Rchemist - 통계 테이블 끝판왕 gtsummary

패키지 소개

데이터 요약 & 회귀분석 결과 테이블 만들기

필요한 패키지 설치

library(gtsummary)
library(gt)

1. Baseline characteristics

data의 일반 특성 요약
집단 별 통계량 표시

기본 테이블

library(gtsummary)
library(gt)
library(survival)
library(flextable)
tbl_summary(
  data = trial,
  by = trt
) |> 
  add_overall() |> 
  as_flex_table()

Characteristic	Overall, N = 2001	Drug A, N = 981	Drug B, N = 1021
Age	47 (38, 57)	46 (37, 59)	48 (39, 56)
Unknown	11	7	4
Marker Level (ng/mL)	0.64 (0.22, 1.39)	0.84 (0.24, 1.57)	0.52 (0.19, 1.20)
Unknown	10	6	4
T Stage
T1	53 (27%)	28 (29%)	25 (25%)
T2	54 (27%)	25 (26%)	29 (28%)
T3	43 (22%)	22 (22%)	21 (21%)
T4	50 (25%)	23 (23%)	27 (26%)
Grade
I	68 (34%)	35 (36%)	33 (32%)
II	68 (34%)	32 (33%)	36 (35%)
III	64 (32%)	31 (32%)	33 (32%)
Tumor Response	61 (32%)	28 (29%)	33 (34%)
Unknown	7	3	4
Patient Died	112 (56%)	52 (53%)	60 (59%)
Months to Death/Censor	22.4 (16.0, 24.0)	23.5 (17.4, 24.0)	21.2 (14.6, 24.0)
1Median (IQR); n (%)

통계량 변경

테이블에 표시되는 숫자의 양식을 변경해줄 수 있습니다. 예를 들어, 연속형(continuous) 변수의 경우 기본적으로 median(IQR)로 표시가 됩니다. tbl_summary()의 statistic 인자를 통해 mean ± sd 형태로 표시할 수 있습니다.

주의해야 할 점은 항상 {}가 붙은 문자 형태("")로 사용해야 한다는 것입니다.

tbl_summary(
  data = trial,
  by = trt,
  include = c(age, marker, stage, grade, response, death, ttdeath),
  type = list(
    c(stage, grade)~ "categorical"
  ),
  statistic = list(
    all_continuous() ~"{mean} ± {sd}",
    all_categorical() ~ "{n} ({p})"
  ),
  digits = list(
    all_continuous() ~ 1,
    all_categorical() ~ c(0,1)
  ),
  missing = "no",
) |> 
  add_overall() |> 
  as_flex_table()

Characteristic	Overall, N = 2001	Drug A, N = 981	Drug B, N = 1021
Age	47.2 ± 14.3	47.0 ± 14.7	47.4 ± 14.0
Marker Level (ng/mL)	0.9 ± 0.9	1.0 ± 0.9	0.8 ± 0.8
T Stage
T1	53 (26.5)	28 (28.6)	25 (24.5)
T2	54 (27.0)	25 (25.5)	29 (28.4)
T3	43 (21.5)	22 (22.4)	21 (20.6)
T4	50 (25.0)	23 (23.5)	27 (26.5)
Grade
I	68 (34.0)	35 (35.7)	33 (32.4)
II	68 (34.0)	32 (32.7)	36 (35.3)
III	64 (32.0)	31 (31.6)	33 (32.4)
Tumor Response	61 (31.6)	28 (29.5)	33 (33.7)
Patient Died	112 (56.0)	52 (53.1)	60 (58.8)
Months to Death/Censor	19.6 ± 5.3	20.2 ± 5.0	19.0 ± 5.5
1Mean ± SD; n (%)

P-value

tbl_summary(
   data = trial,
  by = trt,
  include = c(age, marker, stage, grade, response, death, ttdeath),
  statistic = list(
    all_continuous() ~"{mean} ± {sd}",
    # statistic = "{median} ({p25}-{p75})",
    all_categorical() ~ "{n} ({p})"
  ),
  digits = list(
    all_continuous() ~ 1,
    all_categorical() ~ c(0,1)
  ),
  missing = "no",
) |> 
  add_overall() |> 
  add_p(
    test = list(
      all_continuous() ~ "t.test",
      all_categorical() ~ "chisq.test"
    ),
    pvalue_fun = ~style_pvalue(., digits = 3)
  ) |> 
    as_flex_table()

