Autoimmune disease and COVID-19: a multicentre observational study in the United Kingdom

Objective. To establish the demographic characteristics, laboratory ﬁndings and clinical outcomes in patients with autoimmune disease (AD) compared with a propensity-matched cohort of patients without AD admitted with COVID-19 to hospitals in the UK. Methods. This is a multicentre observational study across 26 NHS Trusts. Data were collected both retrospectively and prospectively using a predesigned standardized case record form. Adult patients ( (cid:2) 18years) admitted between 1 April 2020 and 31 July 2020 were included. Results. Overall, 6288 patients were included to the study. Of these, 394 patients had AD prior to admission with COVID-19. Of 394 patients, 80 patients with SLE, RA or aPL syndrome were classiﬁed as severe rheumatologic AD. A higher proportion of those with AD had anaemia [240 (60.91%) vs 206 (52.28%), P ¼ 0.015], elevated LDH [150 (38.08%) vs 43 (10.92%), P < 0.001] and raised creatinine [122 (30.96%) vs 86 (21.83%), P ¼ 0.01], respectively. A signiﬁcantly higher proportion of patients with severe rheumatologic AD had elevated CRP [77 (96.25%) vs 70 (87.5%), P ¼ 0.044] and LDH [20 (25%) vs 6 (7.5%), P ¼ 0.021]. Patients with severe rheumatologic AD had signiﬁcantly higher mortality [32/80 (40%)] compared with propensity matched cohort of patients without AD [20/80 (25%), P ¼ 0.043]. However, there was no difference in 180-day mortality between propensity-matched cohorts of patients with or without AD in general ( P ¼ 0.47). Conclusions. Patients with severe rheumatologic AD had signiﬁcantly higher mortality. Anaemia, renal impairment and elevated LDH were more frequent in patients with any AD while elevated CRP and LDH were more frequent in patients with severe rheumatologic AD both of which have been shown to associate with increased mortality in patients with COVID-19


Introduction
Coronavirus disease 2019 (COVID-19) is a global pandemic leading to an unprecedented health crisis. The World Health Organization (WHO) declared the novel coronavirus outbreak to be a pandemic in March 2020. Although the number of patients with severe infection is gradually decreasing in some countries due to mass vaccination, it remains a global threat.
COVID-19 is associated with increased risk for thrombosis in addition to causing respiratory failure with or without multi-organ failure and death. Some studies found that patients with autoimmune and inflammatory conditions are at increased risk for COVID-19-associated hospitalizations and worse disease outcomes [1]. However, autoimmune diseases (ADs) are a broad category of diseases with differing severity, from requiring no treatment to multiple immunosuppressive treatments. It is likely that the clinical course and the outcomes of COVID-19 vary in patients with AD depending on the severity of the AD and the immunosuppressive treatment. There are >80 autoimmune conditions affecting >4 million people in the UK. ADs such as rheumatoid arthritis (RA), Systemic lupus erythematosus (SLE) and antiphospholipid syndrome (APS) are generally considered to be severe rheumatologic ADs associated with a higher risk of developing thrombosis in addition to their other complications [2]. In a propensity score-matched analysis from a nationwide multicentric research network study assessing the short-term outcome of COVID-19 patients with SLE, the mortality was comparable to that of the general population, but SLE patients had higher risks of hospitalization, admission to an intensive care unit (ICU), mechanical ventilation, stroke, venous thromboembolism (VTE) and sepsis [3].
Additionally, many studies have demonstrated a frequent occurrence of autoantibodies, including aPL, in patients with COVID-19 [4]. The prevalence of aPL was even higher in patients with severe disease, but there was no association between aPL positivity and disease outcomes including thrombosis, invasive ventilation and mortality. As transiently positive aPL is a well-known phenomenon in patients with acute infection, the significance of these antibodies remains to be determined [5], although some studies have demonstrated aPL from patients with COVID-19 caused thrombosis in a mouse model [6].
The aim of this study was to establish the demographic characteristics, laboratory findings and clinical outcomes in patients with AD compared with a propensity-matched cohort of patients with no AD admitted with COVID-19 to hospitals in the UK.

Methods
This study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology statement.
The study was approved by the Human Research Authority (HRA) and Health and Care Research Wales (HCRW) and the local Caldicott Guardian at Scotland (reference 20/HRA/1785).
We included adult patients (!18 years) admitted to hospital during the first wave of the COVID-19 pandemic in the UK between 1 April 2020 and 31 July 2020. This article includes only patients with AD diagnosed prior to admission to a hospital with COVID-19 and an equal size propensity-matched cohort of patients with no AD with COVID-19 admitted to a hospital during the first wave of the COVID-19 pandemic (1 March-31 May 2020). All patients had severe acute respiratory syndrome coronavirus 2 confirmed by realtime PCR on nasopharyngeal swabs or lower respiratory tract aspirates.

