Birmingham Hip Resurfacing - Literature Review

Survivorship
The survivorship estimates were based on the number of patients with no revision. Survivorship analyses were provided for various cohorts and demographic subgroups calculated according to Peto’s adjustment method as follows (Table 15).

¹ For the Overall McMinn cohort (2,385 hips)
² For the X-Ray + Oswestry cohorts (1,626 hips)

There were no statistically significant differences in cumulative 5-year survival (revision-free) probabilities among three study cohorts. The following Figure 1 summarizes these cumulative survival probabilities (all hips):

Figure 1. Cumulative % Revision-Free, BHR

Due to small number of revisions (total 25, < 5-year follow-up) from large numbers in three study cohorts (total of 2,385 hips), there were no statistically significant differences for all pairwise comparisons in 5-year survival (revision-free) probabilities among three cohorts, either by log-rank test, Wilcoxon test, or Cox proportional hazard (PH) regression analysis. Both the Cox PH regression model and the log-rank test require that the two survival probability curves be parallel or nearly parallel (no significant cohort by time crossover).

The above three statistical significance tests were also applied to several clinically important patient covariates, which include age (<65, >65), gender (M, F), reason for resurfacing (AVN, OA, IA, dysplasia, and others; reference group = OA), baseline OSHIP score (yes, no), hips (unilateral, bilateral). The only marginally statistically significant difference in 5-year survival probability was between the patients with Osteoarthritis (98.8%) and Avascular Necrosis (92.1%) as their primary diagnostic indication. The p-values to compare these two % revision-free curves for OA versus AVN comparison are p=0.0415 (Log-rank) and p=0.2282 (Wilcoxon).

Due to non-parallelism of the Oswestry and X-Ray survival curves, careful clinical interpretation is needed. Both log-rank and Wilcoxon test that the two revision-free curves are equal, and the Cox PH model tests that the ratio of the two hazards (probability of revision) is unity. The log-rank test assigns equal weight to all follow-up times and the Wilcoxon test assigns more weight to the earlier follow-up times where more patients are at risk of revision. The log-rank test has optimum statistical power if the parallelism assumption for the two revision-free curves is valid. The Cox PH model is not appropriate here due to obvious non-parallelism of the two curves in Figure 1. The percentages of revisions are 3.1% (3/97) for AVN, 1.1% for dysplasia (4/377), 0.95% (17/1789) for OA, 1.7% (1/57) for Inflammatory arthritis (IA), and 0% for others (0/65), with a combined 1% (25/2,385) revisions over all diagnostic groups, during 5-year follow-up.

Radiographic Data
The clinical data used to support this series contained the results of an independent radiographic review of the X-Ray Cohort, the first 124 procedures performed in the series from July 1997 through December 1997.

Radiographs were taken on 108 of the 118 procedures expected at 5 years postoperatively (91.5%). Six (6) procedures were not expected at 5 years postoperatively because one patient with bilateral hip implants died from a motor neuron disease unrelated to the BHR procedure; and 4 of the 124 BHR procedures (3.2%) have undergone revision: 3 cases were revised for infection, and 1 case required revision because of a femoral neck fracture. Therefore, 118 procedures (124 hips - 2 hips due to death - 4 revisions = 118 procedures) were eligible for 5 year radiographic evaluation of the BHR. Ten other cases were missing due to lost to follow-up or incomplete film records. Therefore, one hundred and eight (108) of the 118 hips surviving to 5 years had 5 year radiographs available for independent review (91.5%). (Note: An additional bilateral patient died 7 years post-op due to stroke but had 5 year x-rays taken).

Baseline films for the purposes of comparisons were made in each of the 108 cases in the postoperative time period (usually within 3 months, but 8 of the 108 procedures had baseline evaluations performed at time points ranging from 110-860 days).

Radiographic Study: 5-Year Radiographic Assessments
The radiographs were assessed for radiolucencies, bone resorption, heterotopic bone, acetabular angle, medial-lateral migration, and other observations to determine whether a revision surgery was necessary.

Femoral radiolucencies: Radiolucencies were graded 0-9 (Amstutz scale). There were femoral radiolucencies found in 4 cases (4.1%) - 1 each with grade 9 (migration), grade 5 (zone 2-3), grade 2 (zone 1) and grade 1 (zone 2). The patient with a grade 9 femoral radiolucency was classified as a radiographic failure.

