Several types of compression hip-screws with a plate have gained increased popularity for the treatment of intertrochanteric fractures. These implants provide secure fixation and controlled impaction of the fracture. The rate of complications is relatively low. The most frequent mode of failure is cutout of the screw from the femoral head8,28. It is much less common for the fixation of the plate to fail or for the plate to pull off from the shaft. Problems may also arise from the need for extensive dissection and the blood loss resulting from this dissection.
The Gamma nail (Howmedica, Rutherford, New Jersey) was developed to circumvent these drawbacks, by combining the advantages of intramedullary fixation with those of a sliding screw. Theoretically, a shorter operative time and decreased blood loss are expected. Mechanically, the shorter lever arm of the Gamma nail decreases the tensile strain on the implant and thus reduces the risk of failure of the implant17,26. Treatment with this nail is not dependent on fixation of a plate to the lateral cortex with screws, which can be difficult in very osteoporotic bone. Moreover, telescoping displacement should be reduced by the intramedullary placement of the nail10. However, as far as we know, these potential advantages have not been reported in the published series of Gamma nails4,5,9,13,18,25,27. Rather, fractures of the nail3,30,33, pain in the mid-portion of the thigh18, and intraoperative2,3,18,20,25 and late2-5,13,14,18,21,25,31 diaphyseal fractures of the femur have been described. The prevalence of diaphyseal fractures has been reported to range from 0 per cent (of forty-three patients)9 to 17 per cent (eight of forty-seven patients)5.
The intramedullary hip-screw (Smith and Nephew Richards, Memphis, Tennessee) also combines a sliding compression screw and an intramedullary nail (Fig. 1). It was designed in an attempt to overcome some of the problems encountered with the Gamma nail.
We report the results of a randomized, prospective study that compared the use of an intramedullary hip-screw with the use of a compression hip-screw with a plate in elderly patients who had an intertrochanteric fracture.
*Although none of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article, benefits have been or will be received but are directed solely to a research fund, foundation, educational institution, or other non-profit organization with which one or more of the authors is associated. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was Smith and Nephew Richards, Memphis, Tennessee.
†Department of Orthopaedic Surgery, University Hospital Saint-Pierre, 290, Rue Haute, B-1000 Brussels, Belgium.
‡Pacheco Institute of Rehabilitation, 7, Rue du Grand Hospice, B-1000 Brussels, Belgium.
Patients
One hundred and five consecutive patients who had an intertrochanteric fracture of the femur, between December 1993 and January 1995, were prospectively randomized into two treatment groups according to the medical record number. Patients who had an even number were managed with an intramedullary hip-screw and those who had an odd number were managed with a compression hip-screw with a plate. The criteria for inclusion were an age of at least sixty years, a non-pathological acute intertrochanteric fracture of the femur, no history of a fracture or operation involving the ipsilateral hip, no history of a fracture of the lower limb during the year before the index procedure, and a femoral anatomy that allowed osteosynthesis with either an intramedullary hip-screw or a compression hip-screw with a plate. Three patients who had a fracture through a metastatic lytic lesion and two patients who had Paget disease were excluded from the study. In the latter two patients, the femora were curved and the canals were obliterated, which prevented any intramedullary procedure; these patients were managed successfully with a long-bone plate. Informed consent was obtained from the remaining 100 patients before inclusion in the study.
The preoperative parameters that were recorded included the age and gender of the patient, side of the fracture, body-mass index, and medical history. The patients were classified into one of three groups, on the basis of the medical history, with use of the system of the American Society of Anesthesiologists1 and the index of Fitts et al.11. Group I included patients who had no or a mild non-progressive associated pathological condition (such as a cholecystectomy or a healed fracture), group II included patients who had a moderately severe associated disease (such as a stroke, blindness, chronic obstructive pulmonary disease, compensated heart failure, hypertension, or senile dementia) that necessitated treatment or affected the quality of life, and group III included patients who had a severe associated disease (such as metastatic cancer or a recent severe myocardial infarction) that was likely to be life-threatening within six months.
