Deep Learning With Electronic Health Records for Short-Term Fracture Risk Identification: Crystal Bone Algorithm Development and Validation.

BACKGROUND: Fractures as a result of osteoporosis and low bone mass are common and give rise to significant clinical, personal, and economic burden. Even after a fracture occurs, high fracture risk remains widely underdiagnosed and undertreated. Common fracture risk assessment tools utilize a subset of clinical risk factors for prediction, and often require manual data entry. Furthermore, these tools predict risk over the long term and do not explicitly provide short-term risk estimates necessary to identify patients likely to experience a fracture in the next 1-2 years. OBJECTIVE: The goal of this study was to develop and evaluate an algorithm for the identification of patients at risk of fracture in a subsequent 1- to 2-year period. In order to address the aforementioned limitations of current prediction tools, this approach focused on a short-term timeframe, automated data entry, and the use of longitudinal data to inform the predictions. METHODS: Using retrospective electronic health record data from over 1,000,000 patients, we developed Crystal Bone, an algorithm that applies machine learning techniques from natural language processing to the temporal nature of patient histories to generate short-term fracture risk predictions. Similar to how language models predict the next word in a given sentence or the topic of a document, Crystal Bone predicts whether a patient's future trajectory might contain a fracture event, or whether the signature of the patient's journey is similar to that of a typical future fracture patient. A holdout set with 192,590 patients was used to validate accuracy. Experimental baseline models and human-level performance were used for comparison. RESULTS: The model accurately predicted 1- to 2-year fracture risk for patients aged over 50 years (area under the receiver operating characteristics curve [AUROC] 0.81). These algorithms outperformed the experimental baselines (AUROC 0.67) and showed meaningful improvements when compared to retrospective approximation of human-level performance by correctly identifying 9649 of 13,765 (70%) at-risk patients who did not receive any preventative bone-health-related medical interventions from their physicians. CONCLUSIONS: These findings indicate that it is possible to use a patient's unique medical history as it changes over time to predict the risk of short-term fracture. Validating and applying such a tool within the health care system could enable automated and widespread prediction of this risk and may help with identification of patients at very high risk of fracture.

Screening, assessment, and treatment of osteoporosis for the nurse practitioner: key questions and answers for clinical practice--a Canadian perspective.

PURPOSE: Using a case-based approach, we review key clinical questions relevant to nurse practitioners (NPs) regarding the screening, assessment, and treatment of patients at risk for osteoporosis and fractures in a Canadian general practice setting. DATA SOURCES: A case presentation with relevant questions and answers to guide management of a patient. CONCLUSIONS: Osteoporosis is a common condition in both the aging male and female populations. Screening, diagnosis, and treatment of osteoporosis is lagging behind relative to other chronic disease states. NPs have a unique opportunity to help reduce this care gap by playing an integral role in the identification, risk stratification, and treatment of patients at risk for osteoporosis and fractures. IMPLICATIONS FOR PRACTICE: This case highlights the important role an NP can have in screening a patient previously not diagnosed or managed for osteoporosis. Performing a focused history and physical exam of the patient to determine appropriate screening tests and fracture risk will help in guiding treatment decisions.

Heterogeneity in skeletal load adaptation points to a role for modeling in the pathogenesis of osteoporotic fracture.

Genetic, environmental, or hormonal factors may cause heterogeneity in skeletal load response. Individuals with reduced sensitivity to load should require higher strains to generate an adaptive response, consequently have weaker bones and fracture more frequently. The purpose of our study was to determine if stresses (proportional to strains) at the femoral neck under equivalent loads were higher in women with a history of fractures compared with women without fractures. We studied postmenopausal women participating in the Canadian Multicentre Osteoporosis Study who had available hip structure analysis data from dual-energy X-ray absorptiometry scans (n = 2168). Women were categorized into 2 groups based on their number of self-reported fractures. We computed stress (megapascals) at the inferomedial margin of the femoral neck in a one-legged stance mode using a 2-dimensional engineering beam analysis. We used linear regression (SAS 9.3) to determine associations between stress, geometry parameters, and number of fractures. Postmenopausal women with 1 or more fractures had higher stress (2.6%), lower narrow neck bone mineral density (4.2%), cross-sectional area (3.9%), and section modulus (9.6%) than postmenopausal women without fractures (all p < 0.05). These findings provide evidence of heterogeneity in load response and suggest an important role for modeling in the pathogenesis of osteoporotic fracture.

