Characteristics and outcomes of reversible platypnoea–orthodeoxia syndrome in COVID-19 pneumonia

Abstract

Introduction:

Platypnoea–orthodeoxia syndrome (POS) is an uncommon clinical entity characterised by dyspnoea and platypnoea (oxygen desaturation that follows the assumption of an upright position from recumbency). Since the coronavirus disease 2019 (COVID-19) outbreak, increasing reports of COVID-19-related POS and its associated morbidity have been reported around the world. We aimed to study the characteristics of COVID-19-related POS and orthodeoxia (including associations leading to a more prolonged orthodeoxia), and the postdischarge functional outcomes of patients with COVID-19-related POS.

Methods:

An observational cohort study was conducted in a tertiary hospital that managed post-COVID-19 patients. Twenty-four participants with severe-to-critical COVID-19 disease/pneumonia and POS, who received inpatient pulmonary rehabilitation, were enrolled. Descriptive analysis of the data was performed to describe POS/orthodeoxia characteristics and functional outcomes in these participants. Correlation analyses were carried out to identify significant factors associated with a prolonged orthodeoxia.

Results:

The mean duration of POS and orthodeoxia was 12.9 ± 8.3 days and 28.5 ± 14.6 days, respectively. All participants demonstrated resolution of POS and orthodeoxia by hospital discharge. On multivariable analysis, intensive care unit admission and maximal level of respiratory support were significantly associated with a prolonged duration of orthodeoxia. One participant was lost to follow-up. The remaining 23 participants achieved independence in self-care. With the exception of one patient, who was recovering from a hip fracture, the rest achieved independence in ambulation and independent community access.

Conclusion:

Resolution of orthodeoxia was observed in all our participants with COVID-19-related POS. Good functional outcome can be attained with timely and effective rehabilitation interventions.

INTRODUCTION

Platypnoea–orthodeoxia syndrome (POS) is a clinical entity characterised by dyspnoea (platypnoea) and oxygen desaturation (orthodeoxia) that occurs following the assumption of an upright position from recumbency.[1] Before the coronavirus disease 2019 (COVID-19) pandemic, POS owing to respiratory conditions (pulmonary parenchymal disease) had already been described in the literature, but this represented a minority of cases (estimated at around 3.7%).[1] The majority of POS cases were attributable to cardiac causes.[1] Following the COVID-19 outbreak, sporadic cases of COVID-19 pneumonia-related POS were increasingly highlighted around the world.[2,3,4] Although uncommon, it is by no means rare. Many case reports and series have provided different aspects of this disease entity.

We collated a cohort of 24 patients with COVID-19 pneumonia-related POS, who had undergone inpatient rehabilitation in the acute–subacute setting. In this article, we report the characteristics of POS observed in this group of patients and describe their functional outcomes after discharge and review in the outpatient setting.

SUMMARY BOX

What is known?

Platypnoea–orthodeoxia syndrome (POS) is a rare condition, but has increasingly been reported in COVID-19 pneumonia cases.

What is new?

Admission to the intensive care unit and a higher level of respiratory support (possibly a reflection of COVID-19 pneumonia/disease severity) were associated with a longer orthodeoxia duration. Orthodeoxia resolution continued gradually, even on exit from POS criterion, encumbering verticalisation and rehabilitative efforts. Good respiratory and functional outcomes were seen on follow-up postdischarge, with early inpatient rehabilitation.

What is the impact?

A high index of suspicion is needed to detect POS/orthodeoxia in COVID-19 pneumonia. Given its prolonged duration (~1 month) and reversibility, early aggressive rehabilitation is vital for preventing attendant deconditioning and functional loss.

