Trend Analysis of Medically Related Osteonecrosis of the Jaw in the United States Using US National Inpatient Data From 2016 to 2022.

Introduction

Medication‐Related Osteonecrosis of the Jaw (MRONJ) is defined by the American Association of Oral and Maxillofacial Surgeons (AAOMS 2022) as “exposed bone or bone that can be probed through an intra‐ or extra‐oral fistula in the maxillofacial region that has persisted for more than eight weeks in a patient with a history of current or previous treatment with antiresorptive or antiangiogenic agents, and without prior radiation therapy to the jaws or obvious metastatic disease to the jaws” [ 1 , 2 ].

MRONJ is a rare, serious adverse effect of antiresorptive, antiangiogenic, and/or immunomodulating drugs (ARAID's) such as bisphosphonates, denosumab, and antiangiogenic agents used to treat osteoporosis, multiple myeloma, and metastatic cancers. The pathophysiology involves suppression of bone remodeling and vascular compromise, resulting in ischemic necrosis of osseous tissue [ 3 , 4 , 5 ].

Clinical symptoms are pain, tooth mobility, mucosal ulceration, purulent discharge, delayed healing, and sequestration of necrotic bone, leading to secondary infection and progressive destruction of alveolar and periodontal structures [ 6 , 7 ]. These symptoms can severely compromise mastication, speech, and oral hygiene, resulting in diminished nutrition, social distress, and significant reduction in quality of life [ 8 ].

The global incidence and prevalence of MRONJ vary substantially across populations, underlying diseases, therapeutic exposures, and definitions. A nationwide Japanese cohort of 2,819,310 individuals reported a prevalence of 0.06% and an incidence of 22.9 per 100,000 person‐years among osteoporosis patients, compared with a prevalence of 1.47% and 1231.7 per 100,000 person‐years among cancer patients receiving antiresorptive therapy [ 9 ]. In Kure City, Japan, Nashi et al. (2023) reported annual incidence rates of 132.5 per 100,000 persons aged ≥ 40 years for low‐dose regimens and 2305.8 per 100,000 for high‐dose regimens [ 10 ]. A South Korean study analyzed 164,926 bisphosphonate users and found an MRONJ incidence of 20.9 per 100,000 person‐years, with a hazard ratio of 3.72 compared with controls [ 11 ].

In Europe, most data originated from oncology cohorts receiving high dose antiresorptive therapy. A 20‐year multicenter study from Tyrol, Austria, reported an overall 8.8% cumulative incidence of MRONJ among breast cancer patients with bone metastases—2.8% among those treated with bisphosphonates alone, 11.6% among those receiving denosumab, and 16.3% among patients who received sequential bisphosphonate to denosumab therapy [ 12 ]. An Italian multicenter analysis found that MRONJ developed in 5.7% of denosumab‐treated cancer patients with bone metastases at 24 months and 9.8% at 48 months [ 13 ]. In Germany, studies reported MRONJ prevalence rates ranging from 28 to 128 per 100,000 bisphosphonate users, with variability attributed to treatment duration and bisphosphonate formulation [ 14 ]. In Sweden, a national clinic‐based survey estimated the incidence of bisphosphonate‐related osteonecrosis of the jaw (BRONJ) among adults aged ≥ 45 years receiving oral bisphosphonates at 67 cases per 100,000 patient‐years (2007) and 69 cases per 100,000 patient‐years (2008) [ 15 ]. In Australia, a prevalence of 28 per 100,000 patients on oral bisphosphonates and 90 per 100,000 for intravenous bisphosphonates was found [ 16 ].

The data from the United States follows global patterns with lower rates in osteoporosis patients when compared to oncology cohorts. Among osteoporosis patients treated with oral bisphosphonates for ≥ 3 years, the prevalence of MRONJ is approximately 0.10% (95% CI: 0.05%–0.20%) [ 17 ]. In patients receiving intravenous bisphosphonates for osteoporosis exhibit a much lower prevalence of 0.017% [ 2 ]. The 2022 AAOMS position paper estimates MRONJ risk at ≤ 0.05% for bisphosphonate users and ~0.3% for denosumab users in this population [ 2 ]. An international systematic review reported prevalence estimates ranging from 1 to 90 per 100,000 patients for oral bisphosphonates and 1000 to 10,000 per 100,000 for cancer patients receiving intravenous bisphosphonates [ 3 ].

The US Food and Drug Administration's Adverse Event Reporting System (FAERS) from 2010 to 2014 identified 17,119 MRONJ cases reported worldwide, of which approximately 70% originated from the United States. The mean patient age was 62 years, and 69.7% were women. Among the 11,690 cases with documented treatment indications, 59% were associated with therapy for prevention of skeletal‐related events whereas 41% involved osteoporosis treatment [ 18 ]. In oncology populations, pooled analyses of three large Phase III clinical trials involving patients with bone metastases showed cumulative MRONJ risks of 0.5% and 0.8% at 1 year, 1.0% and 1.8% at 2 years, and 1.3% and 1.8% at 3 years for zoledronic acid and denosumab, respectively [ 19 ].

These findings indicate that while MRONJ risk remains below 0.1% in osteoporosis patients, it rises up to 8%–10% or higher in cancer populations exposed to high‐dose or sequential antiresorptive regimens [ 1 ].

While the incidence and prevalence of MRONJ are well documented internationally and in the United States, temporal trends in MRONJ hospitalizations at the national level remain unexplored. Other US studies have relied on the FDA Adverse Event Reporting System, which records voluntary adverse event reports rather than actual hospitalizations. No studies have examined temporal patterns in MRONJ hospitalizations using US National Inpatient Sample (NIS) data.

This study addresses this gap by analyzing the trends in MRONJ hospitalizations from 2016 to 2022 using the US NIS, a 20% sample of all US hospital admissions annually [ 20 , 21 , 22 , 23 , 24 , 25 , 26 ]. Analyses included temporal trends in the incidence, demographics, comorbidities, and hospitalization costs. For each study year, distributions by sex, race, length of stay, hospital charges per day, and geographic regions were analyzed. The study hypothesized that MRONJ hospitalization prevalence would demonstrate significant temporal changes from 2016 to 2022, accompanied by shifts in patient demographics, comorbidity profiles, and healthcare resource utilization patterns. Understanding these trends is essential for identifying healthcare access disparities, evaluating disease management strategies, and informing preventive interventions.

Methods

Study Design, Data Source, IRB Approval

This retrospective cross‐sectional study utilized data from the US NIS Core database for the years 2016–2022. The data is completely deidentified, containing no personal identifiers or hospital‐specific information. Diseases and conditions were identified using International Classification of Diseases, Tenth Revision (ICD‐10) diagnostic codes. This study was approved by the Institutional Review Board at the University of Cincinnati (IRB ID: 2024‐0893) in accordance with the Declaration of Helsinki (1964).

