Poor Correlation between Diaphragm Ultrasound and Invasive Gold Standard Technique Derived Respiratory Muscle Strength Assessment in Patients after Hospitalisation for COVID-19

A growing consequence of the COVID-19 pandemic is the occurrence of “long- or post-COVID syndrome.” These new umbrella terms summarise various persistent complaints after the acute COVID-19 disease [1‒6]. One of the most common long-term symptoms is dyspnoea on exertion, which often cannot be explained by reductions in pulmonary and cardiac function [1‒6]. Our group recently showed that diaphragm muscle weakness might explain dyspnoea [6]. At the same time, previous diaphragm ultrasound-based studies suggest that there are COVID-19 survivors who have diaphragm dysfunction (as defined as reduced diaphragm thickening on ultrasound) as a possible cause of dyspnoea [7‒9]. For this reason, sonographic evaluation of the diaphragm has attracted much attention because it is an easy obtainable imaging modality by which both diaphragm thickness and contractility can be evaluated. Furthermore sonographic evaluation with diaphragm thickening ratio (DTR) is the most commonly reported diaphragm ultrasound-derived metric in the literature thought to reflect diaphragm “strength” [10‒14]. DTR is calculated as a quotient from diaphragm thickness at total lung capacity (TLC) and diaphragm thickness at functional residual capacity (FRC). Poulard et al. [15] investigated the relationship between diaphragm thickening fraction as assessed by diaphragm ultrasound and transdiaphragmatic pressure. They found a poor or no correlation between diaphragm thickening fraction (DTF) and Pdi in mechanically ventilated patients. Diaphragm thickening fraction is another way of numerically representing the inspiratory thickening of the diaphragm and is calculated as follows: [(end-inspiratory diaphragm thickness (TLC = total lung capacity) – end-expiratory diaphragm thickness (FRC = functional residual capacity))/end-expiratory diaphragm thickness (FRC = functional residual capacity)] × 100. The correlation between DTR and twitch transdiaphragmatic pressure (twPdi) following magnetic stimulation (MS) of the phrenic nerves in former COVID-19 patients with impaired DTR has not been investigated before.

Although diaphragm ultrasound yields metrics that seem to intuitively reflect its strength (such as diaphragm thickening fraction and maximum excursion velocity), diaphragm ultrasound is not the gold standard technique for evaluation of diaphragm strength. This role is held by assessment of twPdi in response to supramaximal MS of the phrenic nerves [16].

TwPdi following MS of the phrenic nerves is calculated from invasively recorded twitch oesophageal and gastric pressures (twPes and twPgas) [17‒19]. While oesophageal and gastric manometry following MS is non-volitional and reproducible, it is also invasive, time-consuming, technically challenging, and often unpleasant for the individual being assessed [17‒19]. Therefore, it is not widely available or utilised [17‒19].

The comparative diagnostic accuracy of twPdi and diaphragm ultrasound for diagnosing diaphragm weakness has not yet been assessed in the ever-growing number of individuals who survived acute COVID-19 illness. This is an additional post hoc analysis of data previously published [6]. It was hypothesised that diaphragm ultrasound alone would not provide an accurate diagnosis and that twPdi would remain the gold standard for evaluation of diaphragm muscle strength.

This prospective case-control study (ClinicalTrials.gov Identifier: NCT04854863) was approved by the Local Ethics Committee (Ethikkommission an der Medizinischen Fakultät der Rheinisch-Westfälischen Technischen Hochschule Aachen, CTCA-A-Nr. 20-515, AZ EK 443/20). Consecutive individuals who had been hospitalised at the University Hospital RWTH Aachen, Germany, during the first and second waves of COVID-19 illness between 24 February 2020, and 21 April 2021, were included 1 year after discharge.

All investigations were performed in accordance with the ethical standards of the latest revision of the Declaration of Helsinki. Written informed consent was obtained from all patients. Exclusion criteria were diagnosis of an underlying disease that causes acute or chronic type 1 or 2 respiratory failure (especially neuromuscular diseases), body mass index >35 kg/m², spinal disc herniation, epilepsy, alcohol or drug abuse, and non-elective hospitalisation in the 4 weeks prior to the study-specific first examination.

At follow-up, individuals underwent a full set of pulmonary function tests (PFTs). In addition, whole-body plethysmography (MasterLab, Viasys, Hoechberg, Germany) was performed before and after bronchodilation (including diffusing capacity of the lungs for carbon monoxide measurement only after bronchodilation) according to current guidelines and recommendations [20].

A portable ultrasound machine (LOGIQ S8 -XD clear, GE Healthcare, London, UK) with a 10 MHz linear transducer being used for evaluation of diaphragm thickness in the zone of apposition. Measurements were performed on the right hemidiaphragm in the supine position because posture is known to directly affect diaphragm thickness [12]. The patients were all placed in the same standardised way lying on their backs, the headboard 30° angled. All sonographic recordings were saved for later analysis. All measurements were performed 3 times, and the average value for each parameter was calculated.

