1 Packages

Code

pacman::p_load(tidyverse, gtsummary, skimr, 
               broom, sjPlot)

2 Dataset

3 EDA

Code

# df |> 
#   skimr::skim()

Code

df <- df |> 
  select(`Vecums (gados)`, 
         Dzimums, 
         Reģions, 
         `Piena molāru hipomineralizācija`, 
         `Skolas kods (divi cipari)`, 
         Prevalence_primary_d5, 
         d5mfs_primary)

Code

df <- df |> 
  mutate(
    `Prevalence_primary_d5` = as.factor(`Prevalence_primary_d5`),
    `Reģions` = as.factor(`Reģions`),
    `Dzimums` = as.factor(`Dzimums`),
    `Piena molāru hipomineralizācija` = as.factor(`Piena molāru hipomineralizācija`),
    `Skolas kods (divi cipari)` = as.factor(`Skolas kods (divi cipari)`)
  )

3.1 Data cleaning

Code

df <- df |> 
  mutate(
    `Vecums (gados)` =
      str_trim(`Vecums (gados)`) %>%   # remove spaces
      as.character() %>%               # ensure character
      str_replace("^0+", "") %>%       # remove leading zeros (07 -> 7)
      as.numeric()
  )

Code

df <- df |> 
  mutate(
    `Skolas kods (divi cipari)` =
      as.character(`Skolas kods (divi cipari)`),
    
    `Skolas kods (divi cipari)` =
      if_else(
        `Skolas kods (divi cipari)` == "a",
        "01",
        `Skolas kods (divi cipari)`
      ),
    
    `Skolas kods (divi cipari)` =
      as.factor(`Skolas kods (divi cipari)`)
  )

4 Models

4.1 Prevalence

Code

model_prevalence <- glm(
  `Prevalence_primary_d5` ~ 
    `Vecums (gados)` +
    `Reģions` +
    `Dzimums` +
    # `Skolas kods (divi cipari)` +
    `Piena molāru hipomineralizācija` ,
    
  data = df,
  family = binomial(link = "logit")
)

Code

tbl_regression(
  model_prevalence,
  exponentiate = TRUE
) |> 
  add_n(location = "level") |> 
  bold_labels()

Characteristic	N	OR	95% CI	p-value
Vecums (gados)	822	0.89	0.59, 1.33	0.6
Reģions
Pierīga	407	—	—
Rīga	415	0.77	0.57, 1.05	0.10
Dzimums
Meitene	412	—	—
Zēns	410	0.78	0.58, 1.06	0.12
Piena molāru hipomineralizācija
Jā	33	—	—
Nē	789	0.91	0.44, 2.03	0.8
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

Code

plot_model(
  model_prevalence,
  type = "est",
  transform = "exp",
  show.values = TRUE,
  value.offset = 0.3
)

4.2 Severity

Code

model_d5mfs <- glm(
  `d5mfs_primary` ~ 
    `Vecums (gados)` +
    `Reģions` +
    `Dzimums` +
        # `Skolas kods (divi cipari)` +
    `Piena molāru hipomineralizācija` ,
  data = df,
  family = gaussian()
)

Code

tbl_regression(
  model_d5mfs
) |> 
  add_n(location = "level") |> 
  bold_labels()

Characteristic	N	Beta	95% CI	p-value
Vecums (gados)	822	1.9	0.13, 3.7	0.036
Reģions
Pierīga	407	—	—
Rīga	415	0.06	-1.3, 1.4	>0.9
Dzimums
Meitene	412	—	—
Zēns	410	1.6	0.24, 3.0	0.021
Piena molāru hipomineralizācija
Jā	33	—	—
Nē	789	1.4	-2.1, 4.9	0.4
Abbreviation: CI = Confidence Interval

Code

plot_model(
  model_d5mfs,
  type = "est",
  show.values = TRUE,
  value.offset = 0.3
)