Abstract
Using two tables from the long-term care questionnaire of the 2025 Comprehensive Survey of Living Conditions — "level of care needed now × level one year ago" (Table 28) and "change in the level of care needed over the past year" (Table 29) — how care levels moved in one year was counted from two sides, from one year ago and from now, and lined up. It claims no new mathematical theorem and no new law.
First, the tables are consistent. The discrepancies in column sums, row sums, and men + women versus total are all 2 or less per 100,000 (the rounding width) (Section 3).
Second, Table 29's "became heavier 29,556, no change 60,007, became lighter 6,471, unknown 3,966" matches the collapsed Table 28 to within 6. "No change" and "became lighter" are exactly the movements within the seven levels (Support Level 1 to Care Level 5), but "became heavier" contains, besides the 16,586 who moved up a level, 12,972 who a year ago had been assessed as independent or had not applied for certification. Of those who became heavier, 0.439 were not certified a year ago (Section 4).
Third, therefore, "heavier ÷ lighter" is 4.57 in Table 29 as is, but 2.56 among those also certified a year ago. At every age the share becoming heavier is more than twice the share becoming lighter, and the share becoming heavier is largest at 65–69 (0.392) (Sections 4 and 5).
Fourth, this is the core of the paper. Counted by the level one year ago, the share moving up falls from 0.254 at Support Level 1 to 0.085 at Care Level 4, the share moving down grows to 0.176 at Care Level 3 and 0.231 at Care Level 4, and at Care Levels 3, 4, and 5 the share moving down exceeds the share moving up (Section 6).
Fifth, at every level, the share staying at the same level is larger counted from one year ago than counted from now. The gap is largest at Care Level 5: 0.904 from one year ago and 0.629 from now. Of those now at Support Level 1, 0.232 were not certified a year ago (Sections 7 and 8).
Sixth, what separates them is whether one counts from one year ago or from now, and whether the person was certified a year ago. The table keeps only those who, in the survey month, still need care and live in a household; people who died or moved into a facility during the year are not in it.
Scope note: This paper claims no new mathematical theorem and no new law — the two tables are public, and this paper only turned them into shares, lined them up, and collapsed one to compare it with the other. This paper is a structural interpretation of the shape of numbers in public aggregate data; it is not medical advice, not a diagnosis, and not a treatment recommendation. The author is not a physician. Decisions about individual health should be discussed with a physician or a medical institution. This is a preprint that has not been peer reviewed. It does not go into ways to lighten care levels or into whether care is good or bad. The Comprehensive Survey of Living Conditions is a sample survey, and its figures are estimates. The figures in this paper are "per 100,000 persons needing care," not numbers of people. The care level is a survey answer, not the certification record itself. Residents of social welfare facilities and long-term hospital inpatients are excluded from the survey (according to the survey outline). The table keeps only those who, in the survey month (June), live in a household and need care; people who died, moved into a facility, or lost certification during the year are not in it. At levels where the share moving down looks large, whether that is because people really became lighter or because people who became heavier left the table cannot be separated with this table. Considering that family caregivers and the people receiving care may read it, it is not written as an outlook for any one person. About the data: the Comprehensive Survey of Living Conditions (e-Stat) is used under the Government Standard Terms of Use (Version 2.0), and the shares and ratios in this paper were created by processing it. What this paper adds: dividing "became heavier over the year" into those who moved up a level and those not certified a year ago, confirming that the two tables match when collapsed, and showing that "the share staying at the same level" differs by 0.071–0.275 depending on whether it is counted from one year ago or from now.
1. Notation and question
- The seven levels: Support Levels 1 and 2 and Care Levels 1–5, ordered from lightest (in tables, Support 1–2 and Care 1–5).
- Counting from one year ago: for each level one year ago, with the number whose current level is known as the denominator, the shares that moved up, stayed, and moved down.
- Counting from now: for each current level, with that level's total as the denominator, the shares whose level one year ago was the same, lighter, heavier, not certified, or other (applying for certification a year ago, or unknown).
- Not certified a year ago: in Table 28, the row for one year ago is "assessed as independent" or "had not applied for certification."
- Per 100,000: the number when the total of persons needing care is set to 100,000.
Question: How much does the care level move in one year? What does a number like "people who became heavier are so many times those who became lighter" count?
