
TSH is unreliable in people also. I am hypo, my T4 and T3 levels were both dropping lower each time I was tested, yet my TSH remained "perfect". If it was such a good indicator, it would have risen as the others fell. It is more reliable than in dogs, but a multi-factor panel of at least free T4, free T3, antibodies and TSH is still required for proper diagnosis.
Going back to dogs; to quote Dr Dodds herself: "TSH is NOT a reliable assay when used by itself in the dog, and generally has poor sensitivity and predictive value in the dog. This is in contrast to the use of endogenous TSH assays in people which work very well."
Or another... "In dogs, only ~70 percent of thyroid hormone regulation is controlled by TSH, so this test has relatively poor predictability of primary hypothyroidism. The remaining 30 percent of canine thyroid regulation is controlled by growth hormone. So, although elevated cTSH usually indicates primary thyroid disease, ~30 percent discordance is observed between expected and actual results in normal dogs, hypothyroid dogs and those with NTI or pituitary-dependent hypothyroidism."