( PDF ) Rev Osteoporos Metab Miner. 2010; 2 (1): 23-8

Ivorra Cortés J, Román-Ivorra JA, Alegre Sancho JJ, Beltrán Catalán E, Chalmeta Verdejo I, Fernández-Llanio Comella N, Muñoz Gil S
Servicio de Reumatología – Hospital Universitario Dr. Peset – Valencia


We calculate specific triage thresholds for the PIXI-LUNAR heel densitometer to give a 90% specificity for osteoporosis and normal bone mineral density (BMD) at the hip or spine.
693 women aged 30-93 years (mean age 58.2 ± 9.6 years) referred for osteoporosis study, underwent hip and spine BMD measurements (HOLOGIC) by dual energy X-ray absortiometry (DXA), also had a peripheral heel DXA densitometry (PIXI-LUNAR). The os calcis T-scores for all woman were subjected to a receiver operator characteristic (ROC) analysis with the definition of osteoporosis (T-score ≤ -2.5) and BMD normal (T-score > -1) made at the the lumbar spine or femoral neck.
Patients with a heel T-score of above +0.6 are very likely to have normal bone density on axial densitometry, whilst patients with heel T-score of below -1.3 are very likely to have osteoporosis at the hip or spine. Only patients whose measurements lie between the thresholds should be referred for axial DXA.

Project financed by the Generalitat Valenciana – Conselleria de Sanitat – DOGV 5337 – 1.09.2007 – Resolution 20 10 2007

Keywords: Bone mineral density, Osteoporosis, Peripheral x-ray absorptiometry.


The prevalence of osteoporosis in white women over 50 years of age is high. In fact the risk of suffering an osteoporotic fracture of the hip, the spinal column or wrist over the course of their lives is 40%1. These facts make osteoporosis a real health problem. Bone mineral density (BMD) is the best prognostic factor of risk of osteoporotic fracture, for which reason densitometry is the fundamental technique for the diagnosis of osteoporosis before fractures appear.
Various techniques are available to measure BMD such as computerised tomography and ultrasound, but that most used currently is dual energy X-ray absorptiometry (DXA). The measurement of BMD through a DXA apparatus at the central level (hip and spinal column) is considered as a gold standard for the diagnosis of osteoporosis. The OMS developed some diagnostic criteria2 based on the lowest of the densitometric results carried out on the hip and spinal column. The measurement of BMD in the peripheral skeleton is related to an added risk of fracture at whatever level3. Peripheral densitometers have the advantage of being of a lower cost to purchase, of needing less space for their installation, the tests of measurement are carried out with greater rapidity, and, due to their small size and weight, they are easy to transport. It has been observed that the cut-off points for the diagnosis of osteoporosis with axial densitometers are not the same as for peripheral densitometers4-6. The NOS (National Osteoporosis Society) recommends that peripheral densitometers are use as a screening tool with two cut-off points which identify those patients with spinal and/or hip osteoporosis, with a sensitivity and specificity of 90%7. In such a way the patients with peripheral BMD with a T-score below the lower cut-off point would have a high possibility of having osteoporosis in the hip or spine, and those having a T-score above the higher cut-off point would seldom have osteoporosis in the spine or hip. However, the cut-off points are different for the different peripheral densitometers8. It is not known if the cut-off points can change according to the population studied or whether it is dependent on the model of central densitometer with which it is compared.
The objective of this work is to find a diagnostic algorithm for postmenopausal osteoporosis in our population, combining a DXA peripheral densitometer of the calcaneum (PIXI-LUNAR) and a HOLOGIC central densitometer.

Material and methods

Central (of hip and spine) and peripheral (calcaneum) densitometries were carried out consecutively in 693 women referred to the rheumatology clinic for a study of postmenopausal osteoporosis. The study was approved by the scientific committee of our hospital. The central bone mineral density was measured with a Hologic ExploerTM ExplorerR Series densitometer in the left femur and in the lumbar vertebra – L1 to L4. The bone mineral density in the calcaneum was measured in the left foot with a PIXI-Lunar densitometer. It was considered that a patient had osteoporosis if the T-score in the whole hip or in the lumbar region (L1-L4) was ≤ -2.5. A patient was classified as having normal BDM if the T-score was > 0 both in the hip and in the lumbar region. Using the statistical package SPSS 15.1 the sensitivity and specificity for the different T-scores, obtained through the peripheral densitometer, were calculated for the diagnosis of osteoporosis or normality by means of a ROC (receiver operator characteristic) curve, as were the Pearson correlation coefficient between the peripheral T-scores and those obtained in the hip and lumbar vertebrae. With this data the optimum cut-off points for screening were chosen. These points would have to comply at least with the recommendations of the NOS7,8, which is to say a specificity and a sensitivity for the diagnosis of osteoporosis of 90%, with an interval of confidence which does not surpass the lower limit of 80%. In our diagnostic algorithm we consider it important to detect densitometries classified as normal, therefore for the upper cut-off point we would consider the T-score which would have a specificity of 90% to classify a patient with a normal central densitometry. In addition, the positive and negative predictive value of the cut-off points obtained for osteoporosis in our population would be calculated.