Social functioning was defined according to the Jensen index15. Group 1 included patients who were independent and potentially able to work; group 2, patients who were able to manage a household but needed meals-on-wheels and four hours of home care a week or less; group 3, patients who needed at least five hours of home care a week and a nurse at home for specific care; and group 4, patients who lived in a nursing home or needed long-term nursing care at home.
Mental status was assessed with the Abbreviated Mental Test Score24. The patient was asked a series of ten questions and received 1 point for each correct answer. The maximum possible score was 10 points.
Walking ability was assessed with the mobility score of Parker and Palmer22, which includes three items—one reflecting the ability to walk indoors and two reflecting the ability to walk outdoors (Table I).
The fractures were classified as either stable (types I and II) or unstable (types III, IV, and V) on the basis of the classification of Jensen and Michaelsen16.
The estimated intraoperative blood loss, operative time, and intraoperative complications were recorded, as were data pertaining to the type of fixation, including the length of the lag-screw, the angle between the nail and the lag-screw, the diameter of the nail, the length of the plate, and the use of distal locking screws. The position of the lag-screw was assessed fluoroscopically in the lateral and frontal planes by dividing each image of the femoral head into nine areas to form a grid.
The type of reduction was also recorded, with reference to four basic modes: anatomical reduction, Wayne-County reduction29, telescoping reduction, and loss of contact. Anatomical reduction is essentially seen only with undisplaced fractures (type I) or slightly displaced fractures (type II). Wayne-County reduction is possible only if the calcar femoralis remains attached to the proximal fragment. This fragment is displaced medially and often is tilted into valgus, whereas the distal fragment is displaced laterally. The portion of the calcar femoralis that remains attached impinges on the cortical shaft medially, much like the situation with an intertrochanteric osteotomy. Collapse is minimum with this construct because the medial femoral cortex impinges on the calcar femoralis, which is displaced medially by the tilting of the proximal fragment into valgus. With telescoping reduction, there is medial displacement of the femoral shaft and controlled collapse at the site of the fracture. The reduction is recorded as loss of contact when a gap of five millimeters or more persists between the proximal end of the medial wall of the femoral shaft and the head-and-neck fragment, often due to a valgus overreduction. Stability of this reduction is frequently restored as the fracture is compressed.
Postoperatively, all patients had closed suction drainage of the wound. They also received antibiotic prophylaxis with intravenous administration of two grams of cefazolin before induction of the anesthesia, followed by one gram every eight hours for three doses. Prophylactic low-molecular-weight heparin (7500 anti-factor-Xa Institute Choay units of calcium nadroparin) was administered once a day for fifteen days in the absence of a history of myocardial infarction or stroke within the previous six months, hemorrhagias (gastrointestinal, gynecological, urological, and so on), a bleeding tendency (a thrombocyte count of less than 150 x 109 per liter or a prothrombin time of less than 65 per cent of normal), or a known allergy to low-molecular-weight heparin. Piritramide (a morphinomimetic agent) was administered for relief of early postoperative pain, and paracetamol was administered in the later recovery period. Intake of these drugs was recorded.
Patients were permitted to get out of bed and sit in a chair on the second postoperative day and were allowed to bear full weight by the fourth postoperative day. Mobility at the time of discharge, the output of the suction drainage, the total number of units of packed red blood cells that were transfused, the level of hemoglobin preoperatively and forty-eight hours postoperatively, and perioperative complications were recorded.
The patients were evaluated at one, three, six, and twelve months postoperatively. Mobility, as assessed with the score of Parker and Palmer22; the range of motion; the living situation; and the level of independence, as determined with the Jensen index15, were recorded. The use of assistive devices when the patient was walking outside the home was also noted. If the patient was unable to walk outside the home, the use of assistive devices inside the home was noted. Pain about the hip and in the mid-portion of the thigh was graded on a 4-point scale (1 point indicated no pain; 2, slight pain that did not affect the ability to walk or necessitate the use of analgesics; 3, moderate pain that affected the ability to walk or necessitated the use of analgesics; and 4, severe intractable pain even in bed).
