Methods to assess relative reliability of diet records: minimum records for monitoring lipid and caloric intake.

A total of 252 diet records of 18 outpatients attending a lipid clinic were analyzed by a computer-assisted method to determine the minimum number of daily diet records that would be reliable for monitoring dietary adherence. Each subject recorded food intake in special diaries for 14 consecutive days between two clinic visits. All possible randomly selected combinations of 3, 4, 5, 7, 9, and 11 consecutive days of records in the 14-day period were analyzed for calories and lipids. Sets of records were said to be within a 95% confidence interval when the information yielded on any parameter differed by 5% or less from the mean values for the entire 14 days' records. All sets of records for 7, 9, and 11 days were in the 95% confidence range; therefore, 7 consecutive days of food recording were considered the minimum requirement for a 95% confidence limit. Out of 11 possible combinations of 4 consecutive day-sets of records, all but 3 sets were within 95% confidence limits. Consequently, 4 consecutive days of records were deemed acceptable as a reasonable compromise for minimal, reliable monitoring of diet compliance in outpatients for the nutrients studied.