Editor—We agree with Tunstall-Pedoe that crude death rates are misleading in comparisons.1 However, clinicians and epidemiologists have different attitudes towards these health indicators.
Mortality from ischaemic heart disease has been low in Japan.2 None the less, its recent trends were interpreted differently by clinicians and epidemiologists. On the basis of their experience, clinicians believed that mortality from ischaemic heart disease was on the rise. Trends in crude rates were compatible with their belief. Epidemiologists argued that to see the secular trend required age adjustment, which reduced the resultant rates. There was no simple answer about the validity of the two interpretations.3 Which was true?
If age adjusted mortality is higher in one population than in another, discovering the cause is a concern of public health. However, age adjusted mortality differs from the crude mortality that directly reflects the real number of deaths because it is a hypothetical value. Crude rates may be used to estimate the extent of the needs for health services. The difference between these two indices may be seen in a community with a high proportion of elderly people and a low age adjusted mortality.
Mortality from ischaemic heart disease has been increasing in Japan, while age adjusted mortality has been decreasing. Therefore, the assumption that an individual is exposed to a higher risk of dying from the disease cannot be validated. On the other hand, the need for improving medical facilities to serve a growing patient population is real because the actual number of patients is increasing, as clinicians noted. A former president of the national cancer centre in Japan described his impression as a clinician: “Age adjusted mortality is overused; an estimated incidence is more important for health care planning.”4 In Japan the population in 1985 replaced that in 1935 as the reference for age adjustment in 1990. Because the aged population was small in 1935, the age adjusted mortality on this basis tends to be unrealistically small.5 The age adjusted mortality may be misleading if it is used to plan for present and future health care because it differs noticeably from the actual situation.
Measuring the frequency of disease in an aged society is important not only for estimating health risks but for allocating health services. Both age adjusted and crude rates, with actual numbers, are important. With the collaboration of clinicians, epidemiologists must convey the meaning of these indices and use them as appropriate.
Footnotes
[Published as a rapid response on our website 21 October 1998]
References
- 1.Tunstall-Pedoe H. Crude rates, without standardisation for age, are always misleading. BMJ. 1998;317:475–476. doi: 10.1136/bmj.317.7156.475b. . (15 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Nakayama T, Yokoyama T, Yoshiike N, Iwaoka H, Zaman MM, Chowdhury AH, et al. Secular trends in death rates from ischemic heart diseases and cerebrovascular diseases in selected countries. J Epidemiol. 1996;6(suppl):S189–S196. doi: 10.2188/jea.6.3sup_49. [DOI] [PubMed] [Google Scholar]
- 3.Shimamoto T, Komachi Y. Epidemiology and aetiology of coronary disease [in Japanese] Nippon Naika Gakkai Zasshi. 1990;79:1485–1490. [PubMed] [Google Scholar]
- 4.Ichikawa H. Re-evaluation of cancer mass screening in Japan [in Japanese] J Jpn Med Assoc. 1992;107:622–625. [Google Scholar]
- 5.Abe Y, Kamiya K, Onodera M, Saito F. Reconsideration on the reference population for age-adjusted death rates [in Japanese] J Health Welfare Stat. 1990;37:28–33. [Google Scholar]