Performance of Osteoporosis Self-Assessment Tool (OST) in Predicting Osteoporosis—A Review

Bone health screening plays a vital role in the early diagnosis and treatment of osteoporosis to prevent fragility fractures among the elderly and high-risk individuals. Dual-energy X-ray absorptiometry (DXA), which detects bone mineral density, is the gold standard in diagnosing osteoporosis but is not suitable for screening. Therefore, many screening tools have been developed to identify individuals at risk for osteoporosis and prioritize them for DXA scanning. The Osteoporosis Self-assessment Tool (OST) is among the first tools established to predict osteoporosis in postmenopausal women. It can identify the population at risk for osteoporosis, but its performance varies according to ethnicity, gender, and age. Thus, these factors should be considered to ensure the optimal use of OST worldwide. Overall, OST is a simple and economical screening tool to predict osteoporosis and it can help to optimize the use of DXA.


Introduction
Osteoporosis is a progressive bone metabolic disease. It is undetectable until a bone fracture occurs. Once osteoporosis has developed, then it is less likely to completely restore the bone strength of the patients [1]. The prevalence of osteoporosis is increasing as the global population ages rapidly. In Asia, the number of osteoporotic hip fractures is expected to rise from 1,124,060 in 2018 to 2,563,488 in 2050 [2]. The Asians also encompassed 55% of the population at risk for fragility fractures worldwide [3]. The escalating morbidity and mortality rates due to osteoporotic fractures have distressed the patients, families, and society [4]. In addition, fragility fractures also contribute to a tremendous healthcare and economic burden. A recent meta-analysis of studies in Asia indicated that the median medical cost for hip fracture was USD 2943, representing around 19% of the gross domestic product of the countries studied in 2014 [5]. Thus, it is crucial to identify individuals at risk for osteoporosis to enable early intervention for fracture prevention.
Dual-energy X-ray absorptiometry (DXA) is the gold standard technique used to detect osteoporosis. According to the World Health Organization (WHO), a bone mineral density (BMD) ≤ −2.5 standard deviations (SD) below the young adult mean (or a T-score ≤ −2.5) indicates osteoporosis, while a T-score value at any site between ≤−1.0 and >−2.5 indicates a low bone mass or osteopenia [6]. Dual-energy X-ray absorptiometry cannot be widely used for osteoporosis screening due to its high cost and limited availability in most developing countries [7]. Quantitative ultrasound (QUS) has been developed as an alternative to DXA for osteoporosis screening [7]. Although QUS is portable and more economical than DXA, it may be unavailable in all primary medical settings.
Various clinical risk assessment tools have been developed to evaluate the risk of osteoporosis [8]. These screening tools help physicians to prioritize high-risk patients for a DXA scan. Some of the screening algorithms are the Fracture Risk Assessment Tool (FRAX), Qfracture algorithm, and Garvan Fracture Risk Calculator (Garvan) [9]. The Osteoporosis Self-assessment Tool (OST) is another predictive algorithm currently in use to predict the risk for osteoporosis [10]. It was first established by Koh et al. (2001) using data of postmenopausal women from eight Asian countries. The screening algorithm was only based on age (years) and body weight (kg): OSTA score = (body weight − age) × 0.2, with three osteoporosis risk categories: low risk (>−1), moderate risk (−1 to −4), and high risk (<−4). It performed well to determine women at risk of osteoporosis [11]. The performance of OST among Asian men was first assessed by Kung et al. (2004) and it demonstrated a moderate performance in predicting osteoporosis [12]. OST has been known as OSTA (OST for Asians) when it is applied to Asian women. The establishment of OSTA only involved postmenopausal women and men from East and Southeast Asia. A recent article by Chin (2017) reviewed the performance of OSTA among various Asian populations, but the performance of OST in non-Asian countries was not examined [13]. Thus, the present review summarized and compared the performance of OST in determining osteoporosis risk among the Asian and non-Asian population.

Literature Search
A literature search was performed from 15 January 2018 to 4 April 2018 using two databases: PubMed and Scopus. Only original articles written in English were included in this review. The search term used was "osteoporosis self-assessment tool". The search revealed 84 articles from PubMed and 65 articles from Scopus, which resulted in a total of 149 articles. After removing 16 duplicated articles, 133 articles were screened based on title and abstract. Only studies investigating the performance of OST against DXA were considered. The present review included 44 relevant articles ( Figure 1).

Performance of OST among Asian Women
The osteoporosis self-assessment tool for Asians (OSTA) was first developed by Koh et al. (2001) using data of postmenopausal women from eight Asian countries. The final algorithm only selected age and body weight as the predictors, creating the formula: OSTA score = (body weight in kg − age in years) × 0.2. Based on the truncated product of this formula, the women could be divided into three risk categories: low-risk (>−1), moderate-risk (−1 to −4), and high-risk (<−4). The predictive values of these scores were good, as indicated by the fact that 61% women categorized as high-risk were osteoporotic compared to 3% in the low-risk group in their study [11]. Its performance (cutoff = −1; sensitivity = 91%; specificity = 45%; AUC = 0.79) was superior than the reported values of the SOFSURF index [14], Osteoporosis Risk Assessment Index (ORAI), and Simple Calculated Osteoporosis Risk Estimation (SCORE) [15]. The performance of OSTA was subsequently validated in other East Asian populations, such as Chinese [16] and Korean women [17], and comparable results were obtained. This is not surprising, considering that a majority of the subjects in the development phase were East Asians [11]. In addition, the validation study by Huang et al. (2015) showed that OSTA performed better when BMD at the femoral neck was used as the reference, and when the women tested were older [16]. The site difference is probably due to the presence of osteophyte at the lumbar spine, which distorts its BMD. This finding was validated in similar studies conducted among Thai women [18,19]. The age difference coincides with the development population, which was elderly women (mean age: 62.3 ± 6.2 years) [11]. The limitation of OSTA in predicting osteoporosis among younger women was also observed in Thai women [19].

