Introduction
In this vignette is assessed how daily dose is calculated in the
DrugUtilisation package. This function is used internally in
addDrugUtilisation().
Daily dose
Daily dose is always computed at the ingredient level. So we can calculate the daily dose for each record in drug exposure table for each given ingredient. Then the first step to calculate the daily dose for a given drug record and an ingredient concept id is to examine the relationship between drug concept id and ingredient concept id through the drug strength table:
library(DrugUtilisation)
cdm <- mockDrugUtilisation(numberIndividuals = 100, seed = 123456)
cdm$drug_strength |>
dplyr::glimpse()
#> Rows: ??
#> Columns: 12
#> Database: DuckDB v1.1.2 [unknown@Linux 6.5.0-1025-azure:R 4.4.2/:memory:]
#> $ drug_concept_id <dbl> 1125315, 1125360, 1503297, 1503327, 150332…
#> $ ingredient_concept_id <dbl> 1125315, 1125315, 1503297, 1503297, 150329…
#> $ amount_value <dbl> NA, 5.0e+02, NA, 1.0e+03, 5.0e+02, NA, NA,…
#> $ amount_unit_concept_id <dbl> 8576, 8576, 8576, 8576, 8576, 8510, NA, NA…
#> $ numerator_value <dbl> NA, NA, NA, NA, NA, NA, 100, 300, NA, NA, …
#> $ numerator_unit_concept_id <dbl> NA, NA, NA, NA, NA, NA, 8510, 8510, NA, NA…
#> $ denominator_value <dbl> NA, NA, NA, NA, NA, NA, NA, 3, NA, NA, NA,…
#> $ denominator_unit_concept_id <dbl> NA, NA, NA, NA, NA, NA, 8587, 8587, NA, NA…
#> $ box_size <lgl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
#> $ valid_start_date <chr> "01/01/1970", "01/01/1970", "01/01/1970", …
#> $ valid_end_date <chr> "31/12/2099", "31/12/2099", "31/12/2099", …
#> $ invalid_reason <lgl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…You can read the documentation of the drug strength table and description of the different fields here: https://www.ohdsi.org/web/wiki/doku.php?id=documentation:cdm:drug_strength.
Not all drug concept ids and ingredient concept ids can be related,
if no relation is found then daily dose is considered as
NA.
Using vocabulary version: “v5.0 31-AUG-23” there exist 2,980,115 relationships between a drug concept id and an ingredient concept id. These relationships can be classified into 128 different patterns. Patterns are identified of combinations of 6 variables:
- amount: Whether the amount_value field is numeric or NA.
- amount_unit: The unit of the amount field.
- numerator: Whether the numerator_value field is numeric or NA.
- numerator_unit: The unit of the numerator field.
- denominator: Whether the denominator_value field is numeric or NA.
- denominator_unit: The unit of the denominator field.
These 128 combinations were analysed to see if they could be used to compute daily dose. 41 viable patterns were identified, these patterns covered a total of 2,514,608 (84%) relationships between drug concept id and ingredient concept id. The patterns were classified into 4 different formulas:
- Time based with denominator
This formula was applied for the following 3 patterns that cover 8,044 (<1%) relationships:
pattern_id |
amount |
amount_unit |
numerator |
numerator_unit |
denominator |
denominator_unit |
|---|---|---|---|---|---|---|
1 |
number |
microgram |
number |
hour |
||
2 |
number |
milligram |
number |
hour |
||
3 |
number |
unit |
number |
hour |
The formula in this case will be as follows:
$$\begin{equation} if (denominator > 24) \rightarrow \textrm{daily dose} = 24 \cdot \frac{numerator}{denominator} \\ if (denominator \leq 24) \rightarrow \textrm{daily dose} = numerator \end{equation}$$
Note that daily dose has always unit associated in this case it will
be determined by the numerator_unit field.
- Time based no denominator
This formula was applied for the following 2 patterns that cover 5,611 (<1%) relationships:
pattern_id |
amount |
amount_unit |
numerator |
numerator_unit |
denominator |
denominator_unit |
|---|---|---|---|---|---|---|
4 |
number |
microgram |
hour |
|||
5 |
number |
milligram |
hour |
The formula in this case will be as follows:
In this case unit will be determined by the
numerator_unit field.
- Fixed amount formulation
This formula was applied for the following 6 patterns that cover 1,102,435 (37%) relationships:
pattern_id |
amount |
amount_unit |
numerator |
numerator_unit |
denominator |
denominator_unit |
|---|---|---|---|---|---|---|
6 |
number |
international unit |
||||
7 |
number |
microgram |
||||
8 |
number |
milliequivalent |
||||
9 |
number |
milligram |
||||
10 |
number |
milliliter |
||||
11 |
number |
unit |
The formula in this case will be as follows:
In this case unit will be determined by the amount_unit
field.
