Dataset with 120 patients comparing perfect test (Sens=1.0, Spec=1.0) against imperfect test (Sens=0.85, Spec=0.88).
Format
A data frame with 120 rows and 4 variables:
- patient_id
Character: Patient identifier (PT001-PT120)
- GoldStandard
Factor: True status ("Negative", "Positive"), 30% positive
- PerfectTest
Factor: Perfect test ("Negative", "Positive"), Sens=1.0, Spec=1.0
- ImperfectTest
Factor: Imperfect test ("Negative", "Positive"), Sens=0.85, Spec=0.88
- age
Numeric: Patient age in years (mean 60, SD 11)
Details
Edge case demonstrating perfect test performance (100% agreement with gold standard) compared to realistic imperfect test.
Examples
data(decisioncompare_perfect)
decisioncompare(data = decisioncompare_perfect, gold = "GoldStandard",
goldPositive = "Positive", test1 = "PerfectTest",
test1Positive = "Positive", test2 = "ImperfectTest",
test2Positive = "Positive", test3Positive = "",
statComp = TRUE)
#>
#> COMPARE MEDICAL DECISION TESTS
#>
#> character(0)
#>
#> Test 1 - Recoded Data
#> ────────────────────────────────────────────────────────────────
#> Gold Positive Gold Negative Total
#> ────────────────────────────────────────────────────────────────
#> Test Positive 40.000000 0.000000 40.00000
#> Test Negative 0.000000 80.000000 80.00000
#> Total 40.000000 80.000000 120.00000
#> ────────────────────────────────────────────────────────────────
#>
#>
#> Test 2 - Recoded Data
#> ────────────────────────────────────────────────────────────────
#> Gold Positive Gold Negative Total
#> ────────────────────────────────────────────────────────────────
#> Test Positive 32.000000 8.000000 40.00000
#> Test Negative 8.000000 72.000000 80.00000
#> Total 40.000000 80.000000 120.00000
#> ────────────────────────────────────────────────────────────────
#>
#>
#> Test 3 - Recoded Data
#> ────────────────────────────────────────────────────────────
#> Gold Positive Gold Negative Total
#> ────────────────────────────────────────────────────────────
#> Test Positive . . .
#> Test Negative . . .
#> Total . . .
#> ────────────────────────────────────────────────────────────
#>
#>
#> Decision Test Comparison
#> ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
#> Test Sensitivity Specificity Accuracy Positive Predictive Value Negative Predictive Value Positive Likelihood Ratio Negative Likelihood Ratio
#> ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
#> PerfectTest 100.00000 100.00000 100.00000 100.00000 100.00000 Inf 0.01227087
#> → Excellent overall performance; Strong positive evidence; Strong negative evidence (zero cell; LR may be unstable)
#> ImperfectTest 80.00000 90.00000 86.66667 80.00000 90.00000 8.000000 0.22222222
#> → Good specificity for confirmation; Moderate positive evidence
#> ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
#>
#>
#> Stratified Diagnostic Accuracy
#> ───────────────────────────────────────────────────────────────────────────────────────────────────────────
#> Subgroup N Test Sensitivity Specificity Accuracy PPV NPV OPA
#> ───────────────────────────────────────────────────────────────────────────────────────────────────────────
#> ───────────────────────────────────────────────────────────────────────────────────────────────────────────
#>
#>
#> Statistical Comparison of Test Accuracy
#> ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
#> Comparison Chi-squared df p-value Clinical Interpretation
#> ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
#> PerfectTest vs ImperfectTest 14.06250 1 0.0001768 Highly significant difference (p<0.001) (Holm-Bonferroni corrected)
#> ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
#> Note. For 2 tests: McNemar's test compares diagnostic CORRECTNESS (agreement with gold standard) between paired tests. For 3+
#> tests: Cochran's Q test provides an overall test, followed by pairwise McNemar's tests with Holm-Bonferroni correction for multiple
#> comparisons. Tests examine discordant pairs (cases where one test is correct and the other is wrong relative to the gold standard)
#> to determine if differences in accuracy are statistically significant.
#>
#>
#> Differences with 95% Confidence Intervals
#> ────────────────────────────────────────────────────────────────────────────────────────
#> Comparison Metric Difference Lower Upper
#> ────────────────────────────────────────────────────────────────────────────────────────
#> PerfectTest vs ImperfectTest Sensitivity 20.00000 ᵃ 7.60410 32.39590
#> PerfectTest vs ImperfectTest Specificity 10.00000 ᵇ 3.42608 16.57392
#> PerfectTest vs ImperfectTest Accuracy 13.33333 ᵈ 7.25124 19.41542
#> ────────────────────────────────────────────────────────────────────────────────────────
#> ᵃ Small paired sample/discordant counts; CI may be unstable (n=40, discordant
#> counts: 32, 8, 0, 0).