METHODS

This observational cohort study was conducted in a tertiary hospital that managed patients with COVID-19 disease/pneumonia. We recruited patients who were recovering from COVID-19 pneumonia, and had persistent dyspnoea with impaired oxygen saturation (hence requiring oxygen supplementation) and required ongoing rehabilitation. These patients who required supplemental oxygen therapy would, by definition, be considered as having severe or critical COVID-19 illness.[5]

These patients also fulfilled the criterion for POS: dyspnoea accompanied by a drop in peripheral oxygen saturation (SpO₂) >5% (measured with pulse oximeter) on assuming an upright position (sitting or standing) from recumbency.[1] Participants were recruited between 15 December 2021 and 31 July 2022. The participants started on pulmonary rehabilitation (with a focus on respiratory and physical conditioning) at the earliest time possible. This was undertaken by a multidisciplinary team consisting of physiotherapists and occupational therapists in consultation with rehabilitation medicine physicians. Supplemental oxygen therapy was used adjunctively during rehabilitation as necessary and gradually weaned thereafter. Participants were reviewed in the outpatient clinics following their discharge home.

The study was approved by the National Healthcare Group Domain Specific Research Board (NHG DSRB 2021/01045). Informed consent was obtained from the participants enrolled in this study. This study adheres to applicable STROBE guidelines.

Data were reviewed and collated from the participants’ medical records. Baseline demographic data and clinical characteristics were obtained. These included age, gender, ethnicity, marital status, employment, smoking status, vaccination status and comorbidities. Comorbidities were also represented using the Charlson Comorbidity Index.[6] Premorbid functional status, including the ability to perform basic activities of daily living (ADLs) and ambulatory ability, was recorded. The latter was reflected as a functional ambulation category (FAC) score,[7] which ranges from 0 (unable to walk or requires more than one-person assist to ambulate) to 5 (able to ambulate independently on any surface, including stairs).

Details of the participants’ inpatient hospitalisation were collected; they included length of stay, maximal respiratory support provided, admission to intensive care unit (ICU) and duration of physical therapy received. We also recorded clinical data such as the lowest noted P/F ratio (i.e., the ratio of partial pressure of oxygen in arterial blood [PaO₂] to the fraction of inspiratory oxygen concentration [FiO₂]), duration of POS and concomitant presence of complications, such as exertional desaturation (defined as a drop in SpO₂ ≥4% during exertion)[8] and postural hypotension (defined as a drop in systolic blood pressure ≥20 mmHg and/or diastolic blood pressure ≥10 mmHg within 3 min of standing).[9] Radiological features of the 16 participants who underwent chest computed tomography were recorded.

Lastly, participants’ discharge destination, functional status at discharge and follow-up were noted. The latter included participants’ ability to perform ADLs, ambulation status (including FAC) and life space mobility.[10] Life space mobility, a concept for assessing functional mobility over time, was defined at three levels: homebound mobility, limited community mobility and unlimited community mobility. Participants who accessed facilities in the vicinity of their home (e.g., local shops, markets) but not farther were deemed to have limited community mobility, while those who ventured to areas beyond the vicinity of their home (usually on public transport or driving) were deemed to have unlimited community mobility.

The primary outcome of interest was the duration of POS and orthodeoxia. Time points for the following were noted for each participant [Figure 1]: (A) onset of COVID-19 symptoms (fever, respiratory symptoms); (B) initial detection of POS (fulfilling the POS criteria); (C) first notable improvement of orthodeoxia (i.e., drop in SpO₂ to 2%–5% with verticalisation); and (D) resolution of orthodeoxia (i.e., 0%–1% change in SpO₂ with verticalisation). The duration between time points B and C represents the duration of POS (as fulfilled by POS definition). The duration between time points C and D denotes the period of gradual resolution of orthodeoxia (but not fulfilling the POS criteria) to normality. Time point D represents normalisation of SpO₂ with postural change. Hence, the period between time points B and D corresponds to the total duration of orthodeoxia experienced by the participant.

Figure 1.

Time points in the recovery of platynoea–orthodeoxia syndrome (POS). SpO₂: peripheral oxygen saturation

A secondary outcome measured was the postdischarge functional status of the participants, as reflected by their ability to be independent in ADLs (including ambulation). Other than ambulation status, life space mobility, which reflects the subjects’ functional mobility in the community, was also noted.

