A biopsychosocial analysis of risk factors for persistent physical, cognitive, and psychological symptoms among previously hospitalized post-COVID-19 patients

Abstract

A significant number of COVID-19 survivors continue to experience persistent physical, cognitive, and psychological symptoms up to one year after discharge. This study aimed to examine the frequency, severity, and progression of and risk factors for these symptoms. This single-centre retrospective cohort study included 126 COVID-19 patients admitted to the VieCuri Medical Centre between 2020 and 2022. Follow-up assessments were conducted at 3 and 12 months postdischarge, including pulmonary function tests, CT scans, bioimpedance analysis, and questionnaires on physical, cognitive, and psychological symptoms. At both follow-up assessments, 31–32% of patients reported moderate to severe physical symptoms, 26–27% reported multiple cognitive symptoms, and 14–18% experienced depressive or posttraumatic stress symptoms (PTSSs). Only anxiety symptoms significantly decreased between the 3-month follow-up and the 12-month follow-up (from 22 to 12%; p = .014). The persistence of symptoms at 12 months was significantly associated with premorbid conditions (chronic respiratory disease, multiple comorbidities), illness severity (infection during the third wave), physical factors (COVID-19-related pulmonary abnormalities, lower total lung capacity, and dyspnoea), and cognitive and psychological factors (cognitive symptoms, anxiety, depression, and PTSS) (p < .05). These findings suggest that a significant proportion of COVID-19 survivors continue to experience persistent symptoms due to biopsychosocial factors, thus emphasizing the need for a biopsychosocial approach in early screening and treatment.

Introduction

In 2019, severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), which is the virus that causes COVID-19, spread across the world and led to a pandemic. Approximately 774 million individuals across the world have been infected with COVID-19, and nearly 12 million people have died from the disease¹. In the Netherlands, an estimated 8.6 million people have been infected by the virus¹. The number of admissions due to this disease was reliably recorded until March 2023, with nearly 20 thousand patients requiring admission to the intensive care unit (ICU) and almost 124 thousand patients requiring admission to a nursing ward². Previous research has shown that a subgroup of COVID-19 patients experience persistent physical, cognitive, and psychological symptoms^3–5, thus posing significant challenges for patients and global health care systems. The World Health Organization (WHO) has defined the presence of symptoms that persist or newly develop within three months after SARS-CoV-2 infection and last for a minimum of two months as post-COVID-19 condition⁶. The global estimated pooled prevalence of post-COVID-19 condition is 54% for those hospitalized, whereas it is 34% for nonhospitalized patients⁷. The most prevalent symptoms of post-COVID-19 condition are dyspnoea (34%), concentration problems (32%), and fatigue (31%)⁸.

The mechanisms underlying these persistent post-COVID-19 symptoms are multifaceted, encompassing both biological and psychological factors^9,10. Previous research has shown that COVID-19 affects various body systems, including the respiratory, cardiovascular, neurological, gastrointestinal, and musculoskeletal systems¹¹. Furthermore, preexisting chronic respiratory diseases, including asthma and chronic obstructive pulmonary disease (COPD), have also been shown to be associated with persistent long-term symptoms post-COVID-19^12,13. However, less is understood about potential psychological mechanisms underlying post-COVID-19 condition^14,15, even though associations have been shown between post-COVID-19 condition and psychological variables, including a history of psychiatric diagnosis, depression, anxiety, and posttraumatic stress disorder among both hospitalized and nonhospitalized patients^16,17.

Understanding post-COVID-19 condition seems to require a perspective that integrates biological, psychological, and social factors⁹. This approach may aid in understanding the aetiology and recovery of post-COVID-19 condition, and it may facilitate the development and optimization of multimodal and interdisciplinary treatment strategies. Identifying modifiable psychological and physiologic factors could enhance the treatment of post-COVID-19 condition, thereby enhancing patient resilience¹⁵. However, most studies predominantly adopt either a biomedical or a psychological approach.

The objectives of the current study were as follows: (1) to explore the frequency, severity, and course of physical, cognitive, and psychological symptoms in the first year after discharge among formerly hospitalized COVID-19 patients; and (2) to identify the differential contributions of biopsychosocial factors that are associated with poor outcomes – i.e., persistent physical and cognitive symptoms, depression, anxiety, and posttraumatic stress symptoms (PTSSs) – at twelve months after discharge.

Methods

Design and participants

This study included a cohort of adult COVID-19 patients admitted to the VieCuri Medical Centre in Venlo, Netherlands, between February 2020 and February 2022 (hospital admission = T0). All discharged patients aged 18 years and above with SARS-CoV-2 infection (confirmed via quantitative polymerase chain reaction) were invited to the post-COVID-19 outpatient clinic at both 3 and 12 months postdischarge (T1 and T2, respectively). Inclusion in the study required completing assessments at both 3 and 12 months.

Procedure

As part of regular care, all patients were invited to undergo a standardized outpatient follow-up assessment 3 months after discharge (T1). Patients exhibiting clinical concerns during T1 were referred to medical and psychological specialists, such as cardiologists or medical psychologists. Patients were then treated according to the current guidelines from the Dutch National Institute for Public Health and the Environment¹⁸. Patients who experienced symptoms at T1 were automatically invited for a standardized 12-month follow-up assessment (T2). This assessment was optional for patients without symptoms at T1. The outpatient follow-up assessments included physical, cognitive, and psychological assessments.

Demographic, premorbid, and COVID-19 illness severity factors (i.e., hospitalization characteristics), as well as the results of the questionnaires, were retrieved from the patients’ electronic medical records.

All experimental protocols were approved by the Medical Ethical Committee of Maastricht University Medical Centre. The Medical Ethical Committee of Maastricht University Medical Centre waived the requirement of informed consent due to the retrospective nature of the study. Patients attending the outpatient clinic were informed that their routinely collected clinical data could be used for research purposes and were given the option to opt out. All study procedures were conducted in accordance with the research guidelines and regulations of VieCuri Medical Centre.

