Oxygen Saturation in Patients with Acute Sickle Chest Syndrome

Sickle cell anemia is a disorder of hemoglobin associated with an amino acid substitution of valine for glutamic acid at the sixth residue of the beta-chain. Patients with sickle cell anemia are prone to recurrent pain crises related to RBC sickling and vaso-occlusion with subsequent tissue hypoxia. Acute pain crises may be precipitated by hypoxemia, infection, dehydration, environmental temperature change, or acidosis. Supportive therapy includes IV fluid hydration, analgesic therapy, and antibiotic therapy when infection is suspected.

Acute sickle chest syndrome (ASCS) can result in significant morbidity and mortality ^[1] in patients suffering an acute pain crisis. It manifests as fever, chest pain, infiltrates on chest radiograph, and hypoxemia. The exact pathophysiologic state of ASCS has not been established, although a correlation with pulmonary infection has been suggested. Embolism of fat, ^[2] marrow, ^[3] or thrombus ^[4] has also been described. Microvascular occlusion of the circulation by sickled hemoglobin ^[5] and rib infarction ^[6] have been found to be associated with ASCS, but a cause-and-effect relationship is lacking.

Since hypoxemia is a trigger for sickling of hemoglobin S (HbS), its avoidance and correction are of major importance in the management of sickle cell disease, particularly in ASCS. Alveolar hypoxia has been shown to be associated with entrapment of sickle cells in the pulmonary microcirculation, ^[7] which may propagate a cycle of further hypoxemia and sickling. RBC exchange transfusion (EXC), although never decisively shown to have a beneficial impact on outcome, is frequently used to relieve ASCS. The exchange of HbS with hemoglobin A (HbA) typically results in prompt resolution of chest symptoms and hypoxemia in patients with ASCS.

Arterial hemoglobin oxygen saturation can be measured by either arterial blood gas analysis or pulse oximetry. Clinically, the most precise method of determining arterial blood saturation is via analysis with a co-oximeter (Co-ox). ^[8] The Co-ox measures light transmission through a blood sample at multiple wavelengths to distinguish oxyhemoglobin from reduced hemoglobin, carboxyhemoglobin, and methemoglobin. This method has been shown to be reliable for analyzing oxygen saturation in both HbA and HbS molecules. ^[9] ^[10] An alternative way of determining arterial hemoglobin oxygen saturation is to directly measure the partial pressure of oxygen in arterial blood and then to calculate the hemoglobin saturation (SaO₂ CALC) by plotting the PaO₂ on the oxygen dissociation curve (ODC). A third method is the use of pulse oximetry (SpO₂ ), which detects hemoglobin saturation on a continuous, noninvasive basis, by measuring transdermal absorption of light by hemoglobin in blood flowing through a fingertip or ear. In patients with normal hemoglobin, in the absence of carbon monoxide or methemoglobin toxicity, Co-ox, SaO₂ CALC, and Sp O₂ have been shown to have a very good correlation. ^[8] ^[11] Previous comparisons of various measurements of hemoglobin saturation during acute sickle cell crisis have been reported. ^[12] ^[13] ^[14] ^[15] Significant discrepancies may exist between these measurements, with some reports noting higher saturation with SaO₂ CALC and SpO₂ vs. Co-ox measurements. ^[13] Others have noted SpO₂ to underestimate SaO₂ CALC^[15] or have noted no overall difference. ^[12] To our knowledge, only one comparative study has looked at adult patients, ^[15] which found pulse oximetry to "underestimate oxygenation." Unfortunately, this study used calculated saturation rather than co-oximetry as the gold standard, independent variable. Overall, this limited literature suffers from a lack of uniform methodology (different age groups, different disease acuities [eg, sickle crisis vs. outpatients in stable condition], and different methods of determining hemoglobin saturation). Our anecdotal experience has been that many patients with ASCS have remarkable discrepancies between various hemoglobin oxygen saturation measurements. To our knowledge, a simultaneous comparison of the three modalities for measuring hemoglobin saturation in sickle cell anemia has not been published. ^[3] ^[16] We wondered whether our observed discrepancy between these measurements might be related to the presence of a high percentage of HbS, resulting in inaccurate results from the calculated hemoglobin saturation. If this were true, we speculated that such discrepancies would abate after EXC, since most of the HbS would be replaced by HbA. Therefore, we sought to answer the following questions: (1) During ASCS, what is the relationship of the three methods of measuring oxygen saturation to one another, and which is more accurate (SaO₂ CALC vs. SpO₂ ) relative to the "gold standard" (Co-ox)? (2) If there is a difference in these measurement methods related to HbS, is it eliminated by EXC?

