Overview

This Dryad deposit contains de-identified data and analysis code associated with the three-year follow-up of the multicentre randomised PATENCY trial comparing no-touch and conventional saphenous vein harvesting for coronary artery bypass grafting.

Associated article: No-touch versus conventional vein in coronary artery bypass grafting: three year follow-up of multicentre randomised PATENCY trial (BMJ 2025;389:e082883; DOI: 10.1136/bmj-2024-082883) .

Files included

• PATENCY_3y_baseline_and_clinical_outcome_ageBMI_NA_agegrouped.csv: participant-level baseline characteristics and three-year clinical outcomes.
• PATENCY_3y_Graft_outcomes_ageBMI_NA_agegrouped.csv: graft-level characteristics and three-year computed tomography angiography outcomes.
• Codes_For_BMJ_PATENCY_3y_English.R: R code documenting the original statistical analysis workflow. Comments and explanatory text are in English.

Missing data coding

All originally empty cells in the two public CSV files are coded as NA. NA means that the information was missing, unavailable, or not recorded. It must not be interpreted as zero, absence of a condition or event, or “not applicable.”

De-identification

To strengthen participant anonymity:

• Exact ages were replaced with the following categories: <50, 50-59, 60-69, 70-79, and >=80 years.
• Exact body mass index values were replaced with the following categorical labels: <20, 20-25, 25-30, 30-35, and >35 kg/m².
• Direct personal identifiers, dates of clinical procedures or examinations, and study-centre identifiers are not included in the public files.

Analysis code and computational environment

The original analyses were performed using R version 4.0.1. The submitted R script, Codes_For_BMJ_PATENCY_3y_English.R, documents the statistical analyses conducted for the three-year follow-up of the PATENCY trial.

The script includes graft-level and anastomosis-level analyses of graft failure and occlusion, generalized estimating equation models, generalized linear models, absolute risk differences with bootstrap confidence intervals, subgroup analyses, forest plots, patient-level graft outcome analyses, Kaplan–Meier analyses, log-rank tests, Cox proportional hazards models, competing-risk analyses, and analyses of leg wound complications.

R packages

The following R packages are used or referenced in the script:

• boot
• broom
• cmprsk
• cmprskcoxmsm
• dplyr
• forestplot
• geepack
• ggplot2
• ggpubr
• grid
• Hmisc
• Matching
• mice
• reshape2
• rms
• survival
• survminer
• survey
• tableone
• twang

Base R functions from packages supplied with R, including stats, utils, and graphics, are also used.

Some packages were loaded as part of the original analysis environment even though package-specific functions may not be called in every section of the script.

Running the R script

Before running the code, users should place the R script and the required input data files in the same working directory. The hard-coded working directory in the script:

/Users/tianmeice/Desktop/Rdata

should be replaced with the path to the user's local working directory. Alternatively, the working directory can be set manually in R before running the script.

The two principal data files deposited with this dataset are:

• PATENCY_3y_baseline_and_clinical_outcome_ageBMI_NA_agegrouped.csv
• PATENCY_3y_Graft_outcomes_ageBMI_NA_agegrouped.csv

The filenames used in the original read.csv() commands may therefore need to be updated to match the filenames of the deposited files.

The deposited CSV files use NA to indicate information that was missing, unavailable, or not recorded. R normally recognises these values as missing data when the files are imported using read.csv().

The deposited CSV files are encoded in UTF-8. If necessary, the fileEncoding = "GBK" argument in the original script should be removed or changed to an appropriate UTF-8 encoding when reading the deposited files.

The script should be run sequentially because some later sections depend on data objects created in earlier sections. Users should install all required packages before running the code.

To protect participant confidentiality, exact age and BMI values in the publicly deposited datasets were replaced with categories. Consequently, code sections that originally used continuous age or BMI values may require adaptation to use the corresponding categorical variables and may not reproduce those adjusted analyses exactly.

Some sections of the original analysis script refer to additional source or derived files used for specific analyses, including anastomosis-level analyses, subgroup analyses, forest plots, patient-level graft analyses, and leg wound outcomes. These sections can only be rerun when the corresponding input files and variables are available.

