Validation of the Fracture Risk Evaluation Model (FREM) in predicting major osteoporotic fractures and hip fractures using administrative health data

Validity of Major Osteoporotic Fracture Diagnoses in the Danish National Patient Registry

Objective

To evaluate the validity of diagnosis codes for Major Osteoporotic Fracture (MOF) in the Danish National Patient Registry (NPR) and secondly to evaluate whether the fracture was incident/acute using register-based definitions including date criteria and procedural codes.

Methods

We identified a random sample of 2400 records with a diagnosis code for a MOF in the NPR with dates in the year of 2018. Diagnoses were coded with the 10th revision of the International Classification of Diseases (ICD-10). The sample included 2375 unique fracture patients from the Region of Southern Denmark. Medical records were retrieved for the study population and reviewed by an algorithmic search function and medical doctors to verify the MOF diagnoses. Register-based definitions of incident/acute MOF was evaluated in NPR data by applying date criteria and procedural codes.

Results

The PPV for MOF diagnoses overall was 0.99 (95% CI: 0.98;0.99) and PPV=0.99 for the four individual fracture sites, respectively. Further, analyses of incident/acute fractures applying date criteria, procedural codes and using patients' first contact in the NPR resulted in PPV=0.88 (95% CI: 0.84;0.91) for hip fractures, PPV=0.78 (95% CI: 0.74;0.83) for humerus fractures, PPV=0.78 (95% CI: 0.73;0.83) for clinical vertebral fractures and PPV=0.87 (95% CI: 0.83;0.90) for wrist fractures.

Conclusion

ICD-10 coded MOF diagnoses are valid in the NPR. Furthermore, a set of register-based criteria can be applied to qualify if the MOF fracture was incident/acute. Thus, the NPR is a valuable and reliable data source for epidemiological research on osteoporotic fractures.

Excess mortality following a first and subsequent osteoporotic fracture: a Danish nationwide register-based cohort study on the mediating effects of comorbidities

OBJECTIVES

This study aimed to examine the risk of mortality following incident and subsequent osteoporotic fractures, the effect of different fracture type combinations, and the mediating role of postfracture morbidity in a Danish population.

METHODS

We used the National Patient Registry to identify patients ≥60 years with incident major osteoporotic fracture of the hip, vertebrae, wrist or humerus between 2013 and 2018, and controls matched 1:10 on age and sex. Possible mediators were identified using International Classification of Diseases, 10th Revision codes registered in the 6 months following index fracture. HRs were estimated using Cox regression analyses with 95% CIs. The effect of possible mediators was estimated using mediation analyses.

RESULTS

The study included 106 303 patients and 1 062 988 controls. Mortality following index fracture was highest in the month following hip fractures (HR 10.98 (95% CI 10.23 to 11.79) in women and HR 16.40 (95% CI 15.00 to 17.93) in men). Subsequent hip fractures resulted in the highest HRs for all fracture type combinations. In women, the highest HR was observed in patients with index wrist/subsequent hip fractures (HR 2.43 (95% CI 2.12 to 2.78)). In men, the highest HR was observed in patients with index humerus/subsequent hip fractures (HR 2.69 (95% CI 2.04 to 3.54)). Pneumonia mediated the largest proportion of mortality, but dehydration, urinary tract infection and sepsis were also important factors.

CONCLUSIONS

The highest mortality risk was found in the month immediately following both index and subsequent fracture. The combination of index and subsequent fractures at different skeletal sites had a substantial impact on the risk of mortality. Postfracture morbidities were found mediate the association.

An enhanced version of FREM (Fracture Risk Evaluation Model) using national administrative health data: analysis protocol for development and validation of a multivariable prediction model

BACKGROUND

Osteoporosis poses a growing healthcare challenge owing to its rising prevalence and a significant treatment gap, as patients are widely underdiagnosed and consequently undertreated, leaving them at high risk of osteoporotic fracture. Several tools aim to improve case-finding in osteoporosis. One such tool is the Fracture Risk Evaluation Model (FREM), which in contrast to other tools focuses on imminent fracture risk and holds potential for automation as it relies solely on data that is routinely collected via the Danish healthcare registers. The present article is an analysis protocol for a prediction model that is to be used as a modified version of FREM, with the intention of improving the identification of subjects at high imminent risk of fracture by including pharmacological exposures and using more advanced statistical methods compared to the original FREM. Its main purposes are to document and motivate various aspects and choices of data management and statistical analyses.

METHODS

The model will be developed by employing logistic regression with grouped LASSO regularization as the primary statistical approach and gradient-boosted classification trees as a secondary statistical modality. Hyperparameter choices as well as computational considerations on these two approaches are investigated by an unsupervised data review (i.e., blinded to the outcome), which also investigates and handles multicollinarity among the included exposures. Further, we present an unsupervised review of the data and testing of analysis code with respect to speed and robustness on a remote analysis environment. The data review and code tests are used to adjust the analysis plans in a blinded manner, so as not to increase the risk of overfitting in the proposed methods.

