LOW COST MASS SCREENING TOOL FOREVALUATING POST-MENOPAUSAL OSTEOPOROSIS : A Breakthrough for The Developing World

M Anburajan, C Rethinasabapathi, M Paul Korath, BG Ponnappa, TMR Panicker, A Govindan, GNS Prasad, WD Evans, K Jagadeesan

*Department of Medical Physics; **Department of Internal Medicine and Nuclear Cardiology; ***Department of Surgery; +Department of Radiology; ++Department of Physics, Anna University, Chennai - 600 025, India; #Department of Medical Physics, University Hospital of Wales, Cardiff, UK; ##Director and Founder, KJ Research Foundation, KJ Hospital, Research and PG Training Center, Chennai - 600 084, India.

The aim of this study was to establish an empirical link between total hip bone mineral density (BMD) by dual energy X-ray absorptiometry (DXA) and a simple clavicle radiogrammetry made from the chest radiograph that gives results of high sensitivity and specificity for predicting total hip BMD from chest radiograph for evaluation of osteoporosis. Fifty women, whose age ranged from 16 and 84 years were studied. Of the 50 women, five had previous fractures. These women were randomly selected from the general medical wards of KJ Hospital, Chennai, South-India during November, 1997 to April, 1998. Patients with secondary bone diseases were excluded. Subjects were divided into groups as follows : i) Pre-menopausal females (n=21, mean æ SD age = 31.0 æ 8.8 years); ii) Post-menopausal females (n=29, mean æ SD age = 64.3 æ 11.2 years). BMD of the right proximal femur was evaluated using a QDR - 1000 DXA bone densitometer (Hologic Inc., Waltham, Massachusetts, USA). Standard PA view chest radiograph was taken in all the cases. Right clavicle radiogrammetry measurements were made on the radiograph. WHO’s diagnostic criteria for osteoporosis was used. Data analysis was done with SPSS/PC statistical software package. In 45 pre-and post-menopausal women without previous fracture, total hip BMD by DXA was correlated significantly (p < 0.001) with combined cortical thickness (CCT) and % CCT of the clavicle (r=0.73 and r=0.74 respectively). From the multiple linear regression equation, KJH; Anburajan’s empirical formula was derived for predicting total hip BMD with good sensitivity (82.4%) and specificity (94.0%) from simple chest radiograph and is given as follows: Y=0.47 + 0.003938 (X1) + 0.420914 (X2) - 5.05139 X 10-4 (X3) where Y = predicted total hip BMD (g cm-2), X1 = % CCT of the clavicle, X2 = CCT of the clavicle and X3 = Patient’s age. Its positive and negative predictive value was 87.5% and 91.2% respectively. The KJH; Anburajan’s empirical formula is useful to predict hip BMD in the evaluation of post-menopausal osteoporosis from an inexpensive and widely available standard chest radiograph with good sensitivity and specificity - a boon to the developing nations.

INTRODUCTIONOsteoporosis is a major public health problem in India and in many other parts of world. It predominantly affects post-menopausal women and the elderly of both sexes and therefore the number of sufferers increases with increase of longevity of the general population. [1] Since bone mineral density (BMD) is the major determinant of the compressive and torsional strength of bone, fracture of bone will occur more easily if there is a substantial bone mineral loss (osteoporosis); these fractures are most typical in hip, spine and forearm. Such fractures are associated with morbidity and mortality. [2] The projected number of osteoporotic hip fractures worldwide by the year 2025 will be 2.78 millions in women and 1.16 million in men; this will be most apparent in Asia. [3] In India, accurate epidemiological data regarding osteoporosis is not available, but it is frequent in India and is common for medical practitioners to see many patients suffering from bone fractures and other complications due to the disease. [4]

Several quantitative or semi-quantitative techniques are available for the measurement of bone mass or BMD; these ranged from simple conventional radiography to sophisticated photon absorptiometry. [5] Measurements of periosteal and endosteal width of tubular bone reflect the rate of bone formation and bone resorption respectively with age. Changes in cortical width show a high degree of correlation with changes in cortical bone mass. One way to measure these widths in any tubular bone is by the radiogrammetry method.6 Clavicle radiogrammetry using a conventional chest radiograph has been found to be very useful in evaluating generalised osteoporosis. [7] Dual energy X-ray absorptiometry (DXA) has been accepted as the "gold standard" for measuring BMD at hip, spine, forearm and total body. [8] In India, DXA densitometers are not widely available and are expensive; only a few centres have obtained such equipment, all of them since 1997. Conventional radiography on the other hand, is widely available and is inexpensive.

