Bone mineral density (BMD) is a measurement of the amount of minerals, primarily calcium and phosphorus, present in an individual’s bones, which is a critical indicator of bone health (1). In Dual Energy X-ray Absorptiometry (DXA), BMD is reported as areal BMD, calculated by dividing the aggregate bone mineral content (BMC) by the scanned area, and expressed in grams per square centimeter (g/cm²). BMD in different populations is determined by numerous factors such as age, gender, ethnicity, race, genetics, nutritional level, lifestyles, medical conditions etc. (2–5).

BMD measurement plays a crucial role in assessing the risk of getting a fracture (6), monitoring the efficacy of the treatments for osteoporosis (7–10), orthopedic surgical planning (11, 12), and evaluating overall bone health (13). Low BMD is associated with an increased likelihood of fragility fractures, particularly in post-menopausal females and older males (14). While osteoporosis and fragility fractures are often emphasized in females, males also face significant risks, with studies indicating that approximately 20–25% of hip fractures globally occur in men, and men have a higher mortality rate following such fractures, compared to women due to greater frailty and comorbidities (15).

The spectrum of BMDs as per the age and gender of a large random sample of a certain community is defined as the normative BMD data of the specific community (16). This data is derived from extensive research, large-scale studies and it summarizes the BMD values considered typical or average for that specific population (16, 17). DXA analysis methods, use reference values from the United States of America or Northern European population by default, causing potential discrepancies and biases when used to interpret BMD of different populations with distinct genetic, geographic, and socioeconomic traits. Currently, DXA scanners in Sri Lanka rely on reference values derived from the National Health and Nutritional Examination Survey (NHANES) database, which predominantly represents Caucasian populations. This reliance introduces significant diagnostic bias, due to ethnic and geographic differences in bone structure and mineralization. Hence Sri Lankans may be misclassified, either underdiagnosed or over diagnosed. Hence to ensure precise diagnosis, it’s imperative to compare an individual’s BMD against the T-score defined for the specific population to which that particular individual belongs.

Population-specific biological and environmental factors may further influence BMD and limit the applicability of non-local reference standards. In this context, approximately 90% of Sri Lankan adults have been reported to have inadequate levels of vitamin D (18, 19), yet vitamin D insufficiency remains a relatively less addressed public health issue in Sri Lanka. As vitamin D plays a critical role in calcium absorption and metabolism, its insufficiency can directly affect BMD, further emphasizing the need for population-specific normative BMD reference values.

Numerous studies have been conducted worldwide to establish population-specific normative BMD data. Most of the studies were aimed at demonstrating the considerable differences in the prevalence of osteoporosis, when population-specific BMD reference data is used instead of the reference values recognized for white populations (13, 20–25). Most of the Sri Lankan community-based studies have been carried out to determine the prevalence of osteoporosis by estimating age-specific BMD. Despite the efforts made by previous research, the Sri Lankan scientific community is still in need of the properly established central BMD normative data. Currently in Sri Lanka, clinicians have to compare the BMD of Sri Lankans against Caucasian reference values for BMD because of the unavailability of normative central BMD data for the Sri Lankan population (26–28). Therefore, there is a great need for normative BMD data for both males and females in Sri Lanka, to provide a scientific basis for optimal bone health management.

This study aimed to establish BMD reference values for the lumbar spine and total hip across different age groups in Sri Lankan adults aged 21 to 80 years. By examining age-, sex-, body mass index (BMI), and menopause-related trends in BMD, the study sought to fill a critical gap in population-specific skeletal health data. Specifically, the objectives were to determine BMD across different age categories for both males and females separately, thereby enabling age specific comparisons and evaluations and to compare BMD values from the current study with existing reference standards.

