Tigger the jury is NOT out. There are some conflated sets of information on the web mixing up the fact that women who have scans often miscarry with scans causing miscarriage. Women who are prone to miscarriage, or experiencing pain, have more scans. Thus there is a correlation but not a causation. It's a classic 'smoke and mirrors' case.
To suggest to already vulnurable women that this harmless diagnostic tool could be causing problems could have horrible psychological results if they blame themselves for a loss afterwards.
Additionally, shying away from ultrasound because of hearsay can place women with pain in a dangerous position: the risks associated with a later diagnosis of ectopic or molar pregnancy for example.
The top anti-miscarriage experts in the world (the Danish team, Lesley Regan, Prof Quenby etc') are almost unanimous in their support of what is commonly called 'TLC care'. That is to say, if a woman prone to miscarriage is given extra reassurance... usually by way of extra ultrasounds in the 1st 12 weeks... it improves outcomes by up to 60%. Research is still going on as to why: all they know is it works. In such cases, ultrasound would cause more losses than in the general population as these women are already prone. In fact, their use improved outcomes (not because of anything physical the ultrasound does, but it provides a psychological boost and enables the consultant to keep a close eye on any diagnosed condition).
I have had 5 scans so far this pregnancy. I am very glad of them, they monitored that my blood medication was working and all was well.
2 of my losses had 3 scans: I know (as I have a diagnosis of a blood condition which caused them and the scans showed no heartbeat from the offset- the others were to confirm the diagnosis and check all materials had come out of me). 1 of them I had 10 scans and 25 blood tests ... it was a partial molar, so the 'reason' was a genetic abnormality. However, compared to 'normal' women, I had 3 losses and a hell of a lot of scans... on paper, that could (incorrectly) suggest a correlation.