The Journal of Maternal-Fetal & Neonatal Medicine

ISSN: 1476-7058 (Print) 1476-4954 (Online) Journal homepage: https://www.tandfonline.com/loi/ijmf20

Accuracy of oxygen saturation index in

determining the severity of respiratory failure
among preterm infants with respiratory distress
syndrome

Nasrin Khalesi, Farhad Abolhasan Choobdar, Mousa Khorasani, Fatemeh

Sarvi, Behzad Haghighi Aski & Mahmoud Khodadost

To cite this article: Nasrin Khalesi, Farhad Abolhasan Choobdar, Mousa Khorasani,
Fatemeh Sarvi, Behzad Haghighi Aski & Mahmoud Khodadost (2019): Accuracy of oxygen
saturation index in determining the severity of respiratory failure among preterm infants with
respiratory distress syndrome, The Journal of Maternal-Fetal & Neonatal Medicine, DOI:
10.1080/14767058.2019.1666363

To link to this article: https://doi.org/10.1080/14767058.2019.1666363

Published online: 19 Sep 2019.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ijmf20
THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE
https://doi.org/10.1080/14767058.2019.1666363

ORIGINAL ARTICLE

Accuracy of oxygen saturation index in determining the severity of

respiratory failure among preterm infants with respiratory
distress syndrome
Nasrin Khalesia, Farhad Abolhasan Choobdara, Mousa Khorasanib, Fatemeh Sarvic,d, Behzad Haghighi Askia
and Mahmoud Khodadoste
a
Department of Pediatrics, Ali Asghar Hospital, Iran University of Medical Sciences, Tehran, Iran; bDepartment of Pediatrics, Ali Asghar
Children Hospital, Tehran, Iran; cLarestan University of Medical Sciences, Larestan, Iran; dDepartment of Biostatistics & Epidemiology,
School of Public Health, Hamadan University of Medical Sciences, Hamadan, Iran; eDepartment of epidemiology, School of Public
Health, Shahid Beheshti University of Medical Sciences, Tehran, Iran

ABSTRACT ARTICLE HISTORY

Background: To evaluate the severity of respiratory failure among newborns with respiratory Received 10 May 2019
distress syndrome (RDS), oxygenation index (OI) has been implemented. In the present study, Revised 4 August 2019
we assessed the accuracy of oxygen saturation index (OSI) in determining the severity of respira- Accepted 8 September 2019
tory failure.
KEYWORDS
Methods: A cross-sectional study was carried out in the NICUs of two Iranian Hospitals (Tehran, Accuracy; oxygen saturation
Iran) in 2018. Preterm neonates with RDS entered the study. Immediately after admission, the index; oxygenation index;
severity of RDS was determined based on RDS scoring system. Then, 2 CC of arterial blood was respiratory distress
withdrawn and sent to laboratory determining blood gases. Simultaneously, the level of periph- syndrome; ROC curve
eral capillary oxygen saturation (SpO2) was read using pulse oximeter and recorded. OI and OSI
were measured using the formulae. Receiver Operating Characteristic curve, Kappa agreement
coefficient and accuracy, sensitivity and specificity was used to compare the OI and OSI results.
Results: In the study, 95 neonates were considered. Based on ROC curves, the appropriate cut
off with AUC ¼ 0.99 for severe respiratory failure was OSI >8. The sensitivity, specificity, negative
predicted value, and positive predicted value for the OSI Cut off >8 were 100, 98, 0.97 and
100%, respectively. The overall accuracy and Kappa agreement between OSI and OI was 0.96
and 0.98%, respectively.
Conclusion: Our results showed that OSI with high sensitivity, specificity values could predict
the severity of respiratory failure in preterm neonates with RDS.

