Article Text

Download PDFPDF

Are there differences in the relationship between respiratory rate and oxygen saturation between patients with COVID-19 and those without COVID-19? Insights from a cohort-based correlational study
Free
  1. Mikael Laredo1,2,
  2. Claire Vandiedonck3,
  3. Òscar Miró4,
  4. Juan González del Castillo5,
  5. Aitor Alquézar-Arbé6,
  6. Javier Jacob7,
  7. Pascual Piñera8,
  8. Bruno Mégarbane1
  1. 1 Réanimation Médicale et Toxicologique, Hôpital Lariboisière, APHP; INSERM UMRS-1144, Université de Paris, Paris, France
  2. 2 Institut de Cardiologie, Groupe Hospitalier Pitié-Salpêtrière, APHP, Sorbonne Université, Paris, France
  3. 3 Université Paris Cité, INSERM UMR-S1151, CNRS UMR-S8253, Institut Necker Enfants Malades, IMMEDIAB Laboratory, Paris, France
  4. 4 Emergency Department, Hospital Clinic, Universitat de Barcelona, Barcelona, Spain
  5. 5 Emergency Department, Hospital Clinico San Carlos, Madrid, Spain
  6. 6 Emergency Department, Universitat Autònoma de Barcelona, Barcelona, Spain
  7. 7 Emergency Department, Hospital Universitari de Bellvitge, Barcelona, Spain
  8. 8 Emergency Department, Hospital Reina Sofia de Murcia, Murcia, Spain
  1. Correspondence to Professor Bruno Mégarbane, Hospital Lariboisière, Paris 75010, France; bruno.megarbane{at}lrb.aphp.fr

Abstract

Background Physicians have observed patients with COVID-19 without respiratory distress despite marked hypoxaemia and extensive radiographic abnormalities, a controversial phenomenon called ‘silent hypoxaemia’. We aimed to compare the relationship between RR and peripheral oxygen saturation (SpO2) in patients with COVID-19 versus patients without COVID-19 when breathing air on admission.

Methods We conducted a retrospective multicentre ED cohort correlational study.

We used the Spanish Investigators on Emergency Situations TeAm network cohort of patients with COVID-19 admitted to 61 Spanish EDs between March and April 2020. The non-COVID-19 cohort included patients with lower respiratory tract bacterial infections admitted between January 2016 and April 2018.

We built a multivariable linear model to investigate the independent predictive factors related to RR and a logistic multivariate regression model to analyse the presence of ‘silent hypoxaemia’.

Results We included 1094 patients with COVID-19 and 477 patients without COVID-19. On admission, RR was lower (20±7 vs 24±8/min, p<0.0001), while SpO2 higher (95±5% vs 90±7%, p<0.0001) in patients with COVID-19 versus patients without COVID-19. RR was negatively associated with SpO2 (RR decreasing with increasing age, beta=−0.37, 95% CI (−0.43; −0.31), p<0.0001), positively associated with age (RR increasing with increasing age, beta=0.05, 95% CI (0.03; 0.07), p<0.0001) and negatively associated with COVID-19 status (RR lower in patients with COVID-19, beta=−1.90, 95% CI (−2.65; −1.15), p<0.0001). The negative RR/SpO2 correlation differed between patients with COVID-19 aged <80 and ≥80 years old (p=0.04). Patients with COVID-19 aged ≥80 years old had lower RR than patients without COVID-19 aged ≥80 years old at SpO2 values <95% (22±7 vs 24±8/min, p=0.004). ‘Silent hypoxaemia’ defined as RR <20/min with SpO2 <95% was observed in 162 (14.8%) patients with COVID-19 and in 79 (16.6%) patients without COVID-19 (p=0.4). ‘Silent hypoxaemia’ was associated with age ≥80 years (OR=1.01 (1.01; 1.03), p<0.0001) but not with gender, comorbidities and COVID-19 status.

Conclusion The RR/SpO2 relationship before oxygen administration does not differ between patients with COVID-19 and those without COVID-19, except in elderly patients.

  • COVID-19
  • respiratory
  • emergency department

Data availability statement

Data are available upon reasonable request.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

  • Patients with COVID-19 may initially present to the ED with minimal to no dyspnoea despite profound hypoxaemia and extensive radiographic abnormalities. This phenomenon has been called ‘silent hypoxaemia’.

WHAT THIS STUDY ADDS

  • This study has demonstrated that the relationship between RR and peripheral oxygen saturation (before oxygen administration) in patients younger than 80 years old is comparable between patients with COVID-19 and those without COVID-19 admitted with respiratory disease.

