RichardW is, of course, quite right in a literal sort of way. And I'm the very first to remind that safety isn't about eliminating risk; it's about the elimination of
unnecessary risks; because some risks are absolutely necessary in certain roles of society (a soldier takes risks that I would never take, as they are unnecessary for me; but they are necessary for him).
My replies in this thread are directed to the restriction of considering 'upside down shoulder holsters'. My personal preference in concealment weapons is not to carry at all; on the other hand, in my environment it's really not a choice but a requirement: no guns here, even the coppers are prohibited from carrying off-duty.
So: 'why' upside down shoulder holsters at all? The original, the B-M, was designed and made for only small revolvers and with very short barrels. That meant 2" J, 2" D, and eventually 3" in both and I think even the M&P might have been added.
I personally would not rely on a B-M of any age. But if I were Bond, and someone handed me one with a loaded 2" .38 Special in it and said, "that's all I've got", I'd be grateful for it!
The upside down holster was supposed to, and does, provide a natural draw that is a sort of 'snatch and shoot' from across the chest. It was meant to accomplish this, by reversing the traditionally difficult way to draw (butt up and forward) and to conceal, by turning the grip to the rear. Perhaps Myres was first to 'think outside the box' with this patented holster by son Bill Myres.
So the B-M was issued with a specific set of instructions about where it was to be worn, and that was not under the armpit: between the armpit and the nipple. Yes, drawing it would direct the muzzle back to the shoulder because to be secure, it is designed to require the pistol be 'levered' out.

This is Jack Martin himself.
So (again), carried pointed away from the wearer, and everyone else except directly above in a multi-story situation, at all times except during the draw which, in an expert, is fractions of a single second.
On the other hand, if one doesn't have a good one (including elastic versions made to this very day by folks who don't understand the physics of the situation) one is risking it falling out every minute you're wearing it! That's bad.
The physics of the old #9R are such that it can't fall out. Ever. Under our gravitational limitations. Dunno about high g's such as a fighter jet launch. If you fly fighter jets, please don't listen to me and instead get your instructions from much higher up.
The 9R, and the better #9R-1 and -2, came about for the very reason that's being queried: an LEO jumped a fence wearing the original #9 and out popped the revolver! So we developed a test that it couldn't pass in the 'lab' (my office) and built a holster that would pass that, and an even greater extreme: we called it 'the snap test'.
I've covered the snap test before; will again if someone insists. But it's quite a boring enunciation and few will actually do it anyway. But those who do, with aged #209s and #9Rs, report 100% passes for them of this test. NO other brand of upside down holster will pass it. In fact, no vertical spring shoulder holster will pass it: takes a safety strap to pass a shoulder holster snap test unless it's a #9R or a #209.
When I carried, in USA with a permit in CA when those were v,v rare, I carried a 44 Bulldog in a #9R (-1 or -2 I dunno). Would do it again. It had the side benefit of being proof against my young children ever getting it out, because its draw is, there's the word, 'counterintuitive' until one puts it on.