DISCUSSION

This protocol specifies the planned tool development to ensure transparency in the modeling approach, hence improving the validity of the enhanced tool to be developed. Through an unsupervised data review, it is further documented that the planned statistical approaches are feasible and compatible with the data employed.

Risk of Fracture After Bilateral Oophorectomy

Fragility fractures, resulting from low-energy trauma, occur in approximately 1 in 10 Danish women aged 50 years or older. Bilateral oophorectomy (surgical removal of both ovaries) may increase the risk of fragility fractures due to loss of ovarian sex steroids, particularly estrogen. We investigated the association between bilateral oophorectomy and risk of fragility fracture and whether this was conditional on age at time of bilateral oophorectomy, hormone therapy (HT) use, hysterectomy, physical activity level, body mass index (BMI), or smoking. We performed a cohort study of 25,853 female nurses (≥45 years) participating in the Danish Nurse Cohort. Nurses were followed from age 50 years or entry into the cohort, whichever came last, until date of first fragility fracture, death, emigration, or end of follow-up on December 31, 2018, whichever came first. Cox regression models with age as the underlying time scale were used to estimate the association between time-varying bilateral oophorectomy (all ages, <51/≥51 years) and incident fragility fracture (any and site-specific [forearm, hip, spine, and other]). Exposure and outcome were ascertained from nationwide patient registries. During 491,626 person-years of follow-up, 6600 nurses (25.5%) with incident fragility fractures were identified, and 1938 (7.5%) nurses had a bilateral oophorectomy. The frequency of fragility fractures was 24.1% in nurses who were <51 years at time of bilateral oophorectomy and 18.1% in nurses who were ≥51 years. No statistically significant associations were observed between bilateral oophorectomy at any age and fragility fractures at any site. Neither HT use, hysterectomy, physical activity level, BMI, nor smoking altered the results. © 2023 The Authors. JBMR Plus published by Wiley Periodicals LLC on behalf of American Society for Bone and Mineral Research.

General Comorbidity Indicators Contribute to Fracture Risk Independent of FRAX: Registry-Based Cohort Study

CONTEXT

FRAX® estimates 10-year fracture probability from osteoporosis-specific risk factors. Medical comorbidity indicators are associated with fracture risk but whether these are independent from those in FRAX is uncertain.

OBJECTIVE

We hypothesized Johns Hopkins Aggregated Diagnosis Groups (ADG®) score or recent hospitalization number may be independently associated with increased risk for fractures.

METHODS

This retrospective cohort study included women and men age ≥ 40 in the Manitoba BMD Registry (1996-2016) with at least 3 years prior health care data and used linked administrative databases to construct ADG scores along with number of hospitalizations for each individual. Incident Major Osteoporotic Fracture and Hip Fracture was ascertained during average follow-up of 9 years; Cox regression analysis determined the association between increasing ADG score or number of hospitalizations and fractures.

RESULTS

Separately, hospitalizations and ADG score independently increased the hazard ratio for fracture at all levels of comorbidity (hazard range 1.2-1.8, all P < 0.05), irrespective of adjustment for FRAX, BMD, and competing mortality. Taken together, there was still a higher than predicted rate of fracture at all levels of increased comorbidity, independent of FRAX and BMD but attenuated by competing mortality. Using an intervention threshold of major fracture risk >20%, application of the comorbidity hazard ratio multiplier to the patient population FRAX scores would increase the number of treatment candidates from 8.6% to 14.4%.

CONCLUSION

Both complex and simple measures of medical comorbidity may be used to modify FRAX-based risk estimates to capture the increased fracture risk associated with multiple comorbid conditions in older patients.

Application of intelligent X-ray image analysis in risk assessment of osteoporotic fracture of femoral neck in the elderly

The paper focuses on establishing a risk assessment model of femoral neck osteoporotic fracture (FNOF) in the elderly population and improving the screening efficiency and accuracy of such diseases in specific populations. In literature research, the main risk factors of femoral neck osteoporosis (FNOP) in the elderly were studied and analyzed; the femur region of interest (ROI) and the hard bone edge segmentation model were selected from the X-ray digital image by using the image depth learning method. On this basis, the femoral trabecular score and femoral neck strength (FNS) in the set region were selected as the main evaluation elements, and the quantitative analysis method was established; an X-ray image processing method was applied to the feasibility study of FNOP and compared with dual-energy X-ray absorptiometry measurements of bone mineral density; Finally, the main risk factors of FNOP were selected and the prediction model of FNOP in the elderly population was established based on medical image processing, machine learning model construction and other methods. Some FNOP health records were selected as test samples for comparative analysis with traditional manual evaluation methods. The paper shows the risk assessment model of FNOF in the elderly population, which is feasible in testing. Among them, the artificial neural network model had a better accuracy (95.83%) and recall rate (100.00%), and the support vector machine prediction model had high specificity (62.50%). With the help of a machine learning method to establish the risk assessment model of FNOF for the elderly, one can provide decision support for the fracture risk assessment of the elderly and remind the clinic to give targeted interventions for the above high-risk groups in order to reduce the fracture risk.