  • ‘Silent hypoxaemia’, a posteriori defined as RR <20/min and a peripheral oxygen saturation <95%, was associated with increasing age, but not with gender, comorbidities or COVID-19.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

  • In COVID-19 and other lower respiratory tract infections, the absence of clear dyspnoea does not rule out the presence of profound hypoxaemia and/or extensive radiographic abnormalities.

Introduction

Mechanisms leading to hypoxaemia in SARS-CoV-2-infected patients are not fully understood. SARS-CoV-2-attributed pneumonia is characterised by atypical presentations of acute respiratory distress syndrome, with a significant number of SARS-CoV-2-infected patients showing severe hypoxaemia despite preserved lung compliance at least at the initial stage when admitted to the hospital.1 Clinicians have observed patients with minimal to no dyspnoea despite profound hypoxaemia and extensive radiographic abnormalities, a phenomenon called ‘silent hypoxaemia’, referring to the absence of respiratory distress.2 3 Various underlying pathophysiological hypotheses have been speculated to explain these observations including inadequate nervous system sensing in response to hypoxaemia,4 micro-thrombi in pulmonary vasculature,5 pulmonary angiopathy with abnormal vasodilatation of capillaries,6 7 and altered relationship between oxygen delivery and transport.8

Aiming to identify possible SARS-CoV-2-attributed particularities in the ventilation pattern observed at the bedside, we studied the relationship between RR and peripheral oxygen saturation (SpO2) measured using pulse oximetry when breathing room air on admission in COVID-19 in comparison with patients without COVID-19.

Methods

Study population

We performed a retrospective multicentre ED cohort correlational study. We used the SIESTA (Spanish Investigators on Emergency Situations TeAm) network cohort, which included 1198 consecutive patients with COVID-19 admitted to 61 Spanish EDs between March and April 2020 for which extensive details have been previously published.9 Diagnosis was obtained using reverse transcriptase-PCR (Cobas SARS-CoV-2 Test, Roche, Spain) performed on nasopharyngeal swabs. The non-COVID-19 cohort included patients with lower respiratory tract bacterial infections admitted to 54 Spanish EDs in a period preceding COVID-19 outbreak, between January 2016 and April 2018.10 Among the SIESTA cohort and the non-COVID-19 cohort, 1094 patients and 477 patients, respectively, had more than 95% of data compiled necessary for the present study (variables shown in table 1) and were subsequently included.

Table 1

Patient characteristics on admission to the ED

SpO2 was measured by pulse oximetry, almost systematically at room air. For each patient, we selected the SpO2 value at ED arrival or, alternatively, the first value registered by emergency medical service on the scene.

Definitions

Silent hypoxaemia was defined a posteriori as SpO2 <95% and RR <20/min based on the haemoglobin dissociation curve and clinical practice.

To assess a possible effect of older age on the RR/SpO2 relationship, an arbitrary cut-off of 80 years, corresponding to the third quartile of the age distribution in the whole study population, was defined. We also stratified the correlation analysis according to gender and presence of silent hypoxaemia.

Statistical analyses

Data are presented as percentages or mean±SD as required. There were no missing data for the variables analysed in regression models (age, gender, COVID-19 status, chronic obstructive pulmonary disease (COPD), chronic heart failure, cerebrovascular disease, RR and SpO2). Univariate comparisons between groups were performed using Fisher’s exact or Student’s t-tests as appropriate. We built a multivariable linear model to investigate the independent predictive factors related to RR and a multivariate logistic regression model to analyse the presence of silent hypoxaemia. Pearson correlations between RR and age, SpO2 and age, and RR and SpO2 were tested in patients with COVID-19 and those without COVID-19. Pearson coefficients of correlation (r) were compared following the Fisher z-transformation. Data manipulation and statistical analyses were performed using the R statistical programming language (R Core Team (2021), V.4.0.4-environment; R: A language and environment for statistical computing (R Foundation for Statistical Computing, Vienna, Austria)). This report was prepared in compliance with the Strengthening the Reporting of Observational Studies in Epidemiology checklist for observational studies. Given the absence of published data regarding a potential difference in the RR/SpO2 relationship between COVID-19 and non-COVID-19, and the nature of the data concerned (comparison of correlation coefficients), no sample size calculation was performed prior to data analysis.