Graft-level data dictionary

File: PATENCY_3y_Graft_outcomes_ageBMI_NA_agegrouped.csv

• ID_Graft: unique coded identifier for each graft vessel.
• ID_Anastomosis: unique coded identifier for each anastomosis.
• ID_Patient: unique coded participant identifier.
• RanID: coded randomisation identifier.
• Group: randomised harvesting technique; 0 = conventional, 1 = no-touch.
• sex: 0 = male, 1 = female.
• age: participant age category; <50, 50-59, 60-69, 70-79, or >=80 years.
• BMI: body mass index category; <20, 20-25, 25-30, 30-35, or >35 kg/m².
• DM: diabetes mellitus; 0 = no, 1 = yes.
• HP: hypertension; 0 = no, 1 = yes.
• GraftType: graft type; 1 = in situ left internal mammary artery, 2 = free left internal mammary artery, 3 = vein graft, 4 = in situ right internal mammary artery, 5 = free right internal mammary artery, 6 = radial artery, 7 = gastroepiploic artery.
• Distal_Anastomosis: distal anastomosis territory; 1 = left anterior descending artery, 2 = diagonal branch, 3 = left circumflex territory, 4 = right coronary artery territory.
• Sequential_bypas: sequential bypass; 0 = no, 1 = yes.
• Graft_mark: numerical marker distinguishing grafts within the same participant.
• Gmark_combine: combined graft identifier used to distinguish each graft within the database.
• 3yCT_result: three-year graft imaging result; 1 = fully patent, 2 = mild-to-moderate stenosis, 3 = severe stenosis, 4 = occlusion.

Participant-level data dictionary

File: PATENCY_3y_baseline_and_clinical_outcome_ageBMI_NA_agegrouped.csv

• disno: number of diseased major coronary vessels; 2 = two-vessel disease, 3 = three-vessel disease.
• ID_patient: unique coded participant identifier.
• group: randomised harvesting technique; 0 = conventional, 1 = no-touch.
• gender: 0 = male, 1 = female.
• age: participant age category; <50, 50-59, 60-69, 70-79, or >=80 years.
• BMI: body mass index category; <20, 20-25, 25-30, 30-35, or >35 kg/m².
• smoking_history: current smoking status; 0 = no, 1 = yes.
• DM: diabetes mellitus; 0 = no, 1 = yes.
• HP: hypertension; 0 = no, 1 = yes.
• Hyperlipidermia: hyperlipidaemia; 0 = no, 1 = yes.
• COPD: chronic obstructive pulmonary disease; 0 = no, 1 = yes.
• stroke: history of stroke; 0 = no, 1 = yes.
• CCS: Canadian Cardiovascular Society angina class; 1 = class I, 2 = class II, 3 = class III, 4 = class IV.
• NYHA: New York Heart Association functional class; 1 = class I, 2 = class II, 3 = class III, 4 = class IV.
• MI: history of myocardial infarction; 0 = no, 1 = yes.
• PCI: history of percutaneous coronary intervention; 0 = no, 1 = yes.
• EF: left ventricular ejection fraction.
• LV: left ventricular end-diastolic diameter.
• Peripheral_artery_disease: peripheral artery disease; 0 = no, 1 = yes.
• syntax: SYNTAX score.
• EuroScore: EuroSCORE II.
• Leftmain: left main coronary artery disease; 0 = no, 1 = yes.
• CPB: cardiopulmonary bypass strategy; 0 = off-pump, 1 = on-pump.
• fu3macce: composite major adverse cardiac and cerebrovascular event during three-year follow-up; 0 = no, 1 = yes.
• fu3macceDay: time in days to the first composite event or censoring.
• fu3die: all-cause death during three-year follow-up; 0 = no, 1 = yes.
• fu3dieD: time in days to all-cause death or censoring.
• fu3cardiacdeath: cardiac death during three-year follow-up; 0 = no, 1 = yes.
• fu3cardieDay: time in days to cardiac death or censoring.
• fu3MI: non-fatal myocardial infarction during three-year follow-up; 0 = no, 1 = yes.
• fu3MID: time in days to non-fatal myocardial infarction or censoring.
• fu3stroke: stroke during three-year follow-up; 0 = no, 1 = yes.
• fu3strokeD: time in days to stroke or censoring.
• fu3rePCI: repeat percutaneous coronary intervention during three-year follow-up; 0 = no, 1 = yes.
• fu3rePCID: time in days to repeat percutaneous coronary intervention or censoring.
• fu3angina: recurrent angina during three-year follow-up; 0 = no, 1 = yes.
• fu3anginaD: time in days to first recurrent angina or censoring.
• fu3readCar: rehospitalisation for a cardiac reason during three-year follow-up; 0 = no, 1 = yes.
• fu3readCar_D: time in days to first cardiac rehospitalisation or censoring.

Human subjects data

The PATENCY trial was approved by the relevant institutional ethics committees, and all participants provided written informed consent for participation in the trial and for the use of their clinical and follow-up data for scientific research purposes.

The dataset deposited in Dryad has been de-identified before publication in the public domain. All direct personal identifiers have been removed, including names, hospital numbers, medical record numbers, national identification numbers, telephone numbers, addresses, and other contact information. Variables that could increase the risk of re-identification have also been removed, recoded, or grouped where appropriate. For example, individual BMI values have been replaced by BMI categories (<20, 20–25, 25–30, 30–35, and >35 kg/m²), and centre- or site-level identifiers have not been included.

The deposited dataset contains only coded clinical, procedural, graft-level, CTA-derived, and follow-up variables required to reproduce the analyses reported in the published article. No personally identifiable information is included in the deposited files. The de-identification process was performed in accordance with applicable ethical and institutional data protection requirements to protect participant confidentiality.