Skip to contents

Classifying substances

classify.Rmd
if ("pak" %nin% loadedNamespaces()) {
  install.packages("pak", quiet = TRUE)
}
pkgs <- c("dplyr", "stringr", "tidyr", "forcats")
pak::pak(pkgs)

This vignette covers classifying ‘adverse analytic findings’ for a single banned substance. Reproducing the tidy_sports dataset from the “clean-sports” article:

usada_raw <- get_sanctions_data()
usada <- process_text(raw_data = usada_raw)

bad_dates <- subset(usada,
  grepl("^original", usada[['sanction_announced']]))
good_dates <- subset(usada,
  !grepl("^original", usada[['sanction_announced']]) & sanction_announced != "")

cleaned_dates <- clean_dates(
  df = bad_dates,
  date_col = "sanction_announced",
  split = "updated",
  pattern = "original")
names(cleaned_dates)[names(cleaned_dates) == 'split_date'] <- 'sanction_date'
names(cleaned_dates)[names(cleaned_dates) == 'pattern_date'] <- 'original_date'
good_dates$sanction_date <- as.Date(x = good_dates[['sanction_announced']],
                                    format = "%m/%d/%Y")
nms <- intersect(names(cleaned_dates), names(good_dates))
usada_dates <- rbind(good_dates, cleaned_dates[nms])

tidy_sports <- clean_sports(df = usada_dates, sport_col = "sport", tidy = TRUE)
str(tidy_sports)
#> 'data.frame':    703 obs. of  9 variables:
#>  $ athlete           : chr  "miller, adam" "zilcosky, chase" "trabing, bert" "cantwell, steven" ...
#>  $ sport             : chr  "field hockey" "weightlifting" "weightlifting" "paralympic snowboarding" ...
#>  $ substance_reason  : chr  "non-analytical: 3 whereabouts failures" "amphetamine" "anastrozole; testosterone" "dehydrochlormethyltestosterone (dhcmt)" ...
#>  $ sanction_terms    : chr  "2-year suspension; loss of results" "2-year suspension; loss of results" "4-year suspension; loss of results" "6-year suspension; loss of results" ...
#>  $ sanction_announced: chr  "07/31/2026" "07/29/2026" "07/22/2026" "07/16/2026" ...
#>  $ sanction_date     : Date, format: "2026-07-31" "2026-07-29" ...
#>  $ support_personnel : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
#>  $ paralympic        : logi  FALSE FALSE FALSE TRUE FALSE FALSE ...
#>  $ multiple_sports   : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...

AAFs vs ADRVs

The sanctions are divided into two categories:

  • analytic: Adverse Analytical Finding, AAF; An AAF is a report from a WADA-accredited laboratory that identifies the presence of a prohibited substance and/or its metabolites or markers in a sample.

  • non-analytic: Non-Analytical Anti-doping Rule Violation ADRV; a non-analytical anti-doping rule violation does not stem from a positive urine or blood sample, but instead originates from, and is substantiated by, other evidence of doping or violations by an athlete or athlete support personnel..

Substance/reason

The substance_reason column contains the details of the sanction, which can include the following information:

  1. The name of the banned substance

  2. A description of the infraction (if non-analytic)

We will use regular expressions to identify the type of substance behind the sanction. See the examples below:

stringr::str_view(tidy_sports[['substance_reason']], 
  "use \\(epo & hgh\\)", match = TRUE)
#> [684] │ erythropoietin (epo) and non-analytical: <use (epo & hgh)>
stringr::str_view(tidy_sports[['substance_reason']],
  "tampering, complicity", match = TRUE)
#> [195] │ non-analytical: <tampering, complicity>

Most of the non-analytic sanctions include the terms non-analytic/non-analytical/etc., as a prefix in the substance_reason column.