Materials and Methods

All adult patients with a diagnosis of ASCS undergoing EXC were eligible for study. The study was approved by the Institutional Review Board at the University of Chicago. The diagnosis of ASCS was made on the following clinical grounds: pleuritic chest pain, dyspnea, fever, and infiltrates on chest radiograph. Demographic data (age, sex, weight, history of previous EXC, mortality) were collected for all patients.

Room air arterial blood was sampled from each patient just prior to EXC. All arterial blood samples were drawn as part of routine patient care as directed by the primary care team. Samples were drawn anaerobically into preheparinized 1-mL syringes (A-LINER; Sherwood Medical; St. Louis, MO). All air bubbles were removed from the syringes and each sample was then taken immediately for analysis. Just before each measurement, the sample was mixed well. All analyses were completed within 5 min of sampling time. Each sample was analyzed for oxygen saturation by calculation (SaO₂ CALC) (Radiometer ABL 500 ABG Analyzer; Radiometer; Copenhagen, Denmark) as well as by co-oximetry (Co-ox) (Radiometer ABL 520 ABG Analyzer; Radiometer). The same blood sample was analyzed twice by each method to assure reproducibility.

Sp O₂ was obtained at the same time as arterial blood sampling. Patients were simultaneously evaluated with three different pulse oximeters to assure reproducibility: Nellcor N-180 (Nellcor; Carlsbad, CA), Ohmeda Biox 3740 (Spacelabs; Redford, WA), and Spacelabs Module 90467 (Ohmeda; Louisville, CO). Sp O₂ values were recorded only after a consistent reading, with a strong arterial waveform signal and a pulse reading identical to the patient's heart rate. All pulse oximeters utilized finger sensors.

Baseline hemoglobin electrophoresis was performed prior to EXC. Each patient then underwent EXC as directed by the primary care team and the blood bank consult team. This

TABLE 1 -- Demographics^*

Characteristic	No.
Age, yr	34.3 ± 13.6
Sex, M/F	2/11
Weight, kg	60.0 ± 12.9
Intubated	2/13 (15.4%)
Previous EXC	12/13 (92.3%)
Mortality	1/13 (7.7%)

*Age and weight values are listed as mean ± SD.

typically involves exchange of an average of 6 U of packed RBCs over approximately 1 to 2 h. At steady state, 48 to 72 h after EXC, arterial blood gas analysis and Sp O₂ measurements were made in a manner identical to the preexchange regimen. A postexchange hemoglobin electrophoresis was also performed in all patients.

Comparison of demographic data was made with an unpaired t test. Hemoglobin saturation data were analyzed using the Bland and Altman ^[17] bias analysis for methods comparison studies and are expressed as mean ± SD. Ninety-five percent confidence intervals for the bias of Co-ox minus Sa O₂ CALC and Co-ox minus SpO₂ were determined.

Results

Thirteen different patients with ASCS were enrolled: 11 patients had SS disease and 2 had SC disease. Table 1 outlines the demographic breakdown.

There were no differences among the three pulse oximeters either before or after EXC (Table 2) . Data are reported using Nellcor-180 pulse oximeter results, but there were no significant differences with the others.

Mean percent HbS was 76.7 ± 14.7% at baseline and 27.9 ± 6.2% after EXC.

At baseline, oxygen saturation measurements ranged from 69.8 to 97.4% as measured by Co-ox. Baseline SaO₂ CALC was higher than Co-ox. At baseline, the bias (mean difference of Co-ox minus SaO₂ CALC) between Co-ox and SaO₂ CALC was -6.78 ± 2.63% (95% confidence interval, -8.37 to -5.19%) (Fig 1) . The baseline bias between Co-ox and SpO₂ was +1.86 ± 3.25% (95% confidence interval, -0.10 to 3.82%).

TABLE 2 -- Pulse Oximeter Data

Pulse Oximeter	Mean Sp O₂ , %
Baseline ^*
Sp O₂ Nellcor	83.7
Sp O₂ Ohmeda	84.0
Sp O₂ Spacelabs	82.8
Postexchange
Sp O₂ Nellcor	92.5
Sp O₂ Ohmeda	92.0
Sp O₂ Spacelabs	91.5

*p = 0.92 by one-way analysis of variance.
p = 0.95 by one-way analysis of variance.

After EXC, oxygen saturation measurements (Co-ox) ranged from 85.3 to 97.4%. SaO₂ CALC was not significantly different than Co-ox. Indeed, after EXC, the bias between Co-ox and SaO₂ CALC was -0.17 ± 1.31% (95% confidence interval, -0.95% to 0.61%) (Fig 2) . The post-EXC bias between Co-ox and SpO₂ was +0.30 ± 1.53% (95% confidence interval, -0.62 to 1.22%).

