Introduction

The COVID-19 pandemic brought taste and smell impairment to the forefront of medicine, revealing considerable gaps in the knowledge of underlying pathophysiological mechanisms. Olfactory and gustatory dysfunction are reported in more than half of the infected individuals during the first wave of COVID-19, often as the first, and sometimes as the sole presentation of the acute disease1. Accordingly, some authors proposed that chemosensory symptoms could be used as early predictors of SARS-CoV-2 spread in the population2,3,4. More recently, smell dysfunction showed two to tenfold lower prevalence among Omicron-infected patients compared to those with other variant of concerns (VoCs), mostly Delta5 and persistence after the resolution of the acute phase in a considerable proportion of individuals6,7,8,9,10,11,12. Furthermore, sustained smell loss at 12 months is suggested to be permanent13, while an improvement in olfactory function has been observed after olfactory training14.

Patients reporting an olfactory and/or gustatory impairment may experience poor appetite and malnutrition. Furthermore, people with anosmia or ageusia are at an increased risk to accidentally consuming spoiled or rancid foods or being unaware when they are breathing toxic, polluted or smoke-filled air. Hence, olfactory and gustatory evaluation by means of validated tests is essential to correctly identify patients with such impairments15. Currently, chemosensory dysfunction is detected both through self-reported methods (including interviews, surveys, and electronic health records) and through validated psychophysical assessment16,17,18.

As a sub-study of the ORCHESTRA (connecting European cohorts to increase common and effective response to SARS-CoV-2 pandemic) Project12,19, aiming at tackling the Coronavirus pandemic to establish an international large-scale-cohort to generate rigorous evidence in the field of prevention and treatment of SARS-CoV-2 infection, the present study offers a comprehensive evaluation of the neurosensorial cluster of post-COVID-19 syndrome (NSc-PCS) across the different stages of COVID-19 disease, including a comparison between self-reporting and psychophysical tests, the characteristics and determinants of chemosensory impairment, the outcome of the olfactory training, and the impact on quality of life (QoL) through the 36-Item Short Form Survey (SF-36) questionnaire.

Results

The cohort included 1187 patients (630, 53% female) diagnosed with a SARS-CoV-2 infection between February 2020 and April 2022 (Fig. 1, Table 1).

Fig. 1
figure 1

Flow chart of the study. NSc-PCS: neurosensorial cluster of post-COVID syndrome, including smell and/or taste impairment; SRC: Self-Reported Chemosensory impairment group; PAC: Psychophysical Assessment of Chemosensory impairment group.

Full size image
Table 1 Demographic characteristics in the UNIVR Orchestra cohort: comparison between patients not reporting chemosensory impairment (patients without NSc-PCS, N = 637) and patients with chemosensory impairment (patients with NSc-PCS, N = 550), either only self-reported (SRC, N = 497) or detected through a psychophysical assessment (PAC, N = 50).
Full size table

Of these, 47% (550) reported smell/taste dysfunction at any time point (NSc-PCS). Persistent symptoms were observed in 7%, while 22 experienced fluctuating symptoms. Over 50% have reported a recovery from smell and taste impairments within 4- and 5-months, respectively. By month 14, reporting recovery rates decreased, with ~ 1% in smell and 2% in taste improvements until month 21 (Supplement, Fig. 1). For ~ 5% patients chemosensory impairment started after the infection.

Patients with NSc-PCS exhibit significantly lower anti-SARS-CoV-2 (RBD) (1243.35 vs 4921.72 BAU; p < 0.001) and neutralization antibodies titers (12.57 vs 26.65 BAU; p < 0.001) compared to patients without PCS, with no relevant differences in laboratory parameters. QoL differences were non-significant, except for lower vitality subscale in NSc-PCS (52.21 vs 55.09, p = 0.011) (Supplement, Figs. 2, 3, 4).

Determinants of self-reported neurosensorial impairment during the acute phase and follow up

Chemosensory impairment prevalence was higher during the July-December 2020, and the lower in early 2022, when Omicron VoC was predominant (Fig. 2). Acute smell and taste impairments were strongly positively associated (Matthews coefficient 0.81; p < 0.001). Multivariable risk factors models for chemosensory symptoms during the acute infection included significant variables from the bivariable analysis (Supplement, Figs. 5, 6). Models yielded AUROC values of 0.81 and 0.82 for smell and taste, respectively. The risk of smell and taste dysfunction during acute COVID-19 was higher for females (smell OR: 3.36; taste: OR: 2.42; p < 0.001) and second wave (OR: 3.83; taste: OR: 3.23; p < 0.001). The probability of smell impairment decreased with age (OR: 0.41; p = 0.014), while that of taste impairment was lower for breakthrough infection (OR: 0.44; p = 0.039) (Table 2).

