Eur J Orthop Surg Traumatol
DOI 10.1007/s00590-015-1691-5
ORIGINAL ARTICLE • GENERAL ORTHOPAEDICS - TRAUMA
The effects of diabetes medications on post-operative long bone
fracture healing
C. Simpson1,4 • D. Jayaramaraju2 • D. Agraharam2 • S. Gudipati1
R. Shanmuganathan2 • P. V. Giannoudis1,3
Received: 3 May 2015 / Accepted: 8 August 2015
 Springer-Verlag France 2015
Abstract
Purpose Diabetes has long been known to have an impact
on bone repair. More recently, however, most diabetic
patients receive medications to normalise this hypergly-
caemic environment. To date, no studies have investigated
the effects of diabetic medications on fracture healing in
humans.
Method Patients were identified from two tertiary trauma
centres. Inclusion criteria were adult patients having sus-
tained a closed diaphyseal femoral or tibial fracture, treated
surgically. Exclusion criteria were open, pathological or
peri-prosthetic fractures, and patients having sustained
polytrauma. Matched non-diabetic controls were identified,
matched for age, sex, fracture classification and osteosyn-
thesis. Output measures were: time to callus first appear-
ance, bridging of involved cortices and time to union, along
with the eventual outcome: union/non-union.
Results A total of 36 (25 males) eligible patients were
identified with a control group of 166 patients (138 males).
ANOVA demonstrated class of medication to have a sig-
nificant effect at two of the three time points and on the
eventual outcome. Multiple regression analysis also
demonstrated significant impact (p = 0.02).
& C. Simpson
christopher.simpson@doctors.org.uk
1
Academic Department of Trauma and Orthopaedics, Leeds
General Infirmary, Great George Street, Leeds, England, UK
2
Academic Department of Trauma and Orthopaedics, Ganga
Medical Centre and Hospitals, Coimbatore, India
3
Leeds Biomedical Research Unit, Chappell Allerton Hospital,
Leeds, England, UK
4
35 Potters Lane, East Leake, Loughborough LE12 6NQ,
England, UK
•
Conclusion All classes of medication demonstrated
anti-osteogenic effects compared to the control cohort.
Biguanides demonstrated this in contrast to the in vitro
evidence to date. Sulphonylureas demonstrated this to a
greater extent; however, no in vitro evidence is available
for comparison within this class. Clinicians should be
aware of these delays in bone healing when treating
diabetic patients and aim for optimal blood glucose
control until such time as further research can be
undertaken.
Keywords Diabetes
Fracture (?/- healing)

Osteogenesis
Insulin
Biguanide
Sulphonylurea
Introduction
The diabetic population of the UK is increasing almost
exponentially. In 2008, the estimated diabetic population of
England was over 2.4 million, equating to 4.8 % of the
population. This is expected to have grown to 3.6 million
(6.4 % of the population) by 2025 [1]. Combining these
figures with the number of trauma and orthopaedic opera-
tions being conducted in England, it can be elucidated that
annually approximately 39,000 diabetic patients are being
surgically treated for trauma and orthopaedic injuries.
Moreover, this figure is expected to exceed 52,000 opera-
tions by 2025 [2].
Diabetes has long been known to have an impact on
bone repair, affecting the fracture healing process at mul-
tiple stages. Cozen [3] demonstrated the diabetic phenotype
to significantly delay the appearance and maturation of
fracture callus at the beginning of the healing process,
whereas Loder [4] reported delay across all aspects of the
healing pathway. These effects of diabetes on the skeletal
123