#> ᵇ Small paired sample/discordant counts; CI may be unstable (n=80, discordant
#> counts: 72, 8, 0, 0).
#> ᵈ Small paired sample/discordant counts; CI may be unstable (n=120, discordant
#> counts: 104, 16, 0, 0).
#>
#>
#> <div style="font-family: Arial, sans-serif; max-width: 800px; margin:
#> 0 auto; padding: 20px;"><h2 style="color: #2c3e50; border-bottom: 2px
#> solid #3498db;"> Clinical Summary
#>
#> Among the tests evaluated, PerfectTest demonstrated optimal diagnostic
#> performance with 100% sensitivity (95% CI: [see confidence interval
#> table]), 100% specificity (95% CI: [see confidence interval table]),
#> 100% positive predictive value, 100% negative predictive value, and
#> 100% overall accuracy. Statistical comparisons using McNemar's test
#> revealed significant differences in test performance (detailed results
#> in comparison tables). The likelihood ratio for positive results was
#> Inf and for negative results was 0.01.<h3 style="color: #27ae60;
#> margin-top: 30px;"> Report Sentences
#>
#> <div style="background-color: #f8f9fa; padding: 15px; border-left: 4px
#> solid #28a745; margin: 15px 0;"><h4 style="margin-top: 0;">Methods
#> Section:
#>
#> <p style="font-style: italic; line-height: 1.6;">We compared the
#> diagnostic performance of 2 tests (PerfectTest, ImperfectTest) against
#> the gold standard reference using diagnostic accuracy analysis. The
#> study included 120 cases with complete data. Performance metrics
#> calculated included sensitivity, specificity, positive and negative
#> predictive values, likelihood ratios, and overall accuracy.
#> Statistical comparisons between tests were performed using McNemar's
#> test comparing diagnostic correctness (agreement with gold standard).
#>
#> <div style="background-color: #e8f4f8; padding: 15px; border-left: 4px
#> solid #3498db; margin: 15px 0;"><h4 style="margin-top: 0;">Results
#> Section:
#>
#> <p style="font-style: italic; line-height: 1.6;">Among the tests
#> evaluated, PerfectTest demonstrated optimal diagnostic performance
#> with 100% sensitivity (95% CI: [see confidence interval table]), 100%
#> specificity (95% CI: [see confidence interval table]), 100% positive
#> predictive value, 100% negative predictive value, and 100% overall
#> accuracy. Statistical comparisons using McNemar's test revealed
#> significant differences in test performance (detailed results in
#> comparison tables). The likelihood ratio for positive results was Inf
#> and for negative results was 0.01.
#>
#> <h3 style="color: #8e44ad; margin-top: 30px;"> Clinical
#> Recommendations
#>
#> <div style="background-color: #fff3cd; padding: 15px; border-radius:
#> 8px;">
#>
#> Clinical Use: PerfectTest shows excellent performance for both
#> screening and confirmatory testing.
#>
#> Implementation Note: Results should be interpreted in the context of
#> disease prevalence in your clinical population. Consider local
#> validation studies before implementation.
#>
#> <div style="font-family: Arial, sans-serif; max-width: 900px; margin:
#> 0 auto; padding: 20px;"><h2 style="color: #2c3e50; text-align: center;
#> border-bottom: 2px solid #3498db; padding-bottom: 10px;"> About
#> Medical Decision Test Comparison
#>
#> <div style="background: linear-gradient(135deg, #e3f2fd 0%, #bbdefb
#> 100%); padding: 20px; border-radius: 10px; margin: 20px 0;"><h3
#> style="color: #1565c0; margin-top: 0;"> What This Analysis Does
#>
#> <p style="line-height: 1.6; color: #333;">This tool compares the
#> diagnostic performance of multiple medical tests against a gold
#> standard reference. It systematically evaluates sensitivity,
#> specificity, predictive values, likelihood ratios, and overall
#> accuracy to help you determine which test performs best for your
#> clinical scenario.
