Abstract
Introduction Coronavirus disease 2019 (COVID-19) is associated with significant acute clinical manifestations, and reports indicate that some patients experience prolonged symptomatology and morbidity. These late clinical manifestations have been termed Post-Acute Sequelae of COVID-19 (PASC) and hypothesised to be associated with clinical severity in the acute infection phase and biochemical abnormalities.
Aim Evaluate the incidence of PASC in previously hospitalised COVID-19 patients and compare the admission and follow-up levels of biochemical parameters stratified according to baseline clinical severity.
Methods N = 168 COVID-19 patients previously hospitalised at the Zan Mitrev Clinic in Skopje, North Macedonia, with matched laboratory data at baseline and follow-up clinical visit > 30 days post-discharge, were stratified according to National Institute of Health clinical severity guidelines as mild, moderate, severe or critical according to admission clinical presentation. We assessed the incidence of PASC and compared the biochemical profile.
Results The median hospitalisation and clinical follow-up period were 11 (9-20) and 53 (30-105) days. The overall incidence of PASC was 56.5% (95/168); most PASC cases were confined to the severe sub-group (61/101, 61.4%). Contrary to mild and moderate cases and a healthy “non-COVID-19” control cohort, we observed that severe COVID-19 cases experienced sustained biochemical disturbances, most notably elevated D-dimers and Ferritin of 600 ng/ml (283-1168) and 432 ng/ml (170-916), respectively.
Conclusions Previously hospitalised severe COVID-19 patients are more likely to experience Post-Acute Sequelae of COVID-19 and prolonged biochemical disturbances, evident by abnormal values of D-dimers and Ferritin.
Introduction
Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is marked by several pathological mechanisms1 and clinical severity phases2 leading to acute morbidity and mortality3. Published studies have elucidated many underlying inflammatory pathologies and coagulopathies that characterise acute infection and the early clinical manifestations4-6. COVID-19 disease may be stratified into five phases7: viral symptom phase, early inflammatory phase, secondary infection phase, multisystem inflammatory phase, and tail phase; the latter characterises disease features experienced beyond 30 days of infection.
Studies8-10 show that a substantial proportion of recovered patients experience persistent clinical symptoms during the tail phase of the disease, most commonly fatigue, dyspnoea, and impaired pulmonary function11 12. These Post-Acute Sequelae of COVID-19 (PASC) have been termed “Long COVID” or “Post-COVID-19 Syndrome” and are found to be more prevalent among hospitalised patients with severe disease13-15. Most research has focused primarily on describing the nature of the PASC experienced by recovered patients4 5; however, the biochemical disturbances observed in the tail phase of COVID-19 are ill-defined. Understanding the long-term impact of COVID-19 according to routine laboratory biomarkers may support clinical decision making and help construct appropriate treatment strategies and rehabilitation approaches. This study evaluates the incidence of PASC and leverages a panel of 22 laboratory parameters to estimate the patient biochemical profile two months post-infection stratified to baseline clinical severity.
Methods
Study design and data sources
In this retrospective single-centre cohort study, we screened all adult patients admitted at the Zan Mitrev Clinic, Skopje, North Macedonia, between June 2020 and June 2021 with confirmed COVID-19, ICD-10-CM U07.1 and ICD-10-CM U07.2. The study design and cohort are presented in the Strengthening the Reporting of Observational Studies (STROBE) diagram (Figure 1).
1063 COVID-19 patients were hospitalised at the Zan Mitrev Clinic between June 2020 and June 2021.
The observed overall mortality rate was 24.3% (N = 258); of these mortality cases, N = 43 (17%) succumbed to their disease within 72 hours of admission, and N = 25 patients were discharged in that time frame following conservative treatment for mild symptoms.
COVID-19, coronavirus disease 2019; RT-PCR, reverse transcription-polymerase chain reaction test.
We excluded patients hospitalised for less than 72 hours and evaluated recovered patients successfully discharged following treatment who presented for follow-up assessment, including blood biochemistry evaluation.
The patient cohort was stratified according to the clinical presentation at admission in severity groups as per NIH guidelines16, namely mild, moderate, severe and critical.