Plain radiographs were made at each follow-up examination. Any change in the position of the screw was noted, as were union of the fracture and shortening of the femur. The length of the root of the lag-screw and that of the thread were measured on every radiograph (Fig. 2). The relative length of the unengaged part of the screw (the length of the root divided by the length of the thread) was recorded. By subtracting the ratio measured on the postoperative radiograph from that measured on the most recent radiograph, it was possible to derive the amount of sliding of the screw.
Implants and Operative Technique
The compression hip-screw with a plate (Osteo Hip Screw; Osteo, Selzach, Switzerland) was inserted with a standard technique by means of a straight lateral incision on the lateral aspect of the thigh, as described by Clawson7. The barrel of the plate was at a 135-degree angle in each patient.
The intramedullary hip-screw features a cannulated intramedullary nail with a 4-degree mediolateral bend to allow insertion through the greater trochanter. The nail is twenty-one centimeters long and is available in three diameters (twelve, fourteen, and sixteen millimeters). The opening for the lag-screw is available in two angles (130 and 135 degrees). It can be locked with one or two 4.5-millimeter-diameter interlocking screws. The nail is used with a lag-screw. A keyed centering sleeve, which is held by a set-screw, passes through the intramedullary nail and over the lag-screw. The sleeve helps to prevent rotation while allowing the lag-screw to slide freely.
The position of the patient for insertion of an intramedullary hip-screw is similar to that used for all types of intramedullary nailing of the femur that are done with the patient supine. The involved femur is slightly adducted to allow access to the trochanteric region. The uninvolved limb is flexed and abducted to allow unimpeded access of the image intensifier between the limbs. The fracture is then reduced in both planes. As in all intramedullary procedures on the femur, the tip of the greater trochanter is exposed through a lateral approach and is then opened with a curved awl. Access to the greater trochanter may be difficult in obese patients unless the limb is placed in marked adduction. However, when the limb is placed in this position, the initial reduction is often lost and a second reduction must be performed after the nail has been inserted. A guide-pin is advanced down the femoral canal well beyond the subtrochanteric area. An eighteen-millimeter-diameter proximal reamer is used to open the proximal portion of the femur to accommodate the proximal portion of the nail, which is 17.5 millimeters in diameter. A twelve-millimeter-diameter trial nail, with an opening to allow insertion of the lag-screw and the sleeve at an angle of 135 degrees, is inserted through the prepared part of the femur to ensure that the implant will fit in the intramedullary canal. This step is important and is never done with use of a hammer. If it is extremely easy to introduce a twelve-millimeter nail into the intramedullary canal, a fourteen-millimeter nail is inserted; if this nail is also loose, a sixteen-millimeter nail is used. If the twelve-millimeter trial nail is tight in the canal and jams distally, two options are available. When only a few millimeters would be needed for sufficient insertion of an intramedullary nail with a 135-degree opening for the lag-screw, an intramedullary nail with a 130-degree opening can be used. If the jamming is sufficient to prevent the insertion of an intramedullary nail with a 130-degree opening to the correct level, power reamers are used to enlarge the medullary cavity to as much as fourteen millimeters to accommodate a twelve-millimeter-diameter nail. The chosen nail is then inserted into the proximal aspect of the femur and is pushed down into the shaft under fluoroscopic control. The nail should not be hammered into the femur. The nail is positioned to ensure that the lag-screw will be placed centrally in the femoral neck in the lateral plane and, ideally, distally in the neck in the frontal plane. The lag-screw is inserted after insertion of a guide-pin and reaming. The keyed centering sleeve is pushed through the lateral cortex of the femur and the nail. After the centering sleeve is rigidly fixed with the set-screw, the lag-screw no longer rotates but is able to slide freely.
Distal locking screws can be inserted with a jig to target the holes in the distal aspect of the intramedullary nail. In the absence of guidelines, we decided to lock all of the nails with one screw unless the medial and posterior walls were both comminuted or rotational instability and subsidence of the nail was expected; in those situations, two locking screws were used. Final impaction of the fracture can be achieved with a compression screw inserted into the lag-screw to pull the lag-screw back into the sleeve. By leaning on the edge of the sleeve, the compression screw may exert a powerful traction on the lag-screw.
