Browsing by Subject "Mixed effects models"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Open Access Statistical methods for the assessment of EQAPOL proficiency testing: ELISpot, Luminex, and Flow Cytometry.(Journal of Immunological Methods, 2014-07) Rountree, Wes; Vandergrift, Nathan; Bainbridge, John; Sanchez, Ana M; Denny, Thomas NIn September 2011 Duke University was awarded a contract to develop the National Institutes of Health/National Institute of Allergy and Infectious Diseases (NIH/NIAID) External Quality Assurance Program Oversight Laboratory (EQAPOL). Through EQAPOL, proficiency testing programs are administered for Interferon-γ (IFN-γ) Enzyme-linked immunosorbent spot (ELISpot), Intracellular Cytokine Staining Flow Cytometry (ICS) and Luminex-based cytokine assays. One of the charges of the EQAPOL program was to apply statistical methods to determine overall site performance. We utilized various statistical methods for each program to find the most appropriate for assessing laboratory performance using the consensus average as the target value. Accuracy ranges were calculated based on Wald-type confidence intervals, exact Poisson confidence intervals, or via simulations. Given the nature of proficiency testing data, which has repeated measures within donor/sample made across several laboratories; the use of mixed effects models with alpha adjustments for multiple comparisons was also explored. Mixed effects models were found to be the most useful method to assess laboratory performance with respect to accuracy to the consensus. Model based approaches to the proficiency testing data in EQAPOL will continue to be utilized. Mixed effects models also provided a means of performing more complex analyses that would address secondary research questions regarding within and between laboratory variability as well as longitudinal analyses.Item Open Access The Immunology Quality Assessment Proficiency Testing Program for CD3⁺4⁺ and CD3⁺8⁺ lymphocyte subsets: a ten year review via longitudinal mixed effects modeling.(Journal of Immunological Methods, 2014-07) Bainbridge, J; Wilkening, CL; Rountree, W; Louzao, R; Wong, J; Perza, N; Garcia, A; Denny, TNSince 1999, the National Institute of Allergy and Infectious Diseases Division of AIDS (NIAID DAIDS) has funded the Immunology Quality Assessment (IQA) Program with the goal of assessing proficiency in basic lymphocyte subset immunophenotyping for each North American laboratory supporting the NIAID DAIDS HIV clinical trial networks. Further, the purpose of this program is to facilitate an increase in the consistency of interlaboratory T-cell subset measurement (CD3(+)4(+)/CD3(+)8(+) percentages and absolute counts) and likewise, a decrease in intralaboratory variability. IQA T-cell subset measurement proficiency testing was performed over a ten-year period (January 2003-July 2012), and the results were analyzed via longitudinal analysis using mixed effects models. The goal of this analysis was to describe how a typical laboratory (a statistical modeling construct) participating in the IQA Program performed over time. Specifically, these models were utilized to examine trends in interlaboratory agreement, as well as successful passing of proficiency testing. Intralaboratory variability (i.e., precision) was determined by the repeated measures variance, while fixed and random effects were taken into account for changes in interlaboratory agreement (i.e., accuracy) over time. A flow cytometer (single-platform technology, SPT) or a flow cytometer/hematology analyzer (dual-platform technology, DPT) was also examined as a factor for accuracy and precision. The principal finding of this analysis was a significant (p<0.001) increase in accuracy of T-cell subset measurements over time, regardless of technology type (SPT or DPT). Greater precision was found in SPT measurements of all T-cell subset measurements (p<0.001), as well as greater accuracy of SPT on CD3(+)4(+)% and CD3(+)8(+)% assessments (p<0.05 and p<0.001, respectively). However, the interlaboratory random effects variance in DPT results indicates that for some cases DPT can have increased accuracy compared to SPT. Overall, these findings demonstrate that proficiency in and among IQA laboratories have, in general, improved over time and that platform type differences in performance do exist.