Unrecognized Pulmonary Venous Desaturation Early After Norwood Palliation Confounds Q̇p:Q̇s Assessment and Compromises Oxygen Delivery
Background—Hemodynamic stability after Norwood palliation often requires manipulation of pulmonary vascular resistance to alter the pulmonary-to-systemic blood flow ratio (Q̇p:Q̇s). Q̇p:Q̇s is often estimated from arterial saturation (Sao2), a practice based on 2 untested assumptions: constant systemic arteriovenous O2 difference and normal pulmonary venous saturation.
Methods and Results—In 12 patients early (≤3 days) after Norwood palliation, simultaneous arterial, superior vena caval (Ssvco2), and pulmonary venous (Spvo2) oximetry was used to test whether Sao2 accurately predicts Q̇p:Q̇s. Stepwise multiple regression assessed the contributions of Sao2, Ssvco2, and Spvo2 to Q̇p:Q̇s determination. Sao2 correlated weakly with Q̇p:Q̇s (R2=0.08, P<0.05). Inclusion of Ssvco2 and Spvo2 improved Q̇p:Q̇s prediction accuracy. Pulmonary venous desaturation (Spvo2 <95%) was observed frequently (30%), especially at Fio2 ≤0.21, but normalized with higher Fio2 or PEEP in all patients. In 6 patients, Fio2 was increased incrementally from 0.17 to 0.50 to determine whether this was an effective means to manipulate Q̇p:Q̇s. Q̇p:Q̇s failed to change predictably with increased Fio2. In 5 of 6 patients, however, higher Spvo2 and Sao2 enhanced systemic oxygen delivery, as demonstrated by improvement in oxygen extraction.
Conclusions—Sao2 correlated poorly with Q̇p:Q̇s because of variability in Ssvco2 and Spvo2. A novel observation was that pulmonary venous desaturation occurred frequently early after Norwood palliation but normalized with higher Fio2 or PEEP. Because unrecognized pulmonary venous desaturation confounds Q̇p:Q̇s assessment and compromises Sao2 and oxygen delivery, judicious use of inspired oxygen and PEEP may be beneficial in selected patients early after Norwood palliation.
Hemodynamic stability in the early postoperative period after Norwood palliation for hypoplastic left heart syndrome is largely dependent on maintaining a balance between pulmonary blood flow (Q̇p) and systemic blood flow (Q̇s).1 Ventilation strategies that alter pulmonary vascular resistance are routinely used to manipulate the ratio of Q̇p to Q̇s (Q̇p:Q̇s).1 2 Estimation of the Q̇p:Q̇s ratio is thus vitally important for accurate assessment of an unstable hemodynamic state and to determine the effects of interventions designed to alter the Q̇p:Q̇s. It has long been accepted that arterial oxygen saturation (Sao2) is reflective of Q̇p:Q̇s after the Norwood operation.2
The assertion that Q̇p:Q̇s can be estimated from Sao2 alone in single-ventricle physiology is based on a simplification of the Fick equation that is not well founded. Because Q̇p is supplied by a systemic-to–pulmonary artery shunt, pulmonary and systemic arterial oxygen saturations are equivalent, so that Q̇p:Q̇s can be calculated as Q̇p:Q̇s=(Sao2−Smvo2)/(Spvo2−Sao2), where mv is mixed-venous and pv pulmonary venous. This equation has been further simplified on the basis of 2 untested assumptions: (1) that systemic arteriovenous oxygen saturation difference (ΔA-V O2) is constant (commonly assumed to be ≈25), and (2) that the pulmonary venous (PV) blood is fully oxygen saturated (ie, Spvo2≥95%).1 These assumptions yield the simplified equation: Q̇p:Q̇s=25/(95−Sao2). This equation has been used to estimate Q̇p:Q̇s from Sao2 alone.
The ability to determine Q̇p:Q̇s from Sao2 is highly desirable, given the ease with which Sao2 can be obtained in clinical practice. The assumptions that underlie this simplified Fick equation may be false, however, in the immediate postoperative Norwood period. ΔA-V O2 will be significantly greater than 25 if systemic blood flow is low, a common occurrence in the postoperative period. Rossi and colleagues3 4 reported the importance of monitoring superior vena cava (SVC) oxygen saturation (Ssvco2) after Norwood palliation as a means of assessing Q̇s. The assumption that PV blood is fully saturated has never been tested in the postoperative Norwood circulation; however, it is a potentially dangerous assumption, given the routine use of minimal inspired oxygen (even subambient oxygen) and controlled hypercarbia in postoperative management of these infants. Failure to account for a decrease in Spvo2 will result in an important underestimation of Q̇p:Q̇s based on Sao2 alone. If reduced levels of inspired oxygen result in PV desaturation without a significant decrease in Q̇p:Q̇s, the associated decrease in Sao2 may be detrimental to overall systemic oxygen delivery.