system were subsequently conveniently packaged as ‘dia-
betic osteopathy’ in 1991 by Bouillon [5].
More recently, however, many diabetic patients receive
medication for their condition, normalising their blood
sugar level and therefore by extension neutralising the
aforementioned effects. It may thus be more appropriate in
the current situation to evaluate the effects of these diabetic
medications on the skeletal system, specifically on fracture
healing following surgical stabilisation of the fractures. A
recent review [6] of this topic concluded that in type 1
diabetic patients, the level of glycaemic control is directly
related to the healing outcome. In type 2 diabetic patients
receiving oral hypoglycaemics, thiazolidinediones (e.g.
pioglitazone) exert an anti-osteogenic effect in comparison
with the observed osteogenic properties of biguanides (e.g.
metformin). All but two of the papers analysed were in vitro
or in vivo animal studies, highlighting the distinct lack of
research in this field undertaken within the clinical setting.
The aim of this study therefore was to investigate the
effects of diabetic medications on the healing of long bone
fractures in humans, in comparison with a pair-matched,
non-diabetic patient cohort. We hypothesise that type 1
diabetics will have delayed fracture healing due to the effects
of medicinal insulin having a less efficacious effect on nor-
malising blood glucose than intrinsic. We also hypothesise
that there will be a significant difference in healing rates
between type 2 diabetic patients taking different classes of
oral hypoglycaemic medications, in line with the findings
reported by the aforementioned literature review [6].
Patients and methods
Following appropriate institutional approval, we performed
a retrospective analysis of adult diabetic patients sustaining
long bone fractures of the tibia and femur, treated opera-
tively between 2004 and 2010. Patients were identified
from two tertiary trauma centres, one in the UK and one in
India.
Within the UK centre, patients were identified from a
prospectively maintained database of theatre operative
activity, cross-matched against pharmacy diabetic medi-
cation dispensary data. Within the Indian centre, patients
were identified from a prospectively maintained database
of operative extremity fractures, which also captured co-
morbidities including diabetes.
Study inclusion criteria (group 1) were adult patients
([18 years of age) having sustained a closed diaphyseal
femoral or tibial fracture who were surgically treated.
Exclusion criteria were open, pathological or peri-
prosthetic fractures, fractures treated non-operatively,
123
Eur J Orthop Surg Traumatol
patients with a systemic inflammatory disease and patients
having sustained polytrauma.
Patient demographics, ISS, AO fracture classification,
osteosynthesis performed, mode of mobilisation, type of
diabetic medication and subsequent local and systemic
complications were recorded alongside the radiographic
time to first callus appearance, time to bridging of involved
cortices, time to fracture union and the eventual outcome of
union or non-union.
Non-diabetic controls were identified for each patient
(group 2). Controls were matched for age, sex, fracture
classification and osteosynthesis. On average, at least four
controls were identified for each diabetic participant based
on the above matching criteria, with all possible pair-
matches being used.
The type of surgical implant selected was based on the
fracture personality. Intramedullary nailing procedures
were performed using a fracture table with traction. For
open reduction internal fixation, minimally invasive plate
osteosynthesis was performed adhering to absolute stability
principles.
Narcotics for pain relief requirements were prescribed
routinely using morphine sulphate preparations, tramadol,
codeine and paracetamol. NSAID’s are prohibited in both
institutions as a mode of analgesia in patients treated for
long bone fractures.
Mobilisation was prescribed according to the fracture
pattern and patient’s individual demands. Patients were
asked initially to mobilise toe touch weight bearing for the
first 2–4 weeks, followed by partial weight bearing up to
8 weeks before progressing to full weight bearing (FWB)
thereafter. Following discharge from the hospital, they
were assessed both clinically and radiologically in outpa-
tient clinics at regular intervals every 4 weeks initially up
to 16 weeks and at 24, 36 and 52 weeks and as it was felt
appropriate thereafter based on their condition.
Fracture union was defined clinically when the patient
was able to bear weight without pain and radiologically
when bridging callus was visible in three out of the four
cortices.
Timing of fracture healing was assessed by two of the
authors (C.M.S. and S.G.) using radiological imaging,
corroborated with clinic letters.
Statistical analysis was performed using the SPSS sta-
tistical package. Descriptive statistics were evaluated for
all groups. One-way ANOVA was used to analyse differ-
ences between medication groups pertaining to multiple
outcome measures. Multiple regression analysis was used
to analyse the impact of each variable on individual out-
come measures. A statistically significant level was taken
as p \ 0.05.
Eur J Orthop Surg Traumatol

Table 1 Patient demographics

Demographic Subset Osteosynthesis Diabetic Non-diabetic

Age 55.6 (25–83) 43.4 (21–83)

Long bone Femur Nail 12 62
MIPO 10 28
Tibia Nail 9 52
MIPO 5 24
Sex Male 25 138
Female 11 28
ASA (average) 1.91 1.51

