In dylanrussellmd/nsqipPancreatectomy: What the Package Does (One Line, Title Case)
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>"
)
options("optmatch_max_problem_size" = Inf)
library(nsqipPancreatectomy)
We will start by applying some exclusion criteria. The following inclusion criteria apply:
- Inpatient
- Admitted from home
- Age, height, weight available from data
- Surgical specialty is general surgery (or missing)
- Surgery is elective (or missing)
- Patient is not on a preoperative ventilator (or missing)
- No disseminated cancer (or missing)
- No preoperative wound infection (or missing)
- No preoperative sepsis (or missing)
- Surgery is not an emergency (or missing)
- ASA class < 5
- Operative time > 20 minutes
- Gastrojejunosotomy or duodenojejunostomy not performed (or missing)
- M stage M0/Mx (or missing)
- Reconstruction not performed (or missing)
- Is not missing both malignant and benign histologic data
- Does not have histologic diagnoses of ampullary carcinoma, distal cholangiocarcinoma, or duodenal carcinoma.
df <- df %>% dplyr::filter(
inout,
transt == 2,
!is.na(age),
!is.na(height),
!is.na(weight),
(surgspec == 2 | is.na(surgspec)),
(electsurg | is.na(electsurg)),
(!ventilat | is.na(ventilat)),
(!discancr | is.na(discancr)),
(!wndinf | is.na(wndinf)),
(!prsepis | is.na(prsepis)),
(!emergncy | is.na(emergncy)),
asaclas < 5,
optime > 20,
(is.na(pan_gastduo) | pan_gastduo == 3),
(pan_mstage != 2 | is.na(pan_mstage)),
(pan_reconstruction == 4 | is.na(pan_reconstruction)),
!(is.na(pan_malig_histologic) & is.na(pan_benign_histologic)),
((pan_malig_histologic != 1 & pan_malig_histologic != 3 &
pan_malig_histologic != 4) | is.na(pan_malig_histologic))
)
I will repeat the analysis I completed previously.
df$pan_approach <- factor(df$pan_approach, levels = c(1,2,3,4,5,6,7,8),
labels = c("Hybrid","Laparoscopic","NOTES","Open (planned)",
"Other","Other MIS","Robotic","SILS"))
df <- df %>%
dplyr::filter(pan_approach %in% c("Laparoscopic","Robotic","Open (planned)"),
!pan_open_assist) %>%
dplyr::mutate(group = pan_approach %in% c("Robotic", "Laparoscopic")) %>%
dplyr::mutate(subgroup = ((pan_approach %in% c("Robotic", "Laparoscopic")) + pan_unplanned_conversion))
Next, we will assign a variable to each patient indicating if they have an additional CPT coding for splenectomy:
spleen_vec <- df %>%
dplyr::rename(othercpt0 = cpt) %>%
dplyr::mutate(dplyr::across(c(dplyr::starts_with("othercpt"), dplyr::starts_with("concpt")), as.character)) %>%
tidyr::pivot_longer(cols = c(dplyr::starts_with("othercpt"), dplyr::starts_with("concpt")), values_to = c("cpt"), names_repair = "unique") %>%
dplyr::filter(cpt %in% c("38129","38100","38101","38102")) %>%
dplyr::pull(caseid) %>%
unique()
df <- df %>%
dplyr::mutate(spleen = dplyr::if_else(caseid %in% spleen_vec, TRUE, FALSE))
df <- df %>%
dplyr::mutate(bmi = bmi(height, weight),
pan_benign_histologic = !is.na(pan_benign_histologic))
df <- df %>%
dplyr::mutate(fnstatus2 = factor(.$fnstatus2, levels = c(1, 2, 3), labels = c("Independent","Partially dependent","Totally dependent")),
pan_ductsize = factor(.$pan_ductsize, levels = c(1, 2, 3), labels = c("<3 mm","3-6 mm",">6 mm")),
wndclas = factor(wndclas),
asaclas = factor(asaclas),
