both a modified Hungerford and Kenna knee rating system and Insall and Crosby grading system, and asking whether surgery had been worthwhile and whether they would go through it again. Serial radiographs were assessed for patellar malalignment, mechanical failure and progressive arthritic change in the knee, and failure was defined as a fairlpoor knee score or revision.
Preoperatively 17 knees had undergone arthroscopy. 36 Lubinus, 17 Cartier and 2 PFV prostheses were used. 5 patients died with 8 PF arthroplasties in situ, 1 patient lost to follow up (these patients are not included further in the analysis). 48 Patellofemoral arthroplasty knees were reviewed. 38 knees were classed as good or excellent, 10 had unsatisfactory results, and 7 were revised. 5 implants were revised to TKR and 2 were revised to PF arthroplasty (for maltracking). Subjectively 41 patients felt they were better, 5 unchanged and 2 worse. Overall we had 69% good or excellent results, 18% poor, and 12% revised. There were no infections, no revision for loosening, and no documented difficulty in revisions. The worst results were obtained in patients with evidence of tiblo-femoral OA preoperatively and in patients with tracking problems.