Danish Medical Bulletin    |    Contact Front page  

 
Prediction of changes in bone mineral in early postmenopausal women

Peter Vestergaard  


  Download file   (3827 KB)

Accepted by: Faculty of Health Aarhus University
Defended on: May 1, 2000
Official opponents: Claus Hagen , Jens Otto Lunde Jørgensen , Lars Hyldstrup
Tutors: Leif Mosekilde , Henning Beck-Nielsen , Mogens Erlandsen

Published in the PhD Database: August 8, 2006


English abstract
Aim: To study the possibilities of establishing a simple, economically affordable, easy-to-use algorithm to select women at high risk of having low bone mineral at menopause or at high risk of losing a large amount of bone after the menopause.
Material and methods: A total of 595 women aged 45-58 years participating in the Danish Osteoporosis Prevention Study at the Aarhus Centre. The algorithms were then validated using 555 comparable subjects.
The associations with a number of biochemical, life-style and antropomorphometric variables were studied both in a cross-sectional design and in a longitudinal design over five years.
Results:
Cross-sectional study:
There was an intricate interaction between the independent variables, especially prominent for the biochemical markers of bone turnover that were significantly intercorrelated.
Lumbar spine: There was a general agreement between the multiple regression, the logistic regression, and the discriminant function that serum osteocalcin, and serum BAP were significantly negatively associated with cross-sectional BMD, while BMI was positively associated. The optimal cut point for separating between those with high and those with low BMD was at a T-score of -0.5
Femoral neck: The optimal cut point in a ROC analysis was a T-score of -1. Age and serum Osteocalcin were generally associated with lower BMD, while BMI was associated with a higher BMD.
Ultradistal forearm: At this site only multiple regression was performed. This showed age, daily alcohol consumption, serum osteocalcin, prior fracture, and serum BAP to be negatively associated with BMD while vitamin D intake and BMI were positively associated with BMD.
Longitudinal study:
This study showed that loss rates were normally distributed without signs of a separate group of ¿fast-losers¿. Correlations between independent variables and BMD changed with time.
Lumbar spine: The optimal cut point analysis lay at a loss of more than -1 standard deviations below the mean. Physical activity was a predictor of a higher loss.
Femoral neck: The optimal cut point was at more than 0.5 standard deviations below the mean loss. Increasing age on inclusion was associated with a smaller bone loss.
Ultradistal forearm: At this site no agreement on any predictor variables among the functions.
Testing:
When trying to implement the developed predictive model on another centre, the model performed poorly with systematic deviations between predicted and observed values.
Conclusions:
It is not possible to predict regional cross sectional bone mineral or the loss rate of bone mineral after the menopause through the use of proxy-variables (biochemical and other variables) to such a degree that it will not be useful in general screening without actually measuring bone mineral content in the subjects.



Danish abstract
Formålet var at studere mulighederne for at opstille en simpel, lettilgængelig metode til at identificere kvinder, der er i høj risiko for at have eller at udvikle lavt knoglemineralindhold efter menopausen. I alt 595 kvinder i alderen 45-58 år, der deltog i Danish Osteoporosis Prevention Study. De opstillede metoder blev siden valideret på en uafhængig gruppe på 555 kvinder. Undersøgelserne blev foretaget i et tværsnits-design, ved opfølgning over 5 år. En række biokemiske, antropomorfometriske og livsstils-parametre blev undersøgt.
Tværsnitsstudie:
Lænderyg: Serum osteocalcin og serum BAP var negativt associerede til tværsnits BMD. BMI var positivt associeret til BMD. Det optimale adskillelses-punkt mellem højt og lavt BMD lå ved -0.5 T-score. Lårbenshals: Det optimale adskillelsespunkt mellem høj og lav BMD lå ved -1.0 T-score. Alder og serum osteocalcin var associeret med lavere BMD. BMI var associeret med højere BMD.
Underarm: Alkoholindtag, serum osteocalcin, tidligere fraktur og serum BAP var forbundet med lavere BMD. Vitamin D indtag og BMI var positivt associerede med BMD.
Longitudinale studie:
Tabshastighederne for BMD var normalfordelte uden tegn på en specifik gruppe med hurtigt knogletab. Lænderyg: Den optimale adskillelse lå 1 SD under gennemsnittet. Fysisk aktivitet var forbundet med højere tabsrate. Lårbenshals: Det optimale afskæringspunkt lå ved mere end ½ SD under middelværdien. Stigende alder var forbundet med lavere knogletab.
De forudsagte værdier afveg systematisk fra det observerede, når de blev overført på Odense centret.
Konklusion: Det er ikke muligt at forudsige BMD eller BMD tabshastigheder efter overgangsalderen med sikkerhed ved brug af biokemiske eller andre variable.