Introduction to epocakir

Introduction

The epocakir package makes clinical coding of patients with kidney disease using clinical practice guidelines easy. The guidelines used are the evidence-based KDIGO guidelines. This package covers acute kidney injury (AKI), anemia, and chronic liver disease(CKD).

Features

aki_staging(): Classification of AKI staging (aki_stages) with automatic selection of:
- aki_bCr(): AKI based on baseline creatinine
- aki_SCr(): AKI based on changes in serum creatinine
- aki_UO(): AKI based on urine output
anemia(): Classification of anemia
Classification of albuminuria (Albuminuria_stages)
- Albuminuria_staging_ACR(): Albuminuria based on Albumin excretion rate
- Albuminuria_staging_AER(): Albuminuria based on Albumin-to-creatinine ratio
eGFR(): Estimation of glomerular filtration rate with automatic selection of:
- eGFR_adult_SCr(): eGFR based on the 2009 CKD-EPI creatinine equation
- eGFR_adult_SCysC(): eGFR based on the 2012 CKD-EPI cystatin C equation
- eGFR_adult_SCr_SCysC(): eGFR based on the 2012 CKD-EPI creatinine-cystatin C equation
- eGFR_child_SCr(): eGFR based on the pediatric creatinine-based equation
- eGFR_child_SCr_BUN(): eGFR based on the pediatric creatinine-BUN equation
- eGFR_child_SCysC(): eGFR based on the pediatric cystatin C-based equation
GFR_staging(): Staging of GFR (GFR_stages)
Multiple utility functions including:
- conversion_factors: Conversion factors used throughout the KDIGO guidelines
- as_metric(): Conversion of a measured value into metric units
- dob2age(): Calculation of age from a date of birth
- binary2factor(): Conversion of binary data into factors based on a column name
- combine_date_time_cols(): Combining separate date and time columns into a single date and time column
- combn_changes: Generating changes between measurements
Automatic conversion of units class objects
Tidy output allowing seamless integration with functions from the tidyverse
Tidyeval via programming with dplyr
Comprehensive tests and coverage

Examples

library(epocakir)
library(dplyr)
library(units)

Clinical Data

Often clinical data must be cleansed and tidied before analysis can begin. To assist in this, several utility functions have been included. To explore these, consider a sample clinical dataset clinical_obvs:

# Example workflow: clinical_obvs <- read.csv("cohort.csv")
glimpse(clinical_obvs)
#> Rows: 3
#> Columns: 9
#> $ `Patient Number` <chr> "p10001", "p10002", "p10003"
#> $ `Admission Date` <chr> "2020-03-05", "2020-03-06", "2020-03-17"
#> $ `Admission Time` <chr> "14:01:00", "09:10:00", "12:48:00"
#> $ Discharge_date   <chr> "2020-03-10", "2020-03-16", "2020-03-18"
#> $ Discharge_time   <chr> "16:34:00", "18:51:00", "09:12:00"
#> $ `Date of Birth`  <chr> "1956-01-09", "1997-12-04", "1973-05-28"
#> $ Male             <lgl> TRUE, FALSE, TRUE
#> $ Height           <dbl> 182, 161, 168
#> $ Surgery          <lgl> FALSE, FALSE, TRUE

tidy_obvs <- clinical_obvs %>%
  combine_date_time_cols() %>%
  mutate(
    Age = dob2age(`Date of Birth`),
    Height = as_metric(height = set_units(as.numeric(Height), "cm"))
  ) %>%
  binary2factor(Male, Surgery)

glimpse(tidy_obvs)
#> Rows: 3
#> Columns: 8
#> $ `Patient Number`     <chr> "p10001", "p10002", "p10003"
#> $ `Admission DateTime` <dttm> 2020-03-05 14:01:00, 2020-03-06 09:10:00, 2020-03…
#> $ Discharge_DateTime   <dttm> 2020-03-10 16:34:00, 2020-03-16 18:51:00, 2020-0…
#> $ `Date of Birth`      <chr> "1956-01-09", "1997-12-04", "1973-05-28"
#> $ Male                 <ord> Male, Not_Male, Male
#> $ Height               [m] 1.82 [m], 1.61 [m], 1.68 [m]
#> $ Surgery              <ord> Not_Surgery, Not_Surgery, Surgery
#> $ Age                  <Duration> 2172960000s (~68.86 years), 850608000s (~26.95 y…

