Dataset DOI: 10.5061/dryad.kprr4xhk9

Description of the data and file structure

1. Study Overview

• Trial Title: Metformin for the prevention of preeclampsia in high-risk pregnant women with overweight: randomised, open-label, controlled trial
• Primary Objective: To investigate the effect of metformin in preventing preeclampsia in high-risk and overweight pregnant women in China.
• Secondary Objectives: To compare the effect of metformin to aspirin in preventing preeclampsia in high-risk and overweight pregnant women in China, with the aspirin group designated as an exploratory analysis arm.
• Trial Design: 1:1:1 random allocation to metformin, aspirin, or control groups
• Participants: 410 deidentified pregnant women (BMI ≥ 24 kg/m²) at high risk for preeclampsia
• Ethics Approval: Peking University Third Hospital (PUTH) Medical Science Research Ethics Committee (M2021307)
• Trial Registration: ChiCTR2100051255 (WHO International Clinical Trials Registry Platform)

2. Datasets and Data Dictionaries

We provide 3 core datasets, each paired with a clear data dictionary. All variables are fully defined in the dictionaries, no extra documentation is needed.

Datasets:

1. Dataset 1: (1)_Data1_maternal_ITT.csv
   • Content: Maternal baseline characteristics, intervention details, and maternal outcomes (including preeclampsia)
   • Data Dictionary: (4)_Variable_dictionary_Maternal_ITT_data.csv
   • Missing Data: Hypoalbuminemia has five missing values, which are coded as NA in the dataset.
2. Dataset 2: (2)_Data2_Newborn.csv
   • Content: Neonatal outcomes (birth weight, NICU admission, complications, etc.)
   • Data Dictionary: (4)_Variable_dictionary_Newborn_data.csv
   • Structurally missing data: IVH_Grade_level indicates the specific grade of neonatal intraventricular haemorrhage (IVH). In this study, the secondary outcome was defined as IVH grade 2 or higher. Therefore, IVH_Grade_level is recorded only when IVH_Grade_2=1. When IVH_Grade_2=0, this variable is left blank because no grade ≥2 IVH was present; this should be interpreted as not applicable, rather than missing data. These structurally missing values are explicitly coded as SNA in our dataset.
   • Neonatal_ID was created to identify individual newborns and link them to their mother. The numeric component of Neonatal_ID is the same as Maternal_ID; the suffix A or B distinguishes infants from twin pregnancies, with A indicating the first infant and B indicating the second infant.
3. Dataset 3: (3)_Data3_survival_curve_HR_value.csv
   • Content: Survival analysis data for preeclampsia (hazard ratios, Kaplan-Meier curve support data)
   • Data Dictionary: (4)_Variable_dictionary_Survival_data.csv

Files and variables

All variables, abbreviations, and units of measurement for the dataset are fully documented in three dedicated data dictionaries (variable cross-reference tables) one paired with each core data file. The dictionaries serve as comprehensive references to eliminate ambiguity, and no additional documentation is required to interpret the data.

Code/software

The datasets are provided in standard .xlsx format and can be used in major statistical software packages, including code-based platforms such as R, Python, SAS, and Stata, as well as graphical-interface software such as SPSS, LibreOffice Calc, and Microsoft Excel. No proprietary software is required specifically for access to the data files. The files can be imported using standard spreadsheet import functions, for example read_csv() in R, pd.read_csv() in Python, or PROC IMPORT in SAS, or opened directly in spreadsheet software. All variables, units, and abbreviations are defined in the accompanying data dictionaries.

Access information

Other publicly accessible locations of the data:

• None. This dataset is exclusively shared via Dryad to comply with BMJ's data sharing policy.

Data was derived from the following sources:

• All data were originally collected from the clinical information system of the Department of Obstetrics and Gynecology, PUTH, including data from routine antenatal visits, delivery hospitalization records, postpartum follow-up assessments, and outcome adjudication forms completed by independent senior obstetricians. No secondary data or external datasets were used in this study.

Human subjects data

This research study obtained ethical approval from the PUTH Medical Science Research Ethics Committee (approval number M2021307). All participating individuals provided written explicit informed consent permitting the deposition and unrestricted public release of fully de-identified study data in the public domain via Dryad.
All human subject data included in this dataset has undergone comprehensive anonymization procedures to eliminate all personally identifiable information (PII). We permanently removed all direct identifiers including full names, ID numbers, residential addresses, contact information, precise birth dates and unique medical record codes. Any indirect quasi-identifiers that could potentially enable re-identification of participants have also been aggregated or masked to rule out risks of individual recognition.
All de-identification work strictly complies with the Declaration of Helsinki, institutional ethical research standards, and Dryad’s requirements for human subject data archiving. No residual information within the shared dataset can be used to trace back to any single participant.