Maps of 35 causes of death, one click apart

Contents
  1. Video
  2. What you can do
  3. The three metrics
  4. Caveats
  5. Sources

Our paper in BMJ Public Health (Koda M, et al. BMJ Public Health 2026;4:e004913) has online supplemental figures S1–S35, one for each of 35 causes of death. Each figure holds nine municipal maps: total, male and female in rows, and three metrics in columns.

I turned these 315 maps into a viewer where you choose a cause, sex and metric and see one map at a time: Cause-specific mortality maps. The data and scripts are on GitHub.

It pairs with the article Rural–urban mortality gaps through a rurality dial.

Video

A narrated walkthrough of about 50 seconds (in Japanese).

Watch on YouTube

The narration is synthesised speech: the voice Ayaka from the ElevenLabs voice library, with the Eleven v4 model.

What you can do

  • Choose all-cause mortality or one of 34 causes, or step through them with the previous and next buttons.
  • Choose total, male or female.
  • Choose one of three metrics.
  • Hover over the map to see the municipality and its value class, or search by name to zoom to it.
  • Scroll or pinch to zoom, drag to pan and double-click to reset.

The current selection is kept in the address, so you can share it. The button at the top right switches to Japanese.

The three metrics

  • EB-smoothed SMR: observed deaths relative to the number expected from national age-specific rates (100 = expected), smoothed with empirical Bayes to reduce noise in small municipalities.
  • RIJ residual RR: the part of the SMR not explained by a log-linear relationship with the Rurality Index for Japan (RIJ). Above 1 means more deaths than rurality alone would predict.
  • Adjusted BYM2 residual RR: the municipal effect left after the main BYM2 spatial model with RIJ and the Area Deprivation Index (ADI). Above 1 means more deaths than the model predicts.

Switching between the three for the same cause shows how much of the geographical pattern rurality and deprivation account for, and what remains.

Caveats

  • Values are shown as 12 colour classes over the same range as the supplemental figures. Exact values are not published.
  • As in the figures, the colour range is the 3rd–97th (SMR) or 4th–96th (residuals) percentile pooled across all causes and sexes. Values outside it take the end colours.
  • Municipal RIJ scores are not published.
  • Areas without data are grey. Hokkaido and Okinawa are shifted, as in the figures.
  • This is a municipality-level ecological study and does not show individual risk or causation.

Sources

English translation of the Japanese article.