JOURNAL OF BONE AND MINERAL RESEARCH Volume 6, Number 3, 1991 Mary Ann Liebert, Inc., Publishers
Classification of Vertebral Fractures RICHARD EASTELL,' SANDRA L. CEDEL,' HEINZ W. WAHNER,3 B. LAWRENCE RIGGS,' and L. JOSEPH MELTON 1112
ABSTRACT Although i t i s a cardinal feature of involutional osteoporosis, there i s often disagreement o n what constitutes a vertebral fracture. We measured vertebrae T4-L5 in 52 healthy women t o develop a normal range (mean f 3 SD) f o r vertebral shape and used these data t o assess the prevalence o f vertebral fractures. We classified vertebral fractures by type o f deformity (wedge, biconcavity, o r compression) and further by the degree of deformity (grades 1and 2). In 195 postmenopausal women who were a n age-stratified random sample o f the Rochester population (ages 47-94), 40 (21%) had vertebral fractures (mean, 2 per person). There was a similar number o f compression and wedge fractures, and grade 2 fractures were as common as grade 1. I n a referral sample o f 74 women with suspected osteoporosis, 62 (84%) had vertebral fractures (mean, 3.3 per person). Wedge fractures were most common, and grade 2 fractures were more common than grade 1. The distribution o f type and grade o f fractures differed between the two patient groups ( P < 0.01). Bone mineral density o f the lumbar spine was related to mean fracture grade ( r = -0.33, P < 0.05) and to fracture number ( r = -0.57, P < 0.001) but not to fracture type. We conclude that a comprehensive approach i s required in describing vertebral fractures. Using this approach we found distortion in the fracture characteristics o f women referred to an osteoporosis clinic compared to women in the community.
INTRODUCTION
(1) some classifications include only biconcavity deforrnityl6 ') or wedge deformity" ' I ; ( 2 ) one system is based on
is a cardinal manifestation of osteoporosis.L1) However, about one-half of such fractures may be asymptomatic, and there is frequent disagreement about whether a fracture is present in mild cases, especially if no previous radiograph is available for comparison. Determining whether a fracture is present is important both on clinical grounds and for research purposes, for example in studying the incidence and prevalence of osteoporosis in the population or as an entry criterion to a clinical trial. The difficulty in deciding whether a vertebra is indeed fractured results from variation in shape from one vertebra to another and also between individuals. This problem has been approached by measuring vertebrae from healthy subjects to establish norms.'z-B) These reports of radiographic standards have the following limitations, however:
comparisons with one vertebra (T4), which itself may be fractured'R';(3) some systems are based on absolute height of vertebra," 5 1 but this is dependent on body height of the individual'5'; (4) the degree of deformity may have been selected arbitrarily'9-"); and ( 5 ) the system of classification may be difficult to apply in practice, such as the calculation of wedge angle.1z' The aims of this study were ( I ) to establish normal ranges for vertebral shape; ( 2 ) to develop a comprehensive classification system that includes the type, degree, and number of deformities; (3) to test the relative importance of these characteristics in predicting lumbar spine bone mineral density; and (4) to use this classification system to compare the fractures of women found by screening the population with those of women referred to an outpatient clinic for metabolic bone disease.
V
ERTEBRAI. FRACTURE
~
'Endocrine Research Unit, Mayo C h i c and Foundation, Rochester MN 55905. 'Department of Health Sciences Research, Mayo Clinic and Foundation, Rochester MN 55905. 'Section of Diagnostic Nuclear Medicine, Mayo Clinic and Foundation, Rochester. MN 55905.
207
~~
~
~
208
EASTELL ET AL.
MATERIALS AND METHODS
Population-based sample We studied 195 postmenopausal women from an agestratified random sample of the Rochester population.('*' The original sample size was 300, but only the 200 women age 50 years and over (or postmenopausal) had radiographs taken of the thoracic and lumbar spine; 5 of the latter group were excluded from this analysis because they had vertebral fractures known to have resulted from significant trauma. The women were ages 47-94 years (mean, 69.7; SD, 11.9). The "normal range" for vertebral shape was established from a subset of 52 women ages 47-87 years (mean, 62.7; SD, 9.4) who met the following criteria: no radiologic evidence for fracture (based on a reading by a radiologist); no use of corticosteroids, anticonvulsant medication, thiazide diuretics, vitamin D in pharmacologic doses, or calcium supplements of more than 500 mg/day; no disease known to influence calcium metabolism; and lumbar spine bone mineral density as expected for age.
Referral sample We also studied 74 women referred for evaluation of postmenopausal osteoporosis to the Mayo Clinic. Their ages ranged from 54 to 75 years (mean, 57.3; SD, 5.2). The bone density values from this group have been reported For the ROC analysis (receiver operating characteristic; see subsequent discussion) they were com-
pared with a group of 42 healthy postmenopausal women, ages 50-75 years (mean, 61.2; SD, 7.7 ), who were free of disease and were not taking any drugs known to affect calcium metabolism. These healthy women were recruited in response to a call for volunteers in the local press.
Measurements Lumbar spine bone mineral density (LS-BMD) was meas ~ r e d " ~for ) vertebrae L2-L4 in all subjects by dual-photon absorptiometry (reproducibility, 2010). Anteroposterior and lateral radiographs of the thoracic and lumbar spine were taken at a standard target-to-film distance of 105 cm. Vertebrae T4-LS on the lateral radiographs were marked with a pencil to allow measurement with a transparent ruler to the nearest mm of anterior (ha) and posterior (hp) heights (Fig. 1). Middle height (hm) was the average of right and left middle heights, and these were identified by examination of anteroposterior as well as the lateral views. One of the authors (Eastell) supervised the marking of x-rays to ensure consistency of approach. The anterior margins were marked according to the method of Jensen and Tougaard.(3)This involved an imaginary line drawn through the midpoints of the anterior surfaces of the vertebrae. The anterior markers were placed where this line intersected the superior and inferior end plates. The posterior markers were made at the base of the posterior angulation of the superior end plate. Care was taken to make sure the posterior markers lined up. The coefficients of variation for ha and hp were 4.7 and 6.3010, respectively.
CLASSIFICATION OF VERTEBRAL FRACTURES Wedge delormlty
Blconcavlty deformity
Grad. 1 4.0
ha < 4.0 hp
Grade 1
Grade 1
< k? < 3.0 SO hp
4.0
hm
< 6< 3.0 SD
Grade 2 hp < 4.0 SD
Compresslon deformlty
4.0
ha) to -15% for vertebra L5 (ha > hp), and the range of S D for wedge deformity was 4.7% (T4) to 7.4% (L4). To estimate the false positive rate for fracture we used the data from the 52 normal women in the diagnostic algorithm, and six vertebral fractures were identified, one in each of six women. To estimate the false negative rate we applied this definition of fracture to the 74 referred patients, and only 62 had one or more vertebral fractures. In the remaining 12 symptoms consistent with vertebral fracture syndrome or progression of spinal deformity were seen in 2. The remaining 10 were considered to have fractures by the radiologic appearance only (without the use of an objective standard) and were asymptomatic.
Characteristics of fractured vertebrae Using the criteria in Table 2 for fracture, 40 women (21%) of the population-based sample had a total of 81 fractured vertebrae (mean, 2.0 per fracture case) and 62 of the 74 women (84%) from the referral sample had a total of 202 fractured vertebrae (mean, 3.3 per fracture case).
TABLE 1. EFFECTOF DECISION THRESHOLD ON THE NUMBEROF WOMENIN THE POPULATION-BASED TO HAVESIGNIFICANT SAMPLE CONSIDERED VERTEBRAL DEFORMITIES ( n = 195)
Threshold 2SD 2.5SD 3SD 3.5SD 4SD 15% 25 qo
Number with vertebral fracture (Vo) 128 (67) 73 (37) 40 (21) 29 (IS) 22 (11) 49 (25) 19 (10)
EASTELL ET AL.
