Since 1974 at our institution we have been using a cemented cobalt-chromium rotation THA, occasionally in combination with an acetabular roof augmentation, using two or three small, triangular-shaped femoral head autologous grafts [12, 43]. In 2004, de Jong et al. reported a long-term followup study of a single-surgeon series of all primary THAs of all available stem sizes performed with this approach [13]. Results at 15 years were good with a cumulative survival with aseptic loosening as an end point of 81.5% for the stem and 89.2% for the socket [13].

In THA for primary osteoarthritis (OA), standard-sized total hip components and a regular surgical technique are sufficient to obtain good long-term results with a low rate of complications. In OA secondary to preexistent anatomic deformities of the hip, eg, developmental dysplasia of the hip (DDH), spondyloepiphyseal dysplasia, juvenile osteonecrosis (ON), and posttraumatic OA, surgery is more difficult [8, 26], clinical results are inferior [11, 21, 23, 45], and the complication rate is greater [11, 21, 24, 26] than in a patient group without these deformities. This may be the result of more frequent bone grafting, hypoplasia of the proximal femur with posterior migration of the greater trochanter requiring a smaller than usual femoral component [18], muscle contractures requiring muscular releases, or dislocation of the hip with difficulty in properly positioning components. The impact of inferior results in these patients is substantial because they typically are younger than the typical patient in whom THA is performed.

The numerous intermediate [1, 14, 21, 24, 47, 52] and long-term [5, 23, 25, 35, 39, 41, 46, 48, 49] followup reports of THA in patients with DDH typically focus on the diminished acetabular coverage in relation to fixation of the cup, and survivorship analyses with aseptic loosening of the cup as an end point have been reported [14, 24, 35, 39, 48, 52] to range between 86% and 97.5% in intermediate-term studies [14, 24, 48, 52] and between 88% and 94% in long-term studies [35, 39]. However, the additional problems of the presence of a hypoplastic femur with a small intramedullary canal and the use of a smaller than usual femoral component seldom are discussed [14, 21, 23, 39, 41, 46, 48]. There are even fewer reports of survivorship analysis of small-sized stems used in patients with DDH [14, 35, 39]. In these reports survival at 15 years with aseptic loosening as an end point has ranged between 92.4% and 96%. The number of reports specifically addressing implantation of a stem component in a hypoplastic femur also is limited [16, 34, 47, 58], and survival as reported in two studies only, was 100% [34] at 9 years and 79% at 15 years [47]. Several studies on small stems report relatively high rates of complications such as perioperative femoral fracture (10%) [58], (partial) nerve paralysis (5%) [21, 58], recurrent dislocation (16%) [58], and fracture of the component (5%) [21]. Clinical outcomes seem to deteriorate with decreasing patient height and diameter of the femur [9]. Only five patients in the report by De Jong et al. [13] had a hypoplastic femur and subsequently received the smallest-sized Weber stem. We could not ascertain whether that subgroup had different results compared with patients with average-sized femora. Thus, there is still debate regarding the ideal choice of stem for femoral hypoplasia, whether cemented or uncemented, off-the-shelf, or customized [3, 4, 7, 26, 34, 47, 58], and some have the opinion that patients with femoral hypoplasia have less good results than patients without femoral hypoplasia [26].

We hypothesized long-term survival, revision rates, clinical scores, and complication rates of the small cemented stem we used in patients with femoral hypoplasia were similar to those with larger-sized stems used in patients with a normal-sized femur.

Since 1978 we have been collecting data for all hip arthroplasties performed in our institution. From this database we identified two groups of patients: a study group of 77 selected patients (86 hips) with femoral hypoplasia who received the smallest available sized stem and a comparison group of 198 patients (231 hips) with normal femoral anatomy who received the identical curved type, but larger-sized stems. For both groups survivorship analyses, revision rates of the stems and cups, clinical scores, and incidence of surgical-related complications were assessed. We compared results between groups and also compared results of the small-stem study group with results of published reports of other small-stem hip systems.

