Accepted Article

Original Article

Glycated Hemoglobin Level and Risk of Hip Fracture in Older People with Type
†
2 Diabetes: A Competing Risk Analysis of Taiwan Diabetes Cohort Study

Chia-Ing Li1,2, Chiu-Shong Liu1,2,3, Wen-Yuan Lin1,2, Ching-Chu Chen 4,5, Sing-Yu
Yang6, Hsuan-Ju Chen7, Cheng-Chieh Lin1,2,3*, Tsai-Chung Li 6,8 *

1. School of Medicine, College of Medicine, China Medical University, Taichung,

Taiwan
2. Department of Medical Research, China Medical University Hospital, Taichung,
Taiwan
3. Department of Family Medicine, China Medical University Hospital, Taichung,
Taiwan
4. Division of Endocrinology and Metabolism, Department of Medicine, China
Medical University Hospital, Taichung, Taiwan
5. School of Chinese Medicine, College of Chinese Medicine, China Medical
University, Taichung, Taiwan
6. Graduate Institute of Biostatistics, College of Management, China Medical
University, Taichung, Taiwan
7. Management Office for Health Data, China Medical University Hospital,
Taichung, Taiwan
8. Department of Healthcare Administration, College of Health Science, Asia
University, Taichung, Taiwan

* Correspondence to: Tsai-Chung Li & Cheng-Chieh Lin

China Medical University, 91 Hsueh-Shih Road, Taichung, 40421, Taiwan, Tel:
886-4-22053366 ext. 6605, Fax: 886-4-22078539, e-mail: cclin@mail.cmuh.org.tw

†
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which
may lead to differences between this version and the Version of Record. Please cite this
article as doi [10.1002/jbmr.2462]

Initial Date Submitted September 9, 2014; Date Revision Submitted January 15, 2015; Date Final

Disposition Set January 16, 2015

Journal of Bone and Mineral Research

This article is protected by copyright. All rights reserved
DOI 10.1002/jbmr.2462

This article is protected by copyright. All rights reserved

Accepted Article
Abstract

Hip fracture, which is associated with substantial morbidity and long-term mortality,
imposes a major burden on the healthcare system. Diabetes is a risk factor for
osteoporosis, which is a crucial risk factor of hip fracture. However, epidemiological
studies investigating the risk of hip fracture among patients with type 2 diabetes are
limited. This study explored the association between hemoglobin A1c (HbA1c) and
the risk of hip fracture in people with type 2 diabetes aged 65 years and older. We
conducted a retrospective cohort study of 20  025 older patients with type 2 diabetes
who participated in the National Diabetes Case Management Program in Taiwan. The
HbA1c level at the baseline and hip fracture incidence over an average of 7.41 years
of follow-up were analyzed (maximum and standard deviation were 10.9 and 2.42
years, respectively). A total of 1514 hip fracture cases were recorded. The incidence
rates of hip fracture were 9.15, 8.02, 9.58, 10.61, 12.51, and 13.43 per 1000
person-years in patients with baseline HbA1c levels of < 6%, 6–7%, 7%–8%, 8%–9%,
9%–10%, and ≥ 10%, respectively. After multivariate adjustment, the risk of hip
fracture increased among patients with HbA1c levels of 9%–10% and ≥ 10.0 %
compared with patients with HbA1c levels of 6–7 % (hazard ratio, 1.24; 95%
confidence interval, 1.02–1.49 and 1.32; 1.09–1.58, respectively). Significant linear
trends among various HbA1c levels were observed (P < 0.05). Patients with type 2
diabetes whose HbA1c levels exceeded 9.0% exhibited an increased risk of hip
fracture, confirming a linear relationship. Our study’s findings demonstrated the
importance of glycemic control for fracture prevention in older adults with type 2
diabetes. This article is protected by copyright. All rights reserved

Keywords: glycated hemoglobin; hip fracture; type 2 diabetes

This article is protected by copyright. All rights reserved

Accepted Article
Introduction

Hip fractures are useful indicators of osteoporosis and account for the majority

of fracture-related healthcare expenditures and mortality in men and women older

than 50. (1) In addition, most patients with hip fracture require hospitalization. Hip

fractures and the resulting postsurgical outcomes are a major public health concern in

numerous countries and contribute to excess mortality in the year following a hip

fracture, (2) substantial functional loss, (3) and long-term excess mortality. (4)

Although the role of diabetes mellitus (DM) as a risk factor of osteoporosis and

hip fracture has been determined, (5,6) uncertainty remains regarding which factors

related to diabetes increase and reduce hip fracture incidence. The glycated

hemoglobin (HbA1c) level, which reflects the average ambient fasting and

postprandial glycaemia over a 2–3-month period, is critical to clinically managing

hyperglycemia and is an essential factor for monitoring and treating patients with

diabetes because determining patients’ HbA1c levels is easier to than performing an

oral glucose tolerance test and is independent of patient prandial status. (7) In addition,

the HbA1c level is the most widely used marker of long-term glycoregulation. (8)

HbA1c characterizes dysglycemia and has been used as a long-term glycemic control

marker that is more efficient than fasting glucose because it can be used for testing in

a nonfasting status and its stability for glucose level. Thus, the HbA1c level is widely

used as a therapeutic guideline to prevent cardiovascular complications in patients

This article is protected by copyright. All rights reserved

Accepted Article
with diabetes. (9) After conducting a comprehensive literature review, only one

previous study was found that have examined HbA1c levels and the incidence of hip

fracture in patients with type 2 diabetes. (10) More data are required to characterize the

effects that the HbA1c level exerts on hip fracture. Therefore, we explored the

association between the HbA1c level and hip fracture in Chinese patients with type 2

diabetes who participated in the National Diabetes Case Management Program

(NDCMP) in Taiwan.