We considered cases with follow up data until the 19th of August 2021; the last patient admitted during the country’s 3rd COVID-19 surge was on the 8th of June. Cases with laboratory data at admission and post-COVID (defined as laboratory data ≥ 30 days post-hospital admission) were considered eligible for the study. Clinical symptomatology was recorded at the most recent follow-up visit. We used the median value for subsequent comparative analysis for individuals with multiple laboratory analyses performed after ≥ 30 days.
Laboratory, clinical, and sociodemographic data were extracted from electronic health records. To account for local, demographic, genetic, nutritional and environmental factors17 we established reference values using laboratory results from self-reported healthy individuals undergoing routine annual check-ups at our clinic.
Statistical analysis
Patient clinical characteristics are presented as median + interquartile range (IQR) or mean ± standard deviation (SD) for continuous variables and absolute values with corresponding percentages for categorical variables.
Continuous variables were evaluated for normality using the Shapiro-Wilk test, and non-parametric tests were selected when at least one group followed a non-Gaussian distribution. The Chi-square or Fisher’s exact test were used to assess the association between categorical variables, clinical severity or PASC. Comparisons of admission versus follow-up data were performed using Wilcoxon matched-pairs signed-rank test. Kruskal-Wallis and Dunn’s multiple comparisons tests were used to assess differences across the three clinical severity groups at admission and follow-up.
Follow-up blood biochemistry values were compared to reference values using a Mann-Whitney test. Comparative results are reported according to their “effect size” (Median/Mean differences + 95% Confidence Interval, Odds ratio (OR), and Eta squared when appropriate).
Biochemistry Analysis
Biochemistry analyses were performed as previously described18. Diagnosis of SARS-CoV-2 infection was defined as a positive outcome to RT-PCR from nasal/oral swab.
Ethical considerations
The present study was approved by the local Ethics Committee of the Zan Mitrev Clinic. As an observational study, enrolled patients did not receive additional medical, pharmacological or behavioural interventions other than those administered in routine clinical practice, [#EBPZ.357, NCT04478539].
This research study was conducted in accordance with the principles laid out in the original Declaration of Helsinki. Written informed or verbal consent was obtained from all patients for publication of this manuscript under condition of full anonymity; the use of all health and medical information for scientific research and manuscript preparation was approved. For those patients that were not traceable, the need for informed consent was waived by the ethical committee.
Results
Patient demographics and clinical characteristics at admission
We hospitalised 1063 patients with confirmed COVID-19 disease at Zan Mitrev Clinic between June 2020 and June 2021. The study cohort details are presented in Figure 1. Most of our patients were from Skopje or referred from peripheral clinics across North Macedonia and Kosovo (Supplemental Figure 1A).
The study cohort (N = 168) – with a median age of 59 (50-68) years of which 75% were male (Table 1) – was stratified into subgroups according to National Institute of Health (NIH) clinical severity stratification and treatment guidelines16 into mild (N = 38, 23%), moderate (N = 29, 17%) and severe (N = 101, 60%). The clinical and demographic profiles of the study cohort and subgroups are shown in Table 1. Subgroups were similar in terms of age, gender proportion, and comorbidity distribution.
The peripheral oxygen saturation (SpO2%), cut-off < 94%, represents a key variable for severity stratification. Further translation of the NIH classification to clinical practice showed that moderate cases would either be admitted on- or required supplementary oxygen (flow rates of 2–15 L/min). Severe cases were characterised by a higher oxygen dependence, continuous positive airway pressure (CPAP) or the need for intubation and mechanical respiratory support during the clinical course.
The “severe” subgroup included one critical patient19 admitted with respiratory and multi-organ failure and discharged after 49 ICU days.
We detected pathological levels for several biomarkers at admission and observed that these biochemical disturbances tended to correlate with clinical severity. For instance, levels of Interleukin 6 (IL-6), C-Reactive Protein (CRP), D-dimers, LDH and Ferritin in severe cases were on average 1.5 to 2.3-fold higher compared to mild or moderate cases (Supplemental Table 1 and 2).