Fig. 2
figure 2

Prevalence of self-reported chemosensory symptoms in the acute phase (top panel) and at the follow-up (bottom panel), by the time of the acute phase and annotated with the waves and variants dominating in Europe, based on the Nextstrain prevalence data. Only groups with at least 10 patients are shown.

Full size image
Table 2 Results of the multivariable models for a binary outcome of experiencing smell (left) and taste (right) impairments during the acute phase.
Full size table

Two multivariable models identified risk factors for NSc-PCS. The first highlighted that smell and taste impairment during the acute phase were significant predictors (smell OR: 16.05, p < 0.001; taste OR: 3.67, p = 0.003) (Table 3). Older age was associated with a lower NSc-PCS risk, while monoclonal antibodies, hospitalization, and VoCs during the third wave reduced olfactory dysfunction risk in the follow-up. The second set of models (both AUROC > 0.80), excluding acute chemosensory symptoms, confirmed age (p < 0.001, both for smell and taste) and monoclonal antibodies (smell: p = 0.025; taste: p = 0.022) to be associated with a lower risk for NSc-PCS, while females had a higher probability to develop chemosensory sequelae (smell: p < 0.001; taste: p = 0.043) (Table 4).

Table 3 Results of the multivariable logistic regression models for a binary outcome of experiencing smell and taste impairments during any time of the follow up (months 3 to 18).
Full size table
Table 4 Results of the multivariable logistic regression models for a binary outcome of experiencing smell and taste impairments during any time of the follow up (months 3 to 18). For these models the acute-phase symptoms were not used.
Full size table

Comparison between patients with psychophysical assessment of chemosensory impairment (PAC, N = 50) and patients who self-reported chemosensory impairment (SRC N = 497)

PAC patients were mostly female (36, 71%), and younger than the SRC (PAC: 45.3 ± 16.81, SRC: 52.48 ± 14.28; p = 0.002), with no significant differences in demographic and epidemiological features (Table 1). More PAC patients were infected during the second (p = 0.004) and fewer during the first waves (p = 0.003) compared to SRC. Smell impairment was equally reported during the acute infection but more prevalent in the PAC during follow up (p < 0.001 for each of all time points). Taste impairment was similar between PAC and SRC during acute infection and the first three months, but PAC showed higher prevalence in the follow up (p < 0.001 for 6-, 12-, and 18-months follow up) (Table 1). PAC patients exhibited a slower resolution of smell and taste impairments compared to SRC. The time between the acute phase and psychophysical assessment varied from 105 to 705 days (average 415 ± 149 days), and difference between the assessment and the follow-up phone-call ranged from 130 to 473 days (average 318 ± 98 days) (Supplement, Fig. 7).

Characteristics of smell and taste function in PAC (N = 50)

Three patients, two with a pre-existing smell disorder, and one with information missing, were removed, resulting in a dataset of 50 patients (Fig. 1). Of these, 96% experienced smell impairment during the acute infection and 98% at the chemosensory evaluation. A lower rate of self-reported smell impairment was detected at the phone-call interview compared to the chemosensory evaluation (74% vs 98%; p < 0.001).

The mean TDI score was 24.1 (SD: 7.51). A quantitative smell deficit, predominantly hyposmia, was detected in 33 (66%) patients. Among patients with smell impairment, half had a qualitative disorder, mostly parosmia and consistently unpleasant. Thirty-five percent of normosmic patients presented either parosmia or phantosmia. At the phone-call interview, 36 (73%) patients still reported smell impairment, with 23% reporting a qualitative disorder. Despite 82% of patients engaging in the olfactory training, and half of them completing the protocol, 75% continued to report smell impairment (Fig. 3, Table 5).

Fig. 3
figure 3

Composite scores for SSET (top panel) and TST (lower panel). Each bar corresponds to a single patient. Colours correspond to the measurements of score composites. Each plot is divided into sections according with the cut offs defining normogeusia and hypogeusia for gustatory function and normosmia, hyposmia and anosmia for olfactory function. Each bar is annotated with the summary score. As the SSET cut-off was adjusted by age (see “Methods”) the plot is not strictly monotonous.

