Lab Test Variability: A Road-Block in Precise, Proactive Diabetes Diagnosis and Intervention

Presentation Number: SH02-4
Date of Presentation: April 2nd, 2017

Florence Comite*1, Lorena Martin2, Jennifer Braun3, Francisco Carreño-Galvez3, Adaobi Onunkwo3 and Joel T Dudley4
1Lenox Hill Hospital, Northwell Health, New York, NY, 2UC San Diego, 3Comite Center for Precision Medicine, New York, NY, 4Icahn Institute for Genomics and Multiscale Biology


Diagnostic testing is a vital component of precision medicine and disease prevention. Biomarker measurements are used in the screening and precise, proactive management of chronic and degenerative disease. Concern about the accuracy of these tests, however, has been raised regarding the variability of testing methodology among laboratory services and the lack of clinical equivalency among the results obtained (1)(2). The present study focuses on diagnostic testing for pre-diabetes and Type 2 Diabetes, which together comprise an epidemic that threatens more than 50% of American adults (3).

We conducted a cohort study of 101 healthy adults to compare the inter-service variability of three key biomarkers commonly used for diabetes screening — hemoglobin A1c (HbA1c), fasting glucose (FG), and fasting insulin — between two companies offering these blood tests. Samples were collected at the Center for Precision Medicine in New York City. Aliquots were subsequently allocated and shipped to Quest Diagnostics and Boston Heart Diagnostics from the same sample in an identical fashion.

When comparing diagnostic results from both commercial laboratories, the disagreement in results classifications ranged from 75% to 82%. The inter-service variability was 71% (p = 10-16) for HbA1c; 75% (p = .001) for FG; and 82% (p< .01) for fasting insulin.

Cohen's κappa was run to determine if there was agreement between the companies. The agreement on HbA1c was κ = .291, p< .001, FG κ= .252, p<10-16, and fasting insulin κ= .182, p< .001.

The high level of test variability found in the present study is a significant detriment to the screening, diagnosis, and management of an individual’s risk of diabetes. While laboratory practice standards exist to control variability, disparities of this degree among testing services could substantially alter clinical interpretation, increase patient anxiety and confusion, and thus, adversely influence medical intervention. Greater transparency and standardization of testing technologies would increase their utility in personalized health management and the practical application of precision medicine in the clinic.


Nothing to Disclose: FC, LM, JB, FC, AO, JTD