Descriptive statistics were used to analyse demographic, clinical characteristics and functional outcomes at/after hospital discharge. Correlation analysis (Spearman’s and Pearson’s correlations) was applied as appropriate to determine the relationships between selected variables and the duration of POS. The variables that were analysed included age, gender, ethnicity, comorbidities, smoking status, and vaccination status, admission to ICU, P/F ratio and maximal respiratory support provided. Linear regression was utilised for multivariable analysis. There were no missing values in the data set. Correlation analysis was applied to determine the relationship between orthodeoxia duration (time points B to D) with the duration of oxygen use, as well as the relationship between POS duration (time points B to C) and the duration taken for orthodeoxia to resolve following emergence from POS criteria (time points C to D) [Figure 1]. A P value <0.05 was considered statistically significant for a two-tailed test. Statistical analyses were conducted using IBM SPSS Statistics version 25.0 (IBM Corp., Armonk, NY, USA).

RESULTS

Twenty-four participants with severe COVID-19 illness/pneumonia and POS were enrolled in this study. The base rate of cases of COVID-19-related POS requiring oxygen supplementation and inpatient pulmonary rehabilitation was 0.37%. Most of the participants were male and of Chinese ethnicity. All of them were premorbidly independent in basic ADLs and ambulation. None required supplemental oxygen at rest or during activities before the COVID-19 illness. Other demographic details are presented in Table 1.

Table 1.

Demographic and clinical data of the study population (N=24).

Characteristic	n (%)
Age, mean±SD (yr)	66.9±11.1
Gender
Male	17 (70.8)
Female	7 (29.2)
Ethnicity
Chinese	21 (87.5)
Malay	2 (8.3)
Indian	1 (4.2)
Marital status
Single	6 (25.0)
Married	17 (70.8)
Widowed	1 (4.2)
Preadmission comorbidities
Hypertension	12 (50.0)
Hyperlipidaemia	9 (37.5)
Diabetes mellitus	5 (20.8)
Previous myocardial infarction	3 (12.5)
Previous stroke	2 (8.3)
Chronic obstructive pulmonary disease	1 (4.2)
Charlson Comorbidity Index
<4	15 (62.5)
≥4	9 (37.5)
Smoking status
Active smoker	2 (8.3)
Ex-smoker	6 (25.0)
Non-smoker	16 (66.7)
COVID-19 vaccination status (at the time of disease contraction)
Non-vaccinated	13 (54.2)
Partially vaccinated	6 (25.0)
Fully vaccinated	5 (20.8)
Premorbid employment
Yes	8 (33.3)
No	16 (66.7)
Premorbid functional status
Independent in basic ADLs	24 (100)
Premorbid functional ambulation category
0–3	0 (0)
4	1 (4.2)
5	23 (95.8)

ADLs: activities of daily living, SD: standard deviation

The clinical details and characteristics, including POS, are presented in Table 2. All the patients required oxygen supplementation and had radiographical evidence of bilateral pneumonia during the COVID-19 illness. Computed tomography of the thorax was performed for 16 of the patients. During hospitalisation, 18 (75.0%) patients required ICU admission, of whom six (25.0%) and ten (41.7%) required mechanical ventilation and high-flow nasal cannula oxygen therapy, respectively. The mean duration from onset of COVID-19 symptoms (i.e., respiratory symptoms and/or fever) to detection of POS was 22.1 days. Duration of POS averaged 12.9 days, with most of the patients exhibiting mild platypnoea (75%). Even with POS resolution, orthodeoxia took a further 17.5 ± 10.5 days to achieve normality. All the patients showed concurrent exertional desaturation. It was notable that exertional desaturation persisted beyond the resolution of POS in these patients [Table 3].

Table 2.

Characteristics of COVID-19 pneumonia and POS in the study population.