Measures

See Table 1 for an overview of all the measures and variables.

Table 1.

Overview of the measures and variables.

Measure	Variables	Timepoint
Outcome measures
4DSQ – somatization subscale	Physical symptoms	T1 and T2
CLCE-24 – cognitive subscale	Cognitive symptoms	T1 and T2
HADS	Depression and anxiety symptoms	T1 and T2 T1 and T2
PTSS-14	PTSS	T1 and T2
Predictor variables
Electronic medical files	Age	T0
Electronic medical files	Sex	T0
Electronic medical files	BMI	T0
Electronic medical files	Comorbidities	T0
CCI	Ten-year mortality prediction	T0
WHO COVID-19 disease severity categorization	Illness severity	T0
Prebronchodilator spirometry	FEV1	T1
Prebronchodilator spirometry	FVC	T1
	FEV1/FVC	T1
Body plethysmography	TLC	T1
Body plethysmography	RV	T1
Single-breath method	DLCO	T1
	DLCO/VA	T1
MicroRPM monitor	MIP	T1
MicroRPM monitor	MEP	T1
CT	COVID-related residual pulmonary abnormalities	T1
Bio impedance analysis	FFMI, kg/m²	T1
Consultation internist and pulmonologist	Fatigue	T1
Consultation internist and pulmonologist	Dyspnoea	T1
CERQ	Cognitive coping	T1

4DSQ, Four-Dimensional Symptom Questionnaire (4DSQ); CLCE-24, Checklist for Cognitive and Emotional Consequences; HADS, Hospital Anxiety and Depression Scale; PTSS-14, Posttraumatic Stress Symptoms Checklist-14; BMI, body mass index; CCI, Charlson Comorbidity Index; WHO, World Health Organization; FEV1, forced expiratory volume in one second; FVC, forced vital capacity; TLC, total lung capacity; RV, residual volume; DLCO, diffusion capacity of the lungs for carbon monoxide; DLCO/VA, DLCO per unit alveolar volume; MIP, maximal inspiratory pressure; MEP, maximal expiratory pressure; MicroRPM, Micro Respiratory Pressure Monitor; CT, computed tomography; FFMI, fat-free mass index; CERQ, Cognitive Emotion Regulation Questionnaire.

Outcome variables at T1 and T2

Physical symptoms.The Four-Dimensional Symptom Questionnaire (4DSQ) measures the tendency to experience distress, somatization, depression, and anxiety¹⁹. For this study, data from the somatization subscale, which assesses physical symptoms over the past weeks, were used. The somatization subscale comprises 16 items (scores range from 0 to 32), with scores between 11 and 20 indicating moderate physical symptoms. Scores > 20 indicate severe physical symptoms²⁰. In response to evolving clinical practices, this questionnaire was added in February 2021.

Cognitive symptoms. The Checklist for Cognitive and Emotional Consequences (CLCE-24) comprises 24 questions screening for cognitive and emotional symptoms, with 13 questions specifically assessing cognitive symptoms²¹. Scores on the 13 cognition-related questions indicate the presence of subjective cognitive problems (yes/no answers), with higher scores indicating more cognitive symptoms in daily life. We used a cut-off of ≥ 6 based on a mean score of 1.9 (SD = 1.9) among healthy controls on the cognitive subscale of the CLCE-24²².

Depression and anxiety. The Hospital Anxiety and Depression Scale (HADS) is a self-assessment tool for depression and anxiety symptoms²³. This scale comprises anxiety (HADS-A) and depression (HADS-D) subscales, each of which includes seven items. The items are scored on a four-point Likert scale (0–3), yielding subscale scores ranging from 0 to 21, with higher scores indicating higher levels of anxiety or depression. A cut-off score of ≥ 8 per subscale indicates clinically relevant symptoms of anxiety or depression²⁴.

Posttraumatic stress symptoms (PTSSs). The Post-Traumatic Stress Syndrome 14-questions Inventory (PTSS-14) is a screening questionnaire that identifies patients at risk of posttraumatic stress disorder (PTSD)²⁵. It includes 14 items on a seven-point Likert scale (1–7), with scores ranging from 14 to 98; higher scores indicate higher levels of PTSS. A cut-off score of ≥ 45 indicated high levels of PTSS²⁵.

Predictor variables at T0

The demographic factors included age and sex. Premorbid factors included body mass index (BMI), comorbidities (e.g., chronic respiratory disease, hypertension, and obesity), and chronic respiratory disease. To reduce the number of independent variables, the Charlson Comorbidity Index (CCI) was calculated with a score ranging from 0 to 37, indicating a prediction of ten-year mortality for a patient who may have a spectrum of comorbid conditions²⁶.

COVID-19 illness severity

COVID-19 illness severity was determined on the basis of hospitalization characteristics, including length of hospital stay, type of ward (ICU or nursing ward), and treatment, including oxygen supplementation and invasive and/or noninvasive ventilation. Patients were categorized as having moderate, severe, or critical disease based on the WHO COVID-19 disease severity categorization²⁷.

We divided patients into different waves of SARS-CoV-2 infection on the basis of admission date (first wave between February and July 2020; second wave between August 2020 and January 2021; third wave between February and September 2021; and fourth wave between October 2021 and February 2022)²⁸. The classification of the waves was based on the presence of the most dominant variants in the Netherlands²⁹, as the type of variance is known to be linked to different acute disease severities in patients with COVID-19³⁰.