Acetabular radiolucency: Radiolucencies were graded 0-9 (DeLee and Charnley scale). There were 2 hips with acetabular radiolucencies, both with grade 8 (zones I-III, complete) findings. One hip had preoperative acetabular cysts that progressed over time, and the other had a preoperative dysplastic acetabulum and developed protrusio. Both were classified as radiographic failures. Three patients had insignificant radiolucencies (grade 1 in two hips and grade 2 in one hip).

Heterotopic bone: There were 21 hips that had Brooker I and 5 hips with Brooker II heterotopic ossification (HO). Only 2 hips had “clinically significant HO,” (i.e., Brooker III or IV). Both had Brooker III HO. Thus, 28 of the 108 procedures evaluated (28.9%) had any heterotopic bone at 5 years and 2.1% had significant HO. None of the cases with heterotopic bone were determined to require a revision.

Acetabular angle: There was only 1 case that had a change in the acetabular angle >5Þ. This patient also had the grade 8 acetabular radiolucency (see above). No cases had a change in acetabular angle that was determined to be an indication for a revision.

Medial / Lateral Migration: There were no procedures with a change in medial/lateral acetabular cup position, and no cases with a change in acetabular position that was determined to be an indication for a revision. Additional observations: Bone resorption at the femoral neck was found in 3 cases. In no case was the resorption associated with any other notable radiographic findings. Bone cysts were found in 2 patients: one, described above, and the other had 3cm cysts associated with a grade 1 acetabular radiolucency. No other significant signs were noted.

Three (3) of the 108 (2.8%) patients for whom radiographs were available were radiographic failures at 5 years (Table 16).

¹ Occurred in the same patient.

Radiographic Study: Comparison to Literature Reference
The radiographic results were compared with the literature reference group (Table 17).

* No radiographic data.
** Revision rates are based on a minimum of 2-year follow-up and available x-rays.
¹ Same femoral component.
² Same acetabular component.

Pain and Function - Oswestry Modified Harris Hip (OSHIP) Score—Unilateral Procedures Only
FDA believes that it is difficult to assess the pain and function of each hip separately in patients with bilateral hip involvement using the Harris Hip Score or the Oswestry-modified Harris Hip Score (OSHIP), because it is difficult to distinguish the contributions of each hip on functional assessments such as walking or support, walking distance, stair-climbing, sitting, and transportation. Therefore, FDA believes only the unilateral patients should be used in an analysis of pain and function for the purposes of evaluating safety and effectiveness.

The mean OSHIP Scores (unilateral procedures only) improved from a baseline mean of 60.1 to 94.8 at 5 years. For the group of patients who had high baseline OSHIP scores (>80), the mean OSHIP scores improved from 84.5 to 99.3. The group of patients who had low baseline OSHIP scores (<80), the mean OSHIP scores also improved from 59.4 to 95.6. At postoperative years 2, 3, 4 and 5, the percentage of cases with good or excellent OSHIP scores was 96.9%, 95.8%, 95.2%, and 92.8%, respectively (Table 18)

* SD = Standard deviation;
**SE = Standard error of sample mean = SD/Vn ; CI = confidence interval of true OSHIP mean.

For the data in the table above regarding the number of procedures who improved >10 pts., maintained, or deteriorated >10 pts., that those patients with no baseline scores were counted as “maintained.” The table below contains an analysis of the number of procedures who improved >10 pts., maintained, or deteriorated >10 pts., when the patients without baseline scores are removed from this analysis and just counted as missing (Table 19).

Pain and Function - Comparison to Literature References
In the literature references, the authors used Harris Hip Score, not OSHIP, to collect pain and function effectiveness data. D’Antonio et al. reported Harris Hip Scores at 2 - 4 year follow up (mean 3 year) for the ceramic-on-ceramic hip procedures as follows:

• ABC System 1 (porous): 95.4 mean score (n=166)
• ABC System 2 (HA): 96.6 mean score (n= 172)

Garino reported an average increase in Harris Hip Score from 44 pre-operatively to a mean of 97 at follow up.

Patient Satisfaction
The patient satisfaction question is not a standard component of the OSHIP assessment but was an additional question asked for this study in the annual, patient-completed, mail-in questionnaire. At 5 years, 99.5% of the procedures in the X-Ray/Oswestry combined cohort were pleased or very pleased with the operation. At 5 years, 99.2% of the unilateral procedures from the X-Ray/Oswestry combined cohort were pleased or very pleased with the operation (Table 20).

Return to BHR Literature Review

Back to Patient Education - Hip

2/4/2007

OUR SERVICES  |  PATIENT EDUCATION  |  ABOUT DR. ROGERSON
CLINIC INFORMATION  |  OUR TEAM  |  CURRICULUM VITAE