ISO 15197:2013 Accuracy Standards Evolution

ISO 15197:2013 Accuracy Standards Evolution

Introduction

ISO 15197 is the international standard that sets how accurate a blood glucose meter and its test strips must be for people testing at home. Its full title covers "blood-glucose monitoring systems for self-testing in managing diabetes mellitus," so the meter and strips are judged together as one system [3]. The current edition, ISO 15197:2013, compares each meter result with a reference value, the reading a laboratory method gives for the same blood, and counts how many results land close enough [1].

History and Development

ISO published the first edition, ISO 15197:2003, in April 2003 and replaced it with the second edition, ISO 15197:2013, in May 2013 [3]. The 2003 version accepted results within ±15 mg/dL of the reference below 75 mg/dL and within ±20% at or above 75 mg/dL [2]. The 2013 edition tightened the percentage limit to ±15% and moved the cutoff up to 100 mg/dL [2]. A third edition is underway. As of October 2026, ISO's catalog lists the 2013 edition as due for revision and shows its replacement, ISO/CD 15197, at the committee-draft stage with no publication date [3].

Key Provisions

A meter has to pass two accuracy tests under the 2013 standard [1]. First, at least 95% of results must fall within ±15 mg/dL of the reference when glucose is below 100 mg/dL, and within ±15% when it is at or above 100 mg/dL [1]. The fixed window is the looser one at low readings: at 60 mg/dL, a ±15% window would be only ±9 mg/dL wide.

Second, at least 99% of results must fall in zones A and B of the Consensus Error Grid [1]. That grid was built by asking 100 diabetes experts to sort meter errors into risk categories by how much harm they could cause [6]; zones A and B are its lowest-risk zones. The standard also checks accuracy in the hands of lay users, separately from the lab-run system test [1].

In the US, FDA asks for more. Its guidance for over-the-counter meters, dated September 29, 2020, says ISO 15197's criteria "are not sufficient to adequately protect lay-users," citing performance in the low-glucose range and across test strip lots [4]. FDA recommends that 95% of results fall within ±15% and 99% within ±20% across the meter's whole claimed measuring range; a meter that can't stay within 15% at very low glucose should raise the bottom of that range instead [4]. Meters used by health care staff face tighter numbers still: 95% within ±12% (±12 mg/dL below 75 mg/dL) and 98% within ±15% (±15 mg/dL below 75 mg/dL) [5]. Both 2020 documents replaced versions FDA first finalized on October 11, 2016 [4][5].

Impact on Manufacturers

The clearest record of how manufacturers met the 2013 edition sits in published accuracy studies of specific systems. In a 2015 study, Bayer's Contour Plus system met both the 2003 and 2013 criteria in the lab, and 99.1% of fingerstick results from untrained users met the 2013 criteria [2]. Roche's improved test cassette for its Accu-Chek Mobile system was evaluated against both the system-accuracy and user-performance clauses of the 2013 edition [1].

Passing at clearance doesn't guarantee every meter on the shelf keeps passing. Here's what surprised us: when researchers bought 18 FDA-cleared meters from consumer outlets and tested them on 1,035 people, only 6 met the study's accuracy standard in all three rounds, and 4 never met it [7]. That pass mark was similar to, but more lenient than, the regulatory standards of the time [7]. FDA's own guidance leaves room here, noting that a meter can sometimes be found substantially equivalent without meeting its accuracy numbers when other features compensate [4].

Conclusion

ISO has revised the standard once since 2003. For over-the-counter meters sold in the US, FDA's 2020 guidance already asks for tighter limits than the 2013 edition [4], so an "ISO 15197:2013" claim on a box describes a minimum, not the strictest test a meter may have faced. What the third edition will change, and when it will be published, is still unknown as of October 2026 [3].