Skip to main content
ARS Home » Plains Area » Grand Forks, North Dakota » Grand Forks Human Nutrition Research Center » Dietary Prevention of Obesity-related Disease Research » Research » Publications at this Location » Publication #76111
Title: PATIENT-SPECIFIC ANALYSIS OF SEQUENTIAL HEMATOLOGICAL DATA BY MULTIPLE LINEAR REGRESSION AND MIXTURE DISTRIBUTION MODELING

Author
item MCLAREN, C - MOORHEAD STATE UNIVERSITY
item KAMBOUR, E - TEXAS A & M UNIVERSITY
item MCLACHLAN, G - UNIVERSITY OF QUEENSLAND
item Lukaski, Henry
item LI, X - VETERANS AFFAIRS
item BRITTENHAM, G - CASE WESTERN RESERVE UNIV
item MCLAREN, G - VETERANS AFFAIRS

Submitted to: Statistics in Medicine
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/16/1996
Publication Date: N/A
Citation: N/A

Interpretive Summary: Physicians rely on blood biochemical measurements and their clinical experience to determine the course of treatment for severe iron deficiency. While this medical practice is well established, it is limited because individuals differ in their response to oral iron therapy. To address this problem, we utilized a standard statistical model that incorporates an individual patient's data on standard blood biochemical measures of iron nutrition to develop biologically important target values that indicate improvement in iron nutritional status even when values are within the range of normal values. The model was evaluated in a sample of adults who were made iron deficient by controlled blood loss. Use of this novel approach showed that a new index of iron status, red blood cell distribution, is a more powerful indicator of individual response to iron therapy than traditional measures used in clinical practice. This approach will be useful to nutritionists studying iron metabolism in humans, and perhaps animals, and to clinicians who treat individuals with iron deficiency anemia.

Technical Abstract: Automated storage and analysis of the results of serial hematologic studies are not technically feasible with current laboratory instruments and devices for data storage and processing. In practice, physicians mentally compare a laboratory result with previous values and use their clinical judgement to determine the significance of any change. We describe a new approach, and its statistical basis, for the detection of significant changes in patient-specific sequential measurements of standard hematologic laboratory tests. These methods include hierarchical multiple regression modeling with construction of confidence regions for changes in mean values over time and a new simultaneous significance test for changes in the mean response. This is the first study to analyze sequential patient-specific distributions of laboratory measurements, utilizing mixture distribution modeling with systematic selection of starting values for the EM algorithm. To evaluate these statistical methods, we studied 11 healthy adult volunteers who were made anemic by serial phlebotomy, then treated with oral iron supplements to replete iron stores and correct the anemia. Application of sequential patient-specific analyses of hemoglobin, hematocrit, and mean cell volume showed that significant departures from past values could be identified, in most cases, even when values were still within the population reference ranges. Additionally, for all subjects sequential alterations in red blood cell volume distributions during development of iron-deficiency anemia could be characterized and quantified. These methods promise to provide more sensitive techniques for improved diagnostic evaluation of developing anemia and serial monitoring of response to therapy.