icd (version 2.2)

icd_is_valid: Check whether ICD-9 codes are syntactically valid

Description

This does not check whether the code corresponds to a real ICD-9-CM billing code, or parent grouping. For that, see icd_is_defined.

Usage

icd_is_valid(x, ...)
# S3 method for icd10
icd_is_valid(x, short_code = icd_guess_short(x),
  whitespace_ok = TRUE, ...)
# S3 method for icd9
icd_is_valid(x, short_code = icd_guess_short(x),
  whitespace_ok = TRUE, ...)
# S3 method for icd_comorbidity_map
icd_is_valid(x, short_code, ...)

Arguments

x

An ICD-9 or 10 code. If the class is set to 'icd9', 'icd10', 'icd10cm' etc then perform appropriate validation.

...

arguments passed on to other functions

whitespace_ok

Single logical, if TRUE, the default, matches for ICD codes will accept leading and trailing white space.

major

character vector of 'major' part of ICD-9 codes, i.e. that part which falls before the decimal point, in decimal notation. (In five digit notation, the 'major' part is be the first three characters (with leading zeroes), and includes V or E prefix.

Value

logical vector with TRUE or FALSE for each ICD code provided according to its validity

Methods (by class)

• icd10: Test whether generic ICD-10 code is valid

• icd9: Test whether generic ICD-10 code is valid

• icd_comorbidity_map: Validate a mapping of ICD codes to comorbidities.

Three-digit validation

icd9_is_valid_major validates just the 'major' three-digit part of an ICD-9 code. This can in fact be provided as a numeric, since there is no ambiguity. Numeric-only codes should be one to three digits, V codes are followed by one or two digits, and E codes always by three digits between 800 and 999.

Class

S3 class of on object in R is just a vector. Attributes are lost with manipulation, with the exception of class: therefore, elements of the class vector are used to describe features of the data. If these are not present, the user may specify (e.g. decimal vs short_code type, ICD-9 vs ICD-10 WHO), but if they are, the correct functions are called without any guess work. There is overlap between sets with combinations of short_code or decimal_code, and ICD-9 or ICD-10 codes, so guessing is never going to be perfect.

Details

Factors are accepted, and since the validation is done with grepl these are handled correctly.

Currently, there is a limitation on NA values. Calling with NA (which is a logical vector of length one by default) fails, because it is not a string. This is rarely of significance in real life, since the NA will be part of a character vector of codes, and will therefore be cast already to NA_character

NA values result in a return value of FALSE.

Leading zeroes in the decimal form are not ambiguous. Although integer ICD-9 codes could be intended by the user, there is a difference between 100, 100.0, 100.00. Therefore a warning is given if a numeric value is provided.

See Also

http://www.stata.com/users/wgould/icd9/icd9.hlp http://www.sascommunity.org/wiki/Validate_the_format_of_ICD-9_codes

Examples

# NOT RUN {
  icd_is_valid(as.icd9(c("", "1", "22", "333", "4444", "123.45", "V",
                     "V2", "V34", "V567", "E", "E1", "E70", "E")))
  # internal function:
  
# }
# NOT RUN {
  icd:::icd_is_valid_major(c("", "1", "22", "333", "4444", "123.45", "V",
                     "V2", "V34", "V567", "E", "E1", "E70", "E"))
  
# }