COVID-19 in Adults With Dementia: Clinical Features and Predictive Factors of Mortality. A Clinical Cohort Study on 125 Patients

BACKGROUND There is limited evidence on the characteristics and outcome of patients with dementia hospitalized for novel coronavirus infection (COVID-19). METHOD We conducted a prospective study in 2 gerontologic Covid Units in Paris, France, from March 14 th 2020 to May 7 th 2020. Patients with dementia hospitalized for conrmed COVID-19 infection were systematically enrolled. A binary logistic regression analysis was performed to identify factors associated with mortality at 21 days. RESULTS We included 125 age was 86 (IQI 82-90); female. Most common causes of were had ≥ 2 40.2% in a long-term The most at were and delirium (82.4%), asthenia (76.8%) and (72.8%) before polypnea (51.2%) and desaturation (50.4%). Falls were frequent at the initial phase of the disease (35.2%). The fatality rate at 21 days was 22.4%. Chronic kidney disease and CRP at admission were independent factors of death. Persisting confusion, mood and behavioral disorders were observed in survivors (19.2%).


Introduction
On January 30 th 2020, the World Health Organization (WHO) drew attention to a new coronavirus disease 2019 (COVID- 19) declaring it a public health emergency of international concern. As of February the 4 th 2021, there had been more than 100 millions of cases and 2 millions deaths worldwide. The clinical spectrum of COVID-19 infection appears broad, encompassing asymptomatic infections, mild upper respiratory tract illness and severe pneumonia with respiratory failure, systemic complications and multi-organ failure [1].
Severe COVID-19 affects elderly with chronic diseases, including cognitive decline, in high proportion compared to general adult population [2,3]. According to recent studies, dementia is a major risk factor for COVID-19 severity [4,5]. Concomitantly, the risk of exposure to the infection is more important in patients with dementia, highly exposed in the context of long-term care facilities, frequent hospitalizations and intellectual decline. So far, speci c clinical features and prognostic factors of COVID-19 in demented patients remain unclear [6,7].
In order to identify speci c features and risk factor of death in demented people, we report a cohort study on 125 patients with dementia hospitalized for a con rmed COVID-19 infection.

Study design
This prospective cohort study systematically included patients over 65 with dementia hospitalized for COVID-19 infection in two centers : Leopold Bellan Hospital,Paris, France and Lariboisère Fernand Widal University Hospital Paris, France, between March 14 th and May 7 th 2020. The de nite diagnosis of COVID-19 was determined through reverse-transcription polymerase chain reaction (RT-PCR) testing of pharyngeal swabs or chest CT. We collected demographics, medical history, clinical presentation, laboratory results, treatment, complications and outcome. The outcome was death at 21 days, to assess short-term case fatality rate. Dementia was de ned according to the American Psychiatric Association's Diagnostic and Statistical Manual (DSM-5) as a signi cant cognitive decline from a previous level of performance in one or more cognitive domains interfering with independence in everyday activities [8]. Etiology of the dementia was collected when available. Comorbidities were evaluated using the Charlson Index [9]. Blood oxygen desaturation was de ned by a saturation under 93 % or a loss of 3% or more. Acute kidney injury was diagnosed according to KDIGO de nition [10]. Cardiac injury was diagnosed clinically or throughout, abnormalities observed on electrocardiography and serum level of troponin and Brain natriuretic Peptide (BNP). COVID-19 severity and Acute Respiratory Distress Syndrome were de ned following the WHO recommendations. None of the patients included in this study were deemed candidates for the intensive care unit after staff discussions with the physician in charge, palliative care team and patients and caregivers.

Statistical analysis
Descriptive data are shown as median (interquartile interval) or percentage (number of subjects). Chisquare test or Fisher test was used to compare qualitative data between groups. Mann Whitney test and Kruskal Wallis were applied to analyze non-normally distributed data. A binary logistic regression analysis was performed to identify clinical and demographic characteristics associated with mortality. We included the variables that had differed in descriptive analysis between survivors and non survivors.
Variables with missing data were excluded from this analysis. A two-tailed p-value < 0.05 was considered signi cant. Statistical analysis was performed using GraphPad Prism and SPSS 85 Statistics version 26.0.
Ethical statement: we obtained the required approval from the Commission Nationale Informatique et Liberté (CNIL) to collect anonymized data.

