Risk of 30-Day All-Cause Readmission in Interstitial Lung Disease Patients after COVID-19: National-Level Data

Abstract

Rationale

Hospital readmission within 30 days poses challenges for healthcare providers, policymakers, and patients because of its impact on care quality, costs, and outcomes. Patients with interstitial lung disease (ILD) are particularly affected by readmission, which is associated with increased morbidity and mortality and reduced quality of life. Because small sample sizes have hindered previous studies, this study seeks to address this gap in knowledge by examining a large-scale dataset.

Objective: To determine the rate and probability of 30-day all-cause readmission and secondary outcomes in patients with coronavirus disease (COVID-19) or ILD admitted to the hospital.

Methods

This study is a nested cohort study that used the PearlDiver patient records database. Adult patients (age ⩾18 yr) who were admitted to hospitals in 28 states in the United States with COVID-19 or ILD diagnoses were included. We defined and analyzed two separate cohorts in this study. The first cohort consisted of patients with COVID-19 and was later divided into two groups with or without a history of ILD. The second cohort consisted of patients with ILD and was later divided into groups with COVID-19 or with a non–COVID-19 pneumonia diagnosis at admission. We also studied two other subcohorts of patients with and without idiopathic pulmonary fibrosis within the second cohort. Propensity score matching was employed to match confounders between groups. The Kaplan-Meier log rank test was applied to compare the probabilities of outcomes.

Results

We assessed the data of 2,286,775 patients with COVID-19 and 118,892 patients with ILD. We found that patients with COVID-19 with preexisting ILD had an odds ratio of 1.6 for 30-day all-cause readmission. Similarly, an odds ratio of 2.42 in readmission rates was observed among hospitalized individuals with ILD who contracted COVID-19 compared with those who were hospitalized for non–COVID-19 pneumonia. Our study also found a significantly higher probability of intensive care admission among patients in both cohorts.

Conclusions

Patients with ILD face heightened rates of hospital readmissions, particularly when ILD is combined with COVID-19, resulting in adverse outcomes such as decreased quality of life and increased healthcare expenses. It is imperative to prioritize preventive measures against COVID-19 and establish effective postdischarge care strategies for patients with ILD.

The coronavirus disease (COVID-19) pandemic has had a significant impact on healthcare systems worldwide, including a substantial impact on the outcomes of other diseases, including interstitial lung disease (ILD) (1, 2).

ILD represents a heterogenous group of diffuse parenchymal lung diseases primarily involving the interstitium, comprising idiopathic pulmonary fibrosis (IPF), sarcoidosis, connective tissue disease– associated ILD, and other conditions. In the United States, the prevalence of ILDs ranges from 67.2 to 80.9 per 100,000, and the incidence is between 26.1 and 31.5 per 100,000 per year. Patients with ILD are at increased risks of hospitalization, mortality, and reduced health-related quality of life (3–6). Previous studies have examined the mortality risk in hospitalized patients with ILD following their initial COVID-19 infection. Across all studies, a consistent finding was a significantly higher risk of mortality in individuals with ILD who were hospitalized because of COVID-19 infection (1, 7–9).

Patients with ILD are susceptible to acute episodes of respiratory symptom exacerbations, which can result in increased healthcare use, including emergency department visits and hospitalizations (10–15). These exacerbations are linked to viral infections and thoracic surgical procedures, although the precise reason for their occurrence is not yet fully understood (3). Because patients with ILD have impaired lung function and are prone to acute exacerbations triggered by viral infections, COVID-19 is of particular concern. Exacerbations represent a significant risk factor for poor outcomes in patients with ILD, including 30-day hospital readmissions (6, 16–20).

Thirty-day readmission is a concern for healthcare providers and policymakers because it can have a notable impact on quality of care, healthcare costs, and patient outcomes. Because patients are frequently readmitted with exacerbations of conditions from the index admission, 30-day readmission rates are often used as a quality indicator and a measurable metric for hospitals (21, 22). Thirty-day readmission increases the burden on the healthcare system, increases morbidity and mortality, and reduces the quality of life of patients. Additionally, readmission may serve as an indicator of inadequate disease management and suggest the necessity for more aggressive treatment (23). There is a scarcity of studies on the impact of COVID-19 on readmissions of patients with ILD, and the available studies have been hindered by small sample sizes. To address this gap, this study aimed to assess the interaction between COVID-19 and ILD regarding patient outcomes, including the rate of 30-day readmission, intensive care unit (ICU) admission, 30-day incidence of pulmonary embolism (PE), and hospice referral after discharge, using a large-scale nationwide dataset.

