Previously independent patients with mild-symptomatic COVID-19 are at high risk of developing cognitive impairment but not depression or anxiety

Abstract

Objectives

The aim of the study was to explore the cognitive functions of a large sample of hospitalised subjects with mild symptomatic Coronavirus Disease (COVID-19) who were previously independent at home and without neurological diseases.

Methods

Patients admitted in a COVID-19 Unit for Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) infection between November 2020 and March 2021 were recruited. Inclusion criteria were: being independent at home before the infection, radiologically confirmed COVID-19 pneumonia, positive reverse transcriptase-polymerase chain reaction nasopharyngeal swab and no oxygen supplementation at the time of evaluation. Exclusion criteria: cognitive impairment or neurological diseases previous to the infection, delirium episodes, and history of any mechanical ventilation use. They were evaluated with Montreal Cognitive Assessment (MoCA), Hamilton Depression Rating Scale (HAM-D) and Hamilton Anxiety Rating Scale (HAM-A).

Results

Out of 522 subjects admitted in the COVID-19 Unit, 90 were enrolled [mean age = 68.32(11.99); 46M/44F]. An impaired MoCA (cut-off < 23) was found in 60 subjects (66.66 %). Pathological scores were obtained by 36.7 % of the subjects with <65 years and 78.3 % of those older than 65 years. A high prevalence of executive function and memory impairment was detected.

Conclusions

The results underline a high rate of cognitive impairment in previously independent mild COVID-19 patients. This might represent a potential threat for the everyday independence of these patients due to the consequences on everyday life activities and work following discharge from hospital. These subjects should, therefore, be monitored in order to allow a better understanding of the progression and consequences of the so-called “Long COVID”.

1. Introduction

During the current Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) pandemic, several extrapulmonary manifestations have been attributed to Coronavirus Disease (COVID-19) (Chen et al., 2020; Gupta et al., 2020). It has already been documented that human Coronaviruses has the capacity to enter the nervous system, provoking both central and peripheral pathologies (Paterson et al., 2020; Ritchie et al., 2020). Besides stroke, Guillain-Barré syndrome, myelitis and, more commonly, anosmia, loss of taste and headache (Paterson et al., 2020), cognitive impairment is another reported complication of COVID-19 (Hosp et al., 2021; Ritchie et al., 2020). A recent study demonstrated that persistent functional impairment due to persistent fatigue and difficulties in the workplace can be encountered in patients who experienced either a complicated or an uncomplicated SARS-CoV-2 infection (Jacobson et al., 2021).

Cognitive impairment in the COVID-19 subacute phase has also been reported in 18 out of 26 COVID-19 subjects with anosmia and/or loss of taste and with pathological findings upon neurological examination (Hosp et al., 2021). The cognitive dysfunction was associated with frontotemporal hypometabolism at Positron Emission Tomography (PET) test (Hosp et al., 2021).

Miskowiak et al. (2021) detected cognitive impairment after four months from SARS-CoV-2 infection in 19 out of 29 patients, with 83 % of them complaining about subjective cognitive difficulties in daily life.

According to these findings, cognitive dysfunction appears to be a further concern, representing a threat for the independence in everyday activities after COVID-19. In particular, COVID-19 survivors reported a high prevalence (38.9 %) of subjective difficulties whilst carrying out work-related activities (Jacobson et al., 2021).

To date, there are no data on cognition in individuals with mild COVID-19 in the acute setting. Thus, the aim of the study is to explore the cognitive function of a large sample of hospitalised subjects with mild symptomatic COVID-19, who were previously fully independent at home.

2. Methods

All consecutive patients admitted for SARS-CoV-2 infection between November 1st 2020 and March 31st 2021 to the COVID-19 Unit of the “Mons. L. Novarese” Rehabilitation Center (Moncrivello, Italy) were prospectively enrolled. The Unit was activated in a rehabilitation facility due to the pandemic emergency demands. Only patients with mild to moderate COVID-19 were referred to this COVID-19 Unit after adequate triage in the Internal Medicine Wards of the local Hospitals (Piedmont region).

Upon admission, a detailed medical history was obtained from all patients and a complete physical and neurological examination was performed.

