Fifty Percent of What?
Asystole

Fifty percent less!

Less of what?

You're ruining the entrance.

Same outcome, same follow-up: give both risks a name;
Your headline took a shortcut and forgot half of the frame.
Ten in a hundred with control over one year;
Five in a hundred treated: bring the numbers here.
Risk ratio: five over ten, one half;
Relative reduction: fifty percent. Nice graph.
Absolute reduction: ten minus five;
Five percentage points. Let both descriptions thrive.
One over point-oh-five gives twenty in this frame;
Treat twenty for a year, compared with control, and claim
One extra event prevented on average in the scene;
No promise which one benefits. Keep that distinction clean.
N-N-T needs a comparator, outcome and a span;
You printed one heroic number on a promotional fan.
I bring the denominators right up to the front;
Your footnote's doing squats beneath the headline stunt.

Fifty percent of what? Start where the risks began.
Give me the absolute change and the follow-up span.
Who benefits, who gets harmed? Let the numbers speak.
Your headline's doing cardio. The baseline looks weak.
Fifty percent of what? Start where the risks began.
Give me the absolute change and the follow-up span.
Who benefits, who gets harmed? Let the numbers speak.
Your headline's doing cardio. The baseline looks weak.

New setting: two in a hundred down to one;
Same year, same relative half; your headline's having fun.
One percentage point absolute; N-N-T one hundred here:
One extra case prevented on average over that same year.
Baseline risk can change the absolute return;
Transporting an effect takes evidence to earn.
If the relative effect holds, the arithmetic can travel;
Whether it holds elsewhere needs checking before the gavel.
Now an adverse event rises from one percent to three;
Two points added harm: that's a cost we need to see.
One over point-oh-two gives fifty for that harm;
Over the same year in this example, keep your palm
Off the benefits-only poster covering the wall;
Severity and patient preferences matter through it all.
You traded one mild event against a serious one;
Two naked N-N-Ts won't tell you which choice won.

Fifty percent of what? Start where the risks began.
Give me the absolute change and the follow-up span.
Who benefits, who gets harmed? Let the numbers speak.
Your headline's doing cardio. The baseline looks weak.
Fifty percent of what? Start where the risks began.
Give me the absolute change and the follow-up span.
Who benefits, who gets harmed? Let the numbers speak.
Your headline's doing cardio. The baseline looks weak.

Control risk minus treatment risk: the absolute fall.
Use proportions in the reciprocal when you calculate it all.
Point-oh-five gives twenty; five would lead you astray.
Percent and percentage points have different jobs today.
These are estimates; show uncertainty with the size.
A confidence interval can complicate the prize.
When the risk difference crosses zero, say what that can mean:
Possible benefit and harm both belong in the scene.

Fifty percent of what? Start where the risks began.
Give me the absolute change and the follow-up span.
Who benefits, who gets harmed? Let the numbers speak.
Your headline's doing cardio. The baseline looks weak.
Fifty percent of what? Start where the risks began.
Give me the absolute change and the follow-up span.
Who benefits, who gets harmed? Let the numbers speak.
Your headline's doing cardio. The baseline looks weak.

I reduced my exaggeration by fifty percent.

From what?

I see this relationship will be difficult.