Data collection
Data were collected both retrospectively and prospectively using a predesigned standardized case record form (CRF) and entered in a central electronic database [Coagulopathy Associated with COVID-19 (CA-COVID-19)] (REDcap version 10.0.10; Vanderbilt University, Nashville, TN, USA) hosted by Imperial College London. At the time of writing the paper, all outcomes had been completed and no patient remained in hospital. As the data were collected by clinicians directly involved in patient care with no breach of privacy or anonymity by allocating a unique study number with no direct patientidentifiable data, consent was waived by the HRA. Baseline patient demographics, comorbidities, haematological and biochemical blood results on the day of admission and clinical outcomes until the day of discharge/death were collected. At the time of writing this paper, all patients had completed follow-up until day 180 post-hospital admission or death.

Thrombosis and bleeding complications
Thrombosis and bleeding complications were identified on clinically indicated computerized tomography (CT) scan or ultrasound scan (US) imaging. Thrombotic events were defined as image-confirmed pulmonary embolism (PE), deep vein thrombosis (DVT) or arterial thrombosis. Bleeding events were defined as major or clinically relevant minor haemorrhages according to International Society on Thrombosis and Haemostasis classification [7] (Supplementary Table S1, available at Rheumatology online).

Multi-organ failure
Multi-organ failure was defined as failure in two or more organ systems that required interventions to maintain homeostasis.
Admission to an ICU This was defined as patients who required continuous positive airway pressure ventilation (CPAP) or mechanical ventilation with or without extracorporeal membrane oxygenation or required other organ support.

Statistical analysis
Propensity score matching was performed using the nearest neighbours method, with a desired ratio of 1:1 between patients with and without AD. Covariates (demographics and comorbidities) used for propensity score matching are summarized in Supplementary Fig.  S1, available at Rheumatology online. Laboratory results at presentation were not included in the propensity matching. Factors for propensity matching were chosen based on factors found to contribute to increased mortality in published studies of patients with COVID-19. Propensity matching was performed for patients with any AD and for patients with severe rheumatologic AD separately. The characteristics of the treated and untreated patients were summarized and compared using descriptive statistics. The probability of survival between patients with and without AD were assessed using Kaplan-Meier curves. Characteristics of patients who had AD were compared with patients who did not have AD using the chi-squared or chi-squared trend test. Propensity score matching and analysis were performed using R (R Foundation for Statistical Computing, Vienna, Austria). Two-tailed P-values <0.05 were considered statistically significant.

Results
Overall, 6288 patients with COVID-19 were admitted to 26 NHS Trusts in the UK between 1 April and 31 July 2020. Of these patients, we analysed 394 classified as having AD prior to admission with COVID-19. The patients with AD group included those with chronic inflammatory arthritis, including RA, PsA and SpA; CTD, including SLE, SS, SSc and PMR; vasculitides and APS (Supplementary Table S2, available at Rheumatology online). Of 394 patients, 80 had SLE, RA or APS and were classified as having severe rheumatologic AD (Fig. 1). These patients are more likely to require immunosuppressive medication and be associated with an increased risk of thrombosis, which may cause severe complications when they develop COVID-19.
All ADs compared with non-ADs prior to propensity matching There was no age difference between patients with and without AD; the median age of patients with AD was 71 years (IQR 61-82) compared with 74 years (IQR 59-83) in patients without AD (P ¼ 0.78). As expected, the majority of AD patients were female [229/394 (58.12%) vs 165/394 (41.88%); P < 0.001], although the majority of the patients admitted to hospitals with COVID-19 were male [3279/5894 (55.6%) male vs 2615/5894 (44.4%) female; P < 0.001]. There were no differences in BMI, ethnicity or comorbidities between patients with and without AD. The majority of patients with AD had below normal haemoglobin at the time of admission to hospital [240/394 (60.91%) vs 2895/5894 (49.12%); P < 0.001]. A higher proportion of patients with AD had elevated creatinine levels while a lower proportion had elevated prothrombin time (PT) compared with those without AD [creatinine above normal: 122/394 (30.96%) vs 1565/5894 (26.56%), P ¼ 0.03; PT above normal: 4330/5894 (73.46%) vs 263/394 (66.75%), P ¼ 0.004]. There were no differences in the other laboratory parameters, notably lactate dehydrogenase (LDH), CRP and D-dimer levels, between patients with and without AD at the time of admission to hospital with COVID-19. Patients' characteristics, comorbidities and laboratory parameters at admission are summarized in Table 1.
All AD patients compared with non-AD patients after propensity matching As expected, there were no differences in the demographics and comorbidities of the patients with and without AD after propensity matching (Table 1). However, even after propensity matching, a higher proportion of patients with any AD had low haemoglobin compared with patients without AD [240 (60.91%) vs 206 (52.28%); P ¼ 0.015]. Furthermore, a higher proportion of patients with AD had elevated LDH and creatine levels [LDH in 150 (38.08%) vs 43 (10.92%), P < 0.001; creatinine in 122 (30.96%) vs 86 (21.83%), P ¼ 0.01]. There were no differences in the other laboratory parameters between the two groups ( Table 1).

