Impact of ISO 15197:2013 on Product Discontinuation

We went looking for meters or test strips that were pulled from the market because of ISO 15197:2013, and we didn't find a single documented case. What the record does show is narrower, and still worth knowing. The 2013 revision made the accuracy test harder, and independent labs found that many systems already on sale couldn't pass it [1]. But in Europe, where the standard carries regulatory weight, it applies to the evaluation a manufacturer runs before a system goes on sale, not to routine checks of strips already on shelves [1].

What the 2013 revision changed

ISO 15197 is the international standard for evaluating blood glucose monitoring systems used for self-testing. The 2003 edition required at least 95% of results to fall within ±15 mg/dL of a laboratory reference below 75 mg/dL, and within ±20% at or above 75 mg/dL [1]. The 2013 edition tightened that to ±15 mg/dL below 100 mg/dL and ±15% at or above 100 mg/dL. It also added a clinical-risk check: at least 99% of results must land in zones A and B of the Consensus Error Grid, a chart that sorts meter errors by how much harm they could cause [1].

The testing got heavier, too. The accuracy study now has to cover three strip lots, and each lot must pass on its own; the 2003 edition could allow a single lot [1]. The revision also added checks on "influence quantities," such as hematocrit (the share of blood made up of red cells) and interfering substances like medications [1]. One worked example in a 2015 review shows how big the jump was. A system with 98% of its results inside the 2003 limits had only 92.5% inside the 2013 limits, which is a fail [1].

Many systems on the market missed the mark

Accuracy problems predate the revision. In a 2012 evaluation of 43 systems carrying the CE mark, the label a device needs before it can be sold in the EU, 7 of the 34 systems that could be fully assessed missed even the looser 2003 limits [2]. A companion study tested four strip lots for each of five systems, where a "system" means the meter plus its strips [3]. The largest gap between two lots of the same system ranged from 1.0% to 13.0%, and only two of the five systems passed with every lot [3]. That is the problem the three-lot rule targets, and we cover it further in our piece on post-market surveillance and lot-to-lot variability.

The 2015 review put the follow-up work bluntly: "a remarkable number of systems failed to meet the more stringent accuracy criteria" of the new edition [1]. US data point the same way. A study of 18 FDA-cleared meters bought from consumer outlets, covering about 90% of the systems in use from 2013 to 2015, scored each reading against the same ±15 mg/dL and ±15% bands the 2013 edition uses [4]. Six of the 18 met the study's pass mark in all three rounds. Four never met it, even though the authors describe that pass mark as more lenient than current regulatory standards [4].

Why failing a study didn't mean leaving the shelf

Here's what surprised us: once a system is on sale in Europe, regular, standardized testing of its strip lots from routine production isn't required [1]. Even before launch, compliance with the 2013 edition was recommended only after a 36-month transition that ended in 2016 [1].

The FDA took its own path. Its over-the-counter guidance says the ISO 15197 criteria "are not sufficient to adequately protect lay-users," in part because the standard doesn't adequately address low readings or performance across strip lots [5]. Instead it asks for 95% of results within ±15% and 99% within ±20% across the meter's whole claimed range [5]. The guidance is written to help manufacturers prepare 510(k) submissions, the premarket filing for these devices. As of October 2026, the current version is dated September 29, 2020, and it replaced a final guidance from October 11, 2016 [5].

Stricter rules also left manufacturers room to adjust rather than withdraw. Reviewing the FDA's draft, the 2015 authors noted that a system unable to meet the criteria at the low end would narrow its claimed measuring range and show an error code outside it [1]. They also warned that tighter criteria have to be weighed against higher costs and possible losses in convenience [1].

A change that did reshape strips came from safety reports

The clearest reformulation we found was driven by interference, not by the 2013 accuracy limits. Strips built on the enzyme GDH-PQQ can read falsely high when the blood contains maltose or certain other sugars. A review of FDA's adverse-event database, written by LifeScan staff, found 82 reports tied to this problem, 16 of them involving a death [6]. Icodextrin dialysis solution and maltose-containing immune globulin were the agents most often involved [6].

Roche's Accu-Chek Aviva Plus strips use a modified, or mutant, GDH-PQQ that FDA cleared in September 2011 [7]. A 2010 laboratory study of Roche's modified strips for the Accu-Chek Performa meter found that maltose and xylose had no effect, although galactose still interfered [8]. That clearance came two years before the revision was published, so the timeline doesn't let us credit the standard. Roche's later Accu-Chek Guide strips, cleared in 2016, use a different enzyme, FAD-GDH [9]. Our article on GOx, GDH-FAD and GDH-PQQ chemistry explains the enzymes in more depth.

What a discontinuation record can and can't tell you

When a meter or strip does leave the US market, FDA's device-identifier database (GUDID) usually records it. Every listing for Roche's Accu-Chek Compact Plus, for example, is marked "Not in Commercial Distribution," with end dates from 2017 to 2022 [10]. The record shows when, but it has no field for why [10].

Linking a discontinuation to ISO 15197:2013 would take a manufacturer notice or regulatory record that names the standard. As of October 2026, we haven't found one. The evidence supports a smaller claim: the 2013 edition raised the bar for systems entering the EU market, and independent studies showed which existing systems fell short. Why individual products disappeared is still undocumented.