Make sure to use set_units() from the units package to convert all measurements into unit objects for automatic unit conversion in epocakir.

AKI Staging

Next consider the sample aki_pt_data dataset. It is possible to use aki_staging() to automatically classify the presence and staging of AKI. If a particular method is required, it is possible to classify AKI using aki_bCr(), aki_SCr() or aki_UO().

# Example workflow: aki_pt_data <- read.csv("aki.csv")
head(aki_pt_data)
#> # A tibble: 6 × 7
#>      SCr_    bCr_ pt_id_ dttm_      UO_ aki_staging_type aki_       
#>   [mg/dl] [mg/dl] <chr>  <dttm> [ml/kg] <chr>            <ord>      
#> 1     2       1.5 <NA>   NA          NA aki_bCr          No AKI     
#> 2     2.5     1.5 <NA>   NA          NA aki_bCr          AKI Stage 1
#> 3     3       1.5 <NA>   NA          NA aki_bCr          AKI Stage 2
#> 4     3.5     1.5 <NA>   NA          NA aki_bCr          AKI Stage 2
#> 5     4       1.5 <NA>   NA          NA aki_bCr          AKI Stage 3
#> 6     4.5     1.5 <NA>   NA          NA aki_bCr          AKI Stage 3

aki_staging(aki_pt_data,
  SCr = "SCr_", bCr = "bCr_", UO = "UO_",
  dttm = "dttm_", pt_id = "pt_id_"
)
#>  [1] No AKI      AKI Stage 1 AKI Stage 2 AKI Stage 2 AKI Stage 3 AKI Stage 3
#>  [7] No AKI      No AKI      AKI Stage 1 No AKI      No AKI      AKI Stage 1
#> [13] No AKI      No AKI      No AKI      AKI Stage 1 No AKI      AKI Stage 2
#> [19] AKI Stage 3 AKI Stage 1 AKI Stage 3 AKI Stage 2 No AKI      AKI Stage 1
#> [25] AKI Stage 3 AKI Stage 3 No AKI     
#> Levels: No AKI < AKI Stage 1 < AKI Stage 2 < AKI Stage 3

aki_pt_data %>%
  mutate(aki = aki_staging(
    SCr = SCr_, bCr = bCr_, UO = UO_,
    dttm = dttm_, pt_id = pt_id_
  )) %>%
  select(pt_id_, SCr_:dttm_, aki)
#> # A tibble: 27 × 5
#>    pt_id_    SCr_    bCr_ dttm_               aki        
#>    <chr>  [mg/dl] [mg/dl] <dttm>              <ord>      
#>  1 <NA>       2       1.5 NA                  No AKI     
#>  2 <NA>       2.5     1.5 NA                  AKI Stage 1
#>  3 <NA>       3       1.5 NA                  AKI Stage 2
#>  4 <NA>       3.5     1.5 NA                  AKI Stage 2
#>  5 <NA>       4       1.5 NA                  AKI Stage 3
#>  6 <NA>       4.5     1.5 NA                  AKI Stage 3
#>  7 pt1        3.4    NA   2020-10-23 09:00:00 No AKI     
#>  8 pt1        3.9    NA   2020-10-25 21:00:00 No AKI     
#>  9 pt1        3      NA   2020-10-20 09:00:00 AKI Stage 1
#> 10 pt2        3.4    NA   2020-10-18 22:00:00 No AKI     
#> # ℹ 17 more rows

