Decision Making in the Clinical Laboratory:
A Statistical Approach for Methods Evaluation

Robert L. Klick

Department of Clinical Laboratory Sciences
State University of New York at Buffalo

November 1997

(Users may earn credits for this talk through November 1999)

Abstract

The presentation will introduce the basic principles and statistical tests which can easily be applied to decision making in the clinical laboratory. Most of the statistical tools to achieve these goals are incorporated into many commercial spreadsheet packages, such as Excel (Microsoft) and QuattroPro (Corel).

Topics to be covered:

What federal CLIA regulations mandate and what the Joint Commission on Accreditation of Healthcare Organizations lists as be essential for quality laboratory service.
Are the variances (measure of precision) of two methods statistically the same or different at some confidence level?
Comparison of sample (population) means from to methods. Are the mean results from one method statistically different or the same as the alternative method?
Analysis of Outliers
Computation of within-run, between-run and day-to-day precision.
Determination of the analytical (reportable) range of an assay.
Regression analysis and identification of types and causes of errors between the compared methods.
Determination of the reference interval (reference range) for a partitioned population.

SPECIAL PRESENTATION BONUS!

Included with this presentation is an Excel spreadsheet template for performing a Demings regression analysis. As outlined in this talk, the Demings model is especially useful for method comparisons in the clinical laboratory. The fully functional spreadsheet code is open and non-protected, allowing customization for your own use.

LEVEL: INTERMEDIATE

Learning Objectives

Upon completion of this presentation the participant will be able to do the following:

Identify the initial step in the process of accepting a new analytical method.
Name an appropriate statistical test used to compare user and manufacturer precision studies.
Describe the criteria for objectively eliminating "outliers".
Describe the necessary data when calculating reproducibility studies.
Determine the acceptability of a method based upon CLIA 88 "acceptable performance criteria".
Discriminate among random, constant or proportional error, given X-Y plots of data.
Calculate %CV, given descriptive statistics for a data set.
Contrast "Deming" regression analysis vs. standard (linear) regression analysis for method comparisons in the clinical laboratory.
Given values for "within-run" and "between-day" precision, calculate Total Precision.

P.A.C.E. Credits and Cost

Succesful completion (by electronic post-quiz) will earn participants 4.0 Contact Hours of P.A.C.E. credit.

This is a FREE presentation for Med TechNet subscribers; others: $29.00 (US Dollars)

Hardcopy Reference Service

For those who simply wish to obtain a copy of the presentation, without accessing the online discussion or earning Continuing Education Credits, we offer a special Reference Service. To receive a bound, full-color copy of this 27 page presentation plus disks with the Electronic Version, send a check or money order, for $24 (US Funds, payable to: WNY MICRO, INC.; NY State residents add local sales tax) and request MTCON19-MethodEvaluation-NOV97. (Prices applicable to US, Canada and Mexico. Overseas customers should check with us first for pricing; send inquiries to services@medtechnet.com.)

Requests should be mailed to:
WNY Micro, Inc./Med TechNet
PO Box 84
East Amherst, NY 14051

--OR--

Use your MasterCard/Visa, NOVUS/Discover, Diners Club or Carte Blanche credit card and call 1-800-836-0720, M-F, 9am - 4pm (Eastern), or use the following order form and FAX 24 hours-a-day.

--OR--

Order Online! from Med TechNet's Online Catalog

HARCOPY REQUEST FOR: MTCON19-MethodEvaluation-NOV97 (includes Deming's Excel spreadsheet)

Name:_____________________________________
Adr1:_____________________________________
Adr2:_____________________________________
City:_____________________________________
State:____________________________________
ZIP:______________________________________

_/ MasterCard/Visa _/ NOVUS/Discover _/ Diners Club _/ Carte Blanche

Card Number:____________________________________________
Expiration :_____________

FAX TO: 1-716-636-8235 or 1-716-688-0724

Format

Med TechNet presentations are distributed in Adobe's PDF format (Portable Document Format) for viewing on PC's, Mac's and other platforms. Platform specific readers (DOS, Windows and Mac) are available at no charge from many sites.

Med TechNet is approved - OR - has applied for approval - as a Provider of continuing education programs in the clinical laboratory sciences by the ASCLS P.A.C.E.Ž Program.

Med TechNet credits are approved for clinical laboratory continuing education requirements in California.

Vendors, Consultants and Organizations!!!

You play a critical part in our presentations. It is only through corporate sponsorship that we are able to keep the costs of our programs low. Support the professionals that support you; become a presentation sponsor!

Med TechNet accepts sponsorship positions for presentations throughout the year, with plans for every budget.

E-Mail to, services@medtechnet.com, or, call 1-800-836-0720 for more information.

Med TechNet Subscription Information

Decision Making in the Clinical Laboratory: A Statistical Approach for Methods Evaluation