A CT-based Deep Learning Model for Predicting Subsequent Fracture Risk in Patients with Hip Fracture.

There are few published long-term data on absolute risk of subsequent fracture (refracture) following initial low-trauma fracture in women and fewer in men. To examine long-term risk of subsequent fracture following initial osteoporotic fracture in men and women. Prospective cohort study (Dubbo Osteoporosis Epidemiology Study) in Australia of 2245 community-dwelling women and 1760 men aged 60 years or older followed up for 16 years from July 1989 through April 2005. Incidence of first (initial) fracture and incidence of subsequent fracture according to sex, age group, and time since first fracture. Relative risk was determined by comparing risk of subsequent fracture with risk of initial fracture. There were 905 women and 337 men with an initial fracture, of whom 253 women and 71 men experienced a subsequent fracture. Relative risk (RR) of subsequent fracture in women was 1.95 (95% confidence interval [CI], 1.70-2.25) and in men was 3.47 (95% CI, 2.68-4.48). As a result, absolute risk of subsequent fracture was similar in women and men and at least as great as the initial fracture risk for a woman 10 years older. Thus, women and men aged 60 to 69 years had absolute refracture rates of 36/1000 person-years (95% CI, 26-48/1000) and 37/1000 person-years (95% CI, 23-59/1000), respectively. The increase in absolute fracture risk remained for up to 10 years, by which time 40% to 60% of surviving women and men experienced a subsequent fracture. All fracture locations apart from rib (men) and ankle (women) resulted in increased subsequent fracture risk, with highest RRs following hip (RR, 9.97; 95% CI, 1.38-71.98) and clinical vertebral (RR, 15.12; 95% CI, 6.06-37.69) fractures in younger men. In multivariate analyses, femoral neck bone mineral density, age, and smoking were predictors of subsequent fracture in women and femoral neck bone mineral density, physical activity, and calcium intake were predictors in men. After an initial low-trauma fracture, absolute risk of subsequent fracture was similar for men and women. This increased risk occurred for virtually all clinical fractures and persisted for up to 10 years.

SummaryThe current study investigated subsequent fracture risk following a forearm fracture in three country of birth categories: Norway, Europe and North America, and other countries. Subsequent fracture risk was modestly higher in Norwegian-born individuals compared to the two other groups. Secondary fracture prevention should be recommended regardless of country background.BackgroundFracture risk is higher in patients with a previous fracture, but whether subsequent fracture risk differs by origin of birth is unknown. This study explores subsequent fracture risk in patients with an index forearm fracture according to region of birth.MethodsNationwide data on forearm fractures in patients ≥ 18 years in 2008–2019 were obtained from the Norwegian Patient Registry and Statistics Norway. Index fractures were identified by ICD-10 code S52, whereas subsequent fractures included any ICD-10 fracture code. Data on country of birth were from Statistics Norway and included three regional categories: (1) Norway, (2) other Europe and North America and (3) other countries. Direct age standardization and Cox proportional hazard regression were used to analyse the data.ResultsAmong 143,476 individuals with an index forearm fracture, 35,361 sustained a subsequent fracture. Norwegian-born forearm fracture patients had the highest subsequent fracture rates (516/10,000 person-years in women and 380 in men). People born outside Europe and North America had the lowest rates (278/10,000 person-years in women and 286 in men). Compared to Norwegian-born individuals, the hazard ratios (HRs) of subsequent fracture in individuals from Europe and North American were 0.93 (95% CI 0.88–0.98) in women and 0.85 (95% CI 0.79–0.92) in men. The corresponding HRs in individuals from other countries were 0.76 (95% CI 0.70–0.84) in women and 0.82 (95% CI 0.74–0.92) in men.ConclusionIndividuals born outside Norway had a lower subsequent fracture risk than Norwegian-born individuals; however, subsequent fracture risk increased with age in all groups. Our results indicate that secondary fracture prevention should be recommended regardless of region of origin.

