To determine whether such an injury leads to osteoporosis, we measured the ... follow-up is needed to determine any increased risk of osteoporotic fractures in ...
JOURNAL OF BONE AND MINERAL RESEARCH Volume 9 , Number 3,1994 Mary Ann Liebert, Inc., Publishers
Osteoporosis in Men with a History of Tibia1 Fracture PEKKA KANNUS,' MARKKU JARVINEN,2 HARRI SIEVANEN,' PEKKA OJA,' and ILKKA VUORI'
ABSTRACT The long-term effects of bone fractures on bone mineral density (BMD) at various skeletal sites are poorly established, although a serious fracture, such as a tibia fracture, followed by long immobility and disuse may lead to permanently decreased BMD and, through this mechanism, may be a risk factor for osteoporotic fractures in later life. To determine whether such an injury leads to osteoporosis, we measured the areal BMD (g/cm2)from the lumbar spine ( L M ) , right distal radius and ulna, and the femoral neck, distal femur, patella, proximal tibia, distal tibia, and calcaneus of both extremities in 14 men with a history of primarily nonunited (finally bone-grafted) shaft fracture of the tibia and 20 men with a history of primary union. For evaluation of the patients' BMD in the spine and distal forearm, 22 age-, weight-, and height-matched normal men were also measured. The average time of immobilization in a long plaster cast was 27 weeks in the former group of patients and 16 weeks in the latter. The measurements were performed an average of 9 years after the fracture using a dual-energy x-ray absorptiometric scanner. Compared with normal men (mean SD = 1.116 f 0.160), the spinal BMDs were significantly lower in men with a history of a primary nonunion (0.979 2 0.100, -12.3%) and union (1.010 0.124, -9.5%). In distal radius and distal ulna, there were no significant differences between the three groups. In the 14 patients with a primary nonunion, the mean BMD of the injured extremity (compared with the uninjured side) was significantly lower in the distal femur (- 10.0%), patella (- 11.2%), proximal tibia (-9.2%), distal tibia (-7.9%), and calcaneus (-5.6%). In the 20 patients with a primary union, the side-to-side difference was significant in the femoral neck (-2.4%), distal femur (-4.0%), patella (-3.7%), and proximal tibia (-5.1 %). The relative BMD of the injured extremity did not correlate with the fracture type, fracture location, or patients' age but did show significant (r = 0.41-0.79) positive correlation with short immobilization time, low pain assessment, good muscle strength, and high functional scores of the injured extremity. In conclusion, the tibia1 shaft fracture may be associated with reduced bone density in the lumbar spine and injured extremity. The observed decrease (1&12% in spine and 4 1 1 % in the knee region of the injured limb) is clinically important with respect to age-related bone loss of 1% per year after the age of peak bone mass. Additional follow-up is needed to determine any increased risk of osteoporotic fractures in spine and injured extremity.
*
*
INTRODUCTION
L
OSS OF BONE MINERAL is a
major risk factor for osteoporotic fractures among the elderly."-'*' Approximately 15-20 million adults in the United States are affected by the disease, with as many as 1.2-1.3 million osteoporotic fractures annually, making osteoporosis an important cause of morbidity and,
indirectly, mortality among these people.".'-' '.''"'' As the number of elderly people in the population continues to increase. the number of osteoporotic problems will increase as well. Many of the risk factors for osteoporosis have been documented in the literature, including positive family history, increasing age, white race, female sex, delayed puberty, various hormonal disturbances and nutritional deficiencies, low body
'UKK Institute for Health Promotion Research and the Tampere Research Station of Sports Medicine. Tampere, Finland. *Department of Surgery, Tampere University Hospital, Tampere. Finland. 423
424
KANNUS ET AL.
