Fasting Insulin and Drugs That Distort This Test

What Fasting Insulin Actually Measures

Fasting insulin is the serum concentration of endogenous insulin after an overnight fast of at least 8 hours. It reflects basal pancreatic beta-cell secretion in the absence of a carbohydrate challenge. The test is not the same as a C-peptide, not the same as a postprandial insulin draw, and not interchangeable with HOMA-IR (though HOMA-IR is calculated directly from fasting insulin and fasting glucose).

Most commercial immunoassays report results in microinternational units per milliliter (mcIU/mL). Reference ranges vary slightly by laboratory, but the majority of US labs place the normal fasting insulin interval at 2.6-24.9 mcIU/mL. The American Association of Clinical Endocrinology (AACE) has noted in its 2023 algorithm that values persistently above 15 mcIU/mL in a fasting, non-diabetic adult raise concern for insulin resistance even when fasting glucose remains below the diagnostic threshold for pre-diabetes. [1]

Why the Test Matters Clinically

Insulin resistance precedes type 2 diabetes by years to decades. A fasting insulin level gives clinicians a window into beta-cell compensation: when peripheral tissues resist insulin signaling, the pancreas compensates by secreting more. HOMA-IR above 2.5 (calculated as fasting insulin mcIU/mL multiplied by fasting glucose mmol/L, divided by 22.5) is one commonly used cut-point for insulin resistance in research settings. [2]

In women with polycystic ovary syndrome (PCOS), elevated fasting insulin is both a diagnostic clue and a therapeutic target. The Endocrine Society's 2023 PCOS clinical practice guideline states that insulin resistance affects approximately 50-70% of women with PCOS regardless of body weight, and that lifestyle plus pharmacologic intervention should address insulin sensitivity directly. [3]

The Problem Nobody Mentions Often Enough

The test is exquisitely sensitive to pre-analytical and pharmacologic confounders. A patient who took a single dose of prednisone the morning of the draw may return a fasting insulin value 40-60% above their true baseline. A patient three months into semaglutide therapy may show a fasting insulin near the lower limit of normal, not because their beta-cell function improved dramatically, but because the drug acutely suppresses basal insulin secretion. Getting the interpretation wrong changes clinical decisions.

Normal Fasting Insulin Range and What Alters It

The 2-25 mcIU/mL reference range used by most US laboratories was built on population-average data that includes many metabolically unhealthy individuals. Functional medicine clinicians and endocrinologists increasingly use a tighter optimal window of 3-8 mcIU/mL for metabolically healthy adults not on any glucose-altering medication. [4]

Factors That Raise Fasting Insulin Physiologically

• Obesity, especially visceral adiposity (each 5-unit increase in BMI correlates with roughly 2-3 mcIU/mL higher fasting insulin in population studies)
• Chronic psychological stress via cortisol-mediated insulin resistance
• Pregnancy (gestational insulin resistance peaks in the third trimester)
• Polycystic ovary syndrome
• Cushing syndrome or exogenous glucocorticoid use (covered separately below)
• Insulinoma (rare, but produces dramatically elevated fasting insulin alongside suppressed fasting glucose)

Factors That Lower Fasting Insulin Physiologically

• Prolonged caloric restriction or very low carbohydrate diets
• Regular aerobic exercise (a 12-week moderate-intensity program reduced fasting insulin by a mean of 3.5 mcIU/mL in a 2019 meta-analysis of 37 RCTs, N=2,432) [5]
• Type 1 diabetes (endogenous insulin is absent or near-absent; exogenous insulin is not measured by standard immunoassays that use antibodies against human insulin, though some cross-reactivity exists)
• Pancreatic insufficiency or surgical pancreatectomy

Drugs That Raise Fasting Insulin (False Elevations)

This is the most clinically consequential category for patients being worked up for insulin resistance or metabolic syndrome. A falsely elevated result can lead to unnecessary medication changes, incorrect diagnoses, or missed reassurance that a patient's own physiology is intact.

Corticosteroids

Prednisone, dexamethasone, methylprednisolone, and inhaled fluticasone at high doses all raise fasting insulin by inducing peripheral insulin resistance, which forces compensatory hyperinsulinemia. The effect is dose-dependent. Prednisone 40 mg/day for 7 days has been shown to raise fasting insulin by 50-100% above baseline in non-diabetic adults. [6] Even inhaled corticosteroids at doses above 1,000 mcg/day of fluticasone-equivalent have been associated with measurable insulin resistance in patients with asthma, per a 2020 cohort analysis published in JAMA Internal Medicine. [7]

Atypical Antipsychotics

Olanzapine, clozapine, and quetiapine are the worst offenders in this drug class. The FDA has issued a class-wide warning that second-generation antipsychotics may cause hyperglycemia and diabetes in some patients. [8] Mechanistically, these agents block hypothalamic H1 receptors, drive weight gain, and directly impair beta-cell function and peripheral glucose uptake. Fasting insulin in patients on long-term olanzapine can run 30-80% above drug-naive comparators matched for BMI.

