Predictors of hospitalization in patients with rheumatic disease and COVID-19 in Ireland: data from the COVID-19 global rheumatology alliance registry

Abstract Objectives Given the limited data regarding the risk of hospitalization in patients with rheumatic disease and coronavirus disease 2019 (COVID-19) in Ireland, we used the COVID-19 Global Rheumatology Alliance (GRA) registry data to study outcomes and their predictors. The primary objective was to explore potential predictors of hospitalization. Methods We examined data on patients and their disease-related characteristics entered in the COVID-19 GRA provider registry from Ireland (from 24 March 2020 to 31 August 2020). Multivariable logistic regression was used to assess the association of demographic and clinical characteristics with hospitalization. Results Of 105 patients, 47 (45.6%) were hospitalized and 10 (9.5%) died. Multivariable logistic regression analysis showed that age [odds ratio (OR) = 1.06, 95% CI 1.01, 1.10], number of co-morbidities (OR = 1.93, 95% CI 1.11, 3.35) and glucocorticoid use (OR = 15.01, 95% CI 1.77, 127.16) were significantly associated with hospitalization. A diagnosis of inflammatory arthritis was associated with lower odds of hospitalization (OR = 0.09, 95% CI 0.02, 0.32). Conclusion Increasing age, co-morbidity burden and glucocorticoid use were associated with hospitalization, whereas a diagnosis of inflammatory arthritis was associated with lower odds of hospitalization.


Introduction
Infection with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and the subsequent coronavirus disease 2019 (COVID-19) has caused considerable morbidity and mortality worldwide. Patients with rheumatic diseases are generally at increased risk of infection owing to both their disease with the associated co-morbidities and their immunomodulatory or suppressive treatments [1]. Whether this applies to SARS-CoV-2 infection and the magnitude of any such risk remains unclear.
The COVID-19 Global Rheumatology Alliance (C19-GRA) aims to address the knowledge gap about COVID-19 for patients with rheumatic disease [2]. The C19-GRA has reported data on the first 600 cases in its physicianreported case registry [3]. In these patients, glucocorticoid doses of prednisolone equivalent 10 mg/day were associated with increased risk of hospitalization, whereas TNF-a inhibitors were associated with a decreased risk of hospitalization [3]. Overall, as with the general population, the greatest risk of negative outcomes was associated with increasing age and comorbidities.
Geographical variations in COVID-19 outcomes have been reported [4][5][6][7][8]. Our aim was to report the outcomes and their determinants in patients in Ireland with rheumatic disease who developed COVID-19.

COVID-19 Global Rheumatology Alliance
Data regarding individuals with rheumatic disease who develop COVID-19 are entered into one of two parallel international data portals hosted in the USA and the UK. Details of the C19-GRA registries have been published previously [2,3,9].

Data collection
Patients in this study were entered into the C19-GRA provider registry from 24 March 2020 to 31 August 2020.
Data collected on baseline rheumatic disease status, including demographic and clinical variables, were as follows: age, sex, smoking status, rheumatic disease diagnosis, disease activity and co-morbidities. Medications were categorized as previously described (Supplementary Data S1, available at Rheumatology Advances in Practice online) [3]. Rheumatic diseases were categorized as inflammatory arthritis, gout, vasculitis, CTDs and others (Supplementary Data S2, available at Rheumatology Advances in Practice online). Data collected regarding COVID-19 infection included the method of diagnosis, place of diagnosis, COVID-19 symptoms and the outcomes of COVID-19 disease, including hospitalization, ventilation and death.
Continuous variables were reported as the median [interquartile range (IQR)]. Categorical variables were reported as the number and percentage. In univariable analyses, differences according to hospitalization status and mortality were compared using v 2 tests or Fisher's exact tests, as appropriate, for categorical variables and Mann-Whitney U-tests for continuous variables. The independent associations between demographic and disease-specific features with the odds of COVID-19 hospitalization were estimated using univariable followed by multivariable-adjusted logistic regression and reported as odds ratios (ORs) and 95% CIs. We constructed two multivariable models. The first model included all variables with significant effect sizes (P < 0.05) in the univariable logistic regression analyses, whereas the second model was the most parsimonious model including age, biologic sex and those variables retaining statistical significance. We did not proceed to logistic regression analyses with mortality as the outcome owing to the small number of deaths.
This study was approved by the Irish National Research Ethics Committee for COVID-19 (20-NREC-COV-010). The committee waived the need for written informed patient consent because the data were fully anonymized.