210 The type of fractures differed between the two groups in that the population-based sample had fewer wedge and biconcavity fractures. This was most apparent when the number of fractures of a given type per patient with fracture was calculated for the two groups (Table 3). The figure for compression fractures was similar, but the number of wedge and biconcavity fractures per patient was twice as
high in the referral group as in the population-based group. The grade of deformity also differed between the two groups: the referral group were more likely to have grade 2 fractures (64% versus 48% in the population-based sample). This difference between groups in fracture type and grade was statistically significant (chi-square, 5.2, P < 0.05; Table 3). However, if patients who were younger
TABLE2. DECISION THRESHOLDS DEFINING GRADE1 AND GRADE2 VERTEBRAL DEFORMITY~
Decision thresholds for fracture (% deformityjb Wedge Vertebra
T4 T5 T6 T7 T8 T9 TI0 TI 1 TI2 L1 L2 L3 L4 L5
Biconcavity
Compression
Grade I
Grade 2
Grade I
Grade 2
Grade I
Grade 2
17 22 26 28 25 24 22 27 21 20 18 13 15 3
22 28 32 35 31 30 29 34 27 26 24 18 22 8
17 21 23 24 21 18 20 21 19 20 20 16 15 24
22 27 29 30 25 22 24 25 22 25 25 20 21 29
20 19 18 17 21 20 22 20 20 16 17 14 26 11
25 25 23 22 26 25 26 24 26 21 22 20 31 18
aUsing 3.OSD as the criterion for a grade 1 fracture and 4.OSD as the criterion for a grade 2 fracture. bValues greater than this indicate fracture. CFor cranial compression, hp' is posterior height of vertebra above; for caudal compression, hp' is posterior height of vertebra below.
TABLE3. TYPE,GRADE,AND NUMBER OF VERTEBRAL FRACTURES AMONG PATIENTS WITH FRACTURE
Population-based group (n Grade 1 2 Total @lo of Total No. fractures per person
1
2 Total Vo of Total No. fractures per person
40)
Compression
Wedge
Biconcavity
Total
Yo of Total
15 16 31 38 0.78
19 16 35 43 0.88
8 7 15 19 0.38
42 39 81
52 48
Referral group ( n Grade
=
=
62)
Compression
Wedge
Biconcavity
Total
Yo of Total
23 30 53 26 0.85
24 77 101 49 1.63
28 25 53 26 0.85
75 132 207
36 64
211
CLASSIFICATION OF VERTEBRAL FRACTURES than 75 years were compared, there was no difference between the groups. The distribution of fractures by vertebrae from T4 to L5 was similar in the two groups (Fig. 2).
Fracture characterislics of individuals The combinations of fracture types in the same individual are shown in Table 4. The referral group differed in that the combinations that included wedge deformity (85% of fractures) were more common than in the populationbased group (56% of fractures). In the population-based group 13 individuals had only one grade 1 fracture (Table 5 ) . These may represent false positives because we would expect 12% false positives due
T-4 T-5 T-6 T-7
-
to the problem of calculating 42 deformity values for each individual (see earlier discussion). However, the mean LSBMD value (0.94 g/cmz; SD, 0.12) in these 13 women was lower than that of the 155 women without fracture (1.00 g/cmz; SD, 0.12), although this was not statistically significant. The effect of age on fracture grade in the populationbased sample is shown in Table 6. In the 195 women from the population-based sample there were positive correlations between age and both the number of fractures per person ( r = 0.35; P < 0.OOOl) and the mean grade of fracture ( r = 0.25, P < 0.001); that is, older women had more fractures, and these were more likely to be grade 2 than in younger women.
T-4
T-5
T-6
T-7
T-8
T-8
T-9
T-9
1 I I
T-10 T-11 T-12 L- 1
L-1
I C I W O B
L-2
L-2
L-3
L-3
L-4
L-4
L-5
L-5 0
2
6
4
8
10
0
2
%
6
4
8
10
YO
FIG. 2. Type and level of vertebral fracture: (Left) 40 women from the population-based sample; (right) 62 women from the referral sample.
