Why does unit conversion matter in health care
Longetic keeps every lab value in the unit and with the reference range the laboratory printed, because the unit is part of the result, not a label on it. The same glucose reads 126 in mg/dL and 7.0 in mmol/L. Mix the two up and a reading looks 18 times too high, a dose can be 10 times too large, and two results that look alike may not describe the same thing.
A number without its unit is not a result. It is half of one. Most of the time the other half travels with it, printed on the report, and nobody notices. The trouble starts when results move: between two laboratories, between two countries, from a report into a spreadsheet, or from a prescription label into a syringe.
The scale of the problem is larger than most people expect. A study published in JMIR Medical Informatics in March 2026 checked 6.34 billion laboratory records from real-world health data. Before the authors cleaned them, only "73.1% (4,634,610,173/6,337,101,453) of records" had correctly assigned units. In the three electronic health record sources they examined, the share ranged from 63.3% to 78.5%.
That is the data layer. The sections below cover the other three places units matter to you directly: your lab reports, your diet, and the medicines you take every day.
What is the difference between conventional and SI units
Conventional units, used on most US lab reports, express many markers as mass per volume, such as milligrams per deciliter. SI units, from the International System of Units, express the same markers as amount of substance per liter, such as millimoles per liter. The number changes and the measurement does not. Glucose at 100 mg/dL is about 5.6 mmol/L.
The factors below come from the SI conversion table in the AMA Manual of Style, the reference JAMA Network journals use. To go from conventional to SI, multiply by the factor. To go back, divide by it. The AMA presents its table as a guide rather than a laboratory standard, and your lab's own report stays the authority for its values.
| Marker | Conventional unit | SI unit | Multiply by |
|---|---|---|---|
| Glucose | mg/dL | mmol/L | 0.0555 |
| Total, LDL and HDL cholesterol | mg/dL | mmol/L | 0.0259 |
| Triglycerides | mg/dL | mmol/L | 0.0113 |
| Creatinine | mg/dL | µmol/L | 88.4 |
| Vitamin D, 25(OH)D | ng/mL | nmol/L | 2.496 |
| Hemoglobin | g/dL | g/L | 10 |
| Ferritin | ng/mL | µg/L | 1 |
| Testosterone | ng/dL | nmol/L | 0.0347 |
| Uric acid | mg/dL | mmol/L | 0.0595 |
The famous 18 for glucose is chemistry, not convention. A mole of glucose weighs about 180.16 grams, according to the NIST Chemistry WebBook. A deciliter is a tenth of a liter, so 180.16 divided by 10 gives the 18 that turns mg/dL into mmol/L. Cholesterol is a heavier molecule, which is why its factor is so different.
A lipid panel therefore needs two factors, not one. At 200 mg/dL, cholesterol is about 5.2 mmol/L and triglycerides are about 2.3 mmol/L. Ferritin sits at the other extreme: its factor is 1, so only the label changes.
Which countries report which units
In the United States, glucose meters read mg/dL and HbA1c is reported as a percentage. In Canada, the standard unit for blood glucose is mmol/L. In the UK, HbA1c is quoted in mmol/mol and laboratories were asked to report hemoglobin in g/L from 2013. Australia reports HbA1c in both systems side by side. With results from two countries, expect one marker in two units.
| Country | What the official source states | Source |
|---|---|---|
| United States | glucose meters report mg/dL; HbA1c stays in NGSP % | FDA, NGSP |
| Canada | standard unit for blood glucose is mmol/L | Health Canada |
| United Kingdom | HbA1c quoted in mmol/mol, with % alongside | Diabetes UK |
| United Kingdom | hemoglobin in g/L, deadline March 31, 2013 | Pathology in Practice |
| Australia | HbA1c reported in % and mmol/mol together | NDSS |
The FDA states it plainly: "In the U.S., meters report results in milligrams of glucose per deciliter of blood, or mg/dL." Health Canada says the opposite for its own market: "In Canada, the standard unit of blood glucose measure is mmol/L."
This is the problem people living between two countries run into first. If you are taking results to a clinician in another country, the unit question sits alongside which medical records to bring to a doctor abroad.