To investigate imminent risk and odds of subsequent osteoporotic fractures and associated risk factors in patients who experienced an initial osteoporotic fracture. Patients aged ≥ 50years with ≥ 1 osteoporotic fracture were analyzed from Japan's Medical Data Vision (MDV) database of claims from acute-care hospitals (January 2012-January 2017). Multivariable models were constructed to explore the impact of key comorbidities and medications on the subsequent fracture risk: Cox proportional hazards model for time to subsequent fracture and logistic regression models for odds of subsequent fracture within 1 and 2years from index fracture. In total, 32,926 patients were eligible with a median follow-up duration of 12.3months. The percentage of patients experiencing subsequent fractures was 14.1% across the study duration, and 10.8% and 18.6% in patients with 1 and 2years of follow-up, respectively. In the Cox proportional hazards model, patients with vertebral or NHNV index fractures had a higher subsequent fracture risk than patients with a hip index fracture (adjusted hazard ratio [aHR] 1.11 and 1.26, respectively); subsequent fracture risk was lower in males than females (aHR 0.89). Patients with baseline claims for tranquilizers and glucocorticoids had a higher subsequent fracture risk than those without (aHR 1.14 and 1.08, respectively). Additionally, baseline claims for anti-Parkinson's medications, alcoholism, and stage 4/5 chronic kidney disease were significantly associated with higher odds of subsequent fracture in the logistic regression models. Several clinical and demographic factors were associated with a higher risk and odds of subsequent fracture. This may help to identify patients who should be prioritized for osteoporosis treatment.

The risk of subsequent fracture is increased after initial fractures; however, proper understanding of its magnitude is lacking. This population-based study examines the subsequent fracture risk in women and men by age and type of initial incident fracture. All incident nonvertebral fractures between 1994 and 2009 were registered in 27,158 participants in the Tromsø Study, Norway. The analysis included 3108 subjects with an initial incident fracture after the age of 49 years. Subsequent fracture (n = 664) risk was expressed as rate ratios (RR) and absolute proportions irrespective of death. The rates of both initial and subsequent fractures increased with age, the latter with the steepest curve. Compared with initial incident fracture rate of 30.8 per 1000 in women and 12.9 per 1000 in men, the overall age-adjusted RR of subsequent fracture was 1.3 (95% CI, 1.2-1.5) in women, and 2.0 (95% CI, 1.6-2.4) in men. Although the RRs decreased with age, the absolute proportions of those with initial fracture who suffered a subsequent fracture increased with age; from 9% to 30% in women and from 10% to 26% in men, between the age groups 50-59 to 80+ years. The type of subsequent fracture varied by age from mostly minor fractures in the youngest to hip or other major fractures in the oldest age groups, irrespective of type and severity of initial fracture. In women and men, 45% and 38% of the subsequent hip or other major fractures, respectively, were preceded by initial minor fractures. The risk of subsequent fracture is high in all age groups. At older age, severe subsequent fracture types follow both clinically severe and minor initial incident fractures. Any fragility fracture in the elderly reflects the need for specific osteoporosis management to reduce further fracture risk.

ABSTRACTFracture liaison services (FLS) are considered to be the most effective organizational approach for secondary fracture prevention. In this study, we evaluated whether FLS care was associated with reduced subsequent fracture and mortality risk over 3 years of follow‐up. In total, 8682 consecutive patients aged 50–90 years with a recent fracture were included. Before FLS introduction, regular fracture treatment procedures were followed (pre‐FLS). After FLS introduction, patients were invited to the FLS and FLS attenders were assessed for osteoporosis, prevalent vertebral fractures, metabolic bone disorders, medication use, and fall risk, and treatment for fracture prevention was initiated according to Dutch guidelines. All fractures were radiographically confirmed and categorized into major/hip (pelvis, proximal humerus or tibia, vertebral, multiple rib, distal femur) and non‐major/non‐hip (all other fractures). Mortality risk was examined using age and sex adjusted Cox proportional hazard models. For subsequent fracture risk, Cox proportional hazard models were adjusted for age, sex, and competing mortality risk (subdistribution hazard [SHR] approach). The pre‐FLS group consisted of 2530 patients (72% women), of whom 1188 (46.9%) had major/hip index fractures, the post‐FLS group consisted of 6152 patients (69% women), of whom 2973 (48.3%) had major/hip index fractures. In patients with a non‐major/non‐hip fracture there was no difference in subsequent non‐major/non‐hip fracture risk or mortality between pre‐FLS and post‐FLS. In patients with a major/hip index fracture, mortality risk was lower post‐FLS (hazard ratio [HR] 0.84; 95% confidence interval [CI], 0.73–0.96) and subsequent major/hip fracture risk was lower in the first 360 days after index fracture post‐FLS compared to pre‐FLS (SHR 0.67; 95% CI, 0.52–0.87). In conclusion, FLS care was associated with a lower mortality risk in the first 3 years and a lower subsequent major/hip fracture risk in the first year in patients with a major/hip index fracture but not in patients with a non‐major/non‐hip fracture. © 2022 The Authors. Journal of Bone and Mineral Research published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research (ASBMR).