Sanction types

I can pass these terms as regular expressions to create the sanction_type variable, which will contain two values: non-analytic and analytic. I’ll save this variable in a new intermediate substances dataset:

substances <- dplyr::mutate(.data = tidy_sports,
    sanction_type = dplyr::case_when(
      stringr::str_detect(string = substance_reason,
        "non-analytical") ~ "non-analytic",
      !stringr::str_detect(substance_reason,
        "non-analytical") ~ "analytic",
      TRUE ~ NA_character_
    )
  )
substances |>
  dplyr::count(sanction_type, sort = TRUE)
#>   sanction_type   n
#> 1      analytic 548
#> 2  non-analytic 155

Now I can filter substances to the analytical sanctions in sanction_type.

How can I identify the single vs. multiple substances?

Let’s take a look at four different sanctions, pulled from the live data (one for each combination of sanction_type and single vs. multiple substance/reason):

substance_type_examples <- substances |>
  dplyr::mutate(substance_cat = dplyr::if_else(
    stringr::str_detect(substance_reason, "; |, | and | & | / "),
    "multiple", "single")) |>
  dplyr::group_by(sanction_type, substance_cat) |>
  dplyr::slice_head(n = 1) |>
  dplyr::ungroup() |>
  dplyr::select(athlete, substance_reason, sanction_type, substance_cat)
substance_type_examples
#> # A tibble: 4 × 4
#>   athlete               substance_reason             sanction_type substance_cat
#>   <chr>                 <chr>                        <chr>         <chr>        
#> 1 trabing, bert         anastrozole; testosterone    analytic      multiple     
#> 2 zilcosky, chase       amphetamine                  analytic      single       
#> 3 anmaw mengesha, adane non-analytical: use (intrav… non-analytic  multiple     
#> 4 miller, adam          non-analytical: 3 whereabou… non-analytic  single

Substance category

We can see analytic and non-analytic sanctions, with both single and multiple substance/reason. Fortunately, the sanctions with multiple items are separated by either semicolons (;), commas (,), or a conjunction (and), and can be separated by a regular expression:

dplyr::mutate(substance_type_examples,
  substance_cat = dplyr::case_when(
    # identify the multiple_sr substances using a regular expression
    stringr::str_detect(substance_reason, "; |, | and | & | / ") ~ 'multiple',
    # negate the regular expression for the single substances
    !stringr::str_detect(substance_reason, "; |, | and | & | / ") ~ 'single',
    TRUE ~ NA_character_)) |>
  dplyr::count(substance_cat, substance_reason) |>
  tidyr::pivot_wider(names_from = substance_cat, values_from = n)
#> # A tibble: 4 × 3
#>   substance_reason                                      multiple single
#>   <chr>                                                    <int>  <int>
#> 1 anastrozole; testosterone                                    1     NA
#> 2 non-analytical: use (intravenous infusion); tampering        1     NA
#> 3 amphetamine                                                 NA      1
#> 4 non-analytical: 3 whereabouts failures                      NA      1

The substance_cat identifier can be used to separate sanctions with multiple substance/reasons from sanctions with a single substance or reason.

substances <- substances |>
  dplyr::mutate(substance_cat = dplyr::case_when(
    stringr::str_detect(substance_reason, "; |, | and | & | / ") ~ 'multiple',
    !stringr::str_detect(substance_reason, "; |, | and | & | / ") ~ 'single',
    TRUE ~ NA_character_))
substances |> 
  dplyr::count(substance_cat, sort = TRUE)
#>   substance_cat   n
#> 1        single 522
#> 2      multiple 181

Single analytic substances

First create a dataset that contains the sanctions with a single substance listed. Store these in single_analytic_substances.

single_analytic_substances <- substances |>
  dplyr::filter(substance_cat == 'single' & sanction_type == "analytic")

View the top ten single analytic substances:

single_analytic_substances |> 
  dplyr::count(substance_reason, sort = TRUE) |> 
  head(10)
#>                 substance_reason  n
#> 1    androgenic anabolic steroid 46
#> 2                   cannabinoids 45
#> 3                       ostarine 40
#> 4                     clomiphene 22
#> 5                      meldonium 14
#> 6                   testosterone 14
#> 7  dehydroepiandrosterone (dhea) 11
#> 8                     stanozolol 11
#> 9                    clenbuterol 10
#> 10                    furosemide 10