Diaphragm thickness was measured as the vertical distance between the pleural and peritoneal layer at both TLC and FRC (Fig. 1). This was done in the zone of apposition with the 10 MHz probe positioned in the posterior axillary line between the eighth and tenth intercostal space (Fig. 1). The DTR was calculated as thickness at TLC divided by thickness at FRC. The lower limits of normal, defined as the 5th percentile, for DTR was 1.8, so a DTR value below 1.8 was considered abnormal based on previous validation from our group [21].

Phrenic nerve conduction studies were performed as previously described [16]. Posterior CMS was performed with the subject in the seated position. Stimuli were delivered using a MagPro Compact™ magnetic stimulator equipped with a 2 T 12 cm C-100 circular coil (MagVenture, Willich, Germany) [16]. For posterior CMS, the coil was placed at C7 and then moved up towards C6 until the highest reproducible twPdi was obtained [17‒19]. At least five stimuli were delivered to achieve the highest possible twPdi showing <10% variation from the preceding two stimulations [16]. Supramaximality of magnetic stimuli (with 0.1 msec duration each and 2.0 T maximum magnetic field output) was achieved judging the relationship between stimulation intensity and amplitude of the twPdi. To avoid twitch potentiation, stimuli were separated by at least 30 s. Stimulation at FRC was determined by visual observation of abdominal movements (Fig. 1) [17‒19]. TwPdi below 16 cm H₂O was considered abnormal, as previously defined by our group [22].

Twitch oesophageal pressure (twPes) and twitch gastric pressure (twPgas) were simultaneously recorded using balloon catheters (Cooper Surgical, Trumbull, CT, USA) transnasally inserted into the stomach and the distal oesophagus as previously described [16]. Balloon catheters were connected to a differential pressure transducer (DPT-100 ™, Utah Medical Products, Athlone, Ireland) and a carrier amplifier (ADInstruments, Oxford, UK) [16]. Pressure data for twPgas, twPes, and twPdi (defined as twPes − twPgas) were continuously displayed using LabChart™ software (ADInstruments, Oxford, UK) [17‒19].

All twPdi curves following CMS were saved for detailed analysis. Subjects were also instructed to repeatedly perform a maximum sniff manoeuvre as a measure of volitional inspiratory muscle strength to achieve maximum deflection of the Pdi curve (Fig. 1) [16, 23]. The highest of five consecutive efforts was taken for analysis [17‒19].

Statistical analyses were performed using Sigma Plot™ software (Version 13.0, Systat, Erkrath, Germany). Normal values for twPdi and DTR were derived from our previous work [21, 22]. Data are expressed as mean ± standard deviation. Pearson product moment correlation was used for simple linear regression analysis to explore associations between twitch pressures (twPdi and its components) and data derived from diaphragm ultrasound (DTR and sniff velocity in particular). Strength of correlation was classified as weak (0.20–0.39), moderate (0.40–0.59), strong (0.60–0.79), or very strong (0.80–1.00). For all analyses, a p value of ≤0.05 was considered statistically significant.

50 COVID-19 survivors (58 ± 12 years, 14 female) without any relevant pulmonary or cardiac impairment were evaluated in the pulmonary disease outpatient clinic 15 months after discharge from the hospital and consented to in-depth analysis of respiratory muscle function in our respiratory physiology laboratory. Baseline characteristics, medical history, and characteristics during initial hospitalisation are described in Table 1.

PFT and capillary blood gas analysis did not show any impairments in the study population (Table 1). Based on these results, moderate restrictive lung disease (defined as FVC <60%) would not have been diagnosed, and diaphragm disease would not have been suspected in any of these individuals [22, 24].

Diaphragm ultrasound showed significant alterations overall and diaphragm contraction compared with previously published normal values (Table 2) [22]. DTR was <1.8 in 12/50 participants (24%). Based on DTR as the most commonly reported diaphragm ultrasound-derived metric for indicating diaphragm “strength,” a diagnosis of diaphragm dysfunction would have been made in 12/50 patients (24%).

DTF, calculated as follows: [(end-inspiratory diaphragm thickness (TLC = total lung capacity) − end-expiratory diaphragm thickness (FRC = functional residual capacity))/end-expiratory diaphragm thickness (FRC = functional residual capacity)] × 100, was on average 98,39.

TwPdi following posterior CMS showed substantial alterations overall and in diaphragm contraction properties. TwPdi was <16 cm H₂O in 30/50 participants (60%). Sniff Pdi was <78 cm H₂O in males or <57 cm H₂O in females in 23/50 participants (46%), and sniff Pes was below −57 cm H₂O in males or −41 cm H₂O in females in 32/50 participants (64%), respectively (Table 2). Based on twPdi a diagnosis of diaphragm dysfunction would have been made in 30/50 participants (60%), a rate that is 36% higher than that obtained based on diaphragm ultrasound.

Neither DTR nor sniff velocity correlated with any invasively obtained metrics of diaphragm strength (Fig. 2). For all comparisons between diaphragm ultrasound and invasively obtained diaphragm strength metrics, correlation coefficients were low (mild to moderate at best) (Fig. 2).