Results
We included a total of 125 patients. Patients demographic and clinical characteristics are summarized in Table 1. The median age was 86 (82-91) year-old (yo), and 41.6% of included subjects were male. All patients had a clinical diagnosis of dementia. 40 % of the patients had received a speci c diagnosis: most frequent cause of dementia was Alzheimer's Disease (AD) (10.4%), vascular dementia (7.2%), multiple etiologies dementia (7.2%). Parkinson's disease and atypical parkinsonian syndromes, frontotemporal dementia alcohol related cognitive impairment and psychosis accounted for other diagnoses. Patients frequently presented with associated mood disorder (31.2%) and psychotic symptoms (12%). A stroke history was noted in 32% of patients. A former history of fall was reported for 38.4% of patients. Sixty percent of patients received psychotropic treatment: antidepressants/mood regulators (35.2%), anxiolytic (30.4%) and neuroleptics (12.0%). Most patients were living at home (59.2%) while 40.8% were living in a long-term care facility.

Clinical presentation and biological ndings
Infection was con rmed in 93.2% of patients by PCR test and in 6.8% by chest CT ( Table 2). The most common initial symptoms were confusion and delirium observed in 82.4% of patients. It was notably the only symptom at onset and during the evolution of the disease for three patients in our cohort, thus COVID-19 infection was screened for considering the pandemic context. The most frequent associated general symptoms were asthenia (76.8%), fever (72.8%) and anorexia (56.8%). Most frequent respiratory symptoms were: polypnea (51.2%), desaturation (50.4%) and cough (49.6%). An initial fall was reported in 35.2%. No patient complained of loss of taste or smell.
The most common biochemical abnormality was lymphopenia (84.8%) with a median value of lymphocytes of 0.99 (0.73-1.29), increased CRP at admission and low albumin. Thrombopenia was observed in 16% of cases. Troponin was available for 22 patients and increased in 35.5% of patients.
Mortality at 21 days was 22.4%. The large majority of deaths (92.9%) were attributed to ARDS. Cardiac failure and osmolar coma were associated causes of death. Non-survivors were overrepresented in men (60.7 % versus 36.1%in survivors, P=0.029), presenting more frequent chronic kidney disease. Respiratory distress, lower lymphocytes count, high CRP and positive troponin were more frequent in non-survivors. No statistically signi cant difference was observed in the rate of death between patients with a diagnosed neurodegenerative disorder compared to the rest of the population (P=1.000) or in those living in long-term facilities (P=0.663).
Male sex, chronic kidney disease, desaturation, dyspnea as well as lymphocytes count, CRP, and serum creatinine at admission were included in a multivariate binary logistic regression model to identify associated factors of death (Table 4). Two variables were independently associated to death at 21 days. CRP at admission: (OR=1.013, P=0.004) and chronic kidney disease (OR=4.631, P=0.025).