Multiple analytic substances

Next create a dataset with the sanctions listing multiple substances in substance_reason. Store these in multiple_analytic_substances.

multiple_analytic_substances <- substances |>
  dplyr::filter(substance_cat == 'multiple' & sanction_type == "analytic")

View the top ten multiple analytic substances:

multiple_analytic_substances |> 
  dplyr::count(substance_reason, sort = TRUE) |> 
  head(10)
#>                                                                           substance_reason
#> 1                                                                       ostarine; lgd-4033
#> 2                                                            clomiphene and its metabolite
#> 3                                 androgenic anabolic steroid & 19-norandrosterone (19-na)
#> 4                                                       benzoylecgonine and methylecgonine
#> 5                                                               ostarine; lgd-4033; gw1516
#> 6                                                                 stanozolol; drostanolone
#> 7                                                     1-androstenedione and its metabolite
#> 8                                                     1-testosterone and 1-androstenedione
#> 9  19-norandrosterone (19-na), 19-noretiocholanolone, and drostanolone and its metabolites
#> 10         19-norandrosterone (19-na), drostanolone and its metabolites, and epitrenbolone
#>    n
#> 1  4
#> 2  3
#> 3  2
#> 4  2
#> 5  2
#> 6  2
#> 7  1
#> 8  1
#> 9  1
#> 10 1

Tidying substances

Tidying the sanctions with multiple WADA banned substances (i.e., one substance per athlete per row) will result in certain athletes appearing in the dataset more than once. The regular expressions below cover a range of semicolons, tabs, and spaces to identify and separate each substance.

add_match_col()

I’ve written the add_match_col() function, which creates a new 'matched' column with the matched regular expression pattern (it’s like stringr::str_view(), but in a data.frame/tibble). I used add_match_col() while determining the correct pattern to match on (i.e., any substances listing metabolites):

dplyr::mutate(substance_type_examples,
  # add matched column
  punct_match = add_match_col(
    string = substance_reason,
    pattern = "[[:punct:]]")) |>
  dplyr::select(substance_reason, dplyr::last_col())
#> # A tibble: 4 × 2
#>   substance_reason                                      punct_match  
#>   <chr>                                                 <chr>        
#> 1 anastrozole; testosterone                             ;            
#> 2 amphetamine                                           NA           
#> 3 non-analytical: use (intravenous infusion); tampering -, :, (, ), ;
#> 4 non-analytical: 3 whereabouts failures                -, :

The rows above are all matching correctly on the regular expression pattern.

Below are a series of regular expression to 1) match the substances that list metabolites 2) differentiate the multiple substances, and 3) trims the white space from the tidy substance_reason column.

The code below tests this pattern on a sample from multiple_analytic_substances before its passed to tidyr::separate_rows():

dplyr::sample_n(multiple_analytic_substances, size = 10, replace = FALSE) |> 
  dplyr::mutate(
    # replace plurals
      substance_reason = stringr::str_replace_all(substance_reason,
        "and its metabolite|and its metabolites|its metabolite",
        "(metabolite)")) |> 
    tidyr::separate_rows(substance_reason, 
        sep = "; |;\t|\\t|, |;| and |and a |and | & | / ") |> 
    dplyr::mutate(substance_reason = trimws(substance_reason, "both")) |> 
    dplyr::select(athlete, substance_reason)
#> # A tibble: 30 × 2
#>    athlete                substance_reason                   
#>    <chr>                  <chr>                              
#>  1 romero noboa, isidro   androgenic anabolic steroid        
#>  2 romero noboa, isidro   cannabinoids                       
#>  3 romero noboa, isidro   anastrozole                        
#>  4 romero noboa, isidro   clomiphene                         
#>  5 romero noboa, isidro   gw1516                             
#>  6 cardoso, josé henrique 19-norandrosterone (19-na)         
#>  7 cardoso, josé henrique 2a-methyl-5a-androstan-3a-ol-17-one
#>  8 cardoso, josé henrique epitrenbolone                      
#>  9 cardoso, josé henrique methasterone                       
#> 10 cardoso, josé henrique testosterone                       
#> # ℹ 20 more rows