Development and external validation of a 1- and 5-year fracture prediction tool based on electronic medical records data: The EPIC risk algorithm

OBJECTIVES

We aimed to develop and validate a fracture risk algorithm for the automatic identification of subjects at high risk of imminent and long-term fracture risk.

RESEARCH, DESIGN, AND METHODS

A cohort of subjects aged 50-85, between 2007 and 2017, was extracted from the Catalan information system for the development of research in primary care database (SIDIAP). Participants were followed until the earliest of death, transfer out, fracture, or 12/31/2017. Potential risk factors were obtained based on the existing literature. Cox regression was used to model 1 and 5-year risk of hip and major fracture. The original cohort was randomly split in 80:20 for development and internal validation purposes respectively. External validation was explored in a cohort extracted from the Spanish database for pharmaco-epidemiological research in primary care.

RESULTS

A total of 1.76 million people were included from SIDIAP (50.7 % women with mean age of 65.4 years). Hip and major fracture incidence rates were 3.57 [95%CI 3.53 to 3.60] and 11.61 [95%CI 11.54 to 11.68] per 1000 person-years, respectively. The derived model included 19 risk factors. Internal validity showed good results on calibration and discrimination. The 1-year C-statistic for hip and major fracture were 0.851 (95%CI 0.853 to 0.864), and 0.717 (95%CI 0.742 to 0.749) respectively. The 5-year C-statistic for hip and major fracture were 0.849 (95%CI 0.847 to 0.852) and 0.724 (95%CI 0.721 to 0.727) respectively. External validation showed good performance for hip and major fracture risk prediction.

CONCLUSIONS

We have developed and validated a clinical prediction tool for 1- and 5-year hip and major osteoporotic fracture risks using electronic primary care data. The proposed algorithm can be automatically estimated at the population level using the available primary care records. Future work is needed on the cost-effectiveness of its use for population-based screening and targeted prevention of osteoporotic fractures.

FREM predicts 10-year incident fracture risk independent of FRAX® probability: a registry-based cohort study

The Danish Fracture Risk Evaluation Model (FREM) was found to predict fracture risk independent of 10-year fracture probability derived with the FRAX® tool including bone mineral density from DXA.

INTRODUCTION

FREM was developed from Danish public health registers without DXA information to identify high imminent risk of major osteoporotic fracture (MOF) and hip fracture (HF), while FRAX® estimates 10-year fracture probability from clinical risk factors and femoral neck bone mineral density (BMD) from DXA. The FREM algorithm showed significant 1- and 2-year fracture risk stratification when applied to a clinical population from Manitoba, Canada. We examined whether FREM predicts 10-year fracture risk independent of 10-year FRAX probability computed with BMD.

METHODS

Using the Manitoba BMD Program registry, we identified women and men aged ≥ 45 years undergoing baseline BMD assessment. We calculated FREM and FRAX scores, and identified incident fractures over 10 years. Hazard ratios (HRs) for incident fracture were estimated according to FREM quintile, adjusted for FRAX probability. We compared predicted with observed 10-year cumulative fracture probability estimated with competing mortality.

RESULTS

The study population comprised 74,446 women, mean age 65.2 years; 7945 men, mean age 67.5 years. There were 7957 and 646 incident MOF and 2554 and 294 incident HF in women and men, respectively. Higher FREM scores were associated with increased risk for MOF (highest vs middle quintile HRs 1.49 women, 2.06 men) and HF (highest vs middle quintile HRs 2.15 women, 2.20 men) even when adjusted for FRAX. Greater mortality with higher FREM scores attenuated its effect on 10-year fracture probability. In the highest FREM quintile, observed slightly exceeded predicted 10-year probability for MOF (ratios 1.05 in women, 1.49 in men) and HF (ratios 1.29 in women, 1.34 in men).

CONCLUSIONS

Higher FREM scores identified women and men at increased fracture risk even when adjusted for FRAX probability that included BMD; hence, FREM provides additional predictive information to FRAX. FRAX slightly underestimated 10-year fracture probability in those falling within the highest FREM quintile.

