key: cord-1028309-hj2pubt5
authors: Abdelwahab, Nermine; Ingraham, Nicholas E; Nguyen, Nguyen; Siegel, Lianne; Silverman, Greg; Sahoo, Himanshu Shekhar; Pakhomov, Serguei; Morse, Leslie R; Billings, Joanne; Usher, Michael G.; Melnik, Tanya E.; Tignanelli, Christopher J.; Ikramuddin, Farha
title: Predictors of Post-Acute Sequelae of COVID-19 Development and Rehabilitation: A Retrospective Study
date: 2022-05-13
journal: Arch Phys Med Rehabil
DOI: 10.1016/j.apmr.2022.04.009
sha: 28c2c0d9259dfa4ff88f14dad9c33f4b61589fd7
doc_id: 1028309
cord_uid: hj2pubt5

Objective Clinical and demographic factors associated with the development, severity, and rehabilitation utilization of patients with Post-Acute Sequelae of COVID-19 (PASC) are not well defined. We examined the frequency of PASC, and the factors associated with rehabilitation utilization in a large adult population with PASC. Design Retrospective study Setting Hospital health system Participants All COVID-19 patients from March 10, 2020 to January 17, 2021 Intervention Not applicable. Main outcome measure Descriptive analyses were conducted across the entire cohort along with an adult subgroup analysis. A logistic regression was performed to assess factors associated with PASC development and rehabilitation utilization. Results In an analysis of 19,792 patients, the frequency of PASC was 42.8% in the adult population. Patients with PASC compared to those without had a higher utilization of rehabilitation services (8.6% vs 3.8%, p<0.001). Risk factors for rehabilitation utilization in patients with PASC included younger age (OR 0.99, 95% CI 0.98-1.00; p=0.01). In addition to several comorbidities and demographics factors, risk factors for rehabilitation utilization solely in the inpatient population included male sex (OR 1.24, 95% CI 1.02-1.50; p=0.03) with patients on angiotensin-converting-enzyme inhibitors or angiotensin-receptor blockers three months prior to COVID-19 infections having a decreased risk of needing rehabilitation (OR 0.80, 95% CI 0.64-0.99; p=0.04). Conclusion Patients with PASC had higher rehabilitation utilization. We identified several clinical and demographic factors associated with the development of PASC and rehabilitation utilization.

There is a growing concern that patients infected with SARS-CoV-2 experience persistent symptoms long after the initial symptomatic phase. (1, 2) Currently, there is no established definition to describe patients with COVID-19 sequelae, however, a commonly proposed characterization describes illness greater than four weeks after acute infection as late sequalae or post-acute sequalae of SARS-CoV-2 (PASC). (3) (4) (5) The frequency and timeline of PASC is unclear and varies widely with estimations as high as 50%. (6, 7) The presentation also varies with multiple different organ systems affected, and in certain cases, the symptoms are severe enough to cause new disability. (8) (9) (10) (11) There is limited information on exacerbating and mitigating factors that would predispose patients to develop PASC. (6, 7) For patients that develop PASC, there is limited information on rehabilitation utilization and efficacy, however, case series have suggested improvement in patient's symptoms with rehabilitation. (2, 12) Given such a high overall reported frequency of PASC, more information is needed to help triage and recruit at risk patients to rehabilitation programs, thus, we thought to examine patient factors that increase the likelihood of development of PASC. In addition, we examined rehabilitation utilization in patients with PASC and the factors associated with the need for rehabilitation services. By better understanding the resource utilization of patients, we can implement patient tailored rehabilitation plans to at risk populations. We hypothesized that patients with more severe disease and more comorbidities would require more rehabilitation services.

This study is a retrospective analysis of data from March 10, 2020, to January 17, 2021, of COVID-19 patients who had their test done at a specific health system. Inclusion criteria included all patients with polymerase chain reaction (PCR)-confirmed COVID-19 treated at a participating hospital. Exclusion criteria included patients that died during their initial acute COVID-19 infection (hospitalized and non-hospitalized patients) and those who opted out of research.

The study database was created from Epic electronic health records (EHRs) and included: patient demographics (age, gender, race/ethnicity), medications, past medical history, and health encounters from January 1, 2019, to March 17, 2021 . Additionally, information regarding state death certificates was obtained from the state's Department of Health.(16) Patients without prior encounters within each hospital were included in the primary analysis as we did not want to exclude previously healthy patients that developed de novo COVID-19 and subsequent chronic disease.

