Pulse Oximetry in the Postoperative Care of Cardiac Surgical Patients· A Randomized Controlled Trial Morris I. Bierman, M.D., F.C.C.f;t Keith L. Stein, M.D., F.C.C.E;+ and ]ames v Snyder; M.D., F.C.C.f§ study objective: To demonstrate the utility of pulse oximetry in detecting clinically unapparent episodes of arterial de saturation in postoperative cardiac surgical patients and to evaluate the effect of pulse oximetry on ordering arterial blood gas analyses. Deaign: Prospective, randomized, partially blinded comparison. Setting: Cardiothoracic surgical intensive care unit. Patienta: 35 patients following elective cardiac surgical procedures. Interventions: All patients were monitored continuously with pulse oximetry throughout their ICU course. In group 1 patients, the SpOt data were available at the bedside. In group 2 patients, the SpOt data were masked at the bedside and monitored at a remote location. MeaauremeratB and reaulta: Utilization of pulse oximetry
allowed a signi6cant reduction in arterial blood gas utilization in group 1 (group I: 12.4±7.5 blood gas analyses per ICU admission vs group 2: 23.1±8.8; p=O.OOO7) without adverse events. Clinically unapparent desaturations were detected in 7 of 15 patients in group 2. Conclusions: Pulse oximetry improves patient safety through the detection of clinically unapparent episodes of desaturation and can allow a reduction in the number of blood gas analyses utilized without adverse effects to the patient. This may allO\lV a potential cost savings to the patient. (Chat 1992; 102:1367-70)
oximetry has become widely accepted as a P ulse means of noninvasively monitoring arterial oxygen
sodes that would otherwise be unapparent and can reduce the need for intermittent blood gas analysis.
saturation (SpOJ in patients in the intensive care unit (ICU). Nevertheless, its purported benefits in this setting-greater patient safety and a cost savings due to fewer arterial blood gas analyses-have not yet been demonstrated in a well controlled manner.v" A single randomized study found that fewer blood gas analyses were required with pulse oximetry and capnography, and that the ICU staff were alerted to episodes of hypoxemia and hypercarbia. Although this study was limited to the period of weaning and by the inherent limitations of capnography, its findings warranted further evaluation of noninvasive respiratory monitoring in the leu. We conducted a prospective, randomized trial of continuous pulse oximetry in patients after elective cardiac surgical procedures. A unique feature of our study was the use of a "blinded" group to investigate the value of the pulse oximeter in detecting clinically unapparent episodes of arterial desaturation. Our results support the hypotheses that when properly utilized, pulse oximetry can detect hypoxemic epi*From the Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh Medical Center, Presbyterian University Hospital, Pittsburgh. t Assistant Professor. *Associate Professor. §Professor. Manuscript received October 14, 1991; revision accepted April 29.
AVR=aortic valve replacement; CABG=coronary artery bypass graft; MVR mitral valve replacement; SpOt oxygen saturation measured by pulse oximetry; SaO I = oxygen saturation measured by CG-oximeter
=
=
MATERIALS AND METHODS
The study was approved by the Institutional Review Board at our institution, and patient consent was obtained preoperatively Forty patients scheduled for elective cardiac surgical procedures were enrolled in the study over a six-month period. On their arrival at the ICU after surgery, patients were randomly assigned to one of two groups. Both groups were monitored continuously by pulse oximetry (Nellcor N-200, NeUcor, Inc). In group 1, the SpOt data were available at the bedside); in group 2, the SpOt data were masked at the bedside and transmitted to a remote site. In both groups, each time an arterial blood gas analysis was performed, a CO-oximeter (Radiometer OSM 3, Radiometer, Inc) determination of the saturation was obtained for comparison with the pulse oximeter. The SpOt values lying outside ± 4 percent were considered a failure of pulse oximetry. For patients in group 1, the nursing staff and the physicians were instructed to use the pulse oximetry data in lieu of arterial blood gas analysis whenever possible. No restriction was placed on the frequency or number of arterial blood gas analyses, and a mandatory analysis was obtained each day before morning rounds. Each time an arterial blood gas analysis was obtained, the physician or nurse was required to record the reason for the analysis. The desaturation alarm on the oximeters was set to sound at values less than or equal to 93 percent. The pulse rate alarm was deactivated, because the patients were also monitored by continuous electrocardiography The nursing staff was required to notify the physician of arterial saturations less than 93 percent or inability to obtain an adequate signal. The physician's response to these conditions was not predetermined. For patients in group 2, blood gas analyses were obtained every 6 h or more frequently as clinically indicated- the standard protocol in our intensive care unit. Again, the reason for each arterial blood CHEST I 102 I 5 I NOVEMBER. 1992
1387
Table 2- Patient ChtJracteriatica
Table I-Criteria For Notification OfBetUide Staff Loss of telemetry signal Inadequate telemetry signal lasting ~5 min 90 percent SSpOI s93 percent lasting for> 10 min SPOI 15
7 1 7 10 4 10 0
5 1 5 6 0 3 0
2 3 2 1 1 2
1 0 1 1 0 0
0
1
2 1 2 1 0 3 0
Desaturations < 1 min duration were excluded.
