[original article presented at American Public Health Association, Gerontological Health Section in Washington DC, 17th November 1998]

Agreement between Self-reporting and Professional Assessment for Estimating the Number of Noninstitutionalized Disabled Elderly

AtoZ Okamoto (Kindai Medical School Department of Public Health)


To verify the accuracy of self-reporting survey for estimating the number of noninstitutionalized elderly population.


In prepartion for the new LTC insurance, which will take effect in April 2000, estimating the number of noninstitutionalized disabled elderly became a policy imperative for making budget, setting premium and assessing the eligibility of the applicants. This estimated must be made not only on a national level but on a local level as well because the new LTC insurance will be administered by municipal governments. The premium paid by the local elderly population will account for only 17% of the total benefit budget and the rest will be financed by national government, a national pooling fund levied by the national working population. Therefore it is necessary to determine the number of eligible disabled elderly to make a fair and equitable fund allocation.

Data and Method

National Household Survey (NHS)

The rough estimates of the number of noninstitutionalized disabled elderly have been made by a questionnaire survey conducted by Ministry of Health at the interval of three years.

[survey period] It is conduced on one day in June in '86, '89, '92, '95 and '98.

[sampling method] stratified random sampling based on the statistical districts for the purpose of the National Census 1990.

[sample size] 5,240 statistical districts, approximately 780,000 household members.

Residents in hospitals or nursing homes are not included.

[date of survey] 4th June 1992

[survey method] Surveyors visit the selected households and fill in the questionnaire by interviewing.

[method of extraporation] ratio estimation using the number of household members as parameter.

[surveyed items] level of activity (house-bound, chair-bound, bed-bound), items of ADL requiring assistance (

Baseline surveys by municipal governments (municipal surveys)

All municipal governments (n=3,249) including cities, towns and villages were ordered to conduct need surveys sometime in 1992 to estimate the number of noninstitutionalized elderly.

[purpose] To establish a baseline status on which the goal of the service level by the year 2000 would be based.

[method] population survey. Exact numbers of disabled elderly who would need care services were surveyed. To reach the disabled elderly, family doctors, social workers, volunteers, civil informants (Minsei-Iin) were mobilized to identify the disabled elderly at home. Uniform criteria to define the level of disability was presented by the government. However the quality and accuracy of the survey varies considerably among municipal governments.

[results] The survey findings were aggregate to the prefectural level (n=47). The results of municipal governments and aggregate prefectural levels were published. Those baseline data were used for the purpose of setting the national goal of expected service levels to be achieved by the year 2000.



Comparison between the NHS and municipal surveys

The author obtained the published reports of 13 major cities and 25 prefectures. After excluding the overlapping cities, 25 prefectures and 5 cities yielded the data covering 58.5% of the national elderly population. The aggregate number of noninstitutionalized disabled elderly estimated by municipal surveys was compared with the estimates by NHS. Also most municipal surveys estimated the number of dementia elderly, who are not surveyed by NHS.


The municipal surveys estimated that there were 173,347 noninstitutionalized disabled elderly either in chair-bound or bed-bound condition out of 9,493,898 population 65 or over. The prevalence rate of disability among elderly at home was calculated as 1.83%. When extraporated to the national level, it will translate into the number of 296,556, which approximates well with the NHS estimate: 289,000.

There was a considerable variances in the prevalence of disability among municipal governments, ranging from the high 4% (Tokushima prefecture) to the low 0.53% (Sapporo city). It is noteworthy that such regional variances are cancelled out when aggregated to the national level and the estimates agree well between the NHS and aggregation of municipal surveys.

Municipal surveys surveyed the prevalence of dementia, which was not included in NHS. Overall prevalence of dementia was 0.75% of the elderly population. There may be some overlapping between the disabled elderly and demented elderly. However many of the municipal survey reports failed to clarify if these two categories overlap. Geographic variances of the prevalence of dementia was smaller than disability but there is still a geographic variances.

A weak positive correlationship was observed between the percent of disability and the percent of dementia (r=0.33).

Conclusions and Discussions

The nationwide project of obliging all municipal governments to conduct need surveys in Japan in 1992 gives a rare opportunity to estimate the number of noninstitutionalized disabled elderly by population survey. Although the number of residents taken care of by nursing homes and geriatric hospitals can be easily monitored, it would be difficult to estimate the number of disabled elderly taken care of at home. Estimating the number of those noninstitutionalized elderly is indispensable for enacting the new LTC insurance whose benefit includes home care.

Those estimates have been conducted as a random sampling survey by way of self-reporting through NHS. However its accuracy has always been a question because the NHS is self-reporting questionnaire survey and lacks professional assessment. The timely coincidence that the NHS and municipal surveys had been conducted at almost the same time gives an opportunity to verify the agreement between the two methodologies.

The findnings suggest that self-reporting questionnaire survey can give as good an estimate as to the number of noninstitutionalized disabled elderly when aggregated to the national number. However the large geographic variances as to the prevalence of disability and dementia will leave ample room for discussion.

For example, Sapporo city whose elderly population 172.300 was found to have the lowest prevalence of disability and dementia. Cautions should be taken when one interprets these data. One should not hastily conclude that elderly in Sapporo enjoys the healthiest condition. Actually Sapporo is known to have the highest rate of institutionalization, i.e. it has highest number of beds of geriatric hospitals, psychiatric hospitals and nursing homes.

Both NHS and municipal surveys focused on noninstitutionalized elderly and do not include disabled and demented elderly already institutionalized. The Sapporo data should be interpreted with caution that majority of those disabled and demented elderly are institutionalized.

GO to CareManager.Net