Tidy Tools for Joinpoint Regression Models

Provides tools to fit joinpoint regression models with a log-linear specification by levels of one or two categorical variable(s) using the grid-search method. It includes functions to estimate the Annual Percent Change (APC) and the Average Annual Percent Change (AAPC), along with their 95% confidence intervals, and to generate formatted summary tables and plots of results.

joinpointR 2.0

🇬🇧 English

The goal of joinpointR is to fit joinpoint regression models by groups and generate tidy summaries of the Annual Percent Change (APC) and the Average Annual Percent Change (AAPC), facilitating trend analysis in epidemiological studies.

Installation

The development version of joinpointR can be installed from Github using the command:

remotes::install_github("https://github.com/datos-ine/joinpointR")

Workflow

The package provides a simple and reproducible workflow:

• Fit joinpoint models by group using the grid-search method.
• Generate summary tables with the fitted joinpoints, annual percent change (APC) and its confidence interval (CI), and average annual percent change (AAPC) and its CI.
• Optionally, extract the APC, AAPC and Bayesian Information Criteria of a single model or a list of models.
• Generate summary plots.

Main functions

• model_jp_grid() or model_jp() → Fits joinpoint regression models by groups of up to two categorical variables.
• get_summary(), get_apc(), and get_aapc() → Returns a table with summary statistics for a model or a list of models of class "model_jp".
• gg_jpoint() → Generate summary plots for a model or a list of models of class "model_jp".

Example

# Load packages
library(joinpointR)

# Load data
data(hiv_data)

data_mod <- hiv_data |>
    dplyr::filter(admin == "ARG")

# Fit the joinpoint models
mods <- model_jp_grid(data = data_mod, rate = "hiv_rate", time = "year", group = "sex")

# BIC of the model
bic_jp(mods)

# Summary table
get_summary(mods)

# Plot results
gg_jpoint(mods)

Notes

• The response variable is log-transformed.
• Model selection is based on the Bayesian Information Criterion (BIC). It can be changed to the penalized BIC (BIC3) or the weighted BIC (WBIC) using the argument method.
• Summary table results are returned in tidy format and can be transformed to flextable objects using the argument as.ft = TRUE.

🇪🇸 Español

El objetivo de joinpointR es ajustar modelos de regresión joinpoint por grupos y generar resúmenes en formato tidy del Cambio Porcentual Anual (APC) y del Cambio Porcentual Anual Promedio (AAPC), facilitando el análisis de tendencias en estudios epidemiológicos.

Instalación

La versión en desarrollo de joinpointR se puede descargar desde Github con el comando:

remotes::install_github("https://github.com/datos-ine/joinpointR")

Flujo de trabajo

El paquete cuenta con un flujo de trabajo simple y reproducible:

• Ajusta modelos de regresión joinpoint usando el método de grid-search.
• Genera tablas de resumen con los joinpoints detectados, el cambio porcentual anual (APC) y su intervalo de confianza (IC), y el cambio porcentual anual promedio (AAPC) y su IC.
• Opcionalmente, se puede extraer el APC, AAPC y Criterio de Información Bayesiano de un modelo o lista de modelos.
• Genera gráficos de resumen.

Funciones principales

• model_jp_grid() y model_jp() → Ajustan modelos de regresión joinpoint según niveles de hasta dos variables categóricas.
• get_summary(), get_apc(), y get_aapc() → Devuelven una tabla resumen para un modelo o lista de modelos de clase "model_jp".
• gg_jpoint() → Genera gráficos de resumen para un modelo o lista de modelos de clase "model_jp".

Ejemplo

# Cargar paquetes
library(joinpointR)

# Cargar datos
data(hiv_data)

data_mod <- hiv_data |>
    dplyr::filter(admin == "ARG")

# Ajustar modelos
mods <- model_jp_grid(data = data_mod, rate = "hiv_rate", time = "year", group = "sex")

# BIC
bic_jp(mods)

# Tabla resumen
get_summary(mods)

# Graficar resultados
gg_jpoint(mods)

Notas

• La variable respuesta se transforma logarítmicamente.
• La selección de modelos está basada en el Criterio de Información Bayesian(BIC). Se puede cambiar al BIC penalizado (BIC3) o BIC ponderado (WBIC) usando el argumento method.
• Las tablas de resumen se presentan en formato tidy y pueden transformarse a objetos flextable usando el argument as.ft = TRUE.

Licence / Licencia

MIT License

Author / Autora

Tamara Ricardo

Instituto Nacional de Epidemiología (INE), Argentina

ORCID: https://orcid.org/0000-0002-0921-2611