Statistical Analysis
The Pearson chi-square test, Student t test, and Yates corrected chi-square test were used for statistical analysis. The difference between the two treatment groups was considered to be significant when p was 0.05 or less.
Fifty patients had insertion of a compression hip-screw with a plate and fifty had insertion of an intramedullary hip-screw. The two treatment groups were comparable with regard to age, gender, side of the fracture, body-mass index, alcohol abuse, medical history according to the index of Fitts et al.11 and the system of the American Society of Anesthesiologists1, level of independence15, living situation before the fracture, mental status14, and mobility score22 (Table II). Sixteen stable and thirty-four unstable fractures were treated with a compression hip-screw, whereas thirteen stable and thirty-seven unstable fractures were treated with an intramedullary hip-screw.
The type of anesthesia was similar in the two treatment groups (Table III). The time needed to insert an intramedullary hip-screw was significantly greater (p = 0.02) than that needed to insert a compression hip-screw. The operative time for the first twenty-five procedures in which an intramedullary hip-screw was used (mean and standard deviation, 79.6 ± 35.2 minutes) was also significantly greater than that for the last twenty-five (61.5 ± 17.0 minutes) (p = 0.02). The mean intraoperative blood loss during the procedures involving an intramedullary hip-screw was less than that during the procedures involving a compression hip-screw (p = 0.011), but the mean number of units of packed red blood cells that were transfused and the mean level of hemoglobin at forty-eight hours postoperatively were similar.
Closed reduction of the fracture, with use of an image intensifier, was initially attempted in all patients. The closed reduction was unsuccessful in two patients who were to be managed with a compression hip-screw, and an open reduction was performed.
The median length of the lag-screw was ninety millimeters in both treatment groups. Forty-four of the plates for the compression hip-screws had four holes, and six had five holes. All of the compression hip-screws and seventeen of the intramedullary hip-screws were inserted at a 135-degree angle. The other thirty-three intramedullary hip-screws were inserted at a 130-degree angle because jamming of the nail had prevented ideal positioning of the lag-screw in the distal third of the neck and head in the frontal plane when an intramedullary nail with a 135-degree opening had been selected. Thirty-six of the intramedullary hip-screws were inserted into a twelve-millimeter-diameter nail; twelve, into a fourteen-millimeter nail; and two, into a sixteen-millimeter nail. Distal locking was performed in forty-six patients (with one screw in twenty-eight patients and with two screws in eighteen). Distal locking screws were not inserted in four patients.
There were five intraoperative complications associated with insertion of the intramedullary hip-screws. Distal locking could not be performed in two patients. Two patients had a fracture of the greater trochanter that did not necessitate additional fixation. One patient had a fracture of the femoral shaft that necessitated additional bone-grafting and insertion of a plate (Fig. 3). The fracture healed uneventfully following non-weight-bearing for three months.
The reduction was considered acceptable in forty-eight patients who had a compression hip-screw and in forty-nine patients who had an intramedullary hip-screw. In the remaining three patients, a residual gap of more than one centimeter persisted between the neck fragment and the medial wall of the femoral shaft.
The mortality rate in the hospital was similar in the two groups: three patients who had a compression hip-screw and five patients who had an intramedullary hip-screw died during the stay in the hospital. The use of analgesics during the stay in the hospital was also similar in the two treatment groups. Eleven patients who had a compression hip-screw and nine patients who had an intramedullary hip-screw did not recover any walking ability during the stay in the hospital. At the time of discharge, sixty-two patients (thirty-one in each group) were able to walk a short distance with use of a walker, nine (five who had a compression hip-screw and four who had an intramedullary hip-screw) were able to walk with two crutches, and one (who had an intramedullary hip-screw) was able to walk with one crutch.
The patients who had an intramedullary hip-screw had significantly less mean drainage from the wound (and standard deviation) (98.7 ± 104.3 milliliters) than the patients who had a compression hip-screw (205.4 ± 165.1 milliliters) (p = 0.0002). All of the wounds healed uneventfully without infection. Four patients who had an intramedullary hip-screw had a wound hematoma. Two of the hematomas were evacuated with needle aspiration without any persistent discharge, and the other two hematomas healed spontaneously.
