Developing a Frailty Cohort in the Ontario COVID-19 Prospective Cohort Study

The novel coronavirus 2019 (COVID-19) can lead to acute respiratory, cardiac, neurologic and renal complications that may warrant hospital and/or ICU admission. Early reports suggest that COVID-19 affects older people with multiple comorbidities more severely than younger people. Frailty is known to be associated with worse outcomes for hospitalized medical patients and patients admitted to intensive care units (ICUs). However, the relationship between COVID-19 and frailty remains poorly characterized.

Current population data suggest that 10-15% of those infected with COVID-19 are hospitalized. Among all hospitalized COVID-19 patients, 60-80% will be discharged from hospital after a few days whereas 20-40% may require ICU care and mechanical ventilation (approximately 4-6% of all COVID-19 positive patients). Older patients have been disproportionately and more severely affected by COVID-19, as evidenced by more than half of Canadian deaths are from those living in long term care facilities and seniors’ residences. It is crucial to understand how pre-existing frailty (which may also affect younger people with comorbidities) is an important determinant of differences in clinical care and the varied short-and longer-term clinical trajectories in those hospitalized with COVID-19.

Our group has contributed foundational ICU outcomes work from the 5-year ARDS, 1-year SARS and RECOVER programs. Each shows similar outcomes for patients, including: neuromuscular/neurocognitive dysfunction, mood disorders and diverse medical problems contributing to complex disability and significant healthcare use. However, the effects of severe respiratory illness among frail medical inpatients, including those who are not admitted to the ICU, are not well characterized.

We recently completed a retrospective cohort study (unpublished) involving 29,124 patients hospitalized on general medicine wards at 7 hospitals in Ontario. Using the Hospital Frailty Risk Score, we found that 51% of patients over 75 years had high or moderate frailty risk prior to admission. Hospital mortality rates in this population were high (28% and 24% for high and moderate risk, respectively) and the vast majority of patients (94%) with high or moderate frailty did not use ICU services. Therefore, understanding the short-term and long-term outcomes of frail hospitalized patients in non-ICU settings who are affected by severe respiratory illness such as COVID-19 remains an important gap.

We hypothesize that frailty will be associated with differences in the clinical presentation and management of COVID-19, and with poorer short-term and long-term trajectories of illness and recovery. We also hypothesize that frailty risk scores will have prognostic value above and beyond typical sociodemographic and clinical predictors of patient outcomes in COVID-19.

We will develop a Frailty Cohort within the recently-funded Ontario COVID-19 Prospective Cohort Study (OnCovP). OnCovP is a multi-centre, one-year follow-up of COVID-19 patients who are hospitalized in acute care hospitals in Ontario. The OnCovP overall objectives are to determine short- (in hospital) and longer-term (1, 3, 6 and 12 months post-acute hospital discharge) outcomes in patients and their caregivers, and the clinical, sociodemographic, genetic/transcriptomic/epigenomic predictors of these outcomes. The OnCovP Frailty Cohort will allow us to answer the following CFN priority questions:

1. Is there a difference in the clinical characterization and management of COVID-19 in frail vs. non-frail patients?
2. What are the short and long-term health implications for frail individuals that survive COVID-19?
3. What factors should be considered when developing and evaluating diagnostic and prognostic tools? Should frailty be considered in prognostic models?

Objectives

1. Characterize association of frailty with differences in clinical presentation (e.g. severity of illness based on vital signs and laboratory data, symptoms at presentation, etc.), processes of care (e.g. pharmacotherapy, use of ICU), and short-term patient outcomes (death, hospital length-of-stay, discharge disposition, readmission) in the overall cohort and prospective sub-sample. Mixed-effects regression models will be used to account for hospital-level clustering.
2. Characterize association of mild, moderate, and severe frailty with different trajectories of illness and recovery over a 1-year period. Mixed-effects regression models will be used to account for repeat measurements over time.
3. Develop and validate prognostic models for short-term and long-term outcomes in frail and non-frail patients with COVID-19. We will validate any previously published COVID-19 outcome prediction models and develop a new model including the 10-12 most promising predictors (a sensible limit for clinical use). We will assess model performance in frail and non-frail patients, determine whether adding frailty indices improves model performance and compare predictive accuracy of various frailty indices. Bootstrap sampling will be used to evaluate out-of-sample performance with respect to classification and calibration.

Deliverables

Through this project we will be able to:

1. Assess the utility of frailty as a prognostic factor for in-hospital and 1-year clinical and functional outcomes, resource utilization, and discharge destination (community, seniors’ residences or long term care) for patients with COVID-19 admitted to acute care hospitals in Ontario.
2. Characterize the clinical presentation and management of hospitalized frail patients with COVID-19.
3. Identify the unique challenges for adults living with frailty and COVID-19 and their caregivers.
4. Inform others regarding the unique considerations for developing interventions in this population. We will present our results at international frailty conferences, publish in peer-reviewed journals, and work with others in our networks as well as CFN to translate our insights into actionable interventions. We will also provide an open-source, freely accessible tool to predict outcomes for frail patients with COVID-19.