Table 2 Summary of data quality assessment domains in studies on research in paramedicine

Completeness

57 (45)

Missingness, adherence, availability, unknown/not reported, granularity

Measure of how often a variable is present when expected. Complement of missingness

Proportion and percent

Raw percent complete, weighted percent complete, percent legible

Linkage

34 (27)

Match

Can records belonging to the same person or event be linked between different databases? How well? By what means?

Probabilities, percent success, sensitivities, specificity, positive predictive value, negative predictive value (and related measures: false positive, false negative)

Match-weight cut-off, match quality

Accuracy

14 (11)

Validity, correctness, concordance, plausibility, ascertainment, capture, incidence, population

Does the variable measure what it claims to measure? Is the result plausible or possible?

Proportion and percent, sensitivity, specificity, positive predictive value, negative predictive value

–

Reliability

10 (8)

Agreement, precision, consistency, variation, aggregation, uniqueness, granularity, quality

Is the measurement free from error and consistent over time and among observers?

Difference in proportations, kappa, intraclass correlation coefficient, correlation, other (Andrews, Reisner)

–

Representativeness

11 (9)

External validity, bias, generalizability, concordance

How well does the data correspond to other data expected to be similar? How well do parts of the data correspond when they are expected to be similar? Is the data biased in some way?

Difference in proportions, correlation, kappa, sensitivity

Proportions, absolute standardized difference, ± 5% difference