The prevention of osteoporotic fractures with consequent reductions in health care costs and excess morbidity and mortality is an important clinical goal. The relevance of detecting patients with osteoporosis before they fracture is well recognised as there are several drugs for prevention. In this context, it would seem appropriate to consider a screening programme to identify individuals with osteoporosis or low bone mass (osteopenia). If such a programme were to be implemented, the screening technique should be a "reliable, easily administered and inexpensive measure of bone mass that has practical significance". Advanced techniques like DXA, quantitative computed tomography (QCT) and quantitative ultrasound (QUS) do not fulfil all the above suggested criteria, especially the cost factor.

This study had two objectives. The first was to assess the correlations among the bone mass measurements by clavicle radiogrammetry and proximal femur BMD by DXA. The second was to establishan empirical link between total hip BMD by DXA and a clavicle radiogrammetry made from the conventional chest radiograph that gives results of high sensitivity and specificity for predicting total hip BMD from chest radiograph for evaluation of osteoporosis.

MATERIAL AND METHODS

Subjects

Fifty women whose age ranged from 16 to 84 years were studied. Of the 50 women, five had previous fractures (three had vertebral fracture and two hip fracture). These women were drawn randomly from general medical practices at KJ Hospital, Chennai, South India, during November, 1997 to April, 1998; patients with secondary bone diseases were excluded. Women were divided into sub-groups as follows : i) Pre-menopausal females (Group I) (n=21, M=31.0 æ 8.8 years), ii) Post-menopausal females (n=29, M=64.3 æ 11.2 years).

In this study, menopause was defined as the absence of menses for at least six months in women over the age of 35 years when other secondary causes of amenorrhoea were ruled out. Of the various DXA region of interest selected in the proximal femur scan, only total hip BMD was used for classifying patients according to the diagnostic criteria for osteoporosis given by WHO.9 The classification by T-score was done using a mean value of 0.965 g cm-2 and a population SD of 0.088 g cm-2 for total hip BMD in young normal South Indian women aged 25-34 years, which was established at our centre in our earlier study.5 The post-menopausal females were sub-divided as follows :

Group-II : Normal post-menopausal females (n=6, M=57.8 æ 9.6 years)

Group-III : Those with osteopenia (n=6, M=64.4 æ 10.5 years)

Group-IV : Those with osteoporosis, but nofracture (n=12, M=67.3 æ 7.0 years)

Group-V : Those with established osteoporosis including fracture (n=5, M=72.4 æ 7.3 years)

Group-VI : A combined osteoporosis group including osteoporotic women with and without fracture (n=17, M=68.8 æ 7.2 years).

Measurements

BMD of the right proximal femur was measured by DXA using a Hologic QDR-1000 densitometer (Hologic Inc., Waltham, Massachusetts; USA). PA view chest radiograph was taken in all the cases. From the radiograph the length of the right clavicle was measured and the mid-point identified. At this point the outer periosteal width and inner endosteal width were measured using a needle-tipped varnier calliper with a readout of 0.01 cm and combined cortical thickness (CCT) and % CCT of the clavicle were calculated.

Statistics

Data analysis was done with the SPSS/PC statistical software package. Mean æ SD values of the measurements were calculated in each group. Association between the variables was investigated by correlation and multiple linear regression analysis. Student’s t-test was used to compare the mean æ SD values of each variable between groups. A paired t-test was used to test the significance of any difference between the observed and predicted values.

RESULTS

Osteoporosis

Table 1 gives the mean æ SD values of all the measured variables studied for the various groups mentioned earlier. In this study, it was found that 58.6% (17/29) of the post-menopausal women were diagnosed as having osteoporosis and 20.7% (6/29) of the post-menopausal women were diagnosed as having osteopenia.

Comparison of healthy post-menopausal women with other groups

The healthy post-menopausal women (Group II) was compared with Groups I, III, IV, V and VI respectively. For these various groups comparison, decrements of the measured variables which showed a statistically significant difference were calculated and were expressed as a percentage relative to the initial value (Table 2). In the healthy post-menopausal women (Group II) the mean values of CCT and % CCT of the clavicle were significantly less than in the pre-menopausal women (Group I). In all osteoporotic women with and without fracture (Group VI), the mean values of body weight, body mass index, total hip BMD, CCT and % CCT of the clavicle were significantly less than in those of Group II. Of these variables, the percentage decrements were greatest for CCT and % CCT of the clavicle (32.7% and 34.0% respectively).

