Post-Market Surveillance and Lot-to-Lot Variability

Post-Market Surveillance and Lot-to-Lot Variability

Every vial of test strips comes from a manufacturing batch called a lot, and two lots of the same strip don't always read the same. In one laboratory study of five meter systems, the largest gap in average bias between two lots of the same system ranged from 1.0% to 13.0%[1]. FDA clearance checks a meter and its strips before they go on sale. Post-market surveillance is everything that checks them afterward, and when we read the research, that second half turned out to be the thinner one.

What lot-to-lot variability looks like

The usual measure is bias: how far, on average, a strip's readings sit above or below a laboratory reference value for the same blood. A team in Ulm, Germany, tested four strip lots for each of five systems, with blood from at least 100 people per lot[1]. Single lots ranged from 17.3% below the reference to 9.5% above it. Only two of the five systems met the accuracy standard then in force with every one of their lots[1].

The spread inside one system could be wide. One lot of a system put 100% of its readings inside the limits of the then-draft revision of ISO 15197; another lot of that same system managed 52.5%[1]. Here's what surprised us: the authors found that some of these lot-to-lot gaps were about as large as the differences between entirely different meter systems[1].

Bias that stays steady matters less than bias that moves. The authors pointed out that people may adjust, over time, to a meter that reads a little high, but they can't adjust to a bias that keeps shifting from lot to lot[1]. The study didn't identify what caused the differences.

How accuracy is tested before clearance

The FDA's guidance for home-use meters, finalized in September 2020, says ISO 15197 alone isn't enough to protect people who test at home, in part because the standard doesn't adequately address performance across strip lots[2]. So the FDA asks for a study in which lay users test themselves with strips from at least 10 vials covering at least 3 lots, shipped and handled the way strips normally are. The target: 95% of results within ±15% of the laboratory comparison and 99% within ±20%, across the meter's whole measuring range[2].

ISO 15197:2013 sets a different bar. At least 95% of results must fall within ±15 mg/dL of the reference below 100 mg/dL and within ±15% at or above 100 mg/dL, and at least 99% must land in zones A and B of the Consensus Error Grid[3]. Studies run under its system-accuracy clause test three reagent lots[3]; while the edition was still a draft, the Ulm team noted that each lot would have to meet the criteria on its own[1]. Our ISO 15197:2013 explainer goes further into the standard.

Lot release: the check on every batch

Clearance studies cover a handful of lots. Lot release criteria cover every lot after that. The FDA guidance asks manufacturers to sample each outgoing lot against criteria designed so released strips perform as labeled in users' hands, and says those criteria typically should be tighter than the ones used in the clearance studies[2].

Manufacturers also have to explain how the system compensates for differences between lots, meaning how the meter is calibrated or coded for each one[2]. On older meters the user did this, by inserting a code chip or code strip or typing in a code number[4]. A 2008 study that deliberately miscoded meters concluded that miscoding creates the potential for insulin dose errors and clinically significant lows or highs, while autocoded meters could not be miscoded at all[4]. Our article on calibration codes vs. no-code technology covers that shift.

What happens after a meter is on the market

In 2016, the Diabetes Technology Society (DTS) published a surveillance protocol, written with co-authors from the FDA, the National Institutes of Health and the Centers for Disease Control and Prevention. Its reason: once the FDA clears a meter, "no systematic postmarket Surveillance Program exists" to monitor how accurately it performs[5].

The first results came out in 2018. Researchers bought 18 FDA-cleared systems from consumer outlets, together about 90% of the systems in use from 2013 to 2015, and tested them on 1,035 people in three blinded studies[6]. The pass mark was "similar to, but more lenient than" the regulatory standards. Six systems passed in all three studies. Four passed in none, and the authors concluded that cleared meters don't always deliver the accuracy required for clearance[6].

The FDA's own monitoring depends mostly on reports. A 2019 commentary noted that the FDA had described its system as "largely passive," relying on device users to report problems, and said it was moving toward active surveillance[7]. Those reports go into the Manufacturer and User Facility Device Experience (MAUDE) database, which received 10,837 reports for blood glucose monitoring systems from January through July 2018[7]. The same commentary shows why testing many lots helps: when a drug trial's meter came under suspicion, every strip lot shipped to the test site from several locations showed the same bias, pointing to a systematic problem rather than improper storage[7].

How problems reach people who test

Some of the most serious problems first showed up as reports. A 2010 review by LifeScan staff searched MAUDE and the medical literature for falsely high readings on GDH-PQQ strips (strips using the enzyme glucose dehydrogenase with the cofactor pyrroloquinolinequinone) when interfering sugars such as maltose were present. It found 82 reports, 16 of them associated with death; most were in hospitals, but at least 14 involved outpatients[8].

Recalls and FDA safety communications are the most direct channel. In April 2026, the FDA warned about a Class I recall, the most serious type, covering all TRUE METRIX, TRUE METRIX AIR, TRUE METRIX GO and TRUE METRIX PRO meters, store brands included[9]. The meters showed the same E-5 error code for a very high glucose level and for a strip error. As of January 16, 2026, Trividia Health had reported 114 serious injuries and one death linked to the issue[9]. The FDA recommended moving to another testing method when possible, without stopping testing in the meantime, and asked people to report problems through MedWatch[9]. Which replacement makes sense is a question for your care team or pharmacist.

The FDA has also cleared a modified TRUE METRIX meter that moves very-high-glucose strip errors from E-5 to a new "EHi" code[10]. That's the whole loop in one case: injury reports, a recall, a public warning, then a design change that went back through FDA review.

What still isn't known is how much ordinary lot-to-lot drift affects the people using the strips. The Ulm authors wrote in 2012 that its clinical impact "has not been studied thoroughly," and they called for strip lots from routine production to be checked by manufacturers or by an independent institution[1].