Patients were selected for the study according to the following inclusion criteria: independence at home before the infection (Barthel Index of 100), radiologically confirmed COVID-19 pneumonia, positive reverse transcription-polymerase chain reaction (RT-PCR) nasopharyngeal swab, mild symptomatic COVID-19, no oxygen supplementation at the time of evaluation, having stabilized blood oxygen saturation (SpO2) values at rest and a stable clinical condition, being able to perform a 6-minute walking test.

Exclusion criteria were: diagnosis of cognitive impairment or pre-existing neurological diseases, delirium episodes or history of any type of mechanical ventilation (MV) during the COVID-19 acute phase.

Independence at home was assessed using the Barthel Index, with information being obtained from the patient's self-report and confirmed by phone interviews to the closest relatives for all subjects. The absence of previous cognitive impairment and/or neurological diseases was ascertained through the medical history, taken from the patients and confirmed by phone interviews to the closest relatives.

In particular, it was also investigated whether the patients had already undergone brain imaging for any reason. Furthermore, a neurological examination on admission was performed by a trained neurologist, in order to exclude the presence of current neurological symptoms (except for those attributable to SARS-CoV-2 infection, such as anosmia, loss of taste and newly reported headache) and unrecognized neurological diseases.

After inclusion, cognitive function was evaluated in all patients using the Italian version of the Montreal Cognitive Assessment (MoCA) (Nasreddine et al., 2005) by a trained neuropsychologist.

In order to determine the presence of depressive and anxious symptoms, the Hamilton Depression Rating Scale (HAM-D; pathological cut-off: 8) (Hamilton, 1960) and the Hamilton Anxiety Rating Scale (HAM-A; pathological cut-off: 17) (Thompson, 2015) were used, respectively.

Before and after the neuropsychological assessment, blood oxygen saturation (SpO2) and heart rate (HR) were determined.

Two different MoCA cut-offs have been proposed to define the presence of cognitive impairment: 26 (Nasreddine et al., 2005) and 23 (Carson et al., 2018). Additionally, the equivalent scores, i.e. the normative data stratified by age and education are available and have been validated in the Italian population (Conti et al., 2015; Santangelo et al., 2015).

These three different methods of classification were used in order to present the prevalence of cognitive impairment in COVID-19 patients.

The type of onset (respiratory/other — fever without dyspnoea, weakness, gastrointestinal symptoms) and the presence of neurological symptoms such as anosmia, loss of taste and headache were recorded.

The study was approved by the local Ethics Committee. All patients gave written informed consent before participation.

2.1. Tests and scales

2.1.1. Montreal Cognitive Assessment (MoCA)

MoCA consists of 12 subtasks exploring the following cognitive domains: (1) memory (score range 0–5), assessed by means of a delayed recall of five nouns after two verbal presentations; (2) visuospatial abilities (score range 0–4), assessed by a clock-drawing task (3 points) and by copying a cube (1 point); (3) executive functions (score range 0–4), assessed by means of a brief version of the Trail Making B task (1 point), a phonemic fluency task (1 point), and a two-item verbal abstraction task (2 points); (4) attention, concentration and working memory (score range 0–6), assessed by means of a sustained attention task (target detection using tapping; 1 point), a serial subtraction task (3 points), and forward and backward span tasks for digits (1 point each); (5) language (score range 0–6), assessed by a naming task with low-familiarity animals (3 points), repetition of two syntactically complex sentences (2 points) and the above-mentioned phonemic fluency task and (6) temporal and spatial orientation (score range 0–6), assessed by means of structured queries (6 points). MoCA total score range from 0 (worst performance) to 30 (best performance) (Nasreddine et al., 2005).

MoCA has a 90 % sensitivity in detecting mild cognitive impairment, which is much higher than that of the MMSE (18 %) (Nasreddine et al., 2005). Initially, Nasreddine et al. (2005) suggested a cut-off score of 26 to separate healthy controls from subjects with mild cognitive impairment. A further correction was applied accounting for the influence of education in individuals with 12 or fewer years of formal education, adding one point to obtain the total score. A meta-analysis (Carson et al., 2018) proposed to use a MoCA cut-off score of 23, in order to reduce the false positive rate and increase the diagnostic accuracy.

Furthermore, studies on MoCA have shown that diagnostic accuracy can be greatly enhanced by using normative data stratified by age and education, published for the Italian population, with correction grids permitting the transformation of raw scores into equivalent scores weighted for age, education and sex (Conti et al., 2015; Santangelo et al., 2015).

