MedDRA (Medical Dictionary for Regulatory Activities) is a standardized medical terminology used in clinical trials and regulatory submissions to classify adverse events. It is organized as a five-level hierarchy:
| Level | Abbreviation | Description |
|---|---|---|
| 1 (broadest) | SOC | System Organ Class |
| 2 | HLGT | High Level Group Term |
| 3 | HLT | High Level Term |
| 4 | PT | Preferred Term |
| 5 (most specific) | LLT | Lowest Level Term |
MedDRA data is proprietary and requires a license from the MedDRA MSSO. This package helps you load and work with your licensed MedDRA data. The examples below use a small, clearly fictional dataset bundled with the package for illustration purposes.
When you download a licensed MedDRA release, it contains two subdirectories:
my_meddra_release/
├── MedAscii/ # Main MedDRA data files (.asc)
│ ├── soc.asc # System Organ Classes
│ ├── hlgt.asc # High Level Group Terms
│ ├── hlt.asc # High Level Terms
│ ├── pt.asc # Preferred Terms
│ ├── llt.asc # Lowest Level Terms
│ ├── hlt_pt.asc # HLT to PT linking table
│ ├── hlgt_hlt.asc # HLGT to HLT linking table
│ ├── soc_hlgt.asc # SOC to HLGT linking table
│ ├── mdhier.asc # Denormalized hierarchy
│ ├── meddra_release.asc # Version information
│ └── ... # Additional files (SMQ, specialties, etc.)
└── SeqAscii/ # Sequential update files (.seq)
├── soc.seq
├── pt.seq
└── ...All files use the $ character as a field separator.
Use read_meddra() pointing to the parent directory that
contains MedAscii and SeqAscii (or
MedSeq) subdirectories. It returns a named list of
data.frames, one per file.
# For your licensed data, replace this path with your actual MedDRA directory:
# example_dir <- "/path/to/your/meddra/release"
# The package includes a small fictional dataset for illustration:
example_dir <- system.file("example_meddra", package = "meddra.read")
meddra_raw <- read_meddra(example_dir)The result is a named list with one data.frame per MedDRA file:
names(meddra_raw)
#> [1] "hlgt_hlt.asc" "hlgt.asc"
#> [3] "hlt_pt.asc" "hlt.asc"
#> [5] "intl_ord.asc" "llt.asc"
#> [7] "mdhier.asc" "meddra_history_english.asc"
#> [9] "meddra_release.asc" "pt.asc"
#> [11] "smq_content.asc" "smq_list.asc"
#> [13] "soc_hlgt.asc" "soc.asc"
#> [15] "spec_pt.asc" "spec.asc"
#> [17] "hlgt_hlt.seq" "hlgt.seq"
#> [19] "hlt_pt.seq" "hlt.seq"
#> [21] "intl_ord.seq" "llt.seq"
#> [23] "mdhier.seq" "pt.seq"
#> [25] "soc_hlgt.seq" "soc.seq"
#> [27] "spec_pt.seq" "spec.seq"Each data.frame corresponds to one of the MedDRA source files. For example, the System Organ Class data:
meddra_raw$soc.asc
#> soc_code soc_name soc_abbrev
#> 1 10000100 Example Nervous System Disorders ExNS
#> 2 10000200 Example Cardiac Disorders ExCDThe Preferred Terms:
meddra_raw$pt.asc
#> pt_code pt_name pt_soc_code
#> 1 10001111 Example Headache 10000100
#> 2 10002111 Example Palpitations 10000200The Lowest Level Terms (note: llt_currency = "Y" means
the term is current; "N" means it is a non-current
synonym):
join_meddra() merges all the hierarchy tables into a
single flat data.frame, making it easy to look up or filter by any level
of the hierarchy:
meddra_df <- join_meddra(meddra_raw)
meddra_df
#> soc_code soc_name soc_abbrev hlgt_code
#> 1 10000100 Example Nervous System Disorders ExNS 10001100
#> 2 10000100 Example Nervous System Disorders ExNS 10001100
#> 3 10000200 Example Cardiac Disorders ExCD 10002100
#> hlgt_name hlt_code hlt_name
#> 1 Example Neurological Disorders 10001110 Example Headache Disorders
#> 2 Example Neurological Disorders 10001110 Example Headache Disorders
#> 3 Example Cardiac Structural Disorders 10002110 Example Arrhythmia Disorders
#> pt_code pt_name pt_soc_code llt_code llt_name
#> 1 10001111 Example Headache 10000100 10001111 Example Headache
#> 2 10001111 Example Headache 10000100 10001112 Example Head Pain
#> 3 10002111 Example Palpitations 10000200 10002111 Example Palpitations
#> llt_currency primary_soc_fg
#> 1 Y Y
#> 2 N Y
#> 3 Y YThe resulting data.frame has one row per LLT (Lowest Level Term) and includes all parent hierarchy levels. The columns are:
| Column | Description |
|---|---|
soc_code, soc_name,
soc_abbrev |
System Organ Class |
hlgt_code, hlgt_name |
High Level Group Term |
hlt_code, hlt_name |
High Level Term |
pt_code, pt_name,
pt_soc_code |
Preferred Term |
llt_code, llt_name,
llt_currency |
Lowest Level Term |
primary_soc_fg |
"Y" if this SOC is the primary (preferred) SOC for the
PT |
To work with terms from a specific SOC:
subset(meddra_df, soc_name == "Example Nervous System Disorders")
#> soc_code soc_name soc_abbrev hlgt_code
#> 1 10000100 Example Nervous System Disorders ExNS 10001100
#> 2 10000100 Example Nervous System Disorders ExNS 10001100
#> hlgt_name hlt_code hlt_name pt_code
#> 1 Example Neurological Disorders 10001110 Example Headache Disorders 10001111
#> 2 Example Neurological Disorders 10001110 Example Headache Disorders 10001111
#> pt_name pt_soc_code llt_code llt_name llt_currency
#> 1 Example Headache 10000100 10001111 Example Headache Y
#> 2 Example Headache 10000100 10001112 Example Head Pain N
#> primary_soc_fg
#> 1 Y
#> 2 YTo find all Lowest Level Terms (including non-current synonyms) for a given PT:
Non-current LLTs (llt_currency = "N") are historical
synonyms. In most analyses you will want to keep only current terms:
Standardized MedDRA Queries (SMQs) are pre-defined sets of terms used
to search for adverse events. The SMQ data is available in
smq_list.asc and smq_content.asc:
meddra_raw$smq_list.asc
#> smq_code smq_name smq_level
#> 1 10000001 Example Headache SMQ 2
#> smq_description smq_source smq_note
#> 1 Example SMQ for headache-related terms Example source Example note
#> MedDRA_version status smq_algorithm
#> 1 0 A N
meddra_raw$smq_content.asc
#> smq_code term_code term_level term_scope term_category term_weight
#> 1 10000001 10001111 4 1 A 0
#> term_status term_addition_version term_last_modified_version
#> 1 A 0 0