After confirming the pattern is working, the output will be stored in tidy_multiple_substances.

tidy_multiple_substances <- dplyr::mutate(multiple_analytic_substances,
  # replace plurals
    substance_reason = stringr::str_replace_all(substance_reason,
      "and its metabolite|and its metabolites|its metabolite",
      "(metabolite)")) |> 
  tidyr::separate_rows(substance_reason, 
      sep = "; |;\t|\\t|, |;| and |and a |and | & | / ") |> 
  dplyr::mutate(substance_reason = trimws(substance_reason, "both"))

With both single and multiple substances in tidy format, they can be combined together into a single tidy_substances dataset.

tidy_substances <- rbind(single_analytic_substances, tidy_multiple_substances)

The top 10 tidy substances are below:

tidy_substances |> 
  dplyr::count(substance_reason, sort = TRUE) |> 
  head(10)
#>               substance_reason  n
#> 1                     ostarine 67
#> 2  androgenic anabolic steroid 60
#> 3                 cannabinoids 47
#> 4                   clomiphene 36
#> 5                 testosterone 29
#> 6                       gw1516 20
#> 7          hydrochlorothiazide 17
#> 8                   stanozolol 16
#> 9                     lgd-4033 15
#> 10                 clenbuterol 14

classify_wada_substances()

To identify the WADA banned substances, I’ve written classify_wada_substances(), a function that scans the substance_reason column and identifies any substances found on the WADA list. See the classify_wada_substances() documentation for more information.

WADA Classes

classify_wada_substances() creates a substance_group variable with each of the WADA classifications (stored in dopingdata::wada_classes):

dopingdata::wada_classes
#>                           Classification
#> 1                     S1 ANABOLIC AGENTS
#> 2     S2 PEP HORMONES/G FACTORS/MIMETICS
#> 3                     S3 BETA-2 AGONISTS
#> 4    S4 HORMONE AND METABOLIC MODULATORS
#> 5            S5 DIURETICS/MASKING AGENTS
#> 6                          S6 STIMULANTS
#> 7                           S7 NARCOTICS
#> 8                        S8 CANNABINOIDS
#> 9                     S9 GLUCOCORTICOIDS
#> 10              S0 UNAPPROVED SUBSTANCES
#> 11              M1 MANIPULATION OF BLOOD
#> 12 M2 CHEMICAL AND PHYSICAL MANIPULATION
#> 13               M3 GENE AND CELL DOPING
#> 14                      P1 BETA-BLOCKERS

dopingdata stores vectors with each substance group in the WADA list (the S1 ANABOLIC AGENTS substances are below):

head(dopingdata::s1_substances, 10)
#>  [1] "3α-hydroxy-5α-androst-1-en-17-one"              
#>  [2] "androgenic anabolic steroid"                    
#>  [3] "androgenic anabolic steroids"                   
#>  [4] "anabolic agent"                                 
#>  [5] "anabolic agents"                                
#>  [6] "anabolic steroid"                               
#>  [7] "anabolic steroids"                              
#>  [8] "androstenedione"                                
#>  [9] "metabolites of androstenedione"                 
#> [10] "1-androstenediol (5α-androst-1-ene-3β,17β-diol)"

The substance group vectors are passed to make_regex() to create a regular expressions (s1_regex), which we can use to match the substance_reason column on (see the example using the live tidy_substances dataset built above):