Discussion
In this study, we aimed at describing speci c clinical features and prognostic factors of mortality in demented patients. Confusion with or without general symptoms is the most frequent initial presentation and was not a predictive factor of death while history of chronic kidney disease and CRP level at admission were signi cantly associated with mortality.
Studies focusing on the neurological features of COVID-19 suggest that confusion occurred in 20-30% of hospitalized patients increasing to 60-70% for severe forms [11,12].. Older adults are more prone to experiencing confusion and delirium and dementia is the higher predisposing risk factor (OR from 2.3-4.7), before age, visual and hearing impairment, and polypharmacy [13,14]. Several series and cases reports highlight COVID-19 infection presenting as isolated and persistent confusion [15][16][17]and as a factor of negative outcome [18]. In our serie, confusion & delirium occured independently of the severity of the COVID-19 infection and did not appear as a factor of negative outcome. The discrepancy between all these results could be explained by the high prevalence of confusion in our cohort that might have not allowed to study his weight on prognosis.
Overall, the prevalence of confusion and delirium was higher than the one reported in older adults admitted with non-COVID-19 pneumonia [19,20] and in patients admitted with severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) [21].
The pathophysiological explanations remain unclear. COVID-19 could have a direct neuronal toxicity through CNS invasion [22]. Cognitive symptoms could also indirectly be related to neuroin ammation, corresponding to "sickeness behavior" to which demented subjects have been shown to present with increased vulnerability [23]. An increase of neuro laments lights and Glial Fibrillary Acid Protein, respectively re ecting neuronal injury and glial activation has been observed in patients with moderate and severe COVID-19 with or without dementia [24]. The neuronal and synaptic fragility in demented patients may be particularly prone to injury induced by COVID-19, either through direct infectious lesion or through indirect in ammatory mechanisms. Another frequently reported neurologic symptom in our series is falls, as already reported [5,25]. General risk of fall is high in demented older adults; gait impairment and falls are more prevalent in dementia than in normal aging and are related to the severity of cognitive impairment [26]. Rupture of homeostasis in the context of viral infection would account for their frequent occurence [27].
Beside confusion, delirum and falls, a braod range of neurological complications, caused directly or indirectly by the virus, including infectious, para-infectious, and post-infectious encephalitis, stroke related to coagulopathy, and acute neuropathies have been reported [28].
Stroke and seizure were observed during disease evolution for respectively 3 of our patients, at a similar rate as the one observed in cases of a cohort of around 5000 subjects [29].Case fatality rate at 21 days was 22.4%, in line with previous ndings in a large cohort [2,5]. A study by Canavelli et al [30] evaluating the prevalence of dementia in a random sample of con rmed COVID-19 infected patients, found that patients with dementia accounted for 15.8% of overall COVID-19 related death. In a meta-analysis of, the mortality of individuals with dementia was increased compared to not demented subjects (OR=5.17) [31].
More speci cally, Matias-Giu et al have shown that AD patients showed a higher risk of death in COVID-19 than patients with fronto-temporal dementia. One explanation could be that the APOE e4 genotype (the highest risk factor for AD) has been reported as a predictive factor of severe COVID-19 [19] and death [20]. In our series, we did not nd any difference between patients with a diagnosed dementia (AD, vascular dementia, alcohol related cognitive impairment) and patients with unexplored cognitive impairment.
History of chronic kidney disease and high CRP at admission were independently associated with death. Our overall data are in accordance with previous reports in which male sex, multiple comorbidities, elevated CRP and low lymphocyte count were observed in the majority of COVID-19 deaths [3,32]. None of our patient was deemed suitable candidate for ICU or invasive ventilation. Patients received variable treatment, associating antibiotherapy, rarely antiviral treatment, high ow oxygen and corticosteroids or immunomodulatory drugs for a small number. Given the size of our cohort, we were not able to assess e ciency or draw any recommendation. Therapeutic essays have been ongoing regarding the potential bene cial effect of serotonin reuptake inhibitor (SSRI) through modulation in ammatory response during sepsis [33]; no difference in outcome was observed regarding treatment by antidepressant in our cohort.
All in all, in the actual context of a second wave of COVID-19, this work demonstrates that special attention should be given to demented patients who manifest confusion with acute behavioral changes and falls, with or without asthenia, fever and lymphopenia.

Limitations
This study has several limitations. Our studied population was drawn from geriatric hospitalized samples of demented patients in need of hospital care, limiting its generalization. Patients presenting with atypical symptoms and lacking respiratory symptoms or fever may not have been identi ed as COVID cases and included in our analysis. Due to low group numbers, we could not assess the difference in severity and mortality between the different etiologies of dementia. We focused on short-term mortality; long term follow-up could allow us to identify additional prognostic factors.

Conclusion
In our cohort of individuals with dementia, we have observed that COVID-19 was characterized by atypical presentation with prevalent nonspeci c symptoms at the initial phase of the disease, namely persisting confusion and behavioral disorders as well as frequent falls. Early recognition of COVID-19 in demented adults should help provide early treatment and adequate care and isolation and COVID-19 testing should be considered in front of any signi cant change from baseline.

Declarations Ethics approval and consent to participate
We obtained the required approval from the Commission Nationale Informatique et Liberté (CNIL) to collect anonymized data.

Consent for publication
Not applicable.

Availability of data and material
The full datasets used during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests