Objective. Individuals with psoriasis (PsO) or psoriatic arthritis (PsA) have an elevated risk of major adverse cardiac events (MACE), which include congestive heart failure (CHF), myocardial infarction (MI), and cerebrovascular accident (CVA). Biologic disease-modifying antirheumatic drugs (bDMARDs) may reduce cardiovascular risk; however, whether MACE risk differs by bDMARD class for this population is unknown. Methods. Usingdata from TriNetX database, we identified patients with PsO/PsA who were new bDMARD users, including tumor necrosis factor inhibitors (TNFi), interleukin (IL)-17A inhibitors (-i), IL-23i, or IL-12/23i. Time-dependent risk for MACE was calculated using weighted multinomial Cox proportional hazards regression with TNFi exposure as the referent. Additional analyses evaluated components of the primary outcome and baseline cardiovascular disease. A negative control outcome was used to assess bias. Results. We identified 32,758 patients with PsO/PsA who were new bDMARD users. Patients had PsO/PsA for a mean of 3.5 (SD 4.5) years prior to starting a biologic, the most common being TNFi (62.9%), followed by IL-17i (15.4%), IL-23i (11%), and IL-12/23i (10.7%). In weighted multinomial Cox proportional hazards regression, the adjusted risk of MACE was similar for IL-17Ai (adjusted hazard ratio [aHR] 0.98, 95% CI 0.73-1.32), IL-23i (aHR 0.84, 95% CI 0.54-1.31), and IL-12/23i (aHR 1.08, 95% CI 0.80-1.47) as compared to TNFi. Subset analyses supported the primary analysis. Negative control outcomes suggested adequate control of bias confounding. Conclusion. MACE risk does not significantly differ across bDMARD classes in patients with PsO/PsA. Therefore, cardiovascular risk should not guide biologic selection in this population.

OBJECTIVES:We aimed to evaluate the comparative efficacy of disease-modifying antirheumatic drugs (DMARDs) for patients with cardiac sarcoidosis. METHODS:We performed a retrospective cohort study of new users of methotrexate, mycophenolate or azathioprine for sarcoidosis using the US-based TriNetX electronic health records database from 2008 to 2023. Hazard ratios were calculated using inverse probability of treatment weighted Cox proportional hazards regressions to compare the efficacy of DMARDs with respect to delaying major adverse cardiac events among patients with cardiac sarcoidosis and preventing cardiac sarcoidosis from developing among patients with non-cardiac sarcoidosis. RESULTS:Among 3441 patients with sarcoidosis, 601 were defined as cardiac sarcoidosis and 2840 as non-cardiac sarcoidosis. The average age of the cohort was 52.1 years (standard deviation 11.9 years) and the majority were female (55.9%) and white (50.0%). Among patients with cardiac sarcoidosis at baseline, the risk of serious cardiac outcomes was similar for patients who initiated therapy with mycophenolate mofetil (HR 0.83, 95% CI 0.43-1.59) or azathioprine (HR 0.74, 95% CI 0.29-1.89) as compared with methotrexate. Among patients who did not have cardiac sarcoidosis at baseline, the risk of developing cardiac sarcoidosis was similar for patients who initiated therapy with mycophenolate mofetil (HR 1.11, 95% CI 0.46-2.66) and azathioprine (HR 0.54, 95% CI 0.15-1.91) as compared with methotrexate. Mycophenolate mofetil (HR 1.83, 95% CI 1.10-3.05) and azathioprine (HR 1.32, 95% CI 0.92-1.89) increased the risk of infection. CONCLUSION:A strategy of methotrexate had a favorable safety profile in terms of infection risk and may be favored over azathioprine or mycophenolate mofetil for patients with sarcoidosis or cardiac sarcoidosis.