s1_regex <- make_regex(x = dopingdata::s1_substances, wb = TRUE)
stringr::str_view(string = tidy_substances$substance_reason,
  pattern = s1_regex, match = TRUE)
#>  [2] │ dehydrochlormethyltestosterone (<dhcmt>)
#>  [6] │ <testosterone>
#>  [7] │ <dehydroepiandrosterone> (<dhea>)
#>  [9] │ <dehydroepiandrosterone> (<dhea>)
#> [11] │ <androgenic anabolic steroid>
#> [20] │ <androgenic anabolic steroid>
#> [23] │ <dehydroepiandrosterone> (<dhea>)
#> [24] │ <testosterone>
#> [25] │ <testosterone>
#> [26] │ <testosterone>
#> [28] │ <dehydroepiandrosterone> (<dhea>)
#> [31] │ <ostarine>
#> [33] │ <drostanolone>
#> [34] │ <androgenic anabolic steroid>
#> [38] │ <ostarine>
#> [40] │ <metenolone>
#> [42] │ <androgenic anabolic steroid>
#> [43] │ <oxandrolone>
#> [46] │ <ostarine>
#> [47] │ <19-norandrosterone>
#> ... and 347 more

The output from classify_wada_substances() can be used to answer questions like: what substance_group’s appear the most?

tidy_substances <- classify_wada_substances(
  usada_data = tidy_substances,
  subs_column = "substance_reason")

UNCLASSIFIED single substances

The following single substances are marked as UNCLASSIFIED:

tidy_substances |>
  dplyr::filter(
      substance_cat == "single" & 
      substance_group == "UNCLASSIFIED" & 
      substance_reason != "") |>
  dplyr::distinct(athlete, substance_reason)
#>             athlete                         substance_reason
#> 1     *name removed                              dorzolamide
#> 2 brinegar, michael erythropoietic stimulating agents (esas)
#> 3   mcmahon, kensey                               vadadustat
#> 4     *name removed                             capromorelin
#> 5    sundeen, tyler                           ritalinic acid
#> 6   rodriguez, yair                   3 whereabouts failures

The final unclassified substance is actually a result from a miss-classified sanction type (for rodriguez, yair).

tidy_substances |>
  dplyr::filter(athlete == "rodriguez, yair") |>
  dplyr::select(athlete, substance_reason, substance_group, sanction_type)
#>           athlete       substance_reason substance_group sanction_type
#> 1 rodriguez, yair 3 whereabouts failures    UNCLASSIFIED      analytic

For this particular athlete, 1) the sanction_type should be non-analytic, and 2) the substance_group should be missing (NA_character_)

tidy_substances <- tidy_substances |>
  dplyr::mutate(sanction_type = dplyr::case_when(
    athlete == "rodriguez, yair" ~ "non-analytic",
    TRUE ~ sanction_type
  )) |> 
  dplyr::mutate(substance_group = dplyr::case_when(
    athlete == "rodriguez, yair" ~ NA_character_,
    TRUE ~ substance_group
  )) 
tidy_substances |>
  dplyr::filter(athlete == "rodriguez, yair") |>
  dplyr::select(athlete, substance_reason, substance_group, sanction_type)
#>           athlete       substance_reason substance_group sanction_type
#> 1 rodriguez, yair 3 whereabouts failures            <NA>  non-analytic

UNCLASSIFIED multiple substances 

tidy_substances |>
  dplyr::filter(
      substance_cat == "multiple" & 
      substance_group == "UNCLASSIFIED" & 
      substance_reason != "") |>
  dplyr::distinct(substance_reason)
#>                                                       substance_reason
#> 1                                                                 hctz
#> 2                                                              rad-140
#> 3                                  2a-methyl-5a-androstan-3a-ol-17-one
#> 4                                                    d-methamphetamine
#> 5                                                           arimistane
#> 6                                                           torasemide
#> 7                                                           possession
#> 8                                                    use/attempted use
#> 9                                            evading sample collection
#> 10                                                               igf-1
#> 11                                  human chorionic gonadotropin (hcg)
#> 12                                                            aod-9064
#> 13                                                                s-23
#> 14                           intact human chorionic gonadtrophin (hcg)
#> 15 thiazide metabolite 4-amino-6-chloro-1,3-benzenedisulfonamide (acb)
#> 16                                                      methylecgonine
#> 17                                                     propylhexadrine
#> 18                                                 androstatrienedione
#> 19                                             androst-(2,3)-en-17-one
#> 20                                      non-anatlyical: administration
#> 21                                                         trafficking
#> 22                                 human chorionic gonadotrophin (hcg)
#> 23                                violation of period of ineligibility