Full size image
Table 5 Chemosensory features of the PAC (Psychophysical Assessment of Chemosensory impairment) cohort (N = 50) at the three timepoints: acute infection, neurosensorial evaluation, phone-call interview.
Full size table

No patient reported taste impairment pre-COVID-19, while most experienced it during acute infection (45, 90%) and at the neurosensorial evaluation (38, 76%). At the phone-call interview, a lower proportion reported taste impairment (53% vs 76%; p = 0.012). The mean TST score was 11.6 (SD: 2.81). Most patients were normogeusic (36, 72%), with the remaining (14, 28%) presenting with hypogeusia. No cases of ageusia emerged. Over half of the patients showed a qualitative alteration (28, 56%), mostly parageusia, always described as unpleasant. More than half (19, 53%) of TST normogeusic patients reported at least one qualitative disorder. During the phone-call interview, 26 (53%) patients still reported taste impairment, especially parageusia with unpleasant valence (14, 29%). Among hypogeusic patients some showed reduced scores for single taste qualities (i.e. sweet, sour, salty, bitter) (Fig. 3, Table 5).

Deficit in nasal chemesthesis, was reported especially at the acute phase, and in 6 (12%) patients persisted until the chemosensory evaluation. 26 (52%) patients reported nasal congestion during acute infection, which was negatively associated with chemesthesis (Cohen’s − 0.27; p = 0.077). Considerably less patients reported deficit in oral compared to nasal chemesthesis at the acute infection (28% vs 76%), while these proportions were similar at the chemosensory evaluation (10% and 12%, respectively) (Table 5).

A combined smell and taste quantitative impairment occurred in 11 (22%) patients, while 14 (28%) had normal function for both smell and taste (Supplement, Fig. 8). A majority (28, 56%) reported impairments in both smell and taste. Among these, 20% experienced persistent taste but fluctuating smell impairment, whereas four patients (8%) reported a fluctuating smell with persistent taste impairments, and in eight patients (16%), both were fluctuating.

The proportion of patients reporting smell and taste impairments decreased between the chemosensory evaluation and the phone-call interview. Parosmia showed a stable decreasing trend (p = 0.100), while phantosmia increased (p = 0.013). The prevalence of parageusia decreased significantly (p < 0.001), while for phantogeusia the decrease was not significant (p = 0.152). Parosmia and phantosmia were positively associated at the acute phase (Cohen's 0.42; p = 0.01) and during the phone-call (Cohen's 0.42; p = 0.005). Phantosmia at the acute phase and chemosensory evaluation were positively associated (Cohen's 0.46; p = 0.005).

Among patients with normal psychophysical assessment, the proportion of patients with qualitative taste was higher compared to qualitative smell disturbances (53% vs 35%). When examining chemosensory patterns, disregarding temporal aspects and considering reports at any time during follow-up, the two most prevalent patterns consisted of: (a) self-reported smell and taste impairments without any qualitative alterations and normal SSET and TST results; and (b) quantitative impairment (either hyposmia or functional anosmia) together with parosmia and parageusia (Supplement, Table 1). Only parosmia and parageusia at any time showed an agreement (Cohen's: 0.52; p < 0.001).

TDI scores were significantly lower in women (22.85 ± 7.36 vs 27.15 ± 6.97; p = 0.050), and 20–40 age group (22.4 ± 6.92 vs 27.24 ± 7.52; p = 0.013). Patients hospitalized during the acute infection presented higher TDI scores (25.26 ± 7.58 vs 19.06 ± 4.49; p = 0.015). Conversely, TST scores were higher in women (12.11 ± 2.73 vs 10.4 ± 2.63; p = 0.030). Arthralgia (10.22 ± 3.14 vs 12.29 ± 2.26; p = 0.025) and taste impairment during the acute infection (11.33 ± 2.83 vs 14.0 ± 0.89; p = 0.033) were associated with lower TST scores (Supplement, Tables 2, 3).