Methods

Study Design

In this study, we used a nested cohort design to examine the interaction between ILD and COVID-19 on patient outcomes using two distinct cohorts. Cohort I, referred to as the COVID-19 cohort, consisted of adult patients admitted to the hospital with a primary diagnosis of COVID-19. A subgroup of patients with a positive history of ILD (COVID + ILD) was identified based on ILD status and matched at a 1:1 ratio with those without ILD from the COVID-19 cohort. The objective of this cohort was to assess the impact of ILD on the outcomes of patients hospitalized for COVID-19.

Cohort II, the ILD cohort, included patients admitted with ILD selected from a database of approximately 151 million subjects. Patients in this cohort were categorized into two groups: patients with ILD who contracted COVID-19 (ILD + COVID) as the exposure group and patients with ILD who were admitted with non–COVID-19 pneumonia as the control group. The exposure and control groups were derived from the main ILD cohort and matched using propensity scores. The aim of investigating this cohort was to determine the effect of COVID-19 infection on the outcomes of patients with ILD.

Additionally, within the second cohort, patients with IPF, which is a specific ILD subtype, and those without IPF were identified to form two distinct subcohorts. An additional analysis was conducted to evaluate the impact of COVID-19 on the outcomes of patients in these specific subcohorts.

Data Source

In this study, we used the PearlDiver patient records database (PearlDiver Inc), a subscription-based research suite, to extract data from hospitalized patients with ILD and/or COVID-19. The PearlDiver database, described in detail elsewhere (24), provides a comprehensive collection of medical information encompassing various types of insurance, including private insurance, Medicare, Medicaid, self-pay, and government insurance, within inpatient settings. Equipped with longitudinal tracking based on unique patient identifiers, the database serves as a valuable resource for medical research.

To extract the required medical data from the database, we employed Boolean operators and International Classification of Diseases (ICD) codes (9th and 10th Revisions, Clinical Modifications). The database’s search functionality, including codes such as Current Procedural Terminology and national medication codes, facilitated the extraction of data for our study. We retrieved all-payer claims administrative data spanning from January 1, 2020, to October 30, 2021. To ensure patient privacy, the queried data were deidentified and compliant with the Health Insurance Portability and Accountability Act of 1996. Consequently, our study adhered to all required ethical standards and regulations governing medical research, leading to the approval of a waiver by the institutional review board for this study.

Study Population

We included adult patients (aged ⩾18 yr) who were admitted to hospitals in the 28 states of the United States accessible through the PearlDiver database. Patients who underwent lung transplantation and pediatric patients (aged <18 yr) were excluded from the study.

To ensure the study’s accuracy, an ILD case was identified when there were at least two claims with a diagnostic ICD code for ILD separated by a period of at least 30 days. The diagnosis and classification of ILDs followed the guidelines set forth by the American Thoracic Society and European Respiratory Society (25). Specifically, ILD was defined based on claims with the ICD diagnosis codes (9th and 10th Revisions, Clinical Modifications) D51, J84, and J70 (see data supplement). COVID-19 admission was defined as any admission coded with ICD-10 U07.1 (26, 27) as the primary diagnosis.

Matching

Propensity-score matching was employed to align the groups regarding baseline demographic variables and comorbidities. To achieve optimal matching, variables such as age, sex, drug abuse, alcohol abuse, liver disease, chronic kidney disease, dementia, smoking, myocardial infarction, hypertension, heart failure, chronic obstructive pulmonary disease (COPD), asthma, history of cancer (excluding basal cell carcinoma of the skin), stroke, coronary artery disease, human immunodeficiency virus, diabetes mellitus, and obesity were considered for matching. Comorbidities were defined by two or more claims under ICD-10 diagnosis codes as major diagnoses during the study period. Alcohol abuse and depression were defined based on the Diagnostic and Statistical Manual of Mental Disorders (28). Asthma and COPD were defined by the criteria established by the Global Initiative for Asthma and the Global Initiative for Chronic Obstructive Lung Disease, respectively (29, 30). Rheumatoid arthritis was defined based on the criteria set by the American College of Rheumatology/European League Against Rheumatism collaborative initiative (31). Obesity was defined according to the World Health Organization criteria (32). For ICD codes used to identify baseline characteristics, see data supplement.

The Elixhauser comorbidity index (ECI) was used to stratify the study populations based on the severity of their other chronic conditions. This index comprises a list of 31 comorbidities such as diabetes, hypertension, congestive heart failure, and cancer. Each comorbidity is assigned a weight based on its association with increased healthcare use and mortality, ranging from 1 (least severe) to 6 (most severe), with higher weights indicating a greater impact on outcomes (25).