Characteristic	n (%)
COVID-19 pneumonia in POS (n=24)
Length of hospital staya (day)	61.8±31.0
Intensive care unit admission	18 (75.0)
Severity classification of ARDS
Mild (P/F ratio 201–300 mmHg)	8 (33.3)
Moderate (P/F ratio 101–200 mmHg)	9 (37.5)
Severe (P/F ratio ≤100 mmHg)	7 (29.2)
Maximal respiratory support during acute stay
Mechanical ventilation	6 (25.0)
High-flow nasal cannula therapy	10 (41.7)
Venturi mask, low-flow nasal cannula oxygen therapy	8 (33.3)
Duration of physical therapy receivedb (day)	41 [143]
COVID-19-related POS (n=24)
Disease duration before detection of POSa (day)	22.1±10.1
Duration of POSa (day)	12.9±8.3
Severity of platypnoea during POS
Mild	18 (75.0)
Moderate–severe	6 (25.0)
Total duration of orthodeoxiaa (day)	28.5±14.6
Normalisation of orthodeoxia before discharge	24 (100.0)
Accompanying complications (n=24)
Exertional desaturation	24 (100.0)
Postural hypotension	14 (58.3)
Chest computed tomography features (n=16)
Ground-glass opacification	16 (100.0)
Interstitial thickening, reticular bands, parenchymal bands	11 (68.8)
Bronchiectasis, traction, architectural changes	6 (37.5)
Pulmonary embolism	2 (6.3)

Data presented as ^amean ± standard deviation and ^bmedian [interquartile range]. ARDS: acute respiratory distress syndrome, P/F ratio: ratio of partial pressure of oxygen in arterial blood to fraction of inspiratory oxygen concentration, POS: platypnoea–orthodeoxia syndrome

Table 3.

Associations with orthodeoxia duration.

Variable	Correlation analysis		Multivariable analysis
	Correlation coefficient	P	Standardised coefficient B	95% CI	P
Patient factor (N=24)
Age (yr)	0.184	0.390	–	–	–
Gender	–0.100	0.643	–	–	–
Ethnicity	–0.249	0.241	–	–	–
Smoking status	–0.068	0.751	–	–	–
COVID-19 vaccination status	0.213	0.312	0.291	–1.641 to 12.024	0.128
Hypertension	0.169	0.430	–	–	–
Hyperlipidaemia	0.299	0.155	–	–	–
Diabetes mellitus	0.059	0.783	–	–	–
Previous myocardial infarction	0.082	0.703	–	–	–
Previous stroke	–0.022	0.919	–	–	–
Chronic obstructive pulmonary disease	0.212	0.574	–	–	–
Charlson Comorbidity Index (<4 vs. ≥4)	0.050	0.817	–	–	–
Clinical factors
P/F ratio	–0.372	0.073	–0.182	–0.140 to 0.070	0.496
Admission to intensive care unit	0.042	0.846	–0.560	–36.023 to –0.884	0.041
Maximal level of respiratory support (IMV vs. HFNC/VM/LFNC)	0.414	0.044	0.762	2.504 to 26.130	0.020

Orthodeoxia duration association with duration of supplemental oxygen use: Pearson’s correlation = 0.695 and P <0.001. Association of platypnoea–orthodeoxia syndrome (POS) duration (time points B to C) with duration that orthodeoxia takes to resolve after falling out of POS criteria (time points C to D): Pearson’s correlation = 0.176 and P = 0.410. CI: confidence interval, HFNC: high-flow nasal cannula, IMV: invasive mechanical ventilation, LFNC: low-flow nasal cannula, P/F ratio: ratio of partial pressure of oxygen in arterial blood to fraction of inspiratory oxygen concentration, VM: venturi mask

On correlation analysis, we found that having the need for more aggressive respiratory support, for example, invasive mechanical ventilation, was significantly associated with a longer duration of orthodeoxia. The P/F ratio had mild negative correlation (correlation coefficient of −0.372), but this did not reach statistical significance. On multivariable analysis, only admission to ICU and maximal level of respiratory support were identified as significant factors for a prolonged duration of orthodeoxia, with the latter being more strongly associated. In addition, orthodeoxia duration was significantly associated with total duration of supplemental oxygen use. However, POS duration was not found to be related to the time taken for orthodeoxia to normalise (following emergence from the POS criteria).