weight, smoking, excess alcohol consumption, many diseases and drugs, and physical inactivity."4.'6-2"' However, the generally recognized risk factors typically account for less than half of the variability in bone mass in different studies,(I2' and therefore it is very likely that many other, previously unidentified risk factors exist. The long-term effects of bone fractures on bone mineral density (BMD) at various skeletal sites are not well established, although an immediate decrease in bone mineral mass after fracture, sometimes noticeable even on a standard radiograph, is a well-known A serious fracture, such as a tibia fracture, followed by long-term immobility and disuse may lead to permanent osteoporosis and, through this mechanism, may be a risk factor for osteoporotic refractures in later life. In a 3 month follow-up study of seven patients with a tibial midshaft fracture, Ulivieri et at. observed that the immediate bone loss in distal tibia averaged as high as 50%.(23'In a 2 year follow-up study of 8 subjects, Andersson and Nilsson measured changes in the bone mineral content of the proximal tibia after a fracture of the tibia shaft.'22' The bone loss was approximately 25% after I year without evidence of further restoration. In a similar study of ankle fractures, Finsen and Benum observed that the average bone loss at 1 year was 4-9% without sign of regeneration during the second year.(2s' It would have been most interesting to know if these bone losses were indeed permanent and if there were losses in other bones (calcaneus, patella, femoral neck, and spine) as well. The objective of our study was therefore to examine with a precise and accurate method (dual-energy x-ray absorptiometry)'I2' the long-term effects of primarily united and nonunited (finally bone-grafted) shaft fractures of the tibia on the BMD of the injured extremity and the spine and to determine whether the results correlate with the patients' age, immobilization time, pain assessment, muscle strength, and functional scores. In addition, the effect of fracture type and location was studied.
MATERIALS AND METHODS Subjects We randomly selected 40 adult men (aged 18-40 years) with a tibial shaft fracture treated conservatively at the Tampere University Hospital during 1980-1986 to participate in this retrospective study. All of them were previously healthy without known disease or medication affecting bone metabolism, and 36 agreed to participate. Of the 36, 2 were excluded because they had received an intramedullary nailing procedure to the same tibia 2 and 7 months after the injury. The mean age of the remaining 34 patients at the time of follow-up was 38 2 7 (standard deviation, SD) years, ranging from 25 to 48 years. They had had 9 t 2 years earlier a unilateral shaft fracture of the tibia treated with a long plaster cast from the toes to the groin. The right limb was involved in 18 patients and the left limb in 16 patients. We divided these 34 men into two groups according to the primary healing of the fracture. The first group consisted of 20 men having a primary union after an immobilization period of 16 2 4 weeks. The second group consisted of 14 men with a primarily nonunited fracture after an immobilization period of
16 % 6 weeks. In this group, a bone-grafting procedure was performed after the primary immobilization period, and postoperatively all of the patients wore a plaster cast for 6-16 weeks. During this second period of immobilization, the fracture united in every case. Altogether, in the bone-grafted patients the injured limb was immobilized for 27 t 7 weeks. To obtain adequate control values for patients' spinal BMD, we recruited 22 age-, weight-, and height-matched healthy men for the study. Characteristics of the patients and these normal men are given in Table 1. There were no significant differences between the groups.
Study protocol Interview and Pain Assessment: All the subjects were fully informed of the study procedure and design, the purpose, and any known risks and gave their informed consent, The patient interview was based on our previously formatted questionnaire on knee ligament injuries.'26' With the aid of the questionnaire, the symptoms and present subjective state of the injured extremity and the current physical activity and occupational status of the patients were recorded in a standardized manner. Pain assessment during physical activity was performed using a I 0 0 mm visual analoq scale: no pain to extremely intense pain.'27' Functional Tests: In the functional evaluation the patient was asked to walk, duck walk, run, run in place, jump on one leg, do a half-squat, and squat fully. Along with information from the questionnaire, we calculated the functional scores of Lysholm and Gillquist,'2x' Tegner and L y ~ h o l m , ' ~and ~ ' the International Knee Documentation Committee (IKDC)'30' for each injured extremity. In the scoring system of Lysholm and Gillquist, the injured extremity is given a score from 0 to 100 (100 the best), in that of Tegner and Lysholm the scores range from 0 to 10 (10 the best), and in that of the IKDC system they range from 0 to 4 (4 the best). Clinical Examination: The clinical examination of the extremities was carried out in a standardized manner with special emphasis on the evaluation of mobility and stability of the hip, knee, and ankle joints.'26' Muscle Strength Measurement: The isometric extension strength was separately measured from the injured and healthy extremity using an isometric dynamometer (Tamtron, Inc.,
TABLE1. CHARACTERISTICS OF THE 34 PATIENTS AND 22 NORMALMEN^ Characteristic
Age, years Weight, kg Height, cm Body mass indexb
Patients
*7
38 83 t 179 t 25.9 t
12 7 3.3
Normal men
38 t 6 83 2 8 181 6 25.3 t 2.4
*
"Values are mean 2 SD. bBody mass index was calculated as weight in kilograms divided by the square of the height in meters.