The purpose of this study was to use simultaneous oximetry of arterial, SVC, and PV blood to test the hypothesis that Sao2 alone is an accurate predictor of Q̇p:Q̇s in the early postoperative Norwood circulation. We thus hoped to evaluate the validity of common assumptions regarding ΔA-V O2 and Spvo2 in postoperative Norwood patients. A second aim was to determine whether changes in fraction inspired oxygen (Fio2) are an effective means to alter Q̇p:Q̇s.
Infants with single-ventricle anatomy and aortic arch obstruction undergoing stage I Norwood palliation with functional SVC and PV catheters were eligible for inclusion. These catheters are routinely placed intraoperatively during the Norwood procedure at our institution. Exclusion criteria included weight <2 kg, clinical or radiographic evidence of pulmonary disease, or the inability to draw samples from SVC or PV catheters. The study was approved by the Institutional Review Board of Children’s Hospital Medical Center, and informed consent was obtained from a parent or legal guardian before the study.
A narcotic anesthetic was used for optimal myocardial function. After sternotomy and bypass initiation, dissection was completed during cooling. An atrial septectomy was performed via atriotomy, the aortic arch was reconstructed with a homograft patch,2 5 and the pulmonary artery bifurcation was closed primarily. Retrograde cardioplegia was administered during circulatory arrest. A Gore-tex shunt was placed during rewarming. Intracardiac catheters were inserted by the transatrial approach and the tips directed into the SVC and left lower pulmonary vein under direct visualization.
Patients were returned to the Cardiac Intensive Care Unit with open sternotomy on a fentanyl infusion and neuromuscular blockade. Postoperative monitoring included atrial and systemic arterial pressures, ECG, pulse oximetry, and end-tidal CO2. Patients were ventilated with volume control ventilation at a tidal volume of 20 mL/kg and positive end-expiratory pressure (PEEP) of 4 cm H2O. The ventilator rate was adjusted to control Paco2. Inotropic support and volume were administered as needed. Diuretic therapy was initiated routinely on the first morning after surgery.
Sao2, Spvo2, Ssvco2, and Q̇p:Q̇s
Simultaneous oximetry of arterial, SVC, and PV blood was performed as clinically indicated in all patients with functional SVC and PV catheters. Because there is no site to sample true mixed-venous blood in patients with single-ventricle physiology, blood sampled from the SVC line was considered the mixed-venous sample. It was also assumed that samples from the left lower pulmonary vein were representative of overall PV blood. To reduce potential erroneous sampling of atrial blood, the SVC and PV lines were cleared, and samples were drawn slowly with 1-mL syringes (unit protocol). True Q̇p:Q̇s was determined as (Cao2−Csvco2)/(Cpvo2−Cao2), where Cao2 is arterial oxygen content (1.36×Hgb×Sao2), Csvco2 is mixed-venous oxygen content (1.36×Hgb×Ssvco2), and Cpvo2 is PV oxygen content (1.36×Hgb×Spvo2+0.003×Ppvo2). Hemoglobin was determined oximetrically for each data set. Estimated Q̇p:Q̇s was determined from Sao2 alone by the formula 25/(95−Sao2). ΔA-V O2 was determined as Cao2−Csvco2, and the oxygen extraction ratio was determined as (Cao2−Csvco2)/Cao2, where Cao2 and Csvco2 represent arterial and mixed-venous oxygen content, respectively.
Fio2 Effects on Q̇p:Q̇s
To determine whether changes in Fio2 are an effective means to alter true Q̇p:Q̇s, data were collected in a subset of patients as Fio2 was systematically manipulated. Patients were studied 48 to 72 hours after surgery, when gas exchange and hemodynamics were steady state, usually just before sternal closure. Chest radiograph confirmed the correct position of intracardiac catheters and absence of pulmonary parenchymal changes. No manipulations in inotropic support or minute ventilation were made during the changes in Fio2. After administration of fentanyl (10 μg/kg) and vecuronium (0.1 mg/kg) and endotracheal tube suctioning, baseline oximetric data (Sao2, Ssvco2, and Spvo2) were obtained at an Fio2 of 0.21. Hemodynamic and oximetric data were subsequently collected as Fio2 was adjusted incrementally to 0.17, 0.21, 0.30, and 0.50. Hemoglobin was measured for each oximetric data set. After each change in Fio2, patients were allowed 10 minutes for equilibration before oximetric data were collected.