Correlation

In 45 pre-and post-menopausal women without osteoporotic fracture, there was a significant (p < 0.001) negative correlation between age and the following variables: total hip BMD, CCT and % CCT of the clavicle (r=-0.64; r=-0.80 and r=-0.82 respectively). Also, total hip BMD was correlated significantly (p < 0.001) with CCT and % CCT of the clavicle (r=0.73 and r=0.74 respectively). Thus the total hip BMD measured by DXA was correlated significantly with patient’s age, CCT and % CCT of the clavicle. The above type of correlations are independent of each variables considered. Hence an attempt was made to adopt a multiple linear regression analysis incorporating all the above mentioned variables in such a way as to predict BMD with good sensitivity and specificity.

KJH; Anburajan’s empirical formula for predicting total hip BMD Clavicle CCT in cm (X2), clavicle % CCT (X1) and patient’s age in years (X3) were used as the independent variables in a multiple linear regression analysis to predict total hip BMD in g cm-2 (Y, the dependent variable) using the following equation:

(a) Y = A + B1.X1 + B2.X2 + B3.X3 The value obtained for the coefficients were : A = 0.505563, B1 = 0.003938, B2 = 0.370914 and B3 = -5.05139 x 10-4. Using this equation, total hip BMD (g cm-2) was predicted in each individual of the population studied. A paired t-test was used to compare the predicted total hip BMD calculated using the multiple linear regression equation (a) with the measured BMD using DXA in the population studied. There was no statistically significant difference between the two groups (t = -0.55, p < 0.6). Using the measured total hip BMD as the reference method, it was found that the multiple linear regression equation (a) had a low sensitivity of 76.5% and a specificity of 97.0% for diagnosing osteoporosis in the group of women studied.

In order to improve the sensitivity further the relative weights of the coefficients were changed by trial and error. Final values of A = 0.47 and B2 = 0.420914 increased the sensitivity of equation (a) to 82.4% with a specificity of 94.0%. A complete modified empirical formula derived in this study for predicting total hip BMD (Y) from clavicle radiogrammetry measurements is termed as "KJH; Anburajan’s empirical formula" and is given as follows :

(b) Y = 0.47 + 0.003938 (X1) + 0.420914 (X2) - 5.05139 x 10-4 (X3)

Where X1 = % CCT of the clavicle, X2 = CCT of the clavicle (cm), X3 = patient’s age (years). Using this formula, total hip BMD (g cm-2) was predicted in each individual of the population studied. Again a paired t-test was used to compare the predicted total hip BMD (calculated using the formula (b) mentioned above) with the measured BMD using DXA in the population studied. It was found that there was no statistically significant difference between the two groups (t=0.27, p < 0.08). The positive predictive value of the KJH; Anburajan’s empirical formula (b) was 87.5%, whereas its negative predictive value was 91.2%.

Validity of the KJH; Anburajan’s empirical formula

The validity of the formula was tested in an

TABLE 1

The characteristics of the individuals studied

Post-menopausal women (n=29)

Variables Pre-menopausal Osteoporosis Osteoporotic Student’s t-test women Normal Osteopenia (without Osteoporotic with and ‘p’ value Group I Group II Group III fractures) fracture without Group IV Group V fracture Group VI

Number of cases 21 6 6 12 5 17

Age (years) 30.9 ‘ 8.8 57.8 ‘ 9.6 64.4 ‘ 10.5 67.3 ‘ 7.0 72.4 ‘ 7.3 68.8 ‘ 7.2 (a,d) : p < 0.01 (a) (b) (c) (d,f) (e) (b,f) : NS (c) : p < 0.05 (e) : P < 0.001

Age range 16 to 46 45 to 70 52 to 76 57 to 80 65 to 82 57 to 82(years)

Body height 150.4 ‘ 6.6 152.3 ‘ 2.9 150.00 ‘ 5.87 152.50 ‘ 5.81 146.0 ‘ 8.5 150.6 ‘ 7.1 (a) to (e) : NS(cm) (a) (b) (c) (d,f) (e)