Pulse Oximetry in Postoperative care of Cardac Surgical Patients (Bierman,SteIn, Snyder)
recordings in these patients revealed numerous transient desaturations (less than I min) in addition to the episodes described above (Table 4). The significance of these brief episodes was unclear, however, as it was not always possible to exclude artifacts attributable to unstable plethysmographic tracings. The number of changes in ventilatory support per ICU stay was not different between the two groups (group I: 10.0±5.6 group 2: 9.8±6.2; p=0.92). The number of ventilator changes per arterial blood gas analysis obtained was, however, significantly different (group 1: 1.23± 0.52 vs group 2: 0.71 ±0.30; p=O.OOI) indicating a greater reliance on the pulse oximeter in making ventilatory adjustments when such data were available at the bedside. The dose of supplemental oxygen (ranging in both groups from 2 Umin by nasal cannula to 50 percent O2 by face mask) was adjusted more frequently in group 2 (group I: 1.25±0.91 manipulations of oxygen dosage per ICU stay vs group 2: 2.53 ± 1.69, p = 0.02). Since virtually all such adjustments were decreases in oxygen supply during this study period, physicians had a greater tendency to reduce the dose of supplemental oxygen in the group whose care was guided by arterial blood gas data (group I: 0.95±O.51 decreases in supplemental oxygen per ICU stay vs group 2: 1.93± 1.16, p=O.OO7). DISCUSSION
Although pulse oximetry is widely used, its utility in the intensive care unit has not been well demonstrated.':" Our data indicate that pulse oximetry may contribute to patient safety by detecting potentially serious desaturations that may not be clinically evident. Our data also indicate that by reducing the use of arterial blood gas analyses, pulse oximetry reduces the associated cost as well as the need for phlebotomy. Furthermore, the reduction in the utilization of arterial blood gas analyses was not associated with adverse patient outcomes. Studies of pulse oximetry have demonstrated that in the clinically relevant range for most critically ill patients (SP02 2:85 percent, corresponding to P0 2 ~ 50 mm Hg), the accuracy of the instruments is ± 4 percent with 95 percent confidence limits. 1-4 A recent study by Jubran and Tobin" explored the reliability of pulse oximetry in titrating oxygen therapy in patients on mechanical ventilation using several protocols," They have demonstrated that in caucasian patients, using a target saturation of 94 percent, the positive predictive value of pulse oximetry for a P0 2 2:60 is 100 percent. Thus, they have suggested that such a limit may actually be too conservative. In black patients, a higher limit (Sa02 ~95 percent) was required to reliably prevent inadvertent hypoxemic episodes.