Predictor variables at T1

Pulmonary function tests were conducted in accordance with the guidelines of the European Respiratory Society³¹ on a MasterScreenTM Body and MasterScreenTM PFT (PanGas, Dagmersellen) via SentrySuite V3.0.5 software. The physical variables included forced expiratory volume in one second (FEV₁), forced vital capacity (FVC), total lung capacity (TLC), residual volume (RV), diffusion capacity of the lungs for carbon monoxide (DLCO), maximal inspiratory pressure (MIP) and maximal expiratory pressure (MEP). The DLCO per unit alveolar volume (DLCO/VA) was also calculated. Pulmonary function parameters are expressed as percentages of the predicted values³². The lower limit of normal (LLN), defined as the 5 th percentile according to the standardized multiethnic reference values for spirometry from the Global Pulmonary Function Initiative, was used to report pulmonary function impairments^33,34. Among the pulmonary function parameters, we included only TLC, DLCO, and MEP in the models since these parameters have been shown to be related to impaired health outcomes post-COVID-19³⁵. The presence or absence of COVID-19-related residual pulmonary abnormalities was determined on the basis of computed tomography (CT) scans. Different radiologists interpreted the images. Other physical variables included fat-free mass index (FFMI), fatigue, dyspnoea, and physical symptoms.

Cognitive and psychological variables

Psychological and cognitive functioning and cognitive coping were assessed. Cognitive coping was measured via the Cognitive Emotion Regulation Questionnaire (CERQ)³⁶. It is a 36-item questionnaire featuring nine conceptually distinct subscales³⁷. The items are scored on a five-point Likert scale (1–5), with subscale scores ranging from 4 to 20. Higher scores indicate more frequent use of coping strategies. Normative data were obtained from the CERQ manual, and Z scores were calculated³⁶. In response to evolving clinical practices, this questionnaire was added in February 2021.

Statistical analysis

Normally and nonnormally distributed variables are expressed as the mean (SD) and median (IQR), respectively. The numbers and proportions are reported for binary variables and variables with cut-off values.

Paired sample t tests (parametric data) or Wilcoxon signed rank tests (nonparametric data) were conducted to investigate the course of physical, cognitive, and psychological symptoms between T1 and T2 (4DSQ, CLCE-24, HADS-anxiety, HADS-depression, and PTSS-14).

Pearson’s correlation analysis (for normally distributed variables) or Spearman’s correlation analysis (for nonnormally distributed variables) were conducted to determine the associations between independent variables at T0 and T1 and outcomes at T2 (4DSQ, CLCE-24, HADS-anxiety, HADS-depression, PTSS-14). Chronic respiratory disease was added as a covariate in all models because of the well-known association between COPD and cognitive impairments, as well as psychological distress (i.e., anxiety and depression)^38–40.

Five hierarchical multiple linear regression analyses were conducted with physical, cognitive, and psychological outcomes at T2 as the dependent variables (4DSQ, CLCE-24, HADS-anxiety, HADS-depression, and PTSS-14). Independent variables significantly associated with dependent variables in the bivariate analyses were entered into multivariable analyses. The following blocks were entered (in sequence): (1) demographic and premorbid factors, (2) COVID-19 illness severity, (3) physical factors at T1, and (4) cognitive and psychological factors at T1. The CERQ and 4DSQ results at T1 were not added to the models, as these variables were added at a later stage, resulting in missing data. The assumptions for linear regression modelling were verified. We transformed the dependent variables via log, square root, and polynomial functions to satisfy the assumption of homoscedasticity.

Sensitivity analyses were performed to assess the robustness of the findings. The analyses explored the contribution of additional physical and psychological variables, specifically the 4DSQ and CERQ subscales, as an additional block. The sensitivity analyses followed the same approach as the primary analyses.

For all analyses, the threshold for statistical significance was a 2-sided alpha level of 0.05. To facilitate comparisons across studies, significance was also reported at commonly used alpha levels of 0.05, 0.01, and 0.001. Data analysis was conducted via SPSS version 26.0⁴¹.

Results

A total of 1,176 patients were admitted to the hospital with SARS-CoV-2 infection between February 2020 and February 2022. Among these patients, 651 attended the outpatient post-COVID-19 clinic at T1, and 448 attended at T2. A total of 126 patients completed the questionnaires at both outpatient assessments and were included in the study (Fig. 1).

Fig. 1.

Flowchart of the (included) study population.

Frequency, severity, and course of physical, cognitive, and psychological symptoms

Table 2 shows the characteristics of the included patients. The median age of the patients was 68 years, and most patients were male (67%). The most common comorbidities were hypertension (42%), obesity (33%), and chronic respiratory disease (29%). Nearly half of the patients were included in the first wave of the study (49%).

Table 2.

General characteristics of the hospitalized COVID-19 patients (n = 126).

Demographic and premorbid factors	N (%)	Median [IQR]
Age in years		68 [61–76]
Male, n (%)	84 (67)
BMI in kg/m2		27.5 [25–31]
Comorbidities, n (%) present
Hypertension	53 (42)
Obesity	41 (33)
Chronic respiratory disease	36 (29)
Type 2 diabetes	30 (24)
Chronic cardiac disease	30 (24)
Autoimmune disorder	19 (15)
Chronic neurologic disease	14 (11)
Rheumatologic disorder	12 (10)
Chronic kidney disease	10 (8)
Malignant neoplasm	8 (6)
CCI score		3 [2–4]
COVID-19 illness severity
Total hospital stay, days		7 [5–14]
ICU admission, n (%)	22 (18)
Length of ICU stay, days		13 [6–34]
Days from discharge to T1		109 [99–129]
Days from discharge to T2		372 [351–405]
Oxygen treatments during hospital stay
Nasal oxygen therapy, n (%)	115 (91)
Noninvasive ventilation, n (%)	7 (6)
Invasive ventilation, n (%)	19 (15)
Severity score, n (%)
Moderate	35 (28)
Severe	68 (54)
Critical	23 (18)
Waves, n (%)
First (Feb-July ’20)	62 (49)
Second (Aug ‘20 - Jan ‘21)	24 (19)
Third (Feb – Sept ’21)	30 (24)
Fourth (Oct ’21 – Feb ‘22)	10 (8)

BMI, body mass index; CCI, Charlson Comorbidity Index; ICU, intensive care unit.