Study Population

Any patient with MRONJ (ICD‐10 code M87.180) listed as their primary discharge diagnosis was extracted from the main NIS dataset and compiled into yearly databases.

Data Processing, Cleaning, and Software

Data rows with non‐finite or nonnumeric values for charges per day were excluded from the final dataset. The yearly datasets (2016–2022) were combined into a single dataset for analyses. The data were analyzed using SAS 9.4 (SAS Institute Inc., Cary, NC) and RStudio version 2025.09.2 (RStudio Team, Boston, MA) [ 27 , 28 ].

Variables and Data Elements

The US NIS dataset comprises variables for patient demographics, hospital characteristics, and financial aspects. Categorical variables were sex, race, hospital division (location of the hospital across nine US geographical regions), annual zip code income quartile, in‐hospital mortality status, elective hospital admission, year of data collection, primary payer, one column of primary diagnosis, and other columns of comorbidity codes (39 for 2016–2022; 29 for 2016). Continuous variables were age, number of diagnoses, number of procedures, length of hospital stay in days, and total charges at discharge in USD (Tables 1 , S1 , and S3 ).

To ensure adequate sample sizes for statistical analysis, certain categorical variables were recategorized into binary groups. These variables have been used throughout the analysis. Race was regrouped as Whites versus non‐Whites (African American, Hispanic, Asian or Pacific Islander, Native American, Other). Primary payer was grouped as Medicare versus non‐Medicare (Medicaid, private insurance, self‐pay, no charge, and other). Since all patients had variable lengths of hospital stay, a new variable, charges per day, defined as follows, Charges per day = total charges at discharge in USD length of hospital stay in days , was created to standardize costs. Some continuous variables (age, total charges, number diagnoses, and procedures) were binarized by calculating the median and grouping the values as low if less than or equal to the median and high if greater than the median, so that these could be included in the analysis of variance (ANOVA) analysis.

Statistical Analysis and Outcome Variables

The analysis was performed on raw sample sizes from the NIS (Table 1 ). Since the NIS dataset is a 20% sample, weighted national estimates and standard errors (SEs) for MRONJ cases for all categorical variables were obtained using the survey design specifications of the NIS (strata, cluster, and discharge weight variables) (Table S1 ). The two major outcome variables were length of hospital stay in days and charges per day in USD (Figures S1 and S2 ). Significant p values were p < 0.05.

Descriptive Statistics

Categorical variables were summarized as frequencies (percentages), and continuous variables were summarized as means ± SEs for each study year (2016–2022) (Tables 1 and 2 ). US National estimates ± SEs for MRONJ cases, including demographics, financial, and hospital characteristics, were calculated annually using methods and SAS codes provided by the NIS. The US national estimates have been analyzed and reported for the total MRONJ cases. The ranges of summaries (mean ± SD) of length of stay in days and charges per day have been reported for each year (Table S1 ).

Prevalence Rate Calculations

Since the NIS MRONJ cases are advanced and need hospitalization and there is no information regarding date of diagnosis, prevalence counts are being analyzed.

Crude Prevalence Rates by Number of Patients Who Took Bisphosphonates

Annual crude prevalence rates were calculated by dividing the MRONJ US national estimate by the corresponding total number of unique patients prescribed bisphosphonates in the United States and multiplying by 100,000 (Table S2 ). Prevalence of MRONJ by number of patients who took bisphosphonates = Number of MRONJ cases Number of patients who took bisphophonates × 100,000

Hospital‐Based Prevalence Rates

Two hospital‐level incidence rates were calculated to provide an overview of MRONJ burden across all inpatient hospitalizations and a clinically relevant incidence within the dental inpatient population:

Crude Prevalence Rate Among Total NIS Admissions

Calculated by dividing the number of MRONJ hospitalization cases by the total size of the NIS dataset for each year and multiplying by 1,000,000 to express incidence per million admissions (Table S2 ). Prevalence of MRONJ among total NIS hospitalization = Number of MRONJ cases Number of dental − related hospitalizations × 1,000,000

Crude Prevalence Rate Among Dental‐Related NIS Admissions

Calculated as the number of MRONJ cases divided by total dental‐related NIS admissions (ICD‐10 codes in the Appendix in the Supporting Information ) for each year, multiplied by 10,000 to yield incidence per 10,000 dental admissions (Tables S2 , S17 ). Prevalence of MRONJ among dental related NIS hospitalizations = Number of MRONJ cases Number of dental − related hospitalizations × 10,000

Trend and Proportion Analyses

Cochran‐Armitage (CA) Trend Tests

The CA trend test (analogous to ANOVA for numerical variables) was used to assess homogeneity and detect significant linear trends in proportions of binary variables (MRONJ National estimates with different populations, sex, race, elective admission status, and primary payer) across the years (Tables 3 and 4 ).

Proportion Tests

Two‐sample proportion tests ( Z tests) were conducted to test equality of proportions between two groups for binary categorical variables (sex, race, elective hospital admissions, primary payer). This was done, both across the entire study period (2016–2022) and within each individual year (Tables 3 and 4 ).

Binomial Tests

Year‐specific one‐sample binomial tests were conducted for binary categorical variables (sex, race, elective hospital admissions, and primary payer) to determine whether the observed proportions differed significantly from the expected value under the null hypothesis ( p = 0.5) (Table 4 ).

Since multiple statistical tests were performed for each variable across and within years, several p ‐values per variable have been generated. Therefore, false discovery rate (FDR) correction was applied to control for multiple testing. Here, only FDR p ‐values < 0.05 were considered statistically significant and the lowest p ‐value for each type of test was reported (Tables 3 and S8 ).

Continuous Outcome Analyses

Two Sample Samples t ‐Tests

Means of numeric outcomes (length of hospital stay in days and charges per day in USD) were compared across levels of binary categorical predictors, both for the combined dataset (2016–2022) and within individual years (Tables S5 and S7 ).

Descriptive Statistics

Mean ± standard deviation was calculated for length of hospital stay and charges per day, stratified by each categorical variable (Tables S4 and S6 ).

Linear Regression Models

Univariate and multivariable linear regression models assessed associations with length of hospital stay (days) and charges per day (USD) for individual years (2016–2022) and combined (Table S18 ). Predictors included sex, race, hospital division, elective admission, primary payer, zip income quartile, mortality (2022 only), year (combined models), age, number of diagnoses, number of procedures, and total charges (length of stay only). All predictors were used in the multiple variable models to adjust for potential confounding, regardless of univariate model significance (Table S10 ).

ANOVA

Multivariable Main Effects ANOVA Models

These were applied to yearly datasets to assess independent associations between continuous outcomes, length of hospital stay in days and charges per day in USD, and categorical predictors (sex, race, hospital division, elective hospital admission, primary payer, zip income quartile, in‐hospital mortality) (for 2022 only), year (in combined year models), factorized age, number of diagnoses, number of procedures, total charges (for length of stay models only). The combined year models examined temporal trends across the study period. All predictors were used to adjust for potential confounding, regardless of univariate model significance (Table S11 ).