Thiazide Diuretics

Hydrochlorothiazide (HCTZ) and chlorthalidone raise fasting insulin through hypokalemia-mediated impairment of insulin secretion and through direct effects on peripheral glucose disposal. The ALLHAT trial (N=33,357) found that chlorthalidone increased the rate of new-onset diabetes over 4 years compared with amlodipine and lisinopril (11.6% vs. 9.8% vs. 8.1%, respectively). [9] Fasting insulin values in hypertensive patients on HCTZ should be interpreted with this pharmacologic confounding in mind.

Niacin (Nicotinic Acid)

Niacin at pharmacologic doses (1,000-2,000 mg/day used for dyslipidemia) worsens insulin sensitivity through mechanisms that include increased free fatty acid mobilization and direct inhibition of insulin signaling in adipocytes. In the AIM-HIGH trial, extended-release niacin 1,500-2,000 mg/day significantly raised fasting glucose and worsened glycemic control, consistent with secondary hyperinsulinemia in insulin-resistant participants. [10]

Other Agents Worth Noting

• Oral contraceptives containing high-dose ethinyl estradiol (above 30 mcg) modestly raise fasting insulin in susceptible women, particularly those with underlying PCOS or family history of type 2 diabetes.
• Tacrolimus and cyclosporine (calcineurin inhibitors used in transplant patients) cause significant beta-cell toxicity, producing a paradoxical picture of low endogenous insulin with high glucose.
• Beta-blockers, especially non-selective agents like propranolol, impair insulin secretion and mask hypoglycemia; fasting insulin may be modestly elevated secondary to glucose counterregulation.

Drugs That Lower Fasting Insulin (False Depressions or Therapeutic Reductions)

A low fasting insulin result is sometimes reported without context, creating confusion about whether it reflects genuine beta-cell failure, physiologic improvement, or drug effect.

GLP-1 Receptor Agonists

Semaglutide (Ozempic, Wegovy), liraglutide (Victoza, Saxenda), tirzepatide (Mounjaro, Zepbound), and dulaglutide (Trulicity) all reduce fasting insulin through glucose-dependent mechanisms. They suppress inappropriate glucagon secretion and improve peripheral insulin sensitivity, meaning the pancreas needs to secrete less insulin to achieve the same glucose disposal. This is a therapeutic reduction, not a pathologic one.

In STEP-1 (N=1,961), semaglutide 2.4 mg weekly produced 14.9% mean body weight loss at 68 weeks versus 2.4% with placebo, accompanied by significant reductions in fasting insulin and HOMA-IR. [11] A clinician interpreting a fasting insulin of 5 mcIU/mL in a patient six months into semaglutide therapy should not interpret this as hypoinsulinism. The drug is doing its job.

SGLT-2 Inhibitors

Empagliflozin (Jardiance), dapagliflozin (Farxiga), and canagliflozin (Invokana) lower fasting insulin by reducing glucotoxicity (less ambient glucose means less demand on the beta cell). The EMPA-REG OUTCOME trial and subsequent mechanistic substudies demonstrated significant reductions in fasting insulin in patients with type 2 diabetes over 3.1 years of follow-up. [12] In non-diabetic patients being screened for metabolic syndrome, SGLT-2 use should be documented before the lab draw.

Metformin

Metformin is the most prescribed oral antidiabetic agent globally. It primarily reduces hepatic glucose production (gluconeogenesis) and improves peripheral insulin sensitivity. These effects together reduce the beta-cell's secretory burden. Fasting insulin levels in patients on metformin 1,000-2,000 mg/day are typically 10-25% lower than in drug-naive controls matched for BMI and HbA1c. [13] This is beneficial, but it confounds baseline screening if a patient starts metformin before a diagnostic fasting insulin is ordered.

Sulfonylureas: The Paradox

This class requires separate handling. Sulfonylureas (glipizide, glimepiride, glyburide) acutely increase insulin secretion, raising post-dose insulin levels. With chronic use, however, they cause beta-cell exhaustion, which over years may reduce fasting insulin. A single-timepoint fasting insulin in a patient on long-term glimepiride tells you very little about their true beta-cell reserve.

Thiazolidinediones

Pioglitazone and rosiglitazone (PPAR-gamma agonists) improve peripheral insulin sensitivity without directly stimulating secretion. The net effect on fasting insulin is a true reduction: the pancreas secretes less because cells respond better. Pioglitazone 45 mg/day reduced HOMA-IR by approximately 25-30% versus placebo in the PROactive trial subgroup analyses. [14] This reduction should be labeled a pharmacologic benefit, not a pathologic finding.

How to Prepare Patients for an Accurate Fasting Insulin Draw

Getting the draw right is just as important as interpreting it correctly. The following preparation protocol minimizes pre-analytical error.

Fast Duration and Dietary Preparation

Patients should fast for exactly 8-12 hours before the draw. A 12-hour fast is preferred for consistency across serial measurements. During the fast, only plain water is permitted. Coffee, even black, causes a cortisol and catecholamine response that may modestly raise fasting insulin in sensitive individuals.

