Project Overview
Research Question
Post-mastectomy radiation therapy (PMRT) and breast reconstruction
constitute two important pillars in the treatment of patients with
breast cancer. While PMRT significantly improves survival and reduces
recurrence, it is associated with increased post-operative complications
and decreased patient-reported outcomes following breast reconstruction.
Strategies aimed at optimizing the survival benefits of PMRT and the
quality of life benefits of breast reconstruction are thus a top
priority for those caring for patients with breast cancer.
Proton therapy is an emerging PMRT modality that may offer such a
balance…“A type of radiation treatment called proton beam radiation
therapy may be safer and just as effective as traditional radiation
therapy for adults with advanced cancer. That finding comes from a study
that used existing patient data to compare the two types of
radiation….Traditional radiation delivers x-rays, or beams of photons,
to the tumor and beyond it. This can damage nearby healthy tissues and
can cause significant side effects. By contrast, proton therapy delivers
a beam of proton particles that stops at the tumor, so it’s less likely
to damage nearby healthy tissues….In the new study, patients treated
with proton therapy were much less likely to experience severe side
effects than patients treated with traditional radiation therapy. There
was no difference in how long the patients lived, however. The results
were published December 26 in JAMA Oncology.” (“Is Proton Therapy Safer
than Traditional Radiation?”, National Cancer Institute)
In this study, we set out to determine if surgical complications
following breast reconstruction were reduced in patients who received
proton PMRT compared to those who received traditional photon PMRT.
Note to self: Consider confounding by indication. Are women who
receive proton therapy different to women who receive photon therapy in
other ways than therapy type alone? What potential confounders are
available in dataset to adjust for?
Methods
A retrospective chart review of consecutive patients who underwent
mastectomies at a single institution was performed by the first-author.
Inclusion criteria included receipt of PMRT and implant-based breast
reconstruction (either staged with the use of a tissue expander [TE] or
direct-to-implant [DTI]). Data collected included patient
characteristics (age, BMI, diabetes, smoking), operative details (type
of implant used, unilateral vs bilateral reconstruction), radiation
details (type of radiation, total dose, total rounds). Outcome variables
included seroma, hematoma, mastectomy skin flap necrosis (MSFN), minor
infection (treated outpatient), major infection (requiring readmission
or reoperation), and implant failure defined as removal of the implant.
Implant failure was further stratified into any-cause and
infection-specific implant failure. Patients who received both proton
and photon radiation, did not complete their PMRT, had incomplete
reconstruction or radiation data, or for whom follow-up time was less
than … at the time of chart review were excluded from the study.
\(~\) \(~\)
\(~\) \(~\)
Analysis
Importing the Data
Data from an excel spreadsheet is read in and copied to the
dataframe, df. Data type is then confirmed for all character, factor,
and date variables.
# Read in the data
df <- read_excel("/Users/skebede/Downloads/dataset_cut_10.10.23.xlsx")
# Ensure that all the data types are correct
df <- df %>%
mutate(
# Converting character columns
Name = as.character(Name),
MRN = as.character(MRN),
FIN = as.character(FIN),
# Converting continuous columns
Age = as.numeric(Age),
BMI = as.numeric(BMI),
XRT_dose = as.numeric(XRT_dose),
XRT_rounds = as.numeric(XRT_rounds),
# Converting categorical columns to factors
DM = factor(DM),
Smoking = factor(Smoking),
XRT_type = factor(XRT_type),
XRT_side = factor(XRT_side),
Recon_type = factor(Recon_type),
Recon_laterality = factor(Recon_laterality),
seroma = factor(seroma),
hematoma = factor(hematoma),
infxn_minor = factor(infxn_minor),
infxn_major = factor(infxn_major),
fail_any = factor(fail_any),
fail_infxn = factor(fail_infxn),
# Converting date columns
DOB = ymd(DOB),
surg_date = ymd(surg_date),
XRT_start = ymd(XRT_start),
XRT_end = ymd(XRT_end)
)
\(~\)
\(~\)
Cleaning the Data
Variable names are change for consistency and ease of. Variable
labels are also created to improve readability.
# Rename variables
df <- df %>%
rename(
name = Name,
age = Age,
bmi = BMI,
diabetes = DM,
smoking = Smoking,
xrt_type = XRT_type,
xrt_side = XRT_side,
xrt_dose = XRT_dose,
xrt_rounds = XRT_rounds,
xrt_start = XRT_start,
xrt_end = XRT_end,
recon_type = Recon_type,
recon_laterality = Recon_laterality,
msfn = MSFN
)
# Add variable labels
label(df$age) <- "Age"
label(df$bmi) <- "BMI"
label(df$diabetes) <- "Diabetes"
label(df$smoking) <- "Smoking"
label(df$surg_date) <- "Date of Surgery"
label(df$xrt_type) <- "Type of Radiation"
label(df$xrt_side) <- "Radiation Side"
label(df$xrt_dose) <- "Total Dose of Radiation"
label(df$xrt_rounds) <- "Total Rounds of Radiaton"
label(df$xrt_start) <- "Radiation Start Date"
label(df$xrt_end) <- "Radiation End Date"
label(df$recon_type) <- "Type of Reconstruction"
label(df$recon_laterality) <- "Reconstruction Laterality"
label(df$seroma) <- "Seroma"
label(df$hematoma) <- "Hematoma"
label(df$msfn) <- "MSFN"
label(df$infxn_minor) <- "Minor Infection"
label(df$infxn_major) <- "Major Infection"
label(df$fail_any) <- "Any-Cause Reconstruction Failure"
label(df$fail_infxn) <- "Infection-Specific Reconstruction Failure"
\(~\)
\(~\)
Descriptive Statistics
Descriptive statistics for all patients (Table 1a) and stratified by
type of radiation (Table 1b).
# All Patients
table_1a <- df %>%
select(age, bmi, diabetes, smoking,
recon_type, recon_laterality,
xrt_side, xrt_dose, xrt_rounds, xrt_type,
seroma, hematoma, msfn, infxn_minor, infxn_major, fail_any, fail_infxn) %>%
tbl_summary(type = all_dichotomous() ~ "categorical") %>%
as_gt() %>%
tab_header(title = md("Table 1a: Descriptive Statistics")) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden",
table.border.bottom.style = "hidden",
heading.title.font.size = "medium")
table_1a
| Table 1a: Descriptive Statistics |
| Characteristic |
N = 81 |
| Age |
55 (49, 63) |
| BMI |
26.9 (22.8, 29.5) |
| Diabetes |
|
| No |
79 (98%) |
| Yes |
2 (2.5%) |
| Smoking |
|
| No |
75 (93%) |
| Yes |
6 (7.4%) |
| Type of Reconstruction |
|
| DTI |
43 (53%) |
| TE |
38 (47%) |
| Reconstruction Laterality |
|
| Bilateral |
50 (62%) |
| Unilateral |
31 (38%) |
| Radiation Side |
|
| Left |
42 (52%) |
| Right |
39 (48%) |
| Total Dose of Radiation |
60 (50, 64) |
| Total Rounds of Radiaton |
30.0 (25.0, 32.0) |
| Type of Radiation |
|
| Photon |
71 (88%) |
| Proton |
10 (12%) |
| Seroma |
|
| No |
72 (89%) |
| Yes |
9 (11%) |
| Hematoma |
|
| No |
80 (99%) |
| Yes |
1 (1.2%) |
| MSFN |
|
| No |
76 (94%) |
| Yes |
5 (6.2%) |
| Minor Infection |
|
| No |
73 (90%) |
| Yes |
8 (9.9%) |
| Major Infection |
|
| No |
74 (91%) |
| Yes |
7 (8.6%) |
| Any-Cause Reconstruction Failure |
|
| No |
63 (78%) |
| Yes |
18 (22%) |
| Infection-Specific Reconstruction Failure |
|
| No |
74 (91%) |
| Yes |
7 (8.6%) |
# Stratified Table
table_1b <- df %>%
select(age, bmi, diabetes, smoking, recon_type, recon_laterality, xrt_side, xrt_dose, xrt_rounds, xrt_type,