Interaction ANOVA Models

These models analyzed whether the combined effects of categorical predictors on length of hospital stay and charges per day differed from the effects of each independent predictor. Interactions were tested only when more than one predictor was significant in main effects ANOVA models. Analyses were conducted separately for each year (2016–2022) and for the combined dataset to check whether these interaction patterns were consistent over time (Table S11 ).

Post Hoc Analysis (Tukey Honestly Significant Difference [ HSD] Tests)

Interaction models showing significant predictors or interactions were analyzed using post hoc Tukey's HSD test to identify specific pairwise differences while controlling the family‐wise error rate. Significant p ‐values and 95% confidence intervals have been reported (Table S11 and Figures S35–S41 ).

Univariate ANOVA Models

These models were used for each predictor variable separately to assess its individual association with length of hospital stay in days and charges per day in USD, respectively. These models were conducted on the yearly and on the combined year dataset (Tables S14 and S15 ).

Effect Size Calculations

Effect sizes were calculated for all ANOVA models to complement ANOVA model p ‐values and evaluate the practical importance of observed associations. Effect sizes were interpreted according to Cohen's conventions [ 29 , 30 ]. Partial eta‐squared ( η 2 p ), partial omega‐squared ( ω 2 ) were reported for multivariable main effects and interaction ANOVA models. Eta‐squared ( η 2 ), omega‐squared ( ω 2 ) were reported for univariate ANOVA. Corresponding 95% confidence intervals and model p ‐values were reported where applicable. This analysis was performed to evaluate statistically significant results and identify those that are not useful in practice (Tables S12–S16 ).

Market Basket Transaction Analysis

Market Basket Analysis was conducted on all ICD‐10 diagnosis codes in MRONJ cases from 2016 to 2022. Frequency tables showing counts (percentages—relative to total MRONJ cases) were reported per year or combined year data. This helps to identify comorbid conditions associated with MRONJ and their trends over time. ICD‐10 codes appearing ≥ 10 transactions per year or ≥ 50 transactions in the combined years dataset are reported (Table S9 ).

Results

The sample sizes of MRONJ cases across the years ranged from 38 (2016) to 40 (2022) (Table 1 ). The US national population estimates with SEs for MRONJ cases ranged from 190 ± 42 (2016) to 200 ± 35 (2022) (Figures 1 , 2 , S1, and S2 and Table S1 ).

The crude incidence rate for MRONJ US National Estimates relative to the number of patients prescribed bisphosphonates increased from 7 (2016) to 13 (2021) and dropped to 10 (2022) per 100,000 individuals (CA, p = < 0.001) [ 21 ] (Tables 3 and 4 ; Figure 3 ). The crude incidence rate for MRONJ sample sizes against the total number of hospital admissions in the NIS dataset ranges from 5 (2016) to 9 (2022) per 1000,000 individuals (CA, p = 0.0371) (Tables 3 and 4 ; Figure 3 ). The crude incidence rate for MRONJ sample sizes against the number of dental‐related hospital admissions ranges from 47 (2016) to 56 (2022) per 10,000 individuals (CA, p = 0.0117). These reveal significant temporal trends in MRONJ incidence (Tables 3 and 4 ; Figures 4 and 5 ).

Sex‐stratified analysis showed that there were more cases among females than males consistently across all years (Table 1 ; Figures S10 , S14 , and S15 ). The proportions were statistically significant in all years except 2021 (Proportion, p = 0.349). There was a statistically significant temporal trend in sex distribution proportions across the years (CA, p = 0.033) (Table 3 ).

Race stratified analysis showed that there were more cases among Whites than non‐Whites consistently across all years (Table 1 ). There was a significant difference in proportions between these two racial groups for all years except 2017, 2018, 2021, and 2022 (Proportion, p > 0.074). There was no change in the trend for White and Non‐White cases over the years (CA, p value = 0.563) (Table 3 ; Figures S9 , S16 , and S17 ).

No single hospital division or income category consistently exhibited the highest or lowest case numbers in all years, and there are more than two levels, trend test is not applicable (Tables 1 , S1 , and S3 ; Figure S12, 20–28, 31–34 ). There was only one death in the cohort in 2022, and the other years showed a 100% survival rate. Therefore, no mortality trend analysis was conducted (Table 1 ).

The number of elective admissions among MRONJ cases consistently exceeded non‐elective admissions across all years. Significant differences in proportions were observed in 2016, 2017, 2019, and 2022. However, there was no statistically significant temporal trend (CA, p = 0.135) (Table 3 ; Figures S8 , S29 , and S30 ).

Across all study years, Medicare‐insured patients consistently outnumbered non‐Medicare patients among MRONJ cases. Significant differences in proportions were observed in all years except 2016 (proportion, p = 0.067). There was no statistically significant temporal trend (CA, p = 0.569 ) (Table 3 ; Figures S11 , S18 , and S19 ).

After applying FDR correction to adjust for multiple hypothesis testing across proportion and binomial tests, the predictors with significant associations were sex (2017, 2019, 2020), race (2016, 2019, 2020), elective hospital admissions (2017, 2019), primary payer (2017–2022). All other variables lost significance after FDR correction, indicating that their associations were not robust when accounting for multiple testing (see Table S8 ).

The average patient age in years ranged from 65 ± 10 (2016) to 70 ± 11 (2022) (Table 2 ; Figure S3A ). The average number of diagnoses ranged from 13 ± 5 (2016) to 15 ± 6 (2022) (Table 2 ; Figure S3 ). The average number of procedures ranged from 3 ± 2 (2016) to 5 ± 4 (2022) (Table 2 ; Figure S5 ). The average length of stay in the hospital in days was approximately 4 ± 3 days (2017) to 8 ± 7 days (2021) with a fall to 5 ± 3 (2022) (Tables 2 and S6 ; Figure S7 ). The average total charges at the time of discharge ranged from $77,880 ± 71,790 (2016) to $133,007 ± 168,325 (2022), with the highest mean recorded in 2021 ($142,201 ± 161,277) (Tables 2 and S4 ; Figure S4 ). The average charges per day followed a similar pattern, ranging from $15,490 ± 16,118 (2018) to $28,016 ± 28,762 (2022), showing an overall upward trend across 2016–2022 (Tables 2 and S4 ; Figures S2 and S6 ). Intermediate years show values between these ranges.

The summary of charges per day across categorical variables (Table S4 ) showed that, among the sexes, males showed an increase from $14,074 ± 10,609 (2016) to a peak of $32,629 ± 39,351 (2020), followed by a decline to $21,904 ± 12,377 (2021) and a rise again to $29,589 ± 35,370 (2022). Female charges ranged from $13,144 ± 6511 (2018) to $28,771 ± 20,148 (2017). Throughout most years, males exhibited higher average charges per day than females.

