Why Your Lactate Readings Are Inconsistent (and How to Fix It)

A 400m runner finishes a five-stage treadmill step test and the printout does exactly what it isn't supposed to do: 1.3, 1.8, 2.7, 5.8, then a dip to 5.2 at the fifth and hardest stage - the one with the highest speed and heart rate on the sheet. Nobody believes blood lactate actually fell while the athlete ran faster. The coach re-lances the same finger for a repeat draw and gets 6.1, and now the question isn't the athlete's fitness anymore, it's whether the meter is broken. It almost never is. The two places this usually goes wrong - which finger got squeezed, and how dry the skin was before the lancet went in - happen in the fifteen seconds around each draw, not inside the device.

Before chasing a technique error, it helps to know what a well-run test is even capable of. Portable lactate analyzers aren't lab-grade instruments, and some scatter is built in even when every step is done correctly.

Bonaventura et al. (2015, Journal of Sports Science and Medicine) tested six handheld analyzers, including devices from the Lactate Pro and Lactate Scout families, against a YSI 2300 STAT laboratory reference and reported test-retest coefficients of variation from roughly 2.7% to 7.9% depending on device and concentration, with error growing from a few tenths of a mmol/L at rest to over 1 mmol/L at the top of the physiological range. They also flagged hematocrit as an independent source of drift - two athletes with identical true lactate but different red-cell counts can read differently on the same device with flawless technique. The limitation: those CVs came from controlled, single-operator sampling, close to a best case; a rushed field test will usually scatter more.

That gives a working rule of thumb. A difference under roughly 10% between two draws at the same stage is unremarkable, inside the equipment's own noise floor. A difference above 20%, or a reading that fails to keep rising as work rate rises within one continuous incremental test, isn't biology - blood lactate doesn't spontaneously fall mid-test in a healthy athlete working progressively harder. If a printout shows that, look at the sampling step, not the athlete.

Where a drop of blood comes from, and how much squeezing it took to get there, changes what's actually in that drop - and it changes more as intensity climbs, exactly when consistency matters most.

Medbo et al. (2000, Scandinavian Journal of Clinical and Laboratory Investigation) compared four blood lactate instruments, including Lactate Pro and Accusport, on identical samples and found close agreement at low concentrations that widened to roughly 2 mmol/L at higher intensities. Part of that gap traced to sample handling rather than electronics: hemolyzed drops, underfilled strips, and blood drawn from a site where local circulation no longer represented core arterial blood. That's part of why standardized protocols call for warming the sampling hand, or using an earlobe once intensity climbs, rather than a cold fingertip already shut down by vasoconstriction. The limitation: much of their comparison used venous blood split across devices rather than repeated fresh punctures, so it doesn't fully capture the noise a new puncture adds stage to stage in a live test.

Squeezing compounds the problem. Milking a fingertip pulls interstitial and lymphatic fluid into the sample, carrying its own local tissue lactate unrelated to what's circulating. At low-intensity stages, where flow is sluggish and the temptation to squeeze is strongest, this inflates readings unpredictably rather than by a fixed offset - exactly why it reads as noise, not something subtractable. Pick one site for the whole test, and don't switch mid-test.

The skin around a sampling site during an exercise test is rarely dry, and that's a bigger problem than it sounds. Eccrine sweat glands secrete lactate as a byproduct of their own local metabolism, and sweat-biomarker research has repeatedly measured lactate in sweat comparable to, and often higher than, circulating blood lactate at moderate intensities. A bead of sweat rolling into a puncture site, or an alcohol swab dragged across sweaty skin before it's wiped dry, doesn't dilute the sample toward a lower reading - it adds lactate from a different compartment entirely, with no way to tell how much from the number alone.

This error is easy to miss because it doesn't look like a mistake in the moment - the site looks clean right after the alcohol wipe. The problem is the order of operations: wiping sweat away and then swabbing is very different from swabbing sweaty skin and calling it clean, since alcohol just spreads the contamination instead of removing it. In hot conditions or with heavy sweaters, that means a dry-gauze pass before every alcohol swab, not just the session's first one - sweat reaccumulates within minutes once exercise resumes between stages.

Past the puncture, a handful of habits around the strip and meter produce readings that look like biological noise but trace back to storage and application.

Bonaventura et al.'s 2015 comparison, cited above for its CV figures, made the same point about hematocrit: an athlete's red-cell count is a variable the meter can't see, and it interacts with strip chemistry in ways that mimic inconsistency. It isn't a handling error to fix, but rule it in before hunting for one that isn't there - unusual hematocrit explains readings offset from a lab reference across a whole history, not one stage jumping out of line.

Most of the errors above share one fix: the same steps, in the same order, on every draw.

1. Warm the sampling hand for a minute, or use the earlobe from the outset, so blood flow is representative rather than vasoconstricted.
2. Wipe visible sweat away with dry gauze before the alcohol swab touches the site - sweat first, alcohol second, never the reverse on wet skin.
3. Swab with alcohol and wait 15-20 seconds for it to evaporate; fanning it works, but don't dab dry with the same used swab.
4. Lance to a consistent depth (typically 1.5-2.1mm for an adult fingertip) and let the first drop form on its own without squeezing.
5. Wipe away that first drop with dry gauze - it carries the most interstitial fluid and surface contamination - and test only the second drop.
6. If a second drop doesn't form within a few seconds, apply gentle pressure below the puncture rather than milking the puncture site itself.
7. Present a full, rounded hanging drop to the strip's capillary channel in one continuous touch, and discard the strip if the fill indicator doesn't confirm a complete draw.
8. Log the site, digit, and drop number alongside the mmol/L value for every draw, so a later outlier traces back to a specific step instead of equipment error.

The table below is the same athlete's five-stage cycling step test run twice: once with the fingertip-milking-plus-mid-test-site-switch technique from the opening example, and once with the eight-step protocol above.

Every old-technique value reads higher than the clean-protocol value until stage five, where it drops instead of rising. Stage four's 5.8 came from an aggressive milk on a cold, sweaty fingertip unnoticed since stage two. Stage five's 5.2 came from switching to the earlobe without logging it - a reasonable-looking number on its own that broke the trend next to the rest of the curve. The clean-protocol column, run with one site throughout, rises monotonically the way a continuous incremental test should, and it's the version worth trusting for locating this athlete's threshold.