Introduction supporting cares and medications have been imple-

Respiratory distress syndrome (RDS) is one of the
major causes of neonatal morbidity and mortality. It
represents in 1% of all births; however, this range rises
to 50 and 90% in newborns with gestational age 30
and 28 weeks [1–3]. Clinical manifestations of RDS
include tachypnea, grunting, chest wall retractions,
nasal flaring, cyanosis, and hypoxia leading to respira-
tory failure. Although neonatal RDS could be diag-
nosed by biochemical (lack of surfactant) and
pathological findings, the diagnosis is established by
clinical signs and chest radiologic features [4–6]. The
chest X-ray exam in neonates with RDS represents a
low lung volume with fine diffuse granular or ground-
glass appearance, air bronchograms and obscure heart
borders [7–9]. In recent decades, considerable
mented to improve the respiratory outcome; prenatal
steroid prophylaxis, early nasal continuous positive air-
way pressure, intubation, surfactant administration
and extubation (INSURE), as well as invasive ventilation
are some of them [10–12].
To evaluate the status of respiratory failure among
NICU hospitalized newborns with RDS, we can use
oxygenation index (OI) [13,14]. OI as a respiratory
index shows the severity of oxygenation failure of the
newborns under mechanical ventilation. Usage of OI
to monitor neonate’s respiratory status could be crit-
ical in dictating management processes such as the
initiation time of surfactant and nitric oxide adminis-
trations. By using an equation [15,16], OI can be
calculated; however, there are some disadvantages

CONTACT Mahmoud Khodadost Mahmodkhodadost@yahoo.com Department of Epidemiology, School of Public Health, Shahid Beheshti
University of Medical Sciences, Tehran, Iran
ß 2019 Informa UK Limited, trading as Taylor & Francis Group
2 N. KHALESI ET AL.
regarding to determination of PaO2. The procedure
requiring indwelling arterial line or arterial puncture is
painful and invasive. The blood of sampling site may
not show the precise levels of PaO2 perfusing other
organs like the heart and brain. In addition, some
complications related to arterial puncture like infec-
tion, iatrogenic anemia and need for blood transfusion
have been noted. Finally, this index could not be con-
tinuously used only if arterial blood gas test is
done [8,15–18].
Based on recent publications [16,17,19], oxygen sat-
uration index (OSI) may also assess the severity of
respiratory failure. For the calculation of OSI, no inva-
sive procedure is needed. Using pulse oximeter deter-
mining OSI and SpO2 can provide a continuous
monitoring of infant’s respiratory status [16,17]. In the
present study, we aimed to evaluate the correlation
between OI and OSI Cut offs corresponding severe
and nonsevere respiratory failure among preterm
infants with RDS. In addition, the accuracy, sensitivity,
specificity, positive and negative predictive values
related to OSI in the prediction of severity of respira-
tory failure among subjects were assessed.
Materials and methods
Design and patients
A cross-sectional study was carried out in the NICUs of
Akbar-Abadi and Ali-Asghar Hospitals affiliated to Iran
University of Medical Sciences (Tehran, Iran) in 2018.
Ninety-five hospitalized preterm neonates due to
respiratory distress syndrome (RDS) entered consecu-
tively within 6 months to the study.
Inclusion criteria were preterm birth (gestational
age <37 weeks) and NICU hospitalization due to RDS
(based on clinical manifestations and chest radio-
graphic findings). All subjects with required intubation
continued mechanical ventilation and surfactant
administration were enrolled in this study.
Admitted neonates with respiratory distress related
to other causes including asphyxia, congenital heart
disease, chest or airways malformation, congenital cys-
tic lung disease, pulmonary air-leak syndromes, meta-
bolic disease and acquired lung diseases were
excluded from the study.
Data sources and measurement
Detailed neonates’ demographic and prenatal data
gathered from medical records. Laboratory findings
and data related pulse oximeters were recorded in
some checklists. Immediately after NICU admission,
the severity of RDS was determined based on RDS
scoring system and recorded. RDS score system was
implemented for evaluation of severity of respiratory
distress with six parameters such as respiratory rate
and grunting [7].
Then, 2 CC of arterial blood was withdrawn from,
labeled and sent to laboratory determining blood
gases. Simultaneously with blood sampling, the level
of SPO2 (peripheral capillary oxygen saturation) was
read using pulse oximeter monitor (Masimo set,
Alborz B9, Saadat CO-Iran) and recorded. The sat O2
measured by pulse oximetry of right hand varies to
other extremities. The medium value of sat O2 was
measured from right hand after stabilization of
infant condition.
Oxygenation index (OI ¼ MAP  FiO2  100/PaO2)
and oxygen saturation index (OSI ¼ MAP  FiO2  100/
SpO2) were measured using the formulae where the
MAP is the mean airway pressure refers to the mean
pressure applied during positive-pressure mechanical
ventilation. The severity of respiratory failure based on
both oxygenation index as a gold standard [8] and
oxygen saturation index was determined. Finally, the
correlation between OI and OSI was assessed. The lev-
els of accuracy, sensitivity, specificity, positive and
negative predictive values related to OSI
were calculated.
Sample size
Based on former investigations by Rawat et al. [16].
and using a power of 95% and an alpha error of 0.05,
the proposed sample size was considered 95 subjects
Primary and secondary outcomes
Our primary objective was determining the accuracy
of OSI in prediction of severity of respiratory failure
among preterm hospitalized infants with RDS. The sec-
ondary objectives were assessment of the levels of
sensitivity, specificity, positive and negative predictive
values related to OSI for severity of RDS.
Ethical considerations
The present study was taken from a medical student
thesis with ID; IR.IUMS.FMD.REC 1397.195. Ethics
approval was obtained from the institutional review
board of Iran University of Medical Sciences according
to Declaration of Helsinki. All participants’ parents
gave written consent before enrollment. All gathered
 THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE 3

