Fraud in clinical trials

The background

Clinical trial fraud involving fabricated patients is a well-documented problem in research integrity. Statistical methods, including data visualisation, can be a first line of detection, flagging sites for deeper investigation. A simulated dataset is provided for a randomised, controlled multi-centre study studying an active treatment versus placebo in blood pressure reduction. One site contains data that is fraudulent.

The challenge

Explore the data using data visualisation and discover which site contains fabricated data
Use one or more data visualisations to provide convincing evidence to influence the study team to conduct further investigation

Further Information

Fabricated data often has the following properties:

Digit preference: fraudsters often favour round numbers
Variability: fraudsters unconsciously produce data that is too consistent
Missing data: fraudsters often underestimate the drop-out rate or level of missing data
Correlation: fabricated longitudinal data often overestimates the correlation between visits
Visit regularity: visit dates are often more variable in real data compared to fabricated data

The data

A standard ADVS data format is provided, including the following key variables:

Variable Name	Variable Label
USUBJID	Subject ID
SITEID	Site ID (important for identification)
TRT01P(N)	Treatment group
ADT	Visit date
AVISIT(N)	Visit ID
PARAMCD	DIABP or SYSBP (Blood pressure)
AVAL	Result (measured value)
BASE	Baseline value
CHG	Change from Baseline
PCHG	% Change from Baseline

``` {r, eval=FALSE}

load data

ADVS <- read_csv(“https://raw.githubusercontent.com/VIS-SIG/Wonderful-Wednesdays/master/data/2026/2026-05-13/ADVS.csv”)

```