aki_pt_data %>%
  mutate(aki = aki_SCr(
    SCr = SCr_, dttm = dttm_, pt_id = pt_id_
  )) %>%
  select(pt_id_, SCr_:dttm_, aki)
#> # A tibble: 27 × 5
#>    pt_id_    SCr_    bCr_ dttm_               aki        
#>    <chr>  [mg/dl] [mg/dl] <dttm>              <ord>      
#>  1 <NA>       2       1.5 NA                  No AKI     
#>  2 <NA>       2.5     1.5 NA                  No AKI     
#>  3 <NA>       3       1.5 NA                  No AKI     
#>  4 <NA>       3.5     1.5 NA                  No AKI     
#>  5 <NA>       4       1.5 NA                  No AKI     
#>  6 <NA>       4.5     1.5 NA                  No AKI     
#>  7 pt1        3.4    NA   2020-10-23 09:00:00 No AKI     
#>  8 pt1        3.9    NA   2020-10-25 21:00:00 No AKI     
#>  9 pt1        3      NA   2020-10-20 09:00:00 AKI Stage 1
#> 10 pt2        3.4    NA   2020-10-18 22:00:00 No AKI     
#> # ℹ 17 more rows

Estimated Glomerular Filtration Rate

Similarly, eGFR() offers the ability to automatically select the appropriate formula to estimate the glomerular filtration rate. If a particular formula is required, then eGFR_adult_SCr, eGFR_adult_SCysC, eGFR_adult_SCr_SCysC, eGFR_child_SCr, eGFR_child_SCr_BUN, or eGFR_child_SCysC can be used.

# Example workflow: aki_pt_data <- read.csv("aki.csv")
head(eGFR_pt_data)
#> # A tibble: 6 × 10
#>    SCr_ SCysC_  Age_ male_ black_ height_  BUN_ eGFR_calc_type_ eGFR_ pediatric_
#>   [mg/… [mg/L] [yea… <lgl> <lgl>      [m] [mg/… <chr>           [mL/… <lgl>     
#> 1   0.5   NA      20 FALSE FALSE       NA    NA eGFR_adult_SCr   139. FALSE     
#> 2  NA      0.4    20 FALSE FALSE       NA    NA eGFR_adult_SCy…  162. FALSE     
#> 3   0.5    0.4    20 FALSE FALSE       NA    NA eGFR_adult_SCr…  167. FALSE     
#> 4   0.5   NA      30 FALSE TRUE        NA    NA eGFR_adult_SCr   150. FALSE     
#> 5  NA      0.4    30 FALSE TRUE        NA    NA eGFR_adult_SCy…  155. FALSE     
#> 6   0.5    0.4    30 FALSE TRUE        NA    NA eGFR_adult_SCr…  171. FALSE

eGFR(eGFR_pt_data,
  SCr = "SCr_", SCysC = "SCysC_",
  Age = "Age_", height = "height_", BUN = "BUN_",
  male = "male_", black = "black_", pediatric = "pediatric_"
)
#> Units: [mL/1.73m2/min]
#>  [1] 139.32466 161.68446 166.81886 150.52336 155.33226 171.35616 139.32466
#>  [8]  66.77365  96.41798 150.52336  64.15027  99.04045  49.63420 161.68446
#> [15]  97.06854  53.62373 155.33226  99.70870  49.63420  66.77365  56.10368
#> [22]  53.62373  64.15027  57.62964 155.99874 173.48118 178.86404 168.53768
#> [29] 166.66552 183.72895 155.99874  71.64555 103.37985 168.53768  68.83077
#> [36] 106.19167  66.06766 173.48118 116.50660  71.37808 166.66552 119.67546
#> [43]  66.06766  71.64555  67.33849  71.37808  68.83077  69.17003  99.12000
#> [50] 148.21219 165.89761