Results Table 2 One-way ANOVA output of diabetic medication effects on

all outcome variables
In total, 36 (25 males) diabetic patients met the inclusion Variable df F Sig. (p)
criteria and formed group 1. They had a mean age of
Callus 5 8.653 \0.001
55.6 years (range 25–83). A total of 22 patients had sus-
Bridging 5 2.939 0.014
tained a femoral fracture and 14 a tibial fracture. The
control group consisted of 166 patients (138 males), with a Union 5 0.93 0.464
mean age of 43.4 years (range 21–83). Ninety patients had Non-union (y/n) 5 3.388 0.006
sustained a femoral fracture and 76 a tibial fracture. Full
patient demographics are illustrated in Table 1. There was
no statistical significance between the two groups regard-
ing patient demographics, indicating no significant selec-
Time to callus first Appearance (days)

180
tion bias in identifying patients. 160
A one-way ANOVA analysis looking at the effects of 140
diabetic medication on time to callus formation, time to 120
bridging of involved cortices, time to union and eventual 100
outcome (union/non-union) demonstrated diabetic medi- 80
cation to have a significant impact on all measures except 60
time to union (Table 2). 40
The mean time to callus formation in diabetic patients 20
taking biguanides (128.8 days) or sulphonylureas 0
(168.0 days) was more than double that found in the con- Control Insulin Biguanide
trol group (60.2 days) (p \ 0.001) (see Fig. 1). Sulphonylurea Biguanide + Sulphonylurea Biguanide + DPP4 inhibitor
Multiple regression analysis of all independent variables
Fig. 1 Time to callus first appearance. *p \ 0.001
measured demonstrated whether a patient was diabetic or
not and which diabetic medication they were taking to have
a significant impact on both time to callus formation and biguanide, sulphonylurea and DPP4 inhibitor classes of
time to bridging of involved cortices (p = 0.02). In terms diabetic medications.
of outcome—union versus non-union, the medication being The presented results should be interpreted with caution
taken, but not simply being diabetic or not, was shown to due to inherent limitations of our study. First, this is a
exert a significant effect. retrospective case–control study with all the inherent lim-
itations associated with this; also the number of cases
included to our study group is small. Secondly, there is an
Discussion inherent variance in time points of clinical and radiological
follow-up, attributable to the retrospective nature of the
Diabetes and the medications used for its treatment have study. Time to outcome was taken at the first observed
been shown to have a significant effect on the skeletal point; however, so, any error incurred should be uniform
system, both in terms of initial risk of fracture and subse- throughout all cohorts. Certain patient demographics are
quent fracture healing [6]. This study aimed to determine also not included in the variable analysis, most notably
differences between diabetics and non-diabetics and also smoking status and BMI when discussing delayed union/
assess the effect on long bone fracture healing, of insulin, non-union. Again this is due to the lack of data points