Patients with severe rheumatologic AD
Comparisons were made between the 80 patients classified as severe rheumatologic AD with a 1:1 propensity matched cohort of patients without AD. As expected, no differences were seen in patient demographics and comorbidities between the two groups following the propensity matching. In patients with severe rheumatologic AD, the female preponderance was higher than in the all-AD group [55/80 (68.75%) female vs 25/80 (31.25%) male] (Table 2). Furthermore, a significantly higher proportion of patients with severe rheumatologic AD had elevated CRP and LDH levels compared with patients without AD [CRP in 77 (96.25%) vs 70 (87.5%), P ¼ 0.044; LDH in 20 (25%) vs 6 (7.5%), P ¼ 0.021]. There were no differences in the other laboratory parameters between the two groups ( Table 2).

Outcomes in patients with any AD compared with non-AD after propensity matching
For the primary outcome, there was no difference in the 180 day mortality between the propensity-matched cohort of all patients with and without AD. The overall mortality in patients with any AD was 121/304 (30.71%) compared with 111/394 (28.17%) in patients with no AD (P ¼ 0.435) ( Fig. 2A).
For the secondary outcomes, there were no differences observed in the rate of thrombosis, major bleeding, the development of MOF or admission to an ICU in patients with any AD compared with those with no AD. There was a trend towards more patients with AD supported with CPAP [29/393 (7.36%) vs 17/394 (4.31%); P ¼ 0.068] ( Table 3).
Outcomes in patients with severe rheumatologic AD compared with non-AD after propensity matching For the primary outcome, in contrast to patients with any AD, those with severe rheumatologic AD had significantly higher mortality [all-cause mortality; 32/80 (40%)] compared with the propensity-matched cohort of

Clinical interventions
There were no differences in the clinical interventions during the hospital admission in patients with or without AD as a whole group or with severe rheumatologic AD, except a significantly higher proportion of patients with any AD or severe rheumatologic AD received steroids compared with patients with no AD [82/394 (20.81%) vs 40/394 (10.15%), P < 0.001 and 18/80 (22.5%) vs 5/80 (6.25%), P ¼ 0.003, respectively] ( Table 3 for any AD and  Table 4 for severe rheumatologic AD).