The adjusted scores were classified into five categories (equivalent score) endowed at least with an ordinal relationship (0 = score lower than the outer 5 % tolerance limits; 4 = score higher than the median value of the sample; 1, 2, 3 = intermediate score between the central value and pathology threshold on a quasi-interval scale) (Conti et al., 2015).

2.1.2. Barthel index

Barthel index assesses the ability of an individual with a neuromuscular or musculoskeletal disorder to take care of himself. It is composed of 10 items investigating the everyday life activities (feeding, bathing, personal grooming, dressing, toilet use, bladder and bowel continence, mobility and rising the stairs). A specific score is assigned to every item on the basis of the execution of the activity (autonomously, with partial or complete dependence). The score ranges from 0 to 100, where 0 is complete dependence in the activities of daily living and 100 is the complete autonomy in these activities (Mahoney and Barthel, 1965). This tool has been validated in a wide cohort of patients, including those admitted to Internal Medicine Wards (Galeoto et al., 2015), and it can be assessed also by phone (Della Pietra et al., 2011).

2.1.3. HAM-D

It is a clinician-based questionnaire with 17 items, each having a three (0–2) or five-point score (0–4). The higher the score of each item, the higher is the intensity of the depressive symptom, with 0 indicating no symptomatology. A total score ≤ 7 represents the absence of depressive symptoms, while higher scores specify mild (8–17), moderate (18–24) and severe (>25) depressive symptoms (Hamilton, 1960).

2.1.4. HAM-A

It is a clinician-based questionnaire consisting in 14 items, comprising psychological and somatic symptoms of anxious mood. Each item has a score ranging from 0 (no symptomatology) to 4 (severe anxiety), with a total sum of scores of 56. A total score ≤ 17 identifies mild anxiety symptoms, while a score ranging from 18 to 24 indicates moderate anxiety symptoms. A score between 25 and 30 indicates severe anxiety (Thompson, 2015).

2.2. Statistical analysis

Parametric tests were used for the analysis due to the normal distribution of data. Continuous variables were compared with the Student-t-test for independent or paired samples, as appropriate. Chi-square test was used to compare categorical variables. Correlations between variables were identified with Pearson correlation analysis and partial correlation analysis.

All calculations were performed with the computer program SPSS (Statistical Package for the Social Science) version 14.0 (SPSS Inc., Chicago, Illinois, USA).

3. Results

3.1. Patient characteristics

Out of 522 subjects admitted to our ward, we recruited 90 patients [46 males and 44 females; mean age (SD) = 68.32(11.99) years; mean days since the first positive RT-PCR nasopharyngeal swab 14.37(10.11)].

Sixty-two patients (68.9 %) had a pure respiratory onset with dyspnoea and low saturation, while 28 (31.1 %) experienced other types of onsets (fever without dyspnoea, weakness, gastrointestinal symptoms). Upon admission, 19 patients (21.1 %) had headache, 24 (26.7 %) anosmia and 20 (22.2 %) anosmia and loss of taste. More than half of the population (49 subjects-54.4 %) had arterial hypertension, 24.4 % had type 2 diabetes mellitus (22 patients) and 11.1 % atrial fibrillation (10 subjects).

Two subjects experienced mild anxiety [mean HAM-A score = 5.73(5.65)] and four mild depressive symptoms [mean HAM-D score = 5.29(4.99)]. The mean MoCA score was 20.18(5.38).

3.2. Frequency of cognitive impairment

At first, when considering a cut-off of 26 (Nasreddine et al., 2005), the prevalence of impairment in the entire population was 84.4 %, with percentages of 66.7 % and 93.3 % in the younger (<65 years) and older groups (>65 years), respectively (Table 1 ).

Table 1.

Clinical characteristics of the study population according to age groups (expressed as means and standard deviation); differences expressed as p value, effect size and relative 95 % CI).