Nitrates and bone turnover (NABT) - trial to select the best nitrate preparation: study protocol for a randomized controlled trial.

BACKGROUND: Organic nitrates uncouple bone turnover, improve bone mineral density, and improve trabecular and cortical components of bone. These changes in turnover, strength and geometry may translate into an important reduction in fractures. However, before proceeding with a large fracture trial, there is a need to identify the nitrate formulation that has both the greatest efficacy (with regards to bone turnover markers) and gives the fewest headaches. Ascertaining which nitrate formulation this may be is the purpose of the current study. METHODS AND DESIGN: This will be an open-label randomized, controlled trial conducted at Women's College Hospital comparing five formulations of nitrates for their effects on bone turnover markers and headache. We will recruit postmenopausal women age 50 years or older with no contraindications to nitroglycerin. Our trial will consist of a run-in phase and a treatment phase. We will enroll 420 women in the run-in phase, each to receive all of the 5 potential treatments in random order for 2 days, each with a 2-day washout period between treatments. Those who tolerate all formulations will enter the 12-week treatment phase and be randomly assigned to one of five groups: 0.3 mg sublingual nitroglycerin tablet, 0.6 mg of the sublingual tablet, a 20 mg tablet of isosorbide mononitrate, a 160 mg nitroglycerin transdermal patch (used for 8 h), and 15 mg of nitroglycerin ointment as used in a previous trial by our group. We will continue enrolment until we have randomized 210 women or 35 women per group. Concentrations of bone formation (bone-specific alkaline phosphatase and procollagen type I N-terminal propeptide) and bone resorption (C-telopeptides of collagen crosslinks and N-terminal crosslinks of collagen) agents will be measured in samples taken at study entry (the start of the run in phase) and 12 weeks. Subjects will record the frequency and severity of headaches daily during the run-in phase and then monthly after that. We will use the 'multiple comparisons with the best' approach for data analyses, as this strategy allows practical considerations of ease of use and tolerability to guide selection of the preparation for future studies. DISCUSSION: Data from this protocol will be used to develop a randomized, controlled trial of nitrates to prevent osteoporotic fractures. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT01387672. Controlled-Trials.com: ISRCTN08860742.

Evidence for impaired skeletal load adaptation among Canadian women with type 2 diabetes mellitus: insight into the BMD and bone fragility paradox.

OBJECTIVE: Recent data suggest that women with type 2 diabetes mellitus (T2DM) might be more susceptible to fractures due to an impaired adaptive response to mechanical load, despite reportedly higher bone mineral density (BMD). The purpose of this study was to use an engineering beam analysis to calculate and compare the load stresses on the femurs of healthy women and women with T2DM and compare these levels to conventional measures of femoral neck BMD. MATERIALS/METHODS: We studied 3658 women who participated in the Canadian Multicentre Osteoporosis Study (CaMos), and who had available Hip Structure Analysis (HSA) data from baseline dual energy x-ray absorptiometry (DXA) scans. Women were categorized into two groups based on the presence or absence of self-reported T2DM. We computed stress in megapascals (MPa) at the infero-medial margin of the femoral neck in a one-legged stance using an engineering beam analysis incorporating dimensions and geometry from DXA scans using the HSA method. We used linear regression (SAS 9.3) to determine the association between T2DM status and stress. We also determined the association between T2DM status and femoral neck BMD. RESULTS: Stresses were 4.5% higher in T2DM women than in non-diabetics (11.03±0.18 vs. 10.56±0.04 MPa; p=0.0093). Femoral neck BMD was 4.2% greater in women with T2DM than in non-diabetics (0.74±0.002 vs. 0.71±0.01 g/cm(2); p=0.0008). CONCLUSIONS: Despite higher femoral neck BMD, higher stress indicates weaker skeletal geometry for a given load, and suggests an impaired skeletal adaptive response to load may be present in women with T2DM.

Organic nitrates for osteoporosis: an update.