Methods

Study population

The Taiwan Diabetes Cohort Study, a national population-based retrospective

cohort study, was conducted using a sample of 63  084 ethnic Chinese patients with

type 2 diabetes who were enrolled in the NDCMP during 2002 to 2004 in Taiwan.

Patient date of entry into the NDCMP was set as the index date. Patients clinically

diagnosed with DM according to the criteria established by the American Diabetes

Association (International Classification of Diseases, Ninth Revision, Clinical

Modification [ICD-9-CM] Code 250) who were aged 65 years and older and could be

identified from registry files during the baseline and subsequent years were included

as our study subjects (n = 23  780). We excluded patients with type 1 diabetes

(ICD-9-CM Code 250.x1/x3), gestational diabetes (ICD-9-CM Code 648.83), a

This article is protected by copyright. All rights reserved

Accepted Article
history of hip fracture, and no HbA1c measurements. These inclusion criteria were

fulfilled by 20  793 enrolled patients with type 2 diabetes. After we excluded patients

for whom data necessary for covariate information were missing, 20  025 patients

were included in the final data analysis (Figure 1). This study was approved by the

Ethical Review Board of China Medical University Hospital (CMUH102-REC1-035).

Data sources for baseline and follow-up assessments

The Taiwanese government launched the National Health Insurance (NHI)

program, a universal insurance system established by the Bureau of NHI, Ministry of

Health and Welfare, in March 1995. The NHI program has covered approximately

99% of the 23.74 million people in Taiwan since 1999.(11) By the end of 2010, more

than 99.62% of residents of Taiwan were enrolled in the program, and the Bureau of

NHI maintained contracts with 100% of the hospitals and 92% of the clinics in the

nation. (12) The Bureau of NHI randomly audits claims data, and expert reviews of

every 50 to 100 ambulatory and inpatient claims from each hospital and clinic are

conducted quarterly to ensure the validity of the claims data. Every false diagnostic

report is subject to a severe penalty enforced by the Bureau of NHI.

In this study, datasets on inpatient care and outpatient care during 2001 to 2011

were used to define baseline comorbidity and endpoint of interest. Unique personal

identification numbers (PINs) for all patients were scrambled by the National Health

This article is protected by copyright. All rights reserved

Accepted Article
Research Institutes (NHRI) to protect security and privacy. The NHI datasets can be

linked to the NDCMP registry by using the scrambled PIN of each patient. The

National Health Insurance Research Database contains demographic data as well as

data on the date and institution of diagnosis, ambulatory care, inpatient admission,

and outpatient and inpatient treatment of beneficiaries of NHI. ICD-9-CM codes were

used to identify disease statuses. Because NHI coverage is comprehensive, the

proportion of enrollees who had withdrawn from the NHI program was low. Therefore,

bias resulting from loss to follow up was negligible.

Upon enrolling in the NDCMP, patients were required to undergo a

comprehensive assessment, including a series of blood tests, urine tests, and body

measurements, as well as an evaluation of disease status and complications. A

standardized, computerized questionnaire was administered by a case management

nurse to evaluate previous or current diseases, medications, and lifestyle behaviors.

Blood was drawn from an antecubital vein the morning after a 12-hour overnight fast

and was sent for analysis within 4 hours postcollection.

The study included all clinical sites in Taiwan. HbA1c measurements were

performed at clinical laboratory of each hospital or independent-run clinical

laboratories. Before using National Glycohemoglobin Standardization Program

(NGSP)-assigned target values to assess accuracy in 2005, these clinical laboratories

This article is protected by copyright. All rights reserved

Accepted Article
are under regulation for external quality control by proficiency test for Taiwan

Society of Laboratory Medicine (TSLM). It has been reported the inter-laboratory CV

of HbA1c was 13.9% and the bias calculated by difference between measure and

target values was 0.26% before adopting NGSP-certified methods in 2005. (13)

Baseline HbA1c measurement had to be tested within one year after entry of the

study.

Outcome assessment

Using the PINs, we linked the study patients to inpatient claim records

(2002–2011) to identify the first events leading to primary or secondary hip fracture

diagnoses (ICD-9-CM 820) and internal fixation or hemiarthroplasty operations

(ICD-9-CM 79.15, 79.35, 81.52) used as the endpoint of this study. We excluded

diagnoses of pathological fracture (ICD-9-CM 733.14 and 733.15), open hip fracture

(ICD-9-CM 820.1, 820.10–820.12, 820.19, 820.9, 820.13, 820.22, 820.3,

820.30–820.32), and transport accidents (E800–E848) from the outcomes of interest.

