Research and Statistics: Study Design and Data Sources

Abstract

1. Anne Palaia, PhD, MPH 1. The Johns Hopkins University School of Medicine, Baltimore, MD. A 12-year-old boy presents for continuity care with his mother. She reports that he has had trouble sleeping, weight loss, and low mood. Which studies are best suited to inform your choices about which measure or assessment to use for diagnosing whether he might be depressed? Which treatment of depression is likely to have the greatest benefit with minimal harm? What is the prognosis for depression in adolescents? This article reviews study designs most appropriate to answer these research questions and discusses how different sources of data may influence study results. You are considering using a short, self-administered screening tool for depression in your pediatric practice. You wonder how well this instrument identifies depression compared with a more formal mental health evaluation. What study design would be most likely to yield valid information about the accuracy of the scale for detecting depression in your patient population? A cross-sectional study that compares a new diagnostic or screening test to the gold standard in the same population allows the researcher to assess its accuracy, including, for example, the sensitivity and specificity of the test. A good study should also consider how and when a test …