Classify stunting in data.frame-like objects with GIGS-recommended growth standards

Classify stunting in data.frame-like objects with GIGS-recommended growth standards

Usage

classify_stunting(
  .data,
  lenht_cm,
  age_days,
  gest_days,
  sex,
  id = NULL,
  .new = c("lhaz", "stunting", "stunting_outliers")
)

Arguments

.data

A data.frame-like tabular object with one or more rows. Must be permissible by checkmate::assert_data_frame(), so you can also supply a tibble, data.table, or similar.

lenht_cm

<data-masking> The name of a column in .data which is a numeric vector of length/height values in cm.

age_days

<data-masking> The name of a column in .data which is a numeric vector of age values in days.

gest_days

<data-masking> The name of a column in .data which is a numeric vector of gestational age(s) at birth in days. This column, in conjunction with the column referred to by age_days, is used to select which growth standard to use for each observation.

sex

<data-masking> The name of a column in .data which is a case-sensitive character vector of sexes, either "M" (male) or "F" (female). By default, gigs will warn you if any elements of sex are not "M" or "F", or are missing (NA). You can customise this behaviour using the GIGS package-level options.

id

<data-masking> The name of a column in .data which is a factor variable with IDs for each observation. When not NULL, this variable is used to ensure that only the first measurement taken from each infant is used as a birth measure. If all your data is from one individual, leave this parameter as NULL. Default = NULL.

.new

A three-length character vector with names for the output columns. These inputs will be repaired if necessary using vctrs::vec_as_names(), which will print any changes to the console. If any elements in .new are the same as elements in colnames(.data), the function will throw an error. Default = c("lhaz", "stunting", "stunting_outliers").

Value

A tabular object of the same class that was provided as .data, with three new columns named according to .new. These columns will be (from left to right):

• lhaz - Numeric vector of length/height-for-age zscores

• stunting - Factor of stunting categories without outlier flagging

• stunting_outliers - Factor of stunting categories with outlier flagging

Details

Cut-offs for stunting categories are:

Category	Factor level	Z-score bounds
Severe stunting	"stunting_severe"	lhaz =< -3
Stunting	"stunting"	-3 < lhaz =< -2
No stunting	"not_stunting"	lhaz > -2
Outlier	"outlier"	abs(lhaz) > 6

Note

Categorical (factor) columns produced here may contain unused factor levels. By default, gigs will inform you if these columns have unused factor levels. You can change this behaviour using the GIGS package-level option .gigs_options$handle_unused_levels.

References

'Implausible z-score values' in World Health Organization (ed.) Recommendations for data collection, analysis and reporting on anthropometric indicators in children under 5 years old. Geneva: World Health Organization and the United Nations Children's Fund UNICEF, (2019). pp. 64-65.

'Percentage of children stunted, wasted, and underweight, and mean z-scores for stunting, wasting and underweight' in Guide to DHS Statistics DHS-7 Rockville, Maryland, USA: ICF (2020). pp. 431-435. https://dhsprogram.com/data/Guide-to-DHS-Statistics/Nutritional_Status.htm

See also

See classify_growth() to run this analysis and others at the same time.

Examples

# This dummy dataset contains data from two people, from birth (<3 days) to
# 500 days of age.
data <- data.frame(
  child = factor(rep.int(c("A", "B"), c(3, 3))),
  agedays = c(0, 100, 500, 2, 100, 500),
  gestage  = c(rep(35 * 7, 3), rep(40 * 7, 3)),
  sex = rep.int(c("M", "F"), c(3, 3)),
  length_cm = rep.int(c(52.2, 60.4, 75), 2)
)

# Use the `id` argument to ensure that `classify_stunting` uses the correct
# standard for each observation
data |>
  classify_stunting(lenht_cm = length_cm,
                    age_days = agedays,
                    gest_days = gestage,
                    sex = sex,
                    id = child)
#> Warning: There was 1 'at birth' observation where `age_days` > 0.5.
#> ℹ This occurred for ID B.
#> ! Unused factor levels kept after stunting categorisation: "stunting_severe".
#> ! Unused factor levels kept after stunting categorisation: "stunting_severe"
#> and "outlier".
#>   child agedays gestage sex length_cm        lhaz     stunting
#> 1     A       0     245   M      52.2  2.82368563 not_stunting
#> 2     A     100     245   M      60.4  0.88764438 not_stunting
#> 3     A     500     245   M      75.0 -2.16900825     stunting
#> 4     B       2     280   F      52.2  1.86816342 not_stunting
#> 5     B     100     280   F      60.4 -0.04512913 not_stunting
#> 6     B     500     280   F      75.0 -1.44230628 not_stunting
#>   stunting_outliers
#> 1      not_stunting
#> 2      not_stunting
#> 3          stunting
#> 4      not_stunting
#> 5      not_stunting
#> 6      not_stunting

# If you don't specify `id`, `classify_stunting` will assume data is from one
# child only
data |>
  classify_stunting(lenht_cm = length_cm,
                    age_days = agedays,
                    gest_days = gestage,
                    sex = sex)
#> ! Unused factor levels kept after stunting categorisation: "stunting_severe".
#> ! Unused factor levels kept after stunting categorisation: "stunting_severe"
#> and "outlier".
#>   child agedays gestage sex length_cm        lhaz     stunting
#> 1     A       0     245   M      52.2  2.82368563 not_stunting
#> 2     A     100     245   M      60.4  0.88764438 not_stunting
#> 3     A     500     245   M      75.0 -2.16900825     stunting
#> 4     B       2     280   F      52.2  1.45276970 not_stunting
#> 5     B     100     280   F      60.4 -0.04512913 not_stunting
#> 6     B     500     280   F      75.0 -1.44230628 not_stunting
#>   stunting_outliers
#> 1      not_stunting
#> 2      not_stunting
#> 3          stunting
#> 4      not_stunting
#> 5      not_stunting
#> 6      not_stunting