Therapy sessions spanned a median of 41 days, commencing at the acute phase of stay. Nineteen (79.2%) participants continued inpatient rehabilitation at the tertiary rehabilitation centre attached to the hospital. One (4.2%) was transferred for further therapy at a community hospital for logistic reasons. But eventually, all the subjects were discharged home. On discharged home from the hospital, 20 (83.3%) and 15 (62.5%) patients achieved independence in ambulation and basic ADLs, respectively [Table 4]. One patient was lost to follow-up, while the remaining twenty-three patients were followed up for 2–26 weeks (mean 88.7 days) after they were discharged home. Except for one patient who was wheelchair bound following a fractured hip after discharge and was confined at home, all other participants who attended follow-ups had regained independence in ambulation and achieved community access of varying degrees.

Table 4.

Functional outcome.

Outcome	n (%)	Outcome	n (%)
Outcome at discharge (n=24)		Tertiary rehabilitation centre	19 (79.2)
Functional status at discharge		Community hospital	1 (4.2)
Independence in all basic ADLs	15 (62.5)	Outcome at follow-up (n=23)a
Requires supervision for at least one basic ADL	8 (33.3)	Symptom onset to follow-up, mean±SD (day)	151.7±57.1
Requires assistance for at least one basic ADL	1 (4.2)	Independence in basic ADLs	23 (100)
Need for home oxygen on discharge	0 (0)	Ambulation status
Functional ambulation category at discharge		Independence without aid	19 (82.6)
0–1	0 (0)	Independence with walking aid	3 (13.0)
2	1 (4.2)	Wheelchair mobility	1 (4.3)b
3	3 (12.5)	Life space mobility
4	6 (25.0)	Homebound	1 (4.3)b
5	14 (58.3)	Limited community access	7 (30.4)
Separation destination		Unlimited community access	15 (65.2)
Home	4 (16.7)

^aMissing data: one subject did not attend follow-up after discharge. ^bOne patient sustained a hip fracture posthome discharge, and hence reported poor ambulation and selfcare performance during follow-up review. ADLs: activities of daily living, SD: standard deviation

DISCUSSION

Platypnoea–orthodeoxia syndrome is a rare clinical entity that was first described in the late 1940s.[11] A meta-analysis published before the global COVID-19 pandemic found that the most common cause of POS was of cardiac origin, in particular, a patent foramen ovale[1] (intracardiac shunting). Four pathological processes underpinning the mechanisms driving POS have been described: intracardiac shunting, ventilation–perfusion, mismatch intrapulmonary shunting or a combination of these processes.[12] The latter processes appear to be responsible for the phenomenon of POS in COVID-19 pneumonia. Tan et al.,[2] who first reported POS in COVID-19, surmised that POS in COVID-19 acute respiratory distress syndrome could be due to a combination of vasculopathy and increased wasted ventilation leading to a gravitational exacerbation of intrapulmonary shunting. Most of the reports of COVID-19-related POS had been associated with more severe pneumonia requiring high respiratory support (e.g., mechanical ventilation, non-invasive ventilation, high-flow nasal cannula).[3]

In our multivariable analysis model, admission to ICU and a greater level of respiratory support (e.g., invasive mechanical ventilation) were found to be associated with a longer duration of orthodeoxia. Correlation analysis of the P/F ratio with orthodeoxia duration indicated a low negative correlation, not amounting to significance. These factors are indicative of severe COVID-19 disease/pneumonia and are in keeping with the cases reported in the literature.[2,3]

Interestingly, we found a subset of patients (29.2%) in our POS cohort with milder pneumonia, who required no more than the use of venturi mask or low-flow nasal cannula oxygen therapy. In spite of the differences in COVID-19 pneumonia severity, all participants with POS had concurrent exertional desaturation. Exertional desaturation was noted to resolve later in the recovery process. Often, distinguishing exertional desaturation/dyspnoea from POS/orthodeoxia/platypnoea can be challenging. Rehabilitation and medical strategies that were applied (e.g., oxygen supplementation, respiratory techniques/manoeuvres, paced activities) targeted this constellation of symptoms as a whole.[13]

