Reference Interchange Format (RIF)

Stef van Buuren

Background

The Reference Interchange Format (RIF) is a simple text format for storing and exchanging growth references. Growth references typically consist of a conditioning variable, usually age or height, and a set of parameter estimates at each tabulated age or height. For example, the first four rows of the WHO Growth Standard for lengths of boys are

library(centile)
ref <- load_reference("who_2006_hgt_male_")
head(ref, 4)
#> # A tibble: 4 × 4
#>        x     L     M      S
#>    <dbl> <dbl> <dbl>  <dbl>
#> 1 0          1  49.9 0.0380
#> 2 0.0027     1  50.1 0.0378
#> 3 0.0055     1  50.2 0.0378
#> 4 0.0082     1  50.4 0.0376

In this example, x stands for age in years and the L, M and S columns stand for the estimated parameters of the LMS model (Cole & Green 1992). The study attributes stores some additional information:

head(attr(ref, "study"))
#>         name         year        yname          sex distribution     citation 
#>        "who"       "2006"        "hgt"       "male"        "LMS"   "WHO 2006"

The RIF format is a simple way to store references as text files that can serve as input for different systems.

General specification

Let us first take a look at an example. The following block contains the first 12 rows of the file who_2006_bmi_female_.txt.

name = who
year = 2006
yname = bmi
sex = female
distribution = LMS
citation = WHO 2006
publication = WHO Multicentre Growth Reference Study Group (2006). WHO Child Growth Standards: Length/height-for-age, weight-for-age, weight-for-length, weight-for-height and body mass index-for-age: Methods and development. Geneva: World Health Organization; pp 312. (web site: http://www.who.int/childgrowth/publications/en/)
[data]
x   L   M   S
0   -0.0631 13.3363 0.09272
0.0027  0.0362  13.3185 0.0936
0.0055  0.1355  13.3006 0.09448
...

A RIF formatted files should follow the following principles:

• A RIF file contains the parameters of one reference distribution only. Thus, separate distributions for males and females go into different files;
• The file starts with a header that describe the contents of the reference table. The section below provides the accepted keywords;
• The = sign separates the keyword and its value;
• A special keyword [DATA] separates the header and data;
• The first line after [DATA] contain the variable names. These names adhere to certain naming conventions given below;
• There is one column named x, the conditioning variable;
• Data columns are separated by tabs.

Keywords

Keyword	Required	Interpretation
name	Yes	two-letter country code, who, eu27, fenton, and so on
year	Yes	four number, may be left blank
yname	Yes	Outcome: hgt, wgt, hdc, wfh, bmi, dsc
sex	No	male or female, may be left blank
sub	No	Subgroup
distribution	Yes	NO, LMS, BCCG, BCPE or BCT
citation	No	Short citation phrase
publication	No	Reference to source documents
remark	No	Free text for additional remarks
tx	No	Transformation of x prior to Z-score calculation
xt	No	Back-transform to x after Z-score calculation
ty	No	Transformation of y prior to Z-score calculation
yt	No	Back-transform to y after Z-score calculation

Any other keywords, each on its own line, can be added and will be saved in attr(ref, "study"). The number of header rows is limited to 24.

Column names

Apart from the column named x two or more column with parameters, depending on the specification of the distribution.

Distribution	Required	Interpretation
NO	mean, sd	Mean, Standard deviation
LMS	M, S, L	Location, Scale, Skewness
BCCG	mu, sigma, nu	Location, Scale, Skewness
BCPE	mu, sigma, nu, tau	Location, Scale, Skewness, Kurtosis
BCT	mu, sigma, nu, tau	Location, Scale, Skewness, Kurtosis

Some other distributions can be specified, but no Z-scores will be calculated:

Distribution	Required	Interpretation
MEA	mean	Mean only
PCT	pxx, pyy	One or more percentiles (xx, yy numeric)
MP	mean, pxx, pyy	Mean + Percentiles (xx, yy numeric)