pan_glandtext = factor(.$pan_glandtext, levels = c(1, 2, 3), labels = c("Soft","Intermediate","Hard")),
pan_malig_histologic = factor(.$pan_malig_histologic, levels = c(2, 5:9), labels = c("Cystadenocarcinoma", "IPMN-invasive", "Neuroendocrine-functioning", "Neuroendocrine-nonfunctioning", "Pancreatic adenocarcinoma", "Other type")),
pan_malig_histologic = forcats::fct_explicit_na(pan_malig_histologic, "No malignancy"),
pan_tstage = factor(.$pan_tstage, levels = c(1:7), labels = c("T0","T1","T2","T3","T4","Tis","Tx")),
pan_tstage = forcats::fct_explicit_na(pan_tstage, "No T stage"),
pan_nstage = factor(.$pan_nstage, levels = c(1:3), labels = c("N0","N1","Nx")),
pan_nstage = forcats::fct_explicit_na(pan_nstage, "No N stage"),
death = !is.na(yrdeath)
) %>% dplyr::mutate(dplyr::across(c("diabetes","smoke","hxcopd","ascites","hxchf","hypermed","dialysis","steroid","wtloss","bleeddis","pan_chemo","pan_radio", "pan_drains"), ~tidyr::replace_na(.x, FALSE)))
Hmisc::label(df$sex) = "Male"
Hmisc::label(df$age) = "Age"
Hmisc::label(df$bmi) = "BMI"
Hmisc::label(df$diabetes) = "Diabetes"
Hmisc::label(df$smoke) = "Smoking"
Hmisc::label(df$fnstatus2) = "Functional Status"
Hmisc::label(df$hxcopd) = "COPD"
Hmisc::label(df$ascites) = "Ascites"
Hmisc::label(df$hxchf) = "CHF"
Hmisc::label(df$hypermed) = "Hypertension medication"
Hmisc::label(df$dialysis) = "Dialysis"
Hmisc::label(df$steroid) = "Steroids"
Hmisc::label(df$wtloss) = "Weight loss"
Hmisc::label(df$bleeddis) = "Bleeding disorder"
Hmisc::label(df$spleen) = "Splenectomy"
Hmisc::label(df$wndclas) = "Wound class"
Hmisc::label(df$asaclas) = "ASA class"
Hmisc::label(df$pan_chemo) = "Chemotherapy"
Hmisc::label(df$pan_radio) = "Radiotherapy"
Hmisc::label(df$pan_ductsize) = "Duct size"
Hmisc::label(df$pan_glandtext) = "Gland texture"
Hmisc::label(df$pan_drains) = "Drains"
Hmisc::label(df$pan_malig_histologic) = "Malignant histology"
Hmisc::label(df$pan_benign_histologic) = "Benign histology"
Hmisc::label(df$pan_tstage) = "T stage"
Hmisc::label(df$pan_nstage) = "N stage"
Hmisc::label(df$supinfec) = "Superficial SSI"
Hmisc::label(df$wndinfd) = "Deep SSI"
Hmisc::label(df$orgspcssi) = "Organ Space SSI"
Hmisc::label(df$pulembol) = "Pulmonary embolism"
Hmisc::label(df$reintub) = "Reintubation"
Hmisc::label(df$failwean) = "Failure to wean"
Hmisc::label(df$renainsf) = "Renal insufficiency"
Hmisc::label(df$oprenafl) = "Renal failure"
Hmisc::label(df$death) = "Death"
We can now perform a match on the following variables:
- Sex
- Age
- BMI
- Diabetes
- Smoking
- Functional status
- History of COPD
- Ascites
- History of CHF
- Use of hypertension medications
- Dialysis
- Pre*operative steroids
- Weight loss > 10% within 6 months
- Bleeding disorders
- Concurrent splenectomy
- Wound class
- ASA class
- Pre-operative radio- and chemotherapy
- Duct size
- Gland texture
- Presence of drains
- Malignant histology
- T stage
- N stage
- The presence of benign histology
I will make the assumption for the following variables that missing data is equivalent to FALSE:
- Diabetes
- Smoking
- History of COPD
- Ascites
- History of CHF
- Hypertension medication use
- Dialysis
- Steroid use
- Weight loss
- Bleeding disorders
- Pre-operative radio- and chemotherapy
- Drain use
This is a big assumption and may be something you want to change.