Oximetric data were analyzed to determine the effects of changes in Fio2 on Q̇p:Q̇s. As surrogates of systemic oxygen delivery, alterations in ΔA-V O2 and the oxygen extraction ratio were examined.
Data Analysis and Statistical Methods
Data were analyzed with Statview software (Abacus Concepts). For the purpose of correlating Sao2 with true Q̇p:Q̇s, all measurements were assumed to be independent, and stepwise linear multiple regression was used to determine the contributions of Sao2, Ssvco2, and Spvo2 to the determination of Q̇p:Q̇s. Data are presented as mean±SD. Statistical significance was defined as a value of P<0.05.
From April 1997 to December 1998, 20 patients underwent Norwood palliation at Children’s Hospital Medical Center. Twelve patients were included in this study. Of the remaining 8 patients, there was 1 intraoperative death, 1 patient weighed <2 kg, and another patient developed Gram-negative sepsis/pneumonia. In 5 patients, the PV or SVC catheter was not placed successfully or could not be used for sampling, including 1 patient who was excluded because of atrial contamination of PV blood, as evidenced by Spvo2 data that closely approximated the Sao2. Oximetric data from the 12 study patients were used to test the assumptions underlying the simplified Fick equation, and in 6 of these patients, Fio2 was systematically altered to determine the effect of Fio2 on Q̇p:Q̇s. One patient was not studied at subambient oxygen (Fio2=0.17) because of significant PV desaturation during baseline room air ventilation.
Patient characteristics are summarized in the Table⇓. The median age at the time of surgery was 7 days (range 3 to 45 days). The median weight was 3.23 kg (range 2.2 to 4.1 kg). During the study period, 14 of 20 neonates (70%) were discharged home and subsequently returned for successful bidirectional Glenn anastomosis. Of the 12 infants included in the study, 2 infants died in the early postoperative period and 1 died 2 months after the Norwood procedure of persistent hepatic failure. All nonsurviving infants demonstrated marked pulmonary overcirculation, demonstrating the highest maximum Q̇p:Q̇s measurements recorded. Excessive pulmonary blood flow in these patients was associated with clinically important PV desaturation. PV desaturation (Spvo2<95%) was observed in 11 of 12 patients but normalized with higher Fio2 or PEEP in all patients.
Prediction of Q̇p:Q̇s From Sao2
A total of 115 complete, simultaneous sets of oximetric data (Sao2, Ssvco2, and Spvo2) were available for analysis. True Q̇p:Q̇s calculated from simultaneous oximetry data ranged from 0.45 to 5.43 and correlated poorly with Sao2 (R2=0.08,P<0.05). There is a particularly poor correlation between true Q̇p:Q̇s and Sao2 at arterial saturations of 65% to 85%, where one might expect a “balanced” circulation on the basis of the usual clinical assumptions (Figure 1⇓).
Stepwise linear regression using Sao2, Ssvco2, and Spvo2 to predict Q̇p:Q̇s showed significant improvement in ability to estimate Q̇p:Q̇s with inclusion of each additional variable. Sao2 alone accounted for only 8% of the variation in Q̇p:Q̇s (R2=0.08), a remarkably low finding. Addition of Ssvco2 increased this to 56%, and the PV component accounted for the remaining variability in Q̇p:Q̇s.
Systemic ΔA-V O2
The mean ΔA-V O2 in the 115 samples was 30±11% (95% CI 8% to 52%). There was wide intrapatient and interpatient variability, with a range of 9% to 50%. Variation in Ssvco2 contributed most significantly to alterations in systemic ΔA-V O2. Consistent with the observations of previous investigators,3 our data indicate that variability in the systemic ΔA-V O2 confounds bedside assessment of Q̇p:Q̇s from Sao2 alone.
PV Desaturation Is Common in the Early Postoperative Period After Norwood Palliation
Measured Spvo2 ranged from 76% to 100%, and PV desaturation (defined as Spvo2 <95%) was observed in 35 of the samples (30%). Spvo2 was <90% in 17 of the 115 samples (15%). All but 1 patient had ≥1 desaturated PV sample, and 8 of 9 samples taken during administration of subambient oxygen were <95%. Although PV desaturation was observed more frequently at lower Fio2, it was also observed in 5 samples at Fio2 of ≥0.30 (Figure 2⇓). Desaturation of PV blood occurred without evidence of pulmonary edema, atelectasis, or infiltrates or catheter migration on repeated chest radiographs.