Several guidelines recommend a bone and fall-related osteoporosis risk assessment in all patients with fracture and age > 50years. In practice, however, there is no consensus whether screening > 85years is useful. To evaluate the subsequent fracture risk in all patient > 85 years, comparing the two populations of Fracture Liaison Service (FLS) attenders and non-attenders. All patients > 85years that presented at the FLS with a non-vertebral fracture were included in the study during a 5-year period (September 2004 and December 2009). Excluded were pathologic fractures, death < 30 days, or patients on osteoporosis treatment. in patients that attended the FLS, assessment of bone mineral density and fall-risk factors were screened. In both the attenders and non-attenders groups, mortality and subsequent fracture rates were scored during a follow-up of 2years. 282 patients fulfilled inclusion criteria for screening, of which 160 (57%) patients did not attend the FLS. 122 patients were screened for osteoporosis and fall-related risk of whom 72 were diagnosed with osteoporosis. Subsequent fracture risk in both groups was 19%. Medical treatment was started in 51 patients, of which 15 patients developed a subsequent fracture. Cox-regression analysis indicated a significantly lower mortality rate, but not a diminished subsequent fracture rate in the FLS screened population compared to the non-attenders. The advantage of a FLS in reducing subsequent fracture risk in patients > 85years seems to be limited. In practice a large proportion of these patients are not screened.

Radiologists often face challenges in differentiating benign from malignant sacral bone lesions due to their similar imaging characteristics. This study aimed to develop an ensemble deep learning (DL) model that can preoperatively distinguish between benign and malignant sacral tumors using noncontrast computed tomography images. Preoperative sacral CT scans from 569 patients with confirmed sacral lesions were analyzed. Data from Center 1 were utilized in model development and internal test via fivefold cross-validation, and those from Centers 2 and 3 were employed in external test. Various ensemble models combining human-readable interpretation and DL were developed. The diagnostic performance of the models and radiologists was assessed using metrics such as precision, recall, accuracy, area under the curve (AUC), F1 score, and confusion matrix. Furthermore, the clinical benefits derived from radiologists' interpretations and supported by the DL model were evaluated. The ensemble model, which integrates 3D-DenseNet121 with human interpretation, exhibited the most robust performance. The ensemble model demonstrated high performance on the internal and external test sets and achieved AUCs of 0.9139 and 0.8713, F1 scores of 0.9054 and 0.8571, precision of 0.9041 and 0.8824, recall of 0.9136 and 0.8333, and accuracy of 0.8630 and 0.8182, respectively. Across the external test cohort, all radiologists experienced improvements in AUC, accuracy, sensitivity, and specificity. Notably, junior radiologists demonstrated significant improvements compared with senior radiologists. The potential clinical application of the DL model lies in its capacity to considerably enhance the diagnostic efficiency of radiologists. This study presents the first ensemble deep learning model integrating 3D-DenseNet121 with radiologists' interpretation for preoperative differentiation of sacral tumors on noncontrast CT that improved diagnostic performance across all experience levels, particularly for junior radiologists. First artificial intelligence-radiologist ensemble for noncontrast computed tomography (NCCT)-based sacral tumor classification. Boosts all radiologists' performance, with the greatest gains for juniors, potentially reducing referrals. Enables reliable NCCT diagnosis, overcoming contrast/magnetic resonance imaging dependency in musculoskeletal oncology.

