The COVID-19 cohort was built from the Senior-COVID-Rea study. The Senior-COVID-Rea study was a retrospective and prospective multicenter study on health data, carried out in 7 ICUs in the Auvergne-Rhône-Alpes region (France) for patients admitted between March 1, 2020 and May 6, 2020 during the first wave of the COVID-19 pandemic. The study protocol (V1.0 of April 7, 2020) was approved by the Ethics Committee of the Hospices Civils de Lyon on May 12, 2020 (IRB number 20_025) and declared on the ClinicalTrials platform on June 9, 2020 (NCT04422340). The detailed protocol was published elsewhere . All patients over 60 years old, admitted to ICU in the participating centers during the study period with a COVID-19 diagnosis confirmed by positive SARS-CoV-2 PCR on nasopharyngeal or lung swabs were included in this cohort. In several centers, patients were contacted between days 173 and 187 after the day of ICU admission by a routine post-ICU teleconsultation (day 180 follow-up). In the centers routinely performing post-ICU teleconsultation, the data collected were incorporated in the Senior-COVID-Rea database.
The control cohort was built from the FRAGIREA study, which was a multicenter prevalence study on the frequency of frailty and the management of older patients . The study was approved by the French data protection agency (Commission nationale de l’informatique et des libertés) and by an ethics committee (Comité de Protection des Personnes Ouest II IRB Number 17.11.66). The protocol was submitted to clinicaltrial.gov (NCT03326635). The study was conducted in 40 French ICUs, with the support of the AZUREA network (Additional file 1: Table S1).
Recruitment was conducted from April 2018 to January 2019. All patients included in the study were followed for 6 months or until death. The 6-month follow-up ended in July 2019. All patients aged 70 years or older, who were hospitalized in an ICU with an expected length of stay of more than 48 h, were eligible for inclusion. During this pre-pandemic study, a teleconsultation was conducted following the same protocol as the one described above for the COVID-19 cohort.
In both cohorts, patients who died before day 180, patients without a systematic consultation on day 180, or patients who were lost to follow-up, were excluded, and for the control cohort, patients admitted for a traumatic or surgical diagnosis were also excluded.
The data collected consisted in social and demographic data (age, sex, body mass index [BMI]), previous clinical autonomy (ADL score before admission), place of living, severity at admission (SAPS II), mechanical ventilation, renal replacement therapy, need of vasopressor, and length of ICU stay. For the day-180 consultation, autonomy (ADL score), quality of life regarding usual activities, anxiety, pain or discomfort, and mobility information were collected according to the EQ5D score (except autonomy assessed with the ADL score). For COVID-19 patients, data collected at day 180 included the IADL (instrumental activities of daily living) score, degree of dyspnea measured using the modified Medical Research Council (mMRC) dyspnea scale, number of medical consultations since discharge, consumption of anxiolytics, antidepressants, and analgesics. Autonomy was measured using 2 scales: a functional evaluation was performed using the ADL scale based on 6 activities of daily living, and a more refined evaluation was performed using the IADL scale, based on 8 more complex tasks using instruments of daily living (measured only for the COVID-19 patients) .
We sought to assess the specific association of ICU admission related to COVID-19 on patient quality of life and autonomy (at day 180 post-ICU admission).
Continuous variables were expressed as median (m) and [interquartile range, IQR], and categorical variables were expressed as count (percentage). Differences between groups were tested using the Wilcoxon rank sum test, chi-square test, or Fisher test.
For the modeling of the association of COVID-19-related ICU stay on the quality of life and autonomy, we used an ordinal logistic regression model on the different variables. Variables were ordered as categorical variable from the highest quality of life to the lowest quality of life. For the ADL score, the variable was ordered from the higher score (6, higher autonomy) to the lower score (0, lower autonomy). The ordered logistic regression allowed to estimate odds ratio (OR) and their associated 95% confidence interval (95% CI). Briefly, the OR derived from an ordinal logistic regression model represent the odds associated with the increase in one level in the ordered factor. As an illustrative example, the association of the COVID-19-related ICU stay on the usual activity (ordered as No problem; Some Problems; A lot of problems) will be, for an OR < 1, a “protective factor” of the increase in the variable (and protective factor of the ability to maintain usual activity), for an OR > 1, a promoting factor of the increase in the variable (and interfering with the ability to maintain usual activity), and for an OR = 1, a factor having no association on the usual activity. Finally, adjusted OR (aOR) were also estimated, they were adjusted on the main baseline characteristic that were not well balanced between both cohorts (p value < 0.05 in univariate analysis). These analyses were also performed on subgroups of patients aged over 70 years old as a sensitivity analysis.
P-values < 0.05 were considered as significant. Analyses were performed using R software version 3.6.4, and the package MASS.