Functional decline in hospitalized older patients with coronavirus disease 2019: a retrospective cohort study

This was a retrospective cohort study conducted in a tertiary care hospital (Kawasaki Municipal Kawasaki Hospital, Kawasaki, Japan). Japanese citizen aged 65 years or older with COVID-19 who had hospitalized to the study facility between February 1, 2020 and February 28, 2021 were included. All patients were followed-up until their discharge from the hospital (last follow-up date: April 26, 2021). Among them, survivor patients were classified into two groups according to the presence or absence of functional decline from before the onset of COVID-19 to discharge from the hospital, as defined in details below. We performed one main analysis and two sensitivity analyses. One sensitivity analysis was performed to evaluate functional decline in patients after excluding those with the Barthel Index (BI) score < 25 (i.e., very severe physical dependency). Another sensitivity analysis was performed to evaluate the composite outcome of functional decline and death in survivors and those who died during hospitalization.

As the index of ADL, we evaluated the BI which consists of 10 items: feeding, transferring, grooming, toilet use, bathing, mobility, climbing up and down stairs, dressing, ability to control bowel function, and ability to control bladder function. The BI was scored from 0 (completely dependent) to 100 (independent) points, with higher score indicating greater ADL independence [9]. Information on the BI before the onset of COVID-19 was generally collected by interviewing patients in routine clinical practice. For patients who could not communicate with medical staff, we recorded the BI either based on the written information from primary caregiver or nursing home staff or through direct telephone interview with them as soon as possible after admission (within 3 days from admission in most cases). Physicians or nurses evaluated the BI score at discharge by direct observation. The outcome measure was the development of functional decline at discharge, which was defined as a decrease in the BI score of at least 5 points from before the onset of COVID-19 to that at discharge [8], although the minimal important change in the BI has not been established. The patients were also classified into four categories according to the severity of ADL dependency at discharge: independent (the BI score, 100), mildly dependent (75–95), moderate-to-severely dependent (25–70), and very severely dependent (0–20) [10].

We also evaluated age, sex, route of admission, days from the onset of COVID-19 to admission, presence of dementia, and a recorded diagnosis of other comorbidities (chronic obstructive pulmonary disease, chronic kidney disease, diabetes, hypertension, and cardiovascular diseases), which were listed as risk factors for severe COVID-19 according to the Japanese guidelines [11]. COVID-19 severity at admission and the highest severity during hospitalization for COVID-19 were determined according to the classification published by the US National Institutes of Health [12].

Serum albumin, serum C-reactive protein (CRP), and plasma D-dimer levels were measured upon admission. The Geriatric Nutritional Risk Index (GNRI) was measured to assess nutritional status and nutrition-related risk [13]. In addition to baseline characteristics, we also measured the implementation of in-hospital rehabilitation, generally including rehabilitation to maintain limb muscle strength and/or pulmonary rehabilitation in an isolation room for 20 to 40 min per day according to the hospital’s protocol, length of stay (regardless of the type of bed), and discharge destination (in patients admitted from home only).

Characteristics of the patients were compared according to the presence or absence of functional decline (functional decline and death for one sensitivity analysis) using the Fisher’s exact test for categorical variables and the Mann–Whitney U test for continuous variables.

Logistic regression analyses were performed to evaluate the associations between characteristics of the patients at admission and functional decline. We first selected the variables to be evaluated in the associations with functional decline based on the evidence from previous studies [6] as well as those which were considered clinically relevant from the data available. Subsequently, we selected some variables included in the multivariable analyses, taking into account the statistically significant differences between those with or without functional decline in the univariable analysis. As an exception, we used low albumin level as an indicator of malnutrition, because of frequently missing data for the GNRI. We dichotomized the explanatory variables for possible clinical relevance and simple interpretation, including age group (< 80 and ≥ 80 years [very old]), sex, BI score before the onset of COVID-19 (< 100 [dependent] and 100 [independent]), dementia, serum albumin level (< 3.3 and ≥ 3.3 mg/dL), and plasma D-dimer level (< 1.0 and ≥ 1.0 μg/mL). Both the cut-off values for age and for baseline ADL impairment based on the BI score before the onset of COVID-19, which implies dependence in at least one ADL, were based on a previous study which investigated the risk for new-onset disability in hospitalized older adults [14]. The cut-off values for serum albumin and plasma D-dimer levels were determined according to previous reports that evaluated the prognosis of COVID-19 [15, 16]. These evaluations were repeated in two sensitivity analyses with different inclusion criteria. Additionally, considering the possibility of uncertainty in the cut-off values of the explanatory variables, we also performed an analysis with age, the BI score, serum albumin level, and plasma D-dimer level as continuous/ordinal scales.

