DLNM Meta-Analysis

To pool DLNM estimates across multiple cities in R, this Skill uses the mixmeta or mvmeta packages to aggregate city-specific reduced coefficients and their variance-covariance matrices via a two-stage meta-analysis workflow.

A two-stage DLNM meta-analysis first fits city-specific exposure-lag-response models, then pools their reduced coefficients to derive an overall estimate and identify location-specific variations. You need it when analyzing epidemiological trends across multiple geographical datasets.

Yes, you can use mixmeta for meta-regression in a DLNM workflow. This Skill supports including location-level covariates in the meta-analysis model to investigate and explain heterogeneity across different geographical locations or datasets.

To compute BLUPs for location-specific DLNM estimates, this Skill calculates Best Linear Unbiased Predictions after pooling city-specific models. This provides refined estimates that borrow strength across studies to capture local variations.

You need R packages like dlnm and mixmeta for statistical modeling, with mvmeta as a potential alternative for pooling. The Skill requires these dependencies to properly analyze multi-city exposure-lag-response relationships.

When using BLUP estimation for multi-country epidemiological meta-analysis, limitations arise if location-level predictors are insufficient to explain heterogeneity. The accuracy of refined location-specific estimates depends heavily on the quality of the variance-covariance matrices from city-specific models.