Files:

This Dryad submission includes the following files:

1. HeadSOAR_deidentified_participant_level_dataset.xlsx
   Deidentified participant-level dataset for the HeadSOAR trial. Each row represents one randomised participant. The dataset includes treatment allocation, baseline characteristics, medical history, imaging and procedural variables, follow-up outcomes, safety outcomes, and analysis set indicators.
2. HeadSOAR_data_dictionary.xlsx
   Data dictionary describing the variables, labels, coding, value definitions, units, and missing value conventions used in the participant-level dataset.
3. HeadSOAR.R
   R code used to reproduce the main analyses, including baseline tables, inverse probability treatment weighting, primary and secondary outcome analyses, safety analyses, subgroup analyses, and sensitivity analyses. The R analyses were performed using R version 4.4.0.
4. MI.sas
   SAS code used for multiple imputation analyses. The SAS analyses were performed using SAS version 9.4.

Missing values and unavailable information

In the participant-level dataset, blank cells indicate values that were unavailable or not recorded in the deidentified dataset and should be treated as missing values in analyses. Blank cells were left blank rather than replaced with text strings such as n/a to preserve the numeric and categorical column formats and to maintain compatibility with the accompanying R and SAS analysis code.

In the data dictionary, “-” in the “Missing values / unavailable information” column indicates that no blank cells were observed for that variable in the submitted participant-level dataset.

Blank cells were present in the baseline glucose variable (GLUCBL), the 90-day outcome variables (MRS3M and EQVAS3M), and selected safety outcome variables (SICH, ANYICH, MLCEREDE, and CEREDE). These blank cells indicate unavailable or unrecorded values for the corresponding variables.

Licence and reuse

The files deposited in Dryad are shared under the Dryad CC0 licence waiver. Users may reuse the deidentified data and code in accordance with the Dryad licence terms.

Although the dataset is made openly available under CC0, the authors encourage users follow standard scholarly and ethical norms when reusing these materials:

1. Cite the associated manuscript and the Dryad dataset DOI in any publication, preprint, presentation, report, or derivative work using these data.
2. Acknowledge the HeadSOAR investigators as the original data creators.
3. Do not attempt to re-identify any participant, study site, investigator, or centre from the deidentified dataset.
4. Do not link these data with external datasets in a way that could reasonably increase the risk of participant or centre re-identification.
5. Contact the corresponding author before initiating major secondary analyses intended for publication, so that overlap with planned HeadSOAR investigator-led analyses can be discussed.
6. Clearly state that any secondary analysis, interpretation, or conclusion is the responsibility of the secondary user and does not necessarily represent the views of the HeadSOAR investigators.
7. Share derived code and sufficiently detailed methods to support transparency and reproducibility.

These requests are intended to promote responsible scientific reuse and do not alter the Dryad CC0 licence waiver.

Human subjects data

This dataset contains deidentified human participant data from the HeadSOAR trial, a multicentre randomised controlled trial conducted at 67 comprehensive stroke centres in China. The study was approved by the Ethics Committee of the Affiliated Hospital of Southwest Medical University (approval No. KY2023285), and written informed consent was obtained from all participants or their legally authorised representatives before participation.

Before deposition, the dataset was deidentified in accordance with applicable ethical and legal requirements. Direct personal identifiers, including names, hospital identification numbers, national identification numbers, telephone numbers, addresses, medical record numbers, and other contact information, were removed. Study identifiers were recoded, and no key linking the deposited dataset to identifiable participants is included in this repository. Dates and times were removed or converted to relative time intervals where applicable. Variables that could increase re-identification risk, including centre-related identifiers, rare categories, and extreme or unusual values, were reviewed and modified, grouped, or removed as appropriate. The deposited dataset does not include free-text clinical notes, imaging files, consent forms, or other documents containing personally identifiable information.

The deidentified participant-level dataset and statistical code are provided to support reproducibility of the analyses reported in the associated manuscript. The authors confirm that the shared files contain no personally identifiable information and are compatible with public sharing under the Dryad CC0 licence waiver.