Participants in National Health and Nutrition Examination Survey (NHANES), a nationally representative survey conducted in 2-year cycles from 1999 to March 2020, provided written informed consent, and the National Center for Health Statistics Research Ethics Review Board approved the study protocols. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE).

Our analysis focused on cycles with complete records in demographics, osteoporosis, body measures, prescription medication, hospital utilization and access to care, and health insurance, as shown in Figure 1. Thus, eight cycles (1999–2000, 2001–2002, 2003–2004, 2005–2006, 2007–2008, 2009–2010, 2013–2014, 2017–March 2020) were included in this study. The study sample was limited to adults aged 50 and older with a definite answer regrading OP diagnosis (yes/no).

Data on prescription medication use, including generic drug name, No. prescription medications, and days taking medications, were obtained from “prescription medication” questionnaires. The individual ingredients of these drugs were extracted from the records under “generic drug name”, as shown in Figure 1. These ingredients were then categorized into therapeutic classes based on the Multum Lexicon Plus drug database (Centers for Disease Control and Prevention CDC, 2007).

Based on the responses to the question “Ever told had osteoporosis/brittle bones (yes/no)”, individuals diagnosed with OP were classified into the OP group, while those who explicitly stated they had not been diagnosed with OP were categorized into the non-OP group. In addition, bone fracture history was determined by aggregating reported cases of hip, wrist, and vertebral fractures.

Clinical and demographic information, including age, sex, body mass index (BMI), race/ethnicity, and self-reported health status, were collected from standardized questionnaires and physical examinations. BMI categories were defined as follows: underweight (<18.5), normal weight (18.5–24.9), overweight (25–29.9), and obese (≥30). Because postmenopausal women with low BMI exhibit osteopenia with predisposition for OP, and fat mass assumes a determining role in predicting the bone mineral density (BMD) of the lumbar vertebrae and femoral neck of postmenopausal women (Wu and Du, 2016). Race/ethnicity analyzed included non-Hispanic white, non-Hispanic black, Hispanic, and Other (including multiple races). This assessment was conducted because of the recognized racial and ethnic disparities in the risk and incidence of OP (Thomas, 2007). Individuals with a confirmed diagnosis may report a lower self-assessment of their health, whereas those who view their health more positively may be underdiagnosed. Therefore, the current health status of the participants was analyzed, self-reported as “excellent or very good”, “good”, or “fair or poor”.

Socioeconomic information, including education level, family income to poverty ratio (PIR), and insurance status, which reflects the medication adherence and accessibility within the population, was collected using standardized questionnaires. The education level of the household head was categorized as less than a high school degree, a high school degree, or higher than a high school degree. PIR was calculated as the ratio of family income to the poverty threshold and categorized as <1, 1–1.9, 2–2.9, 3–3.9, or ≤4. Insurance status was self-reported as either insured (including public and private sources) or uninsured. These characteristics were evaluated to assess the socioeconomic status of households in relation to medication use (Venkatesh et al., 2019; McCabe et al., 2023).

Prevalence rates and 95% confidence intervals (CIs) were reported for categorical variables. The chi-square (χ2) test was used to evaluate the consistency of distributions of categorical covariates between OP and non-OP groups, including the clinical, demographic and socioeconomic characteristics.

Trends of OP diagnosis across the survey cycles and OP prevalence by bone fracture history, No. prescription medications, and days taking prescription medications were calculated using a linear regression model. The combined survey cycle was considered as a continuous variable.

To determine medication strategy influenced by OP diagnosis, logistic regression was used to analyze the association between OP diagnosis and medication ingredient administration. The threshold for statistical significance was set at p < 0.05. Odds ratios (ORs) with 95% CIs were derived from a multivariable logistic regression model to assess the altered medication administration by OP diagnosis. National trends in the prescription of specific medication ingredients associated with OP diagnosis were then outlined using linear regression, adjusting for education level, RIP, and insurance status.

To assess the robustness of the association results, sensitivity analyses were performed by 1) assessing the change in use of each ingredient by OP diagnosis, 2) excluding adults aged ≥80 years or with a BMI ≥30 kg/m² who were more prone to multiple comorbidities.

Data analysis for this study applied rigorous methods tailored to structured survey data, including stratification, clustering, and weighting to ensure nationally representative estimates. SAS software (version 9.4) was used, with statistical significance set at p < 0.05. Python (version 9.3) was used to generate diagrams.