Across hospital divisions, New England consistently recorded the highest charges, increasing from $7740 ± 2873 (2018) to a peak of $50,942 ± 44,826 (2020), while the West North Central region exhibited the lowest, ranging from $0 ± 0 (2016) (due to an absence of cases) to $8295 ± 4452 (2021).

For admission type, elective admissions showed higher daily charges overall, ranging from $20,891 ± 13,228 (2016) to $33,921 ± 32,509 (2022). Non‐elective admissions exhibited lower and more variable daily charges, ranging from $12,311 ± 7188 (2018) to $22,836 ± 38,154 (2016) and reached $17,471 ± 16,740 (2022) (Tables S4 and S5 ).

Race stratification showed that White patients had higher total hospital costs and daily charges across all years. Among non‐Whites, daily charges ranged from $17,657 ± 8376 (2018) to $28,213 ± 20,171 (2017), while Whites ranged from $11,852 ± 2398 (2016) to $33,135 ± 20,864 (2019). The total charges at discharge also showed that with Whites ranging from $77,880 ± 71,790 (2016) to $142,201 ± 161,277 (2021), and non‐Whites from $82,250 ± 59,184 (2017) to $133,007 ± 168,325 (2022) (Tables S4 and S5 ). Racial differences in total hospital charges and daily charges did not exhibit a consistent increasing or decreasing trend across the study period.

No income bracket showed consistently the highest or lowest charges. The $1–$24,999 group had charges of $34,809 ± 40,796 in 2016, declining to $15,803 ± 10,688 in 2021, then rising to $34,094 ± 22,394 in 2022. Middle income brackets showed fluctuations; the > $45,000 group peaked at $39,165 ± 52,821 in 2022.

Primary payer shows that Medicare showed higher daily charges consistently throughout all years, when compared to non‐Medicare, with both groups showing fluctuations. Daily charges for Medicare ranged from $15,683 (2016) to $29,334 (2017), whereas non‐Medicare cases ranged from $23,713 (2021) to $29,867 (2016) (Tables 4 , S4 , and S5 ).

The summary of length of hospital stay across categorical variables (Table S6 ) showed that among the sexes, the average length of stay (LOS) for males ranged from 3 ± 2 days (2017) to 7 ± 8 days (2021), while for females, it ranged from 3 ± 2 days (2016) to 6 ± 3 days (2022), indicating longer hospital stay for males.

Hospital division and income quartile trends showed that there was no single region or income group that consistently exhibited the highest, lowest, or a clear increasing or decreasing trend across all years from 2016 to 2022.

Elective admissions showed an average length of stay ranging from 4 ± 3 days (2016) to a peak of 8 ± 7 days (2021), followed by a slight decline to 6 ± 3 days (2022). Nonelective admissions ranged from 3 ± 2 days (2017) to a peak of 8 ± 8 days (2021), also declining to 5 ± 3 days (2022).

Surviving patients showed an increase in average hospital stay from 4 ± 3 days (2016) to a peak of 8 ± 7 days (2021), followed by a decline to 5 ± 3 days (2022). The single patient death recorded in 2022 had a hospital stay of 10 days.

Among racial groups, the average length of hospital stay for White patients ranged from 3 ± 3 days (2017) to 7 ± 6 days (2018), while non‐White patients had longer stays ranging from 4 ± 3 days (2017) to a peak of 9 ± 8 days (2021), followed by a decline to 6 ± 3 days (2022).

Medicare recipients had a longer average length of hospital stay than non‐Medicare patients, ranging from 3 ± 3 days (2017) to 9 ± 9 days (2021), whereas non‐Medicare patients ranged from 4 ± 2 days (2017) to 7 ± 7 days (2021) (Tables 4 , S6 , and S7 ).

The univariate Welch two‐sample t ‐tests for length of stay showed significant predictors including the number of diagnoses (2021, 2022) and procedures (2017, 2018). For charges per day, significant predictors were race (2016), elective admission (2021), and total charges (2018–2022) (Table S5 ).

The multivariable linear regression analysis for both outcomes, charges per day and length of stay, no single variable remained consistently significant throughout all years. The results revealed that for charges per day, significant predictors were hospital division (2017, 2019, 2020, 2021), sex (2018, 2020), income quartile (2020–2021), and elective admission (2021). In the model with data from all years combined, elective admission and primary payer (Medicare) were significant. For length of hospital stay, significant predictors were hospital division (2017, 2019, 2020, 2021), sex (2018, 2020), and income quartile (2020–2021). In the model with data from all years combined, elective admission, number of diagnoses, number of procedures, and total charges were most significant (Tables S10 and S18 ).

Hospital Length of Stay

ANOVA main effects models without interaction terms showed varying significant predictors across the years. For length of stay in hospital as outcome, some years had one or no significant predictors (2016, 2021). Therefore, no interaction or Tukey analysis was performed. Other years had multiple significant predictors with hospital division, primary payer, total charges (2017); primary payer, total charges (2018); total charges, number of diagnoses (2019); elective admission status, total charges, number of diagnoses (2020); primary payer, income quartile and number of diagnoses (2021); total charges, number of diagnoses, number of procedures, and year (combined years dataset) (Table S11 ).

ANOVA interaction models for length of stay were not performed for some years due to insufficient significant main effect predictors (2016, 2021). Some years showed either significant predictors or interactions such as total charges, number of diagnoses (2017,2019, 2022); primary payer, total charges (2018); elective hospitalization, total charges, number of procedures (2020), total charges, number of diagnoses, number of procedures, year (combined years dataset) (Table S11 ).

Tukey HSD pairwise analysis showed that for 2016, 2021, no analysis was performed as there was no interaction model. In 2017, 2020, there were no significant differences. Other years showed total charges (2018); total charges, number of diagnoses, interaction of low total charges and low diagnoses vs. high total charges and high diagnoses (2019); primary payer and number of diagnoses (2022) (Table S11 ).

ANOVA effect size analyses: partial eta‐squared ( η 2 p ) for multivariable ANOVA main effects models identified hospital division ( η 2 p = 0.16–0.63; 2016–2019, 2021–2022) and total charges at discharge ( η 2 p = 0.24–0.52; 2016–2019, 2021, combined years dataset) as the largest predictors of length of stay after controlling for other predictors. Demographic variables (sex, race, age) consistently demonstrated the smallest partial eta‐squared values ( η 2 p ≤ 0.17) across all study years.

Partial omega‐squared ( ω 2 p ) analyses which are less biased population‐level effect estimates also showed total charges at discharge (2016–2019, 2021, combined years dataset) as the predictor with largest effects ( ω 2 p p = 0.12–0.30). Demographic variables (sex, race, age) and mortality status again exhibited the smallest partial omega‐squared values (partial ω 2 ≤ 0.09) across all study years (Table S12 ).