As shown in Tables 3. and 31% of patients reported moderate to severe physical symptoms at T1, and 32% of patients reported moderate to severe physical symptoms at T2, as measured by the 4DSQ. Patients reported a median of three cognitive symptoms at T1 and two symptoms at T2 on the CLCE-24, with 26% and 27% reporting six or more cognitive symptoms at T1 and T2, respectively. Additionally, 22% and 12% of patients had HADS scores exceeding the cut-off for anxiety at T1 and T2, respectively. A total of 17% and 18% of patients had HADS scores exceeding the cut-off for depression at T1 and T2, respectively. A total of 15% and 14% of the patients had PTSS-14 scores exceeding the cut-off at T1 and T2, respectively.

Table 3.

Results of outcome measures at T1 and T2.

Outcome variables	n	Median [IQR]	>cut-off n(%)	n	Median [IQR]	>cut-off n (%)	p value
	T1			T2
4DSQ	53	5.0 [1.5–13.5]	17 (32)	124	5.5 [2.0–12.0]	38 (31)	0.798
CLCE-24	121	3 [0–6]	31 (26)*	116	2 [0–6]	31 (27)*	0.711
HADS-depression	118	2 [1–6]	20 (17)	125	3 [1–7]	23 (18)	0.265
HADS-anxiety	117	3 [1–7]	26 (22)	125	2 [0–5]	15 (12)	0.014
PTSS-14	115	22 [17–34]	17 (15)	118	21 [16–34]	16 (14)	0.764

IQR, interquartile range; 4DSQ, Four-Dimensional Symptom Questionnaire (4DSQ); CLCE-24, Checklist for Cognitive and Emotional Consequences; HADS, HADS-depression; Hospital Anxiety and Depression Scale-Depression Subscale; HADS-anxiety, Hospital Anxiety and Depression Scale-Anxiety Subscale; PTSS-14, Posttraumatic Stress Symptoms Checklist-14.* ≥6 physical symptoms.

* ≥6 physical symptoms.

Wilcoxon signed-rank tests revealed that the HADS anxiety score decreased in 54 patients, remained stable in 26 patients, and increased in 36 patients (T=−2.542, p =.014). There were no significant changes in the other outcomes between T1 and T2 (p >.05).

Among pulmonary function tests, TLC, DLCO, and MEP were impaired in 18%, 42%, and 29% of patients, respectively. 68% of patients reported fatigue, and 57% reported dyspnoea. See Table 4.

Table 4.

Results of predictor variables at T1.

Physical variables	Mean ± SD, Median [IQR]	n (% impaired)*
FEV1	93.9 ± 21.7%pred	17/126 (14)
FVC	96.9 ± 17.1%pred	11/126 (9)
FEV1/FVC	98 [89–104] %	19/126 (15)
TLC	97 [86–108] %pred	22/122 (18)
RV	90 [81–104] %pred	24/121 (20)
DLCO	77.6 ± 0.5%pred	52/125 (42)
DLCO/VA	86.6 ± 20.1%pred	34/125 (27)
MIP	98 [66–122] %pred	12/125 (10)
MEP	89.6 ± 34.0%pred	36/125 (29)
COVID-related residual pulmonary abnormalities		105/123 (85)
FFMI, kg/m²	19 [17–20]
Fatigue		85/126 (68)
Dyspnoea		72/126 (57)

Psychological factors	Mean ± SD, Median [IQR]	n (%) low**	n (%) high**
CERQ Self-blame	4 [4–7]	0 (0)	2 (4)
CERQ Acceptance	10 [6–13]	6 (11)	1 (2)
CERQ Rumination	7 [5–9]	1 (2)	1 (2)
CERQ Positive refocusing	11.9 ± 4.6	0 (0)	10 (19)
CERQ Planning	9 [7–13]	11 (21)	0 (0)
CERQ Positive reappraisal	10.5 ± 4.0	4 (8)	0 (0)
CERQ Putting things in perspective	12 [9–16]	0 (0)	1 (2)
CERQ Catastrophizing	5 [4–8]	0 (0)	5 (9)
CERQ Other blame	4 [4–5]	0 (0)	1 (2)

%pred, % predicted; IQR, interquartile range; FEV₁, forced expiratory volume in one second; FVC, forced vital capacity; TLC, total lung capacity; RV, residual volume; DLCO, diffusion capacity of the lungs for carbon monoxide; DLCO/VA, DLCO per unit alveolar volume; MIP, maximal inspiratory pressure; MEP, maximal expiratory pressure; FFMI, fat-free mass index; CERQ, Cognitive Emotion Regulation Questionnaire; * Impaired = below the lower limit of normal (LLN); ** low = < 2 SD as compared to normative data; high = > 2 SD as compared to normative data.

Associations with outcomes at 12 months postdischarge

Physical symptoms

Bivariate analyses revealed significant associations between physical symptoms at T2 and chronic respiratory disease (r =.389, p <.001), COVID-19-related residual pulmonary abnormalities, fatigue, dyspnoea, and physical symptoms at T1 (r values ranging from 0.329 to 0.771, p <.001). Additionally, cognitive symptoms, anxiety, depression, PTSS at T1 (r values from 0.601 to 0.714, p <.001), and coping strategies (i.e., acceptance, catastrophizing, and rumination) (r values from 0.405 to 0.465, p <.01) were found to be significantly associated with physical symptoms at T2. Hierarchical regression (Table 5) revealed that chronic respiratory disease, dyspnoea, and increased anxiety at T1 were significantly associated with physical symptoms at T2; specifically, these factors explained 65.6% of the variance in physical symptoms at T2.

Table 5.

Regression analysis with the 4DSQ score at T2 as the dependent variable (n = 100).

Independent variables	β
	Model 1	Model 2	Model 3	Model 4
Demographic and premorbid factors
Chronic respiratory disease	0.36***	NE	0.26**	0.20**
Physical factors
Fatigue	NE	NE	0.30***	0.03
Dyspnoea	NE	NE	0.24*	0.15 *
COVID-related residual pulmonary abnormalities	NE	NE	0.03	0.02
Cognitive and psychological factors
CLCE-24	NE	NE	NE	0.16
HADS-depression	NE	NE	NE	−0.12
HADS-anxiety	NE	NE	NE	0.54 ***
PTSS-14	NE	NE	NE	0.12
R²	0.13		0.31	0.66
Adjusted R²	0.12		0.28	0.63
F change	14.79***		8.03***	22.68***

The square root transformation of the 4DSQ was the dependent variable.