Observation 1.1: M058 compared how fast the share certified as needing care (or support) doubles with age against the death rate at the same ages. M059 lined up, from the same 2025 care questionnaire, the causes for needing care. This paper reads, from the same questionnaire, the table of how each person's level moved in one year.
2. Survey and design
To be honest: the design was decided after seeing the tables.
| Content | |
|---|---|
| Tables | 2025 Comprehensive Survey of Living Conditions, long-term care, Table 28 (level now × sex × level one year ago) and Table 29 (change over the year × sex × age group) |
| Noticed in the survey | Table 29's "no change 60,007" and "became lighter 6,471" were exactly the diagonal sum and the below-diagonal sum over the seven levels in Table 28. So the content of "became heavier" was checked with Table 28 |
| Measures | Matching the collapsed numbers, the shares counted from one year ago and from now, and the gap in the share staying at the same level |
| Checked | The survey coverage (the Ministry's "survey outline"): the care questionnaire excludes persons absent from the household, such as residents of social welfare facilities and long-term hospital inpatients, and is conducted in June |
3. Data and checks
Calculation 3.1 The 3 retrieved files were fixed by SHA-256, and the following was checked.
- The titles and headings of Tables 28 and 29 (Table 28's columns are total, persons needing support (total, Support Levels 1 and 2), persons needing care (total, Care Levels 1–5), and care level unknown; Table 29's columns are total, became heavier, no change, became lighter, and unknown). Rows for total, men, and women are complete, and the total of the total row is 100000 in both.
- Table 28: column- and row-sum discrepancies are at most 2, and men + women versus total at most 1. Table 29: discrepancies in column, age, and restated sums are at most 2, and men + women at most 1. All are within the rounding width of per-100,000 figures.
- On the survey outline page, it was confirmed that "residents of social welfare facilities" and "long-term hospital inpatients" are excluded from the care questionnaire, and that it is conducted in June.
4. What "became heavier" contains — the two tables match when collapsed
Calculation 4.1 Table 28 collapsed, and Table 29 (total, per 100,000)
| Table 28 collapsed | Number | Table 29 | Number |
|---|---|---|---|
| Moved up a level + not certified a year ago | 16,586 + 12,972 = 29,558 | Became heavier | 29,556 |
| Same level | 60,007 | No change | 60,007 |
| Moved down a level | 6,471 | Became lighter | 6,471 |
| Remainder | 3,964 | Unknown | 3,966 |
Observation 4.2: Table 29's four numbers match the collapsed Table 28 to within 6 for total, men, and women alike. Table 29's "became heavier" contains, besides the 16,586 who moved up a level, 661 who had been assessed as independent a year ago and 12,311 who had not applied for certification. Of those who became heavier, 0.439 were not certified a year ago and 0.561 moved up a level.
Calculation 4.3 "Heavier ÷ lighter" under the two ways of counting
| Way of counting | Denominator | Heavier (up) | No change (same) | Lighter (down) | Heavier ÷ lighter |
|---|---|---|---|---|---|
| Table 29 as is (excluding unknown) | 96,034 | 0.308 | 0.625 | 0.067 | 4.57 |
| At one of the seven levels both a year ago and now | 83,064 | 0.200 | 0.722 | 0.078 | 2.56 |
Observation 4.4: "Those who became heavier are 4.57 times those who became lighter" becomes 2.56 times among those also certified a year ago. The share staying at the same level also goes from 0.625 to 0.722.
5. Change by age
Calculation 5.1 Shares by age in Table 29 (total; denominator excludes unknown)
| Age | Became heavier | No change | Became lighter |
|---|---|---|---|
| 40–64 | 0.312 | 0.560 | 0.129 |
| 65–69 | 0.392 | 0.557 | 0.052 |
| 70–74 | 0.272 | 0.650 | 0.077 |
| 75–79 | 0.349 | 0.590 | 0.061 |
| 80–84 | 0.348 | 0.585 | 0.067 |
| 85–89 | 0.287 | 0.659 | 0.054 |
| 90+ | 0.276 | 0.649 | 0.075 |
Observation 5.2: At every age, the share becoming heavier is more than twice the share becoming lighter. The share becoming heavier is largest at 65–69 (0.392) and smallest at 70–74 (0.272), and does not line up in one direction with age. As Section 4 showed, these numbers also include the newly certified.