#>
#> <div style="background-color: #f1f8e9; border: 1px solid #8bc34a;
#> padding: 20px; border-radius: 8px; margin: 20px 0;"><h3 style="color:
#> #4a7c59; margin-top: 0;"> When to Use This Analysis
#>
#> <ul style="line-height: 1.8; color: #4a7c59;">Test Validation:
#> Comparing new diagnostic methods against established standardsMethod
#> Comparison: Evaluating which of several tests performs betterClinical
#> Research: Validating biomarkers, imaging techniques, or clinical
#> assessmentsQuality Assessment: Measuring agreement between different
#> raters or methodsProtocol Development: Optimizing diagnostic
#> workflows<div style="background-color: #fff3e0; border: 1px solid
#> #ff9800; padding: 20px; border-radius: 8px; margin: 20px 0;"><h3
#> style="color: #e65100; margin-top: 0;"> How to Use This Analysis
#>
#> <ol style="line-height: 1.8; color: #e65100;">Select Gold Standard:
#> Choose your most reliable reference test (e.g., biopsy, expert
#> consensus)Choose Tests to Compare: Select 2-3 diagnostic tests you
#> want to evaluateDefine Positive Levels: Specify what constitutes a
#> "positive" result for each testConfigure Options: Enable statistical
#> comparisons, confidence intervals, or visualizations as neededRun
#> Analysis: Review results tables and clinical interpretationsCopy
#> Report: Use the auto-generated sentences for your documentation<div
#> style="background-color: #f3e5f5; border: 1px solid #9c27b0; padding:
#> 20px; border-radius: 8px; margin: 20px 0;"><h3 style="color: #6a1b9a;
#> margin-top: 0;"> Key Metrics Explained
#>
#> <div style="display: grid; grid-template-columns: 1fr 1fr; gap: 15px;
#> color: #6a1b9a;">
#>
#> Sensitivity: Probability test is positive when disease present
#> (rule-out ability)
#>
#> Specificity: Probability test is negative when disease absent (rule-in
#> ability)
#>
#> PPV: Probability of disease when test positive
#>
#> NPV: Probability of no disease when test negative
#>
#> LR+: How much positive test increases odds of disease
#>
#> LR-: How much negative test decreases odds of disease
#>
#> Accuracy: Overall probability of correct classification
#>
#> McNemar Test: Statistical comparison between paired tests
#>
#> <div style="background-color: #e8f5e8; border: 1px solid #4caf50;
#> padding: 20px; border-radius: 8px; margin: 20px 0;"><h3 style="color:
#> #2e7d32; margin-top: 0;"> Clinical Interpretation Guidelines
#>
#> <div style="display: grid; grid-template-columns: 1fr 1fr; gap: 15px;
#> color: #2e7d32;"><h4 style="margin-bottom: 5px;">Screening Tests
#> (Rule-Out):
#>
#> <p style="margin-top: 0;">• Sensitivity >=95%: Excellent
#> • NPV >=95%: High confidence
#> • Goal: Minimize false negatives
#>
#> <h4 style="margin-bottom: 5px;">Confirmatory Tests (Rule-In):
#>
#> <p style="margin-top: 0;">• Specificity >=95%: Excellent
#> • PPV >=90%: High confidence
#> • Goal: Minimize false positives
#>
#> <div style="background-color: #fff8e1; border: 1px solid #ffc107;
#> padding: 20px; border-radius: 8px; margin: 20px 0;"><h3 style="color:
#> #f57f17; margin-top: 0;"> Important Assumptions & Limitations
#>
#> <ul style="line-height: 1.6; color: #f57f17;">Gold Standard: Assumes
#> your reference test is truly accurateSample Size: Results more
#> reliable with larger, representative samplesPrevalence Dependency: PPV
#> and NPV vary with disease prevalenceMcNemar Test: Requires
#> paired/matched data for statistical comparisonsMissing Data: Cases
#> with incomplete data are excluded from analysisConfidence Intervals:
#> Calculated using Wilson method for better accuracy
#>
#> <div style='margin: 10px 0;'><div style='background-color: #eff6ff;
#> border-left: 4px solid #93c5fd; padding: 12px; margin: 8px 0;
#> border-radius: 4px;'><strong style='color: #2563eb;'> Zero Cell
#> Continuity Correction
#> <span style='color: #374151;'>Zero cell detected for PerfectTest.
#> LR+/LR- computed with a 0.5 continuity correction to avoid
#> infinite/undefined values; interpret cautiously.<div
#> style='background-color: #eff6ff; border-left: 4px solid #93c5fd;
#> padding: 12px; margin: 8px 0; border-radius: 4px;'><strong
#> style='color: #2563eb;'> Analysis Completed Successfully
#> <span style='color: #374151;'>2 diagnostic tests compared using 120
#> complete cases. Gold standard identified 40 diseased and 80 healthy
#> cases. Review comparison tables and statistical tests below.