Half of fragility fractures occur in individuals with nonosteoporotic BMD (BMD T-score > -2.5); however, there is no information on postfracture adverse events of subsequent fracture and mortality for different BMD levels. The objective of this work was to determine the risk and predictors of subsequent fracture and excess mortality following initial fracture according to BMD. The subjects were community-dwelling participants aged 60+ years from the Dubbo Osteoporosis Epidemiology Study with incident fractures followed from 1989 to 2011. The outcome measurements were as follows: risk of subsequent fracture and mortality according to BMD categorized as normal (T-score < -1), osteopenia (T-score ≤ -1 and > -2.5), and osteoporosis (T-score ≤ -2.5). There were 528 low-trauma fractures in women and 187 in men. Of these, 12% occurred in individuals with normal BMD (38 women, 50 men) and 42% in individuals with osteopenia (221 women, 76 men). The relative risk (RR) of subsequent fracture was >2.0-fold for all levels of BMD (normal BMD: 2.0 [1.2 to 3.3] for women and 2.1 [1.2 to 3.8] for men; osteopenia: 2.1 [1.7 to 2.6] for women and 2.5 [1.6 to 4.1] for men; and osteoporosis 3.2 [2.7 to 3.9] for women and 2.1 [1.4 to 3.1] for men. The likelihood of falling and reduced quadriceps strength contributed to subsequent fracture risk in women with normal BMD. By contrast with subsequent fracture risk, postfracture mortality was increased particularly in individuals with low BMD (age-adjusted standardized mortality ratio [SMR] for osteopenia 1.3 [1.1 to 1.7] and 2.2 [1.7 to 2.9] for women and men, respectively, and osteoporosis 1.7 [1.5 to 2.0] and 2.7 [2.0 to 3.6] for women and men, respectively). This study demonstrates the high burden of subsequent fracture in individuals with normal BMD and osteopenia, and excess mortality particularly for those with osteopenia (and osteoporosis). These findings highlight the importance of these fractures and underscore the gap in evidence for benefit of antiosteoporotic treatment for fragility fracture, in those with only mildly low BMD.

ObjectiveOsteoporotic fragility fractures, that are common in men and women, signal increased risk of future fractures and of premature mortality. Less than one-third of postmenopausal women and fewer men are prescribed active treatments to reduce fracture risk. Therefore, in this study the association of oral bisphosphonate recommendation with subsequent fracture and mortality over eight years in a fracture liaison service setting was analysed.Materials and methodsIn this prospective cohort study, 5011 men and women aged >50 years, who sustained a clinical fracture, accepted the invitation to attend the fracture liaison service of the West Glasgow health service between 1999 and 2007. These patients were fully assessed and all were recommended calcium and vitamin D. Based on pre-defined fracture risk criteria, 2534 (50.7%) patients were additionally also recommended oral bisphosphonates. Mortality and subsequent fracture risk were the pre-defined outcomes analysed using Cox proportional hazard models.ResultsThose recommended bisphosphonates were more often female (82.9 vs. 72.4%), were older (73.4 vs. 64.4 years), had lower bone mineral density T-score (-3.1 vs. -1.5) and more had sustained hip fractures (21.7 vs. 6.2%; p < 0.001). After adjustments, patients recommended bisphosphonates had lower subsequent fracture risk (Hazard Ratio (HR): 0.60; 95% confidence interval (CI): 0.49–0.73) and lower mortality risk (HR: 0.79, 95%CI: 0.64–0.97).ConclusionOf the patients, who are fully assessed after a fracture at the fracture liaison service, those with higher fracture risk and a recommendation for bisphosphonates had worse baseline characteristics. However, after adjusting for these differences, those recommended bisphosphonate treatment had a substantially lower risk for subsequent fragility fracture and lower risk for mortality. These community-based data indicate the adverse public health outcomes and mortality impacts of the current low treatment levels post fracture could be improved by bisphosphonate recommendation for both subsequent fracture and mortality.

Establishing a hospital based fracture liaison service to prevent secondary insufficiency fractures