TABLE4. COMBINATIONS OF TYPES OF FRACTURE AMONGINDIVIDUALS WITH ONE OR MOREVERTEBRAL FRACTURES Fracture type Compression alone Wedge alone Biconcavity alone Compression + wedge Compression + biconcavity Wedge + biconcavity All three types
Population- based n (%)
Referral
9 (23) 12 (30) 6 (15) 6 (15) 3 (8) 1 (3) 3 (8)
6 (10) 17 (27) 3 (5) 13 (22) 1 (2) 8 (13) 14 (23)
40 (100)
62 (100)
n (%l
212
EASTELL ET AL. TABLE 5. NUMBEROF FRACTURES BY GRADEWITHININDIVIDUALS
Population-based sample (n = 195), Grade I Grade 2
0
2
I
3+
Referral sample (n = 74), Grade I
I
0
2
3+
1 2 5
0 2 8
~~
155 7 6
0 1
2+
13 3 1
4 3 0
1 2 0
12 17 4
5 8 10
TABLE 6. DISTRIBUTION OF FRACTURES BY AGE AND GRADE
Population-based sample Age group
Classification of Vertebral Fractures RICHARD EASTELL,' SANDRA L. CEDEL,' HEINZ W. WAHNER,3 B. LAWRENCE RIGGS,' and L. JOSEPH MELTON 1112
ABSTRACT Although i t i s a cardinal feature of involutional osteoporosis, there i s often disagreement o n what constitutes a vertebral fracture. We measured vertebrae T4-L5 in 52 healthy women t o develop a normal range (mean f 3 SD) f o r vertebral shape and used these data t o assess the prevalence o f vertebral fractures. We classified vertebral fractures by type o f deformity (wedge, biconcavity, o r compression) and further by the degree of deformity (grades 1and 2). In 195 postmenopausal women who were a n age-stratified random sample o f the Rochester population (ages 47-94), 40 (21%) had vertebral fractures (mean, 2 per person). There was a similar number o f compression and wedge fractures, and grade 2 fractures were as common as grade 1. I n a referral sample o f 74 women with suspected osteoporosis, 62 (84%) had vertebral fractures (mean, 3.3 per person). Wedge fractures were most common, and grade 2 fractures were more common than grade 1. The distribution o f type and grade o f fractures differed between the two patient groups ( P < 0.01). Bone mineral density o f the lumbar spine was related to mean fracture grade ( r = -0.33, P < 0.05) and to fracture number ( r = -0.57, P < 0.001) but not to fracture type. We conclude that a comprehensive approach i s required in describing vertebral fractures. Using this approach we found distortion in the fracture characteristics o f women referred to an osteoporosis clinic compared to women in the community.
INTRODUCTION
(1) some classifications include only biconcavity deforrnityl6 ') or wedge deformity" ' I ; ( 2 ) one system is based on
is a cardinal manifestation of osteoporosis.L1) However, about one-half of such fractures may be asymptomatic, and there is frequent disagreement about whether a fracture is present in mild cases, especially if no previous radiograph is available for comparison. Determining whether a fracture is present is important both on clinical grounds and for research purposes, for example in studying the incidence and prevalence of osteoporosis in the population or as an entry criterion to a clinical trial. The difficulty in deciding whether a vertebra is indeed fractured results from variation in shape from one vertebra to another and also between individuals. This problem has been approached by measuring vertebrae from healthy subjects to establish norms.'z-B) These reports of radiographic standards have the following limitations, however:
comparisons with one vertebra (T4), which itself may be fractured'R';(3) some systems are based on absolute height of vertebra," 5 1 but this is dependent on body height of the individual'5'; (4) the degree of deformity may have been selected arbitrarily'9-"); and ( 5 ) the system of classification may be difficult to apply in practice, such as the calculation of wedge angle.1z' The aims of this study were ( I ) to establish normal ranges for vertebral shape; ( 2 ) to develop a comprehensive classification system that includes the type, degree, and number of deformities; (3) to test the relative importance of these characteristics in predicting lumbar spine bone mineral density; and (4) to use this classification system to compare the fractures of women found by screening the population with those of women referred to an outpatient clinic for metabolic bone disease.
V
ERTEBRAI. FRACTURE
~
'Endocrine Research Unit, Mayo C h i c and Foundation, Rochester MN 55905. 'Department of Health Sciences Research, Mayo Clinic and Foundation, Rochester MN 55905. 'Section of Diagnostic Nuclear Medicine, Mayo Clinic and Foundation, Rochester. MN 55905.
207
~~
~
~
208
EASTELL ET AL.
MATERIALS AND METHODS
Population-based sample We studied 195 postmenopausal women from an agestratified random sample of the Rochester population.('*' The original sample size was 300, but only the 200 women age 50 years and over (or postmenopausal) had radiographs taken of the thoracic and lumbar spine; 5 of the latter group were excluded from this analysis because they had vertebral fractures known to have resulted from significant trauma. The women were ages 47-94 years (mean, 69.7; SD, 11.9). The "normal range" for vertebral shape was established from a subset of 52 women ages 47-87 years (mean, 62.7; SD, 9.4) who met the following criteria: no radiologic evidence for fracture (based on a reading by a radiologist); no use of corticosteroids, anticonvulsant medication, thiazide diuretics, vitamin D in pharmacologic doses, or calcium supplements of more than 500 mg/day; no disease known to influence calcium metabolism; and lumbar spine bone mineral density as expected for age.
