BACKGROUND: Physical disability and dependency are serious, andfrequent, adverse health outcomes associated with aging and resulting fromchronic disease. Reasoning has suggested that there might be a preclinical,intermediate phase of disablement which might develop in parallel withprogression of underlying disease and precede and predict disability.Definition of this stage could provide a basis for screening and earlyintervention to prevent disability. The objective of this study was todetermine preclinical functional predictors of incident mobility difficultyand provide evidence for a preclinical stage of disability. METHODS: Aprospective, population-based cohort study was carried out in Baltimore,Maryland, with two evaluations 18 months apart. The participants were 436community-dwelling women, 70-80 years of age at baseline, not cognitivelyimpaired, and reporting difficulty in no areas, or only one area, ofphysical function (primarily mobility), who were participating in theWomen's Health and Aging Study II. Participants were recruited from apopulation-based, age-stratified random sample. Incident mobilitydisability was studied in the subset without such disability at baseline.The main outcome measure was self-reported incident difficulty walking 1/2mile or climbing up 10 steps. RESULTS: At baseline, 69.3% of the cohortreported no difficulty with mobility. After 18 months, 16.0 and 11.7% ofthis group reported incident difficulty walking 1/2 mile or climbing up 10steps, respectively. Those reporting baseline task modification due tounderlying health problems, our measure of preclinical disability, were atthree- to fourfold higher odds of progressing to difficulty than were thosewithout such modification. In multivariate logistic regression analyses,this self-report measure, task modification without difficulty, andobjective measures of performance were independently and jointly predictiveof incident mobility difficulty. Specifically, for incident difficultywalking 1/2 mile, self-reported task modification odds ratio (OR) = 3.67,walking speed (.5 m/s difference) OR = 2.16; for incident difficultyclimbing up 10 stairs, OR for task modification = 3.84, for stair climbspeed (1/3 step/s difference) = 2.08 (95% CI did not include 1 for any).Covariates, age, living alone, number of chronic diseases, depressionscore, knee strength, and balance by functional reach, were not significantpredictors in either model. CONCLUSIONS: Two indicators of functionalchanges in older women without mobility difficulty, self- report ofmodification of method of doing a task in the absence of difficulty andperformance measures, are independent and strong predictors of risk ofincident mobility disability. The self-report measure provides substantialstrength in predicting risk of incident disability across the full range ofperformance, and may identify a vulnerable point at which other riskfactors act to cause transitions to disability. Together, the preclinicalindicators identify a subset of high-functioning older women who are athigh risk of mobility disability, and provide a potential basis forscreening for disability risk and targeting interventions to preventmobility disability.