Article Text

Download PDFPDF

1046 Analysis and a prediction model of pattern of visits to medical institutions among working individuals with lifestyle-related diseases in japan
  1. Go Muto1,2,
  2. Atsushi Goto3,
  3. Mitsuhiko Noda4,
  4. Motoki Endo5,
  5. Hiroshi Fukuda6,
  6. Ryoko Katagiri3,
  7. Kazuhito Yokoyama1
  1. 1Department of Epidemiology and Environmental Health, Juntendo University, Tokyo, Japan
  2. 2Harvard University, T. H. Chan School of Public Health, Boston, USA
  3. 3Metabolic Epidemiology Section, Division of Epidemiology, Centre for Public Health Sciences, National Cancer Centre, Tokyo, Japan
  4. 4Department of Endocrinology and Diabetes, Saitama Medical University, Saitama, Japan
  5. 5Department of Public health, Tokyo Womens Medical University, Tokyo, Japan
  6. 6Department of General Medicine, Graduate school of Medicine, Juntendo University, Tokyo, Japan


Introduction We analysed, developed, and evaluated a prediction model of the pattern of visits to medical institutions after annual health check-ups among Japanese individuals with hypertension (HT), diabetes mellitus (DM), or dyslipidemia (DL).

Methods Using claims and health check-up data maintained by the Japan Medical Data Centre from 2008 to 2016, we identified 5 33 955 individuals (20 to 74 years old) with HT, DM, or DL without claim data for the corresponding diseases for the 4 months prior to the health checkups. We calculated overall and disease-specific cumulative non-visit rates after health check-ups using Kaplan-Meier estimators. The prediction model was derived from randomly collected 1 17 671 individuals with risk factors selected by stepwise logistic regression, followed by validation on another data with the same number.

Results The overall cumulative non-visit rates at 3, 6, 9, and 12 months were 91.4%, 88.2%, 86.2%, and 84.4%, respectively. The disease-specific rates at 12 months were 84.3% for HT, 67.9% for DM, and 86.1% for DL. Limiting the analysis to those with extremely high blood pressure (BP) (systolic BP ≥160 mmHg), blood glucose levels (HbA1c ≥8.4%), or lipid levels (LDL-cholesterol levels ≥160 mg/dL) resulted in a slightly lower overall rate at 12 months (74.0%), with a relatively low rate for DM (51.9%). The prediction model, including factors such as age, working status, dietary habits, and the motivation of behavioural change showed modest discrimination ability for the pattern of visits to medical institutions (AUC=0.63), and well calibrated in validation data (Hosmer-Lemeshow p=0.38).

Conclusion Our study demonstrated that over 80% of Japanese individuals with lifestyle-related diseases did not visit medical institutions for 1 year after health check-ups. The pattern of visits to medical institutions model may be used for precision health counselling guidance for occupational health providers, which is a promising strategy to encourage individuals with lifestyle-related diseases to prevent diseases exacerbations.

  • Prevention of lifestyle-related diseases exacerbation
  • Behavioural change
  • Prediction model for precision medicine

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.