data were considered confidential and no extra cost as diagnostic value for severity of respiratory failure;
was imposed on our participants. p  .0001, 95% CI: (0.958–1.000).
The cutoff values, specificity, sensitivity, positive-
and negative-predictive values (P&N PV) of OSI for the
Statistical analysis
assessment of severity of respiratory failure are shown
Quantitative and qualitative variables were reported in Table 3. Our results have shown that cutoff 8 for
by mean þ SD and percent, respectively. Independents OSI was the best diagnostic value for the severity of
samples t-test was used for comparing quantitative respiratory failure. Sensitivity and specificity with the
data. Chi square test was used to analyze the differen- cutoff 8 were 100 and 98%.
ces between qualitative variables. Receiver Operating Based on mentioned formula (Equation (1)):
Characteristic (ROC) curves and area under the curve
OI ¼ 2:12  OSI2:19 þ ei (1)
(AUC) was used to determine the proper cutoff of OSI
index for determining the severity of RDS. The accur- oxygenation index was obtained about 15
acy, sensitivity, specificity, negative predicted value (Accuracy ¼ 0.98, Kappa; measure of
and positive predictive value of oxygen saturation agreement ¼ 0.96) (Table 4). It means that OSI
index were calculated using the results of oxygenation cutoff ¼ 8 equals to OI cutoff ¼ 15 and in 96% of
index as a gold standard for determining the severity cases, there are an agreement between OI and OSI in
of RDS. The Kappa agreement coefficient was used to assessment of the severity of respiratory failure.
evaluate the consistency between the OSI and OI for Scatter plot shows a linear correlation between OI and
determining the severity of respiratory failure. Data OSI (Figure 2). Also, the result of linear regression
was analyzed using Stata software, version 12
(StataCorp, TX). The p < .05 was considered as the Table 2. Associations between oxygenation and oxygen sat-
level of significance in all analysis. uration indices with gestational age or sex.
Variables Coefficient Standard error p-Value
OIa
Results Sex 1.15 0.91 .21
Gestational age 0.39 0.36 .28
In the study, 95 neonates were considered. Of all, 63 OSIb
subjects (66.3%) were male with the mean gestational Sex 0.5 0.41 .18
Gestational age 0.27 0.16 .09
age 32.12 ± 2.88 weeks, OI; 16.1 ± 10.08 weeks, and a
Oxygenation index.
OSI; 8.69 ± 4.73 weeks. These parameters in 23 female b
Oxygen saturation index.
cases (33.7%) were 31.68 ± 3.18 weeks,
15.24 ± 11.14 weeks, and 8.1 ± 4.8 weeks, respectively.
Results showed that there were no significant differen-
ces between male and female neonates regarding to
gestational age (p ¼ .5) T OI (p ¼ .71) and the OSI
(p ¼ .56) (Table 1).
In multiple regression analysis, with increasing one
unit in gestational age, the level of oxygenation index
and oxygen saturation index were increased by 0.39
and 0.27, respectively, however this association was
not statistically significant (Table 2).
ROC curves, calculating sensitivity and specificity for
OSI is shown in Figure 1; based on ROC curves, AUC
for OSI was 0.999. This level was statistically significant

