Table 1.

Possible sources of measurement heterogeneity in measurements of predictors, illustrated by examples from previously published prediction studies

Type of Predictor	Examples of Predictors	Examples of Measurement Heterogeneity
Anthropometric	Height	Guidelines on imaging decisions in osteoporosis care are established using standardized
measurements	Weight	measurements of height, while in clinical practice height is measured using
	Body circumference	nonstandardized techniques or self‐reported values.8
Physiological	Blood pressure	In scientific studies, blood pressure is often measured by the average of multiple
measurements	Serum cholesterol	measurements performed under standardized conditions, while blood pressure
	HbA1c	measurements in practice deviate from protocol guidelines in various ways due to
	Fasting glucose	variability in available time and devices.9
Diagnosis	Previous/current	The diagnosis “hypertension” can be defined as a blood pressure of ≥140/90 mm Hg
	disease	(without use of anti‐hypertensive therapy) or as the use of anti‐hypertensive drugs.10
Treatment/	Type of drug used	The cut‐off value for an “increased length of stay in the hospital” to predict unplanned
Exposure status	Smoking status	readmission may depend on the country in which the model is evaluated.11
	Dietary intake
Imaging	Presence or size of tissue	In scientific studies, review of FDG PET scans may be protocolized or performed by
	on ultrasound, MRI, CT	a single experienced nuclear medicine physician, blinded to patient outcome.12 In routine
	or FDG PET scans	practice, FDG PET scans may be reviewed under various systematics or by a
		multidisciplinary team.13