Antithrombotic therapy for migraine in patients with patent foramen ovale: A randomized clinical trial
Data files
Jun 25, 2026 version files 560.65 KB
-
code.zip
19.72 KB
-
data.zip
478.26 KB
-
README.md
10.44 KB
-
Variable_dictionary.xls
52.22 KB
Abstract
This dataset contains anonymized individual participant data from the COMPETE trial, an investigator-initiated, multicenter, randomized, active-controlled clinical study evaluating the efficacy and safety of antithrombotic therapy for migraine prevention in patients with patent foramen ovale (PFO).
Participants underwent a 12-week screening period followed by a 12-week treatment phase, during which they were randomly assigned to receive aspirin, clopidogrel, rivaroxaban, or metoprolol. The dataset includes demographic and baseline clinical characteristics, treatment allocation, headache diary–derived outcomes (including migraine days and attacks), response rate (≥50% reduction), adverse events, and quality-of-life measures.
Data were prospectively collected using standardized headache diaries and structured follow-up assessments across 39 study sites. These data support the primary and secondary analyses reported in the associated manuscript, as well as additional subgroup, and sensitivity analyses.
Dataset DOI: 10.5061/dryad.2rbnzs84m
Description of the data and file structure
This dataset was generated from a multicenter, randomized clinical trial evaluating the efficacy and safety of antithrombotic therapies for migraine in patients with patent foramen ovale (PFO).
The dataset includes two subject-level datasets (i.e., ADSL and ADACRM), along with detailed variable definitions and key statistical programs. Analysis-ready data are provided in Microsoft Excel format, and the statistical programs used to generate tables, listings, and figures (TLFs) are provided as SAS scripts.
The datasets capture participant demographics, baseline clinical characteristics, treatment allocation, headache diary–derived outcomes (including migraine days and migraine attacks), and safety outcomes. A comprehensive variable dictionary is included to facilitate interpretation of all variables.
These materials are provided to support transparency and reproducibility of the results reported in the study.
Files and variables
File structure
The dataset is organized into three main components:
- code.zip
Contains all statistical analysis programs written in SAS.- Macros/
allfmt.sas: Defines variable formats used across datasets and analyses.
- SAS macro code:
M_CT_BW_NO.sas: Analysis for baseline continuous variables and secondary continuous variables;M_CG_BW_R_ALL_.sas,M_CG_BW_R_ALL.sas: Analysis for baseline categorical variables and safety outcomes;T_CT_BW_CI.sas,T_CG_BW_R2_diff.sas:Analysis for secondary endpoints (between-group differences);M_CG_BW_R_ALL_NT.sas,T_CG_CI_CMH_CRD.sas: Analysis for primary endpoint (descriptive analysis, between-group differences and comparisons) and secondary categorical variables;T_CG_CI_CMH_CRD_WP.sas: subgroup analysis;blank.sas: A nested macro.
TLF.sas: Key code used to generate Tables, Listings, and Figures (TLFs).
- Macros/
- data.zip
Contains two datasets in Excel format for analysis:adsl.xls: dataset for analysis population determination;adacrm.xls: dataset for analysis of baseline characteristics, primary endpoint, secondary endpoints and safety outcomes.
- Variable dictionary.xls
Provides detailed definitions for all variables included in the datasets, including variable names, labels, variable coding, and derivation rules.
Dataset description
adsl.xls is a participant-level dataset containing treatment group assignment and protocol deviation information. Each row represents one participant.
adacrm.xls is the key analysis dataset, including analysis population flag variable, baseline variables, migraine diary–derived variables, MSQ scores, safety outcomes, derived endpoints, and subgroup variables. Each row represents one participant.
