手法を比較

	Chains-of-Risk Model	Cohort-Sequential Design
分野	Social Epidemiology	Social Epidemiology
系統	Process / pipeline	Process / pipeline
提唱年	2003	1953
提唱者	Diana Kuh & Yoav Ben-Shlomo (life-course glossary and conceptual models)	Richard Q. Bell (convergence/accelerated approach); Yasuo Miyazaki & Stephen Raudenbush (cohort-linkage tests)
種類	Sequential-mediation model of linked life-course exposures	Observational longitudinal design linking overlapping age cohorts
原典	Kuh, D., Ben-Shlomo, Y., Lynch, J., Hallqvist, J., & Power, C. (2003). Life course epidemiology. Journal of Epidemiology & Community Health, 57(10), 778-783.	Bell, R. Q. (1953). Convergence: an accelerated longitudinal approach. Child Development, 24(2), 145-152.
別名	Chain of Risk Model, Accumulation of Risk Model, Risk Chains, Additive vs Trigger Chains	Accelerated Longitudinal Design, Convergence Design, Cohort-Sequential Accelerated Design, Overlapping-Cohort Longitudinal Design
関連	3	3
概要	The chains-of-risk model is the specific life-course mechanism in which adverse exposures are linked in a sequence over time, so that one exposure raises the probability of the next, and the cumulative or final link bears on disease. Set out in Ben-Shlomo and Kuh's 2002 conceptual paper and defined in the Kuh, Ben-Shlomo, Lynch, Hallqvist, and Power 2003 life-course glossary, it models how early disadvantage can cascade — poor early circumstances leading to limited education, then to hazardous work or health behaviors, and finally to disease. Its signature analytic distinction is between an additive chain, in which each link independently adds to risk, and a trigger chain, in which the early links matter only because they lead to a final exposure that is the true cause. Chains-of-risk modeling thus treats the life course as a causal pathway to be decomposed, not a list of independent risk factors.	The cohort-sequential design — also called the accelerated longitudinal design — spans a long age range quickly by following several overlapping age cohorts for a short time each and then statistically linking their trajectory segments into one long developmental curve. Richard Bell introduced the idea in 1953 as 'convergence,' a way to study development over many years without waiting many years. Instead of following one cohort from, say, age 10 to age 20 for a full decade, the design enrolls cohorts aged 10, 12, 14, 16, and 18 and follows each for two or three years, with adjacent cohorts overlapping in age so their pieces can be joined. Yasuo Miyazaki and Stephen Raudenbush later supplied the formal multilevel tests for whether the cohorts can legitimately be linked. The design trades a single continuous within-person record for speed, while using overlap to check that the assembled curve is coherent.