Partial eta‐squared ( η 2 p ) and omega‐squared ( ω 2 p ) analyses for two‐way interaction ANOVA models identified hospital division ( η 2 p = 0.16–0.56; ω 2 p = 0–0.22; 2017–2020, 2022) and total charges at discharge ( η 2 p = 0.21–0.52; ω 2 p p = 0.05–0.28; 2017–2020, 2022, combined dataset) as the largest main effect predictors. Socioeconomic variables showed moderate‐to‐large effects: income quartile ( η 2 p = 0.22–0.48; ω 2 p = 0.05–0.16; 2017–2022) and primary payer in select years ( η 2 p = 0.30–0.41; ω 2 p = 0.12–0.15; 2018, 2022). Demographic factors (sex, race, age) consistently demonstrated negligible effects ( η 2 p , ω 2 p ≤ 0.06) across all years (Table S16 ).

For univariate ANOVA models (length of stay ~ variable), eta‐squared ( η 2 ) effect size analyses showed that total charges at discharge ( η 2 = 0.12–0.32; 2016–2022, combined dataset), number of diagnoses, number of procedures ( η 2 = 0.15–0.35; 2017–2022, combined dataset) as the largest predictors of length of stay. The predictors with lowest effects were sex, race, age, hospital division and primary payer ( η 2 ≤ 0.06–0.07) across all years. Omega‐squared ( ω 2 ) values showed that total charges at discharge was the largest predictor ( ω 2 = 0.11–0.29; 2016–2022, combined dataset), number of diagnoses and number of procedures ( ω 2 = 0.12–0.32). All other variables consistently show small or negligible values ( ω 2 ≤ 0.04) across all years (Table S14 ).

Charges Per Day

Main effects ANOVA models without interaction terms for charges per day (CHGPRDAY) showed that some years had no or a single significant predictor (2016, 2017, 2019, 2020, 2021, 2022), and therefore no interaction or Tukey analyses were performed. Significant predictors are sex, hospital division (2018), elective admission, and primary payer (combined years dataset).

Interaction models for charges per day were not performed for 2016, 2017, 2019, 2020, 2021, and 2022 due to no or one significant predictor. The interaction model in 2018 showed extreme significance for all predictors and interactions with only 22 residual degrees of freedom, indicating overfitting and excessive parameter‐to‐observation ratio that limits the reliability of these findings. The combined years dataset showed elective hospital admission and primary payer as significant predictors. The Tukey HSD was not performed due to an extremely overfitted model.

Effect size analyses using partial eta‐squared ( η 2 p ) for multivariable ANOVA main effects models predicting charges per day identified hospital division ( η 2 p = 0.11–0.59; 2016–2022), elective admission status ( η 2 p = 0.16–0.47; 2018, 2021–2022), and income quartile (partial η 2 = 0.13–0.31; 2016–2018, 2020–2021) as the largest predictors. Other variables demonstrated predominantly small to negligible effects ( η 2 p ≤ 0.09) except for sex in 2018 ( η 2 p = 0.42). Partial omega‐squared ( ω 2 p ) analyses revealed hospital division ( ω 2 p = 0–0.30) and elective admission status ( ω 2 p = 0.04–0.37) showed the largest effects, with peak values in 2018 and 2021. Demographic variables and mortality status consistently exhibited the smallest partial omega‐squared values ( ω 2 p ≤ 0.06) across all years (Table S13 ).

For univariate ANOVA models (charges per day ~ variable), eta‐squared ( η 2 ) effect size analyses showed that primary payer had the strongest effects (2016, 2018, combined years; η 2 = 0.11–0.787; ω 2 = 0.10–0.762), elective admission status (2018 and 2021–2022; η 2 = 0.152–0.213; ω 2 = 0.034–0.196). No variable displayed large effects across more than 3 years. Other predictors consistently produced negligible values ( η 2 , ω 2 ≤ 0.05) (Table S15 ).

Partial eta‐squared ( η 2 p ) and omega‐squared ( ω 2 p ) analyses for two‐way interaction ANOVA models for charges per day, for 2018, the interaction model had overfitting, with almost all predictors showing very large η 2 p and ω 2 values across most predictors (up to η 2 p = 0.93, ω 2 p = 0.81). The combined year model showed elective admission status and hospital division ( η 2 p ≤ 0.06, ω 2 p ≤ 0.05) as the largest predictors. All other predictors had negligible effects.

The market basket analysis showed that the most associated co‐morbidity across all years was essential (primary) hypertension (ICD‐10 = I10) with the highest frequency (except in 2018), present in almost 50% of cases. Other frequent comorbidities were unspecified hyperlipidemia (E78.5), inflammatory conditions of jaws (M27.2), and personal history of nicotine dependence (Z87.891), contact with and (suspected) exposure to COVID‐19 (Z20.822, 2021–2022) (Table S9 ).

Discussion

This study represents an initial foray into conducting a comprehensive temporal analysis of MRONJ hospitalizations using US National hospital admissions data from 2016 to 2022. Other studies relied on voluntary adverse event reports limited to shorter times [ 31 , 32 ]. This investigation utilized hospitalization data analyzed with multivariable linear regression, ANOVA with interaction terms, effect size calculations, and FDR correction for multiple comparisons. This paper contains scientific evidence and all the results are written based on the sound evidence.

Interpretation of Lower US Incidence Rates

The observed prevalence of 7–13 per 100,000 patients who were prescribed bisphosphonates in the United States is lower than international reports, which have reported rates ranging from 21–28 per 100,000 in Australian population‐based studies, 67–69 per 100,000 in Sweden and 28–128 per 100,000 in German cohorts, which maybe due to yearly and methodological differences [ 14 , 15 , 16 ].

The NIS data includes only the most severe, advanced MRONJ cases requiring inpatient hospitalization and surgical intervention, thereby excluding the vast majority of MRONJ cases that are routinely managed in outpatient dental and oral surgery settings [ 2 ]. A scoping review reported mean MRONJ rates as 6.22% (oncology) versus 0.58% (osteoporosis), indicating treatment differences [ 29 , 32 ].

The actual US MRONJ incidence and prevalence would follow international estimates when outpatient cases are additionally considered. Hospital administrative billing data depend on clinician recognition and accurate ICD‐10 coding, which can result in underreporting of cases compared to active surveillance with prospective clinical examination used in other studies; this study addresses multiple risk exposures—osteoporosis (≤ 0.1% risk) and cancer patients receiving high‐dose therapy (8%–10% risk) rather than just one underlying cause [ 2 , 9 , 12 ].