6. Core — counting from one year ago
Calculation 6.1 Level now, by level one year ago (total; denominator is the number whose current level is known)
| Level one year ago | Denominator (per 100,000) | Moved up | Same level | Moved down |
|---|---|---|---|---|
| Support 1 | 16,513 | 0.254 | 0.746 | 0.000 |
| Support 2 | 16,351 | 0.215 | 0.724 | 0.061 |
| Care 1 | 17,811 | 0.239 | 0.702 | 0.059 |
| Care 2 | 16,122 | 0.189 | 0.717 | 0.095 |
| Care 3 | 7,861 | 0.144 | 0.680 | 0.176 |
| Care 4 | 5,166 | 0.085 | 0.684 | 0.231 |
| Care 5 | 3,240 | 0.000 | 0.904 | 0.096 |
Observation 6.2: The share moving up falls with the level, from 0.254 at Support Level 1 to 0.085 at Care Level 4, apart from a swap between Support Level 2 and Care Level 1. The share moving down grows to 0.176 at Care Level 3 and 0.231 at Care Level 4, and at Care Levels 3, 4, and 5 the share moving down exceeds the share moving up. The share staying at the same level is largest at Care Level 5 (0.904). Care Level 5 has no higher level, and below Support Level 1 lies outside the table, so both ends reflect the scale's limits. By sex too, at Care Levels 3, 4, and 5 the share moving down exceeds the share moving up (the share moving up from Support Level 1 is 0.305 for men and 0.231 for women; the share moving down from Care Level 4 is 0.195 for men and 0.253 for women).
Observation 6.3: The table keeps those who, in the survey month one year later, still need care and live in a household. At heavy levels, people who died or moved into a facility during the year have left the table. Whether the pattern of the share moving down exceeding the share moving up is because many became lighter or because those who became heavier left the table cannot be separated with this table.
Figure 1: Shares of the level now, by level one year ago (total). Orange is moved up, navy is same level, green is moved down.
7. Counting from now
Calculation 7.1 Level one year ago, by level now (total; denominator is the total at the current level)
| Level now | Total (per 100,000) | Same level | Lighter a year ago | Heavier a year ago | Not certified a year ago | Other |
|---|---|---|---|---|---|---|
| Support 1 | 18,931 | 0.651 | 0.000 | 0.088 | 0.232 | 0.030 |
| Support 2 | 18,121 | 0.653 | 0.129 | 0.058 | 0.123 | 0.037 |
| Care 1 | 19,999 | 0.625 | 0.138 | 0.088 | 0.132 | 0.017 |
| Care 2 | 18,843 | 0.613 | 0.230 | 0.059 | 0.078 | 0.020 |
| Care 3 | 10,892 | 0.490 | 0.324 | 0.072 | 0.094 | 0.019 |
| Care 4 | 7,028 | 0.503 | 0.351 | 0.015 | 0.118 | 0.013 |
| Care 5 | 4,657 | 0.629 | 0.250 | 0.000 | 0.084 | 0.037 |
Observation 7.2: Of those now at Support Level 1, 0.232 were not certified a year ago. This share is largest at Support Level 1 and smallest at Care Level 2 (0.078). Among those now at Care Levels 3 and 4, those at a lighter level a year ago make up 0.324 and 0.351, second only to those at the same level (0.490 and 0.503).
Figure 2: Shares of the level one year ago, by level now (total). Navy is same level, orange is lighter a year ago, green is heavier a year ago, gold is not certified a year ago, and gray is other or unknown.
8. The share staying at the same level, counted from two sides
Calculation 8.1 Share staying at the same level (total)
| Level | From one year ago | From now | Difference |
|---|---|---|---|
| Support 1 | 0.746 | 0.651 | +0.095 |
| Support 2 | 0.724 | 0.653 | +0.071 |
| Care 1 | 0.702 | 0.625 | +0.077 |
| Care 2 | 0.717 | 0.613 | +0.103 |
| Care 3 | 0.680 | 0.490 | +0.189 |
| Care 4 | 0.684 | 0.503 | +0.181 |
| Care 5 | 0.904 | 0.629 | +0.275 |
Observation 8.2: At all seven levels, the "same level" counted from one year ago is larger than the "same level" counted from now. The difference widens from 0.071 at Support Level 2 to 0.275 at Care Level 5. The numerator is the same number; what differs is the denominator. The denominator counted from now includes those not certified a year ago and those who moved in from other levels. For Care Level 5, "the share that stayed the same" depends on the side it is counted from.
