Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
We thank Rosenbaum et al1 for their correspondence on our article.2 In their reply, they largely report updated results of COVID-19 occurrence and subjective patient-reported COVID-19 severity from their patient survey of adult patients with spondyloarthritis and specifically comment that in their analysis, sulfasalazine did not ‘increase the risk to develop COVID-19’, nor did it increase the severity of those who were infected. Unfortunately, no details of the statistical methods, including any additional statistical adjustments (eg, age, sex, comorbidities and disease activity) or details of missing data, were included, and only a point estimate of risk was provided for each therapy without CIs, only p values, which therefore limits interpretation. While it is important to share reports of outcomes of COVID-19 among our populations of patients with rheumatic and musculoskeletal diseases (RMDs), it must be pointed out that the survey presented, as well as the subsequent analysis, is fundamentally different from our report on COVID-19-related mortality set within the Global Rheumatology Alliance (GRA) provider database and therefore the two outputs should not be compared, either directly or indirectly.
First, the GRA database is based on physician reports of COVID-19 among patients with known rheumatic disease and therefore can only be used to look at the associations between demographic and clinical features and COVID-19 outcomes among those known to have COVID-19. The data cannot be used to look at factors associated with acquisition of SARS-CoV-2 infection or development of COVID-19 disease, which is the focus of the report presented in the correspondence. These factors should not be compared since restricting the analysis to patients with an existing COVID-19 diagnosis may cause relationships between any variables that relate to COVID-19 acquisition to be distorted compared with a more general population, due to collider bias.3 Indeed, factors associated with acquisition of disease might not be found to be factors associated with outcome of disease, or vice versa.
Second, the main outcome in our report by Strangfeld et al is death due to COVID-19, not disease acquisition or severity otherwise. Fatal COVID-19 was an impossible outcome in the report by Rosenbaum et al, as only (subjective) patient-reported severity was collected. The association of factors with mortality and severity, although intuitively should be the same, may not be (particularly when death is an impossible outcome as in Rosenbaum et al, resulting in a selection bias that favours reporting by those with better outcomes). Moreover, the analysis by Rosenbaum et al does not take into account the biggest risk factor for COVID-19 disease, which is exposure to SARS-CoV-2 virus, and measures taken to avoid infection are not discussed nor the level of adherence to infection control and confinement measures, which may vary based on which treatment the patient is receiving. Furthermore, the number of patients who developed COVID-19 and were taking the various immunosuppressive/immunomodulatory medications is very low (ranging from 3 to 19 subjects, except for tumour necrosis factor inhibitors with 81 subjects), resulting in estimates that are neither robust nor generalisable.
Third, the reference medication in our report is methotrexate monotherapy, not ‘none’ as in the correspondence. When using ‘no disease-modifying antirheumatic drug (DMARD)’ as reference category in a sensitivity analysis for our report (data not shown), this change alone caused the effects of several DMARDs not to be significant, including sulfasalazine. This is not surprising as drug-free remission is infrequent in inflammatory rheumatic diseases, and patients not taking DMARDs often have contraindications to taking them, namely, severe comorbidities. In our data, patients not taking DMARDs were older than patients in any other medication category and had the highest rates of cancer, hypertension, cardiovascular and cerebrovascular disease. Patients on ‘no DMARD’ were therefore (expectedly) more likely to die from COVID-19, as shown in the results of our multivariable model (OR 2.11, 95% CI 1.48 to 3.01). Using no DMARD or even ‘no medication’ as reference category would therefore be inappropriate and would result in biased estimates and associations, driven by the underlying characteristics of this subgroup, namely, their severe comorbidity burden and older age: ‘everything else will look better’ when compared with a group that is inherently prone to have a more severe COVID-19 outcome.
These points aside, there are some factual inaccuracies about our report raised in the correspondence by Rosenbaum et al which we feel important to correct. First, the correspondence starts with the following sentence: ‘Strangfeld and colleagues recently reported that sulfasalazine usage was a risk factor for death from COVID-19 with a hazard ratio of 3.6, roughly comparable to the hazard ratio from the use of rituximab’. This sentence is incorrect. First, we do not report HRs; we report ORs, generated here using logistic regression, which compare the odds of something happening in one group with the odds of it happening in another. Conversely, the HR is a measure of how often a particular event happens in one group compared with how often it happens in another group, over time, and is typically calculated using Cox regression analysis (ie, in time-to-event or survival analysis). Our analysis was cross-sectional with no time element.
