D

Data-derived analysis: See unplanned analysis.

Data dredging: Performing many analyses on the data from a study, for example looking for associations among many variables.6 The term is particularly used to refer to unplanned analyses, where there is no apparent hypothesis, and only statistically significant results are reported. Multiple statistical analyses on the same set of data increase the probability of making a Type I error (i.e., attributing a difference to an intervention when chance is a reasonable explanation).

Data mining: Data analysis techniques that use algorithms to detect patterns in large data sets containing numerous variables with unknown complex relations.20

DEcIDE: See Developing Evidence to Inform Decisions about Effectiveness.

Detection bias: Systematic differences between comparison groups in how outcomes are ascertained, diagnosed, or verified.6 Also called ascertainment bias. See also bias.

Developing Evidence to Inform Decisions about Effectiveness (DEcIDE): The DEcIDE Network is a collection of research centers created by the Agency for Healthcare Research and Quality in 2005 to gather new knowledge and information on specific treatments and conduct studies on the outcomes, effectiveness, safety, and usefulness of medical treatments and services.21 

Directness: The extent to which the evidence links the interventions directly to health outcomes.18 Indirect evidence can encompass surrogate outcomes or refers to situations when two or more bodies of evidence are needed to compare interventions.

Double blind: See blinded study.