Variables
subjid indicates the random number.
group indicates treatment group, coded as 1=Aspirin, 2=Clopidogrel, 3=Rivaroxaban, and 4=Metoprolol.
pdtype indicates the type of protocol deviation.
fasfl indicates inclusion in the full analysis set (FAS), coded as 1=Yes and 2=No.
pprotfl indicates inclusion in the per-protocol set (PPS), coded as 1=Yes and 2=No.
age round indicates rounded age in years.
sex indicates gender, coded as 1=Male and 2=Female.
bmi cat indicates body mass index category.
comy indicates duration of migraine in years.
migzz_v1 indicates migraine with aura, coded as 1=Yes and 2=No.
ech_v1 indicates transthoracic echocardiography (TTE), coded as 1=Yes and 2=No.
xztee_v1 indicates transesophageal echocardiography (TEE), coded as 1=Yes and 2=No.
xztee1a_v1 indicates contrast transthoracic echocardiography (cTTE), coded as 1=Yes and 2=No.
xztee1b_v1 indicates contrast transesophageal echocardiography (cTEE), coded as 1=Yes and 2=No.
xztee1c_v1 indicates contrast transcranial Doppler (cTCD), coded as 1=Yes and 2=No.
ech29_v1 indicates right-to-left shunt (RLS) grade, coded as 0=Grade 0, 1=Grade 1, 2=Grade 2, and 3=Grade 3.
migpl1_v1 indicates the number of migraine days per month at baseline.
migpl3_v1 indicates the number of migraine attacks per month at baseline.
msqr_v1 indicates the role restrictive (RR) domain of the Migraine-Specific Quality of Life Questionnaire (MSQ) at baseline.
msqp_v1 indicates the role preventive (RP) domain of MSQ at baseline.
msqe_v1 indicates the emotional function (EF) domain of MSQ at baseline.
mi indicates history of myocardial infarction, coded as 1=Yes and 2=No.
ang indicates angina, coded as 1=Yes and 2=No.
smoke indicates smoking status, coded as 1=Yes and 2=No.
hyp indicates hypertension, coded as 1=Yes and 2=No.
dia indicates diabetes mellitus, coded as 1=Yes and 2=No.
hype indicates hyperlipidemia, coded as 1=Yes and 2=No.
snore indicates snoring, coded as 1=Yes and 2=No.
palp indicates palpitations, coded as 1=Yes and 2=No.
jsxt indicates psychiatric disorders, coded as 1=Yes and 2=No.
tws indicates history of head trauma, coded as 1=Yes and 2=No.
pill indicates contraceptive drug use, coded as 1=Yes and 2=No.
cein indicates cerebral infarction, coded as 1=Yes and 2=No.
tia indicates transient ischemic attack, coded as 1=Yes and 2=No.
wzjb indicates peripheral vascular disease, coded as 1=Yes and 2=No.
rein indicates respiratory insufficiency, coded as 1=Yes and 2=No.
nyha indicates NYHA class I heart function, coded as 1=Yes and 2=No.
atfi indicates atrial fibrillation, coded as 1=Yes and 2=No.
arrhy indicates other arrhythmias, coded as 1=Yes and 2=No.
jwpci indicates previous percutaneous coronary intervention, coded as 1=Yes and 2=No.
jwcabg indicates previous coronary artery bypass grafting, coded as 1=Yes and 2=No.
jwtt indicates previous migraine preventive treatment, coded as 1=Yes and 2=No.
jxyy indicates acute medication use, coded as 1=Yes and 2=No.
migpl3_v1a indicates monthly migraine attacks at baseline and is derived from migpl3_v1.
migpl1_v1a indicates monthly migraine days at baseline and is derived from migpl1_v1.
pl3_v3, pl3_v4, and pl3_v5 indicate monthly migraine attacks at weeks 4, 8, and 12.
pl1_v3, pl1_v4, and pl1_v5 indicate monthly migraine days at weeks 4, 8, and 12.
msqr_v1a, msqp_v1a, and msqe_v1a indicate baseline MSQ domain scores derived from original variables.
pl7_v3–pl7_v5, pl8_v3–pl8_v5, and pl9_v3–pl9_v5 indicate MSQ domain scores at follow-up visits.
msqt1v, msqt3v, msqt4v, and msqt5v indicate total MSQ scores at baseline and follow-up visits.
ae indicates any adverse event, coded as 1=Yes and 2=No.
impae indicates adverse events related to investigational medicinal products (IMPs), coded as 1=Yes and 2=No.
sae indicates any serious adverse event, coded as 1=Yes and 2=No.