Patient and public involvement

Patients and/or the public were not involved in the design or conduct or reporting or dissemination plans of this research. In accordance with the ethical standards of Spanish legislation, informed consent was waived due to the non-interventional study design that did not modify existing diagnostic or therapeutic strategies. Only the non-opposition of the patient or their legal representative was collected.

Results

Patients’ characteristics

We included 1094 consecutive patients with COVID-19 and 477 consecutive patients without COVID-19 (figure 1). Patients with COVID-19 were significantly younger than patients without COVID-19 (63±18 years vs 74±14 years, p<0.0001), with a lower proportion of males (56% vs 64%, p=0.005; table 1). COPD and chronic heart failure were significantly less frequent in patients with COVID-19 versus those without COVID-19.

Figure 1

Study flow chart. RT-PCR, reverse transcriptase-PCR; SIESTA, Spanish Investigators on Emergency Situations TeAm.

Relationship between RR and SpO2

On admission, RR was significantly lower (20±7 vs 24±8/min, p<0.0001), while SpO2 significantly higher (95±5% vs 90±7%, p<0.0001) in patients with COVID-19 versus patients without COVID-19. Both RR and SpO2 were correlated with age, positively for RR (RR increasing with increasing age) (r=0.27, 95% CI (0.22; 0.31), p<0.0001) and negatively for SpO2 (SpO2 decreasing with increasing age) (r=−0.34, 95% CI (−0.39; −0.30), p<0.0001).

Overall, there was a weak negative but highly significant correlation between RR and SpO2 (RR decreasing with increasing SpO2, r=−0.39, 95% CI (−0.43; −0.35), p<0.0001), which remained after inclusion of age, gender and comorbidities as covariates (beta=−0.39, 95% CI (−0.44; −0.33), p<0.0001). In a multivariate model including also COVID-19, RR was again negatively associated with SpO2 (RR decreasing with increasing SpO2, beta=−0.37, 95% CI (−0.43; −0.31), p<0.0001), positively associated with age (RR increasing with increasing age, beta=0.05, 95% CI (0.03; 0.07), p<0.0001) but also negatively associated with COVID-19 status (RR lower in patients with COVID-19, beta=−1.90, 95% CI (−2.65; −1.15), p<0.0001). COPD, chronic heart failure and cerebrovascular disease had no effect in the model.

In patients with COVID-19 and those without COVID-19, SpO2 and RR were negatively correlated (r=−0.33, 95% CI (−0.38; −0.28), p<0.0001 and r=−0.34, 95% CI (−0.42; −0.26), p<0.0001, respectively); the difference between these two correlations was not significant (comparison of Pearson coefficients between groups, p=0.90; figure 2A).

Figure 2

Correlations between RR and pulse oximetry oxygen saturation (SpO2). Pearson correlations between RR and pulse oximetry SpO2 in patients with COVID-19 (orange, triangles) and patients without COVID-19 (purple, circles) in the whole study population (A) and the subgroup of patients aged over 80 years (B). In the whole population, RR and SpO2 negatively correlated (r coefficients, −0.33 and −0.34, respectively) with no significant difference between both groups (Pearson coefficients difference, p=0.90). In the subgroup of patients aged over 80 years, patients with COVID-19 had a significantly lower correlation between RR and SpO2 than patients without COVID-19 (comparison of Pearson coefficients between groups, p=0.04).

After stratification for SpO2 levels using the median value of 95% as threshold, a negative RR/SpO2 correlation was found in both patients with COVID-19 (r=−0.29, 95% CI (−0.36; −0.21), p<0.0001) and patients without COVID-19 (r=−0.27, 95% CI (−0.36; −0.18), p<0.0001) with SpO2 ≤95% (figure 2B). By contrast, no correlation was found in patients with COVID-19 (r=−0.008, 95% CI (−0.07; 0.09), p=0.85) and patients without COVID-19 (r=−0.07, 95% CI (−0.26; 0.12), p=0.47) with SpO2 >95%.

To further explore whether SpO2 level might influence the RR/SpO2 relationship in patients with COVID-19 and those without COVID-19, we performed a multivariate analysis searching for an interaction term between SpO2 and COVID-19 status on RR. No interaction was found in the total patient sample (beta=−0.007, 95% CI (−0.13; 0.11), p=0.90).

