1%), followed by the occurrence of SAEs (22.3%), financial (15.9%) and other reasons. Table 3 shows the most frequent SAEs (per 100 patient-years) that led to withdrawal of the biological agents. The commonest reported SAEs were allergy (2.90), serious infections (1.34), tuberculosis (0.93), infusion/injection site reaction (0.75) and malignancies (0.29). Regarding the incidence of SAEs of the anti-TNFα agents, the rates of serious infections (per 100 patient-years) were 1.99, 0.85, 0.63 and 0.61 for IFX, ETN, ADA and GLM, respectively; whereas

the corresponding incidence of tuberculosis was 1.68, 0.43, 0.85 and 0.61, respectively. Infusion/injection site reaction (per 100 patient-years) was highest with IFX (1.38) and skin allergy/anaphylaxis was also most commonly reported with IFX (3.75). There were a total of 32 cases of tuberculosis (TB) reported this website to our registry, exclusively related to the use of the anti-TNF biological agents. Twelve (37.5%) cases developed TB during the 6 months of treatment, whereas four (12.5%) cases developed this infection between 6 and 12 months of therapy.

Twenty-two (69%) patients had TB localized to the lung but the remaining 10 (31%) cases had disseminated disease (miliary TB). Routine screening for latent TB was performed by the tuberculin skin test (TST). Twenty-four percent of patients were given isoniazid treatment for latent TB before the commencement of anti-TNFα therapies. Forty-six episodes of non-TB serious infections were reported. The commonest sites of infection were check details the lower respiratory

tract (46%), followed by soft tissue/skin (20%) and the upper respiratory tract (9%). Disseminated infection (septicemia) was reported in 7% of these episodes. Bacteria contributed to 74% of these infective episodes, whereas there were seven cases of viral (herpes zoster in three, cytomegalovirus in two and hepatitis B reactivation in two cases) and five cases of fungal infection (candidiasis in four and aspergillosis in one). As GLM, TCZ, ABA and RTX had a relatively short history of use in Hong Kong, we only studied the drug retention rates of IFX, ETN and ADA. As shown in Table 3, users of ADA had shorter total patient-years of follow-up than those of IFX or ETN. The cumulative probability of drug withdrawal due to either inefficacy or SAEs in 5 years Adenosine was highest with IFX (64.5%), followed by ETN (44.2%) and ADA (36.9%) (P < 0.001 for IFX compared to others) (Fig. 1). A similar pattern of drug withdrawal was seen when withdrawal due to inefficacy only was considered (Fig. 2). Table 4 shows the results of Cox regression for factors associated with withdrawal of the biological agents because of either inefficacy or SAEs. Increasing age (hazards ratio [HR] per year 1.01 [1.001–1.016; P = 0.02]), female sex (HR 1.46 [1.18–1.81]; P < 0.001), not having a diagnosis of SpA (HR 0.67 [0.53–0.85]; P = 0.001) and IFX use (vs. non-IFX, HR 1.49 [1.25–1.