Rural–urban mortality gaps through a rurality dial

Contents
  1. What the study found
  2. Video
  3. Rurality dial
  4. How the dial converts the numbers
  5. Caveats
  6. Sources

On 1 July 2026, BMJ Public Health published our paper on how municipal rurality and area deprivation relate to cause-specific mortality in Japan (Koda M, et al. BMJ Public Health 2026;4:e004913).

The paper reports rate ratios per 1-SD (about 29 points) increase in the Rurality Index for Japan (RIJ). These numbers are precise but hard to picture, so I made a dial: turn the RIJ and the ratio for each cause of death moves.

What the study found

We analysed about 4.08 million deaths in 1,890 municipalities from 2017 to 2019. Geography was measured with the RIJ and socioeconomic disadvantage with the Area Deprivation Index (ADI). Both entered the same Bayesian spatial Poisson model for all-cause mortality and 34 specific causes.

After ADI adjustment, 14 outcomes, including traffic accidents, senility and acute myocardial infarction, showed higher mortality in more rural municipalities. Eleven, including tuberculosis and asthma, showed higher mortality in more urban municipalities. All-cause mortality was slightly higher with higher rurality (RR 1.010 per 1 SD, 95% CrI 1.003 to 1.016).

Video

A narrated walkthrough of about 90 seconds (in Japanese) on using the dial and reading the numbers.

Watch on YouTube

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

Rurality dial

Turn the dial or move the slider. The buttons jump to typical areas, and you can also choose the reference area. Set the reference to central city wards and the area to compare to remote islands and mountain towns to compare urban and rural areas directly. With the dial selected, the arrow keys move it by 1 point and Shift with an arrow key by 10.

UrbanRural
RIJ score50
Area to compare
Reference area
Area deprivation (ADI)

Higher than the referenceLower than the referenceValue under the other setting
  • All causes

Higher with higher rurality

  • Traffic accidents
  • Senility
  • Acute myocardial infarction
  • Unintentional injuries
  • Arrhythmias and conduction disorders
  • Gallbladder and biliary tract cancer
  • Heart failure
  • Suicide
  • Subarachnoid haemorrhage
  • Heart diseases (excl. hypertensive)
  • Cerebral infarction
  • Cerebrovascular diseases
  • Stomach cancer

Higher in more urban municipalities

  • Colon cancer
  • Oesophageal cancer
  • Pneumonia
  • Uterine cancer
  • Rectosigmoid cancer
  • Leukaemia
  • Liver disease
  • Liver cancer
  • Breast cancer
  • Asthma
  • Tuberculosis

No clear difference after ADI adjustment

  • Intracerebral haemorrhage
  • COPD
  • Pancreatic cancer
  • Lung cancer
  • All malignant neoplasms
  • Renal failure
  • Diabetes mellitus
  • Aortic aneurysm and dissection
  • Hypertensive diseases
  • Other ischaemic heart diseases

Values are ratios relative to the reference area you choose (an RIJ of 50 at first). They extend the paper’s rate ratios per 1 SD (28.8 points) of RIJ, assuming a log-linear relationship, and are not values reported in the paper. The bar scale widens or narrows to fit the largest difference at the current setting; circles beyond it sit at the edge. The circle shows the value under the other ADI setting.

How the dial converts the numbers

The dial shows each cause as a ratio to the reference area (an RIJ of 50 at first). A value of 1.18× means about 18% more deaths than in the reference area. It extends the per-SD rate ratio assuming a log-linear relationship (mean 50.2, SD 28.8). For traffic accidents (RR 1.114), with the reference at 50, an RIJ of 95 gives about 1.18× and an RIJ of 5 about 0.84×. With the reference at 5 and the dial at 95, traffic accidents are about 1.40× and tuberculosis about 0.62×. These are conversions, not values reported in the paper, and the assumption matters most near either end of the scale.

Switching off ADI adjustment shows the crude estimates. For all-cause mortality the ratio shrinks from 1.025 to 1.010 after adjustment. Suicide shrinks markedly (1.102 to 1.035), while acute myocardial infarction barely changes (1.053 to 1.052).

Caveats

  • This is a municipality-level ecological study and does not show individual risk or causation.
  • The dial shows point estimates only. The paper classified causes by whether the 95% credible interval excluded 1.
  • Values are for the total population. Some causes, such as suicide, differ by sex.

Sources

  • Koda M, Harada N, Nomura S, Tsugawa Y. Association of municipal rurality and area deprivation with cause-specific mortality in Japan: a nationwide ecological study. BMJ Public Health 2026;4:e004913. doi:10.1136/bmjph-2026-004913 (open access)
  • Values on the dial: article text and online supplemental table S3 (total population, with and without ADI adjustment).
English translation of the Japanese article.