seroma, hematoma, msfn, infxn_minor, infxn_major, fail_any, fail_infxn) %>%
tbl_summary(by = xrt_type,
type = all_dichotomous() ~ "categorical") %>%
add_p() %>%
bold_p() %>%
as_gt() %>%
tab_header(title = md("Table 1b: Stratified Descriptive Statistics")) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden",
table.border.bottom.style = "hidden",
heading.title.font.size = "medium")
table_1b
| Table 1b: Stratified Descriptive Statistics |
| Characteristic |
Photon, N = 71 |
Proton, N = 10 |
p-value |
| Age |
55 (50, 63) |
53 (40, 63) |
0.6 |
| BMI |
27.1 (22.9, 30.1) |
25.4 (23.2, 27.3) |
0.4 |
| Diabetes |
|
|
>0.9 |
| No |
69 (97%) |
10 (100%) |
|
| Yes |
2 (2.8%) |
0 (0%) |
|
| Smoking |
|
|
>0.9 |
| No |
65 (92%) |
10 (100%) |
|
| Yes |
6 (8.5%) |
0 (0%) |
|
| Type of Reconstruction |
|
|
>0.9 |
| DTI |
38 (54%) |
5 (50%) |
|
| TE |
33 (46%) |
5 (50%) |
|
| Reconstruction Laterality |
|
|
0.7 |
| Bilateral |
43 (61%) |
7 (70%) |
|
| Unilateral |
28 (39%) |
3 (30%) |
|
| Radiation Side |
|
|
0.015 |
| Left |
33 (46%) |
9 (90%) |
|
| Right |
38 (54%) |
1 (10%) |
|
| Total Dose of Radiation |
60 (51, 64) |
50 (50, 50) |
0.009 |
| Total Rounds of Radiaton |
30.0 (25.0, 32.0) |
25.0 (25.0, 25.0) |
0.018 |
| Seroma |
|
|
0.077 |
| No |
65 (92%) |
7 (70%) |
|
| Yes |
6 (8.5%) |
3 (30%) |
|
| Hematoma |
|
|
>0.9 |
| No |
70 (99%) |
10 (100%) |
|
| Yes |
1 (1.4%) |
0 (0%) |
|
| MSFN |
|
|
>0.9 |
| No |
66 (93%) |
10 (100%) |
|
| Yes |
5 (7.0%) |
0 (0%) |
|
| Minor Infection |
|
|
>0.9 |
| No |
64 (90%) |
9 (90%) |
|
| Yes |
7 (9.9%) |
1 (10%) |
|
| Major Infection |
|
|
0.2 |
| No |
66 (93%) |
8 (80%) |
|
| Yes |
5 (7.0%) |
2 (20%) |
|
| Any-Cause Reconstruction Failure |
|
|
0.2 |
| No |
57 (80%) |
6 (60%) |
|
| Yes |
14 (20%) |
4 (40%) |
|
| Infection-Specific Reconstruction Failure |
|
|
0.2 |
| No |
66 (93%) |
8 (80%) |
|
| Yes |
5 (7.0%) |
2 (20%) |
|
\(~\)
\(~\)
Data Visualization
Simple histograms to visualize distributions.
# Histogram for Continuous Variables
ggplot(df, aes(x = age)) +
geom_histogram(bins = 30, fill = "cyan", color = "black") +
theme_minimal() +
labs(title = "Age Distribution", x = "Age", y = "Frequency")

ggplot(df, aes(x = bmi)) +
geom_histogram(bins = 30, fill = "green", color = "black") +
theme_minimal() +
labs(title = "BMI Distribution", x = "BMI", y = "Frequency")

ggplot(df, aes(x = xrt_rounds)) +
geom_histogram(bins = 30, fill = "yellow", color = "black") +
theme_minimal() +
labs(title = "Total Rounds Distribution", x = "Rounds", y = "Frequency")

ggplot(df, aes(x = xrt_dose)) +
geom_histogram(bins = 30, fill = "orange", color = "black") +
theme_minimal() +
labs(title = "Total Dose Distribution", x = "Dose", y = "Frequency")

\(~\)
\(~\)
Bivariate Analyses
Perform bivariate analyses to see the unadjusted relationship between
each predictor and the outcomes.
\(~\)
Chi-Square
# Define categorical variables and outcome variables
categorical_variables <- c("diabetes", "smoking", "recon_type", "recon_laterality", "xrt_side", "xrt_type")
outcome_variables <- c("seroma", "hematoma", "msfn", "infxn_minor", "infxn_major", "fail_any", "fail_infxn")
# Initialize a list to store results
results_chi <- list()
# Loop through each outcome and categorical variable
for (outcome in outcome_variables) {
p_values_chi <- numeric(length(categorical_variables))
for (i in seq_along(categorical_variables)) {
test_result_chi <- chisq.test(table(df[[categorical_variables[i]]], df[[outcome]]))
p_values_chi[i] <- ifelse(is.na(test_result_chi$p.value), NA, test_result_chi$p.value)
}
# Store the results for each outcome in the list
results_chi[[outcome]] <- data.frame(Variable = categorical_variables, P_Value = p_values_chi)
}
# Combine the results into a single data frame
combined_results_chi <- bind_rows(results_chi, .id = "Outcome") %>%
pivot_wider(names_from = Outcome, values_from = P_Value)
# Replace variable names with their labels
combined_results_chi$Variable <- sapply(combined_results_chi$Variable, function(v) {
label(df[[v]]) %||% v})
colnames(combined_results_chi)[-1] <- sapply(colnames(combined_results_chi)[-1], function(v) {
label(df[[v]]) %||% v})
# Create a gt table with variable labels
gt_table_chi <- gt(combined_results_chi) %>%
tab_header(title = "P-Values for Chi-Square Tests") %>%
cols_label(Variable = "") %>%
fmt_number(columns = c(-Variable), decimals = 3) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden",
table.border.bottom.style = "hidden")
# Print the gt table
gt_table_chi
| P-Values for Chi-Square Tests |
|
Seroma |
Hematoma |
MSFN |
Minor Infection |
Major Infection |
Any-Cause Reconstruction Failure |
Infection-Specific Reconstruction Failure |
| Diabetes |
1.000 |
1.000 |
0.000 |
0.468 |
1.000 |
0.069 |
1.000 |
| Smoking |
1.000 |
1.000 |
1.000 |
1.000 |
1.000 |
0.234 |
1.000 |
| Type of Reconstruction |
0.365 |
1.000 |
0.434 |
0.577 |
0.864 |
1.000 |
0.864 |
| Reconstruction Laterality |
0.443 |
1.000 |
0.132 |
1.000 |
0.884 |
0.831 |
0.884 |
| Radiation Side |
0.555 |
0.970 |
0.402 |
0.793 |
0.491 |
0.533 |
0.491 |
| Type of Radiation |
0.136 |
1.000 |
0.869 |
1.000 |
0.445 |
0.299 |
0.445 |
\(~\)
T-test
# t-tests for continuous variables
continuous_variables <- c("age", "bmi", "xrt_dose", "xrt_rounds")
# Initialize a list to store results
results_ttest <- list()
# Define a minimum number of observations required per group
min_obs <- 5
# Loop through each outcome and continuous variable
for (outcome in outcome_variables) {
p_values_ttest <- numeric(length(continuous_variables))
for (i in seq_along(continuous_variables)) {
if(all(table(df[[outcome]]) >= min_obs)) {
test_result_ttest <- t.test(reformulate(outcome, response = continuous_variables[i]), data = df)
p_values_ttest[i] <- test_result_ttest$p.value
} else {
p_values_ttest[i] <- NA # Not enough data to perform the test
}
}
# Store the results for each outcome in the list
results_ttest[[outcome]] <- data.frame(Variable = continuous_variables, P_Value = p_values_ttest)
}
# Combine the results into a single data frame
combined_results_ttest <- bind_rows(results_ttest, .id = "Outcome") %>%
pivot_wider(names_from = Outcome, values_from = P_Value)
# Replace variable names with their labels
combined_results_ttest$Variable <- sapply(combined_results_ttest$Variable, function(v) {
label(df[[v]]) %||% v })
colnames(combined_results_ttest)[-1] <- sapply(colnames(combined_results_ttest)[-1], function(v) {
label(df[[v]]) %||% v })
# Create a gt table with variable labels
gt_table_ttest <- gt(combined_results_ttest) %>%
tab_header(title = "P-Values for T-Tests") %>%
cols_label(Variable = "") %>%
fmt_number(columns = c(-Variable), decimals = 3) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden",
table.border.bottom.style = "hidden")
# Print the gt table
gt_table_ttest
| P-Values for T-Tests |
|
Seroma |
Hematoma |
MSFN |
Minor Infection |
Major Infection |
Any-Cause Reconstruction Failure |
Infection-Specific Reconstruction Failure |
| Age |
0.810 |
NA |
0.317 |
0.216 |
0.915 |
0.670 |
0.511 |
| BMI |
0.530 |
NA |
0.103 |
0.846 |
0.277 |
0.386 |
0.745 |
| Total Dose of Radiation |
0.081 |
NA |
0.842 |
0.222 |
0.410 |
0.384 |
0.274 |
| Total Rounds of Radiaton |
0.108 |
NA |
0.748 |
0.179 |
0.846 |
0.796 |
0.720 |
\(~\)
\(~\)
Multicolinearity Assessments
Assessment of multicolinearity for each outcome variable.
