Closed analyser lids do not reduce the measurement uncertainty of serum total carbon dioxide

To the Editor,

We read with interest the report by Cuykx et al. discussing possible causes affecting variability of total carbon dioxide (tCO₂) measurements in serum. ¹ We were particularly intrigued by the fact that, in the authors’ experiments, the systematic closure of analyser lids appeared to reduce measurement variance. Therefore, we would like to provide our experience about the estimate of random sources of measurement uncertainty (u_Rw) of tCO₂ in serum performed on different generations of chemistry analysers equipped with different options of closing lids.

In the recent years, our total laboratory automation system was first equipped by two Architect c16000 platforms (in which in daily practice reagent chamber lids were often not closed) and then replaced with two Alinity c analysers and lately by two platforms of the new generation of Alinity cic series, in which the lids close automatically (all instruments by Abbott Diagnostics, IL, US). For the three generations of analysers, Abbott provided the same enzymatic method (phosphoenolpyruvate carboxylase-based) and a dedicated calibrator (ref. 08P72/1E64) with values traceable to NIST SRM 351. As per laboratory accreditation duties, we routinely investigated u_Rw by daily measurements of a third-party liquid frozen commutable internal quality control (IQC) material (Liquichek Chemistry Control Level 2, Bio-Rad Laboratories, CA, US) randomly analysed during ordinary laboratory activities. ² This material was previously tested for stability of tCO₂ measurements to exclude the influence of unknown components potentially deteriorating it (data not shown). According to the ISO/TS 20914:2019, u_Rw should be estimated from six consecutive months IQC daily data to capture systematic sources of uncertainty, such as those caused by different lots of reagents, different calibrations, different environmental conditions, etc. ³ Steps to obtain the estimate of u_Rw have been previously described in detail. ⁴

Table 1 displays results for u_Rw related to the evaluated measuring systems. Although Cuykx et al. reported that new generation platforms (by implementing a systematic closure of analyser lids) may improve tCO₂ measurement stability, our results show that u_Rw was better when tCO₂ was measured by Architect rather than on the newest generations of Abbott platforms. Very importantly, when performance specifications for maximum allowable standard measurement uncertainty (MAU) (derived from biological variation data published by Harris et al. ⁵ ) were considered, that is, 2.10% (desirable quality level) and 3.15% (minimum quality level), only Architect appeared to fulfil the minimum quality goal. Our Architect results were also better than those reported by Cuykx using c8000 version (2.28% vs 3.69%, at similar target tCO₂ concentrations) employed by closing lids under favourable weather conditions.

Table 1.

Standard measurement uncertainty from random sources (uRw) for tCO2 in serum measured using Abbott diagnostics chemistry platforms available in the authors’ laboratory at different times.

Platforms	No. of determinations over a 6-month period	Mean tCO2 concentration (mmol/L)	uRw (%)
Architect c16000	121	29.3	2.28
Alinity c08	133	32.8	4.16
Alinity c09	125	32.4	3.94
Alinity cic1	139	30.6	4.82
Alinity cic2	138	30.1	4.51

Our results show that the main contributor of tCO₂ measurement variability is the reproducibility of the employed measurement system. Particularly, we highlighted the worsening of the analytical performance between successive generations of Abbott platforms. We already recommended that manufacturers, before releasing any kind of technological upgrade, should pay major attention to assuring an equivalent, if not better, analytical quality of provided measurements. ⁶ We therefore expect that the involved manufacturer should direct its efforts on improving performance of tCO₂ measurements in terms of random variability to permit to fulfil suitable MAU.