Clinical course and acute treatment
Respiratory distress is one of the hallmarks of COVID-19; viral propagation may lead to lung tissue injury, pneumonia, and impaired lung functioning. To that end, oxygen support forms a cornerstone of COVID-19 treatment; 90 (54%) patients were admitted on or required supplemental oxygen in the first 24 hours, median 4 L/min (2-10). Another 24 patients (15%) required non-invasive mechanical ventilation within the first 24 hours of admission – these patients were placed on CPAP at a median of 80% (60-100) fraction of inspired oxygen (Table 2). Nevertheless, we observed disease progression in 7 mild to- moderate and 12 severe cases, ultimately leading to ICU admission, intubation and mechanical respiratory support during the clinical course.
Moderate and severe cases received dexamethasone, typically a 10-day course of 6 mg once daily; critical cases or other patients showing signs of hyper inflammation were treated with intravenous dexamethasone at a dose of 8 mg q.d.
Secondary infections are recognised as a common complication of hospitalised COVID-19 patients. We recorded 58 (35%) cases of bacterial infections occurring a median of 13 (5-32) days after admission. The majority of bacterial infections were caused by Enterococcus (21%), Acinetobacter Baumannii (13%) and Escherichia Coli (11%) (Supplemental Figure 2). Fungal infections were also prevalent in our COVID-19 cohort; we observed 57 cases (34%) of secondary infections with Candida species occurring a median of 1.5 (0-12) days after admission.
Finally, total hospitalisation increased on average by 2 days according to the disease severity (Table 2); the median hospitalisation time for severe cases was 13 (10-21) days.
COVID-19 symptomatology and biomarker profile at follow-up
Patients were encouraged to attend follow-up laboratory analyses and clinical examinations at least once every 3 months; 20.9% (168 out of 805) of the patients complied with our recommendations resulting in a median follow-up time of 53 (30-105) days; follow-up compliance was similar across the sub-groups (Table 2). PASC was reported by 95 patients (56.5%), of which severe patients represented most cases experiencing long-term symptoms, 60.4% (61/101). We observed increasing cases of PASC associated with clinical severity, but the Chi-square test for trend was non-significant (Figure 2). Individuals experiencing PASC most commonly reported “Tiredness or fatigue”, “Difficulty breathing or shortness of breath”, and “Heart palpitations” as primary ailments.
Panel A shows pie charts indicating the distribution of reported PASC among the mild, moderate, and severe subgroups. (%) Numbers correspond to individuals who reported said PASC; colours indicate respective PASC.
Panel B depicts a dot plot graph visualising the distribution of % Number of individuals reporting PASC according to clinical disease severity.
Panel C shows the contingency table comparing the proportion of mild- to moderate cases vs severe cases with PASC; numbers in bar represent the patient count, % above the grey and red bar are the reported proportion of no PASC and PASC, respectively.
We next assessed biochemical profiles in each subgroup; we observed that laboratory abnormalities seen in mild and moderate COVID-19 patients during their hospital stay typically resolved within the follow-up period (Table 3) except for D-dimers that remained slightly elevated. In contrast, patients hospitalised with the severe disease showed sustained disturbances of several biomarkers at follow-up (Table 3, Figure 3).
Graphs show scatter plots of selected biomarkers (Table 3) stratified according to clinical severity and time. Subgroups at admission and follow-up are represented by grey and red, respectively; colour shading indicates clinical severity. The follow-up measurements were compared to reference values generated from a healthy control cohort.
Effect sizes were determined using the Mann-Whitney test, and the median difference (Δ) in comparison to the median reference value (horizontal dashed line) (Table 3) is shown above the subgroups. Significance levels between the follow-up subgroups and the reference values are shown; significance is visualized by asterisks * <0.05, ** <0.01, *** <0.001, and **** <0.0001. CRP, C-reactive protein; IL6, interleukin-6; LDH, lactate dehydrogenase, SII, systemic immune-inflammation index; NLR, neutrophil-to-lymphocyte ratio.
Notably, severe cases presented with elevated, Ferritin 432 ng/mL (170-916), D-dimer 600 ng/ml (283-1168) and to a lesser extent, LDH, 243 U/L (206-346), C-Reactive Protein (CRP) 5.7 mg/L (2.5-18.9) and IL-6, 5.1 pg/ml (1.7-28.9) (Table 3). Severe cases also exhibited disruptions in white blood cell indices at follow-up; the Neutrophil-to-Lymphocyte ratio (NLR) and Systemic Immune Inflammation index (SII) were elevated on average by 0.4 [95% CI 0.06 to 0.5] and 114 [95% CI 28.7-147] compared to the control population.