Discussion
In this large multicentre observational study across the UK assessing the clinical characteristics and outcomes of patients with any AD and those with severe rheumatologic AD, we found that the presence of any AD did not increase the risk of mortality or other outcomes (thrombosis, major bleeding, MOF or admission to an ICU) compared with the propensity-matched cohort of patients with no AD. However, patients classified as severe rheumatologic AD (SLE, RA or APS) had significantly higher mortality compared with patients with no AD. No differences were seen in the secondary outcomes between the two groups. Following propensity matching for demographics and comorbidities, a higher proportion of patients with AD had low haemoglobin and elevated LDH   In those with severe rheumatologic AD, elevated CRP and LDH were more common compared with patients without AD. Generally, ADs are more common in women, occurring at a ratio of 2:1 [8], whereas the COVID-19 disease severity and admission rate are higher in men [9]. These differences were preserved in this study.
ADs are a heterogeneous group of conditions typified by dysregulation of the immune system. Most of the patients with AD received or were receiving immunosuppressive medications, which make them more susceptible to infections and complications. Observational studies assessing the risk of acquiring COVID-19 and outcomes in patients with AD reported conflicting results. A cross-sectional study in northeast Italy reported that patients with AD had a similar rate of COVID-19 compared with the general population [10]. Another Italian study also found that the presence of AD did not increase the risk of COVID-19 [11]. Furthermore, they suggested that the outcome of patients with AD did not differ from patients with no AD [11]. However, this study did not perform propensity matching for the study groups, which, as shown in this study, are significantly different in important respects. In contrast, the results of a multicentre retrospective study from China showed that patients with AD might be more susceptible to COVID-19 compared those without [12]. Additionally, a Spanish study that assessed the association between the outcome and the potential prognostic variables, adjusted by COVID-19 treatment in patients with AD compared with a matched cohort (for sex and age and blinded to outcome or other variables but not propensity matching for all comorbidities) of patients with no AD, reported that hospitalized patients with AD have a more severe course [13]. In the current propensity-matched study, we did not observe a difference in the mortality or secondary outcomes between patients with any AD compared with patients with no AD (Table 3). This could be due a higher proportion of patients with any AD being given steroids, which has been shown to improve mortality in patients with COVID-19 [14]. However, the mortality rate was still significantly higher in patients with severe rheumatologic AD despite a higher proportion receiving steroids. Additionally, there was a trend towards higher mortality in patients classified as severe rheumatologic AD compared with patients with other ADs (P ¼ 0.056). The higher mortality in patients with severe rheumatologic AD could indicate that these patients suffer more severe rheumatologic COVID-19, although no differences were seen in the secondary outcomes, including the rate of thrombosis, major bleeding, development of MOF or admission to an ICU. Therefore the cause for increased mortality in patients with severe rheumatologic AD was not clear. It is possible that prior non-steroidal immunosuppressive drugs contributed to the increased mortality in these patients (Supplementary  Table S3, available at Rheumatology online).
Anaemia is a frequent complication in patients with AD. It is generally classified as anaemia of chronic disease and is usually multifactorial. Despite propensity matching for demographics and comorbidities, a higher proportion of patients with any autoimmune disease had anaemia on admission to hospital. However, a significantly higher proportion of patients with AD had elevated LDH, which could be due to ongoing tissue damage associated with AD and in some cases autoimmune haemolytic anaemia. Elevated CRP, a marker of disease severity in many ADs, was observed in a significantly higher proportion of patients with severe rheumatologic AD upon hospital admission compared with patients without AD. Both elevated CRP and LDH on admission are considered predictors of increased mortality in patients with COVID-19 [15,16] and indeed these patients had a higher mortality rate compared with the control group in this study. Finally, serum creatinine was elevated on admission in a higher proportion of patients with AD. Renal failure is a frequent complication in these individuals and may additionally contribute to anaemia.
The main limitation of this study is that some of the data were collected retrospectively, but relevant information and clinical outcomes were recorded directly using a predefined, well-structured electronic CRF. However, this did not include DASs such as SLEDAI or the 28-joint DAS. The classification of RA, SLE and APS as severe rheumatologic AD compared with other ADs in the study may be regarded as arbitrary. It is possible that disease severity of any given AD at the time of admission with COVID-19 has an impact on primary or secondary outcomes beyond the primary AD diagnosis and immunosuppressive medications. As the disease severity scores were not included in the data collection, we were unable to assess the impact of the individual disease severity in the clinical outcomes in this study. Although the number of patients included in the study is relatively small, it comprises patients admitted to 26 NHS Trusts across the UK, providing a representative view of AD in the UK.
In conclusion, we found no differences in the clinical outcomes in patients with any AD compared with patients with no AD admitted to hospitals with COVID-19 from the first wave of the pandemic. However, those with severe rheumatologic AD had significantly higher mortality. Anaemia, renal impairment and elevated LDH were more frequent in patients with any AD, while elevated CRP and LDH were more frequent in patients with severe rheumatologic AD. Although vaccination has reduced the risk of mortality associated with COVID-19,  Author contribution: D.J.A. conceptualized the study, acquired the funding and was responsible for the methodology, project administration, validation, visualization, writing the original draft, reviewing and editing the study and data curation. I.R. was responsible for the formal analysis, software, valuation of the study and reviewing and editing the manuscript. C.P. contributed to data interpretation and reviewing and editing of the manuscript. P.N. was responsible for project administration, data collection and reviewing the manuscript. M.M. was responsible for data curation, project administration, resources, validation of the study and reviewing and editing the manuscript. M.L. was responsible for data interpretation and reviewing and editing the manuscript. All other authors reviewed and approved the final version of the manuscript.
Funding: Bayer supported the study by providing the investigator-initiated funding (P87339) to set up the multicentre database of the study. The funder had no access to data and played no part in the analysis or writing. The corresponding author is responsible for the study design, had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Disclosure statement: D.J.A. received funding from Bayer to set up the multicentre database of the study as investigator-initiated funding and received a research grant from Leo Pharma. M.L. has received consultation and speaker fees from AstraZeneca, Sobi, Leo Pharma, Takeda and Pfizer. P.N. has received research grants from Novartis, Principia and Rigel; unrestricted grants from Sanofi, Chugai and Octapharma and honoraria from Bayer. All other authors have declared no conflicts of interest.

Data availability statement
The data underlying this article will be shared upon reasonable request to the corresponding author.