This large-scale cross-sectional study represents a broad analysis on age- and sex-related changes in BMD among Sri Lankan adults. Men consistently had higher mean lumbar spine and hip BMD compared with women throughout all age groups, similar to NHANES 1999–2004 (31) and 2005–2008 (30), which reported higher BMD in men across multiple skeletal sites. These findings confirm consistent trends across populations of inherently sex-related differences in skeletal mass, likely reflecting variations in bone size and hormonal influences.

In the current study, no significant differences were detected in lumbar spine BMD within females aged 21–50 years of age (p > 0.85; Hedges’ g < 0.1), reflecting stability in early adulthood. The peak lumbar BMD occurred in the 31–40-year age group (0.980 g/cm²) and declined significantly thereafter, with substantial losses beyond age 50 (g > 1.2 between ages 31–40 and 71–80). These results agree with previous studies performed in other populations (21, 24, 32, 33). In Indian populations, Makker et al. (24) and Aggarwal et al. (21) similarly recorded peak lumbar BMD in the 31–40 years of age group, though the BMD remained relatively stable until midlife, followed by a steady decline after menopause. While there was a slight decline in lumbar spine BMD between the 31–40 and 41–50 age groups in this study (mean difference: 0.0103 g/cm²), this change was not statistically significant (p = 0.856; Hedges’ g = 0.08); this may reflect limitations in the ability of cross-sectional studies to detect subtle premenopausal bone loss, which is more readily captured in other longitudinal studies (34, 35). However, the BMD fall from 51 to 80 years was statistically significant, thus indicating accelerated postmenopausal bone loss.

Peak lumbar spine BMD in men occurred at 31–40 years (0.997 g/cm²), similar to Indian data indicating peak values at similar ages (Makker et al. (24); Aggarwal et al. (21)). Despite variation in mean BMD across decades, the overall effect size for Welch’s ANOVA was small (F = 4.73, p < 0.001, η² = 0.006), suggesting relative stability in male lumbar BMD with increasing age. Furthermore, the differences in BMD across age decades at the lumbar spine and hips were more pronounced in women than in men, as the pairwise comparisons demonstrated greater age-group differences in women which is consistent with other longitudinal studies reporting that women experience greater age-related BMD differences compared to men (31, 36, 37). An insignificant increase in male lumbar BMD across the 51–60 to 61–70-year interval (+0.006 g/cm²; p = 0.853; 95% CI = –0.124 to +0.033) would more likely be due to sampling variability than a real physiological trend. However, similar patterns of apparent maintenance or slight increases in BMD have been observed in Taiwanese (38), Singaporean (20), and Austrian (25) cohorts, possibly related to degenerative spinal changes including the formation of osteophytes, joint space narrowing and osteosclerosis in aging men.

In this study, right and left proximal femur BMD showed strong agreement, consistent with cadaveric evidence which found no significant side-to-side differences in geometry, density, or structural rigidity (39). Still, small individual asymmetries in structural stiffness can occur, highlighting the value of assessing both hips in clinical and research settings. Afzelius et al. (40) found slightly lower total hip BMD in the dominant side (p = 0.035), though no difference was seen at the femoral neck (p = 0.754). These findings support measuring BMD in both hips for precise screening and diagnosis, especially in postmenopausal women, to ensure proper preventive and therapeutic interventions.

Among females, hip BMD peaked between 41–50 years, showing significant increases from 31–40 years (right: +0.029 g/cm², p = 0.0096; left: +0.028 g/cm², p = 0.0107). Welch’s ANOVA also confirmed significant differences across age groups for both hips (right: F = 257.14, p < 1.18 × 10^-138, η² = 0.154; left: F = 243.44, p < 1.55 × 10^-133, η² = 0.149). This midlife plateau likely reflects a brief phase of skeletal consolidation before menopausal bone loss, possibly influenced by higher BMI and mechanical loading. After 50 years, BMD declined significantly (right: –0.043 g/cm²; left: –0.045 g/cm² BMD loss between 41–50 and 51–60 years). This pattern is consistent with Korean National Health and Nutrition Examination Survey (41), which demonstrated that Korean women attained maximal BMD in the femoral neck, lumbar spine, and total hip during their 20s, 30s, and 40s, respectively, followed by a rapid decline after 50 years. In addition to that, another Korean population-based study by Cui et al. (42) also demonstrated that peak BMD in women occurred between 40–49 years for the femoral neck and trochanter. On the other hand, the study from the Indian population (24) reported an earlier total hip peak BMD in the 31–40 years age group. It should also be noted that most of the previous studies have analyzed hip subregions such as the femoral neck, trochanter, and Ward’s triangle individually, with the trochanter typically reaching peak BMD later, often in the 40s (42–44). As the present study focused on total hip BMD, future research should include subregional analyses to better characterize site-specific bone trends.