Re-classifying substances

Any substances that are not classified from the existing WADA list can be added with reclass_substance() (these substances can also added to their relative vector and regular expression in data-raw/)

reclass_substance()

reclass_substance() takes a df, substance, and value:

# 2a-methyl-5a-androstan-3a-ol-17-one ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=arimistane&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "^2a-methyl-5a-androstan-3a-ol-17-one$",
  value = "S1 ANABOLIC AGENTS")
# d-methamphetamine ----
# https://www.usada.org/sanction/hillary-tran-accepts-doping-sanction/
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "d-methamphetamine",
  value = "S6 STIMULANTS")
# arimistane ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=arimistane&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "arimistane",
  value = "S4 HORMONE AND METABOLIC MODULATORS")
# torasemide ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=torasemide&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "torasemide",
  value = "S5 DIURETICS/MASKING AGENTS")
# igf-1 ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=igf-1&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "igf-1",
  value = "S2 PEP HORMONES/G FACTORS/MIMETICS")
# human chorionic gonadotropin (hcg)  ----
# https://www.usada.org/athletes/antidoping101/athlete-guide-anti-doping/
# https://www.usada.org/spirit-of-sport/education/wellness-and-anti-aging-clinics/
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "^human chorionic gonadotropin \\(hcg\\)$",
  value = "S2 PEP HORMONES/G FACTORS/MIMETICS")
# intact human chorionic gonadtrophin (hcg)  ----
# https://www.usada.org/athletes/antidoping101/athlete-guide-anti-doping/
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "^intact human chorionic gonadtrophin \\(hcg\\)$",
  value = "S2 PEP HORMONES/G FACTORS/MIMETICS")
# human chorionic gonadotrophin (hcg)  ----
# https://www.usada.org/athletes/antidoping101/athlete-guide-anti-doping/
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "human chorionic gonadotrophin \\(hcg\\)",
  value = "S2 PEP HORMONES/G FACTORS/MIMETICS")
# aod-9064  ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=aod-9064&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "aod-9064",
  value = "S2 PEP HORMONES/G FACTORS/MIMETICS")
# s-23  ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=s-23&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "s-23",
  value = "S1 ANABOLIC AGENTS")
# methenolone  ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=methenolone&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "methenolone",
  value = "S1 ANABOLIC AGENTS")
# thiazide metabolite 4-amino-6-chloro-1,3-benzenedisulfonamide (acb)  ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=thiazide&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "^thiazide metabolite 4-amino-6-chloro-1,3-benzenedisulfonamide \\(acb\\)$",
  value = "S5 DIURETICS/MASKING AGENTS")
# methylecgonine  ----
# https://www.usada.org/sanction/mike-alexandrov-accepts-doping-sanction/
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "methylecgonine",
  value = "S6 STIMULANTS")
# propylhexadrine  ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=propylhexadrine&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "propylhexadrine",
  value = "S6 STIMULANTS")
# androstatrienedione  ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=androstatrienedione&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "androstatrienedione",
  value = "S4 HORMONE AND METABOLIC MODULATORS")
# androst-(2,3)-en-17-one  ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=androst&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "androst-\\(2,3\\)-en-17-one",
  value = "S1 ANABOLIC AGENTS")
# methylclostebol ----
# https://www.wada-ama.org/en/prohibited-list?page=0&q=methylclostebol&all=1#search-anchor
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "methylclostebol",
  value = "S1 ANABOLIC AGENTS")
# promagnon  ----
# https://www.usada.org/sanction/u-s-judo-athlete-ohara-accepts-sanction-rule-violation/
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "promagnon",
  value = "S1 ANABOLIC AGENTS")
# 4-hydroxytriamterene  ----
# 4-Hydroxy Triamterene is a diuretic agent and metabolite of Triamterene
# https://www.scbt.com/p/4-hydroxy-triamterene-1226-52-4
tidy_substances <- reclass_substance(
  df = tidy_substances,
  substance = "4-hydroxytriamterene",
  value = "S5 DIURETICS/MASKING AGENTS")