Characteristic	Overall, N = 2001	Drug A, N = 981	Drug B, N = 1021	p-value2
Age	47.2 ± 14.3	47.0 ± 14.7	47.4 ± 14.0	0.834
Marker Level (ng/mL)	0.9 ± 0.9	1.0 ± 0.9	0.8 ± 0.8	0.116
T Stage				0.866
T1	53 (26.5)	28 (28.6)	25 (24.5)
T2	54 (27.0)	25 (25.5)	29 (28.4)
T3	43 (21.5)	22 (22.4)	21 (20.6)
T4	50 (25.0)	23 (23.5)	27 (26.5)
Grade				0.871
I	68 (34.0)	35 (35.7)	33 (32.4)
II	68 (34.0)	32 (32.7)	36 (35.3)
III	64 (32.0)	31 (31.6)	33 (32.4)
Tumor Response	61 (31.6)	28 (29.5)	33 (33.7)	0.637
Patient Died	112 (56.0)	52 (53.1)	60 (58.8)	0.498
Months to Death/Censor	19.6 ± 5.3	20.2 ± 5.0	19.0 ± 5.5	0.108
1Mean ± SD; n (%)
2Welch Two Sample t-test; Pearson's Chi-squared test

2. Regression table 만들기

단순 회귀분석(Univariable regression)

단순회귀분석의 경우,

library(survival)
uni_tbl <- tbl_uvregression(
  data = trial |> select(age, marker, stage, grade, response, death, ttdeath),
  method = coxph,
  y = Surv(ttdeath, death),
  exponentiate = T,
  hide_n = T
  ) |> 
  modify_footnote(everything() ~ NA)
uni_tbl |> as_flex_table()

Characteristic	HR1	95% CI1	p-value
Age	1.01	0.99, 1.02	0.3
Marker Level (ng/mL)	0.91	0.72, 1.15	0.4
T Stage
T1	—	—
T2	1.18	0.68, 2.04	0.6
T3	1.23	0.69, 2.20	0.5
T4	2.48	1.49, 4.14	<0.001
Grade
I	—	—
II	1.28	0.80, 2.05	0.3
III	1.69	1.07, 2.66	0.024
Tumor Response	0.50	0.31, 0.78	0.003
1HR = Hazard Ratio, CI = Confidence Interval

다변량 회귀분석(Multivariable regression)

다중 회귀분석 테이블을 만드는 함수는 tbl_regression()입니다. 단순 회귀분석을 만드는 tbl_uvregression()과 다르게, 먼저 회귀식을 작성한 뒤에 tbl_regression()의 인자로 사용해야 합니다.

fit <- coxph(Surv(ttdeath, death) ~ age +  marker + stage + grade + response, data = trial)
mult_tbl <- tbl_regression(
  x = fit,
  exponentiate = T,
  pvalue_fun = ~ style_pvalue(., digits = 3),
  estimate_fun = ~style_ratio(., digits = 2)
) |> 
  bold_p() |> 
  # modify_column_merge(pattern = "{estimate} ({conf.low}-{conf.high})",
  #                     rows = !is.na(estimate)) |> 
  add_significance_stars(hide_ci = T, hide_se = T) |> 
  modify_header(
    label = "**Variable**",
    estimate = "**OR (95% CI)**",
    p.value = "**P value**"
  ) |> 
  modify_footnote(everything() ~ NA)
mult_tbl |> as_flex_table()

Variable	OR (95% CI)1	P value
Age	1.02*	0.023
Marker Level (ng/mL)	0.89	0.398
T Stage
T1	—
T2	1.33	0.376
T3	1.60	0.183
T4	3.65***	<0.001
Grade
I	—
II	1.27	0.395
III	1.75*	0.025
Tumor Response	0.41***	<0.001
1HR = Hazard Ratio

두 개의 테이블 합치기

이제 단순 회귀분석 테이블과 다중 회귀분석 테이블을 합쳐보도록 하겠습니다.

tbl_merge(
  list(uni_tbl, mult_tbl),
  tab_spanner = 
    c("**Univariable**", "**Multivariable**")
) |> 
  as_flex_table()

	Univariable			Multivariable
Characteristic	HR1	95% CI1	p-value	OR (95% CI)1	P value
Age	1.01	0.99, 1.02	0.3	1.02*	0.023
Marker Level (ng/mL)	0.91	0.72, 1.15	0.4	0.89	0.398
T Stage
T1	—	—		—
T2	1.18	0.68, 2.04	0.6	1.33	0.376
T3	1.23	0.69, 2.20	0.5	1.60	0.183
T4	2.48	1.49, 4.14	<0.001	3.65***	<0.001
Grade
I	—	—		—
II	1.28	0.80, 2.05	0.3	1.27	0.395
III	1.69	1.07, 2.66	0.024	1.75*	0.025
Tumor Response	0.50	0.31, 0.78	0.003	0.41***	<0.001
1HR = Hazard Ratio, CI = Confidence Interval

참고자료

https://www.danieldsjoberg.com/gtsummary/reference/tbl_summary.html