123

Table 3 Multiple regression output of all independent variables
effect on time to bridging of involved cortices
Independent variable b SE Sig. (p)
(Constant) -21.127 50.737 0.681
Age 0.638 0.632 0.324
Sex -0.821 22.99 0.972
Medication -24.019 9.534 0.02
Diabetic 113.796 35.071 0.004
ASA -16.899 9.793 0.099
ISS 4.471 2.481 0.086
available given this studies retrospective nature. Strengths
of the study include the presence of a matched controlled
group, on average at least four controls for each diabetic
participant. Moreover, this study also provides analysis
over a significant number of parameters associated with the
healing process of long bone fractures not least the pres-
ence of diabetes in the patient’s history as well as the
means for its management (Table 3).
Insulin treatment rectifies the decreased endogenous
insulin levels observed in diabetic patients and has been
shown to normalise indices of callus bone content and
mineralisation [7]. Gross observations have been made
regarding the effects of insulin treatment on fracture
healing in both diabetic rats [8] and mice [9, 10] with
findings including normalised bone-bridging and improved
post-fracture callus bone formation. In this study, we could
not demonstrate such a dramatic normalisation with insulin
treatment, however, outcome measures were nearer to
control measures with insulin treatment than with other
oral hypoglycaemic medications. This may reflect a better
control of the patients’ diabetes and lower blood glucose
levels with insulin treatment compared to other treatment
options. Glycosylated collagen products have been shown
to enhance osteoclast-induced bone resorption [11].
Unfortunately, blood glucose measurements were not
available within this retrospective study that would have
allowed us to elaborate further. Alternative arguments may
include the observation that the majority of insulin-de-
pendent diabetic patients are type 1 diabetics and thus of a
younger mean age. In the regression analysis, however, age
was shown to have no significant impact on any of the
observed outcome measures.
In vitro and in vivo studies of biguanides to date have
demonstrated an osteogenic effect of the medication [12–
15]. Over a short time period, Molinuevo et al. [14]
demonstrated metformin to enhance the expression of
osteoblast-specific transcription factor Runx2/Cbfa1 and to
activate AMPK. Over a longer time period, Gao et al. [15]
observed greater bone density and quality in diabetic rats
123
Eur J Orthop Surg Traumatol
treated with metformin. In this study, however, diabetic
patients treated with metformin were demonstrated to show
significantly delayed fracture healing when compared to
non-diabetic controls in both short- and longer-term mea-
sures of healing. One explanation for this stark contrast is
that studies to date almost unanimously compared diabetic
rodents treated with metformin to untreated diabetic
rodents as opposed to non-diabetics ones. It could thus be
argued that in this study, the proposed osteogenic effects of
metformin were not sufficient to counteract the anti-os-
teogenic effects of a hyperglycaemic environment includ-
ing the aforementioned enhanced osteoclast-induced bone
resorption [11] along with its negative impact on enzymatic
and non-enzymatic crosslinking in bone, impairing its
mechanical properties [16].
No in vitro or in vivo evidence pertaining to the effects
of sulphonylureas on the musculoskeletal system could be
found. All conclusions drawn regarding the effects of this
class of anti-diabetic medication are thus solely from the
evidence collected as part of this study. Sulphonylureas
appear to exhibit an anti-osteogenic effect similar to that of
biguanides over more short-term measures of fracture
healing; however, their impact appears to be reduced over
more long-term measures. Whether this observed differ-
ence is as a result of tighter glycemic control or an as yet
undiscovered or unreported direct impact on the fracture
healing process only further research will determine.
The two other main classes of anti-diabetic medications
not assessed as monotherapy in the study, thiazolidine-
diones (e.g. pioglitazone) and DPP4 inhibitors (e.g. sita-
gliptin), have themselves been shown to have an effect on
the musculoskeletal system. Thiazolidinediones have been
shown in vitro to decrease the activity of Wnt, bone mor-
phogenic protein (BMP) and IGF-1 pathways as well as
inducing production of RANKL, a cytokine that supports
osteoclast development [17]. This translates into the in vivo
observations of decreased tibial bone mineral density in
rats [18] and decreased osteoblast markers in post-meno-
pausal women [19]. The observed effects of thiazolidine-
diones have been succinctly summarised by Lecka-Czernik
[20]. Although the evidence to date regarding DPP4 inhi-
bitors is limited, they may possibly exert an osteogenic
effect on the skeletal system. They have been observed to
not affect femoral bone mineral density, geometry or
