max 250 words) Background: There is no national picture of teaching and training practices nor the communities they serve. We aimed to describe the association between general practices' engagement with education and their characteristics, locality and patients' health-status and satisfaction. Methods: Data linkage study of all English practices. Calculation of odds ratios for teaching and training status and practice, locality and patient variables. Results: Teaching and training practices are

more »

... larger than practices which do neither (mean list size(SD) 7074 (3736), 10112(4934) , and 5327 (3368) respectively, p<0.001 and have fewer patients per GP (1932(951), 1838(544) , and 2117 (1585) respectively, p<0.001). Their localities have a higher proportion of White British residents (77.99%(24.17), 81.66%(20.81), 73.07%(26.91), p<0.001). Practices with more GPs (OR 1.21 (95%CI 1.18-1.20)), fewer male GPs (0.45 (0.36-0.55)) and a higher proportion of White British people in their locality (1.30 (1.06-1.60)) were more likely to teach. Practices in rural areas (1.68 (1.43-1.98)), with more GPs (1.22 (1.27-1.39)), more full time equivalent (FTE) GPs (2.68 (1.64-4.40)), fewer male GPs (0.17 (0.13-0.22)) and a higher proportion of White British people in their locality (1.34 (1.02-1.75)) were more likely to train. Teaching and training practices had higher patient satisfaction (0.293 (0.190, 0.397) and (0.563 (0.442, 0.685)) respectively and QOF scores (0.507 (0.211, 0.804)) and (0.996 (0.650, 1.342) respectively than those which did not. Conclusions: Educationally engaged practices are unrepresentative in serving less ethnically diverse and (for training practices) less urban environments. Investment is needed to increase the proportion of educational practices in diverse urban localities.