Parameter	Population (n = 90)	<65 years (n = 30)	>65 years (n = 60)	p	Effect Size (95 % CI)
Age (years)	68.32(11.99)	53.73(4.81)	75.93(6.60)	0.0001a	−3.611 (−4.296 to −2.917)b
Length of Stay (days)	20.09(11.29)	16.13(8.66)	22.17(12.02)	0.017a	−0.550 (−0.998 to −0.099)b
SpO2 (%)c	94.71(2.12)	94.77(2.28)	94.58(2.00)	0.862a	0.039 (−0.399 to 0.477)b
HR (bpm)c	68.11(12.19)	66.43(13.82)	68.82(11.52)	0.359a	−0.206 (−0.645 to 0.234)b
SpO2 (%)d	95.08(3.06)	95.33(3.55)	94.86(2.81)	0.578a	0.125 (−0.314 to 0.563)b
HR (bpm)d	81.29(13.60)	80.83(13.89)	81.65(13.79)	0.824a	−0.050 (−0.488 to 0.388)b
ΔSpO2 (%)e	0.37(3.13)	0.57(3.06)	0.28(3.28)	0.671a	0.095 (−0.344 to 0.533)b
ΔHR (bpm)f	13.21(14.32)	14.40(15.32)	12.87(13.82)	0.580a	0.124 (−0.315 to 0.562)b
HAM-D	5.35(5.18)	4.90(5.30)	5.68(5.16)	0.558a	−0.067 (−0.519 to 0.308)b
HAM-A	6.01(6.01)	5.93(5.66)	6.10(5.61)	0.931a	−0.019 (−0.458 to 0.419)b
MoCA score	20.18(5.38)	23.50(4.42)	18.49(4.80)	0.0001a	1.002 (0.554 to 1.480)b
Cut-off 23g	58/32	11/19	47/13	0.0009h	0.160 (0.061 to 0.420)i
Cut-off 26g	76/24	20/10	56/4	0.0001h	0.143 (0.040 to 0.507)i
MoCA Eq. 3-4 (n)	31	10	21	/	/
MoCA Eq. 2 (n)	18	8	10	/	/
MoCA Eq. 0-1 (n)	41	12	29	/	/
Sub-itemsg
– Orientation	30/60	5/25	25/38	0.086h	0.345 (0.116 to 1.032)i
– Denomination	17/73	2/28	15/45	0.046h	0.214 (0.046 to 1.009)i
– Visuospatial ability	50/40	8/22	42/18	0.0002h	0.156 (0.059 to 0.415)i
– Executive functions	57/23	17/13	50/10	0.006h	0.262 (0.097 to 0.705)i
– Memory	73/17	23/7	50/10	0.446h	0.657 (0.222 to 1.944)i
– Attention	37/53	10/20	27/33	0.289h	0.611 (0.245 to 1.524)i
– Fluency	61/29	14/16	47/13	0.006h	0.239 (0.089 to 0.639)i
– Calculation	41/49	9/21	32/28	0.036h	0.375 (0.148 to 0.952)i
Comorbidities (yes/no)
– Atrial fibrillation	10/80	3/27	7/53	0.813h	1.189 (0.285 to 4.966)i
– Hypertension	49/41	11/19	38/22	0.017h	2.983 (1.202 to 7.408)i
– T2DM	48/42	5/25	43/17	0.225h	1.977 (0.650 to 6.012)i

SpO2 = blood oxygen saturation; HR = heart rate; MoCA = Montreal Cognitive Assessment; HAM-D=Hamilton Depression Rating Scale; HAM-A = Hamilton Anxiety Rating Scale; T2DM = type 2 diabetes mellitus.

^aStudent t-test (significance for p < 0.05).

^bCohen's d and relative 95 % confidence interval.

^cBefore MoCA.

^dAfter MoCA.

^eΔSpO2 = SpO2 after MoCA-SpO2 before MoCA.

^fΔHR = HR after MoCA-HR before MoCA.

^gNumber of subjects with pathological score/number of subjects with normal score.

^hChi-squared test (significance for p < 0.05).

ⁱOdds ratio and relative 95 % confidence interval.

When applying a cut-off of 23 (Carson et al., 2018), 58 (64.4 %) subjects showed a MoCA score < 23 and 32 (35.6 %) showed a MoCA score ≥ 23.

When considering MoCA sub-items in the whole study sample, the alteration of visuospatial abilities, executive functions, memory and fluency accounted for 55.5 %, 63.3 %, 81.1 % and 67.8 % of impaired domains, respectively.

MoCA Equivalent Scores were calculated: scores of 0 and 1, corresponding to a severe cognitive impairment, were found in 45.5 % of patients (age < 65 years: 40 %; age > 65 years: 48.3 %). A score of 2 (borderline cognitive function) was found in 20 % of the whole population (age < 65 years: 26.7 %; age > 65 years:16.7 %), while scores of 3 and 4, highlighting a normal cognitive function, were present in 34.4 % of the population (age < 65 years: 33.3 %; age > 65 years: 35 %) (Fig. 1 ).

