For those of us who don’t practice pediatrics day in and day out (and, I would imagine, for a fair number who do as well), having to do an IV start on a sick kid is a tough task. It’s both technically and emotionally challenging. I remember my first few times doing this in medical school, and you can’t help but feel bad when your young patient starts crying and fussing as you try to get the IV start going.

A new paper in JAMA Pediatrics suggests that playing music during an IV start in a pediatric ED population may reduce children’s pain. 42 children were enrolled in this study, where 21 were exposed to a few songs during their IV start, and the other 21 received usual care.

This sounds like a great idea (after all, wouldn’t you prefer to listen to some Fleetwood Mac during the procedure if you had to have an IV inserted), and the spin on the paper makes it sound like the intervention should be adopted everywhere tomorrow.

Here’s why that’s not necessarily true. From reading the abstract we can see that analyses were undertaken “when children who had no distress during the procedure were removed from the analysis”.

This is the core problem- it means that mean the authors of the study removed the people whom they didn’t expect to respond positively to the intervention. This increases the risk of bias; that is, that the results of the study will not provide the real effect of the treatment, and may overestimate its effectiveness. In order to reduce this, properly carried out randomized controlled trials are analyzed through an “Intention to Treat” approach, where everyone who starts the study is included in the analysis at the end. That way, regardless of what happens to people (they die, they move away, or they don’t like the music being played), their data is included when calculating the treatment effect.

As if that wasn’t enough, we find out in the full text that some parents of kids in the control group were encouraged by emergency department staff to sing to their children. Again, this contamination between treatment and control groups increases the risk of bias.

Maybe playing music is a good thing, and maybe it isn’t- but this study doesn’t give us the answer, and its methodological flaws unfortunately make it unsuitable to tell us if this intervention works.