impsae indicates serious adverse events related to IMPs, coded as 1=Yes and 2=No.
bleed indicates any bleeding event, coded as 1=Yes and 2=No.
hemor indicates major bleeding, coded as 1=Yes and 2=No.
rrydl indicates the primary endpoint-respondse rate, coded as 1=Yes and 2=No, defined as a ≥50% reduction in monthly migraine days or attacks at 12 weeks compared with baseline.
rrydllo indicates response rate using LOCF imputation.
rdlt indicates response rate based on migraine days only, coded as 1=Yes and 2=No.
rdltlo indicates response rate based on migraine days only using LOCF imputation.
avdm and avcm indicate absolute reductions in migraine days and attacks.
avdmr and avcmr indicate percentage reductions in migraine days and attacks.
coce indicates complete cessation of migraine, coded as 1=Yes and 2=No.
chrr, chrp, and chef indicate changes in MSQ domain scores.
chmsq indicates change in total MSQ score.
pttyx indicates migraine frequency subgroup, coded as 1=<15 days/month and 2=≥15 days/month.
ageyz indicates age subgroup, coded as 1=≥38.5 years and 2=<38.5 years.
Missing data
Missing values are represented by blank cells. In addition, indicator variables (e.g., migpl31v, pl33v, msqr1v) identify missingness at specific time points, coded as 1=Yes and 2=No.
Code/software
The datasets are provided in Microsoft Excel (.xls) format and can be opened using Microsoft Excel, LibreOffice Calc, OpenOffice Calc, or other compatible software.
Statistical analysis code is provided as SAS programs (.sas). Analyses were originally conducted using SAS software.
Recommended software:
- SAS version 9.4 or later for reproducing analyses
- Microsoft Excel or equivalent for viewing datasets
- Any text editor for reviewing SAS code
Workflow
- Open datasets in
data.zip. - Refer to
Variable dictionary.xlsfor variable definitions. - Load macro and format files from
code.zip/Macros/. - Run
TLF.sasto generate tables, listings, and figures. - Individual scripts may also be run separately for specific analyses.
No proprietary software is required to view the datasets; however, SAS is required to reproduce the analyses.
Access information
Other publicly accessible locations of the data: None.
The data are exclusively available through the Dryad Digital Repository associated with this study.
Data source: The data were generated prospectively as part of a multicenter randomized clinical trial conducted by the study investigators.
Human subjects data
All participants provided written informed consent prior to enrollment, which included consent for the use of their data for research purposes and for sharing of de-identified data in the public domain.
Prior to data sharing, all datasets were fully de-identified to protect participant privacy. Direct identifiers (including names, contact information, identification numbers, and exact addresses) were removed. Indirect identifiers were minimized or generalized where necessary, including dates and site-level information, to reduce the risk of re-identification. Each participant was assigned a unique study identification number with no linkage to personal identifiers.
The shared dataset contains no information that could reasonably be used to identify individual participants.
Data were derived from the COMPETE trial, an investigator-initiated, multicenter, randomized, active-controlled clinical study conducted across 39 sites in China. Eligible participants with migraine and patent foramen ovale (PFO) underwent a 12-week screening period followed by a 12-week randomized treatment phase.
Clinical and demographic data were collected at baseline using standardized case report forms. Headache outcomes were prospectively recorded using structured daily headache diaries, in which participants documented headache characteristics, associated symptoms, functional impact, and use of acute medications in real time. Follow-up assessments were conducted at 4, 8, and 12 weeks after randomization via office visits, telephone, or mobile applications. Safety data, including adverse events, were collected throughout the study period.
Data were entered into a centralized electronic data capture system and underwent routine data validation procedures, including range checks, consistency checks, and query resolution. Headache diary data were centrally reviewed by trained neurologists to ensure consistency with diagnostic criteria.
For the shared dataset, all data were de-identified prior to release. Variables were harmonized and coded according to predefined definitions, and derived variables (e.g., changes in monthly migraine days, changes in monthly migraine attacks, and responder status) were calculated based on prespecified analysis rules. Missing data were retained as recorded, with imputation procedures applied only in sensitivity analysis of primary endpoint as described in the associated manuscript.