Background/Purpose: Serious infections (SI) are a major concern in patients treated with tumour necrosis factor inhibitors (TNFi). The RABBIT Risk Score (RRS) (1) allows a calculation of the expected number of SI over a one year period of TNFi treatment according to patient characteristics. This score was developed in a cohort of German patients with rheumatoid arthritis (RA) enrolled in the RABBIT biologics register and then validated in a second cohort from the same register. The aim of the present study is to assess the reliability of the RRS in two cohorts of patients with RA treated with TNFi from the British Society for Rheumatology Biologics Register (BSRBR)-RA.

Methods: The BSRBR-RA is an ongoing national prospective cohort study of subjects with RA starting biologic therapy. Patients were recruited to a first TNFi cohort (old) between 2001 and 2008 and to a second TNFi cohort (new) from 2010 to date. Risk factors included in the RRS include age, Health Assessment Questionnaire (HAQ) score, chronic lung disease, chronic renal disease, previous serious infection, number of treatment failures, mean daily glucocorticoid dose and treatment with TNFi. As the BSRBR-RA does not capture the dose of oral steroids or intramuscular injections (common in the UK), we applied a modified version of the score to both cohorts assuming that all patients receiving steroids would take between 7.5-14 mg of prednisolone per day. We calculated the expected number of patients with at least one SI in the first year of therapy and compared to the observed numbers in each cohort. To evaluate the predictive performance of the RRS we calculated the area under the curve (AUC).

Results: 13,121 patients were recruited to the old cohort and 1,475 patients were recruited to the new cohort. There were significant differences between cohorts (Table). Patients from the old cohort had a longer disease duration, higher DAS28 and HAQ score, and more likely to be receiving steroids compared to newer patients. The crude incidence rate of SI was lower in the new cohort (29.3 new vs 62.3 old/1000 patient-years) as was the expected and observed number of patients with ≥1 SI during the first year of therapy. The predictive performance of the RRS was better for the new cohort (AUC=0.82) than for the old cohort (AUC=0.62).

	Old cohort (n=13,121)	New cohort (n=1,475)	p
Age (years), mean (SD)	56.1 (12.3)	56.5 (12.6)	0.128
Gender: n (%) female	10,010 (76.3)	1,116 (75.7)	0.591
RA disease duration (years), Median (IQR)	11 (6 – 19)	6 (2 – 14)	0.0001
DAS28, mean (SD)	6.5 (1.0)	5.9 (1.0)	0.0001
HAQ score, mean (SD)	2.0 (0.6)	1.6 (0.7)	0.0001
RF +, n (%)	8,493 (64.8)	821 (55.6)	<0.0001
Steroid use (%)	5,867 (44.7)	358 (24.3)	<0.0001
Smoking			0.065
Current, n (%)	2,861 (22.0)	279 (20.2)
Ex-smoker, n (%)	4,922 (37.8)	505 (36.5)
Never smoked, n (%)	5,238 (40.2)	600 (43.4)
More than 5 previous DMARDs, n (%)	325 (2.5)	4 (0.3)	<0.0001
Diabetes, n (%)	750 (5.8)	115 (7.9)	0.001
Chronic lung disease, n (%)	2,060 (15.9)	252 (18.5)	0.012
Chronic renal disease, n (%)	330 (2.5)	28 (1.9)	0.149
Crude incidence rate of serious infections in the first year per 1,000 patient-years (95% Confidece Interval)	63.3 (59.0 – 67.9)	29.3 (20.8 – 41.3)
Number of observed first serious infections in the first year	771	33
Number of expected first serious infections in the first year	496	27
Area under the curve (95% Confidence Interval)	0.62 (0.60 – 0.64)	0.82 (0.73 – 0.91)

Conclusion: TNFi are being used earlier in RA patients with lesser disease activity, steroid use and  disability. These patients are also experiencing lower SI rates. Among these cohorts of UK patients, the RRS underestimated the total number of SI, although some misclassification of steroid dose and differential access to TNFi resulting in other unmeasured patient differences may have contributed.  The predictive power improved in a more recent cohort which strengthens the utility of this score in current routine clinical practice.

Reference: (1) Zink A et al. Evaluation of the RABBIT risk score for serious infections. Ann Rheum Dis 2013 Jun 28 (doi:10.1136/annrheumdis-2013-203341)

---

« Back to 2014 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/evaluation-of-the-rabbit-risk-score-for-serious-infections-in-a-uk-anti-tnf-treatment-cohort/