Why two results in the same unit are not always comparable
A matching unit does not guarantee a matching quantity. A home glucose meter and a lab can both say mmol/L while one sample was whole blood and the other plasma. Lp(a) can be a mass or a particle count. The word "units" on a syringe and "units" of a drug are different things. Before you compare two numbers, check what each one actually counts.
| Case | What looks the same | What actually differs | Source |
|---|---|---|---|
| Glucose, meter and lab | both in mmol/L | whole blood or plasma, factor 1.11 | IFCC 2005 |
| Lp(a) | a concentration | mass in mg/dL or particles in nmol/L | EAS 2023 |
| HbA1c | one number | two scales that are not proportional | NGSP |
| Calcium | a concentration | 1 mmol is 2 mEq for a doubly charged ion | Merck Manual |
| Vitamin D | an amount | 1 mcg equals 40 IU | NIH ODS |
| Compounded semaglutide | "units" | syringe marks, not milligrams of drug | FDA 2024 |
The glucose case is the one most people meet. Whole blood and plasma carry different glucose concentrations, so the International Federation of Clinical Chemistry (IFCC) recommends reporting "the concentration of glucose in plasma (with the unit mmol/L), irrespective of sample type or measurement technique", and notes that "a constant factor of 1.11 is used to convert concentration in whole blood to the equivalent concentration in the pertinent plasma."
Equivalents are the quiet trap. The Merck Manual gives the rule as "mEq = mg/formula wt × valence = mmol × valence". For sodium, potassium or lithium, mmol/L and mEq/L are the same number. For calcium, the mEq/L figure is double.
Differences that come from the laboratory's method or calibration, rather than the unit, are a separate question. We cover them, and how much a number has to move before the change is real, in a separate piece.
Why HbA1c needs an equation, not a factor
HbA1c has two scales that are not proportional. The US reports it as a percentage under the NGSP standard, and the IFCC reports it in millimoles per mole. The NGSP publishes the conversion as an equation: IFCC equals 10.93 times the NGSP percentage, minus 23.50. So 6.5% becomes 48 mmol/mol, and 7.0% becomes 53 mmol/mol.
A single multiplier fails here because of the minus 23.50. Multiply by any one number and you are wrong at one end of the scale or the other. The reverse equation, also from the NGSP, is "NGSP = [0.09148 * IFCC] + 2.152".
If an app or a spreadsheet shows HbA1c in both units, test it with one value. Enter 6.5% and the answer should be 48 mmol/mol. Diabetes UK uses exactly that pair in its own guidance.
Why Lp(a) should not be converted between mg/dL and nmol/L
Lipoprotein(a), or Lp(a), is a common marker with no reliable conversion factor. Some labs report the mass of the particles in mg/dL, others the number of particles in nmol/L, and the particle size differs from person to person. The European Atherosclerosis Society advises against converting with a standard factor. Keep an Lp(a) result in the unit your lab used.
The society's follow-up to its 2022 consensus statement, published in Atherosclerosis in 2023, explains why: "The 2022 consensus statement does not recommend using a standard factor to convert between mg/dL and nmol/L since this would require a linear correlation between measurements in each unit. Instead, this relationship is influenced by the apo(a) isoform size of the measured Lp(a) particle and therefore specific to a particular clinical sample."
The same paper offers a rough approximation, multiplying mg/dL by 2 to 2.5, and labels it "not scientifically accurate". That range is wide. At 50 mg/dL it gives anything from 100 to 125 nmol/L, depending on which end you pick. Two Lp(a) results in different units are not directly comparable, so ask the clinician who reads them whether a repeat at one lab makes sense.
How do units affect you on a diet
Diet tracking crosses more unit systems than lab reports do. Food energy is in kilocalories in some countries and kilojoules in others, vitamin D on a label can be micrograms or international units, blood ketone meters read mmol/L, and body weight may be kilograms, pounds or stones. Each is a clean conversion, and each is easy to get wrong between two apps.
| What you track | Unit you may see | Other unit | Conversion | Source |
|---|---|---|---|---|
| Food energy | kcal | kJ | 1 kcal = 4.184 kJ | NIST |
| Vitamin D on a label | mcg | IU | 1 mcg = 40 IU | NIH ODS |
| Blood ketones, BHB | mmol/L | mg/dL | 1 mmol/L is about 10.4 mg/dL | NIST WebBook |
| Blood glucose | mg/dL | mmol/L | divide by 18 | AMA |
| Body weight | stone | kg | 1 stone = 14 lb, about 6.35 kg | UK Weights and Measures Act |
Labels follow the law of the country that sold the food. EU rules require energy in both kilojoules and kilocalories. In Australia and New Zealand, the Food Standards Code requires kilojoules and allows kilocalories only alongside them. In the United States, the FDA moved vitamin D from IU to micrograms: "For vitamins A and D, labels must declare units of measure using micrograms (mcg) rather than international units (IU)." Large manufacturers had to comply by January 1, 2020.