The increases in US hospital‐based prevalence over the study period (5–9 per million total admissions, p = 0.037; 47–56 per 10,000 dental admissions, CA, p = 0.012) suggest increased severe MRONJ requiring hospitalization, ARAIDs use, aging demographics, or improved clinical recognition and coding. The FAERS database analysis (2010–2023) reported over 22,000 MRONJ cases, peaking in 2014–2018 [ 33 ].

In this study, females had a higher rate of MRONJ hospitalizations consistently across all years, suggesting higher use of antiresorptive medications for postmenopausal osteoporosis (CA, p = 0.033) [ 2 , 17 ]. The FAERS analysis results matched the results in this study (females: 60.3%, mean age 67.33). In this study, males demonstrated higher charges per day ($14,074–$32,629) and longer hospital stays (3–8 days), suggesting important sex‐based differences in physiology, disease severity, complications, or treatment intensity [ 31 ]. White patients consistently had more cases, higher daily charges, and longer hospital stays than non‐White patients, likely reflecting the demographic composition of the United States [ 3 , 31 ]. The predominance of Medicare as the primary payer and the average age range in the late 60s to early 70s confirm the concentration of disease in older adults with multiple chronic conditions. However, effect size analyses revealed that demographic factors such as sex, race, and age had minimal practical influence on length of stay or charges after adjustment for clinical and institutional variables [ 4 ].

Healthcare Utilization and Institutional Variation

The multivariable effect size analyses indicated that both length of hospital stay and daily hospital charges among MRONJ inpatients were most strongly influenced by institutional factors and overall clinical complexity. Total charges, hospital division, payer status, institutional and clinical complexity factors exerting the largest effects (partial η 2 p = 0.24–0.63, partial ω 2 p = 0.05–0.36). Patients who had a higher number of procedures and diagnoses, and total charges experienced longer hospital stays. Elective admissions showed higher daily charges, possibly due to resource‐intensive planned interventions. Regional variation in outcomes suggests heterogeneity in hospital practice patterns, availability of specialty services, and potentially coding differences, supporting the need for standardized care pathways and consensus protocols to optimize inpatient management and resource allocation [ 3 , 5 ]. Guidelines from AAOMS (2022), MASCC/ISOO/ASCO, and Italian SIPMO‐SICMF emphasize multidisciplinary care coordination, yet implementation remains variable [ 2 , 34 , 35 ]. Total hospital charges increased 71% from 2016 to 2022, with charges per day rising 81%, indicating increasing case complexity, adoption of advanced surgical techniques or adjunctive therapies, or broader healthcare cost trends.

Comorbidity Patterns and Pathophysiological Considerations

The study shows frequent co‐occurrence of cardiovascular comorbidities like essential hypertension in approximately 50% of cases and hyperlipidemia agrees with MRONJ pathophysiology involving compromised bone vascularity [ 6 ]. Bisphosphonates inhibit osteoclast function and suppress bone remodeling, while anti‐angiogenic agents impair neovascularization, which is critical for wound healing and bone repair. In these patients, the bone's capacity to respond to injury or infection is diminished, heightening susceptibility to MRONJ [ 4 , 7 ]. The high burden of multiple chronic conditions in this hospitalized MRONJ cohort (average 13–17 diagnoses) confirms the substantial number of comorbidities typical of affected patients. To prevent MRONJ in this population, guidelines from AAOMS, MASCC, SIPMO‐SICMF, etc. emphasize pretreatment dental screening, extraction of teeth with poor prognosis, optimization of periodontal health, and patient education regarding oral hygiene and risk factor modification [ 2 , 34 , 35 , 36 , 37 , 38 , 39 , 40 ].

Methodological Considerations

This study's strengths are its large, US nationally representative sample over 7 years. Detailed analytical methods such as multivariable modeling, interaction analyses, effect size reporting, and multiple testing correction, temporal trends, demographic patterns, and economic burdens. This study evaluates both length of hospital stay and daily charges allowing assessment of healthcare resource utilization and economic burden. The study examines actual hospitalizations documented through administrative claims rather than voluntary adverse event reports.

Several limitations are that the study includes only severe inpatient hospitalization cases, excluding the substantial proportion of mild–moderate MRONJ managed in outpatient settings. This means the study findings focus on complicated MRONJ requiring hospitalization. The administrative billing data introduces potential underreporting and misclassification compared to prospective clinical surveillance. Absence of clinical MRONJ staging, specific medications, dosing regimens, surgical interventions, and patient outcomes are unavailable. The study cannot establish causal relationships due to retrospective design. Geographic variations may reflect differences in patient populations, disease severity, or healthcare system characteristics.

Conclusions

This trend analysis reveals an increasing burden of severe MRONJ hospitalizations in the United States, characterized by substantial cost escalation, geographic variation, and institutional rather than demographic determinants of resource utilization. The lower population‐level incidence compared to international cohort studies reflects methodological differences—restriction to severe hospitalized cases, total population denominators, and passive administrative case ascertainment—rather than fundamental disease frequency variation. Significant increases over time in hospital‐based incidence, along with rising costs, show a growing public health and economic burden. These findings provide essential baseline data for monitoring MRONJ trends, evaluating preventive interventions, and informing evidence‐based clinical practice and health policy decisions.

Funding

The authors have nothing to report.

Ethics Statement

The Institutional Review Board at the University of Cincinnati had determined that the study is not research involving human subjects as defined by DHHS, DOJ, and FDA regulations on 11/01/2024.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: jphd70040‐sup‐0001‐Supinfo1.docx.

Acknowledgments

We extend our deepest gratitude to all the authors and contributors who dedicated their time, expertise, and effort to this manuscript. Their commitment to rigorous research methods and insightful analysis has been instrumental in shaping the content of this study.

We thank Dr. M.B. Rao and Dr. D.G. Krishnan for providing their expertise and guidance to the conception, data acquisition, design, and drafting and critically reviewing the manuscript. We thank Nisha Sheshashayee for contributing to the conception, design, data acquisition and analysis, interpretation of the results, and drafting the manuscript. We thank Eric Holmes for contributing to conception, design, data acquisition, drafting, editing, and critically reviewing the manuscript.

We also thank the Healthcare Cost and Utilization Project (HCUP) and the Agency for Healthcare Research and Quality (AHRQ) for providing access to the National Inpatient Sample (NIS) dataset, which was used to source the data for the study.

The collective effort of everyone involved has made this work possible and we sincerely appreciate all the hard work that has resulted in this manuscript. All authors gave their final approval and agree to be accountable for all aspects of the work.

Data Availability Statement

The data that support the findings of this study are available from the Healthcare Cost and Utilization Project (HCUP). Restrictions apply to the availability of these data, which were used under license for this study. Data are available at https://hcup‐us.ahrq.gov/nisoverview.jsp#purchase with the permission of the HCUP.

