Reagent lot verification that fails at the clinically important low end
The chemistry laboratory of a 300-bed hospital runs creatinine on two identical analysers and reports eGFR with every result. During verification of a new reagent lot, pooled serum was run 20 times at two levels: 0.6 mg/dL (within-run CV 2.1%, mean bias +0.03 mg/dL versus the laboratory's reference method) and 4.0 mg/dL (CV 1.4%, bias +2.0%). The laboratory has adopted a total allowable error of 8% for creatinine, based on biologic variation. Nephrology has complained that results run 'about 0.1 mg/dL higher' than those from the commercial laboratory across town. The laboratory is carrying two vacancies, and the night shift has a single generalist technologist responsible for both analysers, coagulation and blood gases.
1. Calculate total error at both levels and decide whether the lot can be accepted. What are the clinical consequences of the failure?
Read the worked answer
Use total error = |bias| + 1.96 x CV, expressed as a percentage of the target.
Low level (0.6 mg/dL): bias is 0.03/0.6 = 5.0%; 1.96 x 2.1% = 4.1%; total error = 9.1%, which exceeds the 8% allowable error. Sigma metric = (8 - 5)/2.1 = 1.4, well below the 4-5 usually required for a stable process without extensive quality control.
High level (4.0 mg/dL): bias 2.0% + 1.96 x 1.4% = 4.7%, which passes comfortably (sigma approximately 4.3).
The lot cannot be accepted as verified because it fails exactly where clinical decisions are made. A 9% positive error at 1.0 mg/dL puts a reported value of 1.1 mg/dL where the true value is approximately 1.0, moving a patient across the CKD-EPI eGFR boundary (e.g. from an eGFR of 60 to below 60 mL/min/1.73 m2) and potentially triggering a new diagnosis of chronic kidney disease, drug dose adjustment or referral. Reporting a value biased against the reference method also explains, at least partly, the nephrology complaint, which should not be dismissed as a comparator-method artefact.
2. Given the analyser and staffing constraints, what do you do next, and what must be documented?
Read the worked answer
Do not go live on the new lot. Continue with the current, verified lot while the problem is corrected; if that is impossible, arrange a reference-laboratory pathway or clearly defined interim reporting with clinician notification rather than releasing unverified results.
Corrective options in order of preference: recalibrate against the reference method and repeat precision and bias at 0.6 and 1.0 mg/dL; if the bias persists, investigate the calibrator assignment and reagent lot, and consider a method less prone to low-end bias (enzymatic creatinine) for the reporting range that drives eGFR. A bias correction should be applied only if it is validated, not improvised. Then perform the standard lot-to-lot comparison on at least 20 patient samples spanning the range, including several below 1.0 mg/dL, since pooled material alone does not demonstrate commutability.
Staffing is part of the risk assessment: with two vacancies and a single night generalist, do not schedule the verification on nights. Use day-shift staff, and ensure quality control review is by a qualified supervisor. Document the failed verification, the root-cause analysis, the corrective action, the repeat verification data, and the medical director's approval, and communicate the outcome to nephrology with the actual bias data.