The primary outcome was the development of PASC, which was defined as any patient that had PASC symptoms 31 days or more after COVID-19 and did not have these symptoms at baseline (i.e., a patient with chronic obstructive lung disease (COPD) and a chronic cough that has a cough after COVID-19 would not be considered PASC). This was done to reduce possible confounding factors. Resource utilization related to PASC was categorized using variables (i.e., physical medicine and rehabilitation referrals, pulmonary and cardiac rehabilitation) shown in Supplementary 

The university's Natural Language Processing and Information Extraction Laboratory used the list of PASC and COVID-19 symptoms that was developed for this study to extract symptoms from health encounter visits. (17) This process is referred to as the creation of a rule-based gazetteer and relied on linguistic rules constructed from the lexicon to match any mentions of the symptoms and their linguistic variants in notes. Each symptom mention was marked as positive or negative based on whether it occurred in a negated context (e.g., "denies cough" would be marked as a negative instance of the cough symptom). The overall performance of the gazetteer was validated against a reference standard set of manually annotated ED clinical notes and yielded a precision of 0.90, recall of 0.87, and f1-score of 0.88. (17) (18) (19) Statistical analysis was conducted by an independent investigator not involved in variable selection. For descriptive purposes, data were expressed as median and IQR for continuous variables with a skewed observed distribution and as percentages for categorical variables.

Student's t-tests, Mann-Whitney U tests, and Pearson χ2 tests were used in the preliminary analyses as appropriate for the assumed variable distribution. Multivariable logistic regression was performed to evaluate the independent association of variables of interest on the need for rehabilitation services in patients with PASC. Subgroup analyses for rehabilitation utilization were conducted on adults that were hospitalized during their initial COVID-19 infection. An additional adjustment was performed on this population to account for confounding variables (i.e., demographics, comorbidities, medications, inpatient data) which can be also found in Supplementary Table 2 .

All statistical analysis was performed using Stata-MP Version 16 (StataCorp, College Station, TX). Goodness of fit was assessed with Hosmer-Lemeshow tests, where a p-value < 0.1 was considered statistically significant. All other tests were two sided and significance was defined with an alpha of < 0.05.

Overall Population 19,792 patients were included in the analysis ( Figure 1 ). In the adult population, the age range was 18 years to 90 years or older. The median age of patients who developed PASC was 51.4 years old (32.8-66.4) with 38% identifying as male. The characteristics of patients with PASC compared to those without are shown in Table 1 . The frequency of PASC was 42.8% in the adult population. Table 1 shows the outpatient rehabilitation services that were analyzed which include physical therapy, occupational therapy and speech language pathology. Patients with PASC compared to those without PASC had a higher frequency of rehabilitation services during COVID-19 (8.6% vs 3.8%, p<0.001), after COVID-19 (8.4% vs 3.0%, p<0.001) as well as outpatient physiatry referrals (3.1% vs 1.7%, p<0.001) ( Table 1) .

The factors associated with the development of PASC in all patients can be found in Table 2 19 ; p<0.001) and those who required any rehabilitation program prior to COVID-19 illness were also at higher risk of developing PASC (OR 1.91, 95% CI 1.78-2.05; p<0.001). Several comorbidities and medications that patients were on 3 months prior were associated with an increased risk of PASC (Table 2) .

Risk factors for need for rehabilitation in patients with PASC included younger age (OR 0.99, 95% CI 0.98-1.00; p=0.01, those who were pregnant (OR 3.30, 95% CI 1.92-5.66; p<0.001), as well as other comorbidities which can be found in Table 3 . Risk factors for rehabilitation utilization solely in the inpatient populations were also explored and can be found in Supplementary compared to non-users of ACE inhibitors or ARBs.

The purpose of this study was to explore rehabilitation utilization for patients with PASC and identify mitigating and protective factors associated with the development of PASC. In our study, we identified three key findings. First, there were high rates of PASC in our patient population. Second, in patients with PASC, younger patients had higher rehabilitation utilization and several comorbidities were found to be risk factors for rehabilitation utilization, especially in cases of severe COVID-19. Third, patients on ACEI/ARB had decreased risk of requiring rehabilitation resource in the inpatient population.

The frequency of PASC in the adult population was 42.8%. Because of the lack of a standardized definition, the rates reported in other studies often ranges between the low teens to up to more than half of the population. PASC was present in both mild and severe disease, however, having severe disease, defined as requiring hospital admission, was a risk factor for development of PASC. Several comorbidities were found to be risk factors including patients with hypertension, chronic kidney disease, and asthma, which is similar to risk factors for acute COVID-19

illness. (20) There were several medications associated with an increased risk of PASC in the patients that were taking it prior to acute illness; this is likely due to the association of those medications with comorbidities. In our study, non-English speaking populations and being Asian or Black were a risk factor for the development of PASC and being Asian or Hispanic was a risk factor for rehabilitation utilization within the inpatient population. Given this increased risk among non-English speakers and patients of color, information regarding PASC needs to be culturally and linguistically accessible as a possible tool to help mitigate this discrepancy.

Resource utilization was high in patients with PASC. Specifically, there was a higher number of therapy sessions and physical medicine and rehabilitation referrals . PASC patients underwent more therapy focused on activities of daily living, cognitive function, and neuromuscular education. Younger patients utilized rehabilitation services more overall (in combined severe and non-severe cases), but for patient's requiring inpatient admission, being older was a risk factor for needing rehabilitation services, and not surprisingly, those who required hospital admission made up the majority of patients needing rehabilitation. In the inpatient population, several demographic factors including male sex were risk factors for need for rehabilitation services. In addition, many comorbidities like hypertension, chronic obstructive lung disease, liver disease, and autoimmune disorders were also associated with increased rehabilitation utilization in the inpatient population. These data highlight not only the high rehabilitation utilization of patients with PASC but also speak to the effect PASC can have on society and the workforce.