Collectively, these results show that more than half of hospitalised COVID-19 report long-term symptoms and everyday ailments. Moreover, severe COVID-19 cases also present with prolonged biochemical disturbances and disruption of immune homeostasis.
Discussion
COVID-19 disease caused by SARS-CoV-2 is an international public health emergency because of the significant mortality in the early inflammatory disease phase7. This study builds on recent observations of prolonged morbidity and frailty in the tail phase of the disease (> 30 days post-infection), termed Post-Acute Sequelae of COVID-19 (PASC)7.
In this observational study, we report the incidence of PASC by clinical disease severity in a previously hospitalised COVID-19 cohort in Skopje, North Macedonia. Secondly, we estimated the extent of post-discharge biochemical disturbances in each subgroup by assessing deviations from reference values generated from a healthy control cohort of non-COVID patients attending the outpatient clinic for their annual check-up.
Disease classification at admission represents a “snapshot”, an attractive starting point for comparative and association analyses. To that end, we observed notable differences in biochemical profile and PASC with increasing disease severity. However, COVID-19 is notorious for its unpredictable clinical course; some individuals experienced acute worsening and adverse outcomes following admission. Nonetheless, we also observed remarkable cases of clinical recovery from severe disease leading to independence from supplemental oxygen and hospital discharge; the latter group formed the basis for the study cohort and long-term observation.
Residual symptomatology of SARS-CoV-2 infection was observed in 56.5% of patients, comparable to the findings of previous studies 11 20-22. There were no major differences in PASC according to baseline severity – despite a trend, the correlation between PASC rates and severe disease did not reach statistical significance.
In contrast, follow-up laboratory data suggests a sustained disruption of biochemical, immunological and coagulation pathways most apparent in the severe subgroup. Particularly, Ferritin, D-dimers, and to an extent LDH, IL-6 and CRP remained abnormal in severe cases up to 38 (24–75) days after discharge (Table 3).
These observations hold true when compared to mild- to moderate COVID-19 cases but also a healthy control cohort; collectively, the results show that severe SARS-CoV-2 infection triggers prolonged disruption of host homeostasis.
We conclude that the pathological mechanisms driving severe disease in the viral symptom and early inflammatory phase (Supplemental Table 1 and 2) may persist long after discharge in select individuals. We observed a trend between COVID-19 coagulopathy and early clinical severity. Elevated levels of D-dimer and fibrin/fibrinogen degradation products were observed in our cohort during hospitalisation (Table 3) as reported by others23, with D-dimer levels selected to differentiate between severe and mild cases24 25. Interestingly, at follow-up, sustained abnormal D-dimer levels were observed in the presence of normalised fibrinogen levels, and platelet counts similar to previous findings26-28.
Ferritin was substantially increased at admission and follow-up, particularly in severe cases (Supplemental Table 1 and 2). This marker is an established independent prognostic factor for clinical severity in COVID-19 patients29 30, and recent literature31 suggests that unbound iron may play a role in the inflammation and hypercoagulation (= ↑ D-dimers) found in severe COVID-19.
Frequently observed comorbidities, such as hypertension (48%) and Type II Diabetes (18%), may also contribute to the inflammatory cycle32 and further disrupt interconnected coagulation pathways33.
Residual lung tissue injury and delayed repair34 might explain the, to an extent, sustained LDH levels at follow-up; recent work found that severe patients who required prolonged hospitalisation (≥ 17 days) were more prone to suffer from long-term tissue damage and persistent post-COVID-19 symptomatology35. Furthermore, severe cases were characterised by critically low levels of SpO2%, 88% (80-92), a need for respiratory support and on average 4 more hospitalisation days than mild cases. This early stage of (peripheral) hypoxia may induce chronic inflammation independent of viral antigens32. In addition, severe cases presented a more pronounced inflammatory profile than the mild and moderate subgroups (Supplemental Table 2). At follow-up, low-grade inflammation seemed persistent, indicated by comparatively elevated CRP and IL-6 levels (Table 3).
