Lactate Pro 2 Analysis: Where a 0.2 mmol/L Reading Error Actually Hurts
Lactate Pro 2 analysis explained: how to log readings, how many strips to budget, and why a 0.2 mmol/L error moves LT1 by 10 watts but LT2 by under 5.
A Lactate Pro 2 gives you one number per stage. Everything that deserves the name *analysis* happens after that: pairing each reading with the workload it belongs to, fitting a curve, and reading LT1 and LT2 off the shape. The meter does not calculate a threshold, and a better meter will not rescue a test where the resting sample was rushed.
That makes the useful question narrower than "how accurate is this device". The question is which of your readings the threshold calculation is actually leaning on. In the worked test below, it is not the stage where lactate finally climbs. It is the flat section near the bottom that most people sample once and then forget about.
What the meter contributes, and where the protocol takes over
Arkray describes the LT-1730, sold as Lactate Pro 2, as a palm-sized analyser that measures from a 0.3 µL blood sample (Arkray product page). Distributor listings put the on-screen result at around 15 seconds (Praxisdienst listing). Treat spec numbers like that as a prompt to check the measuring range, resolution and operating temperature printed on your own unit and your own strip vial, since those details vary by market and lot. The analysis below assumes readings to one decimal place, which is what step-test math is normally built on.
Two independent evaluations are worth reading before deciding how much weight a single value can carry. Bonaventura and colleagues compared six hand-held analysers in 2015 and reported the Lactate Pro 2, together with the Edge, as the better of the portable devices tested at concentrations above 15 mmol/L, while the original Lactate Pro, the Scout+ and the Xpress each showed root-mean-square errors in the region of 4–10 mmol/L in that range (PMC4306774). A 2021 paper examined the Lactate Pro 2 specifically for reliability and validity, together with the workload, heart rate and oxygen consumption recorded at lactate threshold (Crotty et al., 2021). Note where the 2015 headline lives: above 15 mmol/L. An incremental step test spends most of its life between 1 and 5 mmol/L, so device rankings drawn from the top of the range describe a part of the curve you will rarely visit.
The worked test
The numbers below are a constructed example rather than a real athlete: a cycling step test, 3-minute stages, 20 W increments, one capillary sample at the end of each stage.
| Stage | Power (W) | Lactate (mmol/L) | HR (bpm) | |---|---|---|---| | Rest | – | 1.1 | 62 | | 1 | 140 | 1.2 | 118 | | 2 | 160 | 1.3 | 128 | | 3 | 180 | 1.4 | 137 | | 4 | 200 | 1.9 | 146 | | 5 | 220 | 2.8 | 156 | | 6 | 240 | 4.3 | 165 | | 7 | 260 | 6.6 | 174 |
Two thresholds, two rules: LT2 by the fixed 4 mmol/L convention, LT1 as the resting value plus 0.4 mmol/L, which is 1.5 mmol/L here. Both use straight-line interpolation between the two stages that bracket the target — the same arithmetic any calculator runs before it shows you a zone table. Method choice moves the answer as well, and by more than measurement noise does: on our four-method comparison of a single dataset the spread between methods was 33 W, while the measurement noise below moves the answer by 4 to 16 W.
LT2: the steep part is forgiving
220 + 20 × (4.0 − 2.8) / (4.3 − 2.8) = 236 W
Now assume one of those two readings is off by 0.2 mmol/L, in either direction:
- stage 6 reads 4.1 → 238.5 W
- stage 6 reads 4.5 → 234.1 W
- stage 5 reads 3.0 and stage 6 reads 4.1 → 238.2 W
- stage 5 reads 2.6 and stage 6 reads 4.5 → 234.7 W
Total spread: 234.1 to 238.5 W. That is under 4.5 W, about 1.9 % of the threshold power. The curve is climbing 1.5 mmol/L per stage there, so a small vertical error barely shifts the horizontal crossing point.
LT1: the flat part is not
180 + 20 × (1.5 − 1.4) / (1.9 − 1.4) = 184 W
The same ±0.2 mmol/L, applied to the two bracketing stages:
- stage 3 reads 1.3 and stage 4 reads 1.7 → 190 W
- stage 3 reads 1.5 and stage 4 reads 2.1 → 180 W
Spread: 10 W, about 5.4 %, more than twice the LT2 spread from an identical measurement error. The difference is slope. Between 180 and 200 W the curve rises 0.5 mmol/L; between 220 and 240 W it rises 1.5. Dividing by a small number magnifies whatever noise sits in the numerator.