With duct size and gland texture
matchdf <- df %>%
dplyr::select(caseid, group, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio,
pan_ductsize, pan_glandtext, pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage)
Many of the patients are missing data in the duct size and gland texture. Unfortunately, matching requires complete data, so I will first run a match using duct size and gland texture then a larger match without those variables.
match1 <- matchdf %>% dplyr::filter(complete.cases(.))
m1.data <- MatchIt::matchit(group ~ sex + age + bmi + diabetes + smoke + fnstatus2 + hxcopd + ascites + hxchf + hypermed + dialysis + steroid + wtloss + bleeddis + spleen + wndclas + asaclas + pan_chemo + pan_radio + pan_ductsize + pan_glandtext + pan_drains + pan_malig_histologic + pan_benign_histologic + pan_tstage + pan_nstage, data = match1) %>% MatchIt::match.data()
Note that for the tables below, "group" refers to MIS or open. "Yes" refers to the MIS group, "No" refers to the open group. I can format this better for the final document of course.
suppressWarnings(nsqipBileSpill::tableone(m1.data, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio, pan_ductsize, pan_glandtext, pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage, group = group))
m1.data <- dplyr::left_join(m1.data, df)
And we can now compare the outcomes:
suppressWarnings(nsqipBileSpill::tableone(m1.data, supinfec, wndinfd, orgspcssi, pulembol, reintub, failwean, renainsf, oprenafl, death, group = group))
Without duct size and gland texture
matchdf2 <- df %>%
dplyr::select(caseid, group, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio,
pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage)
match2 <- matchdf2 %>% dplyr::filter(complete.cases(.))
m2.data <- MatchIt::matchit(group ~ sex + age + bmi + diabetes + smoke + fnstatus2 + hxcopd + ascites + hxchf + hypermed + dialysis + steroid + wtloss + bleeddis + spleen + wndclas + asaclas + pan_chemo + pan_radio + pan_drains + pan_malig_histologic + pan_benign_histologic + pan_tstage + pan_nstage, data = match2, method = "full") %>% MatchIt::match.data()
suppressWarnings(nsqipBileSpill::tableone(m2.data, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio, pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage, group = group))
m2.data <- dplyr::left_join(m2.data, df)
And we can now compare the outcomes:
suppressWarnings(nsqipBileSpill::tableone(m2.data, supinfec, wndinfd, orgspcssi, pulembol, reintub, failwean, renainsf, oprenafl, death, group = group))
dylanrussellmd/nsqipPancreatectomy documentation built on Sept. 16, 2020, 12:46 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
knitr::opts_chunk$set( collapse = TRUE, comment = "#>" ) options("optmatch_max_problem_size" = Inf)
library(nsqipPancreatectomy)
We will start by applying some exclusion criteria. The following inclusion criteria apply:
- Inpatient
- Admitted from home
- Age, height, weight available from data
- Surgical specialty is general surgery (or missing)
- Surgery is elective (or missing)
- Patient is not on a preoperative ventilator (or missing)
- No disseminated cancer (or missing)
- No preoperative wound infection (or missing)
- No preoperative sepsis (or missing)
- Surgery is not an emergency (or missing)
- ASA class < 5
- Operative time > 20 minutes
- Gastrojejunosotomy or duodenojejunostomy not performed (or missing)
- M stage M0/Mx (or missing)
- Reconstruction not performed (or missing)
- Is not missing both malignant and benign histologic data
- Does not have histologic diagnoses of ampullary carcinoma, distal cholangiocarcinoma, or duodenal carcinoma.
df <- df %>% dplyr::filter( inout, transt == 2, !is.na(age), !is.na(height), !is.na(weight), (surgspec == 2 | is.na(surgspec)), (electsurg | is.na(electsurg)), (!ventilat | is.na(ventilat)), (!discancr | is.na(discancr)), (!wndinf | is.na(wndinf)), (!prsepis | is.na(prsepis)), (!emergncy | is.na(emergncy)), asaclas < 5, optime > 20, (is.na(pan_gastduo) | pan_gastduo == 3), (pan_mstage != 2 | is.na(pan_mstage)), (pan_reconstruction == 4 | is.na(pan_reconstruction)), !(is.na(pan_malig_histologic) & is.na(pan_benign_histologic)), ((pan_malig_histologic != 1 & pan_malig_histologic != 3 & pan_malig_histologic != 4) | is.na(pan_malig_histologic)) )
I will repeat the analysis I completed previously.