Additionally, this research can provide an introductory framework for hospital systems to implement rehabilitation programs targeting patients with multiple risk factors.

ACEIs/ARBs were associated with decreased risk of needing rehabilitation services in the inpatient population. This supports previous data suggesting possible protective benefits of ACEIs/ARBs on mortality for COVID-19 patients. (21) (22) (23) Possible mechanisms for both medication categories include improved blood pressure control and potential downregulation of the Renin-angiotensin-aldosterone system (RAAS) with chronic use leading to decreased inflammation. (24) (25) (26) These data may suggest the benefit is more long-term and may reside across those with more severe disease. (27) 

This study has many strengths. The large sample size, inclusion of both inpatient and outpatient participants, and extensive but relevant clinical variables allowed for a broader analysis of factors associated with PASC and rehabilitation. Additionally, our definition of PASC as new symptoms not present at baseline as well as the additional adjustments on the inpatient population reduced possible confounders. The study has several limitations. Our results do not suggest a casual inference and could be subject to residual confounding. Only patients diagnosed with COVID-19 at the healthcare system were included, and thus the population is not indicative of the whole healthcare system's patient population. There was a lack of a control non-COVID-19 cohort making us unable to compare the frequency and symptoms of PASC to a general postviral illness syndrome. The information was also extracted from the electronic medical record from one hospital system, and taken from problem lists and notes, making data collection not standardized and possibly clinician dependent. Medications listed for patients don't ascertain actual medication use. Patients could have also received care at different healthcare systems and that information would not have been included. Additionally, there was no objective data collection analyzed (i.e., pulmonary function tests or computerized tomography (CT) imaging).

Finally, the overall missingness of data was relatively low; only three variables had any missingness > 0.2%: 17.8% of patients were missing body mass index, 4% were missing comorbidity data, and 6% of patients were missing race/ethnicity data. Given the low rate of missingness, multiple imputation was not done and a complete case analysis was conducted for multivariable analysis. (28) 

Our study demonstrated a high frequency of PASC. Patients with PASC had a high amount of resource utilization and there were several demographic features and comorbidities that were associated with greater rehabilitation utilization. This study highlights the need for continued development of interdisciplinary teams and care facilities to address the needs of patients post-COVID-19 and provides a starting point for hospital systems to help triage at risk patients.

Additional studies are needed that include a non-COVID-19 control group to accurately assess incidence, symptom presentations, and factors specific to PASC and patient's rehabilitation needs compared to general viral illnesses. 

The lasting misery of coronavirus long-haulers

Rehabilitation and COVID-19: a rapid living systematic review by Cochrane Rehabilitation Field updated as of December 31st, 2020 and synthesis of the scientific literature

National Institutes of Health (NIH)

A Proposed Framework and Timeline of the Spectrum of Disease Due to SARS-CoV-2 Infection: Illness Beyond Acute Infection and Public Health Implications

Management of post-acute covid-19 in primary care

Attributes and predictors of Long-COVID: analysis of COVID cases and their symptoms collected by the Covid Symptoms Study App. medRxiv

Post-acute COVID-19 syndrome. Incidence and risk factors: A Mediterranean cohort study

6-month consequences of COVID-19 in patients discharged from hospital: a cohort study. The Lancet

for the Gemelli Against COVID-19 Post-Acute Care Study Group. Persistent Symptoms in Patients After Acute COVID-19

6-month neurological and psychiatric outcomes in 236 379 survivors of COVID-19: a retrospective cohort study using electronic health records

The impact of COVID-19 critical illness on new disability, functional outcomes and return to work at 6 months: a prospective cohort study

COVID-19 related fatigue: Which role for rehabilitation in post-COVID-19 patients? A case series

Pre-and Peri-Operative Factors Associated with Chronic Critical Illness in Liver Transplant Recipients

National Trends and Variation of Functional Status Deterioration in the Medically Critically Ill

Division of Viral Diseases. Symptoms of COVID-19

NLP Methods for UMLS Concept Extraction of Symptoms from Unstructured Data for use in Prognostic COVID-19 Analytic Models

Natural Language Processing / Information Extraction COVID symptom gazetter

A Rule-based System for Covid-19 Symptom Identification and Classification

Does comorbidity increase the risk of patients with COVID-19: evidence from meta-analysis. Aging

Association of Use of Angiotensin-Converting Enzyme Inhibitors and Angiotensin II Receptor Blockers With Testing Positive for Coronavirus Disease 2019 (COVID-19)

Antihypertensive drugs and risk of COVID-19?

Understanding the renin-angiotensin-aldosterone-SARS-CoV axis: a comprehensive review

Renin-Angiotensin-Aldosterone System Inhibitors and Risk of Covid-19

Beta-Adrenergic Blockers as a Potential Treatment for COVID-19

Immunomodulation in COVID-19

A multi-center phase II randomized clinical trial of losartan on symptomatic outpatients with COVID-19

Don't Do Imputation: Dealing with Informative Missing Values in EHR Data Analysis