To verify that these were true PV samples and not samples contaminated with atrial blood, an effort was made to normalize Spvo2 in all patients who exhibited PV desaturation. By use of a combination of increased Fio2 and PEEP, Spvo2 was raised to ≥95% in all patients, and Ppvo2 was >200 mm Hg in all patients studied at an Fio2 of 0.50. The minimum and maximum Spvo2 values for each individual patient are displayed in the Table⇑, and the interventions used to normalize Spvo2 are noted.
Fio2 and Q̇p:Q̇s
As Fio2 was increased, Sao2 increased in all patients (Figure 3⇓, top). The corresponding PV blood, which was desaturated in 5 of 6 patients at 0.17 Fio2, became fully saturated as Fio2 was increased (Figure 3⇓, second panel). Somewhat surprisingly, no significant relationship between Fio2 and Q̇p:Q̇s was observed with systematic manipulation of Fio2 (Figure 3⇓, middle panel). As Fio2 was increased from 0.17 to 0.50, only 1 patient (▴ symbols) demonstrated a clinically important increase in Q̇p:Q̇s with apparent reduction in systemic O2 delivery (increased ΔA-V O2, decreased O2 extraction ratio). Although the systemic-pulmonary shunt in this patient (patient 1, Table⇑) was not excessively large, it is relevant that this patient had the highest Q̇p:Q̇s on subambient oxygen of all patients studied, suggesting excessive baseline pulmonary blood flow. In the remaining 5 patients, systematic increases in Fio2 failed to cause a corresponding change in Q̇p:Q̇s. In these patients, changes in Q̇p:Q̇s and ΔA-V O2 with Fio2 were minimal, but a reduced O2 extraction ratio in these patients suggested more luxuriant systemic O2 delivery (Figure 3⇓, lower 3 panels). Improvement in Spvo2 at higher Fio2 helped to enhance oxygen delivery, demonstrating the importance of fully saturated PV blood to help optimize Sao2 and systemic oxygen delivery. These data support the supposition that decreases in pulmonary vascular resistance may have minimal effect on Q̇p:Q̇s in patients in whom the aortopulmonary shunt limits pulmonary blood flow mechanically.
The fragile early postoperative course after Norwood palliation is exquisitely sensitive to the balance of systemic and pulmonary blood flow.1 2 6 7 Bedside estimation of Q̇p:Q̇s from Sao2 alone has been advocated on the basis of 2 untested assumptions: constant systemic ΔA-V O2 and normal Spvo2 (≥95%). This study demonstrates that Sao2 is a poor predictor of Q̇p:Q̇s after Norwood palliation; Sao2 alone predicted only 8% of the variability in Q̇p:Q̇s. Inaccuracy in Q̇p:Q̇s estimation was caused by variability in both Smvo2 and Spvo2, thus challenging the dual assumptions used by clinicians for rapid bedside assessment of Q̇p:Q̇s.
Although others have demonstrated variability in Smvo2 after Norwood palliation,3 8 this is the first study to demonstrate that PV desaturation is also common. PV desaturation was observed in 30% of PV samples even though there was no evidence of pulmonary edema, atelectasis, or infiltrates on chest radiograph. The data thus suggest that postoperative Norwood patients may have pulmonary gas exchange abnormalities resulting from cardiopulmonary bypass, circulatory arrest, and the extensive surgical procedure. Furthermore, ventilatory maneuvers used to manipulate pulmonary vascular resistance in these patients (ie, controlled hypoventilation, low Fio2) may predispose patients to microatelectasis and ventilation-perfusion mismatch.
The presence of unrecognized PV desaturation may contribute to substantial errors in Q̇p:Q̇s estimated from Sao2 alone. The data indicate that Q̇p:Q̇s is underestimated at higher levels of true Q̇p:Q̇s, when it is most clinically relevant (Figure 4A⇓). Our observations (Figure 4B⇓) provide clinical evidence to substantiate the computer modeling predictions of Barnea and colleagues.6 7 If PV blood is assumed to be fully saturated (≥95%), the error in Q̇p:Q̇s estimation can be substantial if true Spvo2 falls below 95%. Q̇p:Q̇s is underestimated by 20% to 25% when Spvo2 is 90%, and the underestimation error is as large as 30% to 70% if Spvo2 falls below 90%. The data also suggest that the frequency and degree of PV desaturation may be intensified by use of reduced levels of Fio2 (Figures 2⇑ and 3⇑), as commonly used in the perioperative management of single-ventricle patients. Reduced or subambient Fio2 can lower Sao2 by decreasing Spvo2 without a concomitant decrease in Q̇p:Q̇s. The resultant decrease in systemic oxygen delivery may not be recognized, because the patient is assumed to be “better” on the basis of the decrease in Sao2.