df$pan_approach <- factor(df$pan_approach, levels = c(1,2,3,4,5,6,7,8), labels = c("Hybrid","Laparoscopic","NOTES","Open (planned)", "Other","Other MIS","Robotic","SILS")) df <- df %>% dplyr::filter(pan_approach %in% c("Laparoscopic","Robotic","Open (planned)"), !pan_open_assist) %>% dplyr::mutate(group = pan_approach %in% c("Robotic", "Laparoscopic")) %>% dplyr::mutate(subgroup = ((pan_approach %in% c("Robotic", "Laparoscopic")) + pan_unplanned_conversion))
Next, we will assign a variable to each patient indicating if they have an additional CPT coding for splenectomy:
spleen_vec <- df %>% dplyr::rename(othercpt0 = cpt) %>% dplyr::mutate(dplyr::across(c(dplyr::starts_with("othercpt"), dplyr::starts_with("concpt")), as.character)) %>% tidyr::pivot_longer(cols = c(dplyr::starts_with("othercpt"), dplyr::starts_with("concpt")), values_to = c("cpt"), names_repair = "unique") %>% dplyr::filter(cpt %in% c("38129","38100","38101","38102")) %>% dplyr::pull(caseid) %>% unique() df <- df %>% dplyr::mutate(spleen = dplyr::if_else(caseid %in% spleen_vec, TRUE, FALSE))
df <- df %>% dplyr::mutate(bmi = bmi(height, weight), pan_benign_histologic = !is.na(pan_benign_histologic))
df <- df %>% dplyr::mutate(fnstatus2 = factor(.$fnstatus2, levels = c(1, 2, 3), labels = c("Independent","Partially dependent","Totally dependent")), pan_ductsize = factor(.$pan_ductsize, levels = c(1, 2, 3), labels = c("<3 mm","3-6 mm",">6 mm")), wndclas = factor(wndclas), asaclas = factor(asaclas), pan_glandtext = factor(.$pan_glandtext, levels = c(1, 2, 3), labels = c("Soft","Intermediate","Hard")), pan_malig_histologic = factor(.$pan_malig_histologic, levels = c(2, 5:9), labels = c("Cystadenocarcinoma", "IPMN-invasive", "Neuroendocrine-functioning", "Neuroendocrine-nonfunctioning", "Pancreatic adenocarcinoma", "Other type")), pan_malig_histologic = forcats::fct_explicit_na(pan_malig_histologic, "No malignancy"), pan_tstage = factor(.$pan_tstage, levels = c(1:7), labels = c("T0","T1","T2","T3","T4","Tis","Tx")), pan_tstage = forcats::fct_explicit_na(pan_tstage, "No T stage"), pan_nstage = factor(.$pan_nstage, levels = c(1:3), labels = c("N0","N1","Nx")), pan_nstage = forcats::fct_explicit_na(pan_nstage, "No N stage"), death = !is.na(yrdeath) ) %>% dplyr::mutate(dplyr::across(c("diabetes","smoke","hxcopd","ascites","hxchf","hypermed","dialysis","steroid","wtloss","bleeddis","pan_chemo","pan_radio", "pan_drains"), ~tidyr::replace_na(.x, FALSE))) Hmisc::label(df$sex) = "Male" Hmisc::label(df$age) = "Age" Hmisc::label(df$bmi) = "BMI" Hmisc::label(df$diabetes) = "Diabetes" Hmisc::label(df$smoke) = "Smoking" Hmisc::label(df$fnstatus2) = "Functional Status" Hmisc::label(df$hxcopd) = "COPD" Hmisc::label(df$ascites) = "Ascites" Hmisc::label(df$hxchf) = "CHF" Hmisc::label(df$hypermed) = "Hypertension medication" Hmisc::label(df$dialysis) = "Dialysis" Hmisc::label(df$steroid) = "Steroids" Hmisc::label(df$wtloss) = "Weight loss" Hmisc::label(df$bleeddis) = "Bleeding disorder" Hmisc::label(df$spleen) = "Splenectomy" Hmisc::label(df$wndclas) = "Wound class" Hmisc::label(df$asaclas) = "ASA class" Hmisc::label(df$pan_chemo) = "Chemotherapy" Hmisc::label(df$pan_radio) = "Radiotherapy" Hmisc::label(df$pan_ductsize) = "Duct size" Hmisc::label(df$pan_glandtext) = "Gland texture" Hmisc::label(df$pan_drains) = "Drains" Hmisc::label(df$pan_malig_histologic) = "Malignant histology" Hmisc::label(df$pan_benign_histologic) = "Benign histology" Hmisc::label(df$pan_tstage) = "T stage" Hmisc::label(df$pan_nstage) = "N stage" Hmisc::label(df$supinfec) = "Superficial SSI" Hmisc::label(df$wndinfd) = "Deep SSI" Hmisc::label(df$orgspcssi) = "Organ Space SSI" Hmisc::label(df$pulembol) = "Pulmonary embolism" Hmisc::label(df$reintub) = "Reintubation" Hmisc::label(df$failwean) = "Failure to wean" Hmisc::label(df$renainsf) = "Renal insufficiency" Hmisc::label(df$oprenafl) = "Renal failure" Hmisc::label(df$death) = "Death"