mechanical properties and even increase vertebral bone
mineral density in a mouse model of diabetes [21].
Other classes of diabetic medication including thiazo-
lidinediones have been shown to have an effect on the
skeletal system [20] but have not been evaluated in our
cohort of patients, owing to no individuals receiving these
classes being identified within our cohort.
Eur J Orthop Surg Traumatol
Conclusion
Diabetic patients, whether they are receiving insulin or oral
hypoglycaemic medications, be that monotherapy or
polytherapy, progress slower towards long bone fracture
union compared to their non-diabetic counterparts, in line
with our original hypothesis. The proportion of this delay
attributable to the diabetic medication and the proportion
attributable to the hyperglycaemic environment are yet to
be ascertained.
Further research is necessary to elucidate the con-
tributing factors in each case and to guide future clinical
practice. Until such results are available, clinicians should
take this fact into account during their consultation, choice
of fixation methods and biologic augmentation strategies.
Close collaboration of the surgical team with their internal
medicine and endocrinology colleagues will allow for
optimisation of the patients’ glucose control and fracture
management.
Compliance with ethical standards
Conflict of interest None.
References
1. NHS (2011) Diabetic prevelence. NHS England, UK
2. NHS (2011) NHS hospital episode statistics. NHS England, UK
3. Cozen L (1972) Does diabetes delay fracture healing? Clin
Orthop Relat Res 82:134–140
4. Loder RT (1988) The influence of diabetes mellitus on the
healing of closed fractures. Clin Orthop Relat Res 232:210
5. Bouillon R (1991) Diabetic bone disease. Calcif Tissue Int
49(3):155–160
6. Simpson CM, Calori GM, Giannoudis PV (2012) Diabetes and
fracture healing: the skeletal effects of diabetic drugs. Expert
Opin Drug Saf 11(2):215–220
7. Beam HA, Parsons JR, Lin SS (2006) The effects of blood glu-
cose control upon fracture healing in the BB Wistar rat with
diabetes mellitus. J Orthop Res 20(6):1210–1216
8. Follak N, Klöting I, Wolf E, Merk H (2004) Improving metabolic
control reverses the histomorphometric and biomechanical
abnormalities of an experimentally induced bone defect in
spontaneously diabetic rats. Calcif Tissue Int 74(6):551–560
9. Santana RB, Xu L, Chase HB, Amar S, Graves DT, Trackman PC
(2003) A role for advanced glycation end products in diminished
bone healing in type 1 diabetes. Diabetes 52(6):1502–1510
10. Kayal RA, Alblowi J, McKenzie E, Krothapalli N, Silkam L,
Gerstenfeld L et al (2009) Diabetes causes the accelerated loss of
cartilage during fracture repair which is reversed by insulin
treatment. Bone 44(2):357–363
11. Yamagishi S, Nakamura K, Inoue H (2005) Possible participation
of advanced glycation end products in the pathogenesis of
osteoporosis in diabetic patients. Med Hypotheses 65(6):1013
12. Cortizo AM, Sedlinsky C, McCarthy AD, Blanco A, Shurman L
(2006) Osteogenic actions of the anti-diabetic drug metformin on
osteoblasts in culture. Eur J Pharmacol 536(1):38–46
13. Kanazawa I, Yamaguchi T, Yano S, Yamauchi M, Sugimoto T
(2008) Metformin enhances the differentiation and mineralization
of osteoblastic MC3T3-E1 cells via AMP kinase activation as
well as eNOS and BMP-2 expression. Biochem Biophys Res
Commun 375(3):414–419
14. Molinuevo MS, Schurman L, McCarthy AD, Cortizo AM, Tolosa
MJ, Gangoiti MV (2010) Effect of metformin on bone marrow
progenitor cell differentiation: in vivo and in vitro studies. J Bone
Miner Res 25(2):211–221
15. Gao Y, Li Y, Xue J, Jia Y, Hu J (2010) Effect of the anti-diabetic
drug metformin on bone mass in ovariectomized rats. Eur J
Pharmacol 635(1):231–236
16. Saito M, Fujii K, Moti Y, Marumo K (2006) Role of collagen
enzymatic and glycation induced cross-links as a determinant of
bone quality in spontaneously diabetic WBN/Kob rats. Osteo-
poros Int 17(10):1514–1523
17. Shockley KR, Rosen CJ, Churchill GA, Lecka-Czernik B (2008)
PPARc2 nuclear receptor controls multiple regulatory pathways
of osteoblast differentiation from marrow mesenchymal stem
cells. J Cell Biochem 106(2):232–246
18. Lin TH, Yang RS, Tang CH, Lin CP, Fu WM (2007) PPARc
inhibits osteogenesis via the down-regulation of the expression of
COX-2 and iNOS in rats. Bone 41(4):562–574
19. Grey A, Bolland M, Gamble G, Wattie D, Horne A, Davisdon J,
Reid IR (2007) The peroxisome proliferator-activated receptor-c
agonist rosiglitazone decreases bone formation and bone mineral
density in healthy postmenopausal women: a randomized, con-
trolled trial. J Clin Endocrinol Metab 92(4):1305–1310
20. Lecka-Czernik B (2010) Bone loss in diabetes: use of antidiabetic
thiazolidinediones and secondary osteoporosis. Curr Osteoporos
Rep 8(4):178–184
21. Kyle KA, Willett TL, Baggio LL, Drucker DJ, Grynpas MD
(2011) Differential effects of PPAR-c activation versus chemical
or genetic reduction of DPP-4 activity on bone quality in mice.
Endocrinology 152(2):457–467
123