Out of curiosity and in the interest of moving this slightly away from "anecdote" towards "data", I did a quick grobbling over a 22-ish year old codebase and out of the 1.8M lines I found only about 1.55k instances of /\bsubstr\b/ using ripgrep -c (caveat though that line count is a raw wc -l not SLOC).

Of those matches about 900 were in two files of generated code (made by Parse::RecDescent), and a good chunk (~150 matches) of the others were false hits (e.g. the "substr" it found were actually in embedded SQL code in heredocs and not live perl code). Similar small amounts for "index" but that's even worse for this simple checking because "index" seems to be popularly used as part of a method name (i.e. I looked for /\bindex\(/ there to try and get a closer count but my gut feel (at the risk of re-approaching anecdote :) is the few hundred matches I found still are overcounting).