Fixed or loose? Dichotomy in RSA data for cemented cups

Abstract

Background and purpose Roentgen stereometric analysis (RSA) cannot discern whether a single prosthesis is fixed or migrating below the detection level. Samples of patients usually show migration values that appear to be continuously distributed. Is there such continuity, or is there a dichotomy between stable and migrating prostheses? The hypothesis of a dichotomy has, to our knowledge, not been tested. We present an exploratory evaluation of such a dichotomy using a mixture distribution algorithm.

Methods We analyzed the migration (as determined by RSA) of 147 cemented acetabular cups of 7 different designs by using a new set of algorithms for frequency distribution analysis called Rmix.

Results We first analyzed a migration vector, regardless of direction. After 2 years there was a significant dichotomy between 2 lognormal subgroups within the sample. Although some types of cups were over‐represented in one of the subgroups, neither cup design, sex, nor operating department could explain the dichotomy into two groups, which appears to reflect the existence of two basically different types of behavior of the cups. We next analyzed the migration along the 3 axes in space, and found a similar dichotomy. During the second year, around 80% of the patients belonged to a distinct, normally distributed subgroup with a mean not different from 0 mm and a small variation. The remainder differed significantly from this subgroup and showed migration.

Interpretation There is a dichotomy in migration pattern. During the second postoperative year, most cups belonged to a subpopulation that appeared stable. The remainder is probably at risk of loosening. For a single type of prosthesis, the relative size of the stable subgroup may be a good index of the expected performance. The possibility of detecting subgroups within a seemingly continuous sample might be useful in many fields of medicine.

Contributions of authors

PA suggested the study, interpreted the statistical data, and wrote the manuscript. PW performed the statistical analysis and assisted in writing. KGN supplied most of the raw data and assisted in interpretation and writing. JR outlined the statistical analysis and supervised it.