Discussion
This is the largest report of people with rheumatic disease and COVID-19 from Ireland. We identified factors associated with higher odds of hospitalization, including increasing age, number of co-morbidities, a diagnosis of gout and baseline glucocorticoid use. A diagnosis of inflammatory arthritis showed a protective association with hospitalization.
Our findings of the association of increasing age and glucocorticoid use with hospitalization are consistent with the findings from C19-GRA [3]. This highlights that glucocorticoid use should be limited in favour of longerterm disease-modifying therapies. The C19-GRA reported bDMARDs as being associated with a reduced risk of hospitalization, which has some biologic plausibility owing to the potential for reduction of the hyperinflammatory state seen in severe COVID-19 [3,10,11]. Several bDMARDs have been proposed as treatment for severe COVID-19, but results of many controlled trials have been disappointing [12,13]. The RECOVERY and REMAP-CAP trials have demonstrated evidence of modest benefit from the use of the IL-6 receptor antagonists tocilizumab and sarilumab in COVID-19 [14,15]. In our population, the apparent benefit of both csDMARDs and bDMARDs in the univariable analysis disappeared after controlling for the underlying diagnostic group. These agents, and particularly bDMARDs, are more likely to be used in patients with a diagnosis of inflammatory arthritis, and the reduced risk of hospitalization with these agents might represent confounding by indication. The results of ongoing randomized controlled trials of other bDMARDs in COVID-19 will provide clarity.
Our study has several limitations. Although the low number of cases of people with rheumatic diseases and COVID-19 in Ireland entered in the C-19 GRA registry limits the power of our study, this study explores associations between patient and disease characteristics and outcomes in Ireland. Potential overfitting of multivariable models can occur with small sample sizes and is one reason we did not proceed to multivariable analysis for mortality. Once further cases are entered into the GRA registries, we will be able to conduct more detailed analyses in the future. As a physician-entered registry, the C19-GRA is limited by selection bias, with more severe cases likely to be entered, which is likely to explain why 100% of gout cases were hospitalized. Given that there is no denominator population for these data, no inferences can be drawn about the incidence of COVID-19 in patients with rheumatic diseases. The C19-GRA is restricted to patients who have both rheumatic diseases and COVID-19; therefore, no comparisons can be made between this group and the general rheumatic disease population or with patients without rheumatic disease who develop COVID-19. Our finding of gout as a risk factor for hospitalization, although highly significant because all gout cases were hospitalized, might reflect selection bias and requires future evaluation. Gout is usually managed in primary care in Ireland, and given that the C19-GRA is a rheumatologist-entered registry these cases could be less likely to be entered. Patients with rheumatic diseases might have behaved differently with regard to risk behaviour during COVID-19. In Ireland, a strict national lockdown was in place for a large part of the study period (including when most  cases were reported). Individuals judged to be at high risk at the start of the pandemic, including many rheumatic diseases, were specifically advised to stay at home as much as possible and to avoid social interactions [16].
In conclusion, our findings confirm that increasing age, co-morbidity burden and glucocorticoid use are associated with hospitalization with COVID-19 in patients with rheumatic diseases. An underlying diagnosis of inflammatory arthritis appears to be associated with lower risk of hospitalization. The previously reported beneficial effects of bDMARDs might be mediated by underlying diagnosis rather than a specific effect. R.C., C.L., K.L., J.G.R., R.K., A.D.F., J.J.C., P.O.C., R.M.F., R.H.M., D.J.K. and G.M.M. contributed to data collection, data quality control, data analysis and interpretation. They drafted, and revised, the manuscript critically for important intellectual content and gave final approval of the version to be published. E.N. and C.A.D. contributed to the analysis and interpretation of the data. They drafted, and revised, the manuscript critically for important intellectual content and gave final approval of the version to be published. P.C.R., J.W.L. and R.G. directed the work and contributed to the analysis and interpretation of the data. They drafted, and revised, the manuscript critically for important intellectual content and gave final approval of the version to be published.
Funding: No specific funding was received from any funding bodies in the public, commercial or not-for-profit sectors to carry out the work described in this manuscript.