This study identified a final sample size of 21,965 individuals, representing 669,561,146 noninstitutionalized US adults aged 50 and older (Supplementary Table S1). Among them, 2,408 individuals were diagnosed with OP, constituting 11.06% [95% CI, 10.38%–11.75%] of the participants (Table 1). Among OP patients, 24.11% [95% CI, 22.10%–26.11%] had a history of bone fractures, 19.43% (17.70–21.16) were aged 80–89 years, and 30.26% [95% CI, 27.95%–32.58%] were of obesity. OP patients reported similar distributions across health status categories: excellent or very good (33.00% [95% CI, 30.10%–35.90%]), good (34.68% [95% CI, 32.10%–37.26%]), and fair or poor (32.32% [95% CI, 29.46%–35.18%]). OP patients had a higher prevalence of using ≥8 prescription medications compared to non-OP individuals (36.86% [95% CI, 33.96%–39.76%] vs. 23.22% [95% CI, 22.12%–24.32%]), as well as a longer duration of taking prescription medications (64.04% [95% CI, 61.07%–67.00%] vs. 51.76% [95% CI, 50.48%–53.03%]). And the vast majority were insured, with a rate of 96.18% [95% CI, 95.02%–97.34%] in OP group (Supplementary Table S2).

The prevalence of OP diagnosis increased from 9.00% [95% CI, 7.83%–10.17%] in the 1999–2002 cycle to 11.78% [95% CI, 10.31%–13.25%] in the 2003–2006 cycle, decreased to 10.75% [95% CI, 9.44%–12.06%] in the 2007–2010 cycle, plateaued until the 2013–2014 cycle and then increased again to 13.23% [95% CI, 11.57%–14.89%] in the 2017–March 2020 cycle (Figure 2; Supplementary Table S3). The prevalence of OP patients with a history of sustained bone fractures increased from 1.87% [95% CI, 1.43%–2.32%] to 3.84% [95% CI, 3.00%–4.68%].

The prevalence of prescription medication use in the OP diagnosed population was increased, indicated by an rising trend in OP patients using ≥8 medications, rising from 2.48% [95% CI, 1.94%–3.01%] in the 1999–2002 cycle to 5.02% [95% CI, 3.95%–6.90%] in the 2017–March 2020 cycle (p-trend = 0.00), and the increase in the proportion of elderly adults who taking prescription medications for more than 500 days (p-trend <0.0001), from 4.72% [95% CI, 3.86%–5.57%] to 9.86% [95% CI, 8.29%–11.43%].

Significant correlations were found between OP diagnosis and 34 medication ingredients, among the 211 ingredients extracted from the “generic drug name” field, categorized into 22 subcategories across 10 agents (Table 2). And the national trends of these ingredients were explored further (Figure 3; Supplementary Table S4). Of these, 7 ingredients were used sparingly in osteoporotic patients, including an anticonvulsant topiramate (OR 0.18 [95% CI, 0.09–0.37], p < 0.0001), an angiotensin-converting enzyme inhibitor (ACEI) quinapril (OR 0.59 [95% CI, 0.38–0.92], p = 0.02), a nasal decongestant pseudoephedrine (OR 0.26 [95% CI, 0.10–0.69]), p = 0.01), two alpha-blockers tamsulosin (OR 0.37 [95% CI, 0.21–0.67], p = 0.00) and terazosin (OR 0.40 [95% CI, 0.22–0.72], p = 0.00), a 5-alpha reductase (AR) inhibitor finasteride (OR 0.51 [95% CI, 0.27–0.95]), p = 0.04), a non-steroidal anti-inflammatory drug (NSAID) indomethacin (OR 0.15 [95% CI, 0.05–0.47], p = 0.00); and, the remaining 27 ingredients were commonly used in OP cases. Among them, 13 ingredients showed increasing trends, including cyclobenzaprine (p-trend = 0.01), gabapentin (p-trend <0.0001), oxycodone (p-trend = 0.00), losartan (p-trend <0.0001), famotidine (p-trend = 0.01), pantoprazole (p-trend <0.0001), omeprazole (p-trend <0.0001), oxybutynin (p-trend = 0.05), (p-trend <0.0001), tamsulosin (p-trend <0.0001), finasteride (p-trend <0.0001), thyroid desiccated (p-trend <0.0001), levothyroxine (p-trend = 0.00), and meloxicam (p-trend = <0.0001); six showed decreasing trends, including quinapril (p-trend = <0.0001), pseudoephedrine (p-trend = <0.0001), brompheniramine (p-trend = 0.05), terazosin (p-trend = 0.00), raloxifene (p-trend = <0.0001), rofecoxib (p-trend = <0.0001); the prevalence of six ingredients initially increased and decreased in recent years, including carisoprodol (p-trend = 0.00), pregabalin (p-trend <0.0001), topiramate (p-trend <0.0001), lovastatin (p-trend <0.0001), alendronate (p-trend <0.0001), risedronate (p-trend <0.0001); and, the consumption of remind nine ingredients remained stable (p-trend >0.05). Through the 2017–March 2020 cycle, levothyroxine (12.33% [95% CI, 10.54%–14.12%]) in the hormones/hormone modifiers, omeprazole (10.62% [95% CI, 8.86%–12.38%]) in the gastrointestinal agents, and losartan (10.58% [95% CI, 9.05%–12.11%]) in the cardiovascular agents were among the most common OP-related ingredients administrated to 10% of US adults aged 50 and older.

