Objective: The objectives of this study were (a) to elucidate the methodological problems arising when examining lifetime symptom data by exploring the accuracy of recall of medically unexplained symptoms (MUS) and medically explained symptoms (MES) in the general population, based on interviews using the Composite International Diagnostic Interview (CIDI) Somatoform Section C, in 1990 and 2001, and (b) to find predictors for failure at follow-up to recall symptoms reported previously at baseline (i.e., symptoms "lost").
Methods: Four hundred twenty-one persons (response rate, 69.6%; 242 women and 179 men) were reinterviewed in 2001 from a baseline population of 605. Thirty-eight clinically significant MUS and MES were assessed. Linear multiple regression analyses with the numbers of MUS-lost (medically unexplained symptoms lost to recall) and MES-lost (medically explained symptoms lost to recall) as dependent variables were undertaken to find factors affecting symptom loss.
Results: A wide range of individual symptoms (22-100%) were lost to recall at follow-up, indicating a large degree of measurement error, mainly due to faulty recall. The number of symptoms recalled when they were grouped was better (approximately 50% for "1-3" symptoms). Recall variability and MUS/MES transition over time undermined the credibility of this distinction. Gender and age emerged as significant (P<.01) predictors for MUS-lost, and a decrease in physical morbidity for MES-lost. Men tended to forget more symptoms than women, and younger respondents with high levels of baseline MUS remembered slightly better at follow-up.
Conclusion: Lifetime symptom data elicited in community surveys by such instruments as the CIDI should be viewed with caution. Methodological errors weakening data credibility could lead to false impressions of true change over time. A MUS/MES distinction is difficult to maintain.