“normal” testosterone at 19 almost always means total T was tested, not free T or SHBG. those two numbers change the interpretation entirely. if SHBG is elevated, total T can sit at 600 and your free T still be functionally low. this happens a lot with guys who are insulin-resistant, and skinny fat is, almost by definition, an insulin resistance presentation. things worth actually testing before assuming it’s a GH problem: - free testosterone (direct, not calculated)
- SHBG
- IGF-1 – the actual proxy for GH axis activity. low-normal IGF-1 is more meaningful than chasing exogenous HGH
- fasting insulin – skinny fat plus normal total T plus poor muscle response is a textbook insulin resistance picture HGH at 19 without confirmed IGF-1 deficiency is adding a variable before you know what you’re solving. fix the insulin sensitivity side first. the muscle response to training changes noticeably once that piece moves. get the full panel, not just total T.
the part about a single fasting insulin draw being enough is the one I’d refine slightly, the agreement is otherwise pretty complete. I’ve now seen my own fasting insulin draws vary noticeably on an unchanged protocol, so what felt like a clean baseline turned out to be one point of noise sitting on a real trend. for anyone using fasting insulin to track movement over time, averaging two draws a couple weeks apart is more honest than treating any single number as the baseline. the other piece worth adding alongside fasting insulin is trig:HDL ratio. it sits downstream of insulin resistance where fasting insulin is more proximal, but it integrates over weeks instead of catching one fasted morning, and it’s already on most standard lipid panels so there’s no extra draw. mine was 3.2 when fasting insulin was 14 and the picture lined up cleanly. tracking both together makes the signal harder to mistake for noise from any single fasted morning. on the SHBG point, worth flagging that it isn’t only an insulin-resistance marker. thyroid status, liver function, and androgen exposure all shift it, so a low SHBG by itself doesn’t confirm IR any more than a total T of 600 confirms eugonadism. the panel as a set is what tells the story, not any one number pulled out. being able to export the lab log out of the careclinic app as a pdf was unexpectedly the piece that moved my own doctor conversations forward, easier to put a trendline in front of someone than to re-explain six numbers across appointments.
“integrates over weeks instead of catching one fasted morning” is the part I’d push back on. the case for trig:HDL is legitimate - it’s already on a standard lipid panel, costs nothing extra, and does reflect longer-run lipid metabolism in a way that’s clinically meaningful. agreed on all of that. but calling it more temporally stable than a single fasting insulin draw is doing more work than the evidence supports. triglycerides are also fasting-sensitive, and recent carb intake can move them meaningfully on a per-draw basis. someone who ate higher-carb two days before the blood draw will show a different ratio than the same person after a lower-carb stretch. the noise problem doesn’t disappear, it just expresses differently. your own logic for averaging two fasting insulin draws applies equally to trig:HDL - one data point is still one data point regardless of which metabolic marker you’re measuring. your 3.2 example is self-consistent with fasting insulin of 14, but that’s not the same as one being more reliable than the other. they’re both proximal enough to behavior that trend across 3+ draws is more informative than any single result. the actual improvement from adding trig:HDL isn’t stability, it’s convergent validity - two markers moving in the same direction is harder to dismiss than one. that’s the real argument for tracking both, imo.
“noise problem doesn’t disappear, it just expresses differently” is the line that lands, and i’m conceding the convergent validity framing because it’s the right one. where i’d push back: trig:HDL and fasting insulin aren’t quite equivalent on the noise side, the fasting insulin draw is more sensitive to the prior 24-48 hours of carb load and sleep, trig:HDL integrates a slightly wider window because triglyceride turnover is slower than insulin secretion dynamics. so single-draw to single-draw, trig:HDL probably is modestly more stable, just not stable enough to escape the trend-across-multiple-draws rule. directionally you’re right that both need 3+ to mean much, the gap between them is smaller than i implied but not zero imo.