Search

Abstract Alendronate and raloxifene are among the most popular anti-osteoporosis medications. However, there is a lack of head-to-head comparative effectiveness studies comparing the two treatments. We conducted a retrospective large-scale multicenter study encompassing over 300 million patients across nine databases encoded in the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). The primary outcome was the incidence of osteoporotic hip fracture, while secondary outcomes were vertebral fracture, atypical femoral fracture (AFF), osteonecrosis of the jaw (ONJ), and esophageal cancer. We used propensity score trimming and stratification based on an expansive propensity score model with all pre-treatment patient characteritistcs. We accounted for unmeasured confounding using negative control outcomes to estimate and adjust for residual systematic bias in each data source. We identified 283,586 alendronate patients and 40,463 raloxifene patients. There were 7.48 hip fracture, 8.18 vertebral fracture, 1.14 AFF, 0.21 esophageal cancer and 0.09 ONJ events per 1,000 person-years in the alendronate cohort and 6.62, 7.36, 0.69, 0.22 and 0.06 events per 1,000 person-years, respectively, in the raloxifene cohort. Alendronate and raloxifene have a similar hip fracture risk (hazard ratio [HR] 1.03, 95% confidence interval [CI] 0.94–1.13), but alendronate users are more likely to have vertebral fractures (HR 1.07, 95% CI 1.01–1.14). Alendronate has higher risk for AFF (HR 1.51, 95% CI 1.23–1.84) but similar risk for esophageal cancer (HR 0.95, 95% CI 0.53–1.70), and ONJ (HR 1.62, 95% CI 0.78–3.34). We demonstrated substantial control of measured confounding by propensity score adjustment, and minimal residual systematic bias through negative control experiments, lending credibility to our effect estimates. Raloxifene is as effective as alendronate and may remain an option in the prevention of osteoporotic fracture.

Predicting health outcomes from longitudinal health histories is of central importance to healthcare. Observational healthcare databases such as patient diary databases provide a rich resource for patient-level predictive modeling. In this paper, we propose a Bayesian hierarchical vector autoregressive (VAR) model to predict medical and psychological conditions using multivariate time series data. Compared to the existing patient-specific predictive VAR models, our model demonstrated higher accuracy in predicting future observations in terms of both point and interval estimates due to the pooling effect of the hierarchical model specification. In addition, by adopting an elastic-net prior, our model offers greater interpretability about the associations between variables of interest on both the population level and the patient level, as well as between-patient heterogeneity. We apply the model to two examples: 1) predicting substance use craving, negative affect and tobacco use among college students, and 2) predicting functional somatic symptoms and psychological discomforts.

BACKGROUND: Of the ~50 million cases of epilepsy worldwide, an estimated 80% originate from cortical areas implicated in language. Although the precise language loci can vary significantly across individuals, electrical stimulation mapping for eloquent areas has become standard of care in resective surgery for frontotemporal epilepsies. Although considerable work has been done to establish the minimum necessary resection distance from these sites to preserve language, no previous work has shown how these representations are affected by proximal resections. METHODS: Between 1967 and 2005, 22 patients [seizure onset (y): 11.5 (0.2–33); age at initial resection (y): 27.7 (10–39); time between operations (y): 8.4 (1–20.3); sex: 14 females; hemisphere: 21 left] underwent repeated perisylvian resective epilepsy surgeries of the language-dominant hemisphere. Each set of operations comprised intraoperative language mapping and cortical photographs. Using this data, a Bayesian hierarchical model was used to estimate the variability of language localization pre-resection Versus post-resection. RESULTS: The statistical model shows the posterior median difference in cortical location of language sites preresection Versus postresection is 0.6 cm, with a posterior 95% CI of 0.4 cm, 0.9 cm. CONCLUSION: This work suggests permanence in cortical language centers following resection of infringing cortex, while providing a reasonable statistical method to impute unobserved sites during the mappings, and confirming the validity of using proximity sites defined by shortest distance in the current literature.