Preoperative diagnoses in the hypoplastic small-stem group included: (1) DDH in 71 hips (83%); classified according to Crowe et al. [11] as Grade I in 52 hips, Grade II in 13 hips, Grade III in two hips (Fig. 1), and Grade IV in four hips and with a mean CCD angle of 139° (range, 125°–170°); (2) primary OA in five hips and posttraumatic OA in five hips (12% total); (3) ON in four hips (5%), and (4) rheumatoid arthritis in one hip (1%). Thirty-five hips (41%) had previous surgery, including 27 femoral osteotomies, three pelvic osteotomies, and five other hip-related procedures. In the normal anatomy group preoperative diagnoses included primary OA in 148 hips (25%), DDH in 58 hips (25%), posttraumatic OA in seven hips (3%), ON in two hips (1%), and rheumatoid arthritis in 58 hips (25%). Previous surgery was performed in 53 hips (23%) (Table 1). Twenty-one patients (21 hips) died after an average of 8.5 years postoperatively (range, 2.5–23.2 years), and one of these patients had a revision of both components after 7.8 years. Ten patients (12 hips or 14%) were lost by last followup in the hypoplastic group and 43 patients (52 hips or 22%) in the normal anatomy group; these patients were included in the survival analyses. Minimum followup in the hypoplastic group was 4.2 years (mean, 11.7 years; range, 4.2–20.3 years), and in the normal anatomy group 4.7 years (mean, 16.1 years; range, 4.7–23.7 years) (Table 1).

Fig. 1 This preoperative radiograph shows the left hip of a 49-year-old woman with bilateral femoral hypoplasia in the presence of developmental dysplasia of the hip and secondary OA.

---

Table 1 Patient demographics

Demographics	Hypoplastic group	Normal anatomy group
Number of patients (hips)	77 (86)1	198 (231)
Man : Woman	4 : 732	81 : 117
Median age, years (range)	65 (23–80)3	69 (36–85)
Mean followup, years (range)	11.7 (4.2–20.3)4	16.1 (4.7–23.7)
Median weight, kg (range)	64 (38–111)5	68 (46–115)
Median height, cm (range)	160 (143–189)6	168 (144–192)
Median body mass index	25.2 (16.9–41.1)7	24.5 (18.4–37.2)
Preoperative diagnosis (%)	8
DDH	71/86 (83%)	58/231 (25%)
Primary OA	5/86 (6%)	148/231 (64%)
Posttraumatic OA	5/86 (6%)	7/231 (3%)
Osteonecrosis	4/86 (5%)	2/231 (1%)
Reumatoid arthritis	1/86 (1%)	58/231 (25%)
Other	0/86 (0%)	8/231 (3%)
Number of previous hip surgeries (%)	35/86 (41%)9	53/231 (23%)

Statistical differences between the two groups:

¹ Student’s t-test, p < 0.01.

² chi square, p < 0.01.

³ Student’s t-test, p < 0.01.

⁴ Student’s t-test, p < 0.01.

⁵ Mann-Whitney U, p < 0.03.

⁶ Student’s t-test, p < 0.01.

⁷ chi square, p = 0.58.

⁸ chi square, p < 0.01.

⁹ chi square, p < 0.01.

We used an anterolateral (Watson-Jones) approach in all patients. An osteotomy of the greater trochanter, in the oblique sagittal plane, was performed in 22 hips (26%) in the hypoplasia group and in 30 hips (13%) in the normal anatomy group (Table 2). Fixation was accomplished with two lag screws perpendicular to the osteotomy plane and/or cerclage wiring from around the trochanteric tip, in a figure-eight shape around the proximal femur. The psoas tendon was released in all patients. The stem of the off-the-shelf Weber Rotation Modular prosthesis (Allopro, Baar, Switzerland) we used is made of wrought CoNiCrMo alloy (Protasul^® 10; Sulzer AG, Winterthur, Switzerland) and has a cylindrical neck (the trunnion), which is made of a cast CoCrMo alloy (Protasul^® 2) composite welded to the stem (Fig. 2). The cross section of the stem is trapezoidal with rounded corners. It is available in a curved shape in four different sizes and in a straight shape in one size. At surgery, all patients of the hypoplastic group had an intramedullary canal, which was either too small for the standard 103-sized rasp (length, 10.9 cm from medial collar to the tip; width at the middle level, 11 mm) to be used, or the shape of the proximal femur forced the 103 rasp into increased antetorsion and varus. We defined all such femurs as hypoplastic and used the smallest available 102 rasp (length, 9.6 cm from medial collar to the tip; width at the middle level, 9.5 mm) to prepare the canal for a 102 stem. Because of the typical narrow width of the femoral canal in these hypoplastic femurs, it was necessary to enlarge the canal with a special drill before rasping. This drill has sharp edges on the side that cut out the endosteal inner side of the shaft, and a blunt tip that prevents perforation of the cortex. Generally, the cortex was thickened proximally. In the normal anatomy group the larger-sized stems (length, 10.9 cm to 15.2 cm from medial collar to the tip; width at the middle level 11 to 16 mm) were used.