Body weight 48.7 ‘ 7.6 64.50 ‘ 10.77 51.17 ‘ 15.33 50.25 ‘ 10.55 47.00 ‘ 3.90 49.3 ‘ 9.1 (a,d,e) : p < 0.01 (kg) (a) (b) (c) (d,f) (e) (b,f) : NS (c) : p < 0.02

Body mass 21.5 ‘ 3.2 27.79 ‘ 4.56 22.57 ‘ 5.89 21.54 ‘ 4.14 22.2 ‘ 2.8 21.7 ‘ 3.7 (a) : p < 0.01Index (Kg m2) (a) (b) (c) (d,f) (e) (b,f) : NS (c) : p < 0.02 (d) : p < 0.05 (e) : p < 0.01

Time since — 5.5 ‘ 5.3 12.2 ‘ 10.4 17.3 ‘ 6.9 22.4 ‘ 7.3 18.8 ‘ 7.2 (b,f) : NSmenopause (b) (c) (d,f) (e) (c,d,e) p < 0.01(years)

DXA Hip Densitometry

Total hip 0.942 ‘ 0.089 0.928 ‘ 0.041 0.794 ‘ 0.032 0.657 ‘ 0.067 0.646 ‘ 0.055 0.654 ‘ 0.062 (a,f) : NSBMD(g cm-2) (a) (b) (c) (d,f) (e) (b) : < 0.01 (c,d,e) : p < 0.001

Clavicle Radiogrammetry

CCT (cm) 0.58 ‘ 0.09 0.49 ‘ 0.05 0.42 ‘ 0.08 0.31 ‘ 0.09 0.36 ‘ 0.07 0.33 ‘ 0.09 (a) : p < 0.05 (a) (b) (c) (d,f) (e) (b,f) : NS (c,e) : p < 0.001 (d) : p < 0.01

% CCT 60.13 ‘ 7.05 50.92 ‘ 12.17 46.02 ‘ 8.11 32.25 ‘ 9.44 34.86 ‘ 9.01 33.60 ‘ 10.0 (a) : p < 0.05 (a) (b) (c) (d,f) (e) (b,d,f) : NS (c,e) : p < 0.01

Where NS - Not significant

(a) Group - I Vs Group - II; (c) Group - II Vs Group - IV; (e) Group - II Vs Group - VI

(b) Group - II Vs Group - III; (d) Group - II Vs Group - V; (f) Group - IV Vs Group – V

TABLE 2

Percentage decrements of the mean values of variables studied in the inter-group analysis*

Variables Percentage decrements

Group I Vs Group II

CCT of clavicle 15.5

%CCT of clavicle 15.3

Group II Vs Group III

Total hip BMD (g cm-2) 14.4

Group II Vs Group IV

Body weight (kg) 22.1

Body mass index (kg m-2) 22.5

Total hip BMD (g cm-2) 29.2

CCT of clavicle 36.7

%CCT of clavicle 36.7

Group II Vs Group V

Body weight (kg) 27.1

Body mass index (kg m-2) 20.1

Total hip BMD (g cm-2) 30.4

CCT of clavicle 26.5

%CCT of clavicle 31.5

Group II Vs Group VI

Body weight (kg) 23.6

Body mass index (kg m-2) 21.9

Total hip BMD (g cm-2) 29.5

CCT of clavicle 32.7

%CCT of clacvicle 34.0

All the variables analysed were found to have a statistically significant difference when assessed by student’s t-test as mentioned in Table 1.

Group I - Pre-menopausal women; Group II - Normal post-menopausal women; Group III - Osteopenic women; Group IV-Osteoporotic women without fracture; Group V - Osteoporotic fracture women; Group VI - Osteoporotic women with and without fractures.

TABLE 3

Measured variables in osteoporotic women with and without fractures (Group - VI) in relation to pre-menopausal women (Group - I)

Group - VIOsteoporotic women with and withoutfractures (n=17, mean age = 68.8 ‘ 7.2 years)

Variable (Method) Deviation (in SD Number of women > mean pre-menopausal pre-menopausal women women (%)

Total hip BMD (DXA) -3.2 58.8

CCT of clavicle -2.7 52.9(Clavicle Radiogrammetry)

%CCT of clavicle -3.7 70.6(Clavicle Radiogrammetry)

tion studied. There was no statistically significant difference between the two groups (t = -0.55, p < 0.6). Using the measured total hip BMD as the reference method, it was found that the multiple linear regression equation (a) had a low sensitivity of 76.5% and a specificity of 97.0% for diagnosing osteoporosis in the group of women studied.