TIBIAL FRACTURE AND OSTEOPOROSIS Tampere, Finland). The knee and ankle angles were fixed at 90". Three maximum efforts were allowed, and the median value was recorded. Bone Mineral Density Measurement: The bone mineral density (g/cm2) was measured from the lumbar spine (L2-4), right distal radius and ulna, and the femoral neck, distal femur, patella, proximal tibia, distal tibia, and calcaneus of both lower extremities using a Norland XR-26 dual-energy x-ray absorptiometric scanner (Norland, Inc., Fort Atkinson, WI; Fig. 1). All the measurements were performed by the same laboratory technician. Her personal day-to-day coefficient of variation for the repeated measurements of the same test persons has been shown to be low, between 0.7 and 1.9% depending on the anatomic site of measurement."'.'2' The coefficients for the sites used in this study were 1.7% for the lumbar spine, 0.7% for the distal radius and ulna, 1.3% for the femoral neck, 1.2% for the distal femur, 1.0% for the patella, 0.7% for the proximal tibia, 0.7% for the distal tibia, and 1.3% for the calcaneus.
Statistical analysis The data were analyzed by an IBM-compatible 386 microcomputer using the 1990 version of the statistical package of the BMDP.'~" Comparison of the BMD of the injured and uninjured sides was performed first by analysis of variance (Hotelling T' test) and then with the matched, paired ?-test. In the comparison of the groups with a primary union and nonunion with regard to the relative BMD of the injured side, Student's nonpaired t-test was used. The associations between age, immobilization time, pain assessment, muscle strength, the three functional scores, and the relative BMD in the injured extremity were analyzed by calculating the Pearson product moment correlation coefficient r for each comparison. The effect of fracture type and location on the relative BMD in the injured extremity was determined by an analysis of variance. In the spine, distal radius, and distal ulna, the absolute BMD between the normal men, men with a primarily united tibia fracture, and men with a primary nonunion were compared with analysis of variance. When this test indicated a significant difference (P < 0.05), Tukey's post-hoc analysis was performed.
425
*
The results are expressed as the mean standard deviation throughout the study. The given significance levels refer to two-tailed tests. An a level of less than 5% (p < 0.05) was considered significant. Preliminary calculation suggested that, to obtain a power of 80% in detecting a clinically important (in our estimation 5%) difference in BMD between the injured and uninjured extremity, significant at the a level, we would need 30 patients in the study.
RESULTS Lower extremities The absolute (g/cm2) and relative (side-to-side difference in percentage) BMD in the lower extremities are reported in Tables 2 and 3. For the 20 patients with the primarily united tibia1 shaft fracture, there were significant differences between the injured and uninjured extremity (Table 2). The mean BMD of the injured extremity was significantly lower in the femoral neck (-2.4%, P = 0.0241), distal femur (-4.0%, P = 0.0035). patella (-3.7%, P = 0.0089), and proximal tibia (-5.1%, P = 0.0053). In the distal tibia and calcaneus the difference was not significant. For the 14 patients with the primarily nonunited fracture, there were also clear side-to-side differences (Table 3). The mean BMD of the injured extremity was significantly lower in the distal femur (--lO.O%, P = 0.0001), patella (-11.2%, P = 0.0003), proximal tibia (-9.2%, P = 0.0005), distal tibia (-7.9%. P = 0.0022), and calcaneus (-5.6%, P = 0.0175). In the femoral neck the difference (-2.3%) was not significant. The relative BMD in the injured extremity differed between the groups of primarily united and nonunited fractures (Table 4). With the exception of femoral neck, the bone loss was greater in the latter group, the difference being significant in distal femur, patella, and distal tibia. Neither the fracture type (transverse, oblique, or comminuted) nor location (middle or lower third) affected the outcome. The patients' age showed no significant correlation with the relative BMD of the injured extremity (Table 5). However, immobilization time, pain assessment, muscle strength, and all
FIG. 1. Anatomic sites of the bone mineral density measurements in the lower extremities. The numbers refer to millimeters.
426
KANNUS ET AL.
