Five-year fracture risk assessment in postmenopausal women, using both the POL-RISK calculator and the Garvan nomogram: the Silesia Osteo Active Study

Abstract

Summary The study project was designed to assess the concordance of clinical results in the assessment of 5-year fracture risk of any fracture, carried out by two methods: the Garvan algorithm and the POL-RISK model. The study group included 389 postmenopausal women of Caucasian race. The concordance of results, obtained by those two models, turned out to be moderate, and the threshold for high fracture risk group was 11% in the POL-RISK model. Purpose The goal of the study was to evaluate the concordance of results in fracture risk assessments between the Garvan Fracture Risk Calculator and POL-RISK, a new Polish algorithm, and to define an optimal threshold for intervention. Methods The study was a part of the Silesia Osteo Active Study. A group of 389 postmenopausal women, aged 65.2±6.9 years (mean ± SD), was randomly selected from the general population of Zabrze, Poland. All the participants had bone densitometry examination to assess the bone mineral density of the femoral neck. The mean femoral neck T-score was (−0.99) ± 1.05 SD. 6.4% of the women revealed osteoporosis. Five-year risk of any fracture was assessed, using the Garvan and POL-RISK calculators. The performance of each model was evaluated by the area under the receiver operating characteristic curve (AUC). Results The median 5-year risk of any fracture was 7% (range 1–54%) in the Garvan model and 8.8% (range 1.1–45.5%) in the POL-RISK algorithm. There was a significant correlation between the results obtained by both methods (r=0.6, p<0.005). For the thresholds, assumed at 8% and 13% (according to recommendation derived from Garvan tool), the rates of concordance of results between both calculators were 76% and 84%, respectively. In ROC analysis for the POL-RISK method, performed with reference to the Garvan method at two different cut-offs, assumed to be high fracture risk indicators (8% and 13%), the AUC values were 0.865 and 0.884, respectively. The optimal threshold for high fracture risk in the POL-RISK algorithm was ≥ 11%, which yielded a sensitivity of 0.94 and a specificity of 0.71. Conclusion The obtained data demonstrate a moderate concordance of results between the POL-RISK algorithm and the Garvan model, illustrated by low and high fracture risk cut-offs, established in ROC analysis. In addition, the threshold of 11% in the POL-RISK method was the optimal level for “high risk”.

Predicting medication nonadherence risk in the Chinese type 2 diabetes mellitus population – establishment of a new risk nomogram model: a retrospective study

“…The nomogram is a graph with visualized line segments that is based on multifactor regression analysis, integrates multiple clinical indicators and is used to predict the probability of a certain clinical outcome. Although there are many research reports of this kind, 33,34 none of the research reports can really be used for practical clinical needs. 35 The nomogram model of this current research can be used by clinicians for the evaluation of nonadherence risk in the Chinese type 2 diabetes mellitus population.…”

Predictive value of machine learning on fracture risk in osteoporosis: a systematic review and meta-analysis

“…After removing duplicates and screening titles and abstracts, 378 articles remained. According to a full-text review, 53 articles13–67 were included. Fifty-three articles presented the development of one or more prediction models for osteoporotic fracture, while 26 articles described the validation of one or more models.…”

POL-RISK: an algorithm for 10-year fracture risk prediction in the postmenopausal women from the RAC-OST-POL Study

“…Using the longitudinal data, collected during five years of the observation, a new algorithm was introduced in 2017 to determine the five-year fracture risk in Polish postmenopausal women over the age of 55 years [6]. That algorithm, being based on the first five-year follow-up,, has since then been available at the www.fracture-risk.pl webpage to be later referred to as POL-RISK [7]. Afterwards, the follow-up period was extended to ten years.…”