Study design and setting
This was a multicenter cohort study of patients in three acute geriatric units (AGUs) from Assistance Publique Hôpitaux de Paris – Sorbonne Université (APHP-SU) hospitals in Paris, France. We included patients aged 75 years and older who were hospitalized in these AGUs for COVID-19 as the main diagnosis or another medical reason from January 1 to June 15, 2021 and who were discharged alive from the hospital. Eligible patients were identified retrospectively from medical records. Follow-up was conducted from September 2022 to February 2023, 18 months after hospitalization. This report follows the STROBE recommendations (Supplementary Methods 1).
moral support
This study was approved by the ethics committee (CPP Ile de France, Paris, France, number 107–2021). All included patients or their close relatives received an information letter specifying their rights and the conditions for the use of their medical data. No objection certificates were collected by the physicians in charge of the patients.
Participants
Included patients were aged 75 years or older. They were transferred to the AGU from the emergency department or intensive care unit. COVID-19 positivity had to be diagnosed by RT-PCR and/or chest CT for SARS-CoV-2, as per WHO interim guidance. [18]Patients were excluded from the scheme if they died while hospitalised, if they were under legal guardianship or if they refused access to their medical data.
data collection
One physician from each ward (MC, LB, AR) collected medical data retrospectively from computer medical records. The data included sociodemographic information (age, sex, place of living), clinical data such as co-morbidities along with the Charlson Comorbidity Index (CCI). [19]Frailty with the Clinical Frailty Scale (CFS) [20]Functional Independence Score of Activities of Daily Living (ADL) [21] Instrumental and Agile Activities of Daily Living (IADL) [22] and polypharmacy, defined as taking five or more chronic medications daily [23]We recorded descriptive data for hospitalisation including principal diagnosis at admission, clinical severity on admission (rapid sepsis-related organ failure assessment) [qSOFA] score [24]) and laboratory data (hemoglobin level, lymphocyte and neutrophil counts, and C-reactive protein [CRP]creatinine and albumin levels). We also collected information on where patients were discharged from hospital (to home, to a rehabilitation service, to other departments of the hospital or admitted to a nursing home).
One physician per ward (MC, LB, AR) followed up the participants via phone calls over 18 months. When the patient did not respond on several occasions, close relatives or a general practitioner were contacted. When no contact data were available or no response was received despite three phone calls, patients were considered lost to follow-up. Data collected at 18 months included patient vital status, new admission to nursing home, hospital readmission, frailty and functional independence (CFS, ADL, IADL scores) and quality of life with the EQ-5D-5L (mobility, independence, daily activities, pain and anxiety/depression scores from 0 to 4 and global assessment of health score from 0 to 100). [25],
Result
The primary outcome was 18-month mortality and associated factors. Secondary outcomes were functional autonomy with ADL and IADL, frailty with CFS, quality of life with EQ-5D-5 L, and readmission rate at 18 months.
Statistical analysis
Demographic data and baseline characteristics are described according to COVID-19 status for all patients. Missing values are specified only if they exist. Data are presented as mean (SD) or median (interquartile range) [IQR]) for continuous variables and numbers (percentages) for categorical variables. Unpaired student-tests were used to compare quantitative variables between hospitalized and non-hospitalized patients with COVID-19. Tea t test or Mann-Whitney test for non-normally distributed data. Normality was assessed by graphical representation of data distribution. Comparison of categorical variables involved the chi-square or Fisher's exact test as appropriate.
Our primary endpoint, death at 18 months, was described as a denominator variable with Kaplan-Meier curves. All included patients with available 18-month follow-up data were analyzed. Comparisons between the two groups involved the log-rank test. For adjusted analysis, we used Cox regression models studying the association between COVID-19 positivity and mortality at 18 months, estimating adjusted hazard ratios (aHR) and 95% confidence intervals (CI). Adjustment factors were selected when they were clinically significant or important P< 0.2 on univariate analysis [26]The proportional hazards assumption was respected.
Secondary endpoints of re-hospitalisation, loss of autonomy (ADL and IADL scores at 18 months and the ratio between 18 months and baseline), quality of life (EQ-5D-5 L score at 18 months) and frailty score (at 18 months and the ratio between 18 months and baseline) were compared with univariate statistical methods (Wilcoxon–Mann Whitney test for quantitative variables, chi-squared or Fisher's exact test for categorical variables).
Statistical analyses were performed with RStudio 2023.06.0+421. All p-values were two-tailed and P< 0.05 was considered statistically significant.