Fig. 1.

Distribution of MoCA equivalent scores in the study population, divided by age groups.

3.3. Cognitive impairment group comparison

The sample was divided according to a MoCA cut-off of 23 (MoCA<23 = Impaired Group — IG; MoCA≥23 = Non-impaired Group — NIG).

The IG and the NIG differed for age [72.33(10.85) vs 61.06(10.58); p = 0.0001] and, as expected, for MoCA score [17.31(4.39) vs 25.41(2.06); p = 0.0001], but not for SpO2 and HR before and after MoCA, ΔSpO2, ΔHR, HAM-D and HAM-A scores.

No differences in terms of type of onset, headache, anosmia and anosmia/loss of taste frequencies were detected in the IG and NIG. When considering comorbidities, the two cohorts were significantly different only for hypertension prevalence (p = 0.017).

3.4. Age group comparison

The study population was also divided for age, on the basis of the conventionally used cut-off of 65 years (<65 years = 30 subjects; >65 years = 60 subjects) (World Health Organization, 1999). The two age groups differed in terms of MoCA scores, which were higher in the younger group, and for days of hospital stay, higher in the older group. The older group showed statistically significantly higher frequencies of impairment in denomination, visuospatial ability, executive functions, verbal fluency and calculation ability, but not in orientation, memory and attention (Table 1).

The two cohorts differed for hypertension prevalence, higher in the older group, while they were similar with respect to type 2 diabetes mellitus and atrial fibrillation frequencies. When considering a MoCA cut-off score of 23, an impaired cognitive function was present in 11 out of 30 (36.7 %) patients of the younger age group and in 47 out of 60 subjects in the older (78.3 %) group. The younger group showed an impairment in executive functions (56.7 %), in attention (33.3 %), in memory (76.7 %) and in fluency (46.7 %). In subjects >65 years, memory and executive functions were compromised in 50 out of 60 patients (83.3 %). Orientation (41.7 %), visuospatial ability (70 %) and fluency (78.3 %) were also affected (Table 1).

3.5. Correlations

Age was correlated with MoCA scores (p = 0.0004; r = −0.535) in the whole population and in the IG (p = 0.002; r = −0.399), without correlation in the NIG (p = 0.329; r = −0.185).

In the entire study sample, age was also correlated with MoCA sub-items (temporo-spatial orientation p = 0.0004, r = −0.446; denomination ability p = 0.033, r = −0.226; executive functions p = 0.0002, r = −0.456; visuospatial ability p = 0.001, r = −0.331; memory p = 0.045, r = −0.212; verbal fluency p = 0.001, r = −0.347; calculation ability p = 0.005, r = −0.291), but not to attention (p = 0.227, r = −0.129).

Furthermore, the length of hospital stay was linked with age (p = 0.0003; r = 0.405) and MoCA (p = 0.008; r = −0.281) in the entire population. The latter relation was not detectable after correcting for age (p = 0.512; r = −0.072). In contrast, when adjusting for MoCA, age and length of stay were still correlated, although modestly (p = 0.003; r = 0.312). No other correlations of interest were found, in particular neither SpO2 nor HR were associated with MoCA, nor with age.

3.6. SpO2 and HR before and after MoCA

When comparing SpO2 and HR determination before and after MoCA, HR increased significantly (p = 0.0004), while SpO2 did not change (p = 0.270) in the whole population, nor in the IG (p = 0.520; p = 0.0005, respectively) or in the NIG (p = 0.319; p = 0.0003, respectively).

4. Discussion

Currently, a substantial part of the scientific and clinical efforts of the healthcare system is focused on the management of the acute COVID-19 complications. However, attention to the subacute or chronic sequelae will play an essential role in the comprehension of the cognitive consequences of SARS-CoV-2 infection.

4.1. Key findings

In this study, conducted in an acute Internal Medicine Ward, it has been demonstrated that cognitive impairment can be detected from the first hospitalization for COVID-19, even in those patients suffering from mild forms of the disease.

When considering a MoCA cut-off of 26 (Nasreddine et al., 2005), 84.4 % of patients were classified as cognitively impaired, with rates of 66.7 % in the younger and of 93.3 % in the older age group.

