Health Data Analytics
What analytics content the exam tests
Candidates must be fluent in measurement before they can interpret it. Know the data chain: source systems (EHR, claims, registries, surveys) to data governance (ownership, definitions, standardization, security) to data quality dimensions (accuracy, completeness, timeliness, consistency, validity) to analysis, display, and decision. Distinguish structure, process, and outcome measures, and always ask about numerator and denominator. Understand risk adjustment and why comparing raw rates across populations misleads.
For statistics, know sensitivity (true positives divided by all with the condition) versus specificity, positive predictive value (which depends on prevalence), reliability versus validity, and sampling error. Know common versus special cause variation and the tools that separate them: run charts and control charts, with signals such as a point beyond the limits, a shift of six or more consecutive points on one side of the centerline, or a trend. Know basic displays (Pareto, histogram, scatter) and that benchmarking compares you to external performance while a target is an internal goal.
Traps
- Confusing sensitivity with predictive value; PPV falls when prevalence falls.
- Reacting to common-cause variation as if special cause, which is tampering.
- Comparing rates built on different denominators, definitions, or case mix.
- Assuming a statistically significant change is a meaningful one.
Rule of thumb
Sensitivity = TP/(TP+FN). If a screen flags 90 of 100 patients who have the condition, sensitivity is 90%. If it also flags 90 of 900 patients without the condition, PPV = 90/180 = 50%. Always state the denominator before interpreting any rate.