Introduction: Existing fatigue studies of ACL fixation have two disadvantages. There is no agreed standard protocol, making comparison of various studies difficult and average results are presented, disregarding data spread. This may be over-optimistic, because approximately half the fixations will not achieve the average level. The effect of data spread can be summarised using the one-sided 80/80 lower tolerance limit (LTL). This LTL indicates the strength that at least 80% of fixations will reach, with an 80% probability. It is commonly used in engineering.
We fatigue-tested a new resorbable composite screw (PLLA/tri-calcium phosphate) and a metal interference screw. We present average data and tolerance limits.
Methods: Porcine BPTB grafts (Ø=9mm) were fixed inside tibial tunnels (Ø=10mm) using composite or metal screws. Each screw was tested for static pull-out strength (n=6) and cyclic loading to failure at 330N and 415N (n=5 each level)
Means and standard deviations of pullout strength were compared. Log-log curves were fitted between force level and cycles to failure. LTLs were calculated.
Results: During static loading, all repairs failed by graft pullout or tissue failure. During cyclic loading, all except one graft fixed with composite screws failed by pullout. Grafts fixed with metal screws failed by bone fracture in 60% of the cases. A composite screw loaded at 300N would last on average 272 cycles or at least (LTL) 7 cycles. At 200N the average and LTL were 38,218 and 966 cycles. Corresponding values for the metal screw were 263 (mean) and 12 (LTL) at 300N; and 12,454 and 564 at 200N.
Discussion and Conclusions: Repairs with metal screw had higher pullout strength, but proved more prone to fatigue. Higher incidence of bone graft fracture in fatigue testing with metal screws suggests that their sharp threads act as stress risers. Fatigue testing of ACL reconstructions shows wide variation, due to several factors. Average levels are therefore over-optimistic and tolerance limits gives a better indication of screw performance. We suggest that tolerance limits should be reported in future studies.