The prevalence of perioperative complications was similar in the two treatment groups. The most frequent complications were bronchopneumonia (six patients who had a compression hip-screw and four who had an intramedullary hip-screw), cardiac failure (five and seven patients, respectively), and urinary tract infection (four and one patient, respectively). Clinical deep-vein thrombosis was noted in two patients who had a compression hip-screw and in one who had an intramedullary hip-screw, and one patient who had a compression hip-screw had a clinical episode of pulmonary embolism.
One compression hip-screw pulled out on the seventh postoperative day; it was replaced with an intramedullary hip-screw. This failure was related to the use of a plate that was too short for the posteriorly comminuted intertrochanteric fracture, which extended two centimeters distal to the lesser trochanter.
The fracture had healed in all but one of the seventy patients who were still alive at twelve months. The one non-union was in a patient who had a compression hip-screw. This patient had persistent tenderness in the region of the hip eleven months postoperatively. The implant was removed and a hemiarthroplasty was performed.
Twenty-four per cent (twelve) of the patients who had a compression hip-screw and 26 per cent (thirteen) of the patients who had an intramedullary hip-screw had died by the time of the six-month evaluation. The mortality rate at one year was 30 per cent (fifteen) in each group. No death was directly related to the fixation of the fracture.
The functional outcome was assessed at each follow-up interval through a review of the charts of the living patients. Eighty-nine charts were available at one month; seventy-seven, at three months; seventy-five, at six months; and seventy, at twelve months. No patient was lost to follow-up, except through death.
The mean mobility score22 was significantly greater at one month (p < 0.0001) and three months (p = 0.0013) for the patients who had an intramedullary hip-screw. The mean score was also greater at six and twelve months, but the difference between the groups could not be shown to be significant with the numbers available (Fig. 4). The differences at one and three months were significant with regard to the item reflecting the ability to walk inside (p < 0.0001 at one month and p = 0.002 at three months) and with regard to the two items reflecting the ability to walk outside (p = 0.02 at one month and p = 0.003 at three months). It is noteworthy that, although the total mobility score was similar in the two treatment groups at six and twelve months, the ability to walk outside was significantly better at those time-periods (p = 0.05 for both) for the patients who had an intramedullary hip-screw.
The use of assistive devices at twelve months was not found to differ between the two treatment groups. Of the patients who had a compression hip-screw, three were able to walk without any support, twelve used one crutch, ten used a walker and possibly needed the help of another person to walk, and ten were not able to walk. The corresponding numbers for the patients who had an intramedullary hip-screw were eight, sixteen, eight, and three.
Social functioning (as determined with the Jensen index15) in the two treatment groups did not differ markedly at any of the follow-up intervals. All fifty patients who had had group-4 social functioning preoperatively returned to the nursing home (or died), regardless of their actual level of functioning. Of the fifty patients (twenty-four with a compression hip-screw and twenty-six with an intramedullary hip-screw) who had had group-1, 2, or 3 social functioning preoperatively, two (one in each treatment group) had improvement from group 2 to group 1 at the twelve-month evaluation, fourteen (five with a compression hip-screw and nine with an intramedullary hip-screw) recovered their preoperative level of social functioning, and twenty-two (twelve with a compression hip-screw and ten with an intramedullary hip-screw) had a decrease of least one social-functioning group. The remaining six patients in each treatment group had died during the follow-up period.
Of the fifty patients (twenty-four with a compression hip-screw and twenty-six with an intramedullary hip-screw) who lived at home and were able to walk before the fracture, forty-four (twenty [83 per cent] with a compression hip-screw and twenty-four [92 per cent] with an intramedullary hip-screw) were able to walk at the one-year follow-up examination. With the numbers available, this difference was not found to be significant.
