11.07.2015 Views

Mohammed T. Abou-Saleh

Mohammed T. Abou-Saleh

Mohammed T. Abou-Saleh

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

68 PRINCIPLES AND PRACTICE OF GERIATRIC PSYCHIATRY5. Stoller EP. Interpretations of symptoms by older people. A healthdiary study of illness behavior. J Aging Health 1993; 5(1): 58–81.6. Verbrugge LM. Longer life but worsening health? Trends in healthand mortality of middle-aged and older persons. Milbank Mem FundQu/Health Society 1984; 62: 475–519.7. Shock NW, Greulich RE, Andres R et al. Normal Human Aging: TheBaltimore Longitudinal Study of Aging. Washington, DC: USGovernment Printing Office, 1984.8. Busse EW, Maddox GL. The Duke Longitudinal Studies of NormalAging, 1955–1980. An Overview of History, Design and Findings. NewYork: Springer, 1985.9. Rabbitt P. Applied cognitive gerontology: some problems,methodologies and data. Appl Cogn Psychol, 1990; 4: 225–46.10. Risch SC, Lewine RJ, Jewart RD et al. Ensuring the normalcy of‘‘normal’’ volunteers. Am J Psychiat 1990; 147: 682–3.11. Halbreich U. Dr Halbreich replies. Am J Psychiat 1990; 147: 683.12. Todd M, Davis KE, Cafferty TP. Who volunteers for adultdevelopmental research? Research findings and practical steps toreach low volunteering groups. Int J Aging Hum Dev, 1984; 18:177–84.13. Fox NC, Freeborough PA, Rossor MN. Visualisation andquantification of rates of atrophy in Alzheimer’s disease. Lancet1996; 348: 94–7.14. Rodin J, Timko C, Harris S. The construct of social control:biological and psychosocial correlates. In Lawton MP, Maddox GL,eds. Annual Review of Gerontology and Geriatrics, Vol. 5. New York:Springer, 1985.15. Thornton EW. Exercise and Ageing. An Unproven Relationship.Liverpool: Institute of Human Ageing, 1984.16. Rubin EH, Storandt M, Miller JP et al. A prospective study ofcognitive function and onset of dementia in cognitively healthy elders.Arch Neurol 1998; 55(3): 395–401.17. Plemons JK, Willis SL, Baltes PB. Modifiability of fluid intelligencein aging: a short-term longitudinal approach. J Gerontol 1978; 33:224–31.18. Rabbitt PMA. A fresh look at changes in reaction times in old age. InStein DG, ed. The Psychobiology of Aging. New York: Elsevier/NorthHolland, 1980.19. Sternberg RJ, Wagner RK (eds). Practical Intelligence, Nature andOrigins of Competence in the Everyday World. Cambridge: CambridgeUniversity Press, 1986.Cohort StudiesPatrick RabbittAge and Cognitive Performance Research Centre, University of Manchester, UKThe word ‘‘cohort’’ originally designated a Roman military unitbut has now become a technical term in population studies as acollective noun for any peer group, band or sub-set of individualsunder investigation. In current usage, a cohort is any group ofindividuals who are linked in some way. This link may beexperience of common life events, such as particular pathologies,or life transitions, such as menopause, or experience of aparticular historical event or socio-economic condition. Ingerontology the most common defining factor is age group, andthis should be assumed unless some other usage is specified. Ingerontology, ‘‘cohort’’ has become a more acceptable term tospecify chronological age than the often misused ‘‘generation’’,which has an equally precise and different meaning in studies ofkinship terminology.Note that ‘‘age group’’ may be very loosely defined. Forexample cohort members may have all been born, or died, in thesame week, month, year, decade or even century. The definingboundaries of cohorts cause methodological difficulties becausethe effects of age are confounded with those of birth cohort andperiod. That is, groups of people born at the same time (birthcohorts) are by definition all of the same age, and have all livedthrough the same historical period. Groups of the same age (agecohorts) are not necessarily born at the same time, and so mayhave experienced different historical periods and events such aswars, with attendant differences in social circumstances. Groupsof different ages (different age cohorts) also have not sharedparticular historical events or periods and, to the extent that theirexperience of a historical event has overlapped, they have beenaffected by it at different ages.Cohort analysis is the methodology of designing and analysingstudies to make inferences about the behaviour or condition of aparticular sub-group without the necessity for studying themagain after one or more successive time periods. It is now the mostcommon methodology used to study changes in behaviour orattitudes, biological and cognitive effects of human ageing andsocial, political and cultural change. Cohort analyses in developmentaland ageing studies are distinguished from longitudinalanalyses, in which the same birth cohorts are re-examined atintervals over a period of time. A compromise is cross-sequentialanalyses, in which different age groups and birth cohorts arerepeatedly re-examined and compared with themselves and withothers at different measurement points.Unfortunately, although cohort analyses always allow us toidentify and study each of these effects, they do not provide anydirect way of examining all of them, independently, in a singlestudy. One reason for this is sampling variability. That is, anycohort samples we can obtain and compare are unlikely to beprecisely comparable. A related difficulty, which usually guaranteesthat samples in gerontological comparisons will not becomparable, is sample attrition. As cohorts age, they losemembers and so alter in terms of their credibility as representativesamples of the populations from which they were initially selected.The most important limitation on cohort analyses is that there isno way to avoid confounding at least two of the three variables inwhich we are usually interested: age, cohort and period. Thisimposes very inconvenient restrictions on statistical analysis; forexample, in a multivariate regression analysis, all three variablescannot simultaneously be entered as variables in a regressionequation. Although many attempts have been made, and muchhas been written on the subject, there are still no statisticalmethods that can clearly separate the effects of birth cohort andperiod from the effects of ageing. Perhaps the most lucid andhelpful discussions have been by Costa and McCrae 1 andPalmore 2 . The current consensus is that decisions must rest onscientific judgement as to which two of the three possible effectsare likely to be important, so that the third can be omitted. The

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!