The date of the clinical endpoint of interest was the first day of hospitalization, and

the study period was from index date to December 31, 2011. By linking the PIN

numbers with this computerized file, 1514 patients with hip fracture within an average

of 7.41-years of follow up were identified from the cohort. The patients were

followed up from the index date to December 31, 2011, or until hip fracture events,

This article is protected by copyright. All rights reserved

Accepted Article
death, or withdrawal from the NHI program. Data pertaining to chronic medical

conditions and hypoglycemic drugs were collected for the 12-month period prior to

cohort entry by using outpatient and inpatient claims data. Comorbidities, namely,

coronary artery disease (ICD-9-CM Codes 410–413, 414.01–414.05, 414.8, and

414.9), congestive heart failure (ICD-9-CM Codes 428, 398.91, and 402.x1), cancer

(ICD-9-CM Codes 140–149, 150–159, 160–165, 170–175, 179–189, 190–199, 200,

202, 203, 210–213, 215– 229, 235–239, 654.1, and 654.10–654.14), atrial fibrillation

(ICD-9-CM Code 427.31), hyperlipidemia (ICD-9-CM Code 272), hypertension

(ICD-9-CM Codes 401–405), stroke (ICD-9-CM Codes 430–438), chronic hepatitis

(ICD-9-CM Codes 571, 572.2, 572.3, 572.8, 573.1–573.3, 573.8, and 573.9),

obstructive pulmonary disease (490–496), diabetic retinopathy (yes, no),

hypoglycemia (ICD-9-CM Codes 2510–2512), peripheral neuropathy (yes, no), and

alcohol dependence (ICD-9-CM Codes 303 and 305), were identified.

Statistical analysis

The baseline HbA1c levels of each patient were determined using datasets

derived from electronic lab records. The patients were divided into groups according

to baseline HbA1c levels of < 6%, 6–7%, 7%–8%, 8%–9%, 9%–10%, and 10%. An

extended Cox proportional hazards model applied using the Lunn-McNeil approach (a

modified Cox proportional hazards model that consider competing risks) was

This article is protected by copyright. All rights reserved

Accepted Article
employed to evaluate the association between baseline HbA1c level categories and

hip fracture risk. (14) We analyzed whether people at high risk of hip fracture were also at

high risk for death, even after controlling for covariates. We verified the proportional

hazards assumption according to the graph of the log (−log(survival)) versus the log

of the survival time graph by adjusting for all remaining covariates and by testing the

statistical significance of a covariate that allowed HbA1c to have a time varying effect.

(13)
The product terms of five dummy variables representing HbA1c categories with

time t were added into the full multivariate model and the likelihood ratio test was

used to test its significance. We observed no statistically significant violation. We

calculated hazard ratios (HRs) and 95% confidence intervals (CIs) by adjusting for

age, sex, and multiple variables. To examine linear trend of HbA1c, baseline HbA1c

levels of < 6%, 6–7%, 7%–8%, 8%–9%, 9%–10%, and 10% were coded as 0, 1, 2,

3, 4, and 5, respectively. Three multivariate models were employed: the first model

adjusted for age (continuous) and sex (male, female); the second model additionally

adjusted for smoking (yes, no), alcohol consumption (yes, no), the duration of

diabetes (continuous), and type of hypoglycemic drug (no medication, metformin

monotherapy, sulfonylurea monotherapy, oral antidiabetes drug [OAD] monotherapy

other than metformin or sulfonylurea [OAD-other], a combination of metformin and

sulfonylurea, a combination of metformin and OAD-other, a combination of

This article is protected by copyright. All rights reserved

Accepted Article
sulfonylurea and OAD-other, a combination of 2 OAD-others, a combination of 3

OADs, a combination of more than 3 OADs, insulin monotherapy, insulin and one

OAD, insulin and more than one OAD, and short-time OAD), antihypertensive

treatment (yes, no), obesity (yes, no), and baseline fasting plasma glucose

(continuous). The third model additionally adjusted for baseline comorbidities,

namely coronary artery disease (yes, no), congestive heart failure (yes, no), cancer

(yes, no), hyperlipidemia (yes, no), hypertension (yes, no), atrial fibrillation (yes, no),

stroke (yes, no), chronic hepatitis (yes, no), chronic obstructive pulmonary disease

(yes, no), diabetic retinopathy (yes, no), hypoglycemia (yes, no), peripheral

neuropathy (yes, no), and alcohol dependence (yes, no). All analyses were performed

using SAS V9.3 (SAS, Cary, NC, USA). All P values were 2-tailed, and P < .05 was

considered to indicate statistical significance.

Results

During an average of 7.41 years of follow up, a total of 1514 cases of hip fracture

were identified in patients with type 2 diabetes (crude incidence rate, 10.20/1000

person-y; 14.07 for men, 5.54 for women). The incidences of hip fracture were 9.15,

8.02, 9.58, 10.61, 12.51, and 13.43 per 1000 person-years in patients with baseline

HbA1c levels of < 6%, 6–7%, 7%–8%, 8%–9%, 9%–10%, and 10%, respectively.

Table 1 summarizes the baseline sociodemographic and clinical factors of patients

This article is protected by copyright. All rights reserved

Accepted Article
grouped according to hip fracture status. Patients with hip fracture were likely to have

a higher mean age and a longer diabetes duration; be female, nonsmokers, and alcohol

drinkers; use more than 3 oral hypoglycemic drugs, insulin injections, and insulin

injections and oral hypoglycemic drugs; have diabetic retinopathy and peripheral

neuropathy; and not be obese (Table 1). Figure 2 shows the Kaplan-Meier cumulative

incidence curves for hip fracture in the subgroups defined according the HbA1c level.

Patients with an HbA1c level of 10% exhibited the highest risk (log-rank P < .001).