The observation that high Fio2 and/or a mild respiratory alkalosis after Norwood palliation may not result in significant hemodynamic instability is supported by the work of Mosca and colleagues.9 Our data support the conjecture that Q̇p:Q̇s may be insensitive to manipulations in pulmonary vascular resistance when the systemic-to-pulmonary shunt limits pulmonary blood flow mechanically. As Fio2 was systematically increased, we did not observe consistent changes in Q̇p:Q̇s, and the oxygen extraction ratio improved in 5 of 6 patients (Figure 3⇑). These findings support the judicious use of higher PEEP and/or Fio2 in select patients after Norwood palliation. In a similar manner, it may not be beneficial to aggressively induce respiratory acidosis in all postoperative Norwood patients, because controlled hypoventilation may predispose patients to microatelectasis and ventilation-perfusion mismatch. Because clinical recognition of PV desaturation is difficult in the absence of a pulmonary vein catheter, the consequences of any ventilatory maneuvers must be assessed cautiously by determining their effect on Sao2 and Smvo2 and global end-organ perfusion.
The conclusions of this study are contingent on accurate sampling of PV blood. To verify that atrial contamination of PV samples was minimal, an effort was made to normalize Spvo2 in all patients exhibiting PV desaturation. With a combination of increased Fio2 and PEEP, Spvo2 was raised to ≥96% in all patients, and the PV Po2 was >200 mm Hg in all patients studied at an Fio2 of 0.50 (Table⇑). The substantial improvements in PV Po2 with increases in Fio2 and/or PEEP indicate minimal atrial contamination of PV samples. Another limitation of the study is the assumption that SVC and PV samples reflect Smvo2 and mixed Spvo2, respectively. Although these assumptions are commonly used during catheterization of single-ventricle patients, they may be inaccurate. Finally, we would caution that the findings should not be generalized too broadly. These data represent only a small sample of postoperative Norwood patients, and the observations reported are specifically relevant to the early postoperative period. The data should be interpreted carefully and should not be applied broadly to all single-ventricle patients. Low-level or subambient inspired oxygen is presumably not detrimental for patients with preoperative single-ventricle physiology and healthy lungs in whom pulmonary blood flow is limited by pulmonary vascular resistance alone. We do not advocate use of high Fio2 in single-ventricle patients with obvious pulmonary overcirculation, especially those with preoperative single-ventricle physiology. Similarly, unnecessarily high PEEP in single-ventricle patients may reduce venous return and thus be detrimental to overall cardiac output.
In summary, this study demonstrates that systemic Sao2 alone is a poor predictor of Q̇p:Q̇s in the early postoperative period after Norwood palliation. Because inaccuracy in Q̇p:Q̇s estimation resulted from variability in both Ssvco2 and Spvo2, the data challenge the dual assumptions used by clinicians for rapid bedside assessment of Q̇p:Q̇s. The novel observation that PV desaturation is common in the early postoperative period after Norwood palliation raises the possibility that undetected pulmonary gas exchange defects may adversely affect systemic oxygen delivery in some patients after Norwood palliation. Because unrecognized PV desaturation confounds Q̇p:Q̇s assessment and compromises Sao2 and oxygen delivery, judicious use of inspired oxygen and PEEP in the early postoperative period after Norwood palliation may be beneficial in selected patients.
This study was supported by The Children’s Heart Association of Cincinnati.
- Received February 15, 2001.
- Revision received March 21, 2001.
- Accepted March 26, 2001.
- Copyright © 2001 by American Heart Association
Wernovsky G, Bove EL. Single ventricle lesions. In: Chang AC, Hanley FL, Wenovsky G, et al, eds. Pediatric Cardiac Intensive Care. Baltimore, Md: Williams & Wilkins; 1998:271–287.
Barnea O, Santamore WP, Rossi A, et al. Estimation of oxygen delivery in newborns with a univentricular circulation. Circulation. 1998;98:1407–1413.
Mosca RS, Bove EL, Crowley DC, et al. Hemodynamic characteristics of neonates following first stage palliation for hypoplastic left heartsyndrome. Circulation. 1995;92(suppl II):II-267–II-271.