Notably, mild platypnoea was observed in the majority (75%) of POS subjects, and hence, shortness of breath did not present as a major issue as compared to the degree of orthodeoxia during rehabilitation. With improvement of POS orthodeoxia to <6% of SpO₂ (and hence falling out of POS definition criteria), milder orthodeoxia persisted and took an average of slightly more than 2 weeks to resolve to normality (defined as a drop in SpO₂ of 0%–1% with verticalisation from recumbency). Interestingly, POS duration was not found to be predictive of this duration [i.e., time points C to D in Figure 1]. This persistent orthodeoxia, however, continued to be an obstacle to rehabilitation, as functional tasks and activities require assumption of a vertical position (i.e., sitting or standing). Keeping SpO₂ >90% remains a priority during therapy, and this was principally achieved with the judicious use of supplemental oxygen during physical therapy. In fact, the duration of orthodeoxia was directly correlated to the duration of supplemental oxygen use [Table 3].

COVID-19-related POS in our cohort reversed uneventfully, regardless of pneumonia severity, in tandem with the resolution of underlying pneumonia. This also applies to patients who showed fibrotic and/or architectural changes on thoracic computed tomography.

On follow-up, 22 out of 23 patients showed good functional recovery (i.e., independence in basic ADLs and ambulation) [Table 4]. Nearly all (95.7%) demonstrated independence in community accessibility. None needed supplemental oxygen during ambulation or performance of self-care tasks. One patient, however, had poor functional outcome, which was attributable to a hip fracture sustained postdischarge (and not due to COVID-19 disease). Despite significant functional improvements, it should be recognised that some level of reduced physical performance still lingered in some patients. Three patients required the use of walking aid for ambulation, while seven had reduced life space mobility (excluding the patient with hip fracture). These were distinct changes in function following recovery from COVID-19. It is, however, uncertain if these changes were due to post-COVID syndrome or hospital-associated deconditioning[14] or post-hospital syndrome.

There are some limitations to our study. As with other studies, incidences of orthodeoxia/POS in our participants were first revealed with attempted verticalisation at the initial physiotherapy reviews. The true commencement of POS may have predated our recorded onset.[2,13] The utilisation of mechanical ventilation and high-flow oxygen therapy could also have the same effect, thus delaying the onset of POS detection. Secondly, investigations were not routinely conducted to exclude a cardiac cause for POS. However, the resolution of POS in the participants indirectly eliminates the likelihood of underlying cardiac causes, which would otherwise have led to persistent POS in the absence of specific interventions.

To the best of our knowledge, this study is the largest single-centre study of COVID-19-related POS to date. Though uncommon, POS in COVID-19 disease may not be as rare as initially thought. The reversibility of POS in our cohort of patients mirrors other COVID-19 POS reports in the literature.[2,3,4,13] Our study also documented the reversibility and duration taken for orthodeoxia to normalise beyond recovery from POS. Though this portends well for afflicted patients, a longer-term follow-up may be necessary to track the functional outcomes of these individuals. Further, studies with a larger sample size may be needed for better generalisation of results.

In conclusion, we studied a cohort of 24 patients with severe-to-critical COVID-19 pneumonia who were afflicted with POS. We found that a longer duration of orthodeoxia was related to ICU admission and increased respiratory support. Notwithstanding orthodeoxia duration, all cases had resolution of POS and orthodeoxia. Even though good functional recovery was observed in all cases, some participants were still functioning below premorbid levels at follow-up. We believe that individualised and targeted rehabilitative therapy, which can safeguard function and prevent deconditioning, delivered while the afflicted individual is coping with POS is key to ensuring good functional outcome. Continued surveillance and rehabilitative interventions may be of benefit to such individuals.

Conflicts of Interest

There are no conflicts of interest.

Funding Statement

Nil.