# Fit a model for the outcome 'seroma'
model_seroma <- glm(seroma ~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_dose + xrt_rounds + xrt_type, data = df, family = binomial())
# Calculate VIF
vif_values_seroma <- vif(model_seroma)
# Create a data frame with VIF values
vif_df_seroma <- data.frame(
Variable = names(vif_values_seroma),
VIF = vif_values_seroma,
HighVIF =ifelse(vif_values_seroma >= 5, "Yes", "")
)
# Replace variable names in 'Variable' column with their labels
vif_df_seroma$Variable <- sapply(vif_df_seroma$Variable, function(v) {
label(df[[v]]) %||% v # Use the label if it exists, otherwise use the variable name
})
# Create a gt table with VIF values
gt_table_seroma <- gt(vif_df_seroma) %>%
tab_header(title = "Variance Inflation Factor (VIF) for Seroma") %>%
cols_label(Variable = "Coefficient", VIF = "VIF", HighVIF = "High Multicollinearity") %>%
fmt_number(columns = VIF, decimals = 2) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden", table.border.bottom.style = "hidden")
# Print the gt table
gt_table_seroma
| Variance Inflation Factor (VIF) for Seroma |
| Coefficient |
VIF |
High Multicollinearity |
| Age |
1.27 |
|
| BMI |
1.43 |
|
| Diabetes |
1.00 |
|
| Smoking |
1.65 |
|
| Type of Reconstruction |
1.23 |
|
| Reconstruction Laterality |
1.25 |
|
| Radiation Side |
1.45 |
|
| Total Dose of Radiation |
19.59 |
Yes |
| Total Rounds of Radiaton |
21.16 |
Yes |
| Type of Radiation |
1.71 |
|
# Fit a model for the outcome 'hematoma'
model_hematoma <- glm(hematoma ~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_dose + xrt_rounds + xrt_type, data = df, family = binomial())
# Calculate VIF
vif_values_hematoma <- vif(model_hematoma)
# Create a data frame with VIF values
vif_df_hematoma <- data.frame(
Variable = names(vif_values_hematoma),
VIF = vif_values_hematoma,
HighVIF =ifelse(vif_values_hematoma >= 5, "Yes", "")
)
# Replace variable names in 'Variable' column with their labels
vif_df_hematoma$Variable <- sapply(vif_df_hematoma$Variable, function(v) {
label(df[[v]]) %||% v # Use the label if it exists, otherwise use the variable name
})
# Create a gt table with VIF values
gt_table_hematoma <- gt(vif_df_hematoma) %>%
tab_header(title = "Variance Inflation Factor (VIF) for Hematoma") %>%
cols_label(Variable = "Coefficient", VIF = "VIF", HighVIF = "High Multicollinearity") %>%
fmt_number(columns = VIF, decimals = 2) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden", table.border.bottom.style = "hidden")
# Print the gt table
gt_table_hematoma
| Variance Inflation Factor (VIF) for Hematoma |
| Coefficient |
VIF |
High Multicollinearity |
| Age |
525.26 |
Yes |
| BMI |
1,039.33 |
Yes |
| Diabetes |
1.59 |
|
| Smoking |
1.10 |
|
| Type of Reconstruction |
51.62 |
Yes |
| Reconstruction Laterality |
56.11 |
Yes |
| Radiation Side |
11.14 |
Yes |
| Total Dose of Radiation |
48.91 |
Yes |
| Total Rounds of Radiaton |
438.21 |
Yes |
| Type of Radiation |
26.65 |
Yes |
# Fit a model for the outcome 'hematoma'
model_infxn_minor <- glm(infxn_minor ~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_dose + xrt_rounds + xrt_type, data = df, family = binomial())
# Calculate VIF
vif_values_infxn_minor <- vif(model_infxn_minor)
# Create a data frame with VIF values
vif_df_infxn_minor <- data.frame(
Variable = names(vif_values_infxn_minor),
VIF = vif_values_infxn_minor,
HighVIF =ifelse(vif_values_infxn_minor >= 5, "Yes", "")
)
# Replace variable names in 'Variable' column with their labels
vif_df_infxn_minor$Variable <- sapply(vif_df_infxn_minor$Variable, function(v) {
label(df[[v]]) %||% v # Use the label if it exists, otherwise use the variable name
})
# Create a gt table with VIF values
gt_table_infxn_minor <- gt(vif_df_infxn_minor) %>%
tab_header(title = "Variance Inflation Factor (VIF) for Minor Infection") %>%
cols_label(Variable = "Coefficient", VIF = "VIF", HighVIF = "High Multicollinearity") %>%
fmt_number(columns = VIF, decimals = 2) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden", table.border.bottom.style = "hidden")
# Print the gt table
gt_table_infxn_minor
| Variance Inflation Factor (VIF) for Minor Infection |
| Coefficient |
VIF |
High Multicollinearity |
| Age |
1.64 |
|
| BMI |
1.34 |
|
| Diabetes |
1.57 |
|
| Smoking |
1.43 |
|
| Type of Reconstruction |
1.29 |
|
| Reconstruction Laterality |
1.47 |
|
| Radiation Side |
1.45 |
|
| Total Dose of Radiation |
37.46 |
Yes |
| Total Rounds of Radiaton |
38.33 |
Yes |
| Type of Radiation |
1.61 |
|
# Fit a model for the outcome 'hematoma'
model_infxn_major <- glm(infxn_major ~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_dose + xrt_rounds + xrt_type, data = df, family = binomial())
# Calculate VIF
vif_values_infxn_major <- vif(model_infxn_major)
# Create a data frame with VIF values
vif_df_infxn_major <- data.frame(
Variable = names(vif_values_infxn_major),
VIF = vif_values_infxn_major,
HighVIF =ifelse(vif_values_infxn_major >= 5, "Yes", "")
)
# Replace variable names in 'Variable' column with their labels
vif_df_infxn_major$Variable <- sapply(vif_df_infxn_major$Variable, function(v) {
label(df[[v]]) %||% v # Use the label if it exists, otherwise use the variable name
})
# Create a gt table with VIF values
gt_table_infxn_major <- gt(vif_df_infxn_major) %>%
tab_header(title = "Variance Inflation Factor (VIF) for Major Infection") %>%
cols_label(Variable = "Coefficient", VIF = "VIF", HighVIF = "High Multicollinearity") %>%
fmt_number(columns = VIF, decimals = 2) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden", table.border.bottom.style = "hidden")
# Print the gt table
gt_table_infxn_major
| Variance Inflation Factor (VIF) for Major Infection |
| Coefficient |
VIF |
High Multicollinearity |
| Age |
1.32 |
|
| BMI |
1.34 |
|
| Diabetes |
1.00 |
|
| Smoking |
1.35 |
|
| Type of Reconstruction |
1.30 |
|
| Reconstruction Laterality |
1.22 |
|
| Radiation Side |
1.39 |
|
| Total Dose of Radiation |
5.76 |
Yes |
| Total Rounds of Radiaton |
5.48 |
Yes |
| Type of Radiation |
1.20 |
|
# Fit a model for the outcome 'hematoma'
model_fail_any <- glm(fail_any ~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_dose + xrt_rounds + xrt_type, data = df, family = binomial())
# Calculate VIF
vif_values_fail_any<- vif(model_fail_any)
# Create a data frame with VIF values
vif_df_fail_any <- data.frame(
Variable = names(vif_values_fail_any),
VIF = vif_values_fail_any,