The women studied had an average age ± standard deviation (SD) of 58.19 ± 9.61 years (30 and 93 years). The average height of the population studied was 155.8 ± 6.2 cm, the average weight 64.5 ± 10.7 kg and the BMI 26.9 ± 4.5 kg/m2 (Table 1). The Pearson correlation coefficient was 0.616 between the T-scores obtained in the calcaneum and the whole hip, and 0.535 obtained with the lumbar vertebrae. According to the results of the axial densitometer (DXA) 29% of the women were osteoporotic, 47% osteopenic and 27% had normal levels of bone mass. By means of the ROC curve the sensitivity and specificity of the T-score obtained with the peripheral densitometer was calculated, to establish the diagnosis of osteoporosis and normality (Table 2). If a T-score of -2.5 SD is used in the peripheral densitometry for the diagnosis of osteoporosis the specificity is high, but the sensitivity is only 8%, which means that only 3% of densitometries will avoid being carried out. The specificity continues to be higher than 90% up to a cut-off point of T-score -1.3 SD, from which point the loss of specificity is significant (Figure 1A and Table 2). With this screening point only 4% of patients with normal axial densitometry would be classified as osteoporotic. Following the same criteria a T-score, using the PIXI, equal to or higher than 0.6 has a specificity of 90% to identify the normal central densitometries, and a sensibility of 97% to detect osteoporosis (Figure 1B and Table 2). With an algorithm based on the aforementioned cut-off points the positive predictive values for the diagnosis of osteoporosis and normality would be 80% and 78% respectively in our population. The negative predictive value for the diagnosis of osteoporosis is 98% and for the diagnosis of normality, 98%. With this 43% ± 4% of central densitometries would be avoided. If we consider the recommendations of the NOS the cut-off point would be a T-score of -0.2 and less than -1.3 (Tables 2 and 3).


The scarcity of axial densitometers has led to the use of peripheral densitometers in clinical practice, above all in the area of primary care. However, the clinical trials which have demonstrated the efficacy of different drugs in the treatment of osteoporosis, have been based on the selection of patients with low bone mineral density measured in the spine or hip. For this reason it is important to analyse the use of peripheral densitometers in the diagnosis of osteoporosis. The NOS8 (Table 3) evaluated different models of peripheral densitometers and calculated the T-score cut-off points with which a sensibility and specificity of 90% for the diagnosis of osteoporosis with an axial densitometer would be obtained (Table 3). For each model two cut-off points were established, a T-score below which 90% of patients with osteoporosis centrally are classified and another T-score above which are found all those patients without densitometric osteoporosis, that is to say, with a T-score above -2.5 with the central densitometer. Central densitometries would only be carried out in those cases situated between the two cut-off points. The results of this work show that each model of peripheral densitometer had different cut-off points and, indeed, that they vary with age. Subsequently other authors have published results with other models of peripheral densitometers following the same methodology. McCauley et al.9 determined the screening points of the Apollo Normand calcaneum densitometer with respect to a central lunar DPX-IQ densitometer. The T-score values were 9-1, 2 and -2.2.
Our results show different cut-off points with respect to a similar densitometer (PIXI-Lunar) analysed in the work of the NOS8 (Table 3), above all in the lower cut-off point, which in our case was situated at -1.3 as opposed to at -2 obtained in the work of the NOS. There are various differences between the two works which could explain the discrepancies found. The women in our study had an average age of 58 ± 9 years (Table 1), slightly lower than the average of 62 years for the Pixi-Lunar group analysed in the NOS work. In their work they confirmed that in increasing the age of the population the cut-off points tend to result in a lower T-score8. On the other hand the central densitometers were different, a HOLOGIC in our work and a Lunar in the NOS study. Forham et al.4, using a methodology similar to ours by means of ROC curves arrive at a cut-off point for detecting osteoporosis identical to ours, which is to say a T-score of -1.3 with a sensitivity of 69.6% for detecting osteoporosis and a specificity of 82.6%. They did not research the higher cut-off point. Pérez-Castrillón et al.5 in a Spanish population arrived at the conclusion that the best cut-off point for the diagnosis of osteoporosis with a PIXI-LUNAR densitometer of the calcaneum is a T-score of -1.6 SD, even though their results are based on 58 patients on whom they carried out central and peripheral densitometry. The discrepancies which are observed between the different studies, in addition to being explained by the differences in models of peripheral and central densitometers used in each work, could relate to other variables such as age, number of patients included or prevalence of osteoporosis in the population studied10. For this reason screening points for different age ranges for each population where the peripheral densitometer might be used should be calculated.
On the other hand we preferred to change the criteria for determining the upper cut-off point with respect to that used by the NOS7. The risk of fracture is not a dichotomous variable, rather, it is continuous and the current trend is to calculate the absolute risk of fracture, with densitometry being an additional test, such as it is evaluated in the FRAX index11. With the algorithm proposed by the NOS7,8 the cut-off point with a T-score of -0.2 in our peripheral densitometer has a specificity for normality of 70%, which is to say 30% of patients with osteopenia or osteoporosis are classified as non-osteoporotic, a fact which may restore credibility to the test among doctors and patients. In addition, a high percentage of fractures are produced in osteopenic patients and it is important to have this group of patients well classified12. With our cut-off point at a T-score of + 0.6, less than 10% of those patients with reduced BMD are classified as normal.
The National Osteoporosis Foundation13 recommends treating those patients with osteoporosis diagnosed by central densitometry, and uses the risk of fracture calculated through the FRAX index to select those patients with osteopenia who are given treatment. Our algorithm is adapted to this scheme because it serves to detect with high sensitivity and specificity the patients with central osteoporosis by means of peripheral densitometry, while the majority of those who have osteopenia are evaluated through central densitometry.
A limitation in our study, equal to that of most studies published to date, is that the bone mineral density in the femoral neck has not been evaluated, nor that in the whole hip, and we do not know in which way this may modify the results.
The International Society for Clinical Densitometry (ISCD) recommends that the use of peripheral densitometry is limited to those cases in which access to central densitometry is lacking10. There are practically no studies on its utility in male osteoporosis and it is not valid to evaluate the efficacy of the treatment10. This fact contrasts with its wide use, for which reason the NOS7 and the ISCD10 have diffused the methodology to be applied, to ensure its trustworthiness as a screening tool in the diagnosis of osteoporisis, and it is in this context in which our study should be placed.
In conclusion, in our population of postmenopausal patients referred for the study of osteoporosis, a diagnostic algorithm of BMD based on two densitometers, one peripheral – PIXI-LUNAR, and the other central – HOLOGIC, enables 43% fewer central densitometries to be carried out. Those patients with a T-score with the peripheral densitometer between -1.3 SD and +0.6 SD are referred. The sensitivity of the algorithm for the detection of densitometric osteoporosis is 97% and of normality, 96%. With a specificity for both of 90%.

