Patient Outcomes and Lessons Learned From Treating Patients With Severe COVID-19 at a Long-term Acute Care Hospital: Single-Center Retrospective Study

Abstract

Background: With the continuation of the COVID-19 pandemic, shifting active COVID-19 care from short-term acute care hospitals (STACHs) to long-term acute care hospitals (LTACHs) could decrease STACH census during critical stages of the pandemic and maximize limited resources.

Objective: This study aimed to describe the characteristics, clinical management, and patient outcomes during and after the acute COVID-19 phase in an LTACH in the Northeastern United States.

Methods: This was a single-center group comparative retrospective analysis of the electronic medical records of patients treated for COVID-19-related impairments from March 19, 2020, through August 14, 2020, and a reference population of medically complex patients discharged between December 1, 2019, and February 29, 2020. This study was conducted to evaluate patient outcomes in response to the holistic treatment approach of the facility.

Results: Of the 127 total COVID-19 admissions, 118 patients were discharged by the data cutoff. At admission, 29.9% (38/127) of patients tested positive for SARS-CoV-2 infection. The mean age of the COVID-19 cohort was lower than that of the reference cohort (63.3, 95% CI 61.1-65.4 vs 65.5, 95% CI 63.2-67.8 years; P=.04). There were similar proportions of males and females between cohorts (P=.38); however, the proportion of non-White/non-Caucasian patients was higher in the COVID-19 cohort than in the reference cohort (odds ratio 2.79, 95% CI 1.5-5.2; P=.001). The mean length of stay in the COVID-19 cohort was similar to that in the reference cohort (25.5, 95% CI 23.2-27.9 vs 29.9, 95% CI 24.7-35.2 days; P=.84). Interestingly, a positive correlation between patient age and length of stay was observed in the COVID-19 cohort (r²=0.05; P=.02), but not in the reference cohort. Ambulation assistance scores improved in both the reference and COVID-19 cohorts from admission to discharge (P<.001). However, the mean assistance score was greater in the COVID-19 cohort than in the reference cohort at discharge (4.9, 95% CI 4.6-5.3 vs 4.1, 95% CI 3.7-4.7; P=.001). Similarly, the mean change in gait distance was greater in the COVID-19 cohort than in the reference cohort (221.1, 95% CI 163.2-279.2 vs 146.4, 95% CI 85.6-207.3 feet; P<.001). Of the 16 patients mechanically ventilated at admission, 94% (15/16) were weaned before discharge (mean 11.3 days). Of the 75 patients admitted with a restricted diet, 75% (56/75) were discharged on a regular diet.

Conclusions: The majority of patients treated at the LTACH for severe COVID-19 and related complications benefited from coordinated care and rehabilitation. In comparison to the reference cohort, patients treated for COVID-19 were discharged with greater improvements in ambulation distance and assistance needs during a similar length of stay. These findings indicate that other patients with COVID-19 would benefit from care in an LTACH.

Keywords: COVID-19; SARS-CoV-2; long-term acute care hospital; occupational therapy; physical therapy; postacute care; post–COVID-19; pulmonary; rehabilitation; respiratory therapy; speech therapy; speech-language pathology; subacute COVID-19.

Figure 1

Study cohorts (COVID-19 cohort and reference cohort). FIM: Functional Independence Measure; NOMS: National Outcomes Measurement System.

Figure 2

Trends in patient admission and length of stay (LOS) during the COVID-19 pandemic. (A) Patient admission from March 19, 2020, to August 14, 2020. (B) Nonlinear regression analysis for the correlation between patient long-term acute care hospital (LTACH) LOS and short-term acute care hospital (STACH) LOS. The solid regression line shows the correlation coefficient, and the dotted lines show the 95% CI. (C) Scatter plot for the comparison of the LTACH LOS between the reference and COVID-19 cohorts. The colored lines represent the median and interquartile range.

Figure 3

Age as a risk factor for prolonged COVID-19 illness. (A) Scatter plot showing the age distribution in the reference and COVID-19 cohorts. The colored lines represent the median and interquartile range. (B) Nonlinear regression analysis showing the correlation between patient age and long-term acute care hospital length of stay (LOS) in the overall COVID-19 cohort. Solid regression lines show the correlation coefficient surrounded by the 95% CI as dotted lines. (C, D) When evaluated by sex, this pattern was also observed in COVID-19 males alone (C), but was not present in COVID-19 females alone (D). Solid regression lines show the correlation coefficient surrounded by the 95% CI as dotted lines.

Figure 4

BMI as a risk factor for prolonged COVID-19 illness. (A) Scatter plot showing the distribution of BMI in the reference and COVID-19 cohorts. Lighter colored lines represent the median and interquartile range. (B) Nonlinear regression analysis showing the correlation between COVID-19 patient BMI and long-term acute care hospital length of stay (LOS). Solid regression lines show the correlation coefficient surrounded by the 95% CI as dotted lines.

Figure 5

COVID-19 patient respiratory and cognitive-communication outcomes. (A) Scatter plot showing the comparison of ventilator wean times among patients mechanically ventilated during fiscal year 2019 (October 2018 through September 2019) (n=37), the reference cohort (n=7), and the COVID-19 cohort (n=15). The colored lines represent the median and interquartile range. (B) Evaluation of the cognitive communication score of COVID-19 patients recommended for speech-language pathology services (n=75) at admission and discharge. NOMS: National Outcomes Measurement System.

Figure 6

Functional Independence Measure (FIM) assistance scores and gait distances as measures of functional ability. (A and C) For both the reference (n=90) and COVID-19 (n=99) cohorts, FIM assistance scores and gait distances were collected at admission and discharge. In-group and between-group comparisons were made using the Šídák multiple comparisons test following a 2×2 two-way mixed effects analysis of variance test for main effects associated with group and time. Box plots represent the median and the 25% and 75% quartiles. The whiskers extend 1.5 and -1.5 of the interquartile range; circle symbols reflect data points beyond the 1.5 interquartile ranges; and the “+” symbol represents the mean. (B and D) Changes in FIM assistance scores and gait distances were then compared using a nonparametric Mann-Whitney U test. B, Violin plot with medium smoothing to show the distribution of FIM score changes; the colored lines represent the median and interquartile range. D, Scatter plot, with the colored lines representing the median and interquartile range.