Four patients remained hypoxemic (Co-ox <90%, ie, on the steep portion of the ODC) even after exchange transfusion. In this subgroup, the baseline bias between Co-ox and SaO₂ CALC was -7.40 ± 2.46% (95% confidence interval, -11.31 to -3.49%), while the baseline bias between Co-ox and SpO₂ was +1.00 ± 1.15% (95% confidence interval, -0.84% to 2.84%). The post-EXC bias between Co-ox and SaO₂ CALC in this subgroup was -0.30 ± 2.14% (95% confidence interval, -3.70 to 3.10%), while the post-EXC bias between Co-ox and SpO₂ was +0.33 ± 0.78% (95% confidence interval, -0.91 to 1.57%).

Text Box:
Figure 1. Co-ox vs. SaO2 CALC--preexchange.

Discussion

We have demonstrated a significant discrepancy between arterial hemoglobin oxygen saturation as determined by a calculation algorithm vs. direct measurement by co-oximetry. The calculated method greatly overestimates true hemoglobin saturation during ASCS. Co-oximetry is the most accurate, clinically available test used to determine hemoglobin oxygen saturation ^[8] and has been widely used to for studying both HbA and HbS. ^[9] ^[10] The Radiometer ABL 520 co-oximeter measures light transmission through the blood sample at six different wavelengths (535, 560, 577, 622, 636, and 670 nm). Oxyhemoglobin, deoxyhemoglobin, methemoglobin, and carboxyhemoglobin each have different absorbances at these six wavelengths. ^[18] Many blood gas analyzers do not have co-oximeters, relying instead on measurement of PaO₂ and calculation of arterial saturation. This calculated method follows an algorithm to determine the ODC based on blood pH and temperature. Inherent in this algorithm is the expectation that oxygen binding of the specimen is that of HbA. The ODC of HbS is significantly different from that of HbA, with a considerable shift to the right. ^[19]

SpO₂ utilizes the principles of spectrophotometry and plethysmography. An electro-optical sensor with two light-emitting diodes (LEDs) and one photodiode as a photodetector is used. One LED emits red light at 660 nm, while the other emits infrared light at 920 nm. These LEDs pass transdermally (through a digit or ear) where part is absorbed and part transmitted. A photodetector measures transdermal red and infrared absorption. Since oxyhemoglobin and deoxyhemoglobin differ in their relative red and infrared absorption, arterial oxygen saturation can be determined. Pulse oximeters have been shown to be inaccurate in the presence of abnormal hemoglobin species such as methemoglobin and carboxyhemoglobin. Circumstances that reduce finger pulsation amplitude such as hypotension, hypothermia, or administration of vasoconstricting drugs also effect pulse oximeter accuracy. ^[20] All of our patients had strong arterial waveforms on the pulse oximeter with pulse readings consistent with heart rate. Pulse oximeters use a plethysmographic waveform analysis that is used to distinguish the pulsatile (arterial) signal from the nonpulsatile (venous) signal. Since, under most circumstances, the differences among these three methods of determining oxygen saturation are minimal, ^[8] such differences may not be Text Box:
Figure 2. Co-ox vs. SaO2 CALC--postexchange.
appreciated readily.

Our patients all suffered from ASCS, a complication of sickle cell anemia that has a high morbidity. Assuring that oxygen saturation is beyond the steep portion of the ODC (ie, > 90%) is critical in the management of sickle cell crisis--particularly during ASCS. Our data suggest that blood gas analysis using the calculation method may lead to inappropriate decisions regarding the treatment of these patients. In the 13 patients that we studied, 7 had a baseline SaO₂ CALC > 90% while simultaneous Co-ox was < 90%. This may lead the clinician to a false sense of security and inadequate oxygen therapy to manage hypoxemia. However, SpO₂ was comparable to co-oximetry, with a difference of +1.86% when comparing the baseline mean values. Clinicians often use pulse oximetry as a screen, with arterial blood gas analysis, when clinically indicated, as a more definitive, follow-up test. SpO₂ clearly has limitations, particularly during states of systemic hypoperfusion or in patients with severe peripheral vascular disease, as noted above. ^[20] ^[21] ^[22] Such limitations are often the impetus for ordering a follow-up arterial blood gas analysis. Our study shows that in the setting of ASCS, such "follow-up" arterial blood gases--if done with calculated oxyhemoglobin saturation--may have led to inappropriate action in at least 7 of our 13 patients. Our data show that, in patients with ASCS, SpO₂ is a more accurate measure of oxyhemoglobin saturation than is SaO₂ CALC.