HighVIF =ifelse(vif_values_fail_any >= 5, "Yes", "")
)
# Replace variable names in 'Variable' column with their labels
vif_df_fail_any$Variable <- sapply(vif_df_fail_any$Variable, function(v) {
label(df[[v]]) %||% v # Use the label if it exists, otherwise use the variable name
})
# Create a gt table with VIF values
gt_table_fail_any <- gt(vif_df_fail_any) %>%
tab_header(title = "Variance Inflation Factor (VIF) for Any-Cause Reconstruction Failure") %>%
cols_label(Variable = "Coefficient", VIF = "VIF", HighVIF = "High Multicollinearity") %>%
fmt_number(columns = VIF, decimals = 2) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden", table.border.bottom.style = "hidden")
# Print the gt table
gt_table_fail_any
| Variance Inflation Factor (VIF) for Any-Cause Reconstruction Failure |
| Coefficient |
VIF |
High Multicollinearity |
| Age |
1.20 |
|
| BMI |
1.35 |
|
| Diabetes |
1.00 |
|
| Smoking |
1.31 |
|
| Type of Reconstruction |
1.24 |
|
| Reconstruction Laterality |
1.21 |
|
| Radiation Side |
1.29 |
|
| Total Dose of Radiation |
4.82 |
|
| Total Rounds of Radiaton |
4.98 |
|
| Type of Radiation |
1.28 |
|
# Fit a model for the outcome 'hematoma'
model_fail_infxn <- glm(fail_infxn ~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_dose + xrt_rounds + xrt_type, data = df, family = binomial())
# Calculate VIF
vif_values_fail_infxn<- vif(model_fail_infxn)
# Create a data frame with VIF values
vif_df_fail_infxn <- data.frame(
Variable = names(vif_values_fail_infxn),
VIF = vif_values_fail_infxn,
HighVIF =ifelse(vif_values_fail_infxn >= 5, "Yes", "")
)
# Replace variable names in 'Variable' column with their labels
vif_df_fail_infxn$Variable <- sapply(vif_df_fail_infxn$Variable, function(v) {
label(df[[v]]) %||% v # Use the label if it exists, otherwise use the variable name
})
# Create a gt table with VIF values
gt_table_fail_infxn <- gt(vif_df_fail_infxn) %>%
tab_header(title = "Variance Inflation Factor (VIF) for Infection-Specific Reconstruction Failure") %>%
cols_label(Variable = "Coefficient", VIF = "VIF", HighVIF = "High Multicollinearity") %>%
fmt_number(columns = VIF, decimals = 2) %>%
opt_align_table_header(align = "left") %>%
tab_options(table.border.top.style = "hidden", table.border.bottom.style = "hidden")
# Print the gt table
gt_table_fail_infxn
| Variance Inflation Factor (VIF) for Infection-Specific Reconstruction Failure |
| Coefficient |
VIF |
High Multicollinearity |
| Age |
1.45 |
|
| BMI |
1.50 |
|
| Diabetes |
1.00 |
|
| Smoking |
1.69 |
|
| Type of Reconstruction |
1.46 |
|
| Reconstruction Laterality |
1.31 |
|
| Radiation Side |
1.50 |
|
| Total Dose of Radiation |
7.09 |
Yes |
| Total Rounds of Radiaton |
7.19 |
Yes |
| Type of Radiation |
1.28 |
|
\(~\)
\(~\)
Logistic Regression
Sample models for each outcome variable with varying number of
covariates.
Model 1
Includes age, BMI, diabetes, smoking, reconstruction type,
laterality, radiation side, radiation dose, radiation rounds, and
radiation type
# Make new df_model dataframe where outcome variables are coded as 0 / 1
df_model <- df
outcome_variables <- c("seroma", "hematoma", "msfn", "infxn_minor", "infxn_major", "fail_any", "fail_infxn")
for(var in outcome_variables) {
df_model[[var]] <- ifelse(df_model[[var]] == "Yes", 1, 0)
}
# Initialize a list to store tables
tables_list1 <- list()
# Function to create a title with variable labels
create_title <- function(var) {
label_text <- label(df[[var]])
paste("<div style='text-align: left; font-weight: bold; color: grey'> Table 3. Multivariate Statistics for", label_text, "</div>")
}
# Loop to create models and tables
for (var in outcome_variables) {
# Create model
model_formula <- as.formula(paste(var, "~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_dose + xrt_rounds + xrt_type"))
model_name <- paste("m", var, sep = "_")
assign(model_name, glm(model_formula, data = df_model, family = binomial(link = "logit")))
# Create table with labeled title
table_name <- paste("table1", var, sep = "_")
table <- tbl_regression(get(model_name), exponentiate = TRUE) %>%
add_global_p() %>%
modify_caption(create_title(var)) %>%
as_gt() %>%
gt::tab_header("Model 1")
# Store the table in the list
tables_list1[[table_name]] <- table
}
m1_seroma <- tables_list1$table1_seroma
m1_hematoma <- tables_list1$table1_hematoma
m1_msfn <- tables_list1$table1_msfn
m1_minor <- tables_list1$table1_infxn_minor
m1_major <- tables_list1$table1_infxn_major
m1_fail_any <- tables_list1$table1_fail_any
m1_fail_infxn <- tables_list1$table1_fail_infxn
# Print all tables
m1_seroma
Table 3. Multivariate Statistics for Seroma
| Model 1 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.00 |
0.93, 1.08 |
>0.9 |
| BMI |
1.08 |
0.93, 1.23 |
0.3 |
| Diabetes |
|
|
0.5 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.14 |
| No |
— |
— |
|
| Yes |
11.4 |
0.36, 295 |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
3.29 |
0.54, 26.7 |
|
| Reconstruction Laterality |
|
|
0.10 |
| Bilateral |
— |
— |
|
| Unilateral |
4.29 |
0.74, 31.8 |
|
| Radiation Side |
|
|
0.5 |
| Left |
— |
— |
|
| Right |
0.52 |
0.06, 3.99 |
|
| Total Dose of Radiation |
0.93 |
0.73, 1.33 |
0.7 |
| Total Rounds of Radiaton |
0.95 |
0.57, 1.28 |
0.8 |
| Type of Radiation |
|
|
0.075 |
| Photon |
— |
— |
|
| Proton |
7.58 |
0.81, 107 |
|
m1_hematoma
Table 3. Multivariate Statistics for Hematoma
| Model 1 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
0.00 |
0.00, 11,321,943,280,277,891,182,536,295,405,167,251,076,604,521,177,000,159,022,781,590,878,936,560,638,313,585,667,457,805,002,346,923,106,323,644,832,854,252,665,397,768,343,900,454,151,340,666,863,134,862,723,375,453,285,730,843,047,580,783,332,767,306,355,135,912,778,352,798,244,883,752,221,623,545,583,632,384 |
0.009 |
| BMI |
763,020,919,100,322,480,128 |
0.00, Inf |
0.018 |
| Diabetes |
|
|
>0.9 |
| No |
— |
— |
|
| Yes |
Inf |
0.00, Inf |
|
| Smoking |
|
|
>0.9 |
| No |
— |
— |
|
| Yes |
345,824,594,467,641,500,722,553,563,074,054,765,990,592,161,119,700,441,552,830,269,264,160,571,258,158,391,036,496,465,050,206,638,911,547,745,717,731,397,920,293,878,162,517,198,835,392,249,331,712 |
0.00, Inf |
|
| Type of Reconstruction |
|
|
>0.9 |
| DTI |
— |
— |
|
| TE |
0.00 |
0.00, Inf |
|
| Reconstruction Laterality |
|
|
0.026 |
| Bilateral |
— |
— |
|
| Unilateral |
0.00 |
0.00, Inf |
|
| Radiation Side |
|
|
0.011 |
| Left |
— |
— |
|
| Right |