The night before, patients should avoid high-carbohydrate meals (above 150 g of carbohydrate at a single sitting) and vigorous exercise within 24 hours. Acute high-intensity exercise can lower fasting insulin transiently for 12-24 hours post-exercise through GLUT-4 upregulation independent of insulin signaling.

Medication Documentation

Every prescription medication, over-the-counter drug, and supplement the patient takes should be documented on the lab requisition. The ordering clinician needs to decide, in conversation with the patient, whether to hold confounding medications the morning of the draw. For most stable patients, holding a corticosteroid or antipsychotic for a single morning is not safe without medical supervision, so the better approach is to note the drug at a standardized dose and repeat the test during any future medication-free window.

Timing Relative to Medication Changes

When possible, order a baseline fasting insulin before starting any medication that meaningfully alters the result. The AACE 2023 diabetes management algorithm recommends baseline metabolic labs (including fasting insulin in appropriate clinical contexts) before initiating any agent with known insulin-sensitizing or insulin-stimulating properties. [1]

Interpreting Fasting Insulin in Special Populations

Women With PCOS

The Endocrine Society's 2023 PCOS guideline does not include fasting insulin as a required diagnostic criterion under Rotterdam criteria, but the guideline explicitly acknowledges its clinical value for risk stratification and treatment planning. [3] A fasting insulin above 12-15 mcIU/mL in a lean woman with PCOS is clinically meaningful even if it falls within the laboratory's reference range. Metformin is the first-line pharmacologic option for PCOS-associated insulin resistance in women who are not attempting conception, per that same guideline.

Patients on Hormone Therapy

Testosterone therapy in men lowers insulin resistance over 12-24 months, with fasting insulin typically declining 10-20% in hypogonadal men who achieve mid-normal total testosterone levels (400-600 ng/dL) on therapy. A 2021 Cochrane review of testosterone replacement in men with hypogonadism (27 RCTs, N=2,599) found modest but statistically significant improvements in HOMA-IR. [15] Interpreting fasting insulin in a man who recently started TRT requires knowing his testosterone trajectory.

Estrogen-based HRT in postmenopausal women has a more nuanced effect: transdermal estradiol at physiologic doses tends to maintain or modestly improve insulin sensitivity, while oral conjugated equine estrogen at higher doses may worsen it. Progestogen type matters too. Micronized progesterone appears metabolically neutral; medroxyprogesterone acetate is associated with mild insulin resistance.

Pediatric and Adolescent Patients

Reference ranges for fasting insulin in children and adolescents are age and Tanner-stage dependent. Insulin levels peak physiologically during mid-puberty (Tanner stages 2-4) independent of diet or body composition, a phenomenon sometimes called "pubertal insulin resistance." A value of 15 mcIU/mL in a 13-year-old at Tanner stage 3 does not carry the same clinical weight as the same value in a 35-year-old adult. [16]

How to Lower Fasting Insulin Without Drugs

Lifestyle interventions produce clinically meaningful reductions in fasting insulin that are durable when maintained. The ADA's 2024 Standards of Medical Care emphasize that structured lifestyle intervention reducing body weight by 5-10% improves insulin sensitivity comparably to many pharmacologic agents in pre-diabetic adults. [17]

• A low-carbohydrate diet (below 130 g/day total carbohydrate) reduced fasting insulin by a mean of 7.1 mcIU/mL versus a low-fat diet in a 2021 meta-analysis of 23 RCTs (N=1,357). [18]
• Time-restricted eating (16:8 intermittent fasting) reduced fasting insulin by approximately 20% over 12 weeks in a 2020 JAMA Internal Medicine RCT (N=116). [19]
• Resistance training 3 times per week for 12 weeks lowered fasting insulin by 3-5 mcIU/mL independent of weight change, through increased muscle GLUT-4 expression.
• Magnesium supplementation at 300-400 mg/day improved insulin sensitivity markers in a 2016 meta-analysis (N=1,368) of patients with hypomagnesemia, a finding that is particularly relevant because thiazide diuretics deplete magnesium and compound their insulin-raising effect. [20]

Putting It All Together: A Clinical Decision Checklist

Before signing off on a fasting insulin result, the ordering clinician should confirm the following five points:

1. Was the patient truly fasted for 8-12 hours with no caloric intake?
2. Are any of the insulin-elevating drugs (corticosteroids, atypical antipsychotics, thiazides, niacin) documented in the chart at the time of the draw?
3. Are any insulin-lowering agents (GLP-1 agonists, SGLT-2 inhibitors, metformin, TZDs) active?
4. Does the clinical picture (BMI, waist circumference, HbA1c, fasting glucose, lipids) corroborate or contradict the insulin value?
5. Was the sample hemolyzed? Hemolysis causes insulin degradation in vitro and produces spuriously low results.

The AACE position paper on insulin resistance assessment states: "Fasting insulin concentration, while not a perfect measure of insulin sensitivity, remains a practical and clinically accessible tool that should be interpreted alongside the full metabolic panel and the patient's medication list." [1]

A fasting insulin value without its drug context is incomplete clinical data. Order the draw before starting metabolically active medications whenever possible, and recheck 8-12 weeks after any significant medication change to distinguish drug effect from true physiologic shift.