To accommodate for the emergence of new severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants over time, the timing of the index events was matched accordingly between groups. Standardized mean difference, variance ratios, and empirical cumulative distribution function statistics were used before and after matching to assess covariate balance and to optimize matching. A standardized mean difference less than 0.1 and variance ratios close to 1 indicated an insignificant difference between groups (33, 34). To enhance the statistical precision of our study, individuals with ILD and COVID-19 (referred to as ILD + COVID or COVID + ILD depending on cohort) underwent a propensity-matching technique with control groups. The matching process considered all Elixhauser comorbidities as well as key demographic variables such as age, sex, and timing.

Primary and secondary outcomes

The primary outcome of this study was risk-standardized 30-day all-cause hospital readmission, whether to the same hospital or another applicable acute care hospital.

Secondary outcomes included admission to the ICU, referral to hospice upon discharge, and incidence of PE diagnosed by contrast-enhanced computed tomographic angiography within 30 days after the index event. To prevent biased estimation, only the first admission per outcome was considered. Patients with active records who remained enrolled with their insurance carrier during the follow-up period were included in the determination of outcomes.

Statistical Analysis

Baseline characteristics were compared using Pearson’s χ² analysis for categorical variables and a Student’s t test for continuous variables. Kaplan-Meier analysis with the log-rank test was employed to compare outcomes, and the daily hazard for each outcome after admission was estimated. The degree of freedom was 1 among all the groups. For the observed results to theoretically expected results for each outcome (O − E)² / E (i.e., χ² tests; “O,” observed; “E,” expected), see data supplement. The analysis was conducted using R software (version 4.0.3, 2020; R Foundation for Statistical Computing). Both cohorts’ records were examined from the index event up to 30 days. Statistical significance was assumed for a P value less than 0.05.

Results

We assessed medical data for 151,463,675 patients across 28 states using the PearlDiver database. From this dataset, a total of 2,286,775 patients with COVID-19 (cohort I) and 118,892 patients with ILD (cohort II) met the inclusion and exclusion criteria, forming the first and second cohorts. Within the subcohorts with and without IPF, 28,532 and 17,477 patients, respectively, met the eligibility criteria. In this section, we will present the baseline characteristics and outcome measures for each cohort separately.

COVID-19 Cohort

Within the COVID-19 cohort, a total of 7,649 patients were identified with a history of ILD, defined as the exposure group (i.e., COVID + ILD). The mean age of patients in this group was 68 years ± 13 (standard deviation [SD]), and 60.26% were women (39.73% men). The three most prevalent comorbidities in the COVID + ILD group were hypertension (88.74%), COPD (64.80%), and diabetes mellitus (58.29%). The mean ECI score in this group was 10.4 ± 4.5. Basic demographic and comorbidity variables of the COVID + ILD group compared with the rest of the COVID-19 cohort with no history of ILD are presented in Table 1. The control group consisted of 7,649 patients with COVID-19 without a history of ILD extracted from the baseline COVID-19 cohort and matched at a 1:1 ratio with the COVID + ILD group using propensity scores.

Table 1.

Baseline characteristics of patients in the COVID-19 based on history of ILD

Characteristic	COVID-19 and Positive History of ILD (n = 7,649)	COVID-19 and No History of ILD (n = 2,279,126)	OR (95% CI)
Age, yr	68 ± 13	60 ± 16.8	—
Male sex	3,039 (39.73%)	969,030 (42.51%)	0.93 (0.89–0.95)
Drug and smoking history
Drug abuse	1,137 (14.86%)	207,698 (9.11%)	0.61 (0.58–0.64)
Alcohol abuse	637 (8.32%)	161,426 (7.08%)	0.83 (0.77–0.91)
Smoker	4,044 (52.86%)	762,381 (33.45%)	0.63 (0.62–0.64)
Elixhauser comorbidity index	10.4 ± 4.5	6.6 ± 4.2	—
Comorbidities
Myocardial infarction	938 (12.26%)	137,008 (6.01%)	0.45 (0.42–0.49)
Hypertension	6,788 (88.74%)	1,822,140 (79.94%)	0.90 (0.89–0.91)
Heart failure	1330 (17.38%)	176,762 (7.75%)	0.45 (0.42–0.46)
Pulmonary embolism	291 (3.80%)	37,909 (1.66%)	0.43 (0.39–0.49)
COPD	4,957 (64.80%)	695,494 (30.51%)	0.47 (0.46–0.48)
Asthma	2,457 (32.12%)	405,636 (17.79%)	0.55 (0.53–0.57)
Any type of cancer	1,628 (21.28%)	303,006 (13.29%)	0.62 (0.59–0.65)
Cerebrovascular accident	2,746 (35.90%)	544,282 (23.88%)	0.66 (0.64–0.68)
Coronary artery disease	3,391 (44.33%)	632,594 (27.75%)	0.54 (0.52–0.55)
Chronic kidney disease	2,819 (36.85%)	439,731 (19.29%)	0.52 (0.50–0.54)
Liver disease	966 (12.62%)	101,977 (4.47%)	0.50 (0.46–0.54)
Human immunodeficiency virus	58 (0.75%)	23,979 (1.05%)	1.39 (1.07–1.79)
Diabetes	4,459 (58.29%)	1,063,358 (46.65%)	0.80 (0.78–0.81)
Depression	3,625 (47.39%)	861,391 (37.79%)	0.79 (0.77–0.81)
Obesity	3,891 (50.86%)	1,009,554 (44.29%)	0.87 (0.85–0.89)
Rheumatoid arthritis	1,132 (14.79%)	86,464 (3.79%)	0.25 (0.24–0.27)
Dementia	804 (10.51%)	203,554 (8.93%)	0.83 (0.77–0.89)

Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; COVID-19 = coronavirus disease 2019; ILD = interstitial lung disease; OR = odds ratio.

Data presented as mean ± standard deviation where applicable.

Comparing the outcome variables between the two groups (COVID-19 with vs. without ILD) indicated a higher prevalence of 30-day readmission in patients with a positive history of ILD compared with those without ILD (14.95% vs. 9.84%; odds ratio, 1.61; 95% confidence interval [CI], 1.46–1.77; P < 0.0001).

Kaplan-Meier analysis revealed that the probability of 30-day readmission in the COVID + ILD group was 0.154 (95% CI, 0.146–0.162), significantly higher than in patients with COVID-19 without ILD, who had a 30-day readmission probability of 0.101 (95% CI, 0.094–0.108; P < 0.0001). ICU admission was also more probable in the COVID + ILD group (0.098; 95% CI, 0.091–0.105) than in the other group (0.057; 95% CI, 0.051–0.062; P < 0.0001). No significant differences between the two groups were shown in patient referral to hospice upon discharge or the incidence of PE (P > 0.05) (Figure 1A and Table 2).

Figure 1.

Kaplan-Meier curves for the probability of outcomes in the coronavirus disease 2019 cohort (A), ILD cohort (B), idiopathic pulmonary fibrosis (IPF) cohort (C), and non–IPF cohort (D). Outcomes from left to right include 30-day readmission, discharge to hospice, intensive care unit admission, and pulmonary embolism. COVID-19 = coronavirus disease 2019; ICU = intensive care unit; ILD = interstitial lung disease.

Table 2.

Survival analysis of the probabilities of positive and negative ILD outcomes in the cohort of patients with COVID-19

	History of ILD Probability (95% CI)		P Value
	Positive	Negative
30-d readmission	0.154 (0.146–0.162)	0.101 (0.094–0.108)	<0.0001
Hospice	0.009 (0.007–0.012)	0.008 (0.006–0.010)	0.3
ICU admission	0.098 (0.091–0.105)	0.057 (0.051–0.062)	<0.0001
Pulmonary embolism	0.001 (0.000–0.002)	0.001 (0.000–0.001)	0.6

Definition of abbreviations: COVID-19 = coronavirus disease 2019; ICU = intensive care unit; ILD = interstitial lung disease.

ILD Cohort

Among the total of 118,892 patients with a history of ILD, 7,649 patients who were admitted with a primary diagnosis of COVID-19 were selected, forming the ILD + COVID group. Among this group, 3,093 patients (40.43%) were male, with a mean ECI score of 10.43 ± 4.51 (SD). Regarding smoking history, 3,605 patients (47.13%) reported a positive history. The most prevalent comorbidities in this group were hypertension (88.74%) and COPD (64.80%). Table 3 outlines the baseline characteristics of the patients with ILD and COVID-19 compared with the rest of the main ILD cohort, which includes patients with ILD admitted to the hospital for reasons other than COVID-19. The control group (patients with ILD admitted for non–COVID-19 pneumonia) was extracted from the baseline ILD cohort and matched using propensity scores.

Table 3.