Table 2 lists the HRs of hip fracture and all-cause mortality among the patients

grouped according to various HbA1c levels. Compared with patients with HbA1c

levels of 6%–7%, patients with HbA1c levels of 8%–9%, 9%–10%, and > 10%

exhibited age- and sex-adjusted HRs of hip fracture of 1.28 (1.09–1.51), 1.51

(1.26–1.80), and 1.73 (1.47–2.03), respectively. When lifestyle factors and medication

were considered in the multivariate analysis, the effect exerted by HbA1c was slightly

attenuated; however, 9%–10% and > 10% HbA1c levels remained significant. After

we adjusted for comorbidities and complications, patients with HbA1c levels of

9%–10% and 10% exhibited a higher risk of hip fracture (adjusted HR: 1.24

[1.02–1.49], and 1.32 [1.09–1.58], respectively) compared with patients with an

HbA1c level of 6%–7%. An increasing trend was observed between the levels of

HbA1c and hip fracture incidence (P < .05). Similarly, patients with HbA1c levels of

This article is protected by copyright. All rights reserved

Accepted Article
7%–8%, 8%–9%, 9%–10%, and > 10% exhibited age- and sex-adjusted HRs of

all-cause mortality of 1.10 (1.02–1.18), 1.25 (1.15–1.35), 1.40 (1.28–1.53), and 1.62

(1.50–1.75), respectively. When lifestyle factors and medication were considered in

the multivariate analysis, the effect exerted by HbA1c was slightly attenuated;

however, 8%–9%, 9%–10% and > 10% HbA1c levels remained significant. After we

adjusted for comorbidities and complications, patients with HbA1c levels of 9%–10%

and 10% exhibited a higher risk of all-cause mortality (adjusted HR: 1.13

[1.03–1.24], and 1.26 [1.15–1.37], respectively) compared with patients with an

HbA1c level of 6%–7%. An increasing trend was observed between the levels of

HbA1c and all-cause mortality (P < .001).

Sensitivity analyses were conducted to eliminate potential bias caused by the

existence of comorbidities, and patients with stroke, coronary artery disease,

peripheral neuropathy, diabetic retinopathy, diabetic ketoacidosis, hyperglycemic

hyperosmolar nonketotic coma, and liver disease were excluded (n = 9852). A

similarly significant HR for hip fracture was determined among patients with an

HbA1c level 9 % (1.32 [1.03–1.68]). The results of further stratification according

to sex, obesity, and insulin use (Figure 3) were generally consistent with those

determined in the initial analysis.

This article is protected by copyright. All rights reserved

Accepted Article
Discussion

In this study, we observed an increasing trend between the HbA1c level and hip

fracture incidence in patients with type 2 diabetes aged 65 years and over. The risk of

hip fracture was 24%–31% higher among patients with HbA1c levels 9% than

among patients with HbA1c levels of 6%–7% after we adjusted for numerous risk

factors for fracture. After patients with existing chronic diseases and acute

complications were excluded, data stratification according to sex, BMI, and insulin

use revealed a similar association between the HbA1c level and hip fracture

incidence.

Older people with type 2 diabetes have HbA1c levels higher than 9%, which is

associated with a markedly high risk of hip fracture. Compared with research

pertaining to the incidence of microvascular and macrovascular outcomes, (15-17) the

effect of glycemic control on the risk of hip fracture in patients with type 2 diabetes

has seldom been explored. Previous studies have reported that type 2 diabetes was

associated with an increased risk of hip fracture (relative risk, 1.7; 1.4–2.0) in women

from the Nurses’ Health Study, (18) a relative risk of 1.28 (1.21–1.34) in Chinese men,

and a relative risk of 1.72 (1.66–1.78) in Chinese women. (19) These results indicated

that fracture prevention strategies are required to assist people with type 2 diabetes.

Although previous studies have shown an association between the use of DM

This article is protected by copyright. All rights reserved

Accepted Article
medications and hip fracture, (20,21) there are inconsistencies in results for the effect of

glycemic control on fracture or hip fracture risk. (10,22-24) One study focused on the

effect of tight glycemic control on hip fracture (10) by applying a case-control

approach; tight glycemic control (HbA1c < 7%) was associated with greater risk of

hip fracture in patients treated for type 2 DM. (21) In the Rotterdam Study, Oei et al.

examined the effect of inadequately glucose control (HbA1c 7.5%), and found

participants with inadequately glucose control had 62% higher fracture risk than

diabetic patients with adequately glucose control. (22) Similarly, in the Atherosclerosis

Risk in Communities (ARIC) Study, Schneider et al. reported that there was a

significantly increased risk of fracture-related hospitalization among persons with

diagnosed diabetes with HbA1c 8% compared with those with HbA1c <8%. (23) On

the contrary, in the Action to Control Cardiovascular Risk in Diabetes (ACCORD)

randomized trial, Schwartz et al. did not observe increase fracture or fall risk in group

of standard glycemic control. (24) Although the effect of an HbA1c level < 6% on

increased risk of hip fracture was borderline significant in the present study, our

results verified that a higher risk of hip fracture is associated with higher levels of

HbA1c. In consideration of the substantial morbidity and mortality associated with

hip fracture (25,26) our findings have crucial clinical implications for preventing hip

fracture.