At three months postoperatively, seven of the forty surviving patients who had a compression hip-screw and four of the thirty-seven surviving patients who had an intramedullary hip-screw had pain in the hip while walking. At one year, only two patients in each group had pain in the hip. However, at one year, two patients who had a compression hip-screw and seven who had an intramedullary hip-screw had pain in the mid-portion of the thigh while walking, which resulted in a decrease in the ability to walk (to 3 points). Both of the patients who had a compression hip-screw and three of the seven who had an intramedullary hip-screw were pain-free after removal of the implant. One of the two patients who had a compression hip-screw had a bone spur at the end of the plate and a calcified hematoma within the soft tissues. The other patient had no radiographic abnormalities. Of the seven patients with an intramedullary hip-screw who had pain in the thigh, six had cortical hypertrophy at the level of the tip of the nail (Fig. 5); five had had the nail locked with two screws and one had not had the nail locked. The remaining patient had an intraoperative femoral fracture, as mentioned previously, and was therefore excluded from the analysis regarding the effect of locking screws. Of the twenty-eight patients who did not have pain in the mid-portion of the thigh, ten had been managed with two locking screws; seventeen, with one screw; and one, with no screws. Pain in the mid-portion of the thigh was more likely when two distal locking screws were used (p = 0.022).
Because the presence and severity of pain is difficult to assess in elderly patients who have senile dementia, the presence of cortical hypertrophy at the level of the tip of the nail was also considered. Of the thirty-five patients with an intramedullary hip-screw who were still alive at one year, fourteen (six who had pain in the mid-portion of the thigh and eight who were pain-free) had radiographic evidence of cortical hypertrophy at the level of the tip of the nail and twenty did not. (Again, the patient who had an intraoperative fracture of the femoral shaft was excluded.) The classification of the fracture (stable or unstable), features of the implants, mobility score22, and amount of sliding of the lag-screw were compared between these two groups (Table IV). Two factors (the number of distal locking screws and the postoperative mobility score) were significantly associated with cortical hypertrophy (p = 0.02 and p = 0.01, respectively). The nail was locked with two screws in ten of the fourteen patients who had cortical hypertrophy and in only five of the twenty patients who did not.
The four basic modes of reduction were unevenly distributed between the two treatment groups (p = 0.028). An anatomical reduction was seen in twenty-five patients who had a compression hip-screw and in twenty-seven patients who had an intramedullary hip-screw; a Wayne-County reduction, in fourteen and eighteen patients, respectively; a telescoping reduction, in eight and zero patients, respectively; and a gap of five millimeters or more, in three and five patients, respectively.
On the average, the lag-screw of the compression hip-screw device slid significantly more (10.2 ± 11.76 millimeters) than the lag-screw of the intramedullary hip-screw device (5.6 ± 4.32 millimeters) (p = 0.012) (Figs. 6-A and 6-B). The length of the involved limb was measured on the radiographs of sixty-four patients (thirty-seven who had a compression hip-screw and twenty-seven who had an intramedullary hip-screw) at the time of consolidation. The involved limb was a mean of 1.3 ± 1.08 centimeters shorter than the uninvolved limb after treatment with the compression hip-screw and a mean of 0.6 ± 0.69 centimeter shorter after treatment with the intramedullary hip-screw. This difference was significant (p = 0.019). There was no telescoping reduction and less sliding after insertion of an intramedullary hip-screw because the proximal end of the intramedullary nail was at the level of the greater trochanter. When telescoping of the lag-screw occurs, the neck fragment abuts the intramedullary nail, thus preventing further collapse of the fracture (Figs. 7-A and 7-B). No lag-screw cut out and there were no late fractures of the femoral shaft.
The position of the tip of the lag-screw in relation to the central axis of the femoral head differed significantly between the two groups (Fig. 8). The lag-screws of the compression hip-screw devices had been placed more inferiorly in the head (p = 0.005). Few screws in either group were in a bad position, but eight intramedullary hip-screws and four compression hip-screws were placed superiorly in the head. This difference was not found to be significant with the numbers available.