Baseline characteristics of patients in the ILD cohort based on COVID-19 findings

Characteristic	ILD Patients Admitted for COVID-19 (n = 7,649)	ILD Patients Admitted for Other Reasons (n = 111,243)	OR (95% CI)
Age, yr	68.43 ± 12.89	69.69 ± 12.14	—
Male sex	3,093 (40.43%)	47,220 (42.44%)	1.08 (1.03–1.13)
Drug and smoking history
Drug abuse	1,137 (14.86%)	13,208 (11.87%)	0.77 (0.72–0.82)
Alcohol abuse	637 (8.32%)	7,450 (6.69%)	0.79 (0.72–0.85)
Smoker	3,605 (47.13%)	4,044 (3.63%)	0.90 (0.86–0.94)
Elixhauser comorbidity index	10.43 ± 4.51	8.98 ± 4.18	—
Comorbidities
Myocardial infarction	433 (5.66%)	4,832 (4.34%)	0.75 (0.68–0.83)
Hypertension	6,788 (88.74%)	96,310 (86.57%)	0.81 (0.76–0.87)
Heart failure	3,564 (46.59%)	42,429 (38.14%)	0.70 (0.67–0.74)
Pulmonary embolism	291 (3.80%)	3,517 (3.16%)	0.82 (0.73–0.93)
COPD	4,957 (64.80%)	66,769 (60.02%)	0.81 (0.77–0.85)
Asthma	2,457 (32.12%)	28,507 (25.62%)	0.72 (0.69–0.76)
Any type of cancer	1,628 (21.28%)	24,156 (21.71%)	1.02 (0.96–1.08)
Cerebrovascular accident	2,746 (35.90%)	35,732 (32.12%)	0.84 (0.80–0.88)
Coronary artery disease	3,931 (51.39%)	53,251 (47.86%)	0.86 (0.82–0.90)
Chronic kidney disease	2,820 (36.86%)	34,240 (30.77%)	0.76 (0.72–0.79)
Human immunodeficiency virus	58 (0.75%)	701 (0.63%)	0.82 (0.63–1.08)
Diabetes	4,459 (58.29%)	55,023 (49.46%)	0.70 (0.66–0.73)
Depression	2,878 (37.62%)	35,387 (31.81%)	0.77 (0.73–0.81)
Obesity	3,891 (50.86%)	47,956 (43.10%)	0.73 (0.69–0.76)
Rheumatoid arthritis	1,132 (14.79%)	14,337 (12.88%)	0.85 (0.79–0.90)
Dementia	804 (10.51%)	7,290 (6.55%)	0.59 (0.55–0.64)

Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; COVID-19 = coronavirus disease 2019; ILD = interstitial lung disease; OR = odds ratio.

Data presented as mean ± standard deviation where applicable.

The rate of 30-day all-cause hospital readmission was significantly higher among patients with ILD admitted for COVID-19 compared with patients with ILD admitted for non–COVID-19 pneumonia (14.72% vs. 11.75%; P < 0.0001). Kaplan-Meier analysis demonstrated a higher probability of 30-day readmission for patients with ILD admitted for COVID-19 compared with the control group (P < 0.0001). The probability of 30-day readmission for patients with ILD who were COVID-19–positive was 0.103 (95% CI, 0.097–0.110), compared with 0.006 (95% CI, 0.003–0.008) for patients with ILD admitted for non–COVID-19 pneumonia.

In this cohort, the probability of ICU admission was another outcome variable that differed between the two groups (P = 0.004). The probability of discharge to hospice and the incidence of PE were not different between the two groups (P > 0.05). The results of the Kaplan-Meier analysis are presented in Table 4 and Figure 1B.

Table 4.

Survival analysis of the probability of outcomes in positive and negative COVID-19 groups in the cohort of patients with ILD

	COVID-19 Probability (95% CI)		Log-Rank P Value
	Positive	Negative
30-d readmission	0.103 (0.097–0.110)	0.006 (0.003–0.008)	<0.0001
Hospice	0.010 (0.007–0.012)	0.009 (0.006–0.012)	0.7
ICU admission	0.069 (0.063–0.075)	0.055 (0.048–0.062)	0.004
Pulmonary embolism	0.003 (0.002–0.004)	0.002 (0.001–0.004)	0.5

Definition of abbreviations: COVID-19 = coronavirus disease 2019; ICU = intensive care unit; ILD = interstitial lung disease.

IPF subcohort

Among the 28,532 patients diagnosed with IPF, 1,397 patients were admitted for COVID-19 (i.e., IPF + COVID). The mean age of patients in this group was 72 ± 10 (SD) years, with approximately 49% being female. The smoking rate in this group was 54.97%, and approximately 11% had a history of drug abuse. The most common comorbidity among patients with IPF admitted for COVID-19 was hypertension, with a prevalence of 89%. To calculate the propensity score, significantly different variables were applied based on the comparison of basic characteristics of this group with the rest of the IPF cohort who tested negative for COVID-19 (Table 5). The control group was selected among patients with IPF who were admitted for reasons other than COVID-19 and matched at a 1:1 ratio using propensity scores.

