Background One important aim of modern pharmacologic treatment in patients with rheumatoid arthritis (RA) is to prevent joint destruction and reduce the need for orthopedic surgery.
Objectives To investigate the association between anti-TNF treatment and orthopedic surgery procedures in a population-based sample of patients with RA.
Methods The study was based on a dynamic cohort, established in 1997, of all known patients from a defined geographical area with a validated clinical diagnosis of RA. Data on orthopedic surgery were retrieved from a regional health care register, which contains information on all inpatient and outpatient procedures in the area. The cohort was also linked to a regional register of patients with arthritis treated with biologics. Questionnaires on patient reported outcomes (PROs) and medication were sent to the RA patients in the cohort in 1997, 2002, 2005 and 2009 (response rates 62-74%), and data on PROs were also retrieved from the regional biologics register at the start of anti-TNF therapy. The date of the first available health assessment questionnaire (HAQ) score was used as the index date. Anti-TNF exposure was assessed as a time dependent variable, with the end of exposure defined as 30 days after stopping treatment. Patients were followed to death, first relevant procedure (used as outcome) or Dec 31, 2011. Cox regression models were adjusted for HAQ as a time dependent variable, for a propensity score for anti-TNF treatment, based on demographics and baseline clinical characteristics, and for category (quartile) of age at baseline. Analyses were stratified by quartile of disease duration at the index date.
Results A total of 1653 patients (72% female, 71% RF positive) were included. 541 patients (33%) were treated with TNF-inhibitors at some time during the study period. These patients were younger and more likely to be RF positive, and had worse PROs, at the index date. Orthopedic surgery procedures occurred in 499 patients (30%) during the follow-up. Patients who were treated with TNF-inhibitors were more likely to have orthopedic surgery (hazard ratio (HR) 1.37; 95% confidence interval (CI) 1.06-1.75). In individuals with a short disease duration at the index date (<3.5 years; n=417), there was no significant association between anti-TNF treatment and surgery (HR 1.08; 95% CI 0.65-1.80). Anti-TNF treated patients were overall more likely to have foot surgery (p<0.001) and knee surgery (p=0.04), but there was no such association for hip surgery (HR 0.84; 95% CI 0.53-1.33). In patients with a long disease duration (>16.5 years; n=422), treatment with TNF-inhibitors was associated with an increased risk of knee surgery (HR 3.58; 95% CI 1.74-7.36), but a reduced risk of hip surgery (HR 0.27; 95% CI 0.08-0.91).
Conclusions The association between anti-TNF treatment and orthopedic surgery overall most likely reflects confounding by indication in patients with severe, refractory RA. This pattern was not seen in early disease, where the benefits of TNF-inhibitors on joint damage may be greater. Anti-TNF treatment may have a different effect on disease progression in hip joints compared to knees and feet.
Disclosure of Interest None declared