NE, not entered; 4DSQ, Four-Dimensional Symptom Questionnaire (4DSQ); CLCE-24, Checklist for Cognitive and Emotional Consequences; HADS, HADS-depression; Hospital Anxiety and Depression Scale-Depression Subscale; HADS-anxiety, Hospital Anxiety and Depression Scale-Anxiety Subscale; PTSS-14, Posttraumatic Stress Symptoms Checklist-14.

*p <.05; ** p <.01; ***p <.001.

Cognitive symptoms

Bivariate analyses indicated that cognitive symptoms at T2 were significantly associated with BMI, the CCI, chronic respiratory disease (r =.198 to 0.368, p <.05), wave 3 vs. wave 1 (r =.275, p <.01), dyspnoea, fatigue, physical symptoms at T1 (r =.266 to 0.539, p <.01), and several psychological measures (i.e., cognitive symptoms, anxiety, depression, PTSS, and rumination) (r values from 0.293 to 0.708, p <.05). Hierarchical regression (Table 6) revealed that chronic respiratory disease, wave 3 vs. wave 1, and cognitive symptoms at T1 were significantly associated with more cognitive symptoms at T2, and these factors explained 63.2% of the variance in the dependent variable.

Table 6.

Regression analysis with CLCE-24 at T2 as the dependent variable (n = 93).

Independent variables	β
	Model 1	Model 2	Model 3	Model 4
Demographic and premorbid factors
BMI	0.20*	0.21*	0.16	0.04
CCI	0.13	0.14	0.19*	0.16*
Chronic respiratory disease	0.32**	0.31 **	0.22*	0.20**
COVID-19 illness severity
Wave 3 vs. wave 1	NE	0.29**	0.27**	0.17*
Physical factors
Fatigue	NE	NE	0.34***	0.15
Dyspnoea	NE	NE	0.21*	0.13
Cognitive and psychological factors
CLCE-24	NE	NE	NE	0.50***
HADS-depression	NE	NE	NE	−0.00
HADS-anxiety	NE	NE	NE	−0.16
PTSS-14	NE	NE	NE	0.22
R²	0.17	0.25	0.43	0.63
Adjusted R²	0.14	0.22	0.39	0.59
F change	6.00***	10.06**	13.28***	11.28***

The log transformation of CLCE-24 was the dependent variable.

NE, not entered; BMI, body mass index; CCI, Charlson Comorbidity Index; CLCE-24, Checklist for Cognitive and Emotional Consequences; HADS, HADS-depression; Hospital Anxiety and Depression Scale-Depression Subscale; HADS-anxiety, Hospital Anxiety and Depression Scale-Anxiety Subscale; PTSS-14, Posttraumatic Stress Symptoms Checklist-14.

*p <.05; ** p <.01; ***p <.001.

Depression

Bivariate analyses revealed that depressive symptoms at T2 were significantly associated with chronic respiratory disease, wave 3 vs. wave 1 (r =.274 to 0.347, p <.01), TLC, dyspnoea, fatigue, physical symptoms at T1 (r=-.204 to 0.460, p <.05), cognitive symptoms, anxiety, depression, PTSS at T1 (r values from 0.536 to 0.716, p <.001), and coping strategies (i.e., rumination, positive refocusing, putting things in perspective) (r =.277 to − 0.389, p <.05). Hierarchical regression (Table 7.) revealed that chronic respiratory disease, lower TLC, depressive symptoms, and PTSS at T1 were significantly associated with more depressive symptoms at T2, with these factors explaining 62.0% of the variance in the dependent variable.

Table 7.

Regression analysis with HADS-depression (n=103) and HADS-anxiety (n=106) scores at T2 as the dependent variables.

HADS-depression at T2
Independent variables	β
	Model 1	Model 2	Model 3	Model 4
Demographic and premorbid factors
Chronic respiratory disease	0.34***	0.35***	0.32***	0.19*
COVID-19 illness severity
Wave 3 vs. wave 1	NE	0.23*	0.22*	0.13
Physical factors
Fatigue	NE	NE	0.19*	0.01
Dyspnoea	NE	NE	0.14	0.05
TLC	NE	NE	−0.25**	−0.16*
Cognitive and psychological factors
CLCE-24	NE	NE	NE	−0.05
HADS-depression	NE	NE	NE	0.34**
HADS-anxiety	NE	NE	NE	−0.01
PTSS-14	NE	NE	NE	0.38**
R²	0.12	0.17	0.32	0.62
Adjusted R²	0.11	0.15	0.29	0.58
F change	12.53**	5.92*	6.96***	17.24***

HADS-anxiety at T2
Independent variables	β
	Model 1	Model 2	Model 3	Model 4
Demographic and premorbid factors
Chronic respiratory disease	0.23*	0.24*	0.25*	0.12
COVID-19 illness severity
Wave 3 vs. wave 1	NE	0.19	0.17	0.09
Physical factors
Fatigue	NE	NE	0.19	−0.06
TLC	NE	NE	−0.2	−0.13
COVID-related residual pulmonary abnormalities	NE	NE	−0.01	−0.02
Cognitive and psychological factors
CLCE-24	NE	NE	NE	−0.07
HADS-depression	NE	NE	NE	0.06
HADS-anxiety	NE	NE	NE	0.60***
PTSS-14	NE	NE	NE	0.1
R²	0.05	0.09	0.18	0.56
Adjusted R²	0.05	0.07	0.14	0.52
F change	5.53*	3.75	3.52*	19.05***

The square root transformation of HADS-depression scores and the log transformation of HADS-anxiety scores were the dependent variables.

