 17 +/- 3 min). Fifteen minutes after toxic exposure, O2 breathing began and Na2S2O3 (150 mg/kg) was infused. Measurements were repeated 10 and 45 min after treatment. At the end of the CN infusion, QT decreased by 43% and Vo2 decreased by 51%, compared to baseline values. Both variables recovered to baseline by 15 min after stopping toxic exposure. Significant lactic (4.8 +/- 2.9 mM) acidosis (7.14 +/- 0.10) persisted for at least another 10 min. Treatment with oxygen and Na2S2O3 did not hasten the recovery of this lactic acidosis or decrease blood cyanide levels compared to nontreated dogs. However, after treatment, plasma thiocyanate significantly increased from 16.3 +/- 12.5 to 94.4 +/- 72.2 microM, as Na2S2O3 participated in the increased metabolism of cyanide to thiocyanate. We conclude that O2 and Na2S2O3 therapy should be continued during combined CO and HCN poisoning. Oxygen increases CO elimination and can enhance anti-CN treatment. After infusion or inhalation of CN, when most CN has already penetrated the intracellular compartment, postexposure sodium thiosulfate increased the metabolism of CN.