SF-36 QoL assessment conducted close to the chemosensory evaluation (63 ± 50 days), was available for 46 patients. Patients experiencing phantosmia during follow-up showed a decrease in the physical functioning (50.72 vs 57.03; p = 0.022), role limitations due to physical health (47.05 vs 56.85; p = 0.018), and social functioning (43.20 vs 56.85; p = 0.002) scales. Patients reporting phantogeusia presented a significant reduction in the summary physical score (48.99 vs 58.18; p = 0.05), and role limitations due to physical health scales (32.37 vs 56.85; p = 0.012) (Supplement, Figs. 9–12).

Discussion

This study offers a comprehensive analysis of chemosensory impairment across the acute and lingering stages of COVID-19 disease. We analyzed the risk factors for acute smell/taste impairment, and for developing NCs-PCS. We also assessed the differences in the levels of serological and biochemical markers of the patients, and the impact on the quality of life. A subset of patients with NSc-PCS underwent a detailed chemosensory evaluation through the collection of qualitative smell and taste disorders and by means of validated psychophysical tests including for smell, besides identification, also discrimination and thresholds assessment. Overall, we confirmed the high prevalence of chemosensory impairment reported in literature20,21, as almost half of the patients reported a smell/taste dysfunction at any time point, and the possible persistence of symptoms up to 18 months. In a large proportion of patients, we observed a fluctuating trend over time of chemosensory symptoms. This may contribute to inaccurate prevalence estimates of NSc-PCS and undermine the reliability of cross-sectional studies.

Consistently with previously published data12, patients with NSc-PCS did not present an overall reduction in QoL at SF-36 questionnaire. However, an impact of some qualitative disorders on different SF-36 domains was observed, thus suggesting that qualitative alteration might influence specific aspects of QoL. Further research is warranted to enable a better measurement of the effect of the different types of chemosensory impairment on QoL.

In our cohort, being female and VoCs circulating during the second wave of the pandemic (when Delta variant was the dominating strain) were independently associated with a higher probability of acute chemosensory impairment. Conversely, older age was associated with a lower occurrence. Older age and early treatment for SARS-CoV-2 were confirmed to be associated with a lower occurrence of lingering chemosensory symptoms, while females were at higher risk of NSc-PCS. Pre-existing clinical conditions and COVID-19 severity did not show to increase the risk of long-lasting chemosensory impairment. These findings suggest that NSc-PCS is probably not driven by previous clinical conditions or by age, that patients with long-lasting smell/taste dysfunction are more often young and otherwise healthy individuals and, ultimately, that COVID-19 severity does not influence the probability to develop chemosensory impairment. As is recognized, older age is a risk factor for smell and taste reduction. In our cohort, its negative association with chemosensory impairment suggests that the underlying mechanisms for NSc-PCS are different from those responsible for age-dependent smell/taste deterioration.

Females elicit a stronger humoral and cellular immune response compared to men, probably due to sex hormones and genetic factors22, and local inflammation is reported to have a central role in the pathophysiological pathway leading to smell impairment15. Results from the present study and prior evidence12,20 encourage further research to investigate the impact of early therapy for SARS-CoV-2 on acute and chronic chemosensory symptoms. Moreover, the effects of other therapies (i.e. corticosteroids or antivirals) in reducing acute inflammation and preventing PCS deserve to be explored through randomized clinical trials.

As published previously12, data from this sub-study do not show that vaccination before infection has an impact on the prevalence of acute and long-lasting chemosensory symptoms. However, patients with NSc-PCS exhibit lower anti-RBD and neutralizing antibody titer compared to patients without PCS. In literature, the role of vaccination in preventing chemosensory symptoms is controversial. In a cross-sectional study, post-vaccination infections were less likely associated with loss of taste or smell23, while, in another study, chemosensory dysfunctions were observed both in unvaccinated and fully vaccinated individuals24. As per the antibody decay, evidence on the association of a particular serological trend with the occurrence of long-term consequences are inconclusive, adding this aspect to the long list of issues to be addressed.

NSc-PCS has a complex, fluctuating multifaceted presentation. A qualitative disorder was diagnosed in more than one third of normosmic patients self-reporting a smell impairment and in more than half of the normogeusic individuals. This contributes to inaccurate prevalence estimates of NSc-PCS and confirms that patients often fail in distinguishing between quantitative and qualitative disorders, as well as between smell and taste impairment25. Direct (taste receptor cell injury) or indirect infection mechanisms through inflammatory cytokines, as well as the implication of the signalling pathway or impaired taste bud cells renewal, were suggested to be involved in determining gustatory impairment26,27,28,29. Our results underline the need to adopt validated tests to fully quantify and characterize COVID-19-associated chemosensory impairments30.