184,980,745,999,421,216,275,465,106,763,029,657,547,962,286,111,409,961,246,738,371,334,380,241,283,932,183,585,170,113,161,987,267,381,751,281,765,086,908,709,705,762,242,861,954,400,751,696,536,928,256 |
0.00, Inf |
|
| Total Dose of Radiation |
18,336,506,379,421,108 |
0.00, Inf |
0.033 |
| Total Rounds of Radiaton |
1,625,409,369,669,316,378,624 |
0.00, Inf |
0.016 |
| Type of Radiation |
|
|
>0.9 |
| Photon |
— |
— |
|
| Proton |
Inf |
0.00, Inf |
|
m1_msfn
Table 3. Multivariate Statistics for MSFN
| Model 1 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
0.93 |
0.66, 1.22 |
0.6 |
| BMI |
1.54 |
1.00, 4.25 |
0.050 |
| Diabetes |
|
|
0.002 |
| No |
— |
— |
|
| Yes |
221,696,305,144,573,020,078,080 |
0.00, NA |
|
| Smoking |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Type of Reconstruction |
|
|
0.7 |
| DTI |
— |
— |
|
| TE |
0.52 |
0.01, 12.6 |
|
| Reconstruction Laterality |
|
|
0.2 |
| Bilateral |
— |
— |
|
| Unilateral |
51.0 |
0.32, 6,810,658 |
|
| Radiation Side |
|
|
0.008 |
| Left |
— |
— |
|
| Right |
0.00 |
|
|
| Total Dose of Radiation |
1.04 |
0.74, 2.22 |
0.9 |
| Total Rounds of Radiaton |
1.06 |
0.37, 1.98 |
0.9 |
| Type of Radiation |
|
|
0.7 |
| Photon |
— |
— |
|
| Proton |
0.00 |
|
|
m1_minor
Table 3. Multivariate Statistics for Minor Infection
| Model 1 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.04 |
0.96, 1.15 |
0.3 |
| BMI |
0.97 |
0.81, 1.10 |
0.7 |
| Diabetes |
|
|
0.10 |
| No |
— |
— |
|
| Yes |
24.4 |
0.52, 1,389 |
|
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.75 |
0.17, 76.1 |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
3.16 |
0.56, 26.2 |
|
| Reconstruction Laterality |
|
|
0.4 |
| Bilateral |
— |
— |
|
| Unilateral |
0.45 |
0.05, 2.83 |
|
| Radiation Side |
|
|
0.2 |
| Left |
— |
— |
|
| Right |
0.31 |
0.04, 1.97 |
|
| Total Dose of Radiation |
1.10 |
0.83, 1.72 |
0.6 |
| Total Rounds of Radiaton |
0.83 |
0.46, 1.20 |
0.5 |
| Type of Radiation |
|
|
>0.9 |
| Photon |
— |
— |
|
| Proton |
1.20 |
0.04, 19.7 |
|
m1_major
Table 3. Multivariate Statistics for Major Infection
| Model 1 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.94, 1.10 |
0.7 |
| BMI |
0.92 |
0.72, 1.11 |
0.4 |
| Diabetes |
|
|
0.8 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.4 |
| No |
— |
— |
|
| Yes |
3.73 |
0.12, 69.5 |
|
| Type of Reconstruction |
|
|
0.4 |
| DTI |
— |
— |
|
| TE |
2.32 |
0.36, 19.8 |
|
| Reconstruction Laterality |
|
|
0.4 |
| Bilateral |
— |
— |
|
| Unilateral |
0.41 |
0.04, 2.77 |
|
| Radiation Side |
|
|
0.2 |
| Left |
— |
— |
|
| Right |
0.28 |
0.02, 2.15 |
|
| Total Dose of Radiation |
0.83 |
0.68, 0.99 |
0.036 |
| Total Rounds of Radiaton |
1.20 |
0.97, 1.47 |
0.085 |
| Type of Radiation |
|
|
0.7 |
| Photon |
— |
— |
|
| Proton |
1.61 |
0.17, 12.4 |
|
m1_fail_any
Table 3. Multivariate Statistics for Any-Cause Reconstruction Failure
| Model 1 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.96, 1.08 |
0.6 |
| BMI |
1.06 |
0.95, 1.17 |
0.3 |
| Diabetes |
|
|
0.005 |
| No |
— |
— |
|
| Yes |
190,225,829 |
0.00, NA |
|
| Smoking |
|
|
0.023 |
| No |
— |
— |
|
| Yes |
10.9 |
1.40, 96.4 |
|
| Type of Reconstruction |
|
|
0.5 |
| DTI |
— |
— |
|
| TE |
1.65 |
0.44, 6.69 |
|
| Reconstruction Laterality |
|
|
0.2 |
| Bilateral |
— |
— |
|
| Unilateral |
0.39 |
0.08, 1.58 |
|
| Radiation Side |
|
|
0.4 |
| Left |
— |
— |
|
| Right |
0.58 |
0.14, 2.31 |
|
| Total Dose of Radiation |
0.90 |
0.78, 1.03 |
0.12 |
| Total Rounds of Radiaton |
1.14 |
0.96, 1.36 |
0.12 |
| Type of Radiation |
|
|
0.2 |
| Photon |
— |
— |
|
| Proton |
3.05 |
0.56, 16.4 |
|
m1_fail_infxn
Table 3. Multivariate Statistics for Infection-Specific Reconstruction Failure
| Model 1 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.04 |
0.97, 1.14 |
0.3 |
| BMI |
1.02 |
0.81, 1.22 |
0.9 |
| Diabetes |
|
|
0.8 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.2 |
| No |
— |
— |
|
| Yes |
10.2 |
0.28, 322 |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
3.64 |
0.51, 40.4 |
|
| Reconstruction Laterality |
|
|
0.3 |
| Bilateral |
— |
— |
|
| Unilateral |
0.34 |
0.03, 2.52 |
|
| Radiation Side |
|
|
0.2 |
| Left |
— |
— |
|
| Right |
0.21 |
0.02, 1.83 |
|
| Total Dose of Radiation |
0.79 |
0.63, 0.95 |
0.014 |
| Total Rounds of Radiaton |
1.26 |
1.02, 1.61 |
0.035 |
| Type of Radiation |
|
|
0.5 |
| Photon |
— |
— |
|
| Proton |
2.13 |
0.21, 19.0 |
|
Model 2
Includes age, BMI, diabetes, smoking, reconstruction type,
laterality, radiation side, radiation type
# Make new df_model dataframe where outcome variables are coded as 0 / 1
df_model <- df
outcome_variables <- c("seroma", "hematoma", "msfn", "infxn_minor", "infxn_major", "fail_any", "fail_infxn")
for(var in outcome_variables) {
df_model[[var]] <- ifelse(df_model[[var]] == "Yes", 1, 0)
}
# Initialize a list to store tables
tables_list2 <- list()
# Function to create a title with variable labels
create_title <- function(var) {
label_text <- label(df[[var]])
paste("<div style='text-align: left; font-weight: bold; color: grey'> Table 3b. Multivariate Statistics for", label_text, "</div>")
}
# Loop to create models and tables
for (var in outcome_variables) {
# Create model
model_formula <- as.formula(paste(var, "~ age + bmi + diabetes + smoking + recon_type + recon_laterality + xrt_side + xrt_type"))
model_name <- paste("m", var, sep = "_")
assign(model_name, glm(model_formula, data = df_model, family = binomial(link = "logit")))
# Create table with labeled title
table_name <- paste("table2", var, sep = "_")
table <- tbl_regression(get(model_name), exponentiate = TRUE) %>%
add_global_p() %>%
modify_caption(create_title(var)) %>%
as_gt() %>%
gt::tab_header("Model 2")
# Store the table in the list
tables_list2[[table_name]] <- table
}
m2_seroma <- tables_list2$table2_seroma
m2_hematoma <- tables_list2$table2_hematoma
m2_msfn <- tables_list2$table2_msfn
m2_minor <- tables_list2$table2_infxn_minor
m2_major <- tables_list2$table2_infxn_major
m2_fail_any <- tables_list2$table2_fail_any
m2_fail_infxn <- tables_list2$table2_fail_infxn
# Print all tables
m2_seroma
Table 3b. Multivariate Statistics for Seroma
| Model 2 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.95, 1.09 |
0.6 |
| BMI |
1.09 |
0.96, 1.22 |
0.15 |
| Diabetes |
|
|
0.5 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.08 |
0.17, 50.2 |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
2.91 |
0.58, 19.0 |
|