Lack of good correlation coefficients between diaphragm ultrasound-derived metrics and invasively obtained metrics (Fig. 2) also translated into a high rate of false-negative results when plotting patients with DTR below 1.8 against patients with twPDI below 16 cm H₂O (Fig. 3). Diaphragm ultrasound had a sensitivity of 26.67% and a specificity of 80.0% in the detection of diaphragm dysfunction (positive predictive value 66.67%, negative predictive value 42.10%).

This study showed that DTR measured by diaphragm ultrasound was not able to accurately diagnose diaphragm dysfunction, i.e., detect patients with an impaired twPdi <16 cm H₂O at 15 months after discharge in individuals hospitalised with acute COVID-19 disease. Especially the low sensitivity of the use of DTR in the detection of diaphragm dysfunction can be criticised because diaphragm ultrasound has a high rate of false negative results. Therefore, the DTR cannot be used as a screening tool. As diaphragm ultrasound underdiagnoses respiratory muscle impairment, invasive muscle strength measurement using twPdi is definitely needed for confirmation, as otherwise the rate of false-negative diagnoses would be clinically unacceptably high.

The present cohort did not show any abnormalities in standard PFTs that would have raised the suspicion of diaphragm dysfunction. Given that 60% of the study population were objectively diagnosed with diaphragm dysfunction, PFTs cannot be recommended as a diagnostic tool for diagnosing diaphragm weakness, which is consistent with previous data [25]. The high percentage of diaphragm dysfunction in our cohort can be explained by the selection of the patients.

Diaphragm ultrasound is an easy-to-obtain tool for evaluation of diaphragm function at the bedside. Indeed, our group has previously shown that a comprehensive bedside diaphragm ultrasound protocol can be obtained within 20 min [12]. However, the use of the DTR with a cut-off value of <1.8 to detect impaired diaphragmatic function in COVID-19 survivors led to a correct diagnosis in 66.67% of all cases (PPV) and sensitivity was very low (26.67%). The metrics derived from diaphragm ultrasound differ from data obtained with invasive muscle strength measurement. The reason for a 36% lower rate of diaphragm dysfunction on diaphragm ultrasound compared with invasive measurements is that the former yields two-dimensional metrics of the three-dimensional diaphragm contraction process, and only evaluates one part of the diaphragm. Furthermore, unlike invasively obtained pressure recordings, diaphragm ultrasound never actually measures strength in terms of pressure generated (i.e., in cm H₂O). Another consideration is that individuals which do not cooperate fully are likely to be incorrectly diagnosed with diaphragm muscle dysfunction based on this assessment alone. Although all the individuals in this study were motivated to participate in the project and cooperate with study assessments, the influence of non-cooperation on diaphragm ultrasound measures cannot be excluded (and can never be excluded in clinical settings) since it is a volitional method.

The diagnosis of diaphragm weakness in these people as defined as an impaired twPdi <16 cm H₂O cannot be accurately made based on spirometry and/or DTR measured by diaphragm ultrasound alone but requires evaluation of twPdi as the gold standard for assessment of diaphragm strength. This finding is clinically relevant because there is a growing number of individuals with otherwise unexplained dyspnoea on exertion after recovery from acute COVID-19 illness who deserve an accurate diagnostic approach.

The study has some limitations that should be kept in mind when reading the study. In the study, we did not investigate whether other non-invasive non-volitional markers, such as diaphragm EMG or twitch mouth pressure, correlate with invasively measured diaphragmatic strength using twPdi.

TwPdi is the gold standard for assessment of diaphragm strength, but it is invasive, time-consuming, technically challenging, and often unpleasant for the individual being assessed; furthermore, it is not widely available or utilised. Future studies will address this and show whether there are other non-invasive parameters that reliably indicate diaphragm dysfunction.

We gratefully thank all the COVID-19 patients whose cooperation made this study possible. We gratefully acknowledge the help of Faniry Ratsimba in analysing patient-related data. English language editing assistance was provided by Nicola Ryan, independent medical writer. We do also wish to thank Dr. Gerold Kierstein (ADInstruments, Oxford, UK) for his help in performing analysis of twitch transdiaphragmatic pressure gradients following cervical stimulation of the phrenic nerve roots.

Ethical approval was received from the Local Ethics Committee (Ethikkommission an der Medizinischen Fakultät der Rheinisch-Westfälischen Technischen Hochschule Aachen, CTCA-A-Nr. 20-515, AZ EK 443/20). Written informed consent was obtained from all patients prior to inclusion.

The authors declare no conflicts of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

This research received no external funding. Internal funding was provided by the RWTH Aachen Faculty of Medicine (START Grant supporting the junior research group around PD Dr. Jens Spiesshoefer).

J.F., J.S., B.J. and B.R. collected the data. J.F. wrote the manuscript. J.F., J.S. and M.D. contributed significantly to the study design. The manuscript was revised by all other authors.

Janina Friedrich and Binaya Regmi contributed equally to this work.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author, PD Dr. Jens Spiesshoefer (Uniklinik RWTH Aachen, [email protected]), upon reasonable request and with permission of University Hospital RWTH Aachen.