Objective:To quantify the risk of ischemic vision loss among patients with polymyalgia rheumatica (PMR), giant cell arteritis (GCA), and matched population controls.Methods:Retrospective cohort study using the TriNetX database. Adults ≥50 years were included if they had ≥2 ICD-9/10-CM codes for GCA or PMR and a prednisone prescription within 30 days of their first diagnostic code. Matched population control cohorts were constructed using 1:3 nearest-neighbor matching. The primary outcome was ischemic vision loss, defined by ICD-9/10-CM codes for ischemic optic neuropathy and retinal artery occlusion. Multivariable Cox proportional cause-specific adjusted hazard ratios (aHR) and 95% confidence intervals (CI) were calculated.Results:Of 52,287 patients, 16,001 had PMR, 3,054 had GCA, and 33,232 were matched controls. The incidence of ischemic vision loss was highest for GCA (9.2/1,000 person-years) followed by PMR (1.7/1,000 person-years), GCA Controls (1.5/1,000 person-years), and PMR Controls (1.4/1,000 person-years). Ischemic vision loss was increased in GCA as compared to PMR (aHR 5.36, 95% CI 4.09–7.02) and as compared to GCA Controls (aHR 5.69, 95% CI 3.95–8.18). No significant differences were observed for PMR as compared to PMR Controls (aHR 1.10, 95% CI 0.86–1.41). Sensitivity analyses over the first 1 and 5 years of therapy supported the primary analyses.Conclusion:Isolated PMR was not associated with an increased risk in ischemic vision loss, in contrast to the substantially elevated risk in GCA. Our findings do not support routine screening for ischemic vision loss in patients with isolated PMR and instead support a symptom-driven approach to evaluation.

The fragility of randomized controlled trials (RCTs) of large vessel vasculitis (LVV) - defined as the minimum number of outcome events that would need to change to reverse the trial's conclusions - has not been comprehensively studied. We identified relevant RCTs with a systematic literature review till April 2025. The fragility index (FI)/ reverse fragility index (RFI) and fragility quotient (FQ, i.e., FI or RFI divided by number of trial participants) were calculated for primary or key secondary outcomes. Subgroup analyses were based on risk of bias (Cochrane Risk of Bias 2), drug (biologic or targeted synthetic agent versus other), LVV subtype, and time of publication (before/ after 2015). Eighteen RCTs (GCA, n = 14; TAK, n = 4) were analyzed. For trials with significant outcomes, FI ranged from 1 to 12 and FQ from 0.019 to 0.150; 5/9 (56 %) had FI ≤3, and 8/9 (89 %) had FQ ≤0.1. For trials with non-significant primary outcome, RFI ranged from 1 to 9 and FQ from 0.009 to 0.330; 8/12 (67 %) had RFI ≤5, 6/12 (50 %) had FQ ≤0.1, and 4/12 (33 %) had RFI less than the number lost to follow-up. The FI, RFI and FQ were similar for trials based on risk of bias, drug, LVV subtype, or time of publication. The results of most published LVV trials are fragile suggesting treatments are at risk of being misclassified as effective or ineffective. Larger trials with more robust and validated outcome measures or alternate designs should be considered in future LVV trials to improve confidence in their assessments of treatment effects.

Disclosure Form Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Observational research from large population databases may be affected by unmeasured confounding and time-related biases, such as immortal time bias. Modern causal inference practice applies propensity score-based methods, new-user designs, and other strategies to mitigate bias. The degree to which these methodologic approaches adequately address bias for any particular study may be difficult to measure. Recently, the incorporation of positive and negative controls has been identified as a means to assess for the impacts of residual confounding and/or time-related biases. The objective of this commentary is to describe the role of positive and negative controls in observational research. We offer recommendations for incorporating controls into critical appraisal and observational research projects.

