retrospective<\/a> so faster than cohort<\/td>\n<\/tr>\n\n| Quality of evidence<\/td>\n | Lower, hypothesis-generating<\/td>\n | Higher, supports temporal sequence<\/td>\n | Moderate, good for rare outcomes<\/td>\n<\/tr>\n |
\n| Sample research question<\/td>\n | How common is condition X right now in population Y?<\/td>\n | Does exposure X predict outcome Y over time?<\/td>\n | Is prior exposure X more common among people with outcome Y than without it?<\/td>\n<\/tr>\n |
\n| Sample size needs<\/td>\n | Moderate to large for stable prevalence estimates<\/td>\n | Large, especially for rare outcomes<\/td>\n | Can be smaller, efficient for rare outcomes<\/td>\n<\/tr>\n |
\n| Analytical methods<\/td>\n | Prevalence, chi-square, cross-sectional regression<\/td>\n | Incidence rates, survival analysis, relative risk<\/td>\n | Odds ratios, conditional logistic regression<\/td>\n<\/tr>\n |
\n| Watch out for in data and analysis<\/td>\n | Confounding, reverse causation, non-response<\/td>\n | Loss to follow-up, changing exposure status<\/td>\n | Recall bias, choice of comparable controls<\/td>\n<\/tr>\n |
\n| Most common bias<\/td>\n | Selection and recall bias<\/td>\n | Attrition bias<\/td>\n | Recall bias and selection of controls<\/td>\n<\/tr>\n |
\n| Can estimate incidence?<\/td>\n | No<\/td>\n | Yes<\/td>\n | No, only odds<\/td>\n<\/tr>\n |
\n| Typical use case<\/td>\n | Prevalence surveys, needs assessments<\/td>\n | Risk factor studies, natural history of disease<\/td>\n | Rare disease or outcome investigations<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n <\/p>\n |