- Concentration formulation
This formula was applied for the following 30 patterns that cover 1,398,518 (47%) relationships:
pattern_id |
amount |
amount_unit |
numerator |
numerator_unit |
denominator |
denominator_unit |
|---|---|---|---|---|---|---|
12 |
number |
international unit |
number |
milligram |
||
13 |
number |
international unit |
number |
milliliter |
||
14 |
number |
milliequivalent |
number |
milliliter |
||
15 |
number |
milligram |
number |
Actuation |
||
16 |
number |
milligram |
number |
liter |
||
17 |
number |
milligram |
number |
milligram |
||
18 |
number |
milligram |
number |
milliliter |
||
19 |
number |
milligram |
number |
square centimeter |
||
20 |
number |
milliliter |
number |
milligram |
||
21 |
number |
milliliter |
number |
milliliter |
||
22 |
number |
unit |
number |
Actuation |
||
23 |
number |
unit |
number |
milligram |
||
24 |
number |
unit |
number |
milliliter |
||
25 |
number |
unit |
number |
square centimeter |
||
26 |
number |
international unit |
milligram |
|||
27 |
number |
international unit |
milliliter |
|||
28 |
number |
mega-international unit |
milliliter |
|||
29 |
number |
milliequivalent |
milligram |
|||
30 |
number |
milliequivalent |
milliliter |
|||
31 |
number |
milligram |
Actuation |
|||
32 |
number |
milligram |
liter |
|||
33 |
number |
milligram |
milligram |
|||
34 |
number |
milligram |
milliliter |
|||
35 |
number |
milligram |
square centimeter |
|||
36 |
number |
milliliter |
milligram |
|||
37 |
number |
milliliter |
milliliter |
|||
38 |
number |
unit |
Actuation |
|||
39 |
number |
unit |
milligram |
|||
40 |
number |
unit |
milliliter |
|||
41 |
number |
unit |
square centimeter |
The formula in this case will be as follows:
In this case unit will be determined by the
numerator_unit field.
For formulas (3) and (4) quantity is obtained from the
quantity column of the drug exposure table and
time exposed is obtained as the difference in days between
drug_exposure_start_date and
drug_exposure_end_date plus one.
The described formulas and patterns can be found in the exported
patternsWithFormula data set:
patternsWithFormula |>
dplyr::glimpse()
#> Rows: 41
#> Columns: 9
#> $ pattern_id <dbl> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16…
#> $ amount <chr> NA, NA, NA, NA, NA, "number", "number", "number", "nu…
#> $ amount_unit <chr> NA, NA, NA, NA, NA, "international unit", "microgram"…
#> $ numerator <chr> "number", "number", "number", "number", "number", NA,…
#> $ numerator_unit <chr> "microgram", "milligram", "unit", "microgram", "milli…
#> $ denominator <chr> "number", "number", "number", NA, NA, NA, NA, NA, NA,…
#> $ denominator_unit <chr> "hour", "hour", "hour", "hour", "hour", NA, NA, NA, N…
#> $ formula_name <chr> "time based with denominator", "time based with denom…
#> $ formula <chr> "if (denominator>24) {numerator * 24 / denominator} e…The described formulas were validated into 5 different databases and the results were included in an article. Please refer to it for more details on dose calculations: Calculating daily dose in the Observational Medical Outcomes Partnership Common Data Model.
Finding out the pattern information using patternTable() function
The user could also find the patterns used in the
drug_strength table. The output will also include a column
of potentially valid and invalid combinations. The idea of a pattern to
provide a platform to associate each drug in the
drug_strength table with its constituent ingredients.
patternTable(cdm) |>
dplyr::glimpse()
#> Rows: 5
#> Columns: 12
#> $ pattern_id <dbl> 9, 18, 24, 40, NA
#> $ formula_name <chr> "fixed amount formulation", "concentration…
#> $ validity <chr> "pattern with formula", "pattern with form…
#> $ number_concepts <dbl> 7, 1, 1, 1, 4
#> $ number_ingredients <dbl> 4, 1, 1, 1, 4
#> $ number_records <dbl> 169, 34, 32, 35, 25
#> $ amount_numeric <dbl> 1, 0, 0, 0, NA
#> $ amount_unit_concept_id <dbl> 8576, NA, NA, NA, NA
#> $ numerator_numeric <dbl> 0, 1, 1, 1, NA
#> $ numerator_unit_concept_id <dbl> NA, 8576, 8510, 8510, NA
#> $ denominator_numeric <dbl> 0, 1, 1, 0, NA
#> $ denominator_unit_concept_id <dbl> NA, 8587, 8587, 8587, NAThe output has three important columns, namely
number_concepts, number_ingredients and
number_records, which corresponds to count of distinct
concepts in the patterns, count of distinct ingredients involved and
overall count of records in the patterns respectively. The
pattern_id column can be used to relate the patterns with
the patternsWithFormula data set.