Table 1. Comparison of the mean of gestational age, OI and

OSI in male and female subjects.
Male Female
Variables N ¼ 63 N ¼ 32 p-Value
Gestational age; Mean (SD) 32.12 (2.88) 31.68 (3.18) .5
OI; Mean (SD) 16.1 (10.08) 15.24 (11.14) .71
OSI; Mean (SD) 8.69 (4.7) 8.1 (4.8) .56
Oxygenation index.
Oxygen saturation index. Figure 1. ROC curve for severity of respiratory failure among
population study.
4 N. KHALESI ET AL.

analysis shows that there is a significant correlation Discussion

between PaO2 and SatO2 measured by pulse-oxymetry
Respiratory distress syndrome is one of the main
(r ¼ 0.54, p < .001, b ¼ 1.63, p < .001) (Figure 3). For
causes of neonatal morbidity and mortality. Early diag-
determining the severity of respiratory failure based
nosis of RDS and its correlated complications like
on OI; Cutoff 15 and Cut off < 15 represented
respiratory failure would be of value in implementing
severe and nonsevere respiratory failure. These values
some strategies to improve neonatal outcomes [20].
for OSI were Cut off 8 and Cut off < 8.
According to our results, use of oxygen saturation
index (OSI) may also assess the severity of respiratory
Table 3. Sensitivity, specificity, negative- and positivepredic- failure with lower disadvantages in comparison with
tive value of OSI for cut off > 8 determining severity of OI. This index could be easily calculated by SpO2
respiratory failure among population study.
obtained by pulse oximetry. SpO2 with its linear correl-
OSIa
ation to partial pressure of oxygen in the middle por-
Severe Nonsevere Sensitivity Specificity NPVb PPVc
d
tion of oxygen dissociation curve may provide more
OI
Severe 45 0 100% 98% 97.7 100 advantages for OSI. To the best of our knowledge,
Nonsevere 2 48 present study is among the very few studies that
a
Oxygen saturation index. searched for predicting value of OSI in determination
b
Negative predictive value.
c
Positive predictive value. of the severity of respiratory failure. The main
d
Oxygenation index. strengths of this study was the larger sample size in

Table 4. Accuracy and Kappa coefficient for OI.

Variables Nonsevere respiratory failure (n %) Severe respiratory failure (n %) Kappa statistic Accuracy
OIa (Cut off point ¼ 15) 50 (52.6) 45 (47.4) 0.96 0.98
OSIb (Cut off point ¼ 8) 48 (50.5) 47 (49.5)
a
Oxygenation index.
b
Oxygen saturation index. Cutoff point ¼ 15 and OSI; Cutoff point ¼ 8 determining severe and nonsevere respiratory failure.

Figure 2. Scatter plot showing correlation between oxygen saturation index (OSI) and oxygenation index (OI).

