(205) Osteoporosis treatment patterns in France: A Clustering Analysis of the SNDS PMO Cohort Study

Background: Persistence with osteoporotic (OP) treatment is relatively low in France, with high proportions of treatment discontinuations and switches (Belhassen M. et al., DOI: 10.1007/s00198-016-3789-y).

Objectives: Determine rates of OP treatment (the RANKL inhibitor denosumab [Dmab] and bisphosphonates [BPs]) discontinuation and restarting; identify patient groups with similar OP treatment patterns.

Methods: Our retrospective cohort study used the French national claim database SNDS (Système National des Données de Santé) and included women aged ≥60 years with no healthcare resource use (HCRU) markers of Paget disease (bone) or cancer, initiating Dmab or BP between 1st January 2012 and 31st December 2015, irrespective of treatment line. We aimed to describe OP treatment patterns. Long term discontinuation (LTD) was defined as treatment discontinuation of at least one year, occurring after 2 (Dmab) or 3 (BP) consecutive years of treatment. Non-supervised clustering methods (hierarchical ascending classification and k-medoids partition) were used to identify patient clusters with similar OP treatment pathways. The sociodemographic characteristics and medical and treatment history of the identified clusters were described.

Results: A total of 296,675 women were included in the study: median follow-up, 7.3 years. On average (median [Q1, Q3]), they had received index OP treatment for 21.2 (5.5, 52.4) months. Over two-thirds (69.3%) had at least one temporary discontinuation, with a median (Q1, Q3) cumulative duration of 6.8 (2.7-13.0) months. Over three-quarters (79.5%) had LTD. One-fifth (n=47800) of those with LTD resumed OP treatment; in most cases (72.2%) the resumed OP treatment was the one previously discontinued.
Six clusters were identified: 1) 36.4% of the study population, indicating a pattern of “early” discontinuation of oral BP; 2) long term oral BP, 25.9%; 3) “late” discontinuation of oral BP, 19.4%; 4) fragility fracture, 7.6%; 5) intravenous BP, 7.2%; 6) Dmab, 3.4%. The median (Q1, Q3) cumulative duration of oral BP exposure ranged from 2.8 (1.8, 5.5) months for cluster 1 to 46.3 (38.4, 51.8) months for cluster 2. Compared with other clusters, the fragility fracture cluster was older (mean [SD] age, 78.1 [8.6] vs 72.4 [8.0]-74.7 [9.0] years) with higher age-adjusted Charlson comorbidity index. Cluster 6 (Dmab) had received a median of 2 OP treatments versus 1 in all other clusters.

Conclusions: In France, a large proportion of women receiving OP treatment discontinue and few restart. We categorized OP treatment patterns on a national population of PMO women who initiated OP treatment, without any predefined hypothesis to classify patients thank to non-supervised clustering methods.