J Clin Epidembl Vol. 46, No. 9, pp. 10634069, 1993 Printed in Great Britain. All rights reserved 0895-4356/93 $6.00 + 0.00 Copyright 0 1993 Pcrgamon Press Ltd REINTERPRETING MORTALITY STATISTICS: SOME USES OF GOMPERTZIAN ANALYSIS IN EPIDEMIOLOGICAL RESEARCH S. NEILSON and I. ROBINSON The Centre for the Study of Health, Sickness and Disablement, Department of Human Sciences, Brunei, The University of West London, Uxbridge, Middlesex UB8 3PH, England zyxwvutsrqponmlkjihgfe (Received 4 May 1992; received for publication 5 April 1993) Abstract4ompcrtzian analyses of mortality data have recently been undertaken for a number of individual conditions (Riggs JE. Mech Ageag Dev 1990, 55: 207-220 [l]; Neilson S et al. Me& Ageing Dev 1992; 64: 201-216 [2]; Neilson S et al. Acta New01 SC& 1993; 87: 184-l 9 1 [3]). Such analyses are in principle of particular epidemiologic4 value in circumstances where demographic change is occurring and where the balance between mortality from different conditions is subject to change. However, the extent to which a Gompertzian relationship between age and mortality holds for particular conditions has been subject to debate. In this analysis it is demonstrated that even some conditions which do not superficially hold to a Gompertz relationship do in fact do so, if such conditions are considered to be restricted ,to small, inherently susceptible subpopulations. By analysing mortality from a range of neurological conditions within the context of general mortality in England and Wales, conditions with different aetiologies such as Huntington’s chorea, amyotrophic lateral sclerosis and multiple sclerosis can be shown to have a Gompertzian mortality rate distribution. Such analyses are of substantial value in indicating how demographic change affects the balance of mortality between conditions, as well as directing interest to revised aetiological possibilities. Amyotrophic lateral sclerosis Hungtington’s chorea Multiple sclerosis Survival curves Life table analysis Gompertzian analysis INTRODUCTION Despite well documented problems associated with the study of mortality data, such data are still widely employed for epidemiological analy- sis, particularly for those conditions which ex- hibit major changes in mortality rates over time. Conventionally, after using a range of standard- izing procedures, such changes may be ex- plained by two broad factors. The first group of factors are those associated with artifact effects such as modifications in disease classification, modifications in mortality recoiding or improv- ing diagnostic practice, suggesting that changes in recorded mortality rates do not reflect trends in “real” underlying mortality. The second group are those posited on “real” underlying changes in mortality and implying the occur- rence of aetiological or managerial changes which directly increase or reduce the mortality rate. In this analysis it is argued that a third group of factors-associated with changing demographic patterns resulting from changing general mortality-may be equally effective in explaining the changing balance of mortality between particular diseases. The interest in the relationship between changing demographic factors and mortality has increased in recent years as both the number and proportion of deaths in the older age- groups has increased. Brody and Schneider [4] argued in this context for a difference between 1063