The number of osteoporotic fractures is increasing worldwide as populations age. An inexpensive and widely available treatment is necessary to alleviate this increase in fractures. Current treatments decrease fractures at trabecular bone sites (spine) but have limited effects at cortical sites (hip, legs, forearm and upper arm)-the most common sites of osteoporotic fracture. Treatments are also limited by costs, side effects and lack of availability. Nitric oxide (NO) is a novel agent that has the potential to influence cortical bone, is inexpensive, widely available and has limited side effects. In this review, we will evaluate the in vitro and in vivo data that support the concept that NO is important in bone cell function, review the observational, case control and randomized trial data on organic nitrates and the effects of these agents on bone turnover, geometry and strength.

Nitric oxide donors for the treatment of osteoporosis.

The number of osteoporotic fractures is increasing worldwide as populations age. An inexpensive and widely available treatment is necessary to alleviate this increase in fractures. Current treatments decrease fractures at trabecular bone sites (spine) but have limited effects at cortical sites (hip, legs, forearm, and upper arm)-the most common sites of osteoporotic fracture. Treatments are also limited by costs, side effects, and lack of availability. Nitric oxide is a novel agent that has the potential to influence cortical bone, is inexpensive, is widely available, and has limited side effects. In this review we evaluate the in vitro and in vivo data which support the concept that nitric oxide is important in bone cell function, review the observational and case-control studies reporting on subjects taking organic nitrates that act as nitric oxide donors, and review the effects of nitrates on bone mineral density measurements and fracture risk.

Effect of nitroglycerin ointment on bone density and strength in postmenopausal women: a randomized trial.

CONTEXT: Nitroglycerin stimulates bone formation and inhibits bone resorption, is inexpensive, and is widely available. Its effects on bone density, bone structure, and bone strength are unknown. OBJECTIVES: To determine if nitroglycerin increases lumbar spine bone mineral density (BMD) and to evaluate changes in hip BMD, bone geometry, and density at the radius and tibia, and markers of bone turnover. DESIGN, SETTING, AND PARTICIPANTS: A single-center, double-blind, placebo-controlled randomized trial conducted in Toronto, Ontario, Canada, for 24 months starting in November 2005 and completed in March 2010, of 243 postmenopausal women with lumbar spine T scores of between 0 and -2.0 who completed a 1-week run-in period taking nitroglycerin ointment. Intervention Nitroglycerin ointment (15 mg/d) or placebo applied at bedtime for 24 months. MAIN OUTCOME MEASURES: Areal BMD at the lumbar spine, femoral neck, and total hip. Secondary outcomes included indices of bone geometry and strength at the distal radius and tibia, and biomarkers of bone formation (bone-specific alkaline phosphatase) and bone resorption (urine N -telopeptide). RESULTS: At 2 years, women randomized to the nitroglycerin group had significant increases in areal BMD at the lumbar spine (from 1.05 to 1.14 g/cm(2) vs placebo from 1.06 to 1.08 g/cm(2); percentage change, 6.7%; 95% confidence interval [CI], 5.2%-8.2%; P < .001); total hip (from 0.92 to 0.97 g/cm(2) vs placebo from 0.93 to 0.92 g/cm(2); 6.2%; 95% CI, 5.6%-7.0%; P < .001); and femoral neck (from 0.88 to 0.93 g/cm(2) vs placebo from 0.87 to 0.86 g/cm(2); 7.0%; 95% CI, 5.5%-8.5%; P < .001). At 2 years, nitroglycerin also increased volumetric trabecular BMD (11.9% and 8.5%), cortical thickness (13.9% and 24.6%), periosteal circumference (7.4% and 2.9%), polar section modulus (10.7% and 9.8%), and polar moment of inertia (7.3% and 14.5%) at the radius and tibia, respectively (all P < .001); and increased bone-specific alkaline phosphatase by 34.8% and decreased urine N -telopeptide by 54.0% (P < .001). Incidence of serious adverse events did not differ between nitroglycerin (5 [4.2%]) and placebo (5 [4.3%]) groups. Among those women who continued treatment for 24 months, headaches were reported by 40 (35%) in nitroglycerin and 6 (5.4%) in placebo groups during the first month, decreasing substantially after 12 months. CONCLUSION: Among postmenopausal women, nitroglycerin ointment modestly increased BMD and decreased bone resorption. TRIAL REGISTRATION: isrctn.org Identifier: ISRCTN94484747.

Lasofoxifene in osteoporosis and its place in therapy.

