Peer groups splitting in Croatian EQA scheme: a trade-off between homogeneity and sample size number

Area of Science:

• Clinical Chemistry
• Laboratory Medicine
• Biochemistry

Background:

• External quality assessment (EQA) relies on peer group comparisons, assuming matrix effects influence results for analytes lacking commutable reference materials.
• Homogeneous peer groups are crucial for statistical evaluation in EQA schemes with diverse methods and reagents for a single analyte.

Purpose of the Study:

• To develop and apply a multivariate analysis of variance (MANOVA)-based test to assess peer group heterogeneity in the Croatian EQA biochemistry scheme.
• To identify EQA peer groups that could benefit from further splitting to improve laboratory evaluation.

Main Methods:

• EQA biochemistry results were stratified by instrument type for each analyte.
• MANOVA was employed to statistically verify significant differences between subgroups.
• Sample size calculations ensured 90% power and a maximum 5% false flagging rate increase.

Main Results:

• MANOVA identified six out of 29 evaluated peer groups as heterogeneous, warranting further splitting.
• Splitting these groups led to an overall 6% improvement in reported results.
• One specific method showed a substantial improvement of 27.4% in reported results after group splitting.

Conclusions:

• A MANOVA-based approach, combined with sample size planning and flagging rate analysis, effectively identifies heterogeneous peer groups in EQA schemes.
• Splitting identified heterogeneous groups demonstrably improves laboratory evaluation within EQA programs.
• This methodology allows for continuous monitoring of peer group homogeneity in EQA schemes.