Inappropriate Ventilatory Homeostatic Responses in Hospitalized COVID-19 Patients

Abstract

Background: The clinical presentation of COVID-19 suggests altered breathing control - tachypnoea, relative lack of dyspnoea, and often a discrepancy between severity of clinical and radiological findings. Few studies characterize and analyse the contribution of breathing drivers and their ventilatory and perceptual responses.

Aim: To establish the prevalence of inappropriate ventilatory and perceptual response in COVID-19, by characterizing the relationships between respiratory rate (RR), dyspnoea and arterial blood gas (ABG) in a cohort of COVID-19 patients at presentation to hospital, and their post-Covid respiratory sequelae at follow-up.

Methods: We conducted a retrospective cohort study including consecutive adult patients admitted to hospital with confirmed COVID-19 between 1st March 2020 and 30th April 2020. In those with concurrent ABG, RR and documented dyspnoea status on presentation, we documented patient characteristics, disease severity, and outcomes at hospital and 6-week post-discharge.

Results: Of 492 admissions, 194 patients met the inclusion criteria. Tachypnoea was present in 75% pronounced (RR>30) in 36%, and persisted during sleep. RR correlated with heart rate (HR) (r = 0.2674), temperature (r = 0.2824), CRP (r = 0.2561), Alveolar-arterial (A-a) gradient (r = 0.4189), and lower PaO₂/FiO₂ (PF) ratio (r = -0.3636). RR was not correlated with any neurological symptoms. Dyspnoea was correlated with RR (r = 0.2932), A-a gradient (r = 0.1723), and lower PF ratio (r = -0.1914), but not correlated with PaO₂ (r = -0.1095), PaCO₂ (r = -0.0598) or any recorded neurological symptom except for altered consciousness. Impaired ventilatory homeostatic control of pH/PaCO₂ [tachypnoea (RR>20), hypocapnia (PaCO₂ <4.6 kPa), and alkalosis (pH>7.45)] was observed in 29%. This group, of which 37% reported no dyspnoea, had more severe respiratory disease (A-a gradient 38.9 vs. 12.4 mmHg; PF ratio 120 vs. 238), and higher prevalence of anosmia (21 vs. 15%), dysgeusia (25 vs. 12%), headache (33 vs. 23%) and nausea (33 vs. 14%) with similar rates of new anxiety/depression (26 vs. 23%), but lower incidence of past neurological or psychiatric diagnoses (5 vs. 21%) compared to appropriate responders. Only 5% had hypoxia sufficiently severe to drive breathing (i.e. PaO₂ <6.6 kPa). At 6 weeks post-discharge, 24% (8/34) showed a new breathing pattern disorder with no other neurological findings, nor previous respiratory, neurological, or psychiatric disorder diagnoses.

Conclusions: Impaired homeostatic control of ventilation i.e., tachypnoea, despite hypocapnia to the point of alkalosis appears prevalent in patients admitted to hospital with COVID-19, a finding typically accompanying more severe disease. Tachypnoea prevalence was between 12 and 29%. Data suggest that excessive tachypnoea is driven by both peripheral and central mechanisms, but not hypoxia. Over a third of patients with impaired homeostatic ventilatory control did not experience dyspnoea despite tachypnoea. A subset of followed-up patients developed post-covid breathing pattern disorder.

Keywords: COVID-19; breathing pattern disorder; dyspnea; impaired homeostasis; post-covid breathing pattern dysfunction; ventilation.

Figure 1

Panel (A) shows PaCO2 versus RR. Blue dots indicate patients with dyspnoea; red squares, patients with no dyspnoea. Dashed green lines indicate boundaries of normocapnia (4.6kPa and 6.0kPa). Dashed orange line indicates severe hypoxia at 6.6kPa, a level sufficient to drive ventilation. Panel (B) shows the 4-hourly RR (median and interquartile range) over the first 96 hours since admission. Gray-shaded boxes indicate night. Panels (C) show PaCO₂ versus PaO₂ and (D)PaCO₂ versus RR. Panel (E) shows the oxygen dissociation curve (ODC) of Covid-19 patients in this cohort plotted against standard human ODC data from (14), showing no shift.

Figure 2

Breathing Pattern Disorder severity over time. Breathing Pattern Assessment Tool score to rate breathing pattern disorder (BPD) severity, over time since discharge from hospital. Threshold for BPD diagnosis is a score of 4 or more (16).

Figure 3

Components of breathing control in the context of COVID-19.

Figure 4

PaCO₂ thresholds for breathing and % appropriate and inappropriate responders. The normal PaCO₂ threshold for breathing is >4.6 kPa, below which PaCO₂ as a breathing drive would be suppressed. Lowering the PaCO₂ threshold (set-point) for breathing, will decrease the proportion of patients who are considered inappropriate responders for that particular PaCO₂ threshold i.e., still simultaneously have tachypnoea (RR>20) and alkalosis (pH>7.45).