Selective estrogen-receptor modulators (SERMs), which have estrogen-like effects on bone and "antiestrogen effects" on other tissues, have been in development for osteoporosis prevention and treatment in postmenopausal women as a safer alternative to long-term estrogen. We conducted a literature review of the skeletal and extraskeletal effects of lasofoxifene, a new generation SERM approved by the European Commission for osteoporosis treatment. Published data on the effects of lasofoxifene are based on 23 clinical pharmacology studies with over 10,000 participants from 17 phase 2 and 3 randomized controlled trials (RCTs). In RCTs, lasofoxifene decreases bone turnover markers (BTMs), increases bone mineral density (BMD) at the spine and hip, and decreases the incidence of vertebral and nonvertebral nonhip fractures compared with placebo. Compared with raloxifene, lasofoxifene gave greater decreases in BTMs, and greater increases in lumbar spine BMD. Lasofoxifene also decreased the risk of breast cancer, major coronary heart disease events, and stroke, but-similar to raloxifene-there was an increased risk of venous thromboembolism. In one trial, endometrial hypertrophy and uterine polyps were more common with lasofoxifene than with placebo, but endometrial cancer and hyperplasia were not. Lasofoxifene is probably most appropriate for use among women in their early or middle menopausal years (age 55-65) who have, or are at risk of developing, osteoporosis and in particular vertebral fractures. At the time of publication, lasofoxifene is not approved for use by the US Food and Drug Administration, and as such is not used in North America.

Associations between leisure physical activity participation and cortical bone mass and geometry at the radius and tibia in a Canadian cohort of postmenopausal women.

Few studies have examined the effects of leisure physical activity (PA) participation on bone mass and geometry in postmenopausal women using peripheral quantitative computed tomography (pQCT). The purpose of this study was to determine associations between leisure PA participation and bone mass and geometry at the radius and tibia in a Canadian cohort of healthy postmenopausal women (n=234, mean age 62 years). Leisure PA participation was assessed using the Minnesota Leisure Time Physical Activity Questionnaire and by generating a total activity score (TAS, mean=105, range=0-840). Bone mass and geometry at the distal and midshaft sites of the non-dominant radius and tibia were measured using pQCT. Associations between TAS and bone mass and geometry variables were determined using linear regression models adjusted for clinically relevant confounding variables. TAS was positively and significantly associated with total content, total area, cortical content and cortical area at the midshaft sites of the radius and tibia (p<0.05 for all associations). TAS was also positively and significantly associated with bone bending and torsional strength parameters at the midshaft radius and tibia (p<0.05 for all associations). No associations were observed between TAS and trabecular bone parameters. Leisure PA participation is positively associated with cortical bone mass and geometry as well as bending and torsional strength at weight-bearing and non-weight-bearing bone sites and may have the potential to modify bone strength and influence bone fragility in postmenopausal women.

The effects of organic nitrates on osteoporosis: a randomized controlled trial [ISRCTN94484747].

BACKGROUND: Osteoporotic fractures are common and are associated with increased morbidity, mortality and health care costs. The most effective way to moderate increases in health care costs and the sickness and premature death associated with osteoporotic fractures, is to prevent osteoporosis. Several lines of evidence suggest that nitrates, drugs typically prescribed for the treatment of angina, may be effective in preventing postmenopausal osteoporosis. METHODS: We have designed a multicentre randomized controlled trial to determine the effects of nitrates on bone. The trial consists of two studies. The objective of the first study is to determine whether isosorbide mononitrate at 20 mg/day or nitroglycerin ointment at 15 mg/day leads to fewer headaches. The nitrate that is best tolerated will be used in a second study with one main objective: To determine if postmenopausal women with a T-score at the lumbar spine (L1 to L4) between 0 and -2.0 randomized to two years of treatment with intermittent nitrates have a greater increase in spine bone mineral density as compared to women randomized to placebo. We hypothesize that: 1. Women will report fewer headaches when they are randomized to intermittent nitroglycerin ointment at 15 mg/day compared to intermittent oral isosorbide mononitrate at 20 mg/day, and, 2. After two years, women randomized to intermittent nitrates will have a greater percent increase in lumbar spine bone mineral density compared with women randomized to placebo. DISCUSSION: We have completed our pilot study and found that transdermal nitroglycerin was associated with fewer headaches than oral isosorbide mononitrate. We are currently recruiting patients for our second main study.