This article is protected by copyright. All rights reserved

Accepted Article Several possible explanations can account for why patients with type 2 diabetes

and hyperglycemia have an increased risk of hip fracture. First, hyperglycemia may

be associated with factors that influence bone strength and quality, and BMD does not

account for all of the hip fracture risk in the investigated population, because the

population has, on average, higher BMDs than the nondiabetic population. (22)

Hyperglycemia may impair vitamin D and calcium metabolism, (27,28) hindering

calcium deposition and subsequent mineralization and possibly weakening bone

quality and increasing bone fragility, and thicker femoral cortices in narrower bones

independent of BMD, thereby increasing fracture risk. (22,29) In addition,

hyperglycemia may lead to increased accumulation of advanced glycation end

products in bone collagen, inducing osteoblast apoptosis (30) and increasing bone

stiffness and fracture susceptibility. (31,32)

Second, older adults with hyperglycemia have a higher risk of long-term

complications, such as neuropathy and retinopathy, and neuromuscular and vision

impairment may increase the likelihood of falls caused by poor balance and loss of

pressure sensitivity. (33,34) Both of these conditions have been reported to be associated

with higher hip fracture rates. (35, 36) Neuropathy and retinopathy may explain the

effect hyperglycemia exerted in the present study and were slightly more prevalent in

patients with hip fracture. To eliminate the possibility that neuropathy and retinopathy

This article is protected by copyright. All rights reserved

Accepted Article
exerted confounding effects, we adjusted for these variables in the multivariate

analysis and evaluated the independent effect of hyperglycemia.

Third, hyperglycemia is an indicator of multiple comorbidities treated using

multiple medications. Older adults using multiple medications to treat multiple

conditions are often frail and more likely to fall, thereby increasing the risk of hip

fracture. Studies have reported that various comorbidities such as stroke are

associated with an increased risk of hip fracture. (33,34)

Our findings are crucial because computerized evaluation of HbA1c levels can

enable physicians to identify type 2 diabetes patients with a high risk of adverse

outcomes. Current guidelines for glucose control based on HbA1c do not explicitly

consider hip fracture outcome. The results of the present study suggest that primary

care providers and endocrinology professionals should assist patients with type 2

diabetes on high HbA1c levels to prevent hip fracture.

Limitations and strengths of the study

Our study has several limitations. First, only one HbA1c measurement was

available for most of the patients in the study; therefore, we could not evaluate the

effect that changes in the HbA1c level over time exerts on hip fracture risk. Second,

our study cohort comprised type 2 diabetes patients who were enrolled in the NDCMP.

To evaluate the representativeness of this study cohort, we compared the age and sex

This article is protected by copyright. All rights reserved

Accepted Article
of the patients in our study with those of patients enrolled in the NHI program, i.e.,

the entire type 2 diabetes population; similar distributions were observed. The

nondifferential distributions in age and sex indicated that our findings could be

applied generally to the population with type 2 diabetes in Taiwan. Third, we used an

observational longitudinal study design instead of a randomized control trial. Thus, a

confounding effect possibly existed. However, we adjusted for as many confounders

(such as lifestyle behaviors and complications of stroke, neuropathy, and retinopathy)

as possible, enabling us to minimize the effect of potential confounders. In addition,

we conducted sensitivity analysis by excluding patients with comorbidities. However,

we could not exclude the possibility of residual confounding. Finally, there is

potential misclassification error of HbA1c levels due to inter-laboratory variation and

accuracy (difference between measured and target values). However, there is no

evidence indicating this type of misclassification error was differential. This kind of

non-differential misclassification error would result in underestimation of the effect if

the association between HbA1c and hip fracture exists, indicating that the true effect

would be stronger, a lesser threat to the validity of our finding.

Several strengths were noted. First, the study sample comprised a large number of

patients with type 2 diabetes from a nationwide dataset. Second, the NDCMP used a

standardized procedure to collect data, and information on numerous potential

This article is protected by copyright. All rights reserved

Accepted Article
confounding factors featuring low rates of missing data (3.69%) and loss to follow up

of cohort members was available. Finally, our study sample was obtained from a

national population, contributing to the generalizability of the results.

In conclusion, this study indicated an association between higher levels of HbA1c

and an increased risk of hip fracture among older Chinese adults with type 2 diabetes

in Taiwan. After patients with existing chronic diseases were excluded, this

association remained when we examined the data according to sex, BMI, and insulin

use. The consistency of our findings supports the argument that higher HbA1c levels

increase the risk of hip fracture and demonstrates the importance of glycemic control

for fracture prevention in older adults with type 2 diabetes.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

This study was funded primarily by the Bureau of National Health Insurance

(DOH94-NH-1007), the Ministry of Science and Technology of Taiwan (National

Science Council) (NSC 101-2314-B-039 -017 -MY3 & NSC

102-2314-B-039-005-MY2), China Medical University (CMU103-S-10) and Taiwan

Ministry of Health and Welfare Clinical Trial and Research Center of Excellence

(MOHW103-TDU-B-212-113002).

Authors’ roles: Study design: CCL and TCL. Study conduct: CIL. Data analysis:

SYY and TCL. Data interpretation: CSL, WYL and CCC. Drafting manuscript: CCL

and TCL. Revising manuscript content: CCL and TCL. Approving final version of

manuscript: CIL, CSL, WYL and CCC. TCL takes responsibility for the integrity of

the data analysis.

