Screening Programme Analytics: PPV by Prevalence, Yield, Cost per Case & Biases
Analyse a (fictional) cancer screening programme the way a public-health analyst does: build the 2×2 table from programme data, see how PPV moves with prevalence, pick a cut-off with Youden's J and colonoscopy capacity, combine two tests, measure yield, NNS, uptake gaps and cost per QALY, and put numbers on lead-time and length-time bias, every cell checked live.
The first lesson is free — no account needed.
About this course
A screening programme analytics course that analyses one fictional bowel cancer screening programme, Northvale, on synthetic data in a spreadsheet, for public health analysts and students, epidemiology learners, programme managers and health economists. You build a 2×2 table from district reports with sensitivity, specificity, PPV and NPV, show with natural frequencies how PPV rises with prevalence across age bands, pick a test cut-off with Youden's J and the colonoscopy capacity, combine two tests in series, measure the detection rate, the number needed to screen and colonoscopies per cancer, size uptake gaps between deprivation groups in a chart, calculate the cost per cancer detected and a simple cost per QALY, and put numbers on lead-time bias, length-time bias and the false-positive burden over many rounds. Synthetic data, educational only, not medical advice. The first lab is free.
Lessons
- From Programme Data to a 2×2 Table: Sensitivity, Specificity, PPV & NPVFree
- PPV & NPV Across Target Groups: Prevalence Changes Everything (with a Chart) Pro
- Choosing the Cut-off: ROC Points, Youden Index & Colonoscopy Capacity Pro
- Two-Stage Screening: Serial Testing & Combined Accuracy Pro
- Programme Yield: Detection Rate, Number Needed to Screen & Colonoscopies per Cancer Pro
- Uptake & Coverage by Group: Measuring Equity Gaps (with a Chart) Pro
- Cost per Case Detected & a Simple Cost per QALY Pro
- Lead-Time & Length-Time Bias, Interval Cancers & the False-Positive Burden Pro