Table 5.

Baseline characteristics of patients in the ILD cohort IPF subgroup regarding positive or negative COVID-19 results

Characteristic	COVID-19 Positive (n = 1,397)	COVID-19 Negative (n = 27,135)	OR (95% CI)
Age, yr	72 ± 10	73 ± 9	—
Male sex	711 (50.89%)	14,884 (54.85%)	1.17 (1.05–1.3)
Drug and smoking history
Drug abuse	154 (11.02%)	2,073 (7.63%)	0.66 (0.56–0.79)
Alcohol abuse	109 (7.80%)	1,564 (5.76%)	0.72 (0.59–0.88)
Smoker	768 (54.97%)	13,920 (51.29%)	0.86 (0.77–0.96)
Elixhauser comorbidity index	10 ± 4.9	8.47 ± 4	—
Comorbidities
Myocardial infarction	190 (13.60%)	2,477 (9.12%)	0.63 (0.54–0.74)
Hypertension	1,247 (89.26%)	23,762 (87.56%)	0.84 (0.71–1.007)
Heart failure	217 (15.53%)	3,346 (12.33%)	0.76 (0.65–0.88)
Pulmonary embolism	60 (4.29%)	850 (3.13%)	0.72 (0.55–0.94)
COPD	946 (67.71%)	15,975 (58.87%)	0.68 (0.60–0.76)
Asthma	406 (29.06%)	5,354 (19.73%)	0.60 (0.53–0.67)
Any type of cancer	275 (19.68%)	5,737 (21.14%)	1.09 (0.95–1.25)
Cerebrovascular accident	555 (39.72%)	9,023 (33.25%)	0.75 (0.67–0.84)
Coronary artery disease	809 (57.90%)	14,433 (53.18%)	0.82 (0.74–0.92)
Chronic kidney disease	498 (35.64%)	7,915 (29.16%)	0.74 (0.66–0.83)
Liver disease	1,043 (74.65%)	22,321 (82.25%)	0.63 (0.61–0.65)
Human immunodeficiency virus	9 (0.64%)	109 (0.40%)	0.62 (0.31–1.23)
Diabetes	835 (59.77%)	13,663 (50.35%)	0.68 (0.61–0.76)
Depression	584 (41.80%)	9,715 (35.80%)	0.77 (0.69–0.86)
Obesity	655 (46.88%)	10,439 (38.47%)	0.70 (0.63–0.79)
Rheumatoid arthritis	162 (11.59%)	2,344 (8.63%)	0.72 (0.60–0.85)
Dementia	136 (9.73%)	1,568 (5.77%)	0.56 (0.47–0.68)

Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; COVID-19 = coronavirus disease 2019; ILD = interstitial lung disease; IPF = idiopathic pulmonary fibrosis; OR = odds ratio.

Data presented as mean ± standard deviation where applicable.

When comparing outcome variables between these two groups (patients with IPF who were admitted for COVID-19 vs. patients with IPF who were admitted for other reasons), the odds ratio for 30-day readmission was 3.20 (95% CI, 2.49–4.12; P < 0.0001). Furthermore, the P value of the log rank test was less than 0.0001, indicating a significant difference in the probability of readmission between the two groups. Other outcome variables, including referral to hospice upon discharge, ICU admission, and the incidence of PE, did not differ between the two groups (Figure 1C and Table 6).

Table 6.

Survival analysis of the probability of positive and negative COVID-19 outcomes in the IPF subcohort

	COVID-19 Probability (95% CI)		Log-Rank P Value
	Positive	Negative
30-d readmission	0.058 (0.046–0.070)	0.029 (0.020–0.038)	<0.0001
Hospice	0.010 (0.005–0.015)	0.008 (0.003–0.013)	0.6
ICU admission	0.030 (0.021–0.039)	0.021 (0.013–0.028)	0.1
Pulmonary embolism	0.001 (0.000–0.002)	0.001 (0.000–0.002)	1

Definition of abbreviations: CI = confidence interval; COVID-19 = coronavirus disease 2019; ICU = intensive care unit; IPF = idiopathic pulmonary fibrosis.

Non-IPF subcohort

Among the 17,477 patients with ILD with no history of IPF, 913 were admitted with a COVID-19 diagnosis. As outlined in Table 7, the mean age of these patients was 62.86 ± 13.36 (SD) years, with approximately 41% being male. The mean ECI score in this group was 10.31 ± 4.80 (SD), and hypertension was the most prevalent comorbidity, reported in 805 patients (88.17%). Age and sex were used to match with a group of subjects without IPF and with no history of COVID-19.