| Reconstruction Laterality |
|
|
0.2 |
| Bilateral |
— |
— |
|
| Unilateral |
2.59 |
0.51, 14.9 |
|
| Radiation Side |
|
|
0.7 |
| Left |
— |
— |
|
| Right |
0.73 |
0.11, 4.47 |
|
| Type of Radiation |
|
|
0.052 |
| Photon |
— |
— |
|
| Proton |
7.08 |
0.98, 56.2 |
|
m2_hematoma
Table 3b. Multivariate Statistics for Hematoma
| Model 2 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
0.87 |
0.54, 1.10 |
0.3 |
| BMI |
1.06 |
0.62, 1.52 |
0.7 |
| Diabetes |
|
|
>0.9 |
| No |
— |
— |
|
| Yes |
14.1 |
0.00, Inf |
|
| Smoking |
|
|
>0.9 |
| No |
— |
— |
|
| Yes |
0.53 |
0.00, Inf |
|
| Type of Reconstruction |
|
|
0.3 |
| DTI |
— |
— |
|
| TE |
0.00 |
0.00, Inf |
|
| Reconstruction Laterality |
|
|
0.5 |
| Bilateral |
— |
— |
|
| Unilateral |
0.00 |
|
|
| Radiation Side |
|
|
0.3 |
| Left |
— |
— |
|
| Right |
321,820,125 |
0.00, Inf |
|
| Type of Radiation |
|
|
>0.9 |
| Photon |
— |
— |
|
| Proton |
0.16 |
0.00, Inf |
|
m2_msfn
Table 3b. Multivariate Statistics for MSFN
| Model 2 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
0.95 |
0.73, 1.22 |
0.7 |
| BMI |
1.38 |
0.96, 2.55 |
0.084 |
| Diabetes |
|
|
0.003 |
| No |
— |
— |
|
| Yes |
4,154,329,216,695,318,609,920 |
0.00, NA |
|
| Smoking |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Type of Reconstruction |
|
|
0.8 |
| DTI |
— |
— |
|
| TE |
0.68 |
0.02, 13.4 |
|
| Reconstruction Laterality |
|
|
0.2 |
| Bilateral |
— |
— |
|
| Unilateral |
16.1 |
0.31, 20,037 |
|
| Radiation Side |
|
|
0.013 |
| Left |
— |
— |
|
| Right |
0.00 |
|
|
| Type of Radiation |
|
|
0.6 |
| Photon |
— |
— |
|
| Proton |
0.00 |
|
|
m2_minor
Table 3b. Multivariate Statistics for Minor Infection
| Model 2 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.05 |
0.97, 1.16 |
0.2 |
| BMI |
0.98 |
0.82, 1.11 |
0.8 |
| Diabetes |
|
|
0.082 |
| No |
— |
— |
|
| Yes |
27.9 |
0.62, 1,529 |
|
| Smoking |
|
|
0.4 |
| No |
— |
— |
|
| Yes |
3.62 |
0.14, 52.6 |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
3.28 |
0.60, 26.5 |
|
| Reconstruction Laterality |
|
|
0.3 |
| Bilateral |
— |
— |
|
| Unilateral |
0.38 |
0.04, 2.29 |
|
| Radiation Side |
|
|
0.3 |
| Left |
— |
— |
|
| Right |
0.34 |
0.04, 2.09 |
|
| Type of Radiation |
|
|
>0.9 |
| Photon |
— |
— |
|
| Proton |
0.89 |
0.04, 8.80 |
|
m2_major
Table 3b. Multivariate Statistics for Major Infection
| Model 2 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.03 |
0.96, 1.11 |
0.5 |
| BMI |
0.93 |
0.75, 1.10 |
0.5 |
| Diabetes |
|
|
0.7 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.5 |
| No |
— |
— |
|
| Yes |
2.70 |
0.11, 34.2 |
|
| Type of Reconstruction |
|
|
0.6 |
| DTI |
— |
— |
|
| TE |
1.63 |
0.30, 10.2 |
|
| Reconstruction Laterality |
|
|
0.4 |
| Bilateral |
— |
— |
|
| Unilateral |
0.42 |
0.05, 2.49 |
|
| Radiation Side |
|
|
0.3 |
| Left |
— |
— |
|
| Right |
0.39 |
0.04, 2.69 |
|
| Type of Radiation |
|
|
0.4 |
| Photon |
— |
— |
|
| Proton |
2.38 |
0.25, 18.0 |
|
m2_fail_any
Table 3b. Multivariate Statistics for Any-Cause Reconstruction Failure
| Model 2 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.96, 1.08 |
0.6 |
| BMI |
1.04 |
0.95, 1.15 |
0.4 |
| Diabetes |
|
|
0.006 |
| No |
— |
— |
|
| Yes |
148,464,318 |
0.00, NA |
|
| Smoking |
|
|
0.027 |
| No |
— |
— |
|
| Yes |
9.67 |
1.32, 77.9 |
|
| Type of Reconstruction |
|
|
0.6 |
| DTI |
— |
— |
|
| TE |
1.42 |
0.41, 5.18 |
|
| Reconstruction Laterality |
|
|
0.2 |
| Bilateral |
— |
— |
|
| Unilateral |
0.40 |
0.09, 1.52 |
|
| Radiation Side |
|
|
0.5 |
| Left |
— |
— |
|
| Right |
0.64 |
0.16, 2.46 |
|
| Type of Radiation |
|
|
0.10 |
| Photon |
— |
— |
|
| Proton |
3.87 |
0.75, 19.8 |
|
m2_fail_infxn
Table 3b. Multivariate Statistics for Infection-Specific Reconstruction Failure
| Model 2 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.05 |
0.98, 1.15 |
0.2 |
| BMI |
1.00 |
0.83, 1.14 |
>0.9 |
| Diabetes |
|
|
0.7 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
5.05 |
0.19, 75.3 |
|
| Type of Reconstruction |
|
|
0.5 |
| DTI |
— |
— |
|
| TE |
1.93 |
0.34, 13.1 |
|
| Reconstruction Laterality |
|
|
0.3 |
| Bilateral |
— |
— |
|
| Unilateral |
0.38 |
0.04, 2.44 |
|
| Radiation Side |
|
|
0.3 |
| Left |
— |
— |
|
| Right |
0.35 |
0.04, 2.55 |
|
| Type of Radiation |
|
|
0.4 |
| Photon |
— |
— |
|
| Proton |
2.62 |
0.25, 21.9 |
|
Model 3
Includes age, BMI, diabetes, smoking, reconstruction type, radiation
side, and radiation type
# Make new df_model dataframe where outcome variables are coded as 0 / 1
df_model <- df
outcome_variables <- c("seroma", "hematoma", "msfn", "infxn_minor", "infxn_major", "fail_any", "fail_infxn")
for(var in outcome_variables) {
df_model[[var]] <- ifelse(df_model[[var]] == "Yes", 1, 0)
}
# Initialize a list to store tables
tables_list3 <- list()
# Function to create a title with variable labels
create_title <- function(var) {
label_text <- label(df[[var]])
paste("<div style='text-align: left; font-weight: bold; color: grey'> Table 3. Multivariate Statistics for", label_text, "</div>")
}
# Loop to create models and tables
for (var in outcome_variables) {
# Create model
model_formula <- as.formula(paste(var, "~ age + bmi + diabetes + smoking + recon_type + xrt_side + xrt_type"))
model_name <- paste("m", var, sep = "_")
assign(model_name, glm(model_formula, data = df_model, family = binomial(link = "logit")))
# Create table with labeled title
table_name <- paste("table3", var, sep = "_")
table <- tbl_regression(get(model_name), exponentiate = TRUE) %>%
add_global_p() %>%
modify_caption(create_title(var)) %>%
as_gt() %>%
gt::tab_header("Model 3")
# Store the table in the list
tables_list3[[table_name]] <- table
}
m3_seroma <- tables_list3$table3_seroma
m3_hematoma <- tables_list3$table3_hematoma
m3_msfn <- tables_list3$table3_msfn
m3_minor <- tables_list3$table3_infxn_minor
m3_major <- tables_list3$table3_infxn_major
m3_fail_any <- tables_list3$table3_fail_any
m3_fail_infxn <- tables_list3$table3_fail_infxn
# Print all tables
m3_seroma
Table 3. Multivariate Statistics for Seroma
| Model 3 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.96, 1.09 |
0.5 |
| BMI |
1.08 |
0.96, 1.21 |
0.2 |
| Diabetes |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.43 |
0.18, 53.8 |
|
| Type of Reconstruction |
|
|
0.12 |
| DTI |
— |
— |
|
| TE |
3.59 |
0.74, 23.4 |
|
| Radiation Side |
|
|
0.5 |
| Left |
— |
— |
|
| Right |
0.56 |
0.09, 3.06 |
|
| Type of Radiation |
|
|
0.067 |
