Issues Associated with Measurement of Dietary Intake

Measurement Error

There is potential for the occurrence of measurement error with the measurement of any exposure such as when using dietary methods to measure nutritional intake. Errors may arise as a result of flaws in the design of the measurement instrument or during data collection or processing. Measurement error may also occur as a result of individual characteristics of participants in studies. Measurement error can be defined as the difference between the measured exposure (or measure of dietary intake) and the true exposure. All measurement of dietary exposures is subject to some degree of measurement error making it difficult to achieve measurements of true intake.

Efforts to reduce measurement error during data collection and processing should be introduced into the protocol of all studies, however, even if preven-tative measures are taken it is impossible to eliminate it altogether. It is difficult to identify the type and structure of measurement error associated with dietary intake. Measurement error may occur because of inaccurate reporting by respondents. It may also vary according to dietary method, for instance, food items within record methods may be intentionally or unintentionally omitted and with FFQs frequency of consumption may be inaccurately reported. Systematic bias, interviewer bias, recall bias, and social desirability bias have been identified but there are likely to be other sources of error. (Bias can be defined as the modification of a method of measurement by a factor, which influences the measurement in one or more directions.) Measurement error associated with dietary methods may consist of one or more types of error.

Measurement Error in Data Collection and Processing

Systematic bias Systematic bias is a systematic mis-measurement of data and can occur, for instance, if equipment such as weighing scales under- or overestimates values or if an interviewer consistently fails to use questions to probe for consumption of snacks and additional foods. If systematic bias can be identified solutions can be found, for instance, by calibrating equipment or training and monitoring interviewers.

Interviewer bias The behavior of an interviewer can influence the response of interviewees leading to interviewer bias. The degree of rapport between interviewer and respondent also influences results. Bias may occur if interviewers omit responses or record them incorrectly. Trained interviewers should ask open-ended questions in a neutral or nonleading manner, and not imply that a food or beverage should or should not have been consumed and avoid value judgments.

Social desirability bias Social desirability bias can influence dietary measures as respondents strive to report what they think is required not what was actually consumed, for example, reporting less alcohol consumption than is the case or greater consumption of foods with perceived health benefits such as fish, fruit, or vegetables. This is likely to be the cause of mis-reporting, under-reporting, or low energy reporting, which occurs in certain respondents. It is possible to predict how much energy a respondent should report, as this is the amount required to maintain a stable weight. (Weight will be either gained or lost if more or less energy is consumed than required.) As energy intake should equate to energy expenditure, expenditure effectively measures intake. Techniques for measurement of energy expenditure such as whole body calorimetry and doubly labeled water can be used. Using these techniques those individuals classified as low energy reporters are likely to be older, more overweight, and of lower educational and socioeconomic status than the rest of the population. Low energy reporters tend to have lower consumption of foods in the groups cookies, cakes, puddings, confectionery (candy) and sugary foods and, in some populations, lower consumption of spreads, cooking fats, and potato chips. Interviewers should be aware of low energy reporting, aim to be entirely nonjudgemental, and also request participants make complete records of food intake.

Impact of Measurement Error

As the proportion of error within a measurement increases, the accuracy of the measurement decreases and the results using the measurement will become less interpretable. Hence, greater measurement error reduces the likelihood that the truth has been measured with accuracy and increases the likelihood that analyses relating diet to disease status will tend towards null results. The effect of measurement error is to mis-classify an individual within a range of intake.

Validation of Dietary Methods

Validation is used to quantify the measurement error that occurs when measuring dietary intake exposures. It requires two measures: a main measurement and a second measurement subject to less measurement error than the first. The errors of the two measurements should be independent. Validation is used to estimate the proportion of measurement error within the main method by modeling the differences between the main and the secondary measurement. It had been considered that dietary methods had errors independent of each other and that record methods such as 24-HR could be used, but it is now known that the errors are not independent as individuals report in the same way with different methods. Therefore, it is better to use biological variables measurable in blood or urine (also known as biomarkers) as the second measure for dietary validation. Examples are vitamins, minerals, and individual fatty acids in blood such as vitamin C and carotenoids or urinary excretion over 24-h of nitrogen, potassium, and sodium. Examples of validation studies are those performed within EPIC-Europe and the Observing Protein and Energy Nutrition Study (OPEN) in the US. Work is ongoing to extend the number of biomarkers available and to define further and elicit the structure of measurement error.

Use of Calibration Methods to Adjust for Measurement Error

In contrast to validation, which attempts to identify the type and scale of measurement error, calibration is designed to adjust for systematic over- or underestimation in dietary intakes within populations. It may also be used at the individual level to attempt to correct for attenuation bias (or dilution) in relative risk due to errors in dietary measurements. Calibration of data has been proposed for large multicentre nutritional studies that have used different dietary methods to capture population-specific diets. Calibration studies require a highly standardized second dietary measure to be used in a representative sub-sample from each cohort to form a common reference measurement across populations. An example of this approach has been used by the European EPIC (European Prospective Investigations into Cancer and Nutrition) Study using a computerized, standardized 24-h recall in ten countries.

101 Everyday Tips for Losing 10 Pounds

101 Everyday Tips for Losing 10 Pounds

Studies show obesity may soon overtake tobacco as the leading cause of death in world. Who Else Could Use 101 'Everyday' Ways to Lose 10 Pounds or more and Keep it Off! You've been putting it off too long. Hey, everyone needs to lose weight from time to time. You're no different!

Get My Free Ebook


Post a comment