Referral sample We also studied 74 women referred for evaluation of postmenopausal osteoporosis to the Mayo Clinic. Their ages ranged from 54 to 75 years (mean, 57.3; SD, 5.2). The bone density values from this group have been reported For the ROC analysis (receiver operating characteristic; see subsequent discussion) they were com-
pared with a group of 42 healthy postmenopausal women, ages 50-75 years (mean, 61.2; SD, 7.7 ), who were free of disease and were not taking any drugs known to affect calcium metabolism. These healthy women were recruited in response to a call for volunteers in the local press.
Measurements Lumbar spine bone mineral density (LS-BMD) was meas ~ r e d " ~for ) vertebrae L2-L4 in all subjects by dual-photon absorptiometry (reproducibility, 2010). Anteroposterior and lateral radiographs of the thoracic and lumbar spine were taken at a standard target-to-film distance of 105 cm. Vertebrae T4-LS on the lateral radiographs were marked with a pencil to allow measurement with a transparent ruler to the nearest mm of anterior (ha) and posterior (hp) heights (Fig. 1). Middle height (hm) was the average of right and left middle heights, and these were identified by examination of anteroposterior as well as the lateral views. One of the authors (Eastell) supervised the marking of x-rays to ensure consistency of approach. The anterior margins were marked according to the method of Jensen and Tougaard.(3)This involved an imaginary line drawn through the midpoints of the anterior surfaces of the vertebrae. The anterior markers were placed where this line intersected the superior and inferior end plates. The posterior markers were made at the base of the posterior angulation of the superior end plate. Care was taken to make sure the posterior markers lined up. The coefficients of variation for ha and hp were 4.7 and 6.3010, respectively.
CLASSIFICATION OF VERTEBRAL FRACTURES Wedge delormlty
Blconcavlty deformity
Grad. 1 4.0
ha < 4.0 hp
Grade 1
Grade 1
< k? < 3.0 SO hp
4.0
hm
< 6< 3.0 SD
Grade 2 hp < 4.0 SD
Compresslon deformlty
4.0
ha) to -15% for vertebra L5 (ha > hp), and the range of S D for wedge deformity was 4.7% (T4) to 7.4% (L4). To estimate the false positive rate for fracture we used the data from the 52 normal women in the diagnostic algorithm, and six vertebral fractures were identified, one in each of six women. To estimate the false negative rate we applied this definition of fracture to the 74 referred patients, and only 62 had one or more vertebral fractures. In the remaining 12 symptoms consistent with vertebral fracture syndrome or progression of spinal deformity were seen in 2. The remaining 10 were considered to have fractures by the radiologic appearance only (without the use of an objective standard) and were asymptomatic.
Characteristics of fractured vertebrae Using the criteria in Table 2 for fracture, 40 women (21%) of the population-based sample had a total of 81 fractured vertebrae (mean, 2.0 per fracture case) and 62 of the 74 women (84%) from the referral sample had a total of 202 fractured vertebrae (mean, 3.3 per fracture case).
TABLE 1. EFFECTOF DECISION THRESHOLD ON THE NUMBEROF WOMENIN THE POPULATION-BASED TO HAVESIGNIFICANT SAMPLE CONSIDERED VERTEBRAL DEFORMITIES ( n = 195)
Threshold 2SD 2.5SD 3SD 3.5SD 4SD 15% 25 qo
Number with vertebral fracture (Vo) 128 (67) 73 (37) 40 (21) 29 (IS) 22 (11) 49 (25) 19 (10)
EASTELL ET AL.
