Clin Orthop Relat Res. 2019 Dec 13. doi: 10.1097/CORR.0000000000001089. [Epub ahead of print]
Observational studies showed that exposure to exogenous insulin increases fracture risk. However, it remains unclear whether the observed association is a function of the severity of underlying type 2 diabetes mellitus, complications, therapies, comorbidities, or all these factors combined. That being so, and because of the relative infrequency of these events, it is important to study this further in a large-database setting. QUESTION/PURPOSES: (1) Is switching from oral antidiabetic agents to insulin associated with an increased fracture risk? (2) How soon after switching does the increased risk appear, and for how long does this increased risk persist?
Data from healthcare utilization databases of the Italian region of Lombardy were used. These healthcare utilization databases report accurate, complete, and interconnectable information of inpatient and outpatient diagnoses, therapies, and services provided to the almost 10 million residents in the region. The 216,624 patients on treatment with oral antidiabetic therapy from 2005 to 2009 were followed until 2010 to identify those who modified their antidiabetic therapy (step 1 cohort). Among the 63% (136,307 patients) who experienced a therapy modification, 21% (28,420 patients) switched to insulin (active exposure), and the remaining 79% (107,887 patients) changed to another oral medication (referent exposure). A 1:1 high-dimension propensity score matching design was adopted for balancing patients on active and referent exposure. Matching failed for 3% of patients (926 patients), so the cohort of interest was formed by 27,494 insulin-referent couples. The latter were followed until 2012 to identify those who experienced hospital admission for fracture (outcome). A Cox proportional hazard model was fitted to estimate the hazard ratio (HR) for the outcome risk associated with active-exposure (first research question). Between-exposure comparison of daily fracture hazard rates from switching until the 24 successive months was explored through the Kernel-smoothed estimator (second research question).
Compared with patients on referent exposure, those who switched to insulin had an increased risk of experiencing any fracture (HR = 1.5 [95% CI 1.3 to 1.6]; p < 0.001). The same risk was observed for hip and vertebral fractures, with HRs of 1.6 (95% CI 1.4 to 1.8; p < 0.001) and 1.8 (95% 1.5 to 2.3; p < 0.001), respectively. Differences in the daily pattern of outcome rates mainly appeared the first 2 months after switching, when the hazard rate of patients on active exposure (9 cases for every 100,000 person-days) was higher than that of patients on referent exposure (4 cases for every 100,000 person-days). These differences persisted during the remaining follow-up, though with reduced intensity.
We found quantitative evidence that switching from oral antidiabetic therapy to insulin is associated with an increased fracture risk, mainly in the period immediately after the start of insulin therapy. The observed association may result from higher hypoglycemia risk among patients on insulin, which leads to a greater number of falls and resulting fractures. However, although our study was based on a large sample size and highly accurate data, its observational design and the lack of clinical data suggest that future research will need to replicate or refute our findings and address the issue of causality, if any. Until then, though, prescribers and patients should be aware of this risk. Careful control of insulin dosage should be maintained and measures taken to reduce fall risk in these patients.
LEVEL OF EVIDENCE:
Level III, therapeutic study.