According to the cut-off of 23 (Carson et al., 2018), a cognitive impairment was present in approximately two thirds of previously independent subjects affected by mild COVID-19. This was more common in those older than 65 years, who are probably more susceptible to a decrease in cognitive function when compared to the younger COVID-19 patients. In this respect, it is of note that aging is obviously a risk factor for cognitive impairment also in the general population (Iadecola and Gottesman, 2019). In the reported sample, a cognitive impairment could be found not only in the older cohort (78.3 %), but also in those younger than 65 years (36.7 %).

Using the classification of the equivalent scores (Conti et al., 2015; Santangelo et al., 2015), 41 subjects (45.5 %) in the whole population showed a severe cognitive dysfunction, having a MoCA Equivalent score of 0–1 [12 subjects out of 30 (40 %) in the younger population and 29 individuals out of 60 (48.3 %) in the older group]. A MoCA Equivalent Score of 2, which indicates a borderline cognitive function, was instead found in the 20 % of the entire sample of COVID-19 patients.

The most adequate MoCA cut-off for classification of COVID-19 is currently debated and its definition is beyond the aims of this study. We applied three different classification methods (on the basis of the available literature) and, regardless of the used classification, a high percentage of subjects with mild COVID-19 proved to be cognitively impaired.

We chose to divide the population according to the cut-off of 23, as a cut-off of 26 might have led to a higher false positive rate, being inclement in older age and in persons with a lower educational level (Carson et al., 2018; Yeung et al., 2020).

4.2. Comparison with other studies

When using a cut-off of MoCA of 23, the prevalence of cognitive dysfunction in our sample is in line with that described in the paper by Miskowiak et al. (2021), in which cognitive impairment was detected in 65 % of subjects (29 patients tested), although assessed during a more chronic phase. Another recent study (Hosp et al., 2021) reported cognitive impairment on the basis of the MoCA cut-off of 26 in 18 (69.2 %) out of 26 COVID-19 patients in the subacute phase. This frequency of cognitive impairment was lower than the frequency calculated in our sample with the same cut-off of 26 (84.4 %). Hosp et al.(Hosp et al., 2021) selected subjects according to the presence of at least one symptom amongst reduced olfaction and/or gustation and demonstrated pathological frontoparietal hypometabolism through ¹⁸Fluorodeoxyglucose (¹⁸FDG) PET Imaging.

Thus, a crucial finding of the two cited papers (Hosp et al., 2021; Miskowiak et al., 2021) was the alteration of the fronto-temporal functions in the cognitively impaired patients.

In our sample, a large part of the population experienced an impairment of executive functions, visuospatial ability, memory and fluency, suggesting that the cognitive impairment is not only attributable to a frontotemporal dysfunction, but to a more general reduction of the cerebral functions. This alteration was more evident in the older cohort where also orientation and calculation were affected. Another noticeable finding of our study was the absence of association between olfactory and gustative alterations or headache and decline of cognitive functions, which has been previously reported (Nalbandian et al., 2021).

Unexpectedly, neither depression nor anxiety were found in the study population, although a higher prevalence of these diseases could be expected (Nalbandian et al., 2021). In a recent systematic review (Renaud-Charest et al., 2021), more than 50 % of the COVID-19 population presented different degrees of symptoms related to anxiety and depression, while symptoms of fear, worry and anticipatory anxiety were described in more than 70 % of the population. The reported frequency of depressive symptoms following COVID-19, registered after 12 weeks or more, ranged from 11 to 28 % and the frequency of clinically significant depression and/or severe depressive symptoms ranged from 3 to 12 %. However, studies were highly heterogeneous with respect to time and mode of ascertainment, location and age of patients (Renaud-Charest et al., 2021). Furthermore, they were all conducted after 12 weeks or more since COVID-19, in contrast with our population of acute inpatients.

Thus, the study population itself might be an advocate for a lower prevalence of depressive and anxiety symptoms. It might be likely that those who are discharged and return to everyday life (in the post-acute/chronic phase) might experience more depressive symptoms when encountering difficulties in work or other activities. Conversely, when considering the prevalence of anxiety symptoms, they could be low as the enrolled subjects were clinically stable and ready to be discharged.

4.3. Strengths and limitations

The findings of the study are corroborated by the enrolment of an accurately selected population.