This article is protected by copyright. All rights reserved

Accepted Article
Reference
1. Ström O, Borgström F, Kanis JA, et al. Osteoporosis: burden, health care
provision and opportunities in the EU: a report prepared in collaboration with the
International Osteoporosis Foundation (IOF) and the European Federation of
Pharmaceutical Industry Associations (EFPIA). Arch Osteoporos.
2011;6(1-2):59-155.
2. Goldacre MJ, Roberts SE, Yeates D. Mortality after admission to hospital with
fractured neck of femur: database study. BMJ. 2002;325(7369):868-9.
3. Boonen S, Autier P, Barette M, Vanderschueren D, Lips P, Haentjens P.
Functional outcome and quality of life following hip fracture in elderly women: a
prospective controlled study. Osteoporos Int. 2004; 15(2):87-94.
4. Vestergaard P, Rejnmark L, Mosekilde L. Increased mortality in patients with a
hip fracture— effect of pre-morbid conditions and post-fracture complications.
Osteoporos Int. 2007; 18(12):1583-93.
5. Robbins J, Aragaki AK, Kooperberg C, et al. Factors associated with 5-year risk
of hip fracture in postmenopausal women. JAMA. 2007;298(20):2389-98.
6. Janghorbani M, Feskanich D, Willett WC, Hu F. Prospective study of diabetes
and risk of hip fracture: the Nurses' Health Study. Diabetes Care.
2006;29(7):1573-8.
7. Khaw KT, Wareham N, Bingham S, Luben R, Welch A, Day N. Association of
hemoglobin A1c with cardiovascular disease and mortality in adults: the
European prospective investigation into cancer in Norfolk. Ann Intern Med
2004;141(6):413–20.
8. Rahbar S. The discovery of glycated hemoglobin. Ann N Y Acad Sci
2005;1043:9–19.
9. Tahara Y, Shima K. Kinetics of HbA1c, glycated albumin, and fructosamine and
analysis of their weight functions against preceding plasma glucose level.
Diabetes Care 1995;18:440–7.
10. Puar TH, Khoo JJ, Cho LW, et al. Association between glycemic control and hip
fracture. J Am Geriatr Soc. 2012;60(8):1493–7.
11. Wu VC, Huang TM, Wu PC, et al. Preoperative Proteinuria Is Associated with
Long-Term Progression to Chronic Dialysis and Mortality after Coronary Artery
Bypass Grafting Surgery. PLoS One 2012;7(1):e27687.
12. National Health Insurance Administration, Ministry of Health and Welfare, ROC.
The National Health Insurance Statistics, 2010. Available from:
http://www.nhi.gov.tw/English/webdata/webdata.aspx?menu=11&menu_id=296
&WD_ID=296&webdata_id=4010.
13. Shiesh SC, Wiedmeyer HM, Kao JT, Vasikaran SD, Lopez JB; Laboratory

This article is protected by copyright. All rights reserved

Accepted Article Management Committee for the Asian-Pacific Federation of Clinical
Biochemistry. Proficiency testing of Hb A1c: a 4-year experience in Taiwan and
the Asian Pacific region. Clin Chem. 2009;55(10):1876-80.
14. Kleinbaum DG, Klein M. Survival analysis: A self-learning text. 2nd ed. Springer;
2005.
15. Ohkubo Y, Kishikawa H, Araki E, et al. Intensive insulin therapy prevents the
progression of diabetic microvascular complications in Japanese patients with
non-insulin-dependent diabetes mellitus: a randomized prospective 6-year study.
Diabetes Res Clin Pract 1995;28:103–17.
16. Gerstein HC, Swedberg K, Carlsson J, et al. The hemoglobin A1c level as a
progressive risk factor for cardiovascular death, hospitalization for heart failure,
or death in patients with chronic heart failure: an analysis of the Candesartan in
Heart failure: Assessment of Reduction in Mortality and Morbidity (CHARM)
program. Arch Intern Med. 2008;168(15):1699–704.
17. Nathan DM, Cleary PA, Backlund JY, et al. Intensive diabetes treatment and
cardiovascular disease in patients with type 1 diabetes. N Engl J Med.
2005;353(25):2643–53.
18. Janghorbani M, Feskanich D, Willett WC, Hu F. Prospective study of diabetes
and risk of hip fracture: the Nurses' Health Study. Diabetes Care
2006;29(7):1573-8.
19. Chen HF, Ho CA, Li CY. Increased risks of hip fracture in diabetic patients of
Taiwan: a population-based study. Diabetes Care. 2008;31(1):75-80.
20. Monami M, Cresci B, Colombini A, et al. Bone fractures and hypoglycemic
treatment in type 2 diabetic patients: A case-control study. Diabetes Care
2008;31:199–203.
21. Nicodemus KK, Folsom AR. Type 1 and type 2 diabetes and incident hip
fractures in postmenopausal women. Diabetes Care 2001;24:1192–7.
22. Oei L, Zillikens MC, Dehghan A, et al. High bone mineral density and fracture
risk in type 2 diabetes as skeletal complications of inadequate glucose control:
the Rotterdam Study. Diabetes Care. 2013;36(6):1619-28.
23. Schneider AL, Williams EK, Brancati FL, Blecker S, Coresh J, Selvin E.
Diabetes and risk of fracture-related hospitalization: the Atherosclerosis Risk in
Communities Study. Diabetes Care. 2013;36(5):1153-8.
24. Schwartz AV, Margolis KL, Sellmeyer DE, et al. Intensive glycemic control is
not associated with fractures or falls in the ACCORD randomized trial. Diabetes
Care. 2012;35(7):1525-31.
25. Bentler SE, Liu L, Obrizan M, et al. The aftermath of hip fracture: Discharge
placement, functional status change, and mortality. Am J Epidemiol