Figure 3. Scatterplot and the linear regression line for correlation between PaO2 and SatO2.
comparison with other investigations [16,17,21]. In
accordance to our findings, Rawat et al. and
Doreswamy et al. pointed to positive association
between OI and OSI in determination of the severity
of respiratory failure in neonates [16,17,19,22].
The results of present study demonstrated that
based on AUC for OSI; 0.999, OSI can be used as a
noninvasive assessment tool for determination of the
severity of respiratory failure. The best cutoff for OSI
as a diagnostic value for severe respiratory failure was
8 corresponded to OI cutoff 15. Our results have
revealed a linear correlation with measure of
agreement ¼ 0.96 between OI and OSI. Moreover, the
sensitivity and specificity for OSI with the cutoff 8
were high (100 and 98%, respectively). Other studies
have also indicated this positive association between
OI and OSI and near cutoffs points (in predicting the
severe respiratory failure) to our findings; Rawat et al.
revealed that for OSI cutoff 7.5 corresponded to OI
 15, sensitivity and specificity were 83 and 100%. In
addition, for OSI cutoff 10 corresponded to OI  20,
the sensitivity increased to 100% [16]. Doreswamy
et al. showed that for OSI cutoff > 6.5 corresponded
to OI >15, sensitivity and specificity were 100
and 93.7%.
THE JOURNAL OF MATERNAL-FETAL & NEONATAL MEDICINE 5
Moreover, according to the results, high positive
and negative predictive values (100 and 97.7%) related
OSI cutoff >8 may provide a reliable clinical index in
implementing of OSI for the prediction of the severity
of respiratory failure. Rawat et al. have also demon-
strated the highest positive (100%) and negative
(99%) predictive values for OSI 7.5 compared to OSI
cutoff points >5 or >10 [16]. The use of OSI technique
instead of OI have several benefits, for example, it is a
noninvasive method that let us continuous monitoring
instead of intermittent evaluation from blood gas
measurement. Also, this technique does not require
blood sampling thus reduce risks of infection
and anemia.
The main limitation of the present study was
related to pulse oximeter that the alterations in O2
saturation levels in PaO2 more than 60 mmHg are
small. Another limitation is about the Location of the
pulse oximeter probe (preductal: right hand or post-
ductal: other extremities). Also, alteration of relation-
ship between SpO2 and PaO2 by the type of
hemoglobin, PH, temperature, etc. (oxygen dissoci-
ation curve curve) may be another limitation of this
technique. These limitations could influence on our
compared indices like OI.
6 N. KHALESI ET AL.
Conclusions
The early diagnosis of RDS complications certainly
enhances therapeutic efforts leading better outcomes.
Our results showed that oxygen saturation index with
high sensitivity, specificity values and linear correlation
with oxygenation index, could also predict the severity
of respiratory failure in neonates hospitalized due to
RDS. Further studies with larger sample size are
strongly suggested.
Author contributions
Choobdar and Khorasani carried out the design and
coordinated the study, participated in most of the
experiments. Dr Khodadost and Haghighi Aski coordi-
nated and carried out all the experiments, Analysis
of data.
Acknowledgments
The authors would like to thank Ali Asghar Clinical Research
Development Center (AACRDC), for Editorial/Statistical/
Search Assistance.
Disclosure statement
No potential conflict of interest was reported by the authors.
References
[1] Donda K, Vijayakanthi N, Dapaah-Siakwan F, et al.
Trends in epidemiology and outcomes of respiratory