eGFR_pt_data %>%
  dplyr::mutate(eGFR = eGFR(
    SCr = SCr_, SCysC = SCysC_,
    Age = Age_, height = height_, BUN = BUN_,
    male = male_, black = black_, pediatric = pediatric_
  )) %>%
  select(SCr_:pediatric_, eGFR)
#> # A tibble: 51 × 11
#>       SCr_ SCysC_    Age_ male_ black_ height_    BUN_ eGFR_calc_type_     eGFR_
#>    [mg/dl] [mg/L] [years] <lgl> <lgl>      [m] [mg/dl] <chr>               [mL/…
#>  1     0.5   NA        20 FALSE FALSE       NA      NA eGFR_adult_SCr      139. 
#>  2    NA      0.4      20 FALSE FALSE       NA      NA eGFR_adult_SCysC    162. 
#>  3     0.5    0.4      20 FALSE FALSE       NA      NA eGFR_adult_SCr_SCy… 167. 
#>  4     0.5   NA        30 FALSE TRUE        NA      NA eGFR_adult_SCr      150. 
#>  5    NA      0.4      30 FALSE TRUE        NA      NA eGFR_adult_SCysC    155. 
#>  6     0.5    0.4      30 FALSE TRUE        NA      NA eGFR_adult_SCr_SCy… 171. 
#>  7     0.5   NA        20 FALSE FALSE       NA      NA eGFR_adult_SCr      139. 
#>  8    NA      1.2      20 FALSE FALSE       NA      NA eGFR_adult_SCysC     66.8
#>  9     0.5    1.2      20 FALSE FALSE       NA      NA eGFR_adult_SCr_SCy…  96.4
#> 10     0.5   NA        30 FALSE TRUE        NA      NA eGFR_adult_SCr      150. 
#> # ℹ 41 more rows
#> # ℹ 2 more variables: pediatric_ <lgl>, eGFR [mL/1.73m2/min]

eGFR_pt_data %>%
  dplyr::mutate(eGFR = eGFR_adult_SCr(
    SCr = SCr_, Age = Age_, male = male_, black = black_
  )) %>%
  select(SCr_:pediatric_, eGFR)
#> # A tibble: 51 × 11
#>       SCr_ SCysC_    Age_ male_ black_ height_    BUN_ eGFR_calc_type_     eGFR_
#>    [mg/dl] [mg/L] [years] <lgl> <lgl>      [m] [mg/dl] <chr>               [mL/…
#>  1     0.5   NA        20 FALSE FALSE       NA      NA eGFR_adult_SCr      139. 
#>  2    NA      0.4      20 FALSE FALSE       NA      NA eGFR_adult_SCysC    162. 
#>  3     0.5    0.4      20 FALSE FALSE       NA      NA eGFR_adult_SCr_SCy… 167. 
#>  4     0.5   NA        30 FALSE TRUE        NA      NA eGFR_adult_SCr      150. 
#>  5    NA      0.4      30 FALSE TRUE        NA      NA eGFR_adult_SCysC    155. 
#>  6     0.5    0.4      30 FALSE TRUE        NA      NA eGFR_adult_SCr_SCy… 171. 
#>  7     0.5   NA        20 FALSE FALSE       NA      NA eGFR_adult_SCr      139. 
#>  8    NA      1.2      20 FALSE FALSE       NA      NA eGFR_adult_SCysC     66.8
#>  9     0.5    1.2      20 FALSE FALSE       NA      NA eGFR_adult_SCr_SCy…  96.4
#> 10     0.5   NA        30 FALSE TRUE        NA      NA eGFR_adult_SCr      150. 
#> # ℹ 41 more rows
#> # ℹ 2 more variables: pediatric_ <lgl>, eGFR [mL/1.73m2/min]

Reference

See https://alwinw.github.io/epocakir/reference/index.html for more usage details and package reference.

See https://kdigo.org/guidelines/ for full KDIGO guidelines.