At the phone-call interview, half of the patients who completed the training reported persisting smell impairment. The outcome of olfactory training was not assessed through psychophysical tests, hence patients reporting persisting dysfunction may have qualitative rather than quantitative impairment. The effectiveness of olfactory training is probably depending on training duration and aetiology of chemosensory impairment31, hence the outcome in patients with COVID-related smell dysfunction needs to be assessed in large cohort studies32.

This study has several limitations. Firstly, patients evaluated with psychophysical assessment may not be fully representative of the entire patient population. Patients experiencing long-lasting symptoms were probably more motivated to undergo further assessments, while, for the same reason, those asymptomatic were not assessed, thus we could not assess the accuracy of SSET and TST. Additionally, the timing of psychophysical assessment was not standardized. Therefore, this makes it hard to draw definitive conclusions.

However, most of the evidence available in literature presents self-reported data. COVID-19 is probably the first worldwide pandemic being characterised by such an extensive and accurate sharing of symptoms, supported by social media and telecommunication systems3. Self-reported symptoms can help collecting information but, in some cases, fail in giving an exhaustive picture of the problem. COVID-related olfactory and gustatory disturbances are often referred to as a mere reduction of the ability to perceive odours and tastes. However, the shades of different types of chemosensory impairments can vary considerably, and more detailed evaluations would be useful to better understand the problem. Validated tests were available well before COVID-19 but have been rarely used during the pandemic because they are time-consuming and could not be administered during acute infection and quarantine. In such circumstances, on-line questionnaires or phone-call interviews are more feasible, but subjective reports alone fail to measure the true prevalence of chemosensory impairment and to describe its diversity and complexity.

A very recent work by Sharetts et al. reported a nationwide post-Covid-19 study conducted in the US applying direct validated and self-administered taste and smell identification tests33. The mean time between Covid-19 onset and quantitative assessment was 395 days and they found no quantitative taste deficit while in one-third of individuals some smell loss was revealed, pointing out that smell loss could be the reason for taste complaint. A detailed focus on the long-term qualitative disorders is also meaningful especially for taste, considering that we found a qualitative disorder in more than half of the normogeusic individuals. Indeed, the detailed mechanism of COVID-19 dysgeusia remains unknown and taste qualitative disorders are independent symptoms that need to be discussed separately34,35. Therefore, there is still a particular need for well-designed long-term studies analysing the determinants and risk factors for post-COVID-19 chemosensory impairment, combined with the quantitative assessment and detailed qualitative description of its characteristics36.

We believe that this report, even if including a limited number of psychophysical evaluations, is of utmost importance to stress the utility of combining validated tests with self-reported symptoms in the context of post-COVID-19-related smell and taste dysfunction. This approach could fill existing gaps in NSc-PCS knowledge and may contribute to better understand post-infective chemosensory impairment.

Methods

Design of the study, definitions and population

This prospective monocentric cohort included patients with previous laboratory-confirmed SARS-CoV-2 infection enrolled at the University Hospital of Verona, within the ORCHESTRA long-term sequelae study, after a written informed consent was collected. As a sub-study, the present paper focuses on chemosensory evaluation through direct tests in patients diagnosed with NSc-PCS, as defined in the previously published study12 and Supplement. To estimate differences according to the methodology of detection of chemosensory impairment, two groups were identified: one including patients with self-reported chemosensory impairment (SRC) without any further evaluation, and the second including patients who underwent the chemosensory evaluation with a psychophysical assessment (PAC) (Fig. 1).

In addition to the time-points foreseen by the study protocol12, the dates of symptoms onset and end were extrapolated to account for exact duration of symptoms and recovery rates. A chemosensory evaluation, including an anamnestic interview and psychophysical validated tests, was conducted at the Neurology Unit of the University Hospital of Verona by qualified research staff. A subsequent follow up was performed through a phone-call interview.

To account for VoCs-related differences in chemosensory impairment5, the epidemiological wave recorded in Italy at the time of acute infection was specified according to Nextstrain SARS-CoV-2 resources website for Europe37 (Supplement, Definition).

The QoL was assessed through SF-36 questionnaire physical and mental component scores according to the study protocol12. Since the questionnaire was administered at different time points, the closest evaluation to the chemosensory assessment was selected.

