Do PCOS papers say how they measured things?
When a paper reports a testosterone level, an insulin-resistance score or an ovary scan, does it say which assay, formula or criterion produced the value?
38.4%of reported values (201 of 524) come with no stated method. Papers published 2016 to 2026. Interval, resampling whole papers: 29.3 to 47.9. experiments/measurand_tiers/repass/REPORT_v3.md:18-19
45 of 98 papers (46%) leave out at least one method. experiments/measurand_tiers/repass/REPORT_v3.md:20
A formula is not a measurement. An insulin-resistance score such as HOMA-IR counts as stated only if the paper also names the insulin assay behind it; a free androgen index needs the testosterone and SHBG methods. For ovary scans, "an ultrasound was done" is not enough: the paper must give a criterion or the equipment. These rules were written after the pilot results were seen, and both push values towards "not stated". experiments/measurand_tiers/repass/REPORT_v3.md:46-55,134-135
| Reading | Not stated | Rate |
|---|---|---|
| Main reading (formula rule on; a citation alone counts as not stated) | 201/524 | 38.4% |
| Without the formula rule | 179/524 | 34.2% |
| Lenient: a cited methods paper counts as stated | 183/524 | 34.9% |
By what was measured
Papers 2016 to 2026. Small numbers throughout: one or two papers move a row by several points.
| Measure | Values | Not stated | % |
|---|---|---|---|
| Testosterone | 55 | 17 | 30.9% |
| Luteinising hormone | 50 | 22 | 44.0% |
| Fasting insulin | 49 | 20 | 40.8% |
| Follicle-stimulating hormone | 48 | 22 | 45.8% |
| Insulin resistance | 48 | 17 | 35.4% |
| Estradiol | 33 | 17 | 51.5% |
| Sex hormone binding globulin | 29 | 9 | 31.0% |
| Polycystic ovarian morphology (scan) | 27 | 10 | 37.0% |
| Free androgen index | 24 | 10 | 41.7% |
| Hirsutism | 24 | 3 | 12.5% |
| LH/FSH ratio | 24 | 13 | 54.2% |
| Anti-Müllerian hormone | 19 | 10 | 52.6% |
| DHEA sulfate | 19 | 8 | 42.1% |
| Progesterone | 14 | 8 | 57.1% |
Hormone assays (LH, FSH, estradiol, AMH, progesterone) are left out about half the time. Clinical scores such as hirsutism rarely are. Without the formula rule, every insulin-resistance score counted as stated because the formula was named; with it, about a third have no insulin assay behind them. experiments/measurand_tiers/repass/REPORT_v3.md:112-116
Limits
- Open-access papers only. Which papers were free to read differs by period, so the rate by period is not a trend. experiments/measurand_tiers/repass/REPORT_v3.md:72-73,132-133
- "Not stated" refers to the text we stored. A method given only in a supplement we did not fetch counts as not stated. experiments/measurand_tiers/repass/REPORT_v3.md:137-138
- Glucose is not checked, so only the insulin side of HOMA-IR is. experiments/measurand_tiers/repass/REPORT_v3.md:130-131
- It says nothing about the quality of the measurements themselves. experiments/measurand_tiers/repass/REPORT_v3.md:139
- The two readers agree with a kappa of 0.98, but they are two copies of the same model. That measures consistency, not correctness. experiments/measurand_tiers/repass/REPORT_v3.md:120-122