For a prospective randomized clinical trial, we recruited 184 patients between March 2003 and August 2008 at seven centers in North America. Initial inclusion criteria were (1) patients undergoing revision THA, (2) revision of both acetabular and femoral components, (3) acetabular components with a minimum of 50-mm outer diameter, and (4) patients with the ability to reply to questionnaires in French or English. Exclusion criteria were (1) revision for recurrent dislocation, (2) revision of acetabulum requiring structural allograft or a reconstruction cage, (3) revision of the acetabulum with a cemented all-polyethylene cup, or (4) intraoperative decision to use a constrained liner. In the early part of the study, isolated acetabular revisions were not included, as stems often did not have large heads such as Porous-coated Anatomic (PCA®) (Howmedica, Rutherford, NJ, USA). In February 2007, 36-mm heads became available for most stems, and we included patients undergoing isolated acetabular at this point. We recruited patients and obtained consent in accordance with institutional ethics and IRB protocols at all participating sites.
Participating surgeons and their research staff recruited eligible patients at each center. Randomization took place in the operating room after surgeons were satisfied that they could implant either a standard head (32-mm diameter) or large head (36- or 40-mm diameter). Once randomized, patients in the large-head group received a 40-mm head if the stem and cup would accept this diameter. In cases where this was not possible, the patient received a 36-mm head and corresponding acetabular liner. For cup sizes 58 mm and greater, a 40-mm head was used when available [16
]. We stratified patients first by surgeon and then by use of Trabecular Metal™ (Zimmer, Inc, Warsaw, IN, USA) shells. After stratification, the surgeons randomized the patients using sealed envelopes. Single blinding concealed the component design from patient, caregivers, and those related to the study, other than the surgeon and operating room staff. Allocation was a ratio of 1:1, with permuted blocks of four and six. Eleven surgeons at seven centers participated in recruitment and enrollment of patients.
We randomized 184 patients at the seven centers (Table ). Included in these 184 patients were two patients in each group who died before 2 years postoperatively. In addition to the 184 randomized patients, there were six we excluded from the study: two needed a large head, one was removed because the original 32-mm head was used, one was deemed unfit due to possible laryngeal cancer, one was ineligible, and one was removed due to extensive loss of abductors. We followed patients for a minimum of 2 years for the primary outcome of dislocation and the secondary outcome of QOL. For dislocation as outcome, minimum followup was 2 years (mean, 5 years in both groups; range, 2–7 years) (Table ).
Recruitment by research center
Length of followup from index surgery
We based the sample size calculation on the primary end point of this study: the first dislocation. We included patients in the analysis on an intention-to-treat basis and still followed patients who withdrew early for the primary outcome. We performed analysis using Fisher’s exact test. We designed the study to evaluate for a reduction in dislocation rate from 10% in the control (32-mm ball) to 5% in the experimental group (36- and 40-mm ball). We were only interested in superiority of the large femoral head and employed a one-tailed test with α set at 0.05 and power of 80%. Using Fisher’s exact test, we required 381 cases for a balanced design and anticipated a dropout rate of 10%, resulting in a target sample size of 419 patients in each surgical arm of the study. We carried out an interim analysis after enrollment of 175 patients, at which time the safety committee decided to stop further enrollment due to the marked difference in dislocation rates between the two groups.
Preoperatively, we recorded patient demographics, including age, sex, education level, BMI, and occupation (Table ). Checking the randomization analysis with group comparisons at baseline revealed no unexpected results. At baseline, all patients completed a self-administered comorbidity score and completed quality-of-life (QOL) questionnaires. The QOL questionnaires at baseline were the WOMAC [5
] and SF-36 [15
Demographic data for study population (n = 184)
The surgical approach for all cases was left to the discretion of the operating surgeon. The femoral component inserted was the VerSys® beaded full-coated (Zimmer) or ZMR™ (Zimmer) femoral stem or CPT™ (Zimmer) cemented stem, which can all accept the VerSys® 36- or 40-mm cobalt-chrome head (Zimmer). The acetabular component used was the Trilogy® cup (Zimmer) or the Trabecular Metal™ modular shell (Zimmer). Both cups were used in the study, but the decision to use one or the other was left to the operating surgeon. The Longevity® highly crosslinked acetabular liner (Zimmer) was used in both shells. The postoperative protocols were also left to the discretion of the operating surgeon (Table ). We did not collect data regarding the postoperative use of abduction braces.
Surgical variables (n = 184)
We followed patients at 3, 12, 24, and 60 months postoperatively, asking patients to respond to a followup questionnaire at each interval. We contacted those who did not return questionnaires by telephone and encouraged them to complete the submission. At each followup, the questionnaires included the WOMAC and SF-36. Additionally, we questioned patients at each interval as to whether they had a dislocation. Annually, we contacted the family physician and local orthopaedic surgeon to check for any dislocation events. The mean followup for the QOL was 26 months (range, 19–42 months) for the large-head group and 27 months (range, 21–46 months) for the standard-head group (Table ).
The WOMAC osteoarthritis index is the tool recommended for disease-specific outcome measures of hip and knee arthroplasty [6
]. It is a self-administered multidimensional index containing dimensions for pain (five items), stiffness (two items), and function (17 items). Items contain five Likert responses, which may be reported singly and in aggregate. The WOMAC is valid [5
] and reliable in patients with osteoarthritis of the hip and knee. Presently, it is the most frequently used measure of pain and functional disability among arthroplasty patients. WOMAC is scored using normalized data, with a score of 0 being the worst and 100 being the best.
The SF-36 is a self-administered generic measure of QOL with eight subscales: (1) physical functioning, (2) social functioning, (3) role limitations (physical), (4) role limitations (emotional), (5) pain, (6) mental health, (7) vitality, and (8) general health perception. The SF-36 is widely used and is reliable and valid across a broad spectrum of medical conditions [15
We compared baseline values between cohorts regarding age, BMI, comorbidity index, outcome measures of the SF-36 physical and mental scales, and WOMAC function, stiffness, pain and aggregate scales, via two-sample t-tests. Only the SF-36 mental scale differed (at α = 0.05) between cohorts, but the difference was less than ½ of a SD on the 50/10 scale (SD, 4.6). We compared baseline to followup within each cohort on the outcome measures listed above using a one-sample t-tests on differences.