There is a third strip in that equation that people forget: the resting one. Because LT1 is defined here relative to baseline, an error there shifts the target concentration itself. A resting sample reading 1.3 instead of 1.1 moves the target to 1.7 mmol/L and LT1 to 192 W — 8 W from a single strip taken before the athlete started pedalling. Stack all three errors in the same direction (rest 1.3, stage 3 at 1.3, stage 4 at 1.7) and LT1 lands on 200 W, 16 W above nominal.
In heart-rate terms, the 180–190 W band maps to 137–142 bpm on this dataset. If someone sets an easy-endurance ceiling from LT1 heart rate, that is five beats of ambiguity, and it is worth raising with a coach or clinician rather than resolving inside a spreadsheet.
The duplicate strip belongs at the bottom of the curve
Common practice is to spend spare strips around the expected 4 mmol/L crossing. The arithmetic above argues for the opposite. A sampling plan for a seven-stage test:
- 1 strip at rest, taken before the warm-up, seated, with the fingertip clean and dry
- 7 strips, one per stage
- 2 duplicates on the stages you expect to bracket LT1, usually stages 2 to 4
- 2 spares for a smeared drop or a strip that errors out
That is 12 strips for 7 stages, so budget roughly 1.7 strips per stage rather than 1.0. If duplicates do not fit the budget at all, duplicate the resting sample and nothing else; that one strip enters the LT1 calculation twice.
Set the duplicate tolerance before the test, not after you have seen the curve. A workable house rule: if two samples from the same stage agree within 0.3 mmol/L, record the mean; if they disagree by more, flag the stage as uncertain and carry that flag into the write-up. What you should not do is keep whichever value makes the curve look tidier. That is how a test acquires false precision.
Four ways the sheet goes wrong before the maths starts
Baseline taken after the warm-up. A sample drawn after 15 minutes of pedalling is not a resting sample. With a baseline-referenced LT1 rule it inflates the target and pushes LT1 upward, exactly as the 192 W case shows.
Contamination at the fingertip. Sweat and skin residue are a known route for corrupting capillary samples, which is why protocols call for drying the site and discarding the first drop. A single unexplained spike in an otherwise smooth series is the signature.
A changed protocol. Blood lactate at a given stage reflects production and clearance up to the moment you sample. Lengthen the stage from 3 to 5 minutes, or sample 30 seconds later than last time, and you are reading a different point of that balance. Treat the first test after any protocol change as a new baseline series rather than a continuation.
A silently deleted outlier. Removing a point without recording that you removed it makes the next comparison meaningless. Keep the raw column and the analysed column side by side in the same sheet.
Reading the output without over-reading it
Once the values are entered, whether into an online curve fit or your own sheet, you get thresholds in watts or pace, an interpolated heart rate at each, and zone boundaries derived from them. Reasonable things to take from that: the shape of the curve, the direction of change against a previous test run under the same protocol, and where the data is thin.
Less reasonable: reading a 3 W change as progress when your own LT1 spread is 10 W, or treating any of it as a health assessment. Training decisions built on these numbers belong with a qualified coach, and health questions — including whether repeated capillary sampling is appropriate for you if you have a pre-existing condition — with a clinician.
Questions that come up before the first test
Is the Lactate Pro 2 sold in the United States? Availability and regulatory labelling differ by country and change over time, and a sports or research listing is not the same as a clinical one. Check the manufacturer's product page and regional distributors, and confirm which use the listing covers before ordering.
Can I compare my readings with a lab analyser's? Treat them as separate series. The 2015 six-analyser comparison found that error varied by device and by concentration range (PMC4306774), so a gap between a hand-held reading and a lab value is not automatically a fault in either. If you switch systems, restart the trend line.
What lactate level is concerning? That is a clinical question and belongs with a clinician. A sports meter used in a step test is not a diagnostic instrument, and a value that looks dramatic next to a training chart carries no clinical meaning on its own.
Which meter should I buy? It depends on strip supply, sample volume, and how many tests you will genuinely run in a year; our selection framework walks through the trade-offs.
Next test
Run the identical protocol twice inside one week, duplicate the resting sample and the two stages below LT1 in both, then compare the two LT1 values. That spread — your hands, your meter, your strip lot — is the resolution you actually have. Any change smaller than it is not yet news.