Sanni Ali's presentation from Osteoporosis 2016: Antidiabetic medication use and the risk of fracture amongst type 2 diabetic patients: a nested case-control study Find out more at: https://nos.org.uk/conference

1. Antidiabetic Medication Use and Fracture
Risk in T2DM Patients:
a nested case-control study
M Sanni Ali, Eladio L, Berta S, Dídac M, and Daniel Prieto-Alhambra
National Osteoporosis Conference 2016

2. Introduction
 Type 2 diabetes (T2DM) is associated with increased bone mineral
density (BMD) and higher risk of fracture.
 Existing prediction tools such as FRAX underestimate fracture risk.
 Hypoglycaemic treatment could modulate the risk of fractures in
many ways.
 Observational studies on the effect of antidiabetic medications have
conflicting findings.
 Some studies reported higher risk of fractures in insulin-treated
patients:
- Higher prevalence of diabetic complications/comorbidities.
- Higher risk of hypoglycaemic episodes and falls.

3. Introduction
 Only few studies have looked at the comparative safety of different
antidiabetic medications.
Palermo A et al. Osteoporos Intl 2015

4. Introduction
 Aim: To investigate the comparative fracture risk associated with
the different antidiabetic medications in primary care practice.

5. Methods
 Data source: SIDIAP database
- One primary care health provider using one same e-records
software.
- 274 Primary Care Centres in Catalonia
- > 3,400 GPs.
- > 5 million patients (>80% population).
- > 55 million person-years of research-usable data (2005 onwards).
 Study participants:
- Incident T2DM patients registered in SIDIAP, aged 40 years or
older, with an eGFR >15 at T2DM diagnosis.

6. Methods

7. Methods
 Study period: 1st January 2006 - 31st December 2013
 Design: Nested case-control study.
 Exposure: Antidiabetic medications (Reference = Metformin)
 Outcome: Osteoporotic fracture (cases)
 Exclusion Criteria:
- Patients having fractures before entry date.
- T2DM diagnosis date before 1st January 2006 or practice
registration date or within the last year of the study period.
- Patients with chronic kidney failure (eGFR <15).
- Patients < 40 years of age on the date of T2DM diagnosis.

8. Methods
 Cases were (risk set) matched with up to 5 controls on (“epi” package)
• 10-year bands on calendar year of birth
• Calendar year of cohort entry (T2DM cohort)
• Gender
 Multiple imputation with chained equation was used to impute missing
data (“mice” in R).
 Conditional logistic regression was used to estimate OR and 95%CI
adjusting for the confounders listed in the baseline table (ccwc, R).
Potential confounders: demographic variables, comorbidities (eGFR
and HBA1c), comedication, life style factors (Smoking,
Alcohol use, socioeconomic status)

9. Results
Flow chart of data extraction for T2DM cohort and case-control

10. Results
Table: Baseline Characteristics of Cases and Controls
Characteristics Controls (%) Cases (%)
Age (Mean,sd) 72.86 (11.3) 72.90 (11.40)
Gender (%,female) 7277 (0.7) 1457 (0.71)
Stroke (N,%) 605 (0.06) 171 (0.01)
Transient Ischaemic
Attack 207 (0.02) 63 (0.03)
IHD 612 (0.06) 145 (0.07)
Statins 4439 (0.43) 845 (0.41)
ACEi 3280 (0.32) 695 (0.34)
CCB 1987 (0.19) 413 (0.20)
Thiazides 1884 (0.18) 349 (0.17)
Beta.Block 1999 (0.20) 433 (0.21)
Vitamin D 225 (0.02) 51 (0.02)
Calcium Supplement 247 (0.02) 71 (0.03)
Calcium and VitD 1296 (0.13) 384 (0.19)
SERM 98 (0.01) 16 (0.01)
Bisphosphonate 907 (0.09) 278 (0.14)
Strontium renalate 101 (0.01) 31 (0.02)
HBA1c (Mean, sd) 6.65 (1.24) 6.65 (1.28)
eGFR (Mean, sd) 74.04 (22.64) 75.32 (23.86)
BMI (Mean, sd) 30.7 (5.17) 30.25 (5.2)

11. Results
Table: Association Between Different Antidiabetic Medications and Risk of
Fracture
Total
(N= 12,277)
Cases
(N=2,049)
Adjusted
Odds Ratio 95% CI
Metformin 4079 625 REFERENCE
Insulin 546 126 1.63 1.30 2.04
Sulfonylurea 697 119 1.13 0.91 1.41
Gliptins:DPP4 77 14 1.21 0.67 2.19
Insulin +
Metformin
329 58 1.22 0.89 1.65
Metformin and
Sulfonylurea
946 177 1.29 1.07 1.56
Metformin and
DPP4
218 36 1.12 0.77 1.62
Adjusted for age/sex, comedications, comorbidities, BMI, eGFR, HBA1c, smoking,
alcohol use and socioeconomic Status.

12. Results
Table: Timing of Insulin Use and Fracture Risk
Total Cases
Odds
Ratio 95% CI
Never Users Past 2 Years 3933 666 1.00
Current Users- Previous 6
Months 470 116 1.52 1.19 1.93
Recent Users- Between Previous
6 and 12 Months 78 14 1.02 0.57 1.85
Past Users- Between 12 and 18
Months 37 7 1.19 0.51 2.74
Previous Past Users- Between
18 and 24 Months 22 2 0.46 0.10 2.00

13. Discussion
 Insulin treatment is associated with increased risk of
fractures compared to metformin.
 Fracture risk appears to increase during insulin use (6-
months) and disappear with insulin discontinuation.
 Strengths: adjustment was possible for several
confounders including HBA1c, eGFR, among others.
 Limitations: Power and unmeasured confounding could
still explain the observed association.