References

1 A. Cuozzo , V. Iorio‐Siciliano , E. Vaia , L. Mauriello , A. Blasi , and L. Ramaglia , “ Incidence and Risk Factors Associated to Medication‐Related Osteo Necrosis of the Jaw (MRONJ) in Patients With Osteoporosis After Tooth Extractions ,” Journal of Stomatology, Oral and Maxillofacial Surgery 123 , no. 6 ( 2022 ): 616 – 621 , 10.1016/j.jormas.2022.03.020 . 35609780

2 S. L. Ruggiero , T. B. Dodson , T. Aghaloo , E. R. Carlson , B. B. Ward , and D. Kademani , “ American Association of Oral and Maxillofacial Surgeons' Position Paper on Medication‐Related Osteonecrosis of the Jaws—2022 Update ,” Journal of Oral and Maxillofacial Surgery 80 , no. 5 ( 2022 ): 920 – 943 , 10.1016/j.joms.2022.02.008 . 35300956

3 A. A. Khan , A. Morrison , D. A. Hanley , et al., “ Diagnosis and Management of Osteonecrosis of the Jaw: A Systematic Review and International Consensus ,” Journal of Bone and Mineral Research 30 , no. 1 ( 2015 ): 3 – 23 , 10.1002/jbmr.2405 . 25414052

4 T. L. Aghaloo , A. L. Felsenfeld , and S. Tetradis , “ Osteonecrosis of the Jaw in a Patient on Denosumab ,” Journal of Oral and Maxillofacial Surgery 68 , no. 5 ( 2010 ): 959 – 963 , 10.1016/j.joms.2009.10.010 . 20149510 PMC2880179

5 W. S. McLaurin , B. J. Francisco , K. J. Hooker , et al., “ Antiresorptive and Anti‐Angiogenic Drug Therapy in the Pediatric Population With Reference to Medication‐Related Osteonecrosis of the Jaw ,” International Journal of Oral and Maxillofacial Surgery 53 , no. 6 ( 2024 ): 496 – 502 , 10.1016/j.ijom.2023.12.004 . 38030483

6 M. R. Allen and D. B. Burr , “ The Pathogenesis of Bisphosphonate‐Related Osteonecrosis of the Jaw: So Many Hypotheses, So Few Data ,” Journal of Oral and Maxillofacial Surgery 67 , no. 5 Suppl ( 2009 ): 61 – 70 , 10.1016/j.joms.2009.01.007 . 19371816

7 C. A. Migliorati , J. B. Epstein , E. Abt , and J. R. Berenson , “ Osteonecrosis of the Jaw and Bisphosphonates in Cancer: A Narrative Review ,” Nature Reviews Endocrinology 7 , no. 1 ( 2011 ): 34 – 42 , 10.1038/nrendo.2010.195 . 21079615

8 I. R. Suryani , I. Ahmadzai , S. Shujaat , et al., “ Non‐Antiresorptive Drugs Associated With the Development of Medication‐Related Osteonecrosis of the Jaw: A Systematic Review and Meta‐Analysis ,” Clinical Oral Investigations 26 , no. 4 ( 2022 ): 2269 – 2279 , 10.1007/s00784-021-04331-7 . 35013781

9 M. Ishimaru , S. Ono , K. Morita , H. Matsui , Y. Hagiwara , and H. Yasunaga , “ Prevalence, Incidence Rate, and Risk Factors of Medication‐Related Osteonecrosis of the Jaw in Patients With Osteoporosis and Cancer: A Nationwide Population‐Based Study in Japan ,” Journal of Oral and Maxillofacial Surgery 80 , no. 4 ( 2022 ): 714 – 727 , 10.1016/j.joms.2021.12.007 . 35033505

10 T. Kunihara , H. Tohmori , M. Tsukamoto , et al., “ Incidence and trend of antiresorptive agent‐related osteonecrosis of the jaw from 2016 to 2020 in Kure, Japan ,” Osteoporosis International 34 , no. 6 ( 2023 ): 1101 – 1109 . [Correction added on 28 February 2026, after first online publication: Reference 10 has been updated for accuracy in this version. The authors were listed as ‘M. Nashi, K. Fujikawa, N. Okimoto, H. Kishimoto, and A. Sato’ in the originally published version, which was incorrect.] 37002373 10.1007/s00198-023-06732-8 PMC10203006

11 S. H. Kim , Y. K. Lee , T. Y. Kim , Y. D. Kwon , and Y. Rhee , “ Incidence of and Risk for Osteonecrosis of the Jaw in Korean Osteoporosis Patients Treated With Bisphosphonates: A Nationwide Cohort Study ,” Bone 143 ( 2021 ): 115650 , 10.1016/j.bone.2020.115650 . 32956854

12 C. Brunner , M. Arvandi , C. Marth , et al., “ Incidence of Medication‐Related Osteonecrosis of the Jaw in Patients With Breast Cancer During a 20‐Year Follow‐Up: A Population‐Based Multicenter Retrospective Study ,” Journal of Clinical Oncology ( 2024 ). 10.1200/JCO.24.00171 PMC11708989 39163561

13 P. Bracchi , E. Zecca , C. Brunelli , et al., “ A Real‐World Study on the Prevalence and Risk Factors of Medication Related Osteonecrosis of the Jaw in Cancer Patients With Bone Metastases Treated With Denosumab ,” Cancer Medicine 12 , no. 17 ( 2023 ): 18317 – 18326 , 10.1002/cam4.6429 . 37559413 PMC10523953

14 S. Otto , C. Pautke , C. Opelz , et al., “ Osteonecrosis of the Jaw: Effect of Bisphosphonate Type, Local Concentration, and Acidic Milieu on the Pathomechanism ,” Journal of Oral and Maxillofacial Surgery 68 , no. 11 ( 2010 ): 2837 – 2845 , 10.1016/j.joms.2010.07.017 . 20971371

15 M. Ulmner , F. Jarnbring , and O. Törring , “ Osteonecrosis of the Jaw in Sweden Associated With the Oral Use of Bisphosphonate ,” Journal of Oral and Maxillofacial Surgery 72 , no. 1 ( 2014 ): 76 – 82 . 23992777 10.1016/j.joms.2013.06.221

16 T. Mavrokokki , A. Cheng , B. Stein , and A. Goss , “ Nature and Frequency of Bisphosphonate‐Associated Osteonecrosis of the Jaws in Australia ,” Journal of Oral and Maxillofacial Surgery 65 , no. 3 ( 2007 ): 415 – 423 , 10.1016/j.joms.2006.10.061 . 17307586

17 J. C. Lo , F. S. O'Ryan , N. P. Gordon , et al., “ Prevalence of Osteonecrosis of the Jaw in Patients With Oral Bisphosphonate Exposure ,” Journal of Oral and Maxillofacial Surgery 68 , no. 2 ( 2010 ): 243 – 253 , 10.1016/j.joms.2009.03.050 . 19772941 PMC10159647