We chose the 94 percent threshold because of the reported 95 percent confidence intervals of ± 4 percent for pulse oximetry. 1 This corresponds to the potential of an actual saturation of90 percent (Po2= 60 mm Hg) in a patient whose pulse oximeter is reading 94 percent, which we believed to be the lower limit tolerable in patients recovering from cardiac surgery. Our data, obtained almost exclusively in white patients, demonstrate that a Sp02 measurement ~94 percent predicted a P0 2 ~60 mm Hg in 99.7 percent of 526 determinations. This is in agreement with the findings of Jubran and Tobin." There were too few instances of serious hypoxemia where there were simultaneous Sp02 and Sa02 values to allow a proper analysis at the low end of the clinically important range. We agree with these authors that 94 percent may be too conservative a threshold for titration of supplemental oxygen for patients of this type. They have proposed a Sp02 of 92 percent in caucasian patients and 95 percent in black patients as adequate target values for the titration of supplemental oxygen. In group 2 patients, Sp02 measurements revealed arterial desaturations that were clinically unapparent, but potentially important. In this cohort of patients, all of whom had undergone elective surgical procedures, there were few episodes of severe desaturation (Sp02
allowed a signi6cant reduction in arterial blood gas utilization in group 1 (group I: 12.4±7.5 blood gas analyses per ICU admission vs group 2: 23.1±8.8; p=O.OOO7) without adverse events. Clinically unapparent desaturations were detected in 7 of 15 patients in group 2. Conclusions: Pulse oximetry improves patient safety through the detection of clinically unapparent episodes of desaturation and can allow a reduction in the number of blood gas analyses utilized without adverse effects to the patient. This may allO\lV a potential cost savings to the patient. (Chat 1992; 102:1367-70)
oximetry has become widely accepted as a P ulse means of noninvasively monitoring arterial oxygen
sodes that would otherwise be unapparent and can reduce the need for intermittent blood gas analysis.
saturation (SpOJ in patients in the intensive care unit (ICU). Nevertheless, its purported benefits in this setting-greater patient safety and a cost savings due to fewer arterial blood gas analyses-have not yet been demonstrated in a well controlled manner.v" A single randomized study found that fewer blood gas analyses were required with pulse oximetry and capnography, and that the ICU staff were alerted to episodes of hypoxemia and hypercarbia. Although this study was limited to the period of weaning and by the inherent limitations of capnography, its findings warranted further evaluation of noninvasive respiratory monitoring in the leu. We conducted a prospective, randomized trial of continuous pulse oximetry in patients after elective cardiac surgical procedures. A unique feature of our study was the use of a "blinded" group to investigate the value of the pulse oximeter in detecting clinically unapparent episodes of arterial desaturation. Our results support the hypotheses that when properly utilized, pulse oximetry can detect hypoxemic epi*From the Department of Anesthesiology/Critical Care Medicine, University of Pittsburgh Medical Center, Presbyterian University Hospital, Pittsburgh. t Assistant Professor. *Associate Professor. §Professor. Manuscript received October 14, 1991; revision accepted April 29.
AVR=aortic valve replacement; CABG=coronary artery bypass graft; MVR mitral valve replacement; SpOt oxygen saturation measured by pulse oximetry; SaO I = oxygen saturation measured by CG-oximeter
=
=
MATERIALS AND METHODS
The study was approved by the Institutional Review Board at our institution, and patient consent was obtained preoperatively Forty patients scheduled for elective cardiac surgical procedures were enrolled in the study over a six-month period. On their arrival at the ICU after surgery, patients were randomly assigned to one of two groups. Both groups were monitored continuously by pulse oximetry (Nellcor N-200, NeUcor, Inc). In group 1, the SpOt data were available at the bedside); in group 2, the SpOt data were masked at the bedside and transmitted to a remote site. In both groups, each time an arterial blood gas analysis was performed, a CO-oximeter (Radiometer OSM 3, Radiometer, Inc) determination of the saturation was obtained for comparison with the pulse oximeter. The SpOt values lying outside ± 4 percent were considered a failure of pulse oximetry. For patients in group 1, the nursing staff and the physicians were instructed to use the pulse oximetry data in lieu of arterial blood gas analysis whenever possible. No restriction was placed on the frequency or number of arterial blood gas analyses, and a mandatory analysis was obtained each day before morning rounds. Each time an arterial blood gas analysis was obtained, the physician or nurse was required to record the reason for the analysis. The desaturation alarm on the oximeters was set to sound at values less than or equal to 93 percent. The pulse rate alarm was deactivated, because the patients were also monitored by continuous electrocardiography The nursing staff was required to notify the physician of arterial saturations less than 93 percent or inability to obtain an adequate signal. The physician's response to these conditions was not predetermined. For patients in group 2, blood gas analyses were obtained every 6 h or more frequently as clinically indicated- the standard protocol in our intensive care unit. Again, the reason for each arterial blood CHEST I 102 I 5 I NOVEMBER. 1992
1387
Table 2- Patient ChtJracteriatica
Table I-Criteria For Notification OfBetUide Staff Loss of telemetry signal Inadequate telemetry signal lasting ~5 min 90 percent SSpOI s93 percent lasting for> 10 min SPOI 15
7 1 7 10 4 10 0
5 1 5 6 0 3 0
2 3 2 1 1 2
1 0 1 1 0 0
0
1
2 1 2 1 0 3 0
Desaturations < 1 min duration were excluded.