After reclassifying the substances above, remaining UNCLASSIFIED/multiple sanctions are all non-analytic:

tidy_substances |>
  dplyr::filter(
      substance_cat == "multiple" & 
      substance_group == "UNCLASSIFIED" & 
      substance_reason != "") |>
  dplyr::distinct(substance_reason) 
#>                       substance_reason
#> 1                                 hctz
#> 2                              rad-140
#> 3                           possession
#> 4                    use/attempted use
#> 5            evading sample collection
#> 6       non-anatlyical: administration
#> 7                          trafficking
#> 8 violation of period of ineligibility

Changing the sanction_type classification to non-analytic requires missing (NA_character_) values for substance_group:

# Change sanction_type to non-analytic
tidy_substances <- tidy_substances |> 
  dplyr::mutate(
    sanction_type = dplyr::case_when(
      substance_group == "UNCLASSIFIED" & substance_reason == "possession" ~ "non-analytic",
      substance_group == "UNCLASSIFIED" & substance_reason == "use/attempted use" ~ "non-analytic",
      substance_group == "UNCLASSIFIED" & substance_reason == "evading sample collection" ~ "non-analytic",
      substance_group == "UNCLASSIFIED" & substance_reason == "non-anatlyical: administration" ~ "non-analytic",
      substance_group == "UNCLASSIFIED" & substance_reason == "trafficking" ~ "non-analytic",
      TRUE ~ sanction_type
    )
  ) 
# Change substance_group to NA_character_
tidy_substances <- tidy_substances |> 
  dplyr::mutate(
    substance_group = dplyr::case_when(
      sanction_type == "non-analytic" ~ NA_character_,
      TRUE ~ substance_group)
    )

Remove the empty substance_reason values:

tidy_substances <- dplyr::filter(tidy_substances, substance_reason != "")

Tidy substances

Now all the substances have been properly classified as Adverse Analytical Findings.

tidy_substances |> 
  dplyr::count(sanction_type, substance_group) |> 
  tidyr::pivot_wider(names_from = sanction_type, values_from = n)
#> # A tibble: 15 × 3
#>    substance_group                       analytic `non-analytic`
#>    <chr>                                    <int>          <int>
#>  1 M1 MANIPULATION OF BLOOD                     3             NA
#>  2 M2 CHEMICAL AND PHYSICAL MANIPULATION        1             NA
#>  3 P1 BETA-BLOCKERS                             2             NA
#>  4 S0 UNAPPROVED SUBSTANCES                     2             NA
#>  5 S1 ANABOLIC AGENTS                         370             NA
#>  6 S2 PEP HORMONES/G FACTORS/MIMETICS          46             NA
#>  7 S3 BETA-2 AGONISTS                          12             NA
#>  8 S4 HORMONE AND METABOLIC MODULATORS        101             NA
#>  9 S5 DIURETICS/MASKING AGENTS                 70             NA
#> 10 S6 STIMULANTS                               77             NA
#> 11 S7 NARCOTICS                                 2             NA
#> 12 S8 CANNABINOIDS                             47             NA
#> 13 S9 GLUCOCORTICOIDS                           6             NA
#> 14 UNCLASSIFIED                                10             NA
#> 15 NA                                          NA              6

And the non-analytic sanctions are truly Non-Analytical Anti-doping Rule Violations.

dplyr::filter(tidy_substances, sanction_type == "non-analytic") |> 
  dplyr::count(substance_reason, substance_cat) |> 
  tidyr::pivot_wider(names_from = substance_cat, values_from = n)
#> # A tibble: 6 × 3
#>   substance_reason               single multiple
#>   <chr>                           <int>    <int>
#> 1 3 whereabouts failures              1       NA
#> 2 evading sample collection          NA        1
#> 3 non-anatlyical: administration     NA        1
#> 4 possession                         NA        1
#> 5 trafficking                        NA        1
#> 6 use/attempted use                  NA        1