distress syndrome in the United States. Pediatr
Pulmonol. 2019;54(4):405–414.
[2] Thomas NJ, Shaffer ML, Willson DF, et al. Defining
acute lung disease in children with the oxygenation
saturation index. Pediatr Crit Care Med. 2010;11(1):
12–17.
[3] Donn SM, Sinha SK. Respiratory distress syndrome.
Manual of neonatal respiratory care. Switzerland:
Springer; 2012. p. 523–532.
[4] Berger MM. Chest radiography for diagnosing acute
respiratory distress syndrome – fishing in the dark?
Crit Care Med. 2018;46(5):820–821.
[5] Bass CM, Sajed DR, Adedipe AA, et al. Pulmonary
ultrasound and pulse oximetry versus chest radiog-
raphy and arterial blood gas analysis for the diagnosis
of acute respiratory distress syndrome: a pilot study.
Crit Care. 2015;19(1):282.
[6] Peng JM, Qian CY, Yu XY, et al. Does training improve
diagnostic accuracy and inter-rater agreement in
applying the Berlin radiographic definition of acute
respiratory distress syndrome? A multicenter pro-
spective study. Crit Care. 2017;21(1):12.
[7] Goldsmith JP, Karotkin EH, Keszler M, et al. Assisted
ventilation of the neonate. 6th ed. Philadelphia (PA):
Elsevier Health Sciences; 2017.
[8] Martin RJ, Fanaroff AA, Walsh MC. Neonatal-perinatal
medicine: diseases of the fetus and infant. 10th ed.
Maryland Heights (MO): Elsevier; 2015.
[9] Monica NF, Pamela S, Juan QL, et al. Recent under-
standing of pathophysiology, risk factors and treat-
ments of neonatal respiratory distress syndrome: a
review. Sci Lett. 2017;5(1):70–78.
[10] Salvo V, Lista G, Lupo E, et al. Noninvasive ventilation
strategies for early treatment of RDS in preterm
infants: an RCT. Pediatrics. 2015;135(3):444–451.
[11] Rubarth LB, Quinn J. Respiratory development and
respiratory distress syndrome. Neonat Netw. 2015;
34(4):231–238.
[12] Thorkildsen K. The utilization of exogenous surfactant
in the neonate. Lynchburg (VA): Liberty University;
2016.
[13] Chang M, Hongyan L, Xiang H, et al. Effections of dif-
ferent ways of mechanical ventilation with pulmonary
surfactant on respiratory mechanics of patients with
neonatal acute respiratory distress syndrome. J Pract
Med. 2017;33(6):916–919.
[14] Ahn J-H, Jung YH, Shin SH, et al. Respiratory severity
score as a predictive factor for the mortality of con-
genital diaphragmatic hernia. Neonat Med. 2018;25(3):
102–108.
[15] Iyer NP, Mhanna MJ. Non-invasively derived respira-
tory severity score and oxygenation index in venti-
lated newborn infants. Pediatr Pulmonol. 2013;48(4):
364–369.
[16] Rawat M, Chandrasekharan PK, Williams A, et al.
Oxygen saturation index and severity of hypoxic
respiratory failure. Neonatology. 2015;107(3):161–166.
[17] Doreswamy SM, Chakkarapani AA, Murthy P. Oxygen
saturation index, A noninvasive tool for monitoring
hypoxemic respiratory failure in newborns. Indian
Pediatr. 2016;53(5):432–433.
[18] Dechert RE, Park PK, Bartlett RH. Evaluation of the
oxygenation index in adult respiratory failure.
J Trauma Acute Care Surg. 2014;76(2):469–473.
[19] DesPrez K, McNeil JB, Wang C, et al. Oxygenation sat-
uration index predicts clinical outcomes in ARDS.
Chest. 2017;152(6):1151–1158.
[20] Chi M, Mei YB, Feng ZC. [A review on neonatal acute
respiratory distress syndrome]. Zhong Guo Dang Dai
Er Ke Za Zhi. 2018;20(9):724–728.
[21] Claverias L, Marı M, Marın-Corral J, et al. The prognos-
tic value of muscle regional oxygen saturation index
in severe community-acquired pneumonia: a pro-
spective observational study. J Intensive Care. 2016;
4(1):7.
[22] Muniraman HK, Song AY, Ramanathan R, et al.
Evaluation of oxygen saturation index compared with
oxygenation index in neonates with hypoxemic
respiratory failure. JAMA Netw Open. 2019;2(3):
e191179.