Background The evaluation of cortical bone mineral density (BMD) on metacarpal bones by digital radiogrammetry (DXR) has been proven to be a simple, reliable and predictive method to evaluate the severity of the disease in patients with early arthritis (EA). However, DXR is a tool that is not usually available in our environment. By contrast, dual X-ray absorptiometry (DXA) is a more familiar and accessible technique in our clinical practice. Objectives The aim of this study was to compare the association between BMD measurements of the hand by DXR and DXA with parameters of activity and severity at two years of follow-up in a cohort of patients with EA. Methods A prospective longitudinal study of patients with EA was done. DXR was performed in a total of 111 patients (87.4% women) and DXA was implemented in a total of 378 (82% women). Mean age at disease onset was 57 years [46 - 65 (p25 - p50)] in the DXR group and 54 years [44 - 66 (p25 - p50)] in the DXA group. Anthropometric and clinical data were collected per protocol during 2 years of follow-up. Forty-two percent of patients in the DXR group presented citrullinated peptide antibodies and 41.3% in the DXA group. In both, the 57% fulfilled Rheumatoid arthritis (RA) 2010 criteria at the start of follow up (43% undifferentiated arthropathy). Each patient underwent a digital radiograph of both hands (GE © DX Definium 8000) at 0, 3, 12 and 24 months, determining BMD of each hand and the mean of both measures by DXR (Sectra, Linköping, Sweden). Also, DXA of global hand and metacarpophalangeal joints (MCPs) of the nondominant hand were performed and analyzed (Hologic QDR -4500 Elite©) at 0, 6, 12 and 24 months. In addition, a variable that measures the intensity of cumulative treatment received during the 2 year follow-up was specifically generated as a marker of severity. Statistical analysis was performed using the statistical package STATA 12. Results Our data show a good correlation between values of BMD obtained by DXR and DXA in the different locations studied (global hand and MCPs: r=0.830 and 0.718, respectively, p=0.0001), both at the baseline visit and along the two years of monitoring. In the bivariate analysis, a negative association is observed between baseline BMD values measured by DXA and disease activity by DAS28 at 2 yrs, which disappears when adjusting for other variables (age and sex). However, we found an inverse relationship between the intensity of cumulative treatment at two years and baseline BMD measured by DXA, both at global hand (r = -2.51, p=0.041, n=220) and MCPs (r = -3.45, p=0.007, n=221). The DXR association was not significant, probably due to the small sample size (n=32). Conclusions BMD of the global hand and MCPs of the nondominant hand assessed by DXA predicts disease severity and the intensity of cumulative treatment in the first two years of follow-up in a population of patients with EA. By contrast, predictive value of the hand DXR was not demostrated. Further studies with a larger population are needed to obtain ...