Diagnostic Sample Size Planning Examples — diagnostic_sample_size

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.

Usage

diagnostic_sample_size_examples

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% CI width
study_purpose: Study purpose: "diagnostic", "screening_sens", or "screening_spec"
expected_n_sens: Expected sample size for sensitivity estimation
expected_n_spec: Expected sample size for specificity estimation
final_n: Final required sample size (maximum of sensitivity/specificity)
notes: Clinical notes and justification
nonresponse_rate: Expected non-response rate (%)
final_n_adjusted: Final sample size adjusted for non-response

Source

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% Confidence Interval Width. Diagnostics 13(8):1390. doi:10.3390/diagnostics13081390

Details

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% (very low!)
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% (enriched population)
Purpose: Diagnostic with moderate precision
Required N: 88 subjects

Examples

# 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 prevalence final_n
#> 1                 Colorectal Cancer Blood Test       0.10     940
#> 2                  COVID-19 Rapid Antigen Test       0.05    1880
#> 3      AI-Based Diabetic Retinopathy Detection       0.30     147
#> 4 Rare Disease Biomarker (e.g., Fabry Disease)       0.02    7900
#> 5                   Lung Cancer LDCT Screening       0.15    4020
#> 6                Digital Mammography Screening       0.50      88
#>   final_n_adjusted
#> 1             1175
#> 2             2212
#> 3              164
#> 4            10534
#> 5             5025
#> 6              104

# Scenario 1: Colorectal cancer screening
colorectal <- diagnostic_sample_size_examples[1, ]
cat("Scenario:", colorectal$scenario, "\n")
#> Scenario: Colorectal Cancer Blood Test 
cat("Prevalence:", colorectal$prevalence * 100, "%\n")
#> Prevalence: 10 %
cat("Required N:", colorectal$final_n, "\n")
#> Required N: 940 
cat("Adjusted N (20% non-response):", colorectal$final_n_adjusted, "\n")
#> Adjusted N (20% non-response): 1175 

# 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)
#>                                       scenario prevalence final_n
#> 4 Rare Disease Biomarker (e.g., Fabry Disease)       0.02    7900
#> 2                  COVID-19 Rapid Antigen Test       0.05    1880
#> 1                 Colorectal Cancer Blood Test       0.10     940
#> 5                   Lung Cancer LDCT Screening       0.15    4020
#> 3      AI-Based Diabetic Retinopathy Detection       0.30     147
#> 6                Digital Mammography Screening       0.50      88