NE, not entered; TLC, total lung capacity; CLCE-24, Checklist for Cognitive and Emotional Consequences; (HADS), HADS-depression; Hospital Anxiety and Depression Scale-Depression Subscale; HADS-anxiety, Hospital Anxiety and Depression Scale-Anxiety Subscale; PTSS-14, Posttraumatic Stress Symptoms Checklist-14.

*p <.05; ** p <.01; ***p <.001.

Anxiety

Bivariate analyses revealed significant associations between levels of anxiety at T2 and wave 3 vs. wave 1 (r =.198, p <.05), fatigue, physical symptoms, TLC, COVID-19-related pulmonary abnormalities (r values from − 0.193 to 0.472, p <.001), cognitive symptoms, anxiety, depression, PTSS at T1 (r =.497 to 0.716, p <.001), and rumination and catastrophizing (r =.377 to 0.490, p <.01). Hierarchical regression (Table 7.) revealed that higher anxiety at T1 was significantly associated with higher anxiety levels at T2, with these factors explaining 56.3% of the variance in the dependent variable.

PTSS

Bivariate analyses revealed significant associations between PTSS at T2 and chronic respiratory disease (r =.241, p <.01), wave 3 vs. wave 1 (r =.193, p <.05), COVID-19-related residual pulmonary abnormalities, fatigue, dyspnoea, and physical symptoms at T1 (r values from 0.214 to 0.596, p <.05). Depression, anxiety, cognitive symptoms, PTSS (r values from 0.581 to 0.762, p <.001), and coping strategies (i.e., acceptance, rumination, and catastrophizing) (r values from 0.318 to 0.552, p <.001) at T1 were also significantly associated with PTSS at T2. Hierarchical regression analysis (Table 8) revealed that COVID-19-related residual pulmonary abnormalities, increased anxiety, and PTSS at T1 were significantly associated with increased PTSS at T2, with these factors accounting for 68.3% of the variance in the dependent variable.

Table 8.

Regression analysis with PTSS-14 score at T2 as the dependent variable (n = 98).

Independent variables	β
	Model 1	Model 2	Model 3	Model 4
Demographic and premorbid factors
Chronic respiratory disease	−0.26*	−0.25*	−0.18	−0.08
COVID-19 illness severity
Wave 3 vs. wave 1	NE	−0.18	−0.15	−0.10
Physical factors at T1
Fatigue	NE	NE	−0.36***	−0.09
Dyspnoea	NE	NE	−0.13	−0.07
COVID-related residual pulmonary abnormalities	NE	NE	−0.11	−0.13*
Cognitive and psychological factors at T1
CLCE-24	NE	NE	NE	0.01
HADS-depression	NE	NE	NE	−0.02
HADS-anxiety	NE	NE	NE	−0.34*
PTSS-14	NE	NE	NE	−0.39**
R²	0.07	0.10	0.28	0.68
Adjusted R²	0.06	0.08	0.24	0.65
F change	6.58*	3.26	7.78***	27.08***

The polynomial transformation of PTSS-14 was the dependent variable.

NE, not entered; CLCE-24, Checklist for Cognitive and Emotional Consequences; HADS, HADS-depression; Hospital Anxiety and Depression Scale-Depression Subscale; HADS-anxiety, Hospital Anxiety and Depression Scale-Anxiety Subscale; PTSS-14, Posttraumatic Stress Symptoms Checklist-14.

*p <.05; ** p <.01; ***p <.001.

Sensitivity analysis

Adding a fifth predictor block (physical symptoms and coping strategies at T1) did not significantly increase the variance explained in the outcomes. However, a higher number of symptoms at T1 was associated with a higher number of physical symptoms at T2 (β = 0.639, 95% CI [0.028–0.165], p =.008).

Discussion

This study revealed that nearly one-fourth of hospitalized COVID-19 patients reported persistent physical, cognitive, and/or psychological symptoms at 3 and 12 months postdischarge. Only anxiety levels significantly decreased in the first year after discharge. Higher numbers of physical, cognitive, and psychological symptoms 12 months postdischarge were associated with premorbid conditions (chronic respiratory disease, higher CCI), illness severity (being infected during the third wave), physical variables (COVID-19-related pulmonary abnormalities, lower TLC, dyspnoea), and cognitive and psychological variables (cognitive symptoms, anxiety, depressive symptoms, and PTSS levels). Furthermore, a higher number of persistent symptoms at 12 months was associated with higher levels of rumination, catastrophizing, and acceptance as well as lower levels of positive refocusing, putting things into perspective, and a higher number of physical symptoms at three months.

Despite a significant decrease in anxiety symptoms over time, a substantial proportion of patients continued to experience clinically significant levels of anxiety, depression, and PTSS at T1 and T2, with prevalence rates ranging from 15 to 22%. In the literature, there is inconsistency in these rates, with some studies finding similar rates of psychological symptoms⁴²and others finding higher rates. For example, compared with nonhospitalized patients who completed similar measures⁴³, the prevalence of psychological symptoms in the present study was lower but remained higher than that in the general population before the COVID-19 pandemic^44,45. In this study, patients were invited to the COVID-19 clinic regardless of symptoms, while in previous studies, patients were often included due to the presence of persistent symptoms^43,46.

Similar prevalence rates of cognitive symptoms have been reported in other studies. A meta-analysis of 43 studies involving both hospitalized and nonhospitalized patients revealed that approximately one in five COVID-19 patients reported cognitive symptoms 12 or more weeks after infection⁴⁷. However, these symptoms and related cognitive difficulties do not necessarily indicate the presence of cognitive impairment. Klinkhammer et al.⁴²reported that 8–10 months postdischarge, 62% of ICU and general ward COVID-19 survivors reported three or more cognitive symptoms, whereas standard neuropsychological testing revealed cognitive dysfunction in only 12% of patients. Moreover, approximately one-third of the patients reported moderate to severe physical symptoms, which is consistent with findings of a review on post-COVID-19 physical symptoms⁴. These symptoms resemble chronic symptoms after other viral infections. Similar to post-COVID-19 condition, after viral infections, the majority of patients completely recover within several weeks, whereas a small subgroup experiences persistent sequelae. The cognitive behavioural model of chronic fatigue syndrome (CFS) may offer an explanatory framework for how chronic symptoms can develop through reciprocal interactions among physiology, cognition, emotion, and behaviour after a viral infection⁴⁸.