Performance of OST among Asian Men
Men also suffer from osteoporosis and their post-fracture mortality rate is higher than women [20]. Therefore, OSTA was developed for men by Kung et al. in 2004 based on data of community-dwelling Chinese men (age range: 50-93 years) in Hong Kong. The algorithm and cutoff values were the same as reported by Koh et al. (2001), but its performance in the development and validation cohort was not as good as in postmenopausal women (cutoff = −1; sensitivity = 71-73%, specificity = 68%; AUC = 0.780-0.790) [12]. OSTA was subsequently validated in the Chinese Han population [21] and Korean men [22], and both studies obtained a sensitivity > 80%. In contrast, OSTA (cutoff < −1) demonstrated a low sensitivity (27.6-28.5%) and a high specificity (89.2-92.7%) in a large study involving Chinese men of a wide age range (40-96 years) [23]. Sub-analysis revealed that similar to women, OSTA only performed well among older subjects [23].

Performance of OST with Modified Cutoff Values
Since the original cutoff values for OSTA were established in postmenopausal elderly women, predominantly Eastern Asians, its performance, in terms of sensitivity and specificity, may vary according to sex, age, and ethnic groups. Hence, modification to the cutoff values may be necessary to ensure the optimal performance of OSTA. This hypothesis was tested by Bhat et al. (2016) among Indian subjects (aged > 50 years) using a cutoff of 2 and OSTA achieved high sensitivity (sensitivity = 95.7%, specificity = 33.6%, AUC = 0.702) in predicting osteoporosis among the subjects [24]. However, OSTA failed to show similar performance in other populations (Chinese men with the cutoff −3.5, sensitivity = 47.3% [4]; Taiwanese men with the cutoff −1.86, sensitivity 69.2% [25]). In the subsequent discussion, the readers should notice the changes in cutoff values in many studies.

Performance of OST in Comparison with Other Screening Tools
QUS is another popular osteoporosis screening tool [7]. It was reported to have a strong association with BMD and bone mineral content (BMC) measured by DXA [26]. In three studies, OSTA was found to perform equally with QUS [8,27,28]. Thus, OSTA can be used in medical settings without QUS.
However, it should be noted that OSTA and QUS cannot be used interchangeably because the two tools are not equivalent [29].
A summary of the literature on the performance of OST among Asians is listed in Table 1.  To establish a prediction model to identify osteopenia risk in women aged 40-55 years

Performance of OST among Non-Asian Women
Although developed for Asians, the performance of OST has also been validated in non-Asians. The algorithm is the same as reported by Koh et al. (2001), but the cutoff values have been optimized to suit the designated populations. The performance of OST (cutoff < 2) was good in determining Caucasian women at risk of osteoporosis (sensitivity = 86-95.3%; specificity = 39.6-40%; AUC = 0.726-0.82) [10,38]. A study indicated that the performance of OST was similar between younger (cutoff ≤ 1 for 45-64 years) and older women (and cutoff ≤ −1 for >65 years) when different cutoff values were used [39].
Even within the same ethnicity, the cutoff values of OST need to be modified based on sex. A Spanish study showed that OST performed optimally at a cutoff of 2 in women (sensitivity = 94%, specificity = 59%, AUC = 0.762) and 3 in men (sensitivity = 39%, specificity = 86%, AUC = 0.623) [46].
In a group of men with rheumatoid arthritis, OST was weak (sensitivity = 64%; specificity = 54%) in determining those with low BMD even though the cutoff was modified to <4. The researchers indicated that the performance of OST could be limited by the low lean body mass of patients with rheumatoid arthritis [47].
A summary of the literature on the performance of OST among non-Asians is listed in Table 2.

OST for Fracture Prediction
Fragility fracture is one of the most common complications of osteoporosis. Although OST was developed to identify individuals with low BMD, its ability to predict fracture risk was assessed in several studies [54,58,59]. Yang et al. (2013) reported that OSTA (cutoff < −1) performed well in determining new vertebral fractures among postmenopausal Chinese women (sensitivity = 81.7%; specificity = 66%; AUC = 0.812) [59].
A summary of the literature on the performance of OST to predict fracture risk is listed in Table 3.

Conclusions
The performance of OST in predicting osteoporosis has been tested in various Asian and non-Asian populations. It demonstrates good predictive values in terms of sensitivity, specificity, and AUC when BMD is used as the reference. Some modifications in the OST cutoff should be made and tested to optimize its performance prior to its deployment, since the performance may vary according to age, sex, ethnicities, and the site of BMD measurement. Validation studies are necessary before including OST in the national guideline for osteoporosis screening. In most studies, OST demonstrated a high sensitivity and low specificity, which is typical for a screening test. In other words, OST might direct some individuals with normal bone health for an unnecessary DXA scan. At the same time, the number of potential patients subjected to a DXA scan is maximized, allowing the early detection and treatment of osteoporosis. This will reduce the complications and burdens of the disease. Thus, we argue that the benefits of implementing OST will outweigh its cost. As a conclusion, OST is a useful osteoporosis screening tool in prioritizing high-risk individuals for a DXA scan. It enables early disease detection, optimizes the use of the diagnostic facility, and therefore reduces the disease burden of osteoporosis.