Fig. 2 In this picture, the hemispheric (left) and flat (right) socket with the curved stem of the 102 size Weber rotation total hip prosthesis with Protasul (right) and ceramic (left) heads with a trunnion bearing are shown.

---

The all-polyethylene (RCH 1000; Chirulen^®, Hoechst, Germany) sockets had either a hemispheric or a flat outer contour. The latter was designed for a shallow dysplastic acetabulum. In the hypoplastic group the spherical cups had a diameter of 44 mm in 14 cases, 47 mm in 42, 52 mm in 10, 54 mm in one, and 57 mm in two. The flat cups had a diameter of 47 mm in 12 cases and 52 mm in three. In two hips of patients lost to followup, the sizes and shapes of the cups could not be traced. Twenty-two hips (26%) in the hypoplastic group and 64 hips (13%) in the normal anatomy group had deficient support of the socket as suggested by templating on preoperative radiographs, a Sharp angle greater than 42°, and/or less then 90% coverage of the cup observed during surgery. We performed superolateral roofplasty in these 22 hips. We use three triflange corticocancellous bone grafts taken from the resected femoral head and fixed these to the roughened supraacetabular iliac bone with 4.5-cortical lag screws. The interfaces between the iliac bone and the grafts are impacted with cancellous bone [12, 43].

Thirty-two-millimeter metal (Protasul^® 2) and ceramic (Biolox; Feldmühle, Plochingen, Germany) heads (Table 2) were used. The head was placed on a Protasul^® 2 cylinder. The cylinder was placed on the trunnion of the stem, providing a secondary joint with the possibility for rotation and slight axial translation. Three different lengths of cylinders were available, ranging from 2.5 to 3.5 cm (Fig. 2). We used low-viscosity Sulfix^® (Sulzer AG) cement with a second-generation cementing technique, ie, intramedullary restriction of cement and vacuum suction in all patients. The stem was aimed in slight valgus in all patients (Fig. 3).

Fig. 3 The postoperative radiograph shows a 102-sized Weber stem on the left and a larger 103 stem on the right after a trochanteric osteotomy with bilateral 52-mm cemented cups and roofplasties 17 and 18 years, respectively, after the index operation. No demarcation around the components is seen. The acetabular roofplasties are completely integrated and the trochanteric osteotomy is healed. Periacetabular ossifications (Grade II according to Brooker) were asymptomatic. The Harris hip score at the time of the last followup was 87 points for the left side and 100 points for the right side.

---

We (FHRdM, HMVdV, PPB, RKM) clinically examined patients at 6 weeks, 3 months, and 6 months postoperatively and annually thereafter. Patients completed a questionnaire to calculate the Harris hip score [27]. In two patients (three hips), a reliable hip score could not be obtained as a result of severe dementia and recent brain infarction, respectively. Thus, clinical evaluation was possible in 37 patients (43 hips).

Standard anteroposterior and lateral radiographs of the pelvis of the surgically treated hip were obtained at each visit. We (FHRdM, HMVdV, PPB, RKM) analyzed radiographs for signs of osteolysis or loosening according to Harris et al. [28] for the femoral component and Hodgkinson et al. [31] and DeLee and Charnley [15] for the acetabular implant.

Survivorship for the femoral and acetabular components was analyzed using the life-table method [17], with three end points: (1) revision for any reason, (2) revision for aseptic loosening, (3) radiologic loosening (definite loosening of the stem and Grade III or IV loosening of the cup). We compared survivorship for all end points (log rank test) between the groups. We also compared the Harris hip scores (Student’s t-test) and surgery-related complication rates and rates of aseptic loosening of the cup and stem (chi square). Statistical differences in demographic and surgical characteristics between groups were detected (Student’s t-test and chi square for normal distribution and Mann-Whitney U test for nonuniform distribution). Characteristics that were different and potentially could lead to a confounding positive result of aseptic component loosening in the hypoplastic group were tested (Cox regression analysis).

The mean Harris hip score was similar (p = 0.55) in the two groups (mean, 88 and range, 51–100 in the hypoplastic group versus mean, 89 and range, 53–100 in the normal group). Average hip flexion was 104° (range, 70°–130°). A 1-cm leg-length discrepancy was present in five patients and a 4-cm discrepancy was present in one patient.

Surgery-related complications occurred in 5% of patients in both patient groups (Table 2). In the hypoplastic group one patient had a perforation of the femur treated by observation, one had a femoral fracture treated by cerclage wiring, one had a fracture of the greater trochanter after a previous intertrochanteric osteotomy treated with screw fixation, and one had surgical drainage of a hematoma. One patient had a deep calf venous thrombosis. All complications resolved completely after treatment. Late complications included a chronic infection in one patient who had few complaints and a hip score of 88. Therefore, the prosthesis was not revised. After an osteotomy of the greater trochanter, one patient had a nonunion that healed after screw fixation, tension band wiring, and cancellous autologous bone grafting. This patient had a hip score of 87, no Trendelenburg sign, and some residual pain not necessitating revision surgery.

