Analysis code for: "Trends in the Incidence of Childhood Type 1 Diabetes After the COVID-19 Lockdown in Finland: An Observational Study"

Description This repository contains the analysis scripts, statistical code, and aggregated data supporting the manuscript: "Trends in the Incidence of Childhood Type 1 Diabetes After the COVID-19 Lockdown in Finland: An Observational Study". Overview The resources provided here enable the replication of the statistical analyses investigating the incidence rates, longitudinal trends, clinical/metabolic/immunological characteristics, and SARS-CoV-2 antibody and vaccination profiles of children diagnosed with type 1 diabetes across distinct pandemic timelines (before, during, and after the lockdown). Contents & Methods Covered The repository includes the files and scripts used to execute the following analytical steps: Aggregated Data Input: Aggregated Excel data files containing case counts, person-years (PY), and demographic groups. These datasets serve as the direct inputs for the Poisson regression modeling pipelines. Incidence Rate Calculations: Computation of overall, sex-specific, and age-specific (<5 years, 5 to <10 years, and 10 to <15 years) type 1 diabetes incidence rates per 100,000 person-years (PY) utilizing annual population statistics obtained from Statistics Finland. Poisson Regression Models: Multiplicative Poisson regression models applied to aggregated case counts to calculate Incidence Rate Ratios (IRRs) and 95% confidence intervals (CIs) across the pandemic timelines, using the natural logarithm of person-years as an offset term and implementing robust standard errors via the sandwich method (Huber-White estimator). Sensitivity Analyses: Code for the temporal sensitivity analysis comparing post-lockdown incidence rates directly to the most recent preceding 18-month baseline period (March 1, 2018, to August 31, 2019). SARS-CoV-2 Antibody & Vaccination Analyses (R Scripts): S4 Figure Pipeline: Classification and plotting of SARS-CoV-2 infection and vaccination status ('infection', 'no infection', or 'ambiguous') among children diagnosed with type 1 diabetes post-lockdown (09/2021–12/2022) using strict and loose sensitivity boundaries, excluding children sampled <14 days post-vaccination. S5 Figure Pipeline: Graphical and mathematical modeling of longitudinal antibody prevalence (IgG N and IgG S) and vaccination rates between September 1, 2021, and December 31, 2022. This encompasses sliding monthly averages, kinetic modeling of IgG N antibody waning over time (116-day half-life), and 4-fold titer differential tracking (Wuhan vs. non-Wuhan strains) to distinguish vaccine immunity from natural infection profiles. Cohort Characteristic & Confounder Analyses: Linear regression models evaluating continuous variables (age at diagnosis, plasma glucose). Quantile regression models for medians (HbA1c%, HbA1c mmol/mol, blood pH, plasma β-hydroxybutyrate, relative unit autoantibodies). Logistic regression models for dichotomous response outcomes. Multinomial logistic regression models for categorical variables (duration of symptoms, number of family members with type 1 diabetes). Covariate adjustments controlling for the confounding effects of sex and age at diagnosis. Software Requirements The underlying statistical pipelines were developed and executed across the following environments: R Software for Statistical Computing (version 4.4.1) via the R Project SAS 9.4 & SAS Enterprise Guide for Windows (version 7.1) via SAS Support Microsoft Excel (for source data structure review) Note: Due to strict data privacy and legal regulations surrounding clinical health metrics, raw individual-level participant data cannot be made publicly available in this repository. The code is fully structured to run on the aggregated datasets provided or using simulated data structures matching the original schemas.