STUDY OBJECTIVE: To determine if laboratories with disparate alanine aminotransferase (ALT) reference ranges would report different rates of elevation greater than the upper limit of reference range (ULRR) and thus have the potential to alter study results. DESIGN: Interlaboratory evaluation. SETTING: Research department of a poison and drug center. SAMPLES: Sixty-five serum samples collected from a previously published randomized clinical trial that were stored at -57 degrees C for approximately 2 years. INTERVENTION: Stored serum samples were divided and sent to two independent clinical laboratories (laboratory A and laboratory B) that had disparate ALT reference ranges. MEASUREMENTS AND MAIN RESULTS: Samples were stratified by ALT levels measured during the primary study to provide a spectrum of ALT values. The ULRR for serum ALT level at laboratory A was 63 and 54 U/L for male and female subjects, respectively, and for laboratory B was 31 U/L for both male and female subjects. Each laboratory used different instruments and reagents. The primary outcome was the rate of ALT elevation (> 1 x ULRR and > 3 x ULRR) for each study laboratory. The overall mean ALT activity at laboratory A was 33.6 U/L (95% confidence interval [CI] 28.1-39.2 U/L) and at laboratory B was 47.92 U/L (95% CI 40.0-55.8 U/L). Although laboratory A had the higher ULRR (74% higher for female subjects, 103% higher for male subjects), it consistently reported lower ALT values. Laboratory B reported a significantly higher proportion of samples that had an ALT value greater than 1 time the ULRR and greater than 3 times the ULRR. CONCLUSION: Our findings suggest that laboratories systemically report different values for aminotransferase levels and that the choice of laboratory may alter the proportion of patients who have an elevation in ALT levels. Until standardized reference ranges are adopted, studies reporting an outcome that depends on the reference range (such as "abnormal" or > 3 x ULRR) will be subject to misclassification bias based on the laboratory performing the analysis.
STUDY OBJECTIVE: To determine if laboratories with disparate alanine aminotransferase (ALT) reference ranges would report different rates of elevation greater than the upper limit of reference range (ULRR) and thus have the potential to alter study results. DESIGN: Interlaboratory evaluation. SETTING: Research department of a poison and drug center. SAMPLES: Sixty-five serum samples collected from a previously published randomized clinical trial that were stored at -57 degrees C for approximately 2 years. INTERVENTION: Stored serum samples were divided and sent to two independent clinical laboratories (laboratory A and laboratory B) that had disparate ALT reference ranges. MEASUREMENTS AND MAIN RESULTS: Samples were stratified by ALT levels measured during the primary study to provide a spectrum of ALT values. The ULRR for serum ALT level at laboratory A was 63 and 54 U/L for male and female subjects, respectively, and for laboratory B was 31 U/L for both male and female subjects. Each laboratory used different instruments and reagents. The primary outcome was the rate of ALT elevation (> 1 x ULRR and > 3 x ULRR) for each study laboratory. The overall mean ALT activity at laboratory A was 33.6 U/L (95% confidence interval [CI] 28.1-39.2 U/L) and at laboratory B was 47.92 U/L (95% CI 40.0-55.8 U/L). Although laboratory A had the higher ULRR (74% higher for female subjects, 103% higher for male subjects), it consistently reported lower ALT values. Laboratory B reported a significantly higher proportion of samples that had an ALT value greater than 1 time the ULRR and greater than 3 times the ULRR. CONCLUSION: Our findings suggest that laboratories systemically report different values for aminotransferase levels and that the choice of laboratory may alter the proportion of patients who have an elevation in ALT levels. Until standardized reference ranges are adopted, studies reporting an outcome that depends on the reference range (such as "abnormal" or > 3 x ULRR) will be subject to misclassification bias based on the laboratory performing the analysis.