Effect of duration of diabetes on bone mineral density: a population study on East Asian males

The Korea National Health and Nutritional Examination Survey (KNHANES) is a nationwide survey representing the non-institutionalized civilian population of South Korea. The Division of Health and Nutritional Survey of the Korea Centers have periodically conducted it for Disease Control and Prevention (KCDCP) since 1998. The data from KNHANES is open data for research purposes. A complex, stratified, multilevel probability sampling design was used, and sampling units were selected based on geographical area, age and sex [12]. Each sampled participant is assigned a numerical sample weight that measures the number of populations represented by that specific participant [12]. A complex sampling design and sample weights facilitate the production of nationally representative data [13]. The KNHANES consists of three components: a health interview, a nutrition survey, and a health examination. Until the first half of 2008, data were gathered through household interviews and direct standardized physical examinations conducted at simple checkup centers at city government offices and town halls [14]. From the second half of 2008 to 2011, the survey was conducted by introducing mobile inspection vehicles. In July 2008, the whole-body dual-energy x-ray absorptiometry (DXA) survey was newly introduced and tested until May 2011. People who were tested before July 2008 and after May 2011 have not received the DXA. From 2008 to 2011, the KNHANES included 37753 individuals. Of the 5872 male participants who were aged ≥50 years, 2489 were eliminated from the study based on the following exclusion criteria: missing data, fasting time of < 8 h, taking prescription medication for osteoporosis, and the questionnaire answered with having been diagnosed and treated by physicians for the conditions affecting bone metabolism, such as all types of cancer, chronic kidney disease, liver cirrhosis, thyroid disease, or rheumatoid arthritis (Fig. 1). After all exclusions, 3383 participants were finally included in this analysis.

As described previously [12], the demographic and behavior variables were age, monthly house income, smoking (never, past, or current smoker), alcohol drinking (grams of alcohol per day), and physical activity (low, moderate, or high). The average alcoholic beverage intake was assessed by self-reported questionnaire and then converted to the amount of pure alcohol (in grams) consumed per day. Physical activity was quantified as the metabolic equivalent of task (MET) minutes per week calculated using the Korean version score calculation of the short-form International Physical Activity Questionnaire. Therefore, physical activity levels were then classified as low (< 600 MET minutes per week), moderate (600 ≤ − < 3000 MET minutes per week), or high (≥3000 MET minutes per week) [12].

Standardized techniques and equipment were used to measure height and weight, and body mass index (BMI) was calculated by dividing body weight by the square of height (kg/m²) [14]. Blood pressure (BP) was measured by a standard method using a sphygmomanometer in a sitting position. The following definition for hypertension was obtained from the Division of Health and Nutritional Survey under the KCDCP: either by a self-reported history of hypertension diagnosis and current usage of antihypertensive drug or by ≥140 mmHg systolic BP or ≥ 90 mmHg diastolic BP.

During the survey, in the morning, blood samples were collected and analyzed by a certified central laboratory. Plasma total cholesterol (milligrams per deciliter), high-density lipoprotein (HDL) cholesterol (milligram per deciliter), triglycerides (milligrams per deciliter) and fasting glucose levels (milligrams per deciliter) were enzymatically measured using a Hitachi Modular 7600 automatic analyzer (Hitachi Ltd., Tokyo, Japan). The fasting insulin (micro-International units per milliliter) and serum vitamin D (nanograms per milliliter) concentrations were measured using immunoradiometric and radioimmunoassays using a gamma-counter (1470 Wizard; PerkinElmer, Waltham, MA, USA). The direct measurement of low-density lipoprotein (LDL) cholesterol was limited, and it was mainly calculated using the Friedewald equation (i.e., LDL cholesterol = total cholesterol – HDL cholesterol – triglyceride/5) [15]. Insulin resistance was evaluated using the homeostasis model assessment: estimated insulin resistance (HOMA-IR) index (i.e., HOMA-IR = [fasting plasma glucose] × [fasting plasma insulin] × 0.055 /22.5) and homeostasis model assessment: estimated beta cell function (HOMA-beta) index (i.e., HOMA-beta = 20 × [fasting plasma insulin / ([fasting plasma glucose] × 0.055–3.5)] [16].