Despite prevalent anti-osteoporosis medication (AOM) switching in real-world osteoporosis management, few studies have evaluated the impact of persistent AOM treatment, allowing for AOM switching, on the risk of subsequent fracture. We examined the association between persistence in AOM and subsequent fractures, allowing for medication switching among patients with osteoporotic fractures. This retrospective cohort study used Taiwan National Health Insurance claims data to select patients who initiated AOM between 2013 and 2016. Treatment persistence was defined as use of any AOM on a given day of interest with a 45-day grace period. Medication switch was allowed for persistence if remaining on treatment. AOMs with long-lasting inhibition of bone resorption (zoledronate and denosumab) were categorized as high-potency; others as low-potency. Multivariate Cox models were used to evaluate risk of subsequent fractures ≥3 months after initiating AOM. A total of 119 473 patients were included (mean [SD] follow-up 46.4 [15.6] months), and 26.8% switched from the index AOM. Within 1 year, 52% remained persistent with AOM. Compared to patients with persistent AOM, those not persistent had higher risk of subsequent hip (adjusted hazard ratio [aHR] = 1.31; 95% CI, 1.21-1.42), vertebral (aHR = 1.17; 95% CI, 1.13-1.22), and radius fractures (aHR = 1.16; 95% CI, 1.08-1.25). Patients with persistent AOM who switched from high- to low-potency AOM had higher risk of subsequent vertebral fractures than those with persistent AOM and no potency switch (aHR = 1.28; 95% CI, 1.02-1.60). Patients with non-persistent AOM had higher risk of subsequent fractures than persistent users when allowing AOM switch. Switching AOM potency may influence the risk of subsequent vertebral fractures and warrants further investigation.

Sentiment analysis (SA) is a machine learning application that drives people’s opinions from text using natural language processing (NLP) techniques. Implementing Arabic SA is challenging for many reasons, including equivocation, numerous dialects, lack of resources, morphological diversity, lack of contextual information, and hiding of sentiment terms in the implicit text. Deep learning models such as convolutional neural networks (CNN) and long short-term memory (LSTM) have significantly improved in the Arabic SA domain. Hybrid models based on CNN combined with long short-term memory (LSTM) or gated recurrent unit (GRU) have further improved the performance of single DL models. In addition, the ensemble of deep learning models, especially stacking ensembles, is expected to increase the robustness and accuracy of the previous DL models. In this paper, we proposed a stacking ensemble model that combined the prediction power of CNN and hybrid deep learning models to predict Arabic sentiment accurately. The stacking ensemble algorithm has two main phases. Three DL models were optimized in the first phase, including deep CNN, hybrid CNN-LSTM, and hybrid CNN-GRU. In the second phase, these three separate pre-trained models’ outputs were integrated with a support vector machine (SVM) meta-learner. To extract features for DL models, the continuous bag of words (CBOW) and the skip-gram models with 300 dimensions of the word embedding were used. Arabic health services datasets (Main-AHS and Sub-AHS) and the Arabic sentiment tweets dataset were used to train and test the models (ASTD). A number of well-known deep learning models, including DeepCNN, hybrid CNN-LSTM, hybrid CNN-GRU, and conventional ML algorithms, have been used to compare the performance of the proposed ensemble model. We discovered that the proposed deep stacking model achieved the best performance compared to the previous models. Based on the CBOW word embedding, the proposed model achieved the highest accuracy of 92.12%, 95.81%, and 81.4% for Main-AHS, Sub-AHS, and ASTD datasets, respectively.

Nonhip nonvertebral fractures represent half of all osteoporotic fractures; however, their contribution to the burden of refracture and premature mortality is unclear. To examine the risk and burden of subsequent fracture and mortality associated with an initial nonhip nonvertebral fracture. This is a prospective cohort from the Dubbo Osteoporosis Epidemiology Study, 1989-2010 of community dwelling participants aged 60+ with incident fractures. Relative risk of all subsequent fractures and age-adjusted standardized mortality ratios were calculated according to initial fracture type. The total burden of adverse events was assessed using competing risk models with four potential outcomes: mortality after initial fracture, mortality after subsequent fracture, subsequent fracture and alive, or event-free. Of the 952 fractures in women and 343 in men, over half were nonhip nonvertebral fractures (486 in women and 173 in men). Nonhip nonvertebral fractures were associated with increased risk of any subsequent fracture (1.95 [1.67-2.27] for women and 2.47 [1.82-3.35] for men), hip refracture (2.11 [1.04-4.28] for women and 2.63 [1.35-5.13] for men), and vertebral refracture (1.89 [1.43-2.48] for women and 2.13 [1.20-3.79] for men). More importantly, nonhip nonvertebral fractures were associated overall with 20% excess mortality for the first 5 years postfracture, of which approximately half were due to initial fracture and the remaining due to subsequent fractures. Proximal fractures were associated with increased mortality risk per se, whereas distal fractures were associated with increased mortality risk only in the group who sustained subsequent fractures. Nonhip nonvertebral fractures are associated with significant risk of subsequent fracture including hip and vertebral refracture, and premature mortality. Due to their high prevalence, about half of all subsequent fractures and a quarter of all fracture-related excess mortality were attributable to nonhip nonvertebral fracture. Thus nonhip nonvertebral fracture warrants early investigation and appropriate intervention.