This study was conducted in compliance with the principles of the Declaration of Helsinki and current ethical guidelines. This study was approved by the Institutional Review Board of Kawasaki Municipal Kawasaki Hospital (approval number 2020–28) in accordance with the requirements of the institution. The requirement for written informed consent from the patients was waived because of the retrospective nature of this study. The data were analyzed after anonymization.

A total of 171 older patients with COVID-19 were hospitalized during the study period. We excluded five patients (2.9%) who were transferred before recovery from COVID-19. In addition, we excluded patients whose BI score before the onset of COVID-19 were missing (n = 3) or 0 (n = 8), indicating complete dependence of ADL. Of the remaining 155 patients, 23 died (14.8%) during hospitalization. One hundred and thirty-two survivors were included in the main analysis, with a median age of 78.5 years and 49.2% comprising female patients. Seventy-two of the 132 patients (54.5%) were assigned into the functional decline group (FDG), and the remaining 60 into the non-functional decline group (non-FDG).

Demographic and clinical characteristics of the patients are shown in Table 1. Compared with the patients in the non-FDG, those in the FDG were older, had a higher proportion of female patients, had dementia, were admitted from long-term care hospitals/facilities, had lower serum albumin and higher plasma D-dimer levels on admission, and had lower the BI score before the onset of COVID-19. The severity of COVID-19 at admission and serum CRP level at admission were not higher in the FDG.

The associations of baseline characteristics of the patients with a greater risk of functional decline are shown in Fig. 1. In univariable analysis, factors associated with functional decline during hospitalization were age over 80 years (odds ratio [OR], 4.87; 95% confidence interval [CI], 2.28 to 10.43), female sex (OR, 3.33; 95% CI, 1.63 to 6.84), the BI score < 100 before the onset of COVID-19 (OR, 23.75; 95% CI, 6.80 to 82.94), presence of dementia (OR, 9.64; 95% CI, 3.99 to 23.26), and elevation of plasma D-dimer level on admission (OR, 2.66; 95% CI, 1.27 to 5.57).

In multivariable analysis, female sex (adjusted OR, 3.14; 95% CI, 1.25 to 7.94), the BI score < 100 before the onset of COVID-19 (adjusted OR, 13.73; 95% CI, 3.29 to 57.25), and elevation of plasma D-dimer level on admission (adjusted OR, 3.19; 95% CI, 1.12 to 9.07) were independently associated with functional decline.

The clinical course and outcomes are shown in Table 2. The highest severity of COVID-19 during hospitalization was not significantly different between the groups. Most patients in the FDG received rehabilitation services during their hospitalization. The length of stay was longer in the FDG than in the non-FDG. The proportion of patients who were admitted from home but discharged to long-term care hospitals/facilities after recovery from COVID-19 was higher in the FDG than in the non-FDG.

In the FDG, the median decrease in the BI score was 35 points, and the median number of impaired BI components was five. Walking on a level surfaces (69.0%), moving from a wheel chair to a bed and returning (66.2%), and dressing (63.4%) were often impaired in the FDG.

In the non-FDG, two patients changed the two components of the BI without decreasing the total BI score. One patient had an impaired bathing component from independent to dependent, whereas the bladder component improved from incontinent to occasional accident. The other patient had an impaired bowel component from occasional accident to incontinent, whereas the feeding component improved from unable to need help.

The analysis with age and the BI score as continuous/ordinal variables yielded similar results to those of the main analysis: a 10-year increase in age was not associated with a greater risk of functional decline (adjusted OR, 1.04; 95% CI, 0.52 to 2.09), but a 5-point decrease in the BI score was (adjusted OR, 1.16; 95% CI, 1.01 to 1.33). However, the association between dementia and a greater risk of functional decline reached statistical significance (adjusted OR, 4.83; 95% CI, 1.47 to 15.85), whereas a 1 μg/mL increase in plasma D-dimer level was not statistically significantly associated with a greater risk of functional decline (adjusted OR, 1.26; 95% CI, 0.98 to 1.62) (Fig. S1) (Additional file 1).

One sensitivity analysis included 121 patients without very severe ADL dependence before the onset of COVID-19, and of these, 63 (52.1%) developed functional decline. The other sensitivity analysis included 155 survivors and those who died, and 95 patients (61.3%) developed functional decline or death. The differences between the patients with individual outcomes and those without in sensitivity analyses were similar to those in the main analysis, with some exceptions: lower serum albumin level in patients with the outcomes (with statistical significance) and the highest severity of COVID-19 during hospitalization in those with functional decline or death (Tables S1, S2, and S3) (Additional file 1). The sensitivity analyses yielded similar results to those of the main analysis; thus, female sex (not reaching statistically significant difference in the sensitivity analysis including survivors and those who died), the lower BI score before the onset of COVID-19, and elevation of plasma D-dimer level were associated with greater risks for functional decline or functional decline and death (Figs. S2 and S3) (Additional file 1).