Process UKB ICD data for survival analysis
diagnosis_process_icd.RdGenerate icd_code_status and icd_code_time from UKB ICD records.
Usage
diagnosis_process_icd(
data,
icd_code,
icd = 10,
censored = "2025-01-01",
icd_rank = 1,
baseline_date = "p53_i0",
filter_NA = FALSE
)Arguments
- data
A data.frame containing UKB baseline, censoring, ICD code and date columns.
- icd_code
A target ICD code prefix or a character vector of prefixes (e.g., "M10" or c("E10", "E11")).
- icd
ICD version, 10 or 9.
- censored
Censoring date ("YYYY-MM-DD").
- icd_rank
Rank of diagnosis occurrence to consider (1 = first occurrence, 2 = second occurrence, etc.).
- baseline_date
Baseline date column name (default "p53_i0").
- filter_NA
Logical, whether to filter out NA row. Default FALSE (Suitable for comorbidity check).
Examples
df_example <- data.frame(
eid = 1:5,
p53_i0 = as.Date(c("2006-01-01", "2007-06-01", "2005-03-15", "2004-10-30", "2008-05-10")), # baseline
p191 = as.Date(c("2020-01-01", "2019-12-31", "2021-06-30", "2022-07-01", "2023-09-10")), # last follow-up
p40000_i0 = as.Date(c(NA, NA, NA, "2021-06-01", NA)), # death date (field 0)
p40000_i1 = as.Date(c(NA, NA, NA, NA, NA)), # death date (field 1)
p41270 = c("D261|E113|M102", "E113|M101", NA, "D261|M102|M103", "E113|M101|D261"), # ICD10 codes
p41280_a0 = as.Date(c("2010-05-01", "2011-07-20", NA, "2015-03-01", "2017-07-05")), # 1st diagnosis date
p41280_a1 = as.Date(c("2012-03-10", "2014-09-01", NA, "2016-06-01", "2018-08-10")), # 2nd diagnosis date
p41280_a2 = as.Date(c("2015-08-15", NA, NA, "2015-09-10", NA)), # 3rd diagnosis date
p41280_a3 = as.Date(c(NA, NA, NA, NA, NA)) # Additional diagnosis field for testing
)
# Prefix matching: D261 matches only D261 (icd_rank = 1)
diagnosis_process_icd(df_example, icd_code = "D261", icd = 10, icd_rank = 1)
#> ℹ [10:00:28] diagnosis_process_icd: processing ICD10 [D261] using [2025-01-01] as censored date.
#> ℹ [10:00:28] diagnosis_process_icd: converting date columns to <Date>.
#> ℹ [10:00:28] diagnosis_process_icd: combining death dates.
#> ℹ [10:00:28] diagnosis_process_icd: matching ICD10 pattern [^(D261)].
#> ℹ [10:00:28] diagnosis_process_icd: calculating survival time and status.
#> ✔ [10:00:28] diagnosis_process_icd: created [D261_status], [D261_time] - ALL DONE.
#> eid D261_status D261_time
#> 1 1 1 4.331507
#> 2 2 0 12.591781
#> 3 3 0 16.304110
#> 4 4 1 10.339726
#> 5 5 0 15.345205
# Prefix matching: M10 matches M101 / M102 (icd_rank = 1)
diagnosis_process_icd(df_example, icd_code = "M10", icd = 10, icd_rank = 1)
#> ℹ [10:00:28] diagnosis_process_icd: processing ICD10 [M10] using [2025-01-01] as censored date.
#> ℹ [10:00:28] diagnosis_process_icd: converting date columns to <Date>.
#> ℹ [10:00:28] diagnosis_process_icd: combining death dates.
#> ℹ [10:00:28] diagnosis_process_icd: matching ICD10 pattern [^(M10)].
#> ℹ [10:00:28] diagnosis_process_icd: calculating survival time and status.
#> ✔ [10:00:28] diagnosis_process_icd: created [M10_status], [M10_time] - ALL DONE.
#> eid M10_status M10_time
#> 1 1 1 9.624658
#> 2 2 1 7.257534
#> 3 3 0 16.304110
#> 4 4 1 10.868493
#> 5 5 1 10.257534
# Prefix matching: E113 matches E113 only (icd_rank = 1)
diagnosis_process_icd(df_example, icd_code = "E113", icd = 10, icd_rank = 1)
#> ℹ [10:00:28] diagnosis_process_icd: processing ICD10 [E113] using [2025-01-01] as censored date.
#> ℹ [10:00:28] diagnosis_process_icd: converting date columns to <Date>.
#> ℹ [10:00:28] diagnosis_process_icd: combining death dates.
#> ℹ [10:00:28] diagnosis_process_icd: matching ICD10 pattern [^(E113)].
#> ℹ [10:00:28] diagnosis_process_icd: calculating survival time and status.
#> ✔ [10:00:28] diagnosis_process_icd: created [E113_status], [E113_time] - ALL DONE.
#> eid E113_status E113_time
#> 1 1 1 6.191781
#> 2 2 1 4.136986
#> 3 3 0 16.304110
#> 4 4 0 16.597260
#> 5 5 1 9.158904
# Prefix matching: M10 matches M101 / M102 (icd_rank = 2)
diagnosis_process_icd(df_example, icd_code = "M10", icd = 10, icd_rank = 2)
#> ℹ [10:00:28] diagnosis_process_icd: processing ICD10 [M10] using [2025-01-01] as censored date.
#> ℹ [10:00:28] diagnosis_process_icd: converting date columns to <Date>.
#> ℹ [10:00:28] diagnosis_process_icd: combining death dates.
#> ℹ [10:00:28] diagnosis_process_icd: matching ICD10 pattern [^(M10)].
#> ℹ [10:00:28] diagnosis_process_icd: calculating survival time and status.
#> ✔ [10:00:28] diagnosis_process_icd: created [M10_status], [M10_time] - ALL DONE.
#> eid M10_status M10_time
#> 1 1 0 14.00822
#> 2 2 0 12.59178
#> 3 3 0 16.30411
#> 4 4 1 11.59452
#> 5 5 0 15.34521
# Multiple ICD codes: c("M10", "E113") - takes global earliest date among all matches
diagnosis_process_icd(df_example, icd_code = c("M10", "E113"), icd = 10, icd_rank = 1)
#> ℹ [10:00:28] diagnosis_process_icd: processing ICD10 [M10 and E113] using [2025-01-01] as censored date.
#> ℹ [10:00:28] diagnosis_process_icd: converting date columns to <Date>.
#> ℹ [10:00:28] diagnosis_process_icd: combining death dates.
#> ℹ [10:00:28] diagnosis_process_icd: matching ICD10 pattern [^(M10|E113)].
#> ℹ [10:00:28] diagnosis_process_icd: calculating survival time and status.
#> ✔ [10:00:28] diagnosis_process_icd: created [M10_E113_status], [M10_E113_time] - ALL DONE.
#> eid M10_E113_status M10_E113_time
#> 1 1 1 6.191781
#> 2 2 1 4.136986
#> 3 3 0 16.304110
#> 4 4 1 10.868493
#> 5 5 1 9.158904
# Output columns: M10_E113_status, M10_E113_time