Dear David,

I can give you a long list of references. We have a large dataset with data regarding children. Many parameters change appreciably over the course of childhood and adulthood, so you need quite complex methods. It seems to me that if you model age-related reference intervals in adulthood, simpler models may suffice.

Here is the simpler solution: 

Altman DG. Construction of age-related reference centiles using absolute residuals. Statistics in Medicine 12, 917-924 (1993)

And the more complex ones:

Cole TJ, Green P (1992) Smoothing reference centile curves: The LMS method and penalized likelihood. Statistics in Medicine 11: 1305-1319.

Royston P, Wright EM (1998) A method for estimating age-specific reference intervals ('normal ranges') based on fractional polynomials and exponential transformation. Journal of the Royal Statistical Society, Series A  161: 79-101.

Rigby RA, Stasinopoulos DM (2005) Generalized additive models for location, scale and shape. Applied Statistics 54: 507-554.

Borghi E, de Onis M, Garza C et al. for the WHO Multicentre Growth Reference Study Group (in press) Construction of the World Health Organization child growth standards: selection of methods for attained growth curves. Statistics in Medicine.

Something on goodness of fit:

Royston P, Wright EM (2000) Goodness-of-fit statistics for age-specific reference intervals. Statistics in Medicine 19: 2943-2962.

van Buuren S, Fredriks M (2001) Worm plot: a simple diagnostic device for modelling growth reference curves. Statistics in Medicine 20: 1259-1277.

Pan H, Cole TJ (2004): A comparison of goodness of fit tests for age-related reference ranges. Statistics in Medicine 23: 1749-1765.

Greetings,
Angelika
----------------------------------------------------
Angelika Schaffrath Rosario, Stat.
Robert Koch-Institut
Abt. Epidemiologie und Gesundheitsberichterstattung
Seestr. 10, D-13353 Berlin
Tel. ++49 (0)1888 754-3656   Fax -3333
e-Mail:  [log in to unmask]


-----Ursprüngliche Nachricht-----
Von: A UK-based worldwide e-mail broadcast system mailing list [mailto:[log in to unmask]] Im Auftrag von David Crabb
Gesendet: Montag, 19. Dezember 2005 13:22
An: [log in to unmask]
Betreff: Query: age related reference intervals


We are trying to establish an age-related Normal Reference range for a 
parameter from medical imaging. This measurement varies with age and 
another parameter, lets call it V1. If we consider v1 then the 
relationship seems linear for a part of the age range (after taking logs!) 
but the variability increases in the older groups. I think the term is 
heteroscedascity. We have fitted a loess function to this and it mimics 
the apparent relationship well. We would like to establish reference 
intervals (possibly with anon-parametric approach) and the data set is 
probably large enough – Any ideas when using locally weighted regression? 
I would be grateful for any advice, references, papers, software or 
techniques that the list is aware of? Many thanks.