Skip to contents

Diagnostic Sample Size Planning Examples

Source: R/diagnostic_sample_size_examples_documentation.R
diagnostic_sample_size_examples.Rd

Format

A data frame with 6 rows and 13 variables:

scenario

Name of the diagnostic test scenario

population

Target population description

prevalence

Disease prevalence in target population (0-1)

target_sensitivity

Target sensitivity value (0-1)

target_specificity

Target specificity value (0-1)

ci_width

Desired 95\ study_purposeStudy purpose: "diagnostic", "screening_sens", or "screening_spec" expected_n_sensExpected sample size for sensitivity estimation expected_n_specExpected sample size for specificity estimation final_nFinal required sample size (maximum of sensitivity/specificity) notesClinical notes and justification nonresponse_rateExpected non-response rate (\ final_n_adjustedFinal sample size adjusted for non-response

Bujang MA (2023). An Elaboration on Sample Size Planning for Performing a One-Sample Sensitivity and Specificity Analysis by Basing on Calculations on a Specified 95\ doi:10.3390/diagnostics13081390 diagnostic_sample_size_examples Six clinical scenarios demonstrating sample size planning for diagnostic test accuracy studies using Clopper-Pearson exact binomial confidence intervals. Based on Bujang MA (2023) Diagnostics 13(8):1390. The dataset includes six diverse clinical scenarios: 1. Colorectal Cancer Blood Test

  • Population: High-risk patients (age >50, family history)

  • Prevalence: 10\

  • Purpose: Diagnostic (need excellent sensitivity AND specificity)

  • Required N: 940 subjects

2. COVID-19 Rapid Antigen Test

  • Population: General population (asymptomatic screening)

  • Prevalence: 5\

  • Purpose: Screening (emphasize sensitivity)

  • Required N: 1,880 subjects

3. AI-Based Diabetic Retinopathy Detection

  • Population: Diabetic patients

  • Prevalence: 30\

  • Purpose: Diagnostic with moderate precision

  • Required N: 147 subjects

4. Rare Disease Biomarker (Fabry Disease)

  • Population: Suspected patients referred to genetics clinic

  • Prevalence: 2\

  • Purpose: Diagnostic

  • Required N: 7,900 subjects (large due to low prevalence)

5. Lung Cancer LDCT Screening

  • Population: Heavy smokers (>30 pack-years)

  • Prevalence: 15\

  • Purpose: Screening (emphasize specificity to reduce false positives)

  • Required N: 4,020 subjects

6. Digital Mammography Screening

  • Population: Women age 50-70 (recalled for further testing)

  • Prevalence: 50\

  • Purpose: Diagnostic with moderate precision

  • Required N: 88 subjects

# Load the example scenarios data(diagnostic_sample_size_examples) # View all scenarios print(diagnostic_sample_size_examples[, c("scenario", "prevalence", "final_n", "final_n_adjusted")]) # Scenario 1: Colorectal cancer screening colorectal <- diagnostic_sample_size_examples[1, ] cat("Scenario:", colorectal$scenario, "") cat("Prevalence:", colorectal$prevalence * 100, "%") cat("Required N:", colorectal$final_n, "") cat("Adjusted N (20% non-response):", colorectal$final_n_adjusted, "") # Demonstrate impact of prevalence on sample size prevalence_impact <- diagnostic_sample_size_examples[, c("scenario", "prevalence", "final_n")] prevalence_impact <- prevalence_impact[order(prevalence_impact$prevalence), ] print(prevalence_impact) datasets