9. Limitations
- The Comprehensive Survey of Living Conditions is a sample survey, and its figures are estimates. The figures in this paper are per 100,000, not numbers of people. Small cells may lie within the survey's variation.
- Residents of social welfare facilities and long-term hospital inpatients are excluded from the survey. People who died, moved into a facility, or lost certification during the year are not in the table, and this paper does not measure how large that is.
- The care level is a survey answer, not the certification record itself. The level one year ago is an answer given now.
- This is a table for one year (2025), not the same people followed over many years. It does not go into ways to lighten care levels or into whether care is good or bad. Prefectures are not compared. It is not an outlook for any one person.
Conclusion
In the 2025 care questionnaire, care levels "became heavier" over the past year for 29,556 per 100,000 and "became lighter" for 6,471, a ratio of 4.57. Collapsing Table 28, 0.439 of those who became heavier were not certified a year ago, and among those certified a year ago the ratio is 2.56. Counted by the level one year ago, at Care Levels 3, 4, and 5 the share moving down exceeds the share moving up, and the share staying at the same level for Care Level 5 was 0.904 counted from one year ago and 0.629 counted from now.
Three things separate them — counting from one year ago or from now, and whether the person was certified a year ago.
On hearing "those who became heavier in a year are so many times those who became lighter," check together whether newly certified people are included, from which side it was counted, and whether people left the table during the year.
Data and reproduction
| Data | 2025 Comprehensive Survey of Living Conditions, long-term care, Tables 28 and 29 (Ministry of Health, Labour and Welfare; e-Stat), and the survey outline of the Comprehensive Survey of Living Conditions (Ministry of Health, Labour and Welfare) (retrieved 26 September 2026) |
| Terms of use | e-Stat follows the Government Standard Terms of Use (Version 2.0). The shares and ratios in this paper were created by processing it and are not figures produced by the government |
| Scope | Japan, 2025 (the care questionnaire is conducted in June) |
| Checker | 検算/M065.py (--zu also draws the figures). Every number in the text is taken from 検算/M065_output.txt |
| Raw data | The retrieved files and SOURCES.md in data/ (URL, retrieval date, SHA-256, terms of use, definitions and how calculations are set up) |
| Figures | 図/M065_zu1, 図/M065_zu2 (png, svg, English versions) |
References
- Ministry of Health, Labour and Welfare. 2025 Comprehensive Survey of Living Conditions, long-term care, Tables 28 and 29 (in Japanese). e-Stat (accessed 26 September 2026). https://www.e-stat.go.jp/
- Ministry of Health, Labour and Welfare. Comprehensive Survey of Living Conditions, survey outline (in Japanese) (accessed 26 September 2026). https://www.mhlw.go.jp/toukei/list/20-21tyousa.html
- Yamagishi, Y. (2026). The Share Certified as Needing Care Doubles Every 4.26–4.97 Years at Ages 65–84, Faster Than the Death Rate at the Same Ages (5.87–7.90 Years) — The Slowdown from 85 Is in the Support Levels, and Counting Only Care Levels 3–5 It Keeps Doubling Every 4.64–5.61 Years up to 90–94 [M058]. Zenodo. https://doi.org/10.5281/zenodo.22975817
- Yamagishi, Y. (2026). The Leading Main Cause for Needing Care Is Dementia (16,743 per 100,000), but Choosing All That Apply It Is Frailty Due to Old Age (26,100) — By Age It Shifts from Cerebrovascular Disease (65–79) to Dementia (80–89) to Frailty (90+), and for Women the Top Three Are Within 98 [M059]. Zenodo. https://doi.org/10.5281/zenodo.22975921
Reference
Yamagishi, Y. (2026). Over the Past Year, Care Needs "Became Heavier" for 29,556 per 100,000, 4.57 Times "Became Lighter," but 0.439 of Them Were Not Certified a Year Ago [M065]. Zenodo.
DOI: 10.5281/zenodo.22976903
YOI: https://www.yamagishiyuuki.com/yoi/M065
ORCID: 0009-0006-8587-3450
PDF (English) · PDF (Japanese) · Date proof of the English PDF (OpenTimestamps .ots) · SHA256SUMS
The record page for this paper (in Japanese, with the PDFs in each language and the date proof): YOI M065. This paper is a preprint and has not been peer reviewed.