This study assesses prevalence of subsequent fractures during the year after incident osteoporosis-related non-vertebral fractures among privately insured and Medicare populations and compares costs between patients with and without subsequent fractures. Many non-vertebral fracture patients incur subsequent fractures, and those who do are significantly more costly during the year after incident fracture. To estimate the prevalence of subsequent osteoporosis-related non-vertebral (NV) fractures during the year following an incident NV fracture and compare costs between NV fracture patients with and without subsequent fractures. Using insurance claims data (1999-2006), privately-insured (ages 18-64years) and Medicare (ages 65+ years) patients with ≥1 subsequent osteoporosis-related NV fracture within a year of an incident osteoporosis-related NV fracture were matched to controls with incident NV fractures but no subsequent fractures. Subsequent fractures were identified as any claim for an NV fracture occurring >3months after the incident NV fracture (>6months were required for fractures occurring at the same site as the incident fracture). The study assessed prevalence of subsequent fractures and compared costs (from the payer's perspective) between patients with and without subsequent fractures over the year following an incident NV fracture. Among privately insured NV fracture patients, 14.1% had any subsequent NV fractures, 1.6% had subsequent hip fractures, and 13.0% had subsequent non-vertebral, non-hip (NVNH) fractures, while 22.6% of Medicare NV fracture patients had subsequent NV fractures, 9.4% had subsequent hip fractures, and 15.5% had subsequent NVNH fractures. Mean excess health care costs per privately insured subsequent fracture patient were $9,789 ($19,072 vs. $9,914, p < 0.01), while excess medical costs per Medicare subsequent fracture patient were $12,527 ($31,904 vs. $19,377, p < 0.01). NV fracture patients are at substantial risk for subsequent NV fractures within 1year, and patients who incur subsequent fractures are significantly more costly than those who do not during the year following an incident fracture.

SummaryThe absolute 5-year risk of subsequent non-vertebral fractures (NVFs) in 1,921 patients presenting with a NVF was 17.6% and of mortality was 32.3%. These risks were highest within the first year, indicating the need to study which reversible factors can be targeted to immediately minimise subsequent fracture risk and mortality.IntroductionNVFs are the most frequent clinical fractures in patients presenting at the emergency unit because of a clinical fracture. The aim of the study was to determine the 5-year absolute risk (AR) of subsequent NVF and mortality in patients at the time they present with a NVF.MethodsBetween 1999 and 2001, 1,921 consecutive patients 50+ years from a level 1 trauma centre were included. All NVFs were confirmed on radiograph reports, and mortality was checked in the national obituary database. Available potential risk factors for a subsequent NVF and mortality (age, sex and baseline fracture location: major—hip, pelvis, multiple ribs, proximal tibia/humerus and distal femur; minor—all others) were expressed as hazard ratios (HR) with 95% confidence intervals (CI) using multivariable Cox regression analysis.ResultsThe AR for a subsequent NVF was 17.6% and was related to age (HR per decade, 1.44; 95%CI, 1.29–1.60). The AR for mortality was 32.3% and was related to age (HR per decade, 2.59; 95%CI, 2.37–2.84), male sex (HR, 1.74; 95%CI, 1.44–2.10), major fracture at baseline (HR, 5.56; 95%CI, 3.48–8.88; not constant over time) and subsequent fracture (HR, 1.65; 95%CI, 1.33–2.05). The highest risks were found within the first year (NVFs, 6.4%; mortality, 12.2%) and were related to age and, in addition, to baseline fracture location for mortality.ConclusionsWithin 5 years after an initial NVF, nearly one in five patients sustained a subsequent NVF and one in three died. One third of subsequent NVFs and mortality occurred within 1 year, indicating the need to study which reversible factors can be targeted to immediately prevent subsequent fractures and mortality.