The whole-body dual-energy x-ray absorptiometry (DXA) survey begin in July 2008 and ended in May 2011. The BMDs of the lumbar spine, left total hip, and left femoral neck, and body composition including percent fat mass, were measured using DXA (Discovery-W; Hologic Inc., Waltham, MA, USA). As described previously [14], for the lumbar spine, the mean value of BMDs of L1–L4 was selected for analysis according to the recommendation of the International Society for Clinical Densitometry [17]. The BMD of the right femur was used when the left femur could not be measured (e.g., postoperative state, fracture, deformity, or malformation). The technicians performing the BMD measurements tested the precision based on duplicate or triplicate measurements in 30 or 15 participants, respectively. The precision error was calculated as the percentage coefficient of variation (CV %). The precision errors allowed for the total hip, femur neck, and lumbar vertebrae were 1.8%, 2.5%, and 1.9%, respectively [14].

The following definitions were obtained from the Division of Health and Nutritional Survey of the KCDCP. DM was defined by either a self-reported history of DM diagnosis and current usage of insulin or oral hypoglycemic agents, or by a ≥ 126 mg/dL fasting blood glucose level or ≥ 6.5% glycated hemoglobin (HbA1c) level. Prediabetes was defined as ≥100–125 mg/dL fasting blood glucose, or ≥ 5.7–6.5 HbA1c levels. DM duration was calculated by subtracting the age at the time of DM diagnosis from the age at the time of the investigation. Data without time-diagnosed diabetes were excluded.

Statistical analyses were carried out using STATA 14.0 (StataCorp, College Station, TX, USA) with the SVY commands to account for the complex sampling design and include sample weight, which enabled the results to represent the entire national male population > 50 years [13]. All continuous data were presented as the mean ± standard error (SE). All categorical data were presented as numbers and percentages. To compare the participants’ characteristics of the study among control, groups with prediabetes and DM, the analysis of variance (ANOVA) test and χ² test for continuous and categorical variables, respectively, were used. Multivariable linear regression analyses were performed to determine whether the BMD at each measured site differed based on the DM status and duration. In model 1, we adjusted for major factors, such as age and BMI. Model 2 was also adjusted for alcohol consumption; serum vitamin D concentration; smoking status; triglyceride levels, HDL cholesterol levels and total cholesterol levels, hypertension diagnosis, physical activity level, and HOMA-IR index scores. LDL cholesterol was not included as a covariate in model 2 because a significant collinearity was found between LDL and total cholesterol levels. Bonferroni’s correction was performed for comparisons between the control group and groups with prediabetes and DM. The reported probability values are two-sided, and results with a p-value < 0.05 were considered statistically significant.

Among all participants aged ≥50 years of the 2008–2011 KNHANES, 1037 (29.76%) men had prediabetes, and 644 (18.57%) men had DM. Table 1 shows the baseline characteristics of the control, participants with prediabetes and DM. Patients with diabetes were slightly older than men in other groups. Men who had prediabetes and diabetes were more likely to be overweight or obese have hypertension; have higher fasting glucose levels, HOMA-IR index scores, HDL and LDL cholesterol levels, triglyceride levels, and alcohol consumption; and have lower HOMA-beta index scores, than those in the control group (p < 0.05). The control, prediabetic, and diabetic groups differed in terms of BMD based on the World Health Organization’s criteria.

The BMDs at various measured sites in the control, prediabetic, and diabetic groups were compared. Men with prediabetes and diabetes had higher lumbar spine and total hip BMDs than controls in the crude analysis. The prediabetic group had higher femoral neck BMDs than the control group. No significant differences in three BMD sites were found between the prediabetic and diabetic groups. In model 1, after adjusting for age and BMI, the prediabetic group had higher lumbar spine and total hip BMDs than the control group. After further adjustments for all clinically relevant covariates (model 2), the men with DM and men with prediabetes had higher lumbar spine, total hip, and femoral neck BMDs than controls. Similarly, no significant difference at three BMD sites were found between the prediabetic and diabetic groups in both models 1 and 2 (Table 2).

Table 3 shows the baseline characteristics of the participants with diabetes duration of ≤5 years and those with a disease duration of > 5 years. A total of 505 people were identified for type 2 DM duration: 235 participants with a disease duration of ≤5 years and 270 with longer DM duration. No significant differences between the two groups were found in terms of insulin-associated factors, BMI, fat mass, total cholesterol levels, and HDL and LDL cholesterol levels. However, longer DM duration participants were slightly older with a lower percentage of current smokers and had higher fasting glucose and HbA1c levels than those with shorter DM duration.

The BMDs at various measured sites were compared between the participants with DM duration of ≤5 years and those with disease duration of > 5 years. The group with longer DM durations had lower total hip and femoral neck BMDs than those with shorter DM durations in the crude analysis. The strength of the association between total hip BMD and DM duration declined after further adjustments; however, the group with longer DM durations had lower femoral neck BMDs than those with shorter DM durations in both models 1 and 2 (Table 4).