Table 7.

Baseline characteristics of patients in the ILD cohort non-IPF subgroup regarding positive or negative COVID-19 results

	COVID-19 Positive (n = 913)	COVID-19 Negative (n = 16,564)	OR (95% CI)
Age, yr	62.86 ± 13.36	65.35 ± 13.53	—
Male sex	370 (40.52%)	7,021 (42.38%)	1.07 (0.94–1.23)
Drug and smoking history
Drug abuse	125 (13.69%)	1,234 (7.44%)	0.50 (0.41–0.61)
Alcohol abuse	80 (8.76%)	855 (5.16%)	0.56 (0.44–0.72)
Smoker	431 (47.20%)	6,106 (36.86%)	0.65 (0.57–0.74)
Elixhauser comorbidity index	10.31 ± 4.80	7.26 ± 4.01	—
Comorbidities
Myocardial infarction	39 (4.27%)	589 (3.55%)	0.82 (0.59–1.15)
Hypertension	805 (88.17%)	12,967 (78.28%)	0.48 (0.39–0.59)
Heart failure	407 (44.57%)	5,175 (31.24%)	0.56 (0.49–0.64)
Pulmonary embolism	43 (4.70%)	591 (3.56%)	0.74 (0.54–1.02)
COPD	608 (66.59%)	9,740 (58.80%)	0.71 (0.62–0.82)
Asthma	351 (38.44%)	3,061 (18.20%)	0.36 (0.31–0.41)
Any type of cancer	203 (22.23%)	3,604 (21.75%)	0.97 (0.82–1.14)
Cerebrovascular accident	295 (32.31%)	4,234 (25.56%)	0.71 (0.62–0.82)
Coronary artery disease	451 (49.39%)	6,286 (37.94%)	0.62 (0.54–0.71)
Chronic kidney disease	296 (32.42%)	3,479 (21.00%)	0.55 (0.48–0.63)
Liver disease	258 (28.25%)	2,821 (17.03%)	0.52 (0.44–0.60)
Human immunodeficiency virus	12 (1.31%)	133 (0.80%)	0.60 (0.33–1.10)
Diabetes	558 (61.11%)	7,365 (44.46%)	0.50 (0.44–0.58)
Depression	363 (39.75%)	3,266 (19.71%)	0.37 (0.32–0.42)
Obesity	514 (56.29%)	4,970 (30.00%)	0.33 (0.29–0.38)
Rheumatoid arthritis	124 (13.58%)	1,656 (9.99%)	0.70 (0.58–0.85)
Dementia	112 (12.26%)	771 (4.65%)	0.34 (0.28–0.43)

Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; COVID-19 = coronavirus disease 2019; ILD = interstitial lung disease; IPF = idiopathic pulmonary fibrosis; OR = odds ratio.

Data presented as mean ± standard deviation where applicable.

In the group of subjects without IPF and with COVID-19, 155 patients (16.97%) were readmitted, which was a higher incidence than in the control group, with 58 readmissions within 30 days (3.90%; P < 0.0001). Other outcomes were not statistically different between the two groups (P > 0.05). Results of Kaplan-Meier analysis are presented in Table 8 and Figure 1D. For additional details, see data supplement.

Table 8.

Survival analysis of the probability of positive and negative COVID-19 outcomes in the subcohort of non-IPF patients

	COVID-19 Probability (95% CI)		P Value
	Positive	Negative
30-d readmission	0.064 (0.048–0.079)	0.108 (0.092–0.124)	<0.0001
Hospice	0.012 (0.005–0.019)	0.008 (0.003–0.012)	0.3
ICU admission	0.067 (0.051–0.083)	0.074 (0.061–0.088)	0.5
Pulmonary embolism	0.002 (0.000–0.005)	0.004 (0.001–0.008)	0.4

Definition of abbreviations: COVID-19 = coronavirus disease 2019; ICU = intensive care unit; IPF = idiopathic pulmonary fibrosis.

Discussion

This study demonstrated an odds ratio of 1.6 in the rate of 30-day hospital readmission among patients with COVID-19 with preexisting ILD compared with those without ILD and an odds ratio of 2.4 among patients with ILD who were admitted with COVID-19 compared with those admitted with non–COVID-19 pneumonia. Moreover, our findings suggest that the negative outcomes observed in patients with ILD and COVID-19 may be primarily attributed to COVID-19 rather than ILD alone.

