Original articles
Education, aging, and health: to what extent can the rise in educational level relieve the future health (care) burden associated with population aging in the Netherlands?

https://doi.org/10.1016/S0895-4356(99)00232-2Get rights and content

Abstract

This article describes to what extent the expected rise in the educational level of the Dutch population can counterbalance the increases in the prevalence of ill-health and health care utilization based on the aging of the population for the period 1996–2020. Logistic regression models are used to estimate current differences in health (care utilization) by age, sex, and educational level, using data from the Netherlands Health Interview Survey. The current differences in health (care utilization) are applied to national projections of the composition of the population by age, sex, and educational level. Also, scenarios have been made in which the health differences by educational level are assumed to converge and diverge. The rise in the educational level counteracts the expected increases in ill-health based on population aging to a substantial degree (10–100%). We therefore recommend that in projections of ill-health also changes in educational level are taken into account.

Introduction

In projections of future rates of ill-health and health care utilization the association between demographic developments and health outcomes usually is modelled in a very simple way: only changes in the age and sex structure of the population are taken into account. The Population Forecast of Statistics Netherlands shows that the proportion of people older than 65 years of age is expected to increase from 13.3% in 1996 to 18.4% in 2020 [1]. Because of this aging of the population, which is larger in the Netherlands than in most other Western countries, it can be expected that health care needs will increase substantially in the coming decades 2, 3, 4.

However, many other socio-demographic factors have been demonstrated to be causally related to health. In most countries it has for instance been shown that people with a high educational level have lower morbidity rates than people with a lower educational level 5, 6, 7, 8. It can therefore be assumed that future changes in the composition of the population by educational level also will affect the health of the population. As in most other Western countries, large changes in the educational level of the Dutch population are expected in the coming decades. This anticipated rise in educational level might counterbalance (some of) the effects of aging.

In this study we estimate the effect that the expected changes in the composition of the population by educational level will have on the health and health care utilization of the Dutch population during the period 1996–2020. The future composition of the population in the older age groups by educational level can easily be deduced from the current composition by educational level in the young and middle-aged age groups (cohort-wise). The effect of taking changes in the educational level into account, is compared to projections in which only age is modelled. Several scenario projections are made in which it is assumed that current health differences by educational level will remain unchanged, will decrease or will increase.

Section snippets

Data

In this study we consider the health and health care utilization of the population of 25–84 years of age during the period 1996–2020. We chose for the lower limit of 25 years because it can be assumed that the vast majority of people will have attained their final educational level by that age. The upper limit of 84 years was chosen because up to this age reliable estimations of health and health care utilization could be made with the available data.

The Population Forecast of 1996 of

Methods

Logistic regression models were used to determine the current differences in health and health care utilization by age and educational level [14]. Separate models were fitted for men and women. Dummy variables have been constructed for age (12 5-year age groups) and for educational level (4 dummy variables), using the deviation coding scheme. In this coding scheme the effect of each category is compared to the average effect of all categories [15].

In the basic scenario it has been assumed that

Results

In Table 2 the differences in ill-health and health care utilization between people with primary school and people with higher vocational school or university as their highest level of education are shown, which have been applied in the stable, divergent and convergent scenario. For reasons of convenience the figures for people with a lower or intermediate education have been omitted. In general, the differences by educational level are larger among men than among women and larger for the

Discussion

Among Dutch men the expected increases in ill-health in the period 1996–2020, which can be expected to result from the aging of the population, to a large extent might be counteracted by the increase in the educational level. The effects of educational level on the projections of health care utilization are much smaller. Among women the increase in educational level might even nullify the effect of the aging of the population on most outcome measures. Assuming either divergent or convergent

Acknowledgements

This project was supported financially by the Priority Program on Population Research of the Netherlands Organization for Scientific Research (“NWO”). We thank C. W. N. Looman for his statistical advice.

References (31)

  • A.J Fox

    Health Inequalities in European Countries

    (1989)
  • R Illsley et al.

    Social inequalities in health

    Soc Sci Med

    (1990)
  • Mielck A, Rosario Giraldes M do. Inequalities in Health and Health Care. Review of Selected Publications from 18...
  • Beer J de, Roodenburg H. Drie scenario's van de bevolking, huishoudens, opleiding en arbeidsaanbod voor de komende 25...
  • Bevolking en arbeidsaanboddrie scenario's tot 2020 (Population and labour supply: 3 scenarios until 2020)

    (1997)
  • View full text