So a supplement bought in one country and a guideline read in another can use different numbers for the same dose. 1,000 IU of vitamin D is 25 mcg. The number shrinks by 40 times and the amount of vitamin D stays exactly the same.
Glucose meters add one more hazard for anyone who travels with one. Health Canada warned that "in some meters the displayed units of measure can be changed inadvertently if the device is dropped, the batteries are changed, or even without apparent cause". If you track glucose or ketones during a diet or a fast, the markers and apps that do it are covered in our guide to tracking health markers while fasting.
How do units affect medicines you take every day
Most unit errors with medicines are errors of tenfold or more, not small ones. Levothyroxine written as 0.025 mg instead of 25 mcg invites a tenfold slip. A "U" for units can be read as a zero. Teaspoons double the odds of a dosing mistake compared with milliliters. With compounded semaglutide, the FDA reported patients taking five to 20 times the intended dose.
| Medicine | Where the unit trap is | What was reported | Source |
|---|---|---|---|
| Levothyroxine | mg or mcg on the label | tenfold dosing errors when converting 0.025 mg and 25 mcg | ISMP Canada 2017 |
| Compounded semaglutide | mL, mg and syringe "units" | doses five to 20 times the intended amount | FDA 2024 |
| U-500 insulin | U-100 syringe marks | the actual dose was 5 times the measured one | ISMP |
| Liquid medicines for children | teaspoons or milliliters | teaspoon units doubled the odds of an error | Yin 2014 |
| Any order written by hand | "U", "IU", "1.0 mg", ".5 mg" | misread as 0, IV, 10 mg or 5 mg | Joint Commission, ISMP |
| Warfarin | prothrombin time in seconds | INR introduced so labs could agree | WHO 1983, Poller 2004 |
The semaglutide case is the most recent. In its July 2024 alert, the FDA traced errors to "confusion between different units of measurement (e.g., milliliters, milligrams and 'units')". In one pattern, "patients were directed to administer 5 units from a vial. However, these patients mistakenly administered 50 units instead." The FDA repeated its concern about people "measuring and self-administering incorrect doses" in an update current as of September 1, 2026.
The levothyroxine case is quieter and more common. ISMP Canada found that "a common calculation error occurs when converting between 0.025 mg and 25 mcg, causing in a 10-fold error in dosing", and recommends that doses "be expressed consistently in micrograms (mcg), not milligrams (mg)."
For liquids, a study of 287 parents in two emergency departments found that those using teaspoon or tablespoon units "had twice the odds of making an error", and concluded that the findings "support a milliliter-only standard to reduce medication errors."
The warfarin row is the good news. Prothrombin time results once disagreed between laboratories because each used a different reagent. The International Normalized Ratio, adopted by the World Health Organization in 1983, fixed that, and a 2004 review credited it with "greater safety and effectiveness of oral anticoagulant treatment with warfarin and related drugs." It is one of the few cases where medicine solved a unit problem by inventing a new unit.
If you take several medicines long term, write the dose with its unit on your list exactly as the label prints it, and keep it the same everywhere. Our piece on keeping one medication list across several prescribers covers what else belongs on it. Questions about a specific dose go to your pharmacist or prescriber.
What goes wrong when units get mixed up
The best-documented failures involve glucose. A reading in mg/dL is about 18 times larger than the same reading in mmol/L, and Health Canada warned that a meter showing the wrong unit could lead someone to take too much insulin. In the UK, reporting hemoglobin in both g/dL and g/L was flagged as a source of interpretation errors.
Health Canada's notice is specific: "Results in mg/dL may be 18 times higher than the same results expressed in mmol/L. If diabetic patients overestimate their blood glucose reading, they might self-administer too much insulin, which could lead to dangerously low blood glucose levels and loss of consciousness or seizure."
A hazard summary from the National Center for Patient Safety at the US Department of Veterans Affairs warned that a dropped meter could end up with a "changed unit of measure", from mg/dL to mmol/L.
Laboratories had the same problem on paper. Writing in the Annals of Clinical Biochemistry in 2011, Berg and Lane noted that "of particular concern was the reporting of haemoglobin concentration in both g/dL and g/L units, with the potential for clinical interpretation errors." The UK set a deadline of March 31, 2013 for every laboratory to report hemoglobin in g/L.