The current study identified several prognostic factors associated with physical, cognitive, and psychological functioning twelve months postdischarge. Notably, preexisting chronic respiratory disease emerged as a significant predictor of these symptoms. Similarly, a recent meta-analysis revealed that chronic respiratory diseases, including asthma and COPD, are associated with persistent long-term symptoms following COVID-19¹². Moreover, patients with COPD generally experience higher levels of cognitive impairment and psychological distress (i.e., anxiety and depression) than the general population, which may, in turn, contribute to a greater risk of long-term symptoms post-COVID-19^38–40.

While age and sex were not associated with persistent symptoms in the present study, other demographic factors, such as cognitive reserve and lifestyle, have been shown to influence long-term outcomes after COVID-19 infection. For example, Costas-Carrera et al.⁴⁹ reported that cognitive reserve moderated cognitive function in post-ICU patients after severe COVID-19 infection, with greater cognitive reserve providing a protective effect against cognitive impairments. Similarly, Devita et al.⁵⁰reported that cognitive reserve served as a protective factor against psychological distress in COVID-19 survivors one month after hospital discharge. These findings suggest that cognitive reserve may moderate the impact of the disease at different levels. In contrast, an unhealthy lifestyle, characterized by physical inactivity and poor dietary habits, has been associated with a higher risk and prolonged duration of post-COVID-19 condition^51–53. Reduced physical activity levels in individuals with post-COVID-19 condition have been associated with negative psychological outcomes, including diminished self-esteem, heightened frustration, and feelings of guilt^54,55, while maintaining an active lifestyle appears to have a protective effect⁵². Furthermore, diets high in saturated fats and low in nutrient-rich foods may further increase the risk of developing post-COVID-19 condition⁵³.

Among the illness severity variables, only the infection wave was significantly associated with the outcome. Patients infected with COVID-19 during the third wave (predominantly the beta/gamma variant) reported more long-term physical, cognitive, and psychological symptoms than did those infected during the first wave (wild-type variant). While this finding may suggest that the beta/gamma variant leads to more severe long-term symptoms, it is important to note that infection wave was measured indirectly based on the time of admission rather than confirmed variant identification. There are inconsistent findings in the literature regarding the influence of infection waves on outcomes. Some studies suggested that the wild-type variant (first wave) is more strongly associated with severe COVID-19 symptoms, whereas other studies indicated that later waves (Beta/Gamma/Omicron variants) are linked to greater disease severity^56,57. Given these inconsistencies, our findings, alongside existing research, underscore the need for further investigation to clarify the relationship between COVID-19 variants and long-term symptoms. This is particularly relevant, as our study was unable to directly determine the variant, limiting the ability to establish a precise association.

There are also inconsistent findings regarding the severity of acute infection as a potential risk factor for long-term outcomes. A meta-analysis of 10 studies revealed that patients who required ICU admission during the acute phase of SARS-CoV-2 infection had more than twice the risk of developing persistent symptoms compared with those who did not require ICU admission¹². In contrast, other studies did not find an association between the severity of acute infection and long-term neurological and cognitive functioning, emotional distress, or well-being⁴². In the present study, only 18% of patients were admitted to the ICU, which may explain the absence of an association within our cohort. Notably, COVID-19 illness severity in hospitalized patients is often approximated by differentiating between ICU and general ward admissions. While this categorization provides an indication of illness severity, utilizing standardized severity scores such as the WHO COVID-19 disease severity categorization²⁷, as in the present study, can offer a more differentiated measure of disease severity, distinguishing between various levels of severity.

Among the physical symptoms assessed at three months postdischarge, both dyspnoea and fatigue were associated with physical, psychological, and/or cognitive outcomes in the models that did not include psychological or cognitive factors. However, after controlling for psychological and cognitive factors, only dyspnoea was found to be associated with physical symptoms. The lack of association between fatigue and outcomes in the final models may be explained by the overlap between fatigue and cognitive and psychological symptoms. This overlap was also observed in a recently published study, where fatigue was not associated with outcomes after controlling for other factors, such as depression⁵⁸.

The finding that TLC and COVID-19-related pulmonary abnormalities were linked to psychological outcomes contrasts with previous research, which has indicated that pulmonary function impairments and anomalies detected on chest CT scans at three months were not associated with enduring symptoms at twelve months, including cognitive impairments and physical and psychological symptoms post-COVID-19^59,60. However, in our study, the associations were only observed for depressive and PTSS symptomatology, while other lung function variables also showed no correlation with outcomes. Our contrasting findings could arguably be explained by the greater number of patients with chronic respiratory disease in our study than in other studies, potentially influencing the association between TLC impairments and depressive symptomatology. Nevertheless, these findings underscore the intricate aetiology of the multifaceted symptoms following SARS-CoV-2 infection, extending beyond the biological system.