TABLE2. BONEMINERAL DENSITY (G/CM2) IN THE INJURED AND UNINJURED LOWEREXTREMITIES OF 20 PATIENTS WITH PRIMARILY UNITED SHAFT FRACTURE OF THE TIBIA" Anatomic site
Injured
Uninjured
Difference
A
P valueb
% Difference ~~
0.893 -+ 1.312 1.137 0.958 0.882 0.667
Femoral neck Distal femur Patella Proximal tibia Distal tibia' Calcaneus
0.1 I 1
0.919 2 0.134 1.367 k 0.176 1.182 4 0.141 1.006 4 0. I38 0.902 2 0.153 0.678 t 0.093
* 0.185 * 0.136 * 0.166 * 0.122 * 0.107
-0.026 -0.055 -0.045 -0.048 -0.020 -0.011
t 0.046
-2.4 -4.0 -3.7 -5.1 -1.5 -1.8
0.074 2 0.068 t 0.069 2
f 0.064 f 0.028
0.0241 0.0035 0.0089 0.0053 NS NS
"Values are mean 2 SD. "The F value in the analysis of variance (the Hotelling T 2 test) was 2.93 with P = 0.0458. 'Only 16 patients were included in this comparison since in 4 patients the fracture line extended to the measuring area.
three functional scores showed a significant (P < 0.05-0.00l) correlation with the relative BMD (r = 0.41-0.79). The shorter was the immobilization time, the less activity-induced pain in the injured extremity, and the better were the muscle strength and functional scores, the higher was the relative bone mineral density in the same extremity (or the less bone loss due to the injury).
Spine and forearm The BMDs in spine, distal radius, and distal ulna are reported in Table 6 and Fig. 2. Compared with normal men, the spinal BMD were significantly lower in men with a history of a primarily nonunited (mean - 12.3%)and united (mean -9.5%) tibia1 shaft fracture. The mean value of the normal men ( I . I 16 0.160 g/cm2) was
*
very near the manufacturer's age-adjusted reference value for Western European men ( I . 1 1 3 0.120 g/cm2). In the distal radius and distal ulna, there were no significant differences between the three groups.
*
DISCUSSION Local changes of the bone following trauma were observed in humans in the 1860s, decades before the roentgen era.(2')Later, histologic, chemical, and radioisotope studies showed that posttraumatic mineral loss is best described as a high-turnover condition; that is, both bone formation and resorption are increased, but the latter overcomes the former, resulting in net Ioss.'~'.~~' Histomorphologically, this was demonstrated by
DENSITY (G/CM2) IN THE INJURED AND UNINJURED LOWEREXTREMITIES OF 14 PATIENTS WITH TABLE3. BONEMINERAL PRIMARILY NONUNITED SHAFT FRACTURE OF THE TIBIA" Anatomic site
Femoral neck Distal femur Patella Proximal tibia Distal tibia' Calcaneus
injured
* * *
0.902 0.083 1.272 f 0. I27 1.039 0.119 0.934 2 0.130 0.799 0.067 0.613 f 0.063
Uninjured
Difference
P value"
% Difference ~
0.926 2 0.092 1.415 f 0.115 1.169* 0.071 1.029 f 0.123 0.870 t 0.079 0.653 0.084
*
-0.024 2 0.059 -0.143 f 0.095 -0.130* 0.101 -0.095 t 0.069 -0.071 t 0.053 -0.040 0.055
*
-2.3 - 10.0 -11.2 -9.2 -7.9 -5.6
"Values are mean f SD. "The F value in the analysis of variance (the Hotelling T test) was 5.26 with P = 0.0176. 'Only 10 patients were included in this comparison since in 4 patients the fracture line extended to the measuring area. BONEMINERAL DENSITY IN THE INJURED LOWEREXTREMITY OF 20 TABLE4. RELATIVE PATIENTS WITH A PRIMARILY UNITED SHAm FRACTURE OF THE TIBIAAND 14 PATIENTS WITH A PRIMARY NONUNION Anatomic site
Femoral neck Distal femur Patella Proximal tibia Distal tibia Calcaneus
Primarily united (%)
-2.4 -4.0 -3.7 -5.1 -1.5
-1.8
Primarily nonunited (%)
-2.3 - 10.0
-11.2 -9.2 -7.9 -5.6
P value NS 0.0076 0.0059 NS 0.0199 NS
A
~
NS O.OOO1 0.0003 0.0005 0.0022 0.0175
TIBIAL FRACTURE AND OSTEOPOROSIS
421
TABLE5 . CORRELATION BETWEEN THE RELATIVE BONE MINERAL DENSITY I N THE DISTAI.FEMUR, PATELLA, A N D PROXIMAL TIBIA OF THE INJURED EXTREMITY AND THE AGE, IMMOBILIZATION TIME,PAIN ASSESSMENT,MUSCLESTRENGTH, AND FUNCTIONAL SCORES OF THE PATIENTS Distal femur r
Age Immobilization time Pain assessment Muscle strength Functional scoresa Lysholm score Tegner score IKDC score
Proximal tibia
Patella
P
r
P
r
P
NS
NS
