Introduction

Non-invasive measurements of hemoglobin oxygen saturation (SaO2) by pulse oximetry (SpO2) have become routine in emergency settings, and SpO2 has been labeled the “fifth vital sign” [11]. Pulse oximetry is a measurement that reliably estimates SaO2 in arterial blood [6, 7, 9]. Single or serial measurements of SpO2 significantly influence decision-making in regard to workup, treatment, and hospitalization of patients. Many factors have been shown to influence SaO2 including: partial pressure of O2 and CO2 in blood, 2,3 DPG level, blood pH, and temperature [2]. Increased blood temperature decreases the affinity of oxygen to hemoglobin, so that increasing blood temperature leads to a decrease in SaO2.

To the best of our knowledge, the effect of pyrexia upon SpO2 has not been quantified in children. Theoretically, pyrexia may affect the oxygen-hemoglobin dissociation curve directly, but it also may affect SpO2 through indirect mechanisms. For instance, pyrexia is associated with tachypnea which decreases Pco2 causing a shift of the oxygen-hemoglobin dissociation curve and increasing SpO2. Additionally, pyrexia causes tachycardia and increased cardiac output which might shorten blood transit time in the lungs, leading to reduced oxygen uptake.

The aim of the current study was to evaluate the influence of pyrexia on SpO2 in a “real world” setting. For this purpose we measured SpO2 in febrile children presenting to the emergency room before and after antipyretic treatment. We compared our findings to those calculated from the oxygen dissociation curve.

Methods

Patients

Children up to the age of 12 years old, who presented to the emergency department of our institution with pyrexia, were included in the study. Inclusion criteria were body temperature at or above 38.5 °C and intention to use an antipyretic. Exclusion criteria were SpO2 below 90% and the presence of dynamic respiratory disease such as an asthma exacerbation, wheezing, bronchiolitis, or laryngitis. Patients with stable “respiratory disease” such as pneumonia without oxygen supplementation were included in the study.

Antipyretic treatment and vital signs measurements

Oral acetaminophen 15 mg/kg or oral ibuprofen 10 mg/kg was given depending upon the preference of the ER physician. Before administration, we measured baseline vital signs including heart and respiratory rate, temperature, and oxygen saturation. Temperature was measured using a digital thermometer (Welch Allyn SureTemp® Plus 690 with a disposable tip) either orally or rectally, ensuring that the same method was used for all measurements for each child. Approximately 90 min after administration of antipyretics, vital signs including SpO2 and temperature were again measured.

Pulse oximetry measurements

Pulse oximetry was measured using a Masimo Radical-7 motion resistant pulse oximeter (Masimo, Irvine, California, USA) incorporating Masimo SET v7.6.2.1 software. All measurements were conducted by the same investigator (S.H.). Measurements were conducted while the children were sleeping or awake and quiet (not feeding, fussy, or crying) before and after antipyretic therapy. Measurements were performed either on the palm of the hand or the foot (infants), or on a digit in older children. Both measurements for each child were performed at the same site. Using an averaging mode of 12 s, the measurements were conducted for a period of at least 90 s after a stable and sharp pulsatile pulse waveform appeared and the perfusion index (PI) was steadily > 1, all in accordance with the manufacturer’s recommendations. We used trend data for analysis collected with Masimo Trendcom software provided by the manufacturer. SpO2, pulse rate, and perfusion index values (including alerts of low SpO2 signal quality) were recorded at 2-s intervals. Trend data were omitted from analysis at least for the first 12 s of each recording and until 6 s of steady SpO2 recording was obtained. Any recordings that were missing a value or had a low SpO2 signal quality alert were also omitted from analysis. The average of all reliable SpO2 measurements was used.

Predicting the expected influence of blood temperature on SpO2

In order to calculate the expected influence of pyrexia on SpO2, we first calculated the expected alveolar Po2 (PAo2). For this purpose we used the alveolar gas equation: PAo2 = FiO2 (PB − P H20) − PAco2 [FiO2 + (1 − FiO2) / R] (where FiO2 is the fractional inspired oxygen concentration in air, PB is the atmospheric (barometric) pressure, P H20 is the partial pressure of water, and PAco2 is the partial CO2 alveolar pressure, while R is the respiratory quotient [1]). Assuming P H20 = 47 mmHg, PAco2 = 40 mmHg, and R = 0.8, the PAo2 in a healthy examinee breathing room air at sea level (760 mmHg) is expected to be 101.7 mmHg. The current study was done in Jerusalem, which is located at 800 m above sea level, where the barometric pressure is 697 mmHg giving an anticipated PAo2 of 88.5 mmHg. Assuming that the alveolar-arterial gradient is 10 mmHg, the anticipated arterial Po2 (PaO2) is 91.7 mmHg at sea level and 78.5 mmHg in Jerusalem. In order to calculate the SpO2 from PaO2, we used an Internet-based calculator called “The Interactive Oxyhemoglobin Dissociation Curve” [8]. The software requires PaO2, Pco2, pH, and temperature of the patient to calculate SpO2 and is based on the classic oxyhemoglobin dissociation curve equation developed by Kelman [3] and from data obtained by Severinghaus [10]. The calculations were all performed based upon the assumptions that Pco2 is 40 mmHg and pH is at 7.38. The expected changes in SpO2 were calculated separately for each patient according to the temperature measured before and after antipyretic treatment.

Statistical analysis

We used the Minitab 15 statistical software (Minitab, State College, Pennsylvania) for statistical analyses. We assessed the effect of fever reduction on SpO2 and other vital signs using paired t tests. A p value of < 0.05 was considered significant.