Unstable Fractures
Thirty-four of the fractures treated with a compression hip-screw and thirty-seven of the fractures treated with an intramedullary hip-screw were unstable. These two subgroups were comparable, except with regard to gender. The female-to-male ratio in the subgroup managed with a compression hip-screw (1.8:1) was significantly lower than that in the other subgroup (6.4:1) (p = 0.03). The mean operative time was similar in the two subgroups, even though the mean was greater for the patients who had an intramedullary hip-screw when the entire series was considered. Twenty-three of the patients who had an intramedullary hip-screw had the nail inserted at a 130-degree angle and fourteen had it inserted at a 135-degree angle. The nail was locked with one screw in nineteen patients and with two screws in eighteen patients.
The rate of mortality at one year was similar in the two subgroups. Twelve (35 per cent) of the thirty-four patients who had a compression hip-screw and nine (24 per cent) of the thirty-seven who had an intramedullary hip-screw died.
The patients who had an intramedullary hip-screw had a greater mean mobility score22 at each follow-up interval than did the patients who had a compression hip-screw. At one year, the mean total score was 3.4 ± 3.41 points for the patients who had a compression hip-screw and 5.3 ± 3.03 points for those who had an intramedullary hip-screw (p = 0.04). The patients who had an intramedullary hip-screw had a higher score because they performed better when walking outside (3.1 ± 2.14 points) compared with the patients who had a compression hip-screw (1.8 ± 2.42 points) (p = 0.04).
Radiographically, twelve of the twenty-eight patients with an intramedullary hip-screw who were still alive at one year had cortical hypertrophy at the level of the tip of the nail; four of the twelve had pain in the mid-portion of the thigh. The patients who had a compression hip-screw device had more sliding of the lag-screw (19.0 ± 7.84 millimeters) than did those who had an intramedullary hip-screw device (9.1 ± 4.62 millimeters) (p < 0.001). Consequently, the mean limb-length discrepancy after treatment with a compression hip-screw (1.6 ± 0.63 centimeters; range, 0.0 to 3.4 centimeters) was considerably greater than that after treatment with an intramedullary hip-screw (1.0 ± 0.40 centimeter; range, 0.0 to 1.9 centimeters) (p < 0.001).
Some of the benefits that are usually obtained with closed intramedullary nailing of fractures—namely, decreased blood loss and faster rehabilitation—were found in this study of intertrochanteric fractures treated with an intramedullary hip-screw. However, it must be emphasized that the apparent decrease in blood loss was clinically irrelevant as it did not affect the amount of blood that was transfused or the postoperative level of hemoglobin. Furthermore, the mean operative time that was needed to insert an intramedullary hip-screw was significantly greater than that needed to insert a compression hip-screw because the femoral shaft had to be reamed in eighteen patients. The different levels of experience of the two senior operating surgeons (D. C. R. H. and P. E. D.) and the three junior attending surgeons (P.-Y. D., P. K., and L. F.) in the present study and the prolonged learning curve for insertion of intramedullary hip-screws6 may also have affected the operative time.
The better mobility scores22 in the early postoperative period after insertion of the intramedullary hip-screws are difficult to explain. This finding was also reported in some trials that compared the results of treatment with a Gamma nail and those of treatment with a compression hip-screw and a plate18,19. The rehabilitation regimen in the present study was the same in the two treatment groups. To avoid the introduction of a bias when the mobility score was determined, the patients remained dressed so that the examiner (C. L. B.) could not guess the type of implant from the shape and length of the scar. The better mobility after treatment with the intramedullary hip-screw may be explained by the fact that these patients had less limb-shortening; this was particularly true for those who had an unstable fracture. Two centimeters of shortening or more is not uncommon after treatment of a comminuted intertrochanteric fracture with a compression hip-screw, and this shortening may have prevented these older, sometimes senile, patients from recovering the ability to walk. Nevertheless, the advantages of increased mobility, although important, are insufficient to alter the other functional parameters, such as social independence, or mortality.
There was one intraoperative fracture of the femoral shaft in the present study. This devastating complication was due to a technical error at the beginning of our experience. The femoral canal was insufficiently reamed and the nail was hammered down into the femur. These errors have been reported in association with Gamma nails3,25 and may be eliminated by greater familiarity with the implant system.