The negative RR/SpO2 correlation was different between patients with COVID-19 aged <80 and ≥80 years old (comparison of Pearson coefficients between groups, p=0.04; figure 3). In a multivariate model including age and comorbidities, there was an interaction between SpO2 and COVID-19 status (beta=0.25, 95% CI (−0.0003; 0.49), p=0.0499), reaching statistical significance in the patients with COVID-19 aged ≥80 years old. These patients exhibited lower RR than patients without COVID-19 aged ≥80 years old at SpO2 values <95% (22±7 vs 24±8/min, p=0.004).

Figure 3

Comparison of Pearson coefficients between groups. Forest plot of Pearson correlation coefficients between RR and SpO2 in patients with COVID-19 and those without COVID-19. Reported p values are those obtained for comparisons of Pearson coefficients between groups. There were no significant differences in Pearson coefficients between COVID-19 and non-COVID-19 except in patients aged over 80 years, with significantly lower correlation between RR and SpO2 in patients with COVID-19 versus patients without COVID-19 (significantly higher Pearson coefficient, p=0.04). SpO2, peripheral oxygen saturation.

‘Silent hypoxaemia’

Defined as RR <20/min together with SpO2 <95%, ‘silent hypoxaemia’ was found in 162 (14.8%) patients with COVID-19 vs 79 (16.6%) patients without COVID-19 (p=0.4). Overall, 242 (22%) patients had SpO2 <95% with RR ≥20/min in the COVID-19 group vs 262 (55%) in the non-COVID-19 group (p<0.0001). In a multivariate model, silent hypoxaemia was only significantly associated with age (silent hypoxaemia more frequent with increasing age, OR=1.01 (1.01; 1.03), p<0.0001) but not with gender, comorbidities or COVID-19 status.

Discussion

Clinical presentation in COVID-19 is highly variable with some patients exhibiting no respiratory distress despite profound hypoxaemia, a phenomenon referred to as ‘silent hypoxaemia’.2–4 Whether this entity is physiological or paradoxical remains controversial. To date, no study has compared the relationship between hypoxaemia and respiratory status in COVID-19 versus other aetiologies of acute lung injury. Differences could only be investigated in cohorts of patients with COVID-19 and patients without COVID-19 with data obtained at the first medical contact before oxygen administration, like in our series.

Overall, we observed no significant differences in the RR/SpO2 relationship between patients with COVID-19 and those without COVID-19 on admission. However, patients with COVID-19 aged ≥80 years old with low SpO2 exhibited lower variations in RR relative to changes in SpO2. Our findings did not support better tolerance of hypoxaemia in patients with COVID-19 except those aged ≥80 years old.

Study limitations

Our study had limitations. Our patient groups were not admitted in the same period. ABG analysis before oxygen administration was mostly unavailable, and therefore, arterial pressure of carbon dioxide, a major determinant of respiratory function, could not be compared between patients with COVID-19 and those without COVID-19. We used RR as surrogate of respiratory distress, which may not adequately represent clinical tolerance to hypoxaemia. Although our non-COVID-19 group included various aetiologies of lower respiratory tract infections not restricted to viral infections, patterns of hypoxaemia without high RR did not differ between patients with COVID-19 and those without COVID-19. Moreover, measurements of RR relying on human observations are notoriously subject to errors.11 Last, our study population lacked in-depth clinical characteristics. The differences observed in the RR/SpO2 relationship between patients with COVID-10 and those without COVID-19, especially in the elderly group, may be explained by a greater proportion of patients with underlying conditions affecting oxygen and carbon dioxide physiology, as lower SpO2 values have been reporter in older patients.12

Conclusions

The RR/SpO2 relationship before oxygen administration, investigated in a nationwide multicentre ED cohort, does not differ between patients with COVID-19 and patients without COVID-19, except in the elderly patients. Mechanisms underlying these findings are currently unexplained.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study was performed in agreement with the 2013 Declaration of Helsinki of the World Medical Association. The ethics committee for clinical investigation of the University Hospital Clinics of Barcelona, Spain approved the study (protocol no: HCB/2020/0534).

Acknowledgments

The authors would like to thank all emergency physicians who contributed to data collection in the SIESTA (Spanish Investigators on Emergency Situations TeAm) network.

References

Footnotes

  • Handling editor Ed Benjamin Graham Barnard

  • Twitter @LaredoMikael, @CVandiedonck

  • Contributors ML, CV, OM and BM have made substantial contributions to the conception and design of the study. OM, JGdC, AA-A, JJ and PP have contributed to acquisition of data. ML, CV and BM performed the analysis and the interpretation of data. All authors read and approved the final manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Commissioned; externally peer reviewed.

Linked Articles