Data were collected and managed using REDCap electronic data capture tool (Research Electronic Data CAPpture). The study was registered on ClinicalTrials.gov (CT registration number: NCT05097677) and the protocol is available at the ORCHESTRA website19. All investigations were carried out in accordance with the Helsinki Declaration and its later amendments. The protocol was approved by the Ethical Committee of the University Hospital of Verona (ORCHESTRA project-Prot. n. 3199CESC).

Chemosensory evaluation

Exclusion criteria for chemosensory evaluation were conditions known to affect smell and/or taste (e.g., recent traumatic event, otolaryngology disorders, stroke, neurodegenerative diseases). Qualitative disorders were assessed for both olfaction (i.e., phantosmia/parosmia) and taste (i.e. phantogeusia/parageusia). Intensity of perceptions was recorded by means of a Numeric Rating Scale (NRS) scale (0–10). Information was collected for nasal and oral chemesthesis too (i.e. nasal and oral trigeminal somatosensation, Table 5). The Sniffin’ Sticks Extended Test (SSET, Burghart, Germany) was applied for evaluating smell dysfunction. This is a validated olfactory test consisting of odour dispensing pens, forced choice paradigm based, including three subtests assessing odour threshold (T), discrimination (D) and identification (I), respectively. The sum of results obtained in each test defines the ‘‘TDI score’’, indicating the olfactory performance status of the subject. The score was interpreted according to scores adjusted for age, according to literature31,38: between 11 and 20 years the cut-off was 28.5, 21–30: 30.75, 31–40: 30.5, 41–50: 28.15, 51–60: 27.25, 61–70: 24.88, 71–80: 19.2 and >  = 81: 13.

An olfactory training was proposed to patients with smell impairment17,39,40,41.

Gustatory impairment was assessed through the Taste Strips Test (TST, Burghart, Germany), using filter paper strips impregnated with the four taste qualities in four different concentrations (sweet: 0.4, 0.2, 0.1, 0.05 g/ml sucrose; sour: 0.3, 0.165, 0.09, 0.05 g/ml citric acid; salty: 0.25, 0.1, 0.04, 0.016 g/ml sodium chloride; bitter: 0.006, 0.0024, 0.0009, 0.0004 g/ml quinine hydrochloride). Paper strips were placed on the tongue and patients were asked to close the mouth and move each strip on the tongue to check a whole gustatory sensitivity. Then the patient was asked to choose from a list of four tastes (sweet, sour, salty, bitter). The patient is required to rinse the mouth with water before each strip test. One point is assigned to each correct answer; normogeusia is defined for TST ≥ 9, while TST < 9 indicates hypogeusia. Complete ageusia is diagnosed in case of no sensation to the highest concentrations of all the four taste solutions. Taste strips qualities do not include umami taste, which has been found to be poorly conceptualised in European countries42.

Statistical analysis

Proportions and Fisher’s test were used for categorical variables. Median, quartiles and Kruskall Wallis tests were used for ordinal variables. A bivariable analysis of factors associated with acute and long-term self-reported chemosensory impairment was carried out. Odds ratios (OR) with 95% confidence interval (CI) and two-sided Fisher’s exact test’s p-value estimation corrected for multiple testing (Bonferroni correction) were computed with scipy (v.1.10) and statsmodels (v. 0.14.0)43. Variables significantly associated with the outcomes (corrected p <  = 0.05) were selected to compute the multivariable Generalized Linear Model with Binomial linking function. The negative class of the dataset was under-sampled with the Neighbourhood Cleaning Rule44. The fitting of the models was assessed by evaluating the discrimination power and generalizability of the area under the roc curve (AUROC). No missing data imputation was performed; hence only full data-vectors were used for the multivariable models. Time to event analysis and log-rank test for comparing symptoms duration between patients with self-reported chemosensory alteration was performed with STATA v.17. Concordance between self-reported chemosensory impairment and the results of psychophysical tests was assessed by computing Cohen-Kappa coefficients, and possible causes of discordance were investigated. Visualisations were carried out in Python3.11 using matplotlib package v. 3.7.1.

For laboratory values and SF-36 results, the Mann–Whitney test was applied for comparison of the distributions between patients with NSc-PCS and patients without PCS. For biochemistry assessment, outliers (datapoints falling outside the 1.5 interquartile range) were removed.