Skip to contents

Calculate diagnostic likelihood ratios

Source: R/DiagnosticLikelihoodRatioTable.R
DiagnosticLikelihoodRatioTable.Rd

Calculates diagnostic likelihood ratios for categorical diagnostic results and binary outcomes. A likelihood ratio is the probability of a result among outcome-positive participants divided by its probability among outcome-negative participants; this is a diagnostic accuracy measure, not a nested-model test.

Usage

DiagnosticLikelihoodRatioTable(
  data,
  outcome_vars,
  predictor_vars,
  positive_level = NULL,
  predictor_positive_level = NULL,
  stratify_by = NULL,
  confidence_level = 0.95,
  continuity_correction = NULL,
  Relabel = TRUE
)

Arguments

data

A data frame.

outcome_vars

Character vector of binary outcome variable names.

predictor_vars

Character vector of categorical or binary diagnostic predictor names.

positive_level

Optional outcome-positive level specification: NULL, one level for all outcomes, or a named vector keyed by outcome variable.

predictor_positive_level

Optional positive-result specification for binary predictors: NULL, one level for all predictors, or a named vector.

stratify_by

Optional character vector of variables defining strata.

confidence_level

Confidence level for log-scale LR confidence intervals.

continuity_correction

Optional positive value added to all four calculation cells only if a zero cell occurs. NULL preserves zero and infinite estimates.

Relabel

Logical; use attached variable labels when available.

Value

A list containing Results, BinarySummary, compact and expanded gt tables (FormattedTable, LargeTable, BinaryFormattedTable, and BinaryLargeTable), and Metadata.

Details

Binary predictors produce sensitivity, specificity, LR+, and LR-. Predictors with more than two levels return one likelihood ratio for each result level. Numeric predictors with more than two observed values must be categorized first.

See also

PlotDiagnosticLRHeatmap() for matrix views and PlotDiagnosticLRForest() for estimate-and-interval views.

Examples

data(SampleData)
data(SampleVariableTypes)
df_Labelled <- RevalueData(SampleData, SampleVariableTypes)$RevaluedData
df_Labelled$DiagnosisBinary <- factor(df_Labelled$Diagnosis,
  levels = c("Control", "Impaired"))
lr <- DiagnosticLikelihoodRatioTable(df_Labelled, "DiagnosisBinary",
  c("sex", "Genotype"))
#> DiagnosisBinary: 'Impaired' treated as outcome-positive.
lr$FormattedTable



  
Diagnostic likelihood ratios
    
Stratum
      Outcome
      Diagnostic result
      N
      Case %
      Control %
      Likelihood ratio (95% CI)
    
All participants
DiagnosisBinary
Sex: Female
333
51.6
64.9
0.796 (0.639, 0.991)
All participants
DiagnosisBinary
Sex: Male
333
48.4
35.1
1.38 (1.05, 1.81)
All participants
DiagnosisBinary
Genotype: E2E2
333
0.0
0.8
0
All participants
DiagnosisBinary
Genotype: E2E3
333
7.7
12.4
0.621 (0.283, 1.36)
All participants
DiagnosisBinary
Genotype: E2E4
333
1.1
2.9
0.38 (0.0474, 3.05)
All participants
DiagnosisBinary
Genotype: E3E3
333
37.4
55.0
0.68 (0.509, 0.908)
All participants
DiagnosisBinary
Genotype: E3E4
333
45.1
26.9
1.68 (1.23, 2.28)
All participants
DiagnosisBinary
Genotype: E4E4
333
8.8
2.1
4.25 (1.43, 12.7)
PlotDiagnosticLRHeatmap(lr)
#> $DiagnosticLR

#> 
#> $DiagnosticMatrices

#> 
#> $DiagnosticLRData
#> # A tibble: 8 × 30
#>   Outcome         OutcomeLabel    OutcomePositiveLevel Predictor PredictorLabel
#>   <chr>           <chr>           <chr>                <chr>     <chr>         
#> 1 DiagnosisBinary DiagnosisBinary Impaired             sex       Sex           
#> 2 DiagnosisBinary DiagnosisBinary Impaired             sex       Sex           
#> 3 DiagnosisBinary DiagnosisBinary Impaired             Genotype  Genotype      
#> 4 DiagnosisBinary DiagnosisBinary Impaired             Genotype  Genotype      
#> 5 DiagnosisBinary DiagnosisBinary Impaired             Genotype  Genotype      
#> 6 DiagnosisBinary DiagnosisBinary Impaired             Genotype  Genotype      
#> 7 DiagnosisBinary DiagnosisBinary Impaired             Genotype  Genotype      
#> 8 DiagnosisBinary DiagnosisBinary Impaired             Genotype  Genotype      
#> # ℹ 25 more variables: PredictorLevels <int>, ResultLevel <chr>, Stratum <chr>,
#> #   N <int>, NCase <int>, NControl <int>, CaseWithResult <int>,
#> #   ControlWithResult <int>, CaseWithoutResult <int>,
#> #   ControlWithoutResult <int>, CaseProbability <dbl>,
#> #   ControlProbability <dbl>, LikelihoodRatio <dbl>, LRLowerCI <dbl>,
#> #   LRUpperCI <dbl>, Log2LR <dbl>, ZeroCell <lgl>, Corrected <lgl>,
#> #   RowId <chr>, RowLabel <chr>, .OriginalOrder <int>, CIExcludesOne <lgl>, …
#> 
#> $DiagnosticMatrixData
#> # A tibble: 16 × 14
#>    Stratum          Predictor PredictorLabel Outcome    OutcomeLabel ResultLevel
#>    <chr>            <chr>     <chr>          <chr>      <chr>        <chr>      
#>  1 All participants sex       Sex            Diagnosis… DiagnosisBi… Female     
#>  2 All participants sex       Sex            Diagnosis… DiagnosisBi… Male       
#>  3 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E2E2       
#>  4 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E2E3       
#>  5 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E2E4       
#>  6 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E3E3       
#>  7 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E3E4       
#>  8 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E4E4       
#>  9 All participants sex       Sex            Diagnosis… DiagnosisBi… Female     
#> 10 All participants sex       Sex            Diagnosis… DiagnosisBi… Male       
#> 11 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E2E2       
#> 12 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E2E3       
#> 13 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E2E4       
#> 14 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E3E3       
#> 15 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E3E4       
#> 16 All participants Genotype  Genotype       Diagnosis… DiagnosisBi… E4E4       
#> # ℹ 8 more variables: OutcomeCondition <chr>, Count <int>, Denominator <int>,
#> #   Probability <dbl>, CellLabel <chr>, MatrixRow <chr>, MatrixRowLabel <chr>,
#> #   HoverText <chr>
#>