Radiographic analysis revealed one stem was definitively loose and one stem probably was loose, but because both patients had few complaints and with hip scores of 88 and 84, respectively, the stems were not revised. Eight cups had Grade I loosening, but hip scores ranged from 89 to 100 and none was revised. All acetabular roofplasties showed remodeling and incorporation into pelvic bone with reformation of trabeculae.

Total hip arthroplasty in patients with hypoplasia of the femur is technically difficult because of the associated anatomic abnormalities. Several studies report a higher incidence of complications [11, 22, 40, 58] and less successful long-term results [24, 58] regarding stem fixation in comparison to primary THA in osteoarthritic hips of patients without hypoplasia. We asked whether the results in patients with OA treated with the smallest-sized cemented stem because of femoral hypoplasia were different from results in patients treated with a larger stem in an average-sized femur.

A limitation of this study is the lack of data regarding exact size of the proximal femur. Nevertheless, this size can be related to the size of the 102 component used. Another limitation is we did not measure clinical hip scores preoperatively. Because the pain caused by an osteoarthritic hip is the main reason to perform a THR, and pain is a dominant determinant of the Harris hip score and because hip scores at followup were high we can assume patients had substantial pain relief. Another concern is 11% of patients were lost to followup at the time of this study, but such a percentage is common and we believe does not jeopardize the conclusions; a mean of 6 years followup in this subgroup was still available, and patients who were lost to followup do not necessarily have poor results [36].

Reports specifically addressing patients with hypoplastic femurs [16, 34, 47, 58] have advocated off-the-shelf [47, 58] and custom-made stems [16, 34] (Table 5). Huo et al. [34] and Di Fazio et al. [16] used a custom-made varus-offset cemented straight femoral component in 19 small femora. The clinical score (Hospital for Special Surgery) was 35 of 40, no complications were recorded, and the revision rate after a mean followup of 13 years was 6% [16, 34]. Woolson and Harris used a miniature cemented off-the-shelf straight femoral component [58]. The revision rate for aseptic stem loosening after a mean followup of 5 years was 5.5%, the mean Harris hip score was 80, and surgery-related complications were relatively high at 19% [58]. In both studies, the exact stem size was not stated.

In a recent meta-analysis of studies comparing off-the-shelf cemented with uncemented THAs, cemented fixation had superior survival [44]. However, when femoral anatomy is different, some advocate an uncemented custom-made stem [4]. The concept is to completely fill the femur with a component, which would lead to less stress shielding of the proximomedial femur compared with a cemented implant [33, 55]. However, this is difficult to achieve because stress shielding will only decrease when stem fit is precise for the entire proximal femur [7, 20, 54, 56] and the necessary determination of femoral internal geometry is still inaccurate [7]. Moreover, as a result of the smaller diameter at the femoral neck compared with the intertrochanteric area, it theoretically is impossible to insert a component from one side only and have a perfect anatomic fit [56]. Furthermore, fabrication of a custom-made prosthesis is time-consuming and costly, regardless of cement use.

Surgery-related complications were limited to 5%, which equaled the complication rate in the normal anatomy group, whereas these rates in patients with femoral hypoplasia have been reported as high as 19% [11] and even 30% in extremely small femurs as seen in patients with achondroplasia [9]. We emphasize the need for a trochanteric osteotomy to obtain clear surgical exposure in these difficult cases. The incidence of nonunion after trochanteric osteotomy has been reported as high as 15% [10], but only 1% in our series. There were no dislocations, which might be explained by the use of 32-mm heads and the telescopic capacity of the rotating trunnion bearing, experience of the surgeon [29], and the use of an anterolateral approach [57].

In this series of relatively young patients after complex hip reconstructions resulting from a hypoplastic femur and OA, our data suggest the small cemented off-the-shelf Weber curved stem has high survival rates, is easy to use, is associated with few complications, and has outstanding long-term followup. Our results are in concordance with those of others for femoral hypoplasia and cemented off-the-shelf stems [35, 39, 41]. We agree with Capello [7] that if a cemented stem is used in small femora, the cured polymethylmethacrylate becomes part of the construct, which as a whole can be considered a customized implant. We believe there is no need for costly custom-made implants in patients with femoral hypoplasia.