Table 1.  Pooled likelihood ratios (95% confidence intervals; random effects) of negative result with Ottawa ankle rules for subgroups of increasing complexity of methodological quality Stratum Prospective data collection Plus consecutive enrolment Plus blinding Plus radiography as reference standard in all patients All studies Within 48 hours After 48 hours Within 48 hours After 48 hours Within 48 hours After 48 hours Within 48 hours After 48 hours Ankle 0.01 (0.08 to 0.22) 0.09* (0.04 to 0.22) 0.08 (0.02 to 0.39) 0.07 (0.01 to 0.44) 0.08 (0.03 to 0.20)   2×2 tables n=1 n=12 n=5 n=4 n=13 Mid-foot 0.07* (0.03 to 0.21) 0.08 (0.01 to 0.77) 0.08  (0.003 to 1.74) 0.07 (0.03 to 0.21)   2×2 tables n=9 n=4 n=3 n=9 Combined 0.21* (0.12 to 0.38) 0.26 (0.13 to 0.51) 0.29 (0.12 to 0.71) 0.42 (0.21 to 0.81) 0.21 (0.12 to 0.38)   2×2 tables n=10 n=6 n=4 n=1 n=10 Children 0.08 (0.02 to 0.29) 0.06  (0.02 to 0.25) 0.10 (0.01 to 1.64) 0.10 (0.01 to 1.64) 0.07 (0.03 to 0.18)   2×2 tables n=4 n=3 n=1 n=1 n=7  * Larger negative likelihood ratios in studies testing Ottawa ankle rules in mixed populations (ankle and mid-foot versus combined: P<0.001).