Following EXC, the discrepancy between Co-ox and SaO₂ CALC abated. This correlated with the decrease in HbS from a mean of 76.7 to 27.9%. Since most circulating hemoglobin molecules are changed from HbS to HbA, it is reasonable to expect that the calculation algorithm will more closely approximate the true saturation. A possible alternative explanation to our findings is that EXC led to an improvement in hypoxemia (which it clearly did) and that the improvement in PaO₂ moved saturations to the flat portion of the ODC where differences in saturation measurement methods are more difficult to detect. To address this concern, we looked at 4 of 13 patients who remained hypoxemic (Co-ox < 90%, ie, on the steep portion of the ODC) even after exchange transfusion. The mean pre-EXC saturation values in these four patients were 88.7% (SaO₂ CALC), 81.3% (Co-ox), and 80.3% (SpO₂ ). The baseline bias between Co-ox and SaO₂ CALC in this subgroup was large (-7.40 ± 2.46% [95% confidence interval, -11.31 to -3.49%]). This bias also abated after EXC (+0.30 ± 2.14% [95% confidence interval, -3.70 to 3.10%]). Mean post-EXC saturation values were 87.9% (SaO₂ CALC), 87.6% (Co-ox), and 87.3% (SpO₂). Confidence intervals are somewhat wider here because of the small number of observations; nevertheless, it appears that the discrepancy among the three methods of determining saturation abates after HbS is replaced by HbA, independent of position on the ODC.

In conclusion, we have determined the following: (1) SaO₂ CALC overestimates Co-ox during ASCS; (2) the discrepancy between SaO₂ CALC and Co-ox abates after exchange transfusion; and (3) SpO₂ more closely follows Co-ox than does SaO₂ CALC during ASCS.

References

1. Platt OS, Brambilla DJ, Rosse WF, et al. Mortality in sickle cell disease: life expectancy and risk factors for early death. N Engl J Med 1994; 330:1639-1644

2. Vichinsky E, Williams R, Das M, et al. Pulmonary fat embolism: a distinct cause of severe acute chest syndrome in sickle cell anemia. Blood 1994; 83:3107-3112

3. Charache S, Page DL. Infarction of bone marrow in the sickle cell disorders. Ann Intern Med 1967; 67:1195-1200

4. Walker BK, Ballas SK, Burka ER. The diagnosis of pulmonary thromboembolism in sickle cell disease. Am J Hematol 1979; 7:219-232

5. Bhalla M, Abboud MR, McLoud TC, et al. Acute chest syndrome in sickle cell disease: CT evidence of microvascular occlusion. Radiology 1993; 187:45-49

6. Rucknagel DL, Kalinyak KA, Gelfand MJ. Rib infarcts and acute chest syndrome in sickle cell diseases. Lancet 1991; 337:831-833

7. Aldrich TK, Dhuper SK, Patwa NS, et al. Pulmonary entrapment of sickle cells: the role of regional alveolar hypoxia. J Appl Physiol 1996; 80:531-539

8. Forte VA, Malconian MK, Burse RL, et al. Operation Everest II: comparison of four instruments for measuring blood O₂ saturation. J Appl Physiol 1989; 67:2135-2140

9. DeYoung A, Noble RW. Oxygen binding to sickle cell hemoglobin. Methods Enzymol 1981; 76:792-805

10. Sunshine HR, Hofrichter J, Ferrone FA, et al. Oxygen binding by sickle cell hemoglobin polymers. J Mol Biol 1982; 158:251-273

11. Nickerson BG, Sarkisian C, Tremper K. Bias and precision of pulse oximeters and arterial oximeters. Chest 1988; 93:515-517

12. Pianosi TD, Charge TD, Esseltine DW, et al. Pulse oximetry in sickle cell disease. Arch Dis Child 1993; 68:735-738

13. Craft JA, Alessandrini E, Kenney LB, et al. Comparison of oxygenation measurements in pediatric patients during sickle cell crises. J Pediatr 1994; 124:93-95

14. Rackoff WR, Kunkel N, Silber JH, et al. Pulse oximetry and factors associated with hemoglobin oxygen desaturation in children with sickle cell disease. Blood 1993; 81:3422-3427

15. Comber JT, Lopez BL. Evaluation of pulse oximetry in sickle cell anemia patients presenting to the emergency department in acute vasoocclusive crisis. Am J Emerg Med 1996; 14:16-18

16. Jones SW. Pulse oximetry in sickle cell disease [letter]. Arch Dis Child 1994; 70:71

17. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-310

18. ABL520 user's handbook (version 3.0). Copenhagen, Denmark: April 1992

19. Weston Smith SG, Glass UH, Acharya J, et al. Pulse oximetry in sickle cell disease. Clin Lab Haematol 1989; 11:185-188

20. Barker SJ, Tremper KK. Pulse oximetry: applications and limitations. Int Anesthesiol Clin 1987; 25:155-175

21. Severinghaus JW, Spellman MJ. Pulse oximeter failure thresholds in hypotension and vasoconstriction. Anesthesiology 1990; 73:532-537

22. Vegfors M, Lindberg LG, Lennmarken C. The influence of changes in blood flow on the accuracy of pulse oximetry in humans. Acta Anaesthesiol Scand 1992; 36:346-349

Characteristic