Management of patients at very high risk of osteoporotic fractures through sequential treatments

Osteoporosis care has evolved markedly over the last 50 years, such that there are now an established clinical definition, validated methods of fracture risk assessment and a range of effective pharmacological agents. Currently, bone-forming (anabolic) agents, in many countries, are used in those patients who have continued to lose bone mineral density (BMD), patients with multiple subsequent fractures or those who have fractured despite treatment with antiresorptive agents. However, head-to-head data suggest that anabolic agents have greater rapidity and efficacy for fracture risk reduction than do antiresorptive therapies. The European Society for Clinical and Economic Aspects of Osteoporosis, Osteoarthritis and Musculoskeletal Diseases (ESCEO) convened an expert working group to discuss the tools available to identify patients at high risk of fracture, review the evidence for the use of anabolic agents as the initial intervention in patients at highest risk of fracture and consider the sequence of therapy following their use. This position paper sets out the findings of the group and the consequent recommendations. The key conclusion is that the current evidence base supports an "anabolic first" approach in patients found to be at very high risk of fracture, followed by maintenance therapy using an antiresorptive agent, and with the subsequent need for antiosteoporosis therapy addressed over a lifetime horizon.

Evaluating the performance of the Charlson Comorbidity Index (CCI) in fracture risk prediction and developing a new Charlson Fracture Index (CFI): a register-based cohort study

The Charlson Comorbidity Index (CCI) may be applicable for predicting fracture risk since several diagnoses from the index are predictors of fracture. Main results were that the CCI was updated to predict risk of hip fracture with fair precision and that the index could be useful in detecting high-risk individuals.

PURPOSE

Several of the Charlson Comorbidity Index (CCI) diagnoses are validated predictors of fracture. The purpose of this study was to evaluate the performance of the CCI 1987 by Charlson et al. and of the CCI 2011 by Quan et al. in predicting major osteoporotic fracture (MOF) and hip fracture (HF). Furthermore, it was examined whether the index could be modified to improve fracture risk prediction.

METHODS

The study population included the entire Danish population aged 45 + years as per January 1, 2018. The cohort was split randomly 50/50 into a development and a validation cohort. CCI diagnoses and fracture outcomes were identified from hospital diagnoses. The weighting of diagnoses was updated in a new Charlson Fracture Index (CFI) using multivariable logistic regression. Predictive capabilities of the CCI 1987, the updated CCI 2011 and the new Charlson Fracture index were evaluated in the validation cohort by receiver operating characteristics (ROC) curves and area under the curve (AUC).

RESULTS

In the validation cohort, the 1987 and 2011 CCIs resulted in AUCs below or around 0.7 in prediction of MOF and HF in both sexes. The CFI resulted in AUCs < 0.7 in prediction of MOF in both sexes. In prediction of HF, the CFI resulted in AUC of 0.755 (95% CI 0.749; 0.761) in women and 0.782 (95% CI 0.772; 0.793) in men.

CONCLUSION

The 1987 and 2011 CCIs showed overall poor accuracy in fracture risk prediction. The CFI showed fair accuracy in prediction of HF in women and in men.

Prediction of imminent fracture risk in Canadian women and men aged 45 years or older: external validation of the Fracture Risk Evaluation Model (FREM)

The Fracture Risk Evaluation Model (FREM) identifies individuals at high imminent risk of major osteoporotic fractures. We validated FREM on 74,828 individuals from Manitoba, Canada, and found significant fracture risk stratification for all FREM scores. FREM performed better than age alone but not as well as FRAX® with BMD.

INTRODUCTION

The FREM is a tool developed from Danish public health registers (hospital diagnoses) to identify individuals over age 45 years at high imminent risk of major osteoporotic fractures (MOF) and hip fracture (HF). In this study, our aim was to examine the ability of FREM to identify individuals at high imminent fracture risk in women and men from Manitoba, Canada.

METHODS

We used the population-based Manitoba Bone Mineral Density (BMD) Program registry, and identified women and men aged 45 years or older undergoing baseline BMD assessment with 2 years of follow-up data. From linked population-based data sources, we constructed FREM scores using up to 10 years of prior healthcare information.

RESULTS

The study population comprised 74,828 subjects, and during the 2 years of observation, 1612 incident MOF and 299 incident HF occurred. We found significant fracture risk stratification for all FREM scores, with AUC estimates of 0.63-0.66 for MOF for both sexes and 0.84 for women and 0.65-0.67 for men for HF. FREM performed better than age alone but not as well as FRAX® with BMD. The inclusion of physician claims data gave slightly better performance than hospitalization data alone. Overall calibration for 1-year MOF prediction was reasonable, but HF prediction was overestimated.

CONCLUSION

In conclusion, the FREM algorithm shows significant fracture risk stratification when applied to an independent clinical population from Manitoba, Canada. Overall calibration for MOF prediction was good, but hip fracture risk was systematically overestimated indicating the need for recalibration.