BACKGROUND:Patients with polymyalgia rheumatica (PMR) may have subclinical large vessel vasculitis. We compared the incidence of aortic complications in PMR and giant cell arteritis (GCA) to the general population. METHODS:A retrospective cohort study was performed of patients with PMR and GCA identified by ≥2 ICD-9/ICD-10-CM diagnostic codes and concurrent corticosteroid treatment in the US-based TriNetX database (2000-2024). Matched general population controls were identified (1:3 ratio). The primary outcome, aortic complications, was a composite of aortic aneurysm and dissection/rupture. Adjusted hazard ratios (aHR) were calculated using Cox proportional cause-specific hazard models with PMR as the referent category. FINDINGS:Of 57,336 patients, 17,327 had PMR, 4,734 had GCA, and 35,275 were matched controls. Median follow-up time was 3.74 years (interquartile range, 1.8-6.4). The incidence rate of any aortic complication per 1,000 person-years was highest for GCA (11.69), followed by PMR (6.78) and the general population (5.09). Compared to patients with PMR, patients with GCA had a higher risk of any aortic complication (aHR 1.87, 95 % confidence interval (CI) 1.58-2.21); the general population risk was similar (aHR 0.95, 95 % CI 0.84-1.06). In a sensitivity analysis, patients with PMR who later developed GCA had a risk similar to those initially diagnosed with GCA (aHR 0.85, 95 % CI 0.60-1.19). INTERPRETATION:Patients with PMR had a similar risk of large vessel complications compared to the general population and a lower risk compared to those with GCA. These results do not support screening for aortic inflammation among patients with PMR who lack features of GCA.

ObjectiveWe aimed to evaluate the robustness of phase III randomised controlled trials (RCTs) for SLE and lupus nephritis (LN) using the fragility index (FI), the reverse FI (RFI) and the fragility quotient (FQ).MethodsWe searched for phase III RCTs that included patients with active SLE or LN. Data on primary endpoints, total participants and the number of events for each arm were obtained. We calculated the FI score for RCTs with statistically significant results (number of patients required to change from event to non-event to make the study lose statistical significance), the RFI for RCTs without statistically significant results (number of patients required to change from non-event to event to make study gain statistical significance) and the FQ score for both (FI or RFI score divided by the sample size).ResultsWe evaluated 20 RCTs (16 SLE, four LN). The mean FI/RFI score was 13.6 (SD 6.6). There were nine RCTs with statistically significant results (seven SLE, two LN), and the mean FI score was 10.2 (SD 6.2). The lowest FI was for the ILLUMINATE-2 trial (FI=2), and the highest FI was for the BLISS-52 trial (FI=17).Twelve studies had non-statistically significant results (10 SLE, two LN) with a mean RFI score of 15.6 (SD 6.1). The lowest RFI was for the ILLUMINATE-1 trial (RFI=4), and the highest RFI was for the TULIP-1 trial (RFI=27). The lowest FQ scores were found in the ILLUMINATE trials and the highest in the Rituximab trials (EXPLORER and LUNAR), meaning that the last ones were the most robust results after accounting for sample size.ConclusionsThe evidence of therapies for patients with SLE and LN is derived mostly from fragile RCTs. Clinicians and trialists must be aware of the fragility of these RCTs for clinical decision-making and designing trials for novel therapeutics.

Overdiagnosis occurs when patients are diagnosed with a disease that would otherwise never have affected the quality or duration of their lives. This often happens unintentionally through well-meaning screening programmes that aim to detect diseases during so-called subclinical stages. Recently, it has been suggested that patients with polymyalgia rheumatica should be screened for giant cell arteritis to identify those at higher risk of relapse or vascular complications. Screening for interstitial lung disease for patients with rheumatoid arthritis has also been recommended to identify patients who could benefit from pulmonary interventions. These potential benefits must be weighed against foreseeable harms. Such harms include the uncovering of incidental findings that necessitate additional medical follow-up, the financial costs associated with screening initiatives, the risk of overtreatment through increased immunosuppression in patients who might not have otherwise required it, and the psychosocial burden of a new diagnosis. Randomised clinical trials and prospective cohort studies of screening interventions should be conducted to establish the risks and benefits and identify patients most likely to benefit from them. This Viewpoint covers risks that overdiagnosis presents to the field of rheumatology, with focus on rheumatoid arthritis and polymyalgia rheumatica.