The associations between OP diagnosis and administration of individual drug ingredients remained robust, as the direction and magnitude of most changes in the medication ingredient administration remained in the crude weighted logistic regression model adjusted for a single ingredient (Table 2) and in the multivariable model in a population excluding elderly aged ≥80 years (Supplementary Table S5) or obese individuals (Supplementary Table S6).

Though the OP diagnosis was increased, the use of bisphosphonates (alendronate, risedronate) and raloxifene, the FDA-approved OP drugs (LeBoff et al., 2022), was decreased. This may be due to the emergency of anabolic and catabolic treatments such as denosumab (Prolia, 2010), abaloparatide (Tymlos, 2017), and romosozumab (Evenity, 2019) (Kuril et al., 2024). Except for the OP drugs, the ingredients identified in this study can be classified as follows, based on the clinical reports bonding medical application and OP as well as the national trends in medication prescribed:

• Avoided in OP patients due to their adverse effects on bone formation or bone fracture healing, including topiramate, indomethacin, and rofecoxib (Zheng et al., 2020; Heo et al., 2011).

Experimental studies investigating the role of medication ingredients in regulating bone hemostasis have been emerged. And the association found between OP diagnosis and compounds without prior clinical evidence may underlie their effects on bone remodeling, sympathetic regulation, and gastric acidity.

• Bone remodeling The balance between osteoclastogenesis and osteoblast expressing receptor activator of nuclear factor-kappa B ligand (RANKL) and osteoprotegerin (OPG) is regulated by systemic hormones, such as parathyroid hormone and local signaling molecules (Celebi Torabfam and Porsuk, 2024). Carisoprodol, a skeletal muscle relaxant, is associated with increased OP risk by inhibiting osteoblast differentiation and reducing bone density through inhibiting Wnt/beta-catenin signaling pathway (YRKM, 2022). Similarly, histamine promotes bone resorption by inducing osteoclast formation and increasing RANKL expression in osteoblasts and bone marrow cells (Ng et al., 2022; Biosse-Duplan et al., 2009). And histamine receptor H₁ antagonists (brompheniramine and promethazine) and histamine receptor H₂ antagonist (famotidine) contribute to protect against this (Abra et al., 2013). And the alpha-blocker, tamsulosin, exerts significant anti-osteoporotic effects by inhibiting the activity of transmembrane protein 16A (TMEM16A), which reduces the differentiation and function of osteoclasts, thereby decreasing bone resorption (Li et al., 2025).

Note that, OP increases infection risk and antibiotic use. For example, combination drug amoxicillin/clavulanate is often prescribed (Khan, 2023). Hence, a significant correlation of OP with only one specific anti-infective agent, clavulanate, was observed, which is due to the frequent prescription of amoxicillin in common conditions.

This study emphasizes the importance of a multidisciplinary approach for healthcare professionals in managing OP, particularly among the elderly with multiple health issues. Traditional anticonvulsants contribute to OP (LeBoff et al., 2022), while it has been reported in 2016 that the effect of new antiepileptic drugs such as gabapentin and topiramate on bone metabolism and bone density are scanty and controversial (Arora et al., 2016). The finding of this study revealed that pregabalin and gabapentin were still commonly prescribed in elder patients with OP diagnosis while topiramate was avoided. Moreover, the risk of OP varies with lovastatin dosage, i.e., lower doses (up to 10 mg daily) lows OP risk while higher doses increase the risk (Leutner et al., 2019), suggesting need the for examining cumulative drug exposure. This study underscores the urgent need for researchers to explore the mechanisms of action of medications in their intended conditions and OP development which lack of clinical and experimental supports and significantly associated with OP diagnosis, including terazosin, meloxicam, finasteride, metoclopramide, and cyclobenzaprine.

This study has several limitations. 1) Data of 2011–2012 and 2015–2016 OP questionnaires are not collected, limiting the continuity in national trend analysis from 1999 to March 2020. 2) The cross-sectional design limits the ability to establish causality of disease occurrence and medication intake with OP. It also restricts the evaluation of cumulative drug exposure on OP risk, given that OP is a chronic condition (Foessl et al., 2023). 3) There was a potential bias due to the exclusion of participants with missing data. 4) Though the survey staffs aimed to capture all prescription medication use, underreporting is possible, and data on most over-the-counter medications were not collected. 5) Lack of data on medication adherence, which may affect the interpretation of medication use. 6) Reliance on self-reported data for key variables introduces a risk in the recall and social desirability. 7) Limited fracture history data, focusing on self-reported hip, wrist, and vertebral fractures. 8) Residual confounding may arise from unmeasured variables, including smoking, physical activity, diet, and comorbid conditions.