There were no late postoperative fractures of the shaft, as have been reported in association with Gamma nails2-5,14,18,19,21,25,30. Parker and Pryor, in a meta-analysis of ten randomized trials that compared treatment of intertrochanteric femoral fractures with a compression hip-screw with a plate and treatment with a Gamma nail, concluded that the Gamma nail cannot be recommended for routine use in such situations until the problem of fracture of the femoral shaft is resolved23. These fractures may be caused by excessive loads around the end of the nail26. The absence of late fractures in association with intramedullary hip-screws is probably due to the smaller overall valgus angle of the nail (4 degrees over the entire length) compared with the 10-degree angle of the standard Gamma nail. Because the 10-degree angle does not match the shape of the femur, the tip of the Gamma nail usually impinges against the lateral cortex, causing three-point loading4,20,25. The smaller diameter of the locking screws (4.5 millimeters compared with 6.0 or 6.28 millimeters with Gamma nails) is an additional factor.
Nevertheless, with the intramedullary hip-screw device, there is a stress-riser at the transition to the unsupported portion of the femoral shaft, as evidenced by cortical hypertrophy at the level of the tip of the nail in fourteen of the thirty-five patients with an intramedullary hip-screw who were still alive at one year. The nail is stiffer than a bone plate because of the large proximal diameter of the nail (17.5 millimeters), which enhances its resistance to bending. This stiffness appears to shield the proximal-medial cortex from applied load and to displace the stress toward the end of the nail26,27, in a fashion similar to that occurring with a femoral prosthesis. The use of two interlocking screws as well as a larger-diameter nail may also be implicated. Some osteoporotic patients have excessive anterior-posterior curvature of the femur. An abnormally high stress concentration may result from the contact between the tip of the nail (which is straight) and the anterior cortex of the femur (which can be surprisingly bowed in osteoporotic patients). The use of a ten-millimeter-diameter nail might be desirable in patients who have severe osteoporosis, but this type of nail was not available at the time of the present study. We found an association between the use of two locking screws and pain in the mid-portion of the thigh: all but one of the patients who had such pain had the nail locked with two screws. Thus, the use of interlocking screws should be weighed carefully. In the treatment of intertrochanteric fractures, the nail should be left unlocked if there is subtrochanteric extension or a reverse oblique pattern, except if rotational instability is anticipated. If locking is indicated, we recommend that one screw be used.
The lag-screw was noted to be positioned in the superior part of the femoral head in eight of the patients who had an intramedullary hip-screw device and in four of those who had a compression hip-screw device. However, no lag-screw cut out. Proper initial placement of the lag-screw central or inferior in the femoral head as seen on the anteroposterior radiograph, central as seen on the lateral radiograph, and within one centimeter of the subchondral bone decreases the rate of failure due to cutout8,12,32. However, placement of the lag-screw is limited by the intramedullary position of the nail itself. Some authors2,9,13 have drawn attention to the possibility of an increased prevalence of cutout due to technical difficulties encountered while inserting the nail. The nail may jam distally if the femur is narrow or curved. When the positioning of the lag-screw is improper, the use of a nail with a 130-degree opening may be helpful. If this is not sufficient, reaming of the medullary canal is mandatory, to allow the nail to be placed far enough distally. These difficulties accounted for the longer time needed for insertion of the first twenty-five intramedullary hip-screws in the present study and indicate the importance of the learning curve6.
We found less sliding of the lag-screw after the intramedullary hip-screw procedures, as has been noted by other investigators4,10. The intramedullary nail stops the telescoping displacement of the proximal aspect of the femur. In fact, the proximal part of the nail blocks the head-and-neck fragment, preventing its complete impaction. Thus, there is less subsequent shortening of the affected limb.
In view of these results, routine use of intramedullary hip-screws cannot be recommended for the treatment of intertrochanteric femoral fractures. However, because of the decreased shortening of the limb and the possibility of early weight-bearing even after a comminuted fracture with subtrochanteric extension or a reverse oblique pattern, this device is a promising alternative. The issue of pain in the mid-portion of the thigh is of concern, but it can be partially solved by restricting the use of interlocking screws to unstable fractures when rotational instability or subsidence of the nail may be expected. One locking screw instead of two seems advisable for such fractures.