As I delved into a research paper on cardiomyopathy related to methamphetamine abuse, I stumbled upon a table that left me scratching my head. The numbers seemed almost identical, yet the p-value proudly declared a statistical difference at p < 0.001. I'm no statistics expert, but my gut told me something was off. The table compared the length of hospital stays for patients with cardiomyopathy, with and without methamphetamine abuse. The distributions were eerily similar, with differences of less than 0.1%. I couldn't fathom how these nearly identical sets of values could yield a statistically significant result. ## The Table in Question | Length of stay (d) | With meth | Without meth | p-value | | --- | --- | --- | --- | | < 3 d | 1,037,195 (40.34) | 5,098,918.41 (40.39) | < 0.001 | | 4-6 d | 738,610 (28.73) | 3,632,147.96 (28.77) | - | | 7-9 d | 353,964 (13.77) | 1,740,210.64 (13.79) | - | | 10-12 d | 167,402 (6.51) | 822,719.36 (6.52) | - | | > 12 d | 273,942 (10.65) | 1,328,752.52 (10.53) | – |
## The Question Remains
Is this an error? Have I misunderstood something fundamental about statistics? I’d love to hear from a statistician or anyone who can shed some light on this enigma.
You can find the original paper [here](https://www.jacc.org/doi/10.1016/j.jacadv.2024.100840).
What do you think? Am I missing something obvious, or is there more to this story?