| Photon |
— |
— |
|
| Proton |
5.95 |
0.88, 41.2 |
|
m3_hematoma
Table 3. Multivariate Statistics for Hematoma
| Model 3 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
0.85 |
0.52, 1.10 |
0.2 |
| BMI |
1.08 |
0.65, 1.56 |
0.6 |
| Diabetes |
|
|
>0.9 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
>0.9 |
| No |
— |
— |
|
| Yes |
0.35 |
0.00, Inf |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
0.00 |
0.00, Inf |
|
| Radiation Side |
|
|
0.3 |
| Left |
— |
— |
|
| Right |
331,519,241 |
0.00, Inf |
|
| Type of Radiation |
|
|
>0.9 |
| Photon |
— |
— |
|
| Proton |
0.27 |
0.00, Inf |
|
m3_msfn
Table 3. Multivariate Statistics for MSFN
| Model 3 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.03 |
0.87, 1.28 |
0.7 |
| BMI |
1.18 |
0.91, 1.63 |
0.2 |
| Diabetes |
|
|
0.003 |
| No |
— |
— |
|
| Yes |
11,262,384,710,665,029,632 |
0.00, NA |
|
| Smoking |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Type of Reconstruction |
|
|
0.7 |
| DTI |
— |
— |
|
| TE |
0.62 |
0.02, 10.6 |
|
| Radiation Side |
|
|
0.013 |
| Left |
— |
— |
|
| Right |
0.00 |
|
|
| Type of Radiation |
|
|
0.3 |
| Photon |
— |
— |
|
| Proton |
0.00 |
|
|
m3_minor
Table 3. Multivariate Statistics for Minor Infection
| Model 3 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.04 |
0.97, 1.13 |
0.3 |
| BMI |
0.99 |
0.83, 1.12 |
0.9 |
| Diabetes |
|
|
0.13 |
| No |
— |
— |
|
| Yes |
14.8 |
0.40, 612 |
|
| Smoking |
|
|
0.4 |
| No |
— |
— |
|
| Yes |
3.25 |
0.13, 41.3 |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
3.13 |
0.59, 24.3 |
|
| Radiation Side |
|
|
0.4 |
| Left |
— |
— |
|
| Right |
0.49 |
0.08, 2.56 |
|
| Type of Radiation |
|
|
>0.9 |
| Photon |
— |
— |
|
| Proton |
0.95 |
0.04, 9.17 |
|
m3_major
Table 3. Multivariate Statistics for Major Infection
| Model 3 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.95, 1.10 |
0.6 |
| BMI |
0.94 |
0.76, 1.10 |
0.5 |
| Diabetes |
|
|
0.7 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.5 |
| No |
— |
— |
|
| Yes |
2.51 |
0.10, 29.1 |
|
| Type of Reconstruction |
|
|
0.6 |
| DTI |
— |
— |
|
| TE |
1.48 |
0.28, 8.72 |
|
| Radiation Side |
|
|
0.5 |
| Left |
— |
— |
|
| Right |
0.51 |
0.06, 3.21 |
|
| Type of Radiation |
|
|
0.4 |
| Photon |
— |
— |
|
| Proton |
2.62 |
0.28, 19.6 |
|
m3_fail_any
Table 3. Multivariate Statistics for Any-Cause Reconstruction Failure
| Model 3 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.01 |
0.96, 1.07 |
0.7 |
| BMI |
1.05 |
0.96, 1.16 |
0.3 |
| Diabetes |
|
|
0.013 |
| No |
— |
— |
|
| Yes |
77,638,774 |
0.00, NA |
|
| Smoking |
|
|
0.034 |
| No |
— |
— |
|
| Yes |
8.27 |
1.19, 60.7 |
|
| Type of Reconstruction |
|
|
0.7 |
| DTI |
— |
— |
|
| TE |
1.28 |
0.38, 4.41 |
|
| Radiation Side |
|
|
0.7 |
| Left |
— |
— |
|
| Right |
0.79 |
0.21, 2.92 |
|
| Type of Radiation |
|
|
0.076 |
| Photon |
— |
— |
|
| Proton |
4.31 |
0.85, 21.6 |
|
m3_fail_infxn
Table 3. Multivariate Statistics for Infection-Specific Reconstruction Failure
| Model 3 |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.04 |
0.97, 1.13 |
0.3 |
| BMI |
1.01 |
0.84, 1.15 |
0.9 |
| Diabetes |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.46 |
0.18, 58.6 |
|
| Type of Reconstruction |
|
|
0.5 |
| DTI |
— |
— |
|
| TE |
1.71 |
0.32, 10.6 |
|
| Radiation Side |
|
|
0.5 |
| Left |
— |
— |
|
| Right |
0.49 |
0.06, 3.16 |
|
| Type of Radiation |
|
|
0.3 |
| Photon |
— |
— |
|
| Proton |
3.02 |
0.31, 24.4 |
|
Model 4
Includes age, BMI, diabetes, smoking, reconstruction type, and
radiation type
# Make new df_model dataframe where outcome variables are coded as 0 / 1
df_model <- df
outcome_variables <- c("seroma", "hematoma", "msfn", "infxn_minor", "infxn_major", "fail_any", "fail_infxn")
for(var in outcome_variables) {
df_model[[var]] <- ifelse(df_model[[var]] == "Yes", 1, 0)
}
# Initialize a list to store tables
tables_list4 <- list()
# Function to create a title with variable labels
create_title <- function(var) {
label_text <- label(df[[var]])
paste("<div style='text-align: left; font-weight: bold; color: grey'> Table 3b. Multivariate Statistics for", label_text, "</div>")
}
# Loop to create models and tables
for (var in outcome_variables) {
# Create model
model_formula <- as.formula(paste(var, "~ age + bmi + diabetes + smoking + recon_type + xrt_type"))
model_name <- paste("m", var, sep = "_")
assign(model_name, glm(model_formula, data = df_model, family = binomial(link = "logit")))
# Create table with labeled title
table_name <- paste("table4", var, sep = "_")
table <- tbl_regression(get(model_name), exponentiate = TRUE) %>%
add_global_p() %>%
modify_caption(create_title(var)) %>%
as_gt() %>%
gt::tab_header("Multivariate Models")
# Store the table in the list
tables_list4[[table_name]] <- table
}
m4_seroma <- tables_list4$table4_seroma
m4_hematoma <- tables_list4$table4_hematoma
m4_msfn <- tables_list4$table4_msfn
m4_minor <- tables_list4$table4_infxn_minor
m4_major <- tables_list4$table4_infxn_major
m4_fail_any <- tables_list4$table4_fail_any
m4_fail_infxn <- tables_list4$table4_fail_infxn
# Print all tables
m4_seroma
Table 3b. Multivariate Statistics for Seroma
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.96, 1.09 |
0.5 |
| BMI |
1.07 |
0.95, 1.19 |
0.2 |
| Diabetes |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.41 |
0.18, 52.7 |
|
| Type of Reconstruction |
|
|
0.14 |
| DTI |
— |
— |
|
| TE |
3.19 |
0.69, 19.6 |
|
| Type of Radiation |
|
|
0.037 |
| Photon |
— |
— |
|
| Proton |
7.22 |
1.14, 46.3 |
|
m4_hematoma
Table 3b. Multivariate Statistics for Hematoma
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
0.86 |
0.58, 1.10 |
0.2 |
| BMI |
1.06 |
0.69, 1.47 |
0.7 |
| Diabetes |
|
|
0.9 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Type of Reconstruction |
|
|
0.3 |
| DTI |
— |
— |
|
| TE |
0.00 |
|
|
| Type of Radiation |
|
|
0.7 |
| Photon |
— |
— |
|
| Proton |
0.00 |
|
|
m4_msfn
Table 3b. Multivariate Statistics for MSFN
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.01 |
0.88, 1.16 |
>0.9 |
| BMI |
1.05 |
0.90, 1.19 |
0.5 |
| Diabetes |
|
|
0.007 |
| No |
— |
— |
|
| Yes |
4,242,160,977 |
0.00, NA |
|
| Smoking |
|
|
0.5 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Type of Reconstruction |
|
|
0.7 |
| DTI |
— |
— |
|
| TE |
0.62 |
0.03, 7.94 |
|
| Type of Radiation |
|
|
0.4 |
| Photon |
— |
— |
|
| Proton |
0.00 |
|
|
m4_minor
Table 3b. Multivariate Statistics for Minor Infection
