Multilevel Modelling of Patient Satisfaction in Hospitals

This repository contains a detailed multilevel modeling analysis based on a fictitious dataset of patient satisfaction in hospitals. The project is structured as an R Markdown (.Rmd) file that walks through exploratory data analysis, model building, validation, and interpretation of a multilevel model using the lme4 package in R.

📁 Files

• Instructions.Rmd: Main analysis file containing code and interpretation.
• hospSatisfaction.Rda: Simulated dataset used for the analysis.

🧠 Project Overview

The goal is to examine how patient satisfaction varies across hospitals and what individual and institutional-level factors contribute to this variation. The dataset includes both patient-level (e.g., age, sex, readmission) and hospital-level variables (e.g., hospital type, area).

🛠 Methods

• Multilevel models (lmer from the lme4 package)
• Model comparison using ANOVA
• Visual diagnostics (residual plots, caterpillar plot, predicted vs. observed)

🧾 Key Findings

• A random intercept and random slope model (Model 7) best captures variation.
• 65.8% of satisfaction score variance is explained by differences between hospitals.
• Hospital type, area, sex, age, and readmission status significantly influence satisfaction.
• Remote hospitals and private hospitals generally show higher satisfaction levels.

📊 Visual Outputs

• Histograms and boxplots of satisfaction across categories
• Residual analysis
• Caterpillar plots for random effects
• Observed vs. predicted plot for model accuracy

🔍 Citation Reference

Liu, M., et al. (2023). Patient healthcare experiences of cancer hospitals in China: A multilevel modeling analysis based on a national survey. Frontiers in Public Health, 11, 1059878. DOI link

👨‍💻 Author

Parth Chandan
Master of Public Health | Health Data Science
LinkedIn