A review for clinical outcomes research: hypothesis generation, data strategy, and hypothesis-driven statistical analysis.

In recent years, more and more large, population-level databases have become available for clinical research. The size and complexity of these databases often present a methodological challenge for investigators. We propose that a "protocol" may facilitate the research process using these databases. In addition, much like the structured History and Physical (H&P) helps the audience appreciate the details of a patient case more systematically, a formal outcomes research protocol can also help in the systematic evaluation of an outcomes research manuscript.