These results are consistent with current literature indicating that COVID-19 exacerbates outcomes in patients with ILD (1, 8, 9, 35–37). In agreement with our results, Nadimpalli and colleagues reported that 11.1% of patients with COVID-19 who were hospitalized experienced all-cause hospital readmission within 30 days of their initial hospitalization (38). Furthermore, their findings revealed that chronic lung disease was among the factors most associated with readmission (odds ratio, 1.29; 95% CI, 1.24–1.34). Castillo and colleagues investigated the rate of all-cause hospital readmissions within 30 days of an index hospitalization in 98 patients with ILD and reported a readmission rate of 20.2% (23). However, the potential impact of ILD exacerbation on the clinical course of COVID-19 has yet to be investigated.

The subgroup analysis of patients with IPF revealed a higher rate of all-cause hospital readmission in those who contracted COVID-19. However, the groups had no significant difference in hospice referral or ICU admission. This result was unexpected because IPF is regarded as the most severe form of ILD (39). Like patients with IPF, patients with non-IPF ILD who were admitted with COVID-19 showed a higher probability of 30-day readmission compared with the control group. Furthermore, we found no evidence of an increased likelihood of PE occurrence in any of the cohorts or ILD subtypes irrespective of COVID-19 status.

In a multicenter observational nested case-control study conducted by Lee and colleagues (8) in a Korean population, a twofold increase in the odds of COVID-19 diagnosis was found among individuals with ILD. This indicates that ILD may make individuals more susceptible to a more severe and symptomatic disease course, resulting in increased use of O₂ therapy, higher rates of ICU admission, longer durations of mechanical ventilation, and higher mortality rates (13.4% vs. 2.8%). Although this study had the advantage of being a large national project and providing results that can be generalized, it did not examine the rates of 30-day hospital readmission. Moreover, the study had limitations, including a failure to adjust for SARS-CoV-2 variants and relevant baseline comorbidity variables such as smoking and body mass index, which could have introduced confounding effects (40, 41), and no comparison group from the ILD population without COVID-19.

Another international multicenter study involving 349 participants was conducted to assess the in-hospital mortality rate of patients with ILD who were admitted with respiratory SARS-CoV-2 infection. The study aimed to determine whether there was a higher mortality rate in the group with ILD plus COVID-19 compared with the group without a history of ILD. The results showed a higher mortality rate in the ILD plus COVID-19 group (49%) compared with the group without a history of ILD (35%) (1). In a single-center study, Guiot and colleagues (36) suggested that patients with ILD affected by COVID-19 may have similar overall outcomes to the general population. Our study, which had a larger sample size, suggests that patients with ILD who contract COVID-19 have significantly worse clinical outcomes.

The strength of our analysis lies in the employment of a unique design that enabled us to explore the interaction of COVID-19 and ILD on the outcomes of hospitalized patients in a large national-level cohort. We used a nationwide medical record database to examine the primary outcome of 30-day all-cause hospital readmission, along with several secondary outcomes, in two distinct cohorts of patients diagnosed with COVID-19 or ILD upon admission to the hospital. Furthermore, to minimize the potential confounding effect of SARS-CoV-2 variants over time, we matched the study population based on the timing of incidents.

It is important to note that our study has several limitations that should be considered when interpreting our findings. A significant drawback of this study is the low sensitivity of the ICD codes we used (9th and 10th Revisions, Clinical Modifications) for COVID-19 and ILD, which may lead to underdiagnosis of some cases. Although our study highlights the heightened risk of adverse outcomes in patients with ILD who contract COVID-19, we did not assess the clinical severity of ILD or the treatment strategies employed, such as immunosuppressive therapies, which may have impacted the outcomes of COVID-19. We used propensity score matching to control for confounding variables; however, this approach may have excluded patients at the extremes of the propensity score distribution, which could have introduced bias.

Given the potential impact of various SARS-CoV-2 variants on patient outcomes, we conducted time-matched analysis of outcome measures. However, because we did not have access to the SARS-CoV-2 variants in our database, we were unable to directly account for their potential influence on the observed outcomes.

Finally, it is important to mention that the time at risk is an important variable that we could not consider in this study. Future studies may consider using incidence density sampling to select controls that are more reflective of the underlying population and account for the time at risk (42).

Conclusions

Patients with ILD experience a substantial rate of hospital readmissions within 30 days. However, the presence of COVID-19 increases the readmission rates for these patients. These heightened readmission rates are associated with adverse outcomes, including decreased health-related quality of life and increased healthcare costs for patients and hospitals. These findings highlight the importance of implementing a more cautious preventive approach to COVID-19 in patients with ILD and emphasizing the need for effective postdischarge care planning.