We can now perform a match on the following variables:
- Sex
- Age
- BMI
- Diabetes
- Smoking
- Functional status
- History of COPD
- Ascites
- History of CHF
- Use of hypertension medications
- Dialysis
- Pre*operative steroids
- Weight loss > 10% within 6 months
- Bleeding disorders
- Concurrent splenectomy
- Wound class
- ASA class
- Pre-operative radio- and chemotherapy
- Duct size
- Gland texture
- Presence of drains
- Malignant histology
- T stage
- N stage
- The presence of benign histology
I will make the assumption for the following variables that missing data is equivalent to FALSE:
- Diabetes
- Smoking
- History of COPD
- Ascites
- History of CHF
- Hypertension medication use
- Dialysis
- Steroid use
- Weight loss
- Bleeding disorders
- Pre-operative radio- and chemotherapy
- Drain use
This is a big assumption and may be something you want to change.
With duct size and gland texture
matchdf <- df %>% dplyr::select(caseid, group, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio, pan_ductsize, pan_glandtext, pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage)
Many of the patients are missing data in the duct size and gland texture. Unfortunately, matching requires complete data, so I will first run a match using duct size and gland texture then a larger match without those variables.
match1 <- matchdf %>% dplyr::filter(complete.cases(.)) m1.data <- MatchIt::matchit(group ~ sex + age + bmi + diabetes + smoke + fnstatus2 + hxcopd + ascites + hxchf + hypermed + dialysis + steroid + wtloss + bleeddis + spleen + wndclas + asaclas + pan_chemo + pan_radio + pan_ductsize + pan_glandtext + pan_drains + pan_malig_histologic + pan_benign_histologic + pan_tstage + pan_nstage, data = match1) %>% MatchIt::match.data()
Note that for the tables below, "group" refers to MIS or open. "Yes" refers to the MIS group, "No" refers to the open group. I can format this better for the final document of course.
suppressWarnings(nsqipBileSpill::tableone(m1.data, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio, pan_ductsize, pan_glandtext, pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage, group = group))
m1.data <- dplyr::left_join(m1.data, df)
And we can now compare the outcomes:
suppressWarnings(nsqipBileSpill::tableone(m1.data, supinfec, wndinfd, orgspcssi, pulembol, reintub, failwean, renainsf, oprenafl, death, group = group))
Without duct size and gland texture
matchdf2 <- df %>% dplyr::select(caseid, group, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio, pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage)
match2 <- matchdf2 %>% dplyr::filter(complete.cases(.)) m2.data <- MatchIt::matchit(group ~ sex + age + bmi + diabetes + smoke + fnstatus2 + hxcopd + ascites + hxchf + hypermed + dialysis + steroid + wtloss + bleeddis + spleen + wndclas + asaclas + pan_chemo + pan_radio + pan_drains + pan_malig_histologic + pan_benign_histologic + pan_tstage + pan_nstage, data = match2, method = "full") %>% MatchIt::match.data()
suppressWarnings(nsqipBileSpill::tableone(m2.data, sex, age, bmi, diabetes, smoke, fnstatus2, hxcopd, ascites, hxchf, hypermed, dialysis, steroid, wtloss, bleeddis, spleen, wndclas, asaclas, pan_chemo, pan_radio, pan_drains, pan_malig_histologic, pan_benign_histologic, pan_tstage, pan_nstage, group = group))
m2.data <- dplyr::left_join(m2.data, df)
And we can now compare the outcomes:
suppressWarnings(nsqipBileSpill::tableone(m2.data, supinfec, wndinfd, orgspcssi, pulembol, reintub, failwean, renainsf, oprenafl, death, group = group))
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.