Back in the days when junior doctors worked a 1:3 (1:2 when one of you decided to go on holiday), there were only two dieuretics and antibiotics used to kill bacteria, we learned our trade by hard graft. I (honestly) can’t remember the number of lumbar punctures I’ve performed, but I am aware that it’s a fair few more than many of the folk that train with me today will ever do. Part of this is the use of immunisation to reduce the incidence of severe bacterial infection, part is in a switch to a more rational approach to LPs, part from the safety-first removal of intrathecal injection to anyone but those specifically trained in the task. What is clear is that doing fewer makes it harder to get it right. Avoiding this problem can be tackled in a few ways – one of which may be simulation based training, another may be the use of imaging adjuncts, like ultrasound. A systematic review just out in our sister publication, BMJ, shows that the use of ultrasound reduces the chances of failed procedure to about 20%, halves the risk of a traumatic tap and reduces the number of re-passes to be undertaken. What’s unclear is exactly how easy it is to learn to do, what the cost/benefit would be and if we could actually implement it in practice. Or if another approach – like knowing how far to stab depending on how heavy the child was – would be equally good? Follow this link to see it undertaken in Vimeo from Academic Emergency Medicine: Lumbar Puncture in Infants with USS http://vimeo.com/32875951