Ability and Accuracy of Long-term Weight Recall by Elderly Males: The Manitoba Follow-up Study

Purpose

To assess the ability and accuracy of elderly men to recall their weights and determine what characteristics might predict recall ability and accuracy.

Methods

Eight hundred sixty-nine elderly men (mean age, 84 years), participants of the Manitoba Follow-up Study (MFUS), responded to a questionnaire asking them to recall their weights at ages 20, 30, 50, and 65 years. Recalled weights were compared with measured weights collected since MFUS began in 1948. Logistic regression was used to predict ability and accuracy of weight recall.

Results

Only 75% of respondents attempted to recall their weights at all 4 ages. Among men recalling 4 weights, fewer than half were accurate within ±10%, just 7% were within ±5% of their measured weights. Accuracy of recall was significantly and independently associated with body mass index during middle age (5 kg/m²) (odds ratio 0.83, 95% confidence interval: 0.76, 0.90) and weight change. Unmarried men were less likely than married men to attempt recalling all 4 weights. Men overweight at middle age were more likely to underestimate their recalled weights.

Conclusions

Studies relating weight in early adulthood or middle age with outcomes in later life should not rely on elderly male participants recalling those weights.

Introduction

Excessive weight is associated with numerous adverse health conditions, such as cardiovascular disease, hypertension, type 2 diabetes, certain forms of cancer, arthritis, and gall bladder disease (1). The prevalence of overweight and obesity are increasing, with more than a billion individuals worldwide considered overweight or obese (2). Hence our understanding of the relationship between weight history during the life course and outcomes in later life has become increasingly important.

Few longitudinal studies have been able to examine such relationships with prospectively measured weights and documentation of disease over long periods of time 3, 4. Most investigators addressing hypotheses concerning historical body weight must rely on participants being able to recall their weights, potentially a questionable practice if participants are required to recall their weights well into the past. Even relying on elderly individuals to recall current weight has proven problematic (5).

Several studies have compared recalled body weights to previously measured weights 6, 7, 8, 9, 10, 11, 12. Despite the lack of a consistent definition of recall “accuracy” and a common approach to assessing it, there has been general agreement that recalled weights may be used with some caution. Most of these studies have involved middle-aged participants, with only two involving elderly individuals, 65 years or older 6, 8. The longer recalls have ranged from 25 or 30 years 6, 7, 11, 12 to over 50 years (8). Possible predictors of weight recall accuracy have included respondents' age 6, 9, 10, 12, sex 6, 7, 8, 9, 10, race 6, 9, education 6, 7, 9, marital status (10), weight at time of recall or earlier 6, 7, 8, 9, 10, 11, 12, weight change 9, 12, smoking history (10), recent hospitalization (9), self-rated health (10), and physical and mental functioning (6). Although considered in an investigation of self-reported current weight (13), none of these studies examined the effect of end-digit reporting preference, the tendency of subjects to report their weights as a number ending in “0” or “5”. Also, none of these reports investigated the willingness or “ability” of elderly individuals to recall weight.

By using historically recorded weights in the Manitoba Follow-up Study (MFUS) and weights recalled by elderly men (average age 84 years) at 4 distant points in time (20, 30, 50, and 65 years of age), this paper determines the frequency and determinants of two aspects of weight recall: the ability to attempt recall and the accuracy of recall.