Our observations resonate with Liao et al.27, who reported elevated levels of IL-6 and other pro-inflammatory cytokines in mainly severe cases at weeks 4 and 6 post-discharge. Moreover, the authors established an association between sustained IL-6 levels and persistent pulmonary lesions.
In a comparable time window after discharge, Mandal and colleagues36 found that a proportion of patients experienced persistently elevated D-dimer and CRP serum concentrations.
These homeostatic disruptions may function synergistically in severe cases; hence, returning to homeostasis is slower, resulting in prolonged symptomatology known as “Long COVID”.
The incidence of bacterial co-infection (35%) is higher than in published studies37-39. Admittedly, our clinical team faced unprecedented challenges adhering to local antimicrobial stewardship protocols and international COVID-19 treatment guidelines. Empirical antibiotics were administered for a maximum of 48 hours or discontinued immediately following negative representative cultures40; however, unwarranted antibiotic prophylaxis is unfortunately still common clinical practice in the Western Balkans 41-44.
Moreover, the administration of the immunosuppressant dexamethasone, known to disrupt the innate and adaptive immune responses and increase susceptibility to invasive fungal diseases45-47, may explain the early and high incidence of fungal co-infection in our cohort. As reported previously, the proportion of fungal and bacterial co-infections was the highest in severe COVID-19 cases48 49.
Limitations
The study’s observational single-centre design prevents us from establishing a causative relationship between initial disease severity, presence of PASC, and degree of biochemical disturbances. Follow-up visits to our clinic were affected by ongoing lockdowns and public movement restrictions. As a result, we lacked exhaustive follow-up data on the whole “survivors cohort”. Consequently, certain analyses were underpowered due to the sample size and we were unable to connect the reported PASC rates with the observed biochemical profiles.
Nonetheless, this study presents time-series matched data acquired from the clinic’s Electronic Health Records and Laboratory Information System, excluding interlaboratory variations. Of importance, the clinical characteristics and biochemical profile of the study cohort (N = 168) were representative of the whole cohort (N = 1063) (Supplemental Table 3), suggesting that minimal impact on the results related to bias in the study design. Furthermore, comparison to a local reference values accounts for the many factors that affect biochemical pathways – hence, we were able to more accurately estimate deviations from normal values.
To the best of our knowledge, the current work is one of the first to report a matched admission vs follow-up biochemical profile according to NIH severity status guidelines, thus providing real-world insights into Long COVID-19 biochemical disturbances.
Conclusion
This study contributes to the emerging literature on the “Post-Acute Sequelae of COVID-19” phase of COVID-19. We show that whereas mild- and moderate cases recover biochemical homeostasis after discharge, severe cases are characterised by a higher incidence of PASC as well as sustained biochemical disturbances such as elevated D-dimers, Ferritin, LDH, CRP, IL-6 and immune indices NLR and SII; features commonly observed with low-grade inflammation in the elderly population, “inflammaging” 50-54. Our findings suggest that an initial evaluation and a time-series analysis of biochemical markers may aid in the diagnosis, risk stratification, construction of individualised rehabilitation approaches, clinical management, and a better understanding of the mechanism driving PASC.
Data Availability
The data that support the findings of this research study are available from the corresponding author, Rodney Rosalia, upon request.
Supplementary Figure Legends
Supplementary Figure 1 – Geographical location of study cohort
Panel A depicts the geographical location of our study cohort. The percent (%) above the location name represents the proportion of patients; increasing circle size corresponds to number magnitude.
Supplementary Figure 2 – Detected bacterial isolates in the study cohort
The pie chart depicts the distribution of bacterial isolates among the study cohort. (%) Numbers correspond to individuals in which said isolate was detected; colours indicate individual isolates.
Footnotes
Mariija Simjanoska: m.simjanoska{at}columbia.edu, Zan Mitrev: zan{at}zmc.mk, Gianluca Villa: gianluca.villa{at}unifi.it, Daniel O Griffin: dg2810{at}cumc.columbia.edu, Rodney Rosalia: rodney.rosalia{at}zmc.mk
References
