Skip to contents

MetSS

Source: vignettes/articles/metss.Rmd
metss.Rmd

Scope

Compute the Metabolic Syndrome Severity Score (MetSSS) using sex- and race-specific coefficients, with configurable NA/extreme handling and plausibility diagnostics.

When to use

  • You have waist, blood pressure, fasting lipids/glucose, sex, and race and want a continuous MetS severity score.
  • You need built-in NA policies, high-missingness warnings, and optional extreme-value scanning/capping.
  • You are okay with a single parameter key (race/sex) applied to all rows—matching the original implementation.

Inputs

  • data: data.frame/tibble with waist (cm), bp_sys/bp_dia (mmHg), TG/HDL_c/glucose (mmol/L), sex (1=male, 2=female), race in {NHW, NHB, HW, HA} or accepted synonyms.
  • params: named list keyed by RACE_SEX (e.g., NHW_M), each with intercept and component vectors (waist, TG, HDL, glucose, MAP) containing mean, sd, coef.
  • na_action: keep (default) propagates NA; omit drops rows missing required inputs; error aborts; ignore/warn act like keep (warn also adds high-missingness warnings via na_warn_prop).
  • check_extreme: set TRUE to scan inputs; extreme_action (warn/cap/error/ignore/NA) controls handling. Defaults (cm, mmHg, mmol/L): waist 40–200; bp_sys 60–300; bp_dia 30–200; TG 0–20; HDL_c 0–5; glucose 0–40. Override with extreme_rules.
  • diagnostics: if TRUE (default), warns on implausible ranges and invalid sex/race encodings.
  • verbose: log progress and summary.

Quick start 

library(HealthMarkers)
library(tibble)

mets <- tibble::tibble(
  waist = c(95, 108),
  bp_sys = c(120, 138),
  bp_dia = c(80, 86),
  TG = c(1.5, 2.1),
  HDL_c = c(1.2, 0.9),
  glucose = c(5.5, 6.2),
  sex = c(1, 1),
  race = c("NHW", "NHW")
)

metss(
  data = mets,
  params = list(
    NHW_M = list(
      intercept = -2.344,
      waist   = c(mean = 94.0, sd = 12.4, coef = 0.846),
      TG      = c(mean = 1.5,  sd = 0.6,  coef = 0.701),
      HDL     = c(mean = 1.1,  sd = 0.3,  coef = -0.663),
      glucose = c(mean = 5.3,  sd = 0.6,  coef = 0.658),
      MAP     = c(mean = 97,   sd = 11,   coef = 0.466)
    )
  ),
  na_action = "keep"
)
#> # A tibble: 2 × 1
#>   MetSSS
#>    <dbl>
#> 1  -2.43
#> 2   1.01

Screen extremes and cap 

extreme <- mets
extreme$TG[2] <- 25       # high TG
extreme$bp_sys[1] <- 400  # out of range

metss(
  data = extreme,
  params = list(
    NHW_M = list(
      intercept = -2.344,
      waist   = c(mean = 94.0, sd = 12.4, coef = 0.846),
      TG      = c(mean = 1.5,  sd = 0.6,  coef = 0.701),
      HDL     = c(mean = 1.1,  sd = 0.3,  coef = -0.663),
      glucose = c(mean = 5.3,  sd = 0.6,  coef = 0.658),
      MAP     = c(mean = 97,   sd = 11,   coef = 0.466)
    )
  ),
  na_action = "warn",
  check_extreme = TRUE,
  extreme_action = "cap",
  diagnostics = TRUE,
  verbose = TRUE
)
#> # A tibble: 2 × 1
#>   MetSSS
#>    <dbl>
#> 1  0.109
#> 2 21.9

Missing data policy 

missing <- mets
missing$HDL_c[1] <- NA

metss(
  data = missing,
  params = list(
    NHW_M = list(
      intercept = -2.344,
      waist   = c(mean = 94.0, sd = 12.4, coef = 0.846),
      TG      = c(mean = 1.5,  sd = 0.6,  coef = 0.701),
      HDL     = c(mean = 1.1,  sd = 0.3,  coef = -0.663),
      glucose = c(mean = 5.3,  sd = 0.6,  coef = 0.658),
      MAP     = c(mean = 97,   sd = 11,   coef = 0.466)
    )
  ),
  na_action = "omit"
)
#> # A tibble: 1 × 1
#>   MetSSS
#>    <dbl>
#> 1   1.01

metss(
  data = missing,
  params = list(
    NHW_M = list(
      intercept = -2.344,
      waist   = c(mean = 94.0, sd = 12.4, coef = 0.846),
      TG      = c(mean = 1.5,  sd = 0.6,  coef = 0.701),
      HDL     = c(mean = 1.1,  sd = 0.3,  coef = -0.663),
      glucose = c(mean = 5.3,  sd = 0.6,  coef = 0.658),
      MAP     = c(mean = 97,   sd = 11,   coef = 0.466)
    )
  ),
  na_action = "warn"
)
#> # A tibble: 2 × 1
#>   MetSSS
#>    <dbl>
#> 1  NA   
#> 2   1.01

Outputs and expectations

  • Returns a tibble with MetSSS using a single race/sex key derived from the first row (warns if multiple keys present).
  • na_action = "omit" drops rows missing required inputs; otherwise missing inputs propagate NA into MetSSS.
  • Extreme handling can cap/NA/warn/error; diagnostics run independently and can still warn after capping.

Verbose diagnostics 

old_opt <- options(healthmarkers.verbose = "inform")
metss(
  data = mets,
  params = list(
    NHW_M = list(
      intercept = -2.344,
      waist   = c(mean = 94.0, sd = 12.4, coef = 0.846),
      TG      = c(mean = 1.5,  sd = 0.6,  coef = 0.701),
      HDL     = c(mean = 1.1,  sd = 0.3,  coef = -0.663),
      glucose = c(mean = 5.3,  sd = 0.6,  coef = 0.658),
      MAP     = c(mean = 97,   sd = 11,   coef = 0.466)
    )
  ),
  verbose = TRUE
)
#> metss(): preparing inputs
#> metss(): column map: waist -> 'waist', bp_sys -> 'bp_sys', bp_dia -> 'bp_dia', TG -> 'TG', HDL_c -> 'HDL_c', glucose -> 'glucose', sex -> 'sex', race -> 'race'
#> metss(): results: MetSSS 2/2
#> # A tibble: 2 × 1
#>   MetSSS
#>    <dbl>
#> 1  -2.43
#> 2   1.01
options(old_opt)

Tips

  • Confirm sex encoded as 1/2 and race matches a params key; only the first-row key is used.
  • Keep units consistent (waist cm, BP mmHg, lipids/glucose mmol/L); no unit conversion occurs here.
  • Tighten extreme_rules to your cohort before using extreme_action = "cap" or "error".
  • Leave diagnostics = TRUE during QA to catch implausible ranges; silence only if needed for tests.