210 The type of fractures differed between the two groups in that the population-based sample had fewer wedge and biconcavity fractures. This was most apparent when the number of fractures of a given type per patient with fracture was calculated for the two groups (Table 3). The figure for compression fractures was similar, but the number of wedge and biconcavity fractures per patient was twice as
high in the referral group as in the population-based group. The grade of deformity also differed between the two groups: the referral group were more likely to have grade 2 fractures (64% versus 48% in the population-based sample). This difference between groups in fracture type and grade was statistically significant (chi-square, 5.2, P < 0.05; Table 3). However, if patients who were younger
TABLE2. DECISION THRESHOLDS DEFINING GRADE1 AND GRADE2 VERTEBRAL DEFORMITY~
Decision thresholds for fracture (% deformityjb Wedge Vertebra
T4 T5 T6 T7 T8 T9 TI0 TI 1 TI2 L1 L2 L3 L4 L5
Biconcavity
Compression
Grade I
Grade 2
Grade I
Grade 2
Grade I
Grade 2
17 22 26 28 25 24 22 27 21 20 18 13 15 3
22 28 32 35 31 30 29 34 27 26 24 18 22 8
17 21 23 24 21 18 20 21 19 20 20 16 15 24
22 27 29 30 25 22 24 25 22 25 25 20 21 29
20 19 18 17 21 20 22 20 20 16 17 14 26 11
25 25 23 22 26 25 26 24 26 21 22 20 31 18
aUsing 3.OSD as the criterion for a grade 1 fracture and 4.OSD as the criterion for a grade 2 fracture. bValues greater than this indicate fracture. CFor cranial compression, hp' is posterior height of vertebra above; for caudal compression, hp' is posterior height of vertebra below.
TABLE3. TYPE,GRADE,AND NUMBER OF VERTEBRAL FRACTURES AMONG PATIENTS WITH FRACTURE
Population-based group (n Grade 1 2 Total @lo of Total No. fractures per person
1
2 Total Vo of Total No. fractures per person
40)
Compression
Wedge
Biconcavity
Total
Yo of Total
15 16 31 38 0.78
19 16 35 43 0.88
8 7 15 19 0.38
42 39 81
52 48
Referral group ( n Grade
=
=
62)
Compression
Wedge
Biconcavity
Total
Yo of Total
23 30 53 26 0.85
24 77 101 49 1.63
28 25 53 26 0.85
75 132 207
36 64
211
CLASSIFICATION OF VERTEBRAL FRACTURES than 75 years were compared, there was no difference between the groups. The distribution of fractures by vertebrae from T4 to L5 was similar in the two groups (Fig. 2).
Fracture characterislics of individuals The combinations of fracture types in the same individual are shown in Table 4. The referral group differed in that the combinations that included wedge deformity (85% of fractures) were more common than in the populationbased group (56% of fractures). In the population-based group 13 individuals had only one grade 1 fracture (Table 5 ) . These may represent false positives because we would expect 12% false positives due
T-4 T-5 T-6 T-7
-
to the problem of calculating 42 deformity values for each individual (see earlier discussion). However, the mean LSBMD value (0.94 g/cmz; SD, 0.12) in these 13 women was lower than that of the 155 women without fracture (1.00 g/cmz; SD, 0.12), although this was not statistically significant. The effect of age on fracture grade in the populationbased sample is shown in Table 6. In the 195 women from the population-based sample there were positive correlations between age and both the number of fractures per person ( r = 0.35; P < 0.OOOl) and the mean grade of fracture ( r = 0.25, P < 0.001); that is, older women had more fractures, and these were more likely to be grade 2 than in younger women.
T-4
T-5
T-6
T-7
T-8
T-8
T-9
T-9
1 I I
T-10 T-11 T-12 L- 1
L-1
I C I W O B
L-2
L-2
L-3
L-3
L-4
L-4
L-5
L-5 0
2
6
4
8
10
0
2
%
6
4
8
10
YO
FIG. 2. Type and level of vertebral fracture: (Left) 40 women from the population-based sample; (right) 62 women from the referral sample.
TABLE4. COMBINATIONS OF TYPES OF FRACTURE AMONGINDIVIDUALS WITH ONE OR MOREVERTEBRAL FRACTURES Fracture type Compression alone Wedge alone Biconcavity alone Compression + wedge Compression + biconcavity Wedge + biconcavity All three types
Population- based n (%)
Referral
9 (23) 12 (30) 6 (15) 6 (15) 3 (8) 1 (3) 3 (8)
6 (10) 17 (27) 3 (5) 13 (22) 1 (2) 8 (13) 14 (23)
40 (100)
62 (100)
n (%l
212
EASTELL ET AL. TABLE 5. NUMBEROF FRACTURES BY GRADEWITHININDIVIDUALS
Population-based sample (n = 195), Grade I Grade 2
0
2
I
3+
Referral sample (n = 74), Grade I
I
0
2
3+
1 2 5
0 2 8
~~
155 7 6
0 1
2+
13 3 1
4 3 0
1 2 0
12 17 4
5 8 10
TABLE 6. DISTRIBUTION OF FRACTURES BY AGE AND GRADE
Population-based sample Age group