What are the most surprising facts about units in medicine
Normal body temperature is not quite 37°C anymore, the kilogram stopped being a metal cylinder in 2019, and a spacecraft was lost to a pounds-for-newtons mix-up. Inside medicine, one of the most useful units ever made, the INR, exists only because laboratories could not agree on seconds. In 2026, European lab medicine journals are still debating how to harmonize units.
| Fact | The detail | Source |
|---|---|---|
| 37°C is a 19th-century number | US body temperature has fallen 0.03°C per birth decade | Protsiv, eLife 2020 |
| Men born in the early 1800s ran warmer | 0.59°C higher than men today | Protsiv, eLife 2020 |
| The same shift happened fast elsewhere | Tsimane body temperature fell about 0.05°C a year for 16 years | Gurven, Science Advances 2020 |
| Every mmol/L rests on a constant | since May 20, 2019, one mole is exactly 6.022 140 76 × 10²³ entities | BIPM |
| The kilogram is no longer an object | the 1889 prototype definition was abrogated in 2019 | BIPM |
| Units have cost more than doses | Mars Climate Orbiter was lost to pound-seconds used for newton-seconds | NASA |
| Warfarin got its own unit | WHO adopted the INR in 1983 | Poller 2004 |
The temperature finding is the one worth knowing. The canonical figure goes back to Carl Wunderlich in the 19th century. Protsiv and colleagues compared three cohorts spanning 1860 to 2017 and found that "the normal, oral temperature of adults is, on average, lower than the canonical 37°C established in the 19th century." Among the Tsimane in Bolivia, a 2020 study found body temperature "declined as much in the past two decades as among Americans in the past 150 years."
The mole matters more than it sounds. Every result in mmol/L counts molecules, and since 2019 the BIPM defines that count exactly: "One mole contains exactly 6.022 140 76 x 10²³ elementary entities."
The Mars case is outside medicine and still the clearest warning. NASA's investigation found that trajectory data "was to be in metric units of Newton-seconds (N-s). Instead, the data was reported in English units of pound-seconds (lbf-s)." NIST puts the project total at $327.6 million.
The debate is still open. In 2026 Clinical Chemistry and Laboratory Medicine published a run of papers on the subject, including a position paper from the German society for clinical chemistry and laboratory medicine (DGKL) titled "Unity among the units", and another titled "Harmonization of measurement units: mission impossible or ethical imperative?".
How do you standardize units and reference ranges across multiple labs
Keep three things together for every result: the value, the unit and the reference range from the same report. Convert a copy when you need to compare, never the original. A range belongs to the laboratory that set it for its own method, so read a converted value against that lab's converted range, not against another lab's range.
MedlinePlus explains the reason: "labs often use different testing methods, they often use different reference ranges to describe normal results." It adds that "different tests use different units of measurement."
A workable order, if you are building one history from several labs:
- Record the unit and the range printed beside each value.
- Pick one unit system for comparison, usually the one most of your reports use.
- Convert with the factor for that marker, and convert the range with the same factor.
- Use the equation for HbA1c, and leave Lp(a) unconverted.
- Ask the clinician reading the history which unit they prefer.
Which apps handle medical unit conversions
Apps that publish a unit setting or conversion include Carrot Care, which lets you pick a display unit for each biomarker, Health3, Heads Up Health, and Apple Health for blood glucose. MDCalc and Medical Unit Converter are calculators: you type one value and get it back in the other system. Record apps keep a history, while calculators convert one number and keep nothing.
Every row below comes from the vendor's own site or store listing, checked on September 23, 2026. As a disclosure, Longetic is our product, so weigh the first row of the table below accordingly.
| App | What it does with units | Reads lab reports | Platforms | Price as published |
|---|---|---|---|---|
| Longetic | keeps the lab's unit and range | yes, PDF, photo or scan | iPhone | $9.99 a month, $69.99 a year |
| Carrot Care | display unit chosen per biomarker | yes, photos and PDFs | iPhone, Vision Pro | free, paid plans from $6.99 a month |
| Health3 | converts supported marker and unit pairs | not verified | iPhone, Android | $14.99 a month, $99.99 a year |
| Heads Up Health | imperial or metric, set at sign-up | not verified | web, iPhone, Mac | from $250 a month, for practitioners |
| Apple Health | preferred unit for blood glucose | connected providers only | iPhone | built into iOS |
| MDCalc | US and SI toggle inside calculators | no, calculator only | iPhone, iPad, Mac | free |
| Medical Unit Converter | converts one value at a time | no, calculator only | iPhone, iPad, Mac | $3.99 |
If an app converts for you, check a few values by hand against the factors above before you trust a chart built from them. Keeping the original has one advantage: the value, the unit and the range always come from the same report. Longetic is iPhone only. If you need every result in one unit system, pair it with a calculator, or use one of the apps above that publishes a unit setting.