This study confirmed the impact of psychological factors at three months postdischarge, including depression, anxiety, and PTSS, on physical and psychological symptoms at twelve months postdischarge. Moreover, the regression models revealed a considerable increase in the amount of explained variance when psychological factors were added to the models for all the outcomes. Similarly, cognitive factors at three months were associated with cognitive outcomes at twelve months. These findings are consistent with prior research indicating associations between psychological factors, such as anxiety and depression, and post-COVID-19 condition¹². Given the potential impact of acute infection on physical, cognitive, and psychological functioning, these factors may interact with each other, creating a vicious cycle in which physical, cognitive, and psychological factors reinforce each other, leading to persistent symptoms. Coping likely plays a significant role in this cycle, as has been found in other patient populations, such as those with Lyme disease, fibromyalgia, and stroke^61–64. Rumination and catastrophizing have been shown to be maladaptive strategies, whereas positive refocusing and putting things into perspective have been recognized as adaptive strategies. However, the finding that acceptance was associated with maladaptive outcomes is not consistent with previous research. Studies conducted during the pandemic in various populations have shown positive effects of acceptance strategies on quality of life, resilience, and psychological functioning^65,66. However, to our knowledge, few studies have investigated coping among formerly hospitalized COVID-19 patients⁶⁷. Acceptance reflects acknowledging the reality of the situation³⁷. A potential explanation for the negative association between acceptance and physical and psychological symptoms, as observed in the present study, is that the uncertainty of the situation during the COVID-19 pandemic may have led to feelings of hopelessness upon acceptance. It remains unclear whether patients, in addition to accepting the situation, actively committed to living according to their values and resumed daily activities. Avoidance of these activities may contribute to persistent symptoms. However, this finding requires further investigation.

In addition to coping mechanisms, personality traits such as high levels of optimism and low levels of neuroticism and pessimism may serve as protective factors against the impact of illness, as demonstrated in other diseases, such as stroke^63,64. To our knowledge, this association has not yet been explored in post-COVID-19 patients. Furthermore, previous research has shown that greater perceived social support is associated with fewer persistent psychological symptoms, such as anxiety and depression, following SARS-CoV-2 infection⁶⁸. Additional recent studies have supported these findings, demonstrating that greater perceived social support is associated with fewer depressive symptoms in hospitalized patients⁶⁷, as well as with reduced anxiety and depressive symptoms and improved quality of life on average one year after infection⁶⁹. Additionally, among ICU survivors, a lack of social support has been identified as a risk factor for psychological symptoms three months after discharge⁷⁰. These findings provide further insight into the role of psychological resources and coping mechanisms in mitigating the long-term impact of COVID-19.

The current study had several limitations. First, only patients who completed both outcome assessments were included, thus limiting the sample size. Patients who had recovered at three months may have been less inclined to return for the 12-month follow-up appointment. All hospitalized patients were invited to the clinic, and the data were registered as part of regular care to mitigate selection bias. However, the extent to which these findings can be generalized to a broader population of hospitalized post-COVID-19 patients remains unclear. Second, the findings cannot be generalized to nonhospitalized post-COVID-19 patients. Third, factors that have been shown to be associated with persistent symptoms in previous studies, such as psychiatric history, race, cognitive reserve, lifestyle, personality factors, and perceived social support, were not assessed in this study. Additionally, demographic factors were not included in further analyses, as they were not significant in the bivariate analyses. However, we acknowledge, based on current literature, that female sex is a risk factor for post-COVID-19 condition¹¹. Data collection occurred during the early stages of the COVID-19 pandemic, when studies on the associations between potential determinants and outcomes were scarce.

An inherent strength of this study is the broad spectrum of factors that were included in the analysis and the adoption of a biopsychosocial perspective in comprehending persistent sequelae post-COVID-19, which has significant clinical and research implications.

These findings underscore the persistent physical, cognitive, and psychological symptoms experienced by post-COVID-19 patients, thus highlighting the need for targeted interventions to address these sequelae. This study demonstrated that the interplay between biopsychosocial factors and symptomatology post-COVID-19 emphasizes the importance of incorporating biopsychosocial aspects into post-COVID-19 patient care. Understanding these interactions is crucial for effective interventions^9,71. Based on individualized case conceptualization that includes premorbid, physical, psychological, and cognitive factors, a personalized treatment plan can be devised. The finding that symptoms remain relatively stable over time and that early symptoms predict long-term outcomes supports the need for early screening of patients at risk of long-term problems, which could be targeted with treatment.

There is evidence supporting the effectiveness of multidisciplinary treatments that target biological, psychological, and/or social factors in alleviating post-COVID-19 symptoms. For example, a recent study demonstrated that an inpatient multidisciplinary rehabilitation programme incorporating cognitive behavioural therapy and exercise led to reduced symptom severity, improved self-efficacy, and increased activity and participation⁷². Additionally, cognitive behavioural therapy (CBT) has been shown to effectively alleviate fatigue post-COVID-19 and enhance disease coping in both a feasibility study and a randomized controlled trial^10,73. Moreover, the reduction in PTSS following rehabilitation was associated with decreased fatigue⁷⁴. Further research employing a biopsychosocial perspective is warranted to deepen our understanding of the aetiology and treatment of persistent symptoms.

In conclusion, this study underscores the persistent physical, cognitive, and psychological symptoms experienced by COVID-19 patients postdischarge and the need for targeted interventions. The biopsychosocial perspective provides insights into the complex interplay of factors influencing post-COVID-19 sequelae, emphasizing the importance of personalized interventions that focus on biological and psychological factors. These findings have implications for improving the long-term outcomes and mental health of COVID-19 survivors.

Author contributions

G.C.: Conceptualization, Formal analysis, Methodology, Visualization, Writing – original draft, Writing – review & editing; D.G.: Methodology, Writing – original draft, Writing – review & editing; F.H.M.v.O.: Methodology, Writing – review & editingD.V.: Data curation, Writing – review & editing; J.P.v.d.B.: Data curation, Writing – review & editing; V.v.K.: Data curation; R.J.H.C.G.B.: Writing – review & editing; A.M.W.J.S.: Writing – review & editing; E.v.B.: Conceptualization, Data curation, Writing – review & editing; C.M.v.H.: Conceptualization, Methodology, Writing – review & editing; All authors have approved the submitted version.

Funding

This research received financial support from the Science and Innovation Fund of VieCuri Medical Centre, Netherlands, and Maastricht University, Netherlands. This funding source had no role in the study design; collection, analysis or interpretation of the data; writing of the report; or decision to submit the article for publication.

Data availability

The datasets and scripts used in this study are available from the corresponding author upon request.

Declarations

Competing interests

The authors declare no competing interests.