1. Melton LJ 3rd. How many women have osteoporosis now? J Bone Miner Res 1995;10:175-7.
2. World Health Organisation 1994 Assessment of Fracture Risk and its Application to Screening for Postmenopausal Oteoporosis. World Health Organization, Geneva, Switzerland, pp. 1-129.
3. Van der Klift M, Laet CDE, Pols HAP. Assessment of fracture risk: Who should be treated for osteoporosis? Beast Practice & Research Clinical Rheumatology 2005; 19:937-50.
4. Fordham JN, Chinn DJ, Kumar N. Identification of Women with Reduced Bone Density ar the Lumbar Spine and Femoral Neck using BMD at the Os Calcis. Osteoporos Int 2000;11:797-802.
5. Pérez-Castrillón JL, Martín-Escudero JC, del Pino-Montes J, et al. Prevalence of Osteoporosis Using DXA Bone Mineral Density Measurements at the Calcaneus. J Clin Densitom 2005;8:404-8.
6. Faulkner KG, Von Stetton, Miller P. Discordance in patient classification using T–scores. J Clin Densitom 1999;2:343-50.
7. National Osteoporosis Society (2004). Position statement on the use of peripheral x-ray absortiometry in the management of osteoporosis. National Osteoporosis Society, Bath, England.
8. Blake GM, Chin DJ, Steel SA, Patel R, Panayiotou E, Thorpe J, et al. A list of device-specific thresholds for the clinical interpretation of peripheral x-ray absorptiometry examinations. Osteoporos Int 2005;16:2149-56.
9. McCauley E, Mackie A, Elliot D, Chuck A. Heel Bone densitometry: device specific thresholds for the assessment of osteoporosis. Br J Radiol 2006;79:464-7.
10. Hans DB, Shepherd JA, Schwartz EN, Reid DM, Blacke GM, Fordham JN, et al. Peripheral Dual-Energy X- ray Absortiometry in the management of osteoporosis: The 2007 ISCD Official Positions. J Clin Densitom 2008;11:188-206.
11. Kanis JA, Johnell O, Oden A, Johansson H, McCloskey E. FRAX and the assessment of fracture probability in men and women from the UK. Osteoporos Int 2008;19:385-97.
12. Siris E, Miller PD, Barret-Connor E, Faulkner KG, Wehren LE, Abbot TA, et al. Identification and fracture outcomes of undiagnosed low bone mineral density in postmenopausal women. Results from National Osteoporosis Risk Assessment. JAMA 2001;286:2815-22.
13. NOF’s New Clinician’s Guide to Prevention and Treatment of Osteoporosis. www.nof.org/professionals/Clinicians _Guide.htm Accessed April 18, 2008.