Only individuals who were independent at home were recruited and those who were previously diagnosed with neurological diseases and/or dementia were not eligible. In addition, the cognitive tests were administered when the subjects were in a stable clinical condition and with a stable SpO2. This excludes the potential influence of hypoxemia on cognitive performances at the time of evaluation. In particular, this might explain the lack of correlation between SpO2 levels and cognitive function and of changes before and after cognitive testing, as the large majority of the population had a stable SpO2 > 90 %.

According to these data, it is difficult to exclude a direct or indirect role of SARS-CoV-2 infection in the genesis of cognitive impairment. However, when discussing the specificity of the cognitive impairment, a limitation might be represented by the lack of a control group affected with other respiratory illnesses in an acute care hospital.

The influence of length of hospital stay on MoCA was also investigated, due to the possible role of prolonged physical inactivity and hospitalization in the development of cognitive dysfunctions, especially in older subjects (Carda et al., 2020). However, no correlation was found in neither the younger nor the older cohorts.

The main limitation of the study is that pre-existing impaired autonomy and cognitive impairment could not be formally ruled out, as none of the patients had been evaluated accordingly before the infection. Self-sufficiency before COVID-19 was determined on the basis of the medical history and through the administration of the Barthel Index, a highly validated instrument in the determination of individual autonomy (Galeoto et al., 2015). In this respect, it must be noted that Barthel Index can be administered by phone (Della Pietra et al., 2011) and a further confirmation of the patients' independence was obtained in this way (due to the COVID-19-related restrictions) by interviewing the patients' closest relatives.

In parallel, the presence of previous or current neurological diseases was principally ruled out through history taking and the interview of relatives, in addition to an accurate neurological examination performed on admission.

Another point of note is that our patients were in a clinically stable condition at the time of evaluation and that, formerly, they were ready to be discharged to their homes, awaiting a negative nasopharyngeal swab.

4.4. Implications

In October 2020, our Neurorehabilitation Unit was converted into an Internal Medicine ward to face the pandemic and the admitted COVID-19 patients were assisted by personnel already trained in recognizing and assessing cognitive alterations. These considerations represent the “rehabilitative point of view”, which is not currently applicable in every setting where COVID-19 subjects are admitted in the most acute phase. Our study suggests that a cognitive assessment should be included in the evaluation of all COVID-19 patients, since the time of admission, in every department involved in their acute and sub-acute care, even if only mildly symptomatic and regardless of anosmia, headache or loss of gustation.

With regard to the type of cognitive examination, despite MoCA certainly represents a suitable screening tool, it might be less useful for the detection of more subtle cognitive alterations, in particular in the younger group and in those with a better cognitive performance, due to the known ceiling effect. Thus, probably, a more specific cognitive evaluation should be implemented in these patients.

In an ideal world, personnel trained in recognizing and treating cognitive dysfunction, such as rehabilitation physicians, neuropsychologists and occupational therapists, should be involved in the COVID-19 team.

5. Conclusion

According to these findings, including the high prevalence of executive function and memory impairment (Table 1), it may be hypothesized that mild COVID-19 patients might not go on to be completely independent at home after discharge, and that they may possibly have negative consequences.

In fact, a study reported that 11.5 % of patients missed work and that 38.9 % experienced an impairment in work-related activities after 4 months from the acute disease (Jacobson et al., 2021). This constitutes a matter of crucial importance, as younger patients in the productive age represent the work force in the immediate “post-pandemic world”, which might feasibly suffer more severe socioeconomic consequences than those which have already become manifest.

Due to the cross-sectional nature of the study, it is not possible to draw conclusions on the long-term effects of COVID-19 on cognition, although it has been reported that cognitive impairment has been detected until 3–4 months after COVID-19 (Miskowiak et al., 2021).

In this cohort, periodical monitoring of the enrolled subjects is currently undergoing, with the help of the patients and their relatives. Follow-up studies are ongoing to evaluate the evolution of the cognitive dysfunction and the possible consequences of the so-called “long-COVID” in these patients (Nalbandian et al., 2021).

CRediT authorship contribution statement

GG, MFG and CS conceived the study and designed the study protocol. GG, MC and MS recruited the subjects. MFG analyzed the results. GG, MFG, FD and CS wrote the manuscript. All authors have read and approved the final version of the manuscript.

Ethical standards statement

The study was approved by the local Ethics Committee and was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

Funding

None.

Acknowledgements and role of funding

This study has not received funds.

Conflict of interest

The authors declare no conflict of interest relevant to this paper.