In men, both right and left hip BMD remained stable between 21–50 years, with no significant differences across early to mid-adulthood (all p ≥ 0.97; Hedges’ g ≤ 0.12). Mean differences between 21–30 and 31–40 years were minimal (right: +0.0006 g/cm², p = 1.000; left: –0.0120 g/cm², p = 0.989). Although Welch’s ANOVA indicated overall variation (right: F = 9.13, p < 2.8 × 10^-8; left: F = 11.61, p < 1.4 × 10^-10), post-hoc tests showed, these early adulthood increases were not statistically robust. Peak hip BMD appeared at 51–60 years (right: 0.9848 g/cm²; left: 0.9813 g/cm²), followed by notable declines after 60 years (right: –0.0424 g/cm²; left: –0.0459 g/cm²; both p < 3.6 × 10^-9; Hedges’ g > 0.36), with the lowest BMD observed at 71–80 years. This suggests a delayed BMD peak in Sri Lankan men, with bone loss becoming evident after 60. Most previous studies have shown an earlier peak in male hip BMD at 20–30 years (24, 45) or 31–40 years (21). The later peak observed here may reflect lifestyle, BMI, or physical activity factors that help preserve bone mass into later decades. Previous research also supports a strong positive association between BMI and BMD, linking greater body mass with higher skeletal loading and bone strength (46, 47).

Among women, higher BMI consistently corresponded with greater BMD at all skeletal sites, with noticeable age-related decline only after 50 years. Up to midlife, lumbar and hip BMD remained comparable across BMI categories, whereas post-50 declines were significant. In men, age-related differences across BMI groups were minimal, indicating generally stable BMD patterns. Similar findings have been reported in earlier studies, which found a positive association between BMI and BMD in a nationally representative sample of older adults, with no variation by age, sex, or race (48). Consistent with these findings, Looker et al. (49) noted a significant interaction between race and overweight/obesity status in predicting osteoporosis prevalence among postmenopausal women, highlighting racial disparities in bone health outcomes. However, they did not find a significant age interaction, suggesting that BMI–BMD associations may vary by ethnicity and demographic context rather than age alone.

A significant inverse association was observed between BMD and years since menopause, confirmed by Welch’s ANOVA (F = 223.17, p < 8.82 × 10^-190, η² = 0.186), with menopausal duration accounting for almost 19% of the variance in lumbar BMD and 15 to 16% of hip BMD. Over the initial five postmenopausal years, lumbar spine BMD showed a decline of 0.0166 g/cm² per year, slowing down thereafter at approximately 0.0094 g/cm² per year. Hip BMD declined more gradually, around 0.0043 to 0.005 g/cm² per year during early post menopause, but its decline accelerated later to 0.006 to 0.007 g/cm² per year. The findings indicate well-described physiological patterns wherein trabecular-rich regions, like the spine, show early bone loss, while cortical-rich sites, such as the hip, are characterized by a more gradual but progressive decrease of BMD in later life (50–53).