In order to improve the sensitivity further the relative weights of the coefficients were changed by trial and error. Final values of A = 0.47 and B2 = 0.420914 increased the sensitivity of equation (a) to 82.4% with a specificity of 94.0%. A complete modified empirical formula derived in this study for predicting total hip BMD (Y) from clavicle radiogrammetry measurements is termed as "KJH; Anburajan’s empirical formula" and is given as follows :

(b) Y = 0.47 + 0.003938 (X1) + 0.420914 (X2) - 5.05139 x 10-4 (X3)

Where X1 = % CCT of the clavicle, X2 = CCT of the clavicle (cm), X3 = patient’s age (years). Using this formula, total hip BMD (g cm-2) was predicted in each individual of the population studied. Again a paired t-test was used to compare the predicted total hip BMD (calculated using the formula (b) mentioned above) with the measured BMD using DXA in the population studied. It was found that there was no statistically significant difference between the two groups (t=0.27, p < 0.08). The positive predictive value of the KJH; Anburajan’s empirical formula (b) was 87.5%, whereas its negative predictive value was 91.2%.

Validity of the KJH; Anburajan’s empirical formula

The validity of the formula was tested in an other sample of 3 post-menopausal women (whose ages were 45, 51 and 56 years). In subject (i), the predicted value of total hip BMD in g cm-2 was (0.47 + 0.003938 (41.76) + 0.420914 (0.38) - 5.05139 x 10-4) = 0.722, whereas the measured value using DXA was 0.841 g cm-2. The difference between the measured and predicted values was 0.069 g cm-2 (8.2%). In subject (ii), the predicted value of total hip BMD in g cm-2 was (0.47 + 0.003938 (50.54) + 0.420914 (0.47) 5.05139 x 10-4 (51)) = 0.841, whereas the measured value using DXA was 0.870 g cm-2. The difference between the measured and predicted values was 0.029 g cm-2 (3.3%). In subject (iii), the predicted value of the total hip BMD in g cm.-2 was (0.47 + 0.003938 (55.44) + 0.420914 (0.49) - 5.05139 x 10-4 (56)) = 0.866, whereas the measured value using DXA was 0.787 g cm-2. The difference between the measured and predicted values was 0.079 g cm-2 (-9.1%). Thus, the difference between the predicted total hi BMD using the formula and the measured values by DXA ranged from -9.1% to +8.2%.

In each of these three women, the predicted BMD using KJH; Anburajan’s empirical formula was in the range 1.0 to 2.5 SD below the mean value for young adult women. The measured BMD using DXA for all three subjects was also within this range. According to the criteria of the WHO (1994), all three women were diagnosed as having osteopenia using KJH; Anburajan’s empirical formula’s prediction (from clavicle radiogrammetry measurements) and DXA measurements. In the present study, the number of cases studied for confirmation was limited. However, we are in the process of instituting an ongoing trial with larger number of patients at present. The KJH; Anburajan’s empirical formula will be useful in predicting total hip BMD from the conventional chest radiograph even in larger samples.

DISCUSSION

In the study, the mean body weight of the osteoporotic women was significantly less (p < 0.01) than in the health post-menopausal group. This is consistent with the observation that in women, body weight is the principal determinant of bone mass; the greater the body weight, the greater is the bone mass.10 Gupta and co-workers11 reported that radiological evidence of osteoporosis was present in 89% and 18% patients, aged above and below 50 years respectively. Vaishnava and co-workers12 reported that out of 421 patients with hip fracture studied, 33.5% had osteoporosis based on conventional X-ray and iliac crest biopsies; out of these, 10% patients were below the age of 50 years and the remaining 67% patients were over 70 years of age. In this study, it was found that 20.7% (6/29) and 58.6% (17/29) of the South-Indian post-menopausal women were diagnosed as having osteopenia and osteoporosis respectively based on DXA hip bone densitometry and their age ranged from 52-76 years and 57-82 years respectively.