Pulse Oximetry in Postoperative care of Cardac Surgical Patients (Bierman,SteIn, Snyder)
recordings in these patients revealed numerous transient desaturations (less than I min) in addition to the episodes described above (Table 4). The significance of these brief episodes was unclear, however, as it was not always possible to exclude artifacts attributable to unstable plethysmographic tracings. The number of changes in ventilatory support per ICU stay was not different between the two groups (group I: 10.0±5.6 group 2: 9.8±6.2; p=0.92). The number of ventilator changes per arterial blood gas analysis obtained was, however, significantly different (group 1: 1.23± 0.52 vs group 2: 0.71 ±0.30; p=O.OOI) indicating a greater reliance on the pulse oximeter in making ventilatory adjustments when such data were available at the bedside. The dose of supplemental oxygen (ranging in both groups from 2 Umin by nasal cannula to 50 percent O2 by face mask) was adjusted more frequently in group 2 (group I: 1.25±0.91 manipulations of oxygen dosage per ICU stay vs group 2: 2.53 ± 1.69, p = 0.02). Since virtually all such adjustments were decreases in oxygen supply during this study period, physicians had a greater tendency to reduce the dose of supplemental oxygen in the group whose care was guided by arterial blood gas data (group I: 0.95±O.51 decreases in supplemental oxygen per ICU stay vs group 2: 1.93± 1.16, p=O.OO7). DISCUSSION
Although pulse oximetry is widely used, its utility in the intensive care unit has not been well demonstrated.':" Our data indicate that pulse oximetry may contribute to patient safety by detecting potentially serious desaturations that may not be clinically evident. Our data also indicate that by reducing the use of arterial blood gas analyses, pulse oximetry reduces the associated cost as well as the need for phlebotomy. Furthermore, the reduction in the utilization of arterial blood gas analyses was not associated with adverse patient outcomes. Studies of pulse oximetry have demonstrated that in the clinically relevant range for most critically ill patients (SP02 2:85 percent, corresponding to P0 2 ~ 50 mm Hg), the accuracy of the instruments is ± 4 percent with 95 percent confidence limits. 1-4 A recent study by Jubran and Tobin" explored the reliability of pulse oximetry in titrating oxygen therapy in patients on mechanical ventilation using several protocols," They have demonstrated that in caucasian patients, using a target saturation of 94 percent, the positive predictive value of pulse oximetry for a P0 2 2:60 is 100 percent. Thus, they have suggested that such a limit may actually be too conservative. In black patients, a higher limit (Sa02 ~95 percent) was required to reliably prevent inadvertent hypoxemic episodes.
We chose the 94 percent threshold because of the reported 95 percent confidence intervals of ± 4 percent for pulse oximetry. 1 This corresponds to the potential of an actual saturation of90 percent (Po2= 60 mm Hg) in a patient whose pulse oximeter is reading 94 percent, which we believed to be the lower limit tolerable in patients recovering from cardiac surgery. Our data, obtained almost exclusively in white patients, demonstrate that a Sp02 measurement ~94 percent predicted a P0 2 ~60 mm Hg in 99.7 percent of 526 determinations. This is in agreement with the findings of Jubran and Tobin." There were too few instances of serious hypoxemia where there were simultaneous Sp02 and Sa02 values to allow a proper analysis at the low end of the clinically important range. We agree with these authors that 94 percent may be too conservative a threshold for titration of supplemental oxygen for patients of this type. They have proposed a Sp02 of 92 percent in caucasian patients and 95 percent in black patients as adequate target values for the titration of supplemental oxygen. In group 2 patients, Sp02 measurements revealed arterial desaturations that were clinically unapparent, but potentially important. In this cohort of patients, all of whom had undergone elective surgical procedures, there were few episodes of severe desaturation (Sp02