Second, we do not report ‘risk factors for death’. The use of the word ‘risk’ implies causation and should be avoided. This is true for the data by Rosenbaum et al presented in the correspondence as well. In our article, we report associations and caution explicitly against a causal interpretation of the associations described. Association should not be confused with causality and a direct causal link cannot be inferred; the association merely suggests a hypothesis but does not offer proof. In addition, we would like to point out again that in an observational study like ours (and the one by Rosenbaum et al), one can never completely rule out the possibility of bias due to the selection of patients or unmeasured confounding, for example, due to the particularities of health care systems and case reporting differences or confounding by indication. Indeed, while Rosenbaum and colleagues argue that spondyloarthritis is the most common rheumatic disease indication for sulfasalazine, in our global dataset (and many healthcare settings), this was simply not true. Over 50% of patients receiving sulfasalazine in our database did so for rheumatoid arthritis (RA). A very high proportion of these patients were from the UK, where sulfasalazine is a commonly used DMARD due to the guidelines for access to targeted therapies. It is also a therapy in RA used preferentially in patients with RA-associated interstitial lung disease, which is a likely confounder in the relationship between sulfasalazine exposure and fatal COVID-19.
So while we agree that reports of the impact of COVID-19 on populations of people with RMD are critical for helping us understand this infection further, we must take care to not conflate results looking at similar exposures but different outcomes, which will only add to the confusion for our colleagues and patients.
Handling editor Josef S Smolen
Twitter @carmona_loreto, @saskiaamber, @philipcrobinson, @pedrommcmachado
Collaborators No applicable.
Contributors KLH and PMM drafted the first version of the manuscript. All authors revised the manuscript and approved the final version.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests KLH reports she has received non-personal speaker’s fees from Abbvie and grant income from BMS, UCB and Pfizer, all unrelated to this manuscript, and is supported by the NIHR Manchester Biomedical Research Centre. MS has nothing to disclose. AS reports personal fees from lectures for AbbVie, MSD, Roche, BMS and Pfizer, all outside the submitted work. LC has not received fees or personal grants from any laboratory, but her institute works by contract for laboratories among other institutions, such as Abbvie Spain, Eisai, Gebro Pharma, Merck Sharp & Dohme España, S.A., Novartis Farmaceutica, Pfizer, Roche Farma, Sanofi Aventis, Astellas Pharma, Actelion Pharmaceuticals España, Grünenthal GmbH and UCB Pharma. LG reports personal consultant fees from AbbVie, Amgen, BMS, Biogen, Celgene, Gilead, Janssen, Lilly, Novartis, Pfizer, Samsung Bioepis, Sanofi-Aventis and UCB, and grants from Amgen, Lilly, Janssen, Pfizer, Sandoz, Sanofi and Galapagos, all unrelated to this manuscript. EFM reports that LPCDR received support for specific activities: grants from Abbvie, Novartis, Janssen-Cilag, Lilly Portugal, Sanofi, Grünenthal S.A., MSD, Celgene, Medac, Pharmakern, and GAfPA; grants and non-financial support from Pfizer; non-financial support from Grünenthal GmbH, outside the submitted work. SL-T has nothing to disclose. MG reports grants from National Institutes of Health, NIAMS, outside the submitted work. PCR reports personal fees from Abbvie, Eli Lilly, Gilead, Janssen, Novartis, Pfizer, Roche and UCB; non-financial support from BMS; research funding from Janssen, Novartis, Pfizer and UCB, all outside the submitted work. JY reports consulting fees from Astra Zeneca and Eli Lilly, and grants from Pfizer, outside the submitted work. PMM has received consulting/speaker’s fees from Abbvie, BMS, Celgene, Eli Lilly, Janssen, MSD, Novartis, Orphazyme, Pfizer, Roche and UCB, all unrelated to this manuscript, and is supported by the National Institute for Health Research, University College London Hospitals and Biomedical Research Centre.
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Provenance and peer review Commissioned; internally peer reviewed.