Schrank · Archives of endocrinology and metabolism 2024 · Retrospective cross-sectional database study · n=21684

Proposal for fasting insulin and HOMA-IR reference intervals based on an extensive Brazilian laboratory database.

Cited 13 times in the scientific literature.

Level 4 - case-series / case-control

Retrospective cross-sectional laboratory database study establishing diagnostic reference intervals

PubMed 39529982 · doi:10.20945/2359-4292-2023-0483 · record verified 2026-08-28

What was done

Researchers retrospectively extracted serum insulin and fasting glucose data from a private laboratory database in Rio de Janeiro, Brazil, spanning 146,497 individuals aged 20–60 years. Insulin was measured using an electrochemiluminescence immunoassay. After applying exclusion criteria to filter out metabolic confounding, 21,684 individuals (18,576 [86%] women) were included. The 95% reference intervals (RIs) for fasting insulin and HOMA-IR were calculated using a computational data-mining approach in R to evaluate overall and sex-stratified values.

What was found

The 95% RIs for fasting insulin were 2.52–13.14 μU/mL (15.2–79.2 pmol/L) for the overall population, 2.54–13.30 μU/mL (15.3–80.12 pmol/L) in women, and 2.43–11.89 μU/mL (14.6–71.7 pmol/L) in men. The corresponding 95% RIs for HOMA-IR were 0.39–2.86 overall, 0.39–2.86 in women, and 0.38–2.81 in men. Despite statistically significant differences between sexes, the numerical differences were minimal, leading the authors to conclude sex-specific RIs are not justified.

Why it matters

These findings establish assay-specific, population-based cutoffs for Brazilian adults, proposing an upper limit of normal of 13.14 μU/mL for fasting insulin and 2.86 for HOMA-IR to help standardize the diagnosis of insulin resistance.

Limits

The cohort is retrospective and heavily skewed toward female participants (86%). As an indirect reference interval study based on laboratory records, detailed clinical information (such as exact medication use, body mass index, and subclinical conditions) was not directly verifiable from the abstract, and findings from one geographic region and specific assay platform may not generalize globally.

Cited by