Results

Forty-five patients were recruited for the study, and 34 (64.7% males) completed the study protocol. The other 11 children did not complete the study protocol mostly because they were either crying or because they were not in the same state of arousal at the time of the second measurement. One patient was asleep during both measurements and all the others were awake at both measurements.

Twelve patients (35%) were diagnosed with suspected viral infection, nine (26%) with pneumonia, three (9%) with acute gastroenteritis, three (9%) with febrile seizures, and seven (21%) with other diagnoses including tonsillitis, acute otitis media, familial Mediterranean fever, and lymphadenitis.

Mean patients’ age was 37.5 ± 14.8 months, ranging 0.45–192 months, and mean weight was 14.2 ± 10.5 kg. Nineteen of the patients (55.9%) were treated with acetaminophen and the rest (44.1%) with ibuprofen.

Vital signs and oxygen saturation before and after administration of antipyretic agent are depicted in Table 1. In all participants, body temperature decreased, with a mean decrease of 1.71 ± 0.67 °C (p < 0.001). Both respiratory and heart rates decreased significantly. SpO2 increased significantly by 0.95 ± 1.76% (p = 0.0034). SpO2 increased in 73.5% (25/34) of the whole cohort and in 90% (18/20) of children in whom body temperature decreased by 1.5 °C or more. There was a significant correlation between the decrease in temperature and the increase in SpO2 (p = 0.031, R = 0.34, Fig. 1). The mean measured increase in SpO2 was only 56 ± 114% of that expected when using Kelman’s equation. Taking only patients (n = 20) in whom the temperature decreased by ≥ 1.5 °C, the measured increase in SpO2 was much closer to the anticipated increase (90 ± 105%, Table 2).

Table 1 Vital signs and oxygen saturation (mean ± SD) before and after administration of antipyretic medication
Full size table
Fig. 1
figure 1

The association between the change in body temperature and the change in SpO2 (p = 0.031, R = 0.34)

Full size image
Table 2 Comparison of the measured increase in SpO2 after antipyretic treatment to the predicted increase in SpO2. The comparison was performed on the whole cohort and then only on patients with a decrease in body temperature of 1.5 °C or above
Full size table

Discussion

In this study, we measured the SpO2 in febrile children in the emergency room and the influence of defervescence upon SpO2. We also compared our findings to expected changes based upon theoretical calculations using Kelman’s equation. We found that most, but not all patients, had an increase in SpO2 concomitant with the decrease in body temperature. Moreover, a larger increase in SpO2 after defervescence correlated with a greater reduction in temperature. Finally, in patients in whom the decrease in temperature was greater (1.5 degrees and above), the mean increase in SpO2 was very close to the calculated theoretical increase.

To the best of our knowledge, there is only one study in the English written literature quantifying the influence of temperature upon SpO2 in patients with fever [4]. This study involved adults hospitalized in an intensive care unit and included patients receiving ventilator support and supplemental oxygen. An increased median temperature from 38.1 to 39.0 °C was accompanied by a decrease in SpO2 from 98.0 to 97.6%. Since the change in temperature was small and some (numbers are not given) of their patients were on ventilatory support, their findings might not be applicable to children breathing room air and with higher temperature changes. A previous study quantifying the influence of temperature upon SpO2 on 22 children was published by our group in Hebrew [5]. We determined the level of SpO2 by documenting the values as appeared on the pulse oximeter screen and not by averaging recorded values as was done in the current study. The results were similar to the current study. The mean decrease in temperature was 2.03 °C and the average rise in SpO2 was 1.55 ± 1.79% (p = 0.001).

In the current study, by definition, baseline SpO2 was above 90% in all patients. We speculate that the influence of an increase in body temperature on SpO2 in patients with low baseline SpO2 may be greater than that found in our study. This is because the oxyhemoglobin dissociation curve is much steeper when Po2 is lower than 60 mmHg. Indeed, according to Kelman’s equation, when the Po2 is 60 mmHg with a blood temperature of 37 °C, the anticipated SpO2 is 91%. When the blood temperature is increased to 40 °C, the anticipated SpO2 decreases to 85.8%.

It is important to note that the calculation presented above is based upon the assumption that the values of PAco2, R value, P H20, barometric pressure, and the alveolar-arterial oxygen gradient are constant and equal to the values presented in the methods section. In practice, these values differ between patients and days. Moreover, the influence of compensatory mechanisms such as an increase in the respiratory rate is not known. Hence, the influence of an increased body temperature upon oxygen saturation will vary between patients and should be measured individually. We suggest that in a febrile patient, with borderline oxygen saturation, the decision to give supplemental O2, or to hospitalize, should be postponed until after the administration of antipyretics.

Study limitation

The main limitation of the study is the absence of hypoxemic patients. It is anticipated that in those patients the influence of temperature upon SpO2 would be greater. We did not include them in the current study because those patients usually receive supplemental oxygen. Another study incorporating hypoxemic patient is required.

Conclusion

Increased body temperature is associated with decreased SpO2 as suggested by the oxyhemoglobin dissociation curve. The relationship is not completely predictable. Physicians treating patients with fever should be aware of the potential significant influence of temperature on SpO2, especially in patients with borderline saturation. An increase in SpO2 in a patient after fever reduction should not be automatically interpreted as an improvement in the patients’ respiratory condition. Similarly, a low SpO2 in a patient with high fever does not always imply significant respiratory compromise.