Chat to your friends instead and save money.

The NHS spends many millions a year on speech therapy for patients with speech problems after a stroke. Most improve, but would they do just as well chatting to their family?

A nice trial (the statistician was my old friend Andy Vail) in this week’s BMJ says they would. The protocol was registered here, there was independent randomisation, an agreed primary endpoint, and analysis was by intention to treat. Recruitment was slower than anticipated but the decision to stop at 170 participants (85 per group) instead of the planned 600, was made without knowledge of the results.

The intervention group got an average of 22 sessions with a trained speech therapist while controls got the same number with an untrained visitor who just chatted in a general way. Care was taken to ensure the speech therapy was the best possible.  The primary outcome was a measure of patient communication at six months.  To avoid bias participants were videoed having a semi-structured conversation with a research assistant.  An independent person then scored the video without knowing if they were in the treatment or control group.  A high score was good.

After six months average scores were the same in both groups, 3.0 control and 3.3 treatment. The difference is not statistically significant, and when adjusted for chance differences at baseline, disappears altogether.

The trial was shockingly expensive, £1.5M, but if it stops the NHS wasting money on useless speech therapy for stroke patients it will have been money well spent.

Or maybe not.  Way back in 1982 a similar trial (click here) compared speech therapy with chatting to a volunteer and showed the same result – no difference.  Speech therapists ignored it.

It’s tough persuading people with a vested interest that they’re doing no good!

Jim Thornton