Clavicle radiogrammetry made from the chest radiograph has been shown to be useful in evaluating generalised osteoporosis.7 In this study, the mean values of CCT and % CCT of the clavicle in all osteoporotic women (Group VI) were significantly less than in the healthy post-menopausal women (Group II). The mean values of DXA and clavicle radiogrammetry measurements in all osteoporotic women (Group VI) were compared to those in the pre-menopausal women (Group II). It was found that the average difference (in SD units) from the peak value (obtained in the young adult age) was greatest for clavicle radiogrammetry i.e., -3.7 for % CCT. Furthermore, the percentage of all osteoporotic women who were more than 3.0 SD below the mean value for pre-menopausal women was only 59% for total hip BMD by DXA; but 71% for % CCT of the clavicle. Kleerekoper and co-workers13 reported that BMD of phalanges by radiographic absorptiometry showed the greatest average difference (-2.71 SD) from the peak value when compared to BMD of the radius, lumbarspine and femoral neck by DXA and lumbar spine BMD by QCT in women.

In this study, it was found that total hip BMD by DXA was correlated significantly (p < 0.001) with CCT and % CCT of the clavicle in 45 pre-menopausal and post-menopausal women without fractures of the bone. To the best of our knowledge, this study is the first on correlations between total hip BMD by DXA, CCT and % CCT of the clavicle. The evolved KJH; Anburajan’s empirical formula can be used to predict total hip BMD from simple clavicle radiogrammetry indices (measured from chest radiograph) with 82.4% sensitivity and 94.0% specificity. The positive and negative predictive value of the formula was found to be 87.5% and 91.2% respectively. Moreover the PA view chest radiograph, which is used for clavicle radiogrammetry is a standard radiological procedure and is available quite easily in many institutions in India.

In India, women are living longer and many upto the age of 75 years. The population of post-menopausal women in India, which was below 7-8 crores in 1990 is expected to increase to 10 crores by 2000 and to 20 crores by 2030.14 Accurate epidemiological data regarding osteoporosis in India is not available, but in this study it was found that 58.6% and 20.7% of the South Indian post-menopausal women studied were diagnosed as having osteoporosis and osteopenia respectively. This proportion may increase steeply in a decade due to malnutritional diet and lower physical activity thanks to industrialisation and the modern life style of most of the individuals.1 As a result, a screening programme for detecting low bone mass should be considered in India. Advanced sophisticated techniques like DXA are costly and not widely available. Moreover chest radiograph is being taken as a primary investigation for many diseases even at rural medical centre. At this juncture, the evolved KJH; Anburajan’s empirical formula is used to analyse the chest radiograph and thereby evaluate osteoporosis in women with good sensitivity and specificity; So that, it could be used as a mass screening tool for the disease in developing countries like India. An automated computerised clavicle radiogrammetry can also be performed and thus making the process more precise.

CONCLUSION

There is a pressing need for a low cost mass screening tool for evaluating osteporosis in post-menopausal women especially in developing poor countries like India. It is shown that a widely available and inexpensive standard chest radiograph in combination with KJH; Anburajan’s empirical formula established at our centre would give a reliable prediction of bone mineral density with the good sensitivity and specificity in the evaluation of post-menopausal osteoporosis. This serves the purpose instead of going in for a costly gadget like DXA which is not widely available and not affordable in our country.

REFERENCES

1. Teotia SPS, Teotia M. Osteoporosis-India : emerging message. JAPI 1996; 11 : 759-60.

2. Passariello R, Albanese CV, Kvasnova. Bone densitometry in the clinical practice. Eur Radiol 1997; 7 : S2-S10.

3. Cooper C, Campion G, Melton LJ 111. Hip fractures in the elderly : a world-wide projection. Osteoporosis Int 1992; 2 : 285-9.

4. Sastry NV, Sridhar GR, Reddy GN, Davidraju S, Madhavi GV, Nagamani G. Evaluation of osteoporosis in patients with fractures neck of femur using conventional radiography. JAPI 1992; 42 : 209-11.

5. Anburajan M. Evaluation of osteoporosis using conventional radiographic methods and dual energy X-ray absorptiometry. Ph. D. Thesis, 1999; Anna University, Chennai, India.

6. Barnett E, Nordin BEC. The radiological diagnosis of osteoporosis : a new approach. J Clin Radiol 1960; 11 : 166-74.

7. Anton HC. Width of clavicular cortex in osteoporosis. Br Med J 1969; 1 : 409-11.