This article is protected by copyright. All rights reserved

Accepted Article 2009;170:1290–9.
26. Brewer LM, Kelly R, Donegan C, Moore AR, Williams D. Poor return of
functional mobility after hip fracture in older patients—it's time to improve on
hip fracture prevention. J Am Geriatr Soc 2011;59:1562–3.
27. Christiansen C, Christensen MS, McNair P, Nielsen B, Madsbad S. Vitamin D
metabolites in diabetic patients: decreased serum concentration of
24,25-dihydroxyvitamin D. Scand J Clin Lab Invest 1982;42:487–91.
28. Nyomba BL, Verhaeghe J, Thomasset M, Lissens W, Bouillon R. Bone mineral
homeostasis in spontaneously diabetic BB rats. I. Abnormal vitamin D
metabolism and impaired active intestinal calcium absorption. Endocrinology
1989;124:565–72.
29. Balint E, Szabo P, Marshall CF, Sprague SM. Glucose-induced inhibition of in
vitro bone mineralization. Bone 2001;28:21–8
30. Alikhani M, Alikhani Z, Boyd C, et al. Advanced glycation end products
stimulate osteoblast apoptosis via the MAP kinase and cytosolic apoptotic
pathways. Bone 2007;40:345–53.
31. Paul RG, Bailey AJ. Glycation of collagen: the basis of its central role in the late
complications of ageing and diabetes. Int J Biochem Cell Biol
1996;28:1297–310.
32. Vashishth D, Gibson GJ, Khoury JI, Schaffler MB, Kimura J, Fyhrie DP.
Influence of nonenzymatic glycation on biomechanical properties of cortical
bone. Bone 2001;28:195–201.
33. Schwartz AV, Hillier TA, Sellmeyer DE, et al. Older women with diabetes have
a higher risk of falls: a prospective study. Diabetes Care 2002;25:1749–54.
34. Ivers RQ, Cummings RG, Mitchell P, Peduto AJ. Diabetes and risk of fracture:
the Blue Mountains Eye Study. Diabetes Care 2001;24:1198–203.
35. Greenspan SL, Myers ER, Kiel DP, Parker RA, Hayes WC, Resnick NM. Fall
direction, bone mineral density, and function: risk factors for hip fracture in frail
nursing home elderly. Am J Med 1998;104:539–45.
36. Wei TS, Hu CH, Wang SH, Hwang KL. Fall characteristics, functional mobility
and bone mineral density as risk factors of hip fracture in the
community-dwelling ambulatory elderly. Osteoporos Int 2001;12:1050–5.

This article is protected by copyright. All rights reserved

Accepted Article
Figure Legends:

Figure 1: Flowchart of recruitment procedures for the current study

Figure 2 Risks of hip fracture according to HbA1c levels
Figure 3 risks of hip fracture for different HbA1c levels stratified by gender, BMI, and
insulin use.

This article is protected by copyright. All rights reserved

Accepted Article Table 1. The comparisons of baseline sociodemographic factors, life style behaviors,
diabetes-related variables, drug-related variables and co-morbidity according to hip
fracture incidence in elders with type 2 diabetes enrolled in the National Diabetes
Care Management Program, Taiwan (n=20,025)
variables Hip fracture
No (N=18,511) Yes (N=1,514) P value
Sociodemographic factors
Gender <0.001
Female 9687 (52.33) 1142 (75.43)
Male 8824 (47.67) 372 (24.57)
Age (years)† 71.92 (4.85) 73.60 (5.38) <0.001
Lifestyle behaviors
Smoking <0.001
No 16501 (89.14) 1397 (92.27)
Yes 2010 (10.86) 117 (7.73)
Alcohol drinking <0.001
No 17554 (94.83) 1476 (97.49)
Yes 957 (5.17) 38 (2.51)
Diabetes-related variables
Duration of diabetes (years)† 8.16 (7.55) 9.21 (7.90) <0.001
Type of hypoglycemic drug use <0.001
No medication 471 (2.54) 29 (1.92)
One oral hypoglycemic drug
Met 1107 (5.98) 57 (3.76)
Su 2327 (12.57) 160 (10.57)
Meg 148 (0.80) 13 (0.86)
Big 75 (0.41) 2 (0.13)
TZD 30 (0.16) 1 (0.07)
Two oral hypoglycemic drugs
Met+Su 6884 (37.19) 512 (33.82)
Met+Meg 284 (1.53) 23 (1.52)
Met+Big 55 (0.30) 6 (0.40)
Met+TZD 51 (0.28) 0 (0.00)
Su+Meg 99 (0.53) 10 (0.66)
Su+TZD 246 (1.33) 34 (2.25)
Su+Big 248 (1.34) 14 (0.92)
Other 61 (0.33) 4 (0.26)
Three oral hypoglycemic drugs 2826 (15.27) 276 (18.23)
>3 oral hypoglycemic drugs 760 (4.11) 77 (5.09)
Insulin 569 (3.07) 55 (3.63)
Insulin+ oral hypoglycemic drug 2270 (12.26) 241 (15.92)
Drug-related variables
Hypertension drug treatment 0.37
No 9933 (53.66) 831 (54.89)
Yes 8578 (46.34) 683 (45.11)
Comorbidity
Obesity 0.002
No 12469 (67.36) 1078 (71.2)

This article is protected by copyright. All rights reserved

Accepted Article
Yes
CAD
No
6042 (32.64)

16171 (87.36)
436 (28.80)