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.04 |
0.97, 1.12 |
0.3 |
| BMI |
0.99 |
0.84, 1.12 |
>0.9 |
| Diabetes |
|
|
0.14 |
| No |
— |
— |
|
| Yes |
14.3 |
0.38, 594 |
|
| Smoking |
|
|
0.4 |
| No |
— |
— |
|
| Yes |
3.36 |
0.14, 41.1 |
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
2.87 |
0.56, 21.6 |
|
| Type of Radiation |
|
|
0.8 |
| Photon |
— |
— |
|
| Proton |
1.34 |
0.06, 11.1 |
|
m4_major
Table 3b. Multivariate Statistics for Major Infection
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.95, 1.09 |
0.6 |
| BMI |
0.94 |
0.76, 1.09 |
0.5 |
| Diabetes |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.5 |
| No |
— |
— |
|
| Yes |
2.81 |
0.12, 31.4 |
|
| Type of Reconstruction |
|
|
0.7 |
| DTI |
— |
— |
|
| TE |
1.36 |
0.27, 7.72 |
|
| Type of Radiation |
|
|
0.2 |
| Photon |
— |
— |
|
| Proton |
3.49 |
0.42, 22.3 |
|
m4_fail_any
Table 3b. Multivariate Statistics for Any-Cause Reconstruction Failure
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.01 |
0.96, 1.07 |
0.7 |
| BMI |
1.05 |
0.95, 1.15 |
0.3 |
| Diabetes |
|
|
0.013 |
| No |
— |
— |
|
| Yes |
74,666,311 |
0.00, NA |
|
| Smoking |
|
|
0.030 |
| No |
— |
— |
|
| Yes |
8.62 |
1.25, 62.7 |
|
| Type of Reconstruction |
|
|
0.7 |
| DTI |
— |
— |
|
| TE |
1.23 |
0.37, 4.13 |
|
| Type of Radiation |
|
|
0.049 |
| Photon |
— |
— |
|
| Proton |
4.76 |
1.01, 21.7 |
|
m4_fail_infxn
Table 3b. Multivariate Statistics for Infection-Specific Reconstruction Failure
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.04 |
0.97, 1.13 |
0.2 |
| BMI |
1.01 |
0.84, 1.13 |
>0.9 |
| Diabetes |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.76 |
0.19, 59.9 |
|
| Type of Reconstruction |
|
|
0.6 |
| DTI |
— |
— |
|
| TE |
1.53 |
0.29, 9.02 |
|
| Type of Radiation |
|
|
0.2 |
| Photon |
— |
— |
|
| Proton |
4.07 |
0.47, 28.4 |
|
Model 5
Includes age, BMI, smoking, reconstruction type, and radiation
type
# Make new df_model dataframe where outcome variables are coded as 0 / 1
df_model <- df
outcome_variables <- c("seroma", "hematoma", "msfn", "infxn_minor", "infxn_major", "fail_any", "fail_infxn")
for(var in outcome_variables) {
df_model[[var]] <- ifelse(df_model[[var]] == "Yes", 1, 0)
}
# Initialize a list to store tables
tables_list5 <- list()
# Function to create a title with variable labels
create_title <- function(var) {
label_text <- label(df[[var]])
paste("<div style='text-align: left; font-weight: bold; color: grey'> Table 3b. Multivariate Statistics for", label_text, "</div>")
}
# Loop to create models and tables
for (var in outcome_variables) {
# Create model
model_formula <- as.formula(paste(var, "~ age + bmi + smoking + recon_type + xrt_type"))
model_name <- paste("m", var, sep = "_")
assign(model_name, glm(model_formula, data = df_model, family = binomial(link = "logit")))
# Create table with labeled title
table_name <- paste("table5", var, sep = "_")
table <- tbl_regression(get(model_name), exponentiate = TRUE) %>%
add_global_p() %>%
modify_caption(create_title(var)) %>%
as_gt() %>%
gt::tab_header("Multivariate Models")
# Store the table in the list
tables_list5[[table_name]] <- table
}
m5_seroma <- tables_list5$table5_seroma
m5_hematoma <- tables_list5$table5_hematoma
m5_msfn <- tables_list5$table5_msfn
m5_minor <- tables_list5$table5_infxn_minor
m5_major <- tables_list5$table5_infxn_major
m5_fail_any <- tables_list5$table5_fail_any
m5_fail_infxn <- tables_list5$table5_fail_infxn
# Print all tables
m5_seroma
Table 3b. Multivariate Statistics for Seroma
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.96, 1.09 |
0.5 |
| BMI |
1.07 |
0.95, 1.19 |
0.2 |
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.49 |
0.18, 54.3 |
|
| Type of Reconstruction |
|
|
0.12 |
| DTI |
— |
— |
|
| TE |
3.37 |
0.72, 20.7 |
|
| Type of Radiation |
|
|
0.034 |
| Photon |
— |
— |
|
| Proton |
7.44 |
1.17, 48.0 |
|
m5_hematoma
Table 3b. Multivariate Statistics for Hematoma
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
0.86 |
0.58, 1.09 |
0.2 |
| BMI |
1.06 |
0.68, 1.47 |
0.8 |
| Smoking |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Type of Reconstruction |
|
|
0.3 |
| DTI |
— |
— |
|
| TE |
0.00 |
|
|
| Type of Radiation |
|
|
0.7 |
| Photon |
— |
— |
|
| Proton |
0.00 |
|
|
m5_msfn
Table 3b. Multivariate Statistics for MSFN
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.08 |
0.98, 1.22 |
0.14 |
| BMI |
1.04 |
0.93, 1.17 |
0.4 |
| Smoking |
|
|
0.6 |
| No |
— |
— |
|
| Yes |
0.00 |
|
|
| Type of Reconstruction |
|
|
0.2 |
| DTI |
— |
— |
|
| TE |
0.24 |
0.01, 1.93 |
|
| Type of Radiation |
|
|
0.2 |
| Photon |
— |
— |
|
| Proton |
0.00 |
|
|
m5_minor
Table 3b. Multivariate Statistics for Minor Infection
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.05 |
0.98, 1.14 |
0.2 |
| BMI |
1.00 |
0.85, 1.12 |
>0.9 |
| Smoking |
|
|
0.4 |
| No |
— |
— |
|
| Yes |
3.42 |
0.14, 39.8 |
|
| Type of Reconstruction |
|
|
0.4 |
| DTI |
— |
— |
|
| TE |
2.00 |
0.44, 10.8 |
|
| Type of Radiation |
|
|
>0.9 |
| Photon |
— |
— |
|
| Proton |
1.12 |
0.05, 8.89 |
|
m5_major
Table 3b. Multivariate Statistics for Major Infection
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.02 |
0.95, 1.09 |
0.6 |
| BMI |
0.94 |
0.76, 1.09 |
0.5 |
| Smoking |
|
|
0.5 |
| No |
— |
— |
|
| Yes |
2.82 |
0.12, 31.7 |
|
| Type of Reconstruction |
|
|
0.7 |
| DTI |
— |
— |
|
| TE |
1.42 |
0.27, 8.02 |
|
| Type of Radiation |
|
|
0.2 |
| Photon |
— |
— |
|
| Proton |
3.58 |
0.43, 22.9 |
|
m5_fail_any
Table 3b. Multivariate Statistics for Any-Cause Reconstruction Failure
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.03 |
0.98, 1.08 |
0.3 |
| BMI |
1.05 |
0.96, 1.16 |
0.2 |
| Smoking |
|
|
0.030 |
| No |
— |
— |
|
| Yes |
8.48 |
1.24, 61.6 |
|
| Type of Reconstruction |
|
|
>0.9 |
| DTI |
— |
— |
|
| TE |
0.97 |
0.31, 3.03 |
|
| Type of Radiation |
|
|
0.068 |
| Photon |
— |
— |
|
| Proton |
4.18 |
0.89, 19.0 |
|
m5_fail_infxn
Table 3b. Multivariate Statistics for Infection-Specific Reconstruction Failure
| Multivariate Models |
| Characteristic |
OR |
95% CI |
p-value |
| Age |
1.04 |
0.97, 1.12 |
0.3 |
| BMI |
1.01 |
0.84, 1.13 |
>0.9 |
| Smoking |
|
|
0.3 |
| No |
— |
— |
|
| Yes |
4.82 |
0.19, 61.4 |
|
| Type of Reconstruction |
|
|
0.6 |
| DTI |
— |
— |
|
| TE |
1.64 |
0.32, 9.59 |
|
| Type of Radiation |
|
|
0.2 |
| Photon |
— |
— |
|
| Proton |
4.26 |
0.49, 29.7 |
|