18 X. Zhang , I. S. Hamadeh , S. Song , et al., “ Osteonecrosis of the Jaw in the United States Food and Drug Administration's Adverse Event Reporting System (FAERS) ,” Journal of Bone and Mineral Research 31 , no. 2 ( 2016 ): 336 – 340 , 10.1002/jbmr.2693 . 26288087

19 F. Saad , J. E. Brown , C. Van Poznak , et al., “ Incidence, Risk Factors, and Outcomes of Osteonecrosis of the Jaw: Integrated Analysis From Three Blinded Active‐Controlled Phase III Trials in Cancer Patients With Bone Metastases ,” Annals of Oncology 23 , no. 5 ( 2012 ): 1341 – 1347 , 10.1093/annonc/mdr435 . 21986094

20 Agency for Healthcare Research and Quality , Clinical Classifications Software Refined (CCSR) for ICD‐10‐CM, v2021.2 [Software] ( Healthcare Cost and Utilization Project (HCUP) , 2021 ), https://hcup‐us.ahrq.gov/toolssoftware/ccsr/dxccsr.jsp .

21 Agency for Healthcare Research and Quality , Healthcare Cost and Utilization Project (HCUP): 2016 National Inpatient Sample (NIS) ( U.S. Department of Health and Human Services , 2017 ), https://www.hcup‐us.ahrq.gov/nisoverview.jsp .

22 Agency for Healthcare Research and Quality , Healthcare Cost and Utilization Project (HCUP): 2017 National Inpatient Sample (NIS) ( U.S. Department of Health and Human Services , 2018 ), https://www.hcup‐us.ahrq.gov/nisoverview.jsp .

23 Agency for Healthcare Research and Quality , Healthcare Cost and Utilization Project (HCUP): 2018 National Inpatient Sample (NIS) ( U.S. Department of Health and Human Services , 2019 ), https://www.hcup‐us.ahrq.gov/nisoverview.jsp .

24 Agency for Healthcare Research and Quality , Healthcare Cost and Utilization Project (HCUP): 2019 National Inpatient Sample (NIS) ( U.S. Department of Health and Human Services , 2020 ), https://www.hcup‐us.ahrq.gov/nisoverview.jsp .

25 Agency for Healthcare Research and Quality , Healthcare Cost and Utilization Project (HCUP): 2020 National Inpatient Sample (NIS) ( U.S. Department of Health and Human Services , 2021 ), https://www.hcup‐us.ahrq.gov/nisoverview.jsp .

26 Agency for Healthcare Research and Quality , Healthcare Cost and Utilization Project (HCUP): 2021 National Inpatient Sample (NIS) ( U.S. Department of Health and Human Services , 2022 ), https://www.hcup‐us.ahrq.gov/nisoverview.jsp .

27 SAS Institute Inc , SAS® 9.4 [Computer Software] ( SAS Institute Inc. , 2013 ).

28 RStudio Team , RStudio: Integrated Development Environment for R ( RStudio, PBC , 2020 ), https://www.rstudio.com/ .

29 J. Cohen , Statistical Power Analysis for the Behavioral Sciences , 2nd ed. ( Lawrence Erlbaum Associates , 1988 ).

30 S. P. Kane , “ClinCalc DrugStats Database (Version 2025.08),” ClinCalc ( 2025 ), https://clincalc.com/DrugStats .

31 J. R. Curtis , L. Carbone , H. Cheng , et al., “ Longitudinal Trends in Use of Bone Mass Measurement Among Older Americans, 1999‐2005 ,” Journal of Bone and Mineral Research 23 , no. 7 ( 2008 ): 1061 – 1067 , 10.1359/jbmr.080232 . 18302495 PMC2497454

32 A. S. E. Clark and A. M. Glenny , “ The Issue With Incidence: A Scoping Review of Reported Medication‐Related Osteonecrosis of the Jaws (MRONJ) Incidence Around the Globe ,” BMJ Public Health 3 , no. 1 ( 2025 ). 10.1136/bmjph-2024-002009 PMC11973752 40196437

33 Y. Zhong , W. Dai , L. Yin , G. Wu , and X. Wang , “ Real‐World Study of Medication‐Related Osteonecrosis of the Jaw From 2010 to 2023 Based on Food and Drug Administration Adverse Event Reporting System ,” JBMR Plus 9 , no. 3 ( 2025 ): ziaf003 . 39990280 10.1093/jbmrpl/ziaf003 PMC11843447

34 N. Yarom , C. L. Shapiro , D. E. Peterson , et al., “ Medication‐Related Osteonecrosis of the Jaw: MASCC/ISOO/ASCO Clinical Practice Guideline ,” Journal of Clinical Oncology 37 , no. 25 ( 2019 ): 2270 – 2290 , 10.1200/JCO.19.01186 . 31329513

35 A. Bedogni , R. Mauceri , V. Fusco , et al., “ Italian Position Paper (SIPMO‐SICMF) on Medication‐Related Osteonecrosis of the Jaw (MRONJ) ,” Oral Diseases 30 , no. 6 ( 2024 ): 3679 – 3709 , 10.1111/odi.14887 . 38317291

36 Y. S. Chang , S. Nanayakkara , A. Yaacoub , and S. C. Cox , “ Prevention of Medication‐Related Osteonecrosis of the Jaw: Institutional Insights From a Retrospective Study ,” Australian Dental Journal 70 , no. 1 ( 2025 ): 70 – 77 , 10.1111/adj.13050 . 39670507

37 N. Baghalipour , O. Moztarzadeh , W. Samar , J. Gencur , V. Volf , and L. Hauer , “ Comprehensive Review of Prevention and Management Strategies for Medication‐Related Osteonecrosis of the Jaw (MRONJ) ,” Oral Health & Preventive Dentistry 23 ( 2025 ): 403 . 40762678 10.3290/j.ohpd.c_2169 PMC12327072

38 O. Di Fede , V. Panzarella , R. Mauceri , et al., “ The Dental Management of Patients at Risk of Medication‐Related Osteonecrosis of the Jaw: New Paradigm of Primary Prevention ,” BioMed Research International 2018 , no. 1 ( 2018 ): 2684924 . 30306086 10.1155/2018/2684924 PMC6164200

39 A. M. Vandone , M. Donadio , M. Mozzati , et al., “ Impact of Dental Care in the Prevention of Bisphosphonate‐Associated Osteonecrosis of the Jaw: A Single‐Center Clinical Experience ,” Annals of Oncology 23 , no. 1 ( 2012 ): 193 – 200 , 10.1093/annonc/mdr039 . 21427065

40 N. Patel and N. Seoudi , “ Management of Medication‐Related Osteonecrosis of the Jaw: An Overview of National and International Guidelines ,” British Journal of Oral and Maxillofacial Surgery 62 , no. 10 ( 2024 ): 899 – 908 . 39448352 10.1016/j.bjoms.2024.08.008