Contralateral hip comparisons found statistically significant differences only in the youngest age group (21–30 years, p < 0.05), possibly due to early skeletal adaptation, habitual loading, or minor measurement variability. In the elderly, no significant side-to-side difference existed (p > 0.05), which supports earlier research showing strong anatomical symmetry in hip morphology and fracture patterns, especially in older adults (54, 55). From a practical perspective, this symmetry means that measuring just one hip could be sufficient in many cases. Such an approach could make BMD assessments more affordable and accessible, allowing more patients to undergo screening without compromising accuracy, particularly in settings with limited resources.

When compared with the NHANES white reference data, Sri Lankan participants exhibited consistent BMD deficits across all ages and sites, except at the hips of peri- and postmenopausal women, which showed nonsignificant differences. The largest deficits were in lumbar spine BMD of postmenopausal women, reaching up to 15% lower. These findings emphasize the importance of population-specific reference ranges to reduce the misclassification of osteoporosis risk. Similar efforts to establish ethnic-specific BMD databases have been undertaken in Indian (24), Taiwanese (56), Chinese (57), Japanese (58), Spanish (32), Swedish (33), and Italian (59) populations. Notably, Indian women have also been found to have lower mean spinal and hip BMD compared to North American and European reference populations (60, 61). Additionally, the World Health Organization (WHO) and the ISCD, 2019 have recognized the limitations of applying Western thresholds universally, calling for ethnicity-specific reference databases. Our findings also reinforce this recommendation and stress the urgent need for Sri Lankan population specific BMD standards to optimize diagnostic accuracy and treatment decisions.

While robust, the retrospective design and lack of some key lifestyle or biochemical variables, such as diet, vitamin D status, and physical activity, limit causal interpretation. The study focused on the Western Province, which may limit its generalizability to rural populations with different environmental and socioeconomic conditions, despite being justified by its high population density, diverse ethnic representation, and concentration of privately available central DXA facilities. Data were drawn from private healthcare institutions, as most of the DXA scanners are available only in the private healthcare institutions, providing high-quality imaging but potentially introducing selection bias, as individuals undergoing routine or self-referred screening may not represent the general population. Limited availability of DXA in public hospitals further restricts national representativeness. Future studies should incorporate broader geographic coverage once the facilities are more uniformly available, to develop more representative BMD reference values for Sri Lankan adults. Furthermore, in national hospitals, DXA is typically reserved for patients with clinically suspected osteoporosis, resulting in a skewed dataset that does not reflect the general population. Due to these constraints, assembling a large, representative cohort for normative reference development remains challenging. Additionally, the exclusion of subregional hip analyses (e.g., femoral neck, trochanter) limits the depth of clinical insight. Future studies should incorporate subregional skeletal analyses to develop more representative BMD reference values for the Sri Lankan population. Despite these limitations, the large sample size and rigorous exclusion criteria provide a robust foundation for deriving preliminary normative BMD reference values for Sri Lankan adults.

Building on these findings, future research from this study will incorporate advanced machine learning techniques, including random forests and neural networks, to uncover complex, nonlinear relationships that may not be detectable using traditional statistical methods. Additionally, exploring how lifestyle, nutrition, and genetic factors influence bone health could provide deeper insights into population-specific determinants of BMD. Ultimately, these efforts could guide personalized prevention strategies, improve early diagnosis, and strengthen public health initiatives for bone health in Sri Lanka and similar settings.

This study underlines the significant variability in BMD among populations and creates a need for Sri Lankan population specific reference values for the proper diagnosis of osteoporosis, assessment of risks, and planning at the public health level. With a large cohort (n = 10,946) across all adult age groups, and with central skeletal measurements obtained via DXA, this work represents one of the most comprehensive datasets on Sri Lankan bone health, offering greater diagnostic precision than studies using peripheral sites. The findings give added weight to the need for local BMD reference thresholds, rather than Western standards, to avoid diagnostic misclassification. Preventive screening should target middle-aged adults, as early BMD decline begins before advanced age. The strong weight/BMI and BMD association further highlights the importance of optimal nutrition and weight management in maintaining bone health.