Finding out the dose coverage using summariseDoseCoverage() function
This package also provides a functionality to check the coverage daily dose computation for chosen concept sets and ingredients. Let’s take acetaminophen as an example.
summariseDoseCoverage(cdm = cdm, ingredientConceptId = 1125315) |>
dplyr::glimpse()
#> Rows: 56
#> Columns: 13
#> $ result_id <int> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,…
#> $ cdm_name <chr> "DUS MOCK", "DUS MOCK", "DUS MOCK", "DUS MOCK", "DUS …
#> $ group_name <chr> "ingredient_name", "ingredient_name", "ingredient_nam…
#> $ group_level <chr> "acetaminophen", "acetaminophen", "acetaminophen", "a…
#> $ strata_name <chr> "overall", "overall", "overall", "overall", "overall"…
#> $ strata_level <chr> "overall", "overall", "overall", "overall", "overall"…
#> $ variable_name <chr> "number records", "Missing dose", "Missing dose", "da…
#> $ variable_level <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, N…
#> $ estimate_name <chr> "count", "count_missing", "percentage_missing", "mean…
#> $ estimate_type <chr> "integer", "integer", "percentage", "numeric", "numer…
#> $ estimate_value <chr> "78", "0", "0", "14949.800361887", "99310.1784420234"…
#> $ additional_name <chr> "overall", "overall", "overall", "overall", "overall"…
#> $ additional_level <chr> "overall", "overall", "overall", "overall", "overall"…The output will summarise the usage of acetaminophen in the
database. For example, overall there are
records of acetaminophen and for all of them daily dose can be
calculated. By default the output will also include the mean, median,
lower and upper quartiles and standard deviation of the daily dose of
acetaminophen calculated as explained above. The results will
also be stratified by unit, route and pattern (which we saw in
patternsWithFormula data set).
Different routes are documented in the
CodelistGenerator package. Route is defined at the
concept (drug_concept_id) level, there exist an equivalence
between each concept and a route. You can stratify a codelist using the
function: CodelistGenerator::stratifyByRouteCategory().
To better inspect the content of the output of
summariseDoseCoverage() we can create a gt table like
so:
coverageResult <- summariseDoseCoverage(cdm = cdm, ingredientConceptId = 1125315)
tableDoseCoverage(coverageResult)|
Variable
|
||||||||
|---|---|---|---|---|---|---|---|---|
|
number records
|
Missing dose
|
daily_dose
|
||||||
| Database name | Ingredient name | Unit | Route | Pattern id |
Estimate name
|
|||
| N | N (%) | Mean (SD) | Median (Q25 - Q75) | |||||
| DUS MOCK | acetaminophen | overall | overall | overall | 78 | 0 (0.00 %) | 14,949.80 (99,310.18) | 234.38 (19.68 - 1,264.66) |
| milligram | overall | overall | 78 | 0 (0.00 %) | 14,949.80 (99,310.18) | 234.38 (19.68 - 1,264.66) | ||
| oral | overall | 18 | 0 (0.00 %) | 182.38 (294.08) | 41.03 (15.69 - 237.42) | |||
| topical | overall | 60 | 0 (0.00 %) | 19,380.03 (113,070.32) | 308.61 (28.48 - 1,962.38) | |||
| oral | 9 | 18 | 0 (0.00 %) | 182.38 (294.08) | 41.03 (15.69 - 237.42) | |||
| topical | 18 | 34 | 0 (0.00 %) | 32,837.73 (149,637.62) | 1,066.79 (243.71 - 4,339.96) | |||
| 9 | 26 | 0 (0.00 %) | 1,781.50 (6,876.01) | 49.95 (4.96 - 268.87) | ||||
The user also has the freedom to customize the gt table output. For
example the following will suppress the cdmName:
tableDoseCoverage(coverageResult, cdmName = F)|
Variable
|
|||||||
|---|---|---|---|---|---|---|---|
|
number records
|
Missing dose
|
daily_dose
|
|||||
| Ingredient name | Unit | Route | Pattern id |
Estimate name
|
|||
| N | N (%) | Mean (SD) | Median (Q25 - Q75) | ||||
| acetaminophen | overall | overall | overall | 78 | 0 (0.00 %) | 14,949.80 (99,310.18) | 234.38 (19.68 - 1,264.66) |
| milligram | overall | overall | 78 | 0 (0.00 %) | 14,949.80 (99,310.18) | 234.38 (19.68 - 1,264.66) | |
| oral | overall | 18 | 0 (0.00 %) | 182.38 (294.08) | 41.03 (15.69 - 237.42) | ||
| topical | overall | 60 | 0 (0.00 %) | 19,380.03 (113,070.32) | 308.61 (28.48 - 1,962.38) | ||
| oral | 9 | 18 | 0 (0.00 %) | 182.38 (294.08) | 41.03 (15.69 - 237.42) | ||
| topical | 18 | 34 | 0 (0.00 %) | 32,837.73 (149,637.62) | 1,066.79 (243.71 - 4,339.96) | ||
| 9 | 26 | 0 (0.00 %) | 1,781.50 (6,876.01) | 49.95 (4.96 - 268.87) | |||