Thanks for the correction on the n-grams. I was aware of that but wasn't thinking carefully when I made that comment/observation.
I see now that 猫背な is favoured over 猫背の in certain cases so I don't have problem with including adj-na.

~20 instances of adj-na usage in the past 24 hours on Twitter
猫背のあなた	33
猫背なあなた	1760
猫背の私　	525
猫背な私　	400
猫背の感じ	27
猫背な感じ	210
猫背の背中	261

In a forked thread Alan Cheng wrote:
I agree that a lot of the 猫背な hits are from non-adj-na usages, so I dug a little deeper into the data for 
猫背な. FWIW I initially suggested this amendment because I came across it in a book (in the form 猫背な背中).
----
GG5, etc. show it's really 猫背の, but as often is the case a lot of people use ~な. I think it's worth reflecting this usage.

@@ -14,0 +15 @@
+<pos>&adj-na;</pos>

猫背な	10465
猫背など	2040
猫背なら	102
猫背なので	1454
猫背な私	400
猫背なのは	313	  
猫背なハンバーグ	303	 
猫背なのです	254	  
猫背なのに	253	  
猫背なのか	251	  
猫背なのが	219

Re "A lot of those 〜な counts are for など, なん, なら, etc."
This is not actually the case. The n-gram counts are based on morphemes, and など, なら, etc. are distinct morphemes.
A text sequence such as "猫背など姿勢が悪いと" is segmented as "猫背 + など + 姿勢 + が + 悪い + と" and would not be included in the 猫背な count.
The highish 猫背な is probably coming from other sources, some of which I've included. E.g. 猫背なのは is segmented as 猫背 + な + の + は.
(I'll close this comment after a while.)

猫背の人	2208
猫背の男	487
猫背な人	99
猫背な男	No matches

I don't agree in this case. None of the refs have this as adj-na, and adj-na usage isn't especially common.
A lot of those 〜な counts are for など, なん, なら, etc.

@@ -15 +14,0 @@
-<pos>&adj-na;</pos>

猫背	222457	  
猫背の	25474	   
猫背な	10465

@@ -14,0 +15 @@
+<pos>&adj-na;</pos>