1325 (87.52)
0.89

Yes 2340 (12.64) 189 (12.48)

CHF 1.00
No 17840 (96.38) 1459 (96.37)
Yes 671 (3.62) 55 (3.63)
Stroke <0.001
No 17023 (91.96) 1348 (89.04)
Yes 1488 (8.04) 166 (10.96)
Cancer 0.63
No 18030 (97.4) 1471 (97.16)
Yes 481 (2.60) 43 (2.84)
Hyperlipidemia 0.12
No 14536 (78.53) 1215 (80.25)
Yes 3975 (21.47) 299 (19.75)
Hypertension 0.67
No 8119 (43.86) 655 (43.26)
Yes 10392 (56.14) 859 (56.74)
Atrial fibrillation 0.22
No 18322 (98.98) 1504 (99.34)
Yes 189 (1.02) 10 (0.66)
Chronic hepatitis 0.02
No 17149 (92.64) 1428 (94.32)
Yes 1362 (7.36) 86 (5.68)
COPD 0.85
No 17234 (93.10) 1412 (93.26)
Yes 1277 (6.90) 102 (6.74)
Diabetic retinopathy <0.001
No 13977 (75.51) 1071 (70.74)
Yes 4534 (24.49) 443 (29.26)
Hypoglycemia 0.52
No 18385 (99.32) 1501 (99.14)
Yes 126 (0.68) 13 (0.86)
Peripheral neuropathy <0.001
No 16229 (87.67) 1268 (83.75)
Yes 2282 (12.33) 246 (16.25)
Alcohol dependence 0.55
No 18509 (99.99) 1513 (99.93)
Yes 2 (0.01) 1 (0.07)
Differences in continuous variables were tested using the Student’s t-test. Differences in
categorical variables were tested using the Chi-square test. SU: sulfonylurea; Met: metformin;
Meg: meglitinide; Big: biguanide; TZD: thiazolidinedione; CAD: coronary artery disease; CHF:
congestive heart failure; COPD: chronic obstructive pulmonary disease.

This article is protected by copyright. All rights reserved

 Accepted Article
Table 2 The hazard ratios (HRs) of hip fracture and all-cause mortality according to different HbA1c levels in elders with type 2 diabetes
enrolled in the National Diabetes Care Management Program, Taiwan (n=20,025).
Person- Hip fracture (N=1,514)a
Variables n Cases years IR Age & gender-adjusted Multivariate- Adjusted1 Multivariate- Adjusted2
HbA1c (%)
<6 2096 143 15628.15 9.15 1.15 (0.94-1.40) 1.20(0.98-1.46) 1.19 (0.97-1.45)
6~7 4629 282 35168.78 8.02 1.00 1.00 1.00
7~8 4666 337 35184.63 9.58 1.15 (0.98-1.34) 1.08 (0.93-1.27) 1.07 (0.92-1.25)
8~9 3412 268 25260.85 10.61 1.28 (1.09-1.51)** 1.13 (0.95-1.33) 1.12 (0.94-1.32)
9~10 2231 202 16149.25 12.51 1.51 (1.26-1.80)*** 1.26 (1.04-1.52)* 1.24 (1.02-1.49)*
10 2991 282 20996.77 13.43 1.73 (1.47-2.03)*** 1.33 (1.11-1.60)** 1.32 (1.09-1.58)**
P for trend <0.001 0.03 0.05
HbA1c (%) All-cause mortality (N=6,501)
<6 2096 652 15628.15 41.72 1.03 (0.94, 1.13) 1.08 (0.99, 1.19) 1.07 (0.97, 1.17)
6~7 4629 1380 35168.78 39.24 1.00 1.00 1.00
7~8 4666 1441 35184.63 40.96 1.10 (1.02, 1.18)* 1.03 (0.96, 1.11) 1.03 (0.95, 1.10)
8~9 3412 1127 25260.85 44.61 1.25 (1.15, 1.35)*** 1.09 (1.00, 1.18)* 1.08 (1.00, 1.17)
9~10 2231 770 16149.25 47.68 1.40 (1.28, 1.53)*** 1.15 (1.05, 1.26)** 1.13 (1.03, 1.24)*
10 2991 1131 20996.77 53.87 1.62 (1.50, 1.75)*** 1.23 (1.12, 1.34)*** 1.26 (1.15, 1.37)***
P for trend <0.001 <0.001 <0.001
*:p<0.05; **:P<0.01; ***:p<0.001.
a: Considering all-cause death as competing risk.
Multivariate-adjusted1 for age, gender, smoking, alcohol consumption, duration of diabetes, type of hypoglycemic drugs, hypertension drug

This article is protected by copyright. All rights reserved

 Accepted Article
treatment, obesity and baseline fasting glucose.
Multivariate-adjusted2 for coronary artery disease, congestive heart failure, stoke, cancer, hyperlipidemia, hypertension, atrial fibrillation,
chronic hepatitis, chronic obstructive pulmonary disease, diabetic retinopathy, hypoglycemia, peripheral neuropathy, and, alcohol dependence in
addition to the variables in the first multivariate model.
IR: incidence density rate = number of incident cases / person-years*1000.
HR: hazard ratio.

This article is protected by copyright. All rights reserved

Accepted Article

Figure 1

This article is protected by copyright. All rights reserved

Accepted Article

Figure 2

This article is protected by copyright. All rights reserved

Accepted Article

Figure 3

This article is protected by copyright. All rights reserved