The Modified Baecke Questionnaire is designed to measure habitual physical activity in the elderly. Correlation with the DLW method as a reference method, showed fair-to-moderate correlation (Spearman correlation coefficient 0.54). This study shows that the questionnaire does a fairly good job of classifying individuals as low active and high active, but poorly for moderately active.
In the past ten years, several physical activity questionnaires have been validated using the DLW method [3–6, 17–23]. Published correlation coefficients varied widely between the different questionnaires. Also studies investigating the validity of the same questionnaire in different populations resulted in dissimilar correlation coefficients, indicating that one should be cautious when generalizing results to populations with different characteristics.
The Modified Baecke Questionnaire has been validated before by Bonnefoy and coworkers, also using the DLW method . In a healthy population of 19 older men, the Spearman correlation coefficient between the PAR and the Modified Baecke Questionnaire score was 0.14. This poor result is probably due to the way in which the questionnaire was administered. First of all, the study simultaneously validated ten questionnaires, which were completed during a personal interview of 4 hours and, therefore, the questionnaires might have influenced each other. In our study, we also have this potential bias, although we only administered one more questionnaire (PASE) during the interview, instead of nine. Moreover, in the study of Bonnefoy and coworkers, the bias might also be greater since the interviewers reminded the subjects about activities when inconsistent answers between the questionnaires were given. This was not done in our study.
Other physical activity questionnaires that are especially developed for administration in elderly subjects and have been validated using the DLW method are the Physical Activity Scale for the Elderly (PASE) , the Yale Physical Activity Survey (YPAS) , and the Zutphen Physical Activity Questionnaire .
Validation of the PASE with DLW has been performed in the same study population as the current study . The Spearman correlation coefficient between the PASE score and PAR was 0.68 (95% CI 0.35–0.86). The higher correlation coefficient found for the PASE score than for the Modified Baecke Questionnaire score might be explained by the difference in period over which physical activity was assessed. The PASE contains questions referring to activities during the past week and fully covered the period when the DLW measurement was conducted, whereas the Modified Baecke Questionnaire comprises questions referring to activities in the last year. Due to the incongruence of time frames, the correlation can be diluted. However, the DLW method is still the reference standard to measure energy expenditure and it is not feasible to conduct the measurement over a long period.
The validity of the Zutphen Physical Activity Questionnaire score was somewhat higher than for the Modified Baecke Questionnaire score (Spearman correlation coefficient of 0.67, p < 0.01) . It should be noted that this study was performed in a population of only elderly men.
The PASE and YPAS have also been validated by Bonnefoy and coworkers. The Spearman correlation coefficients between PAR and the questionnaire scores were 0.24 and 0.03, respectively. These validation results should be considered with caution because of above mentioned limitations in the conduct of that study.
Strength of our study is that the selection of subjects was based on tertiles of the PASE score. Hence all levels of physical activity were represented in the study population. Because participants were enrolled in an exercise intervention study and therefore where more aware of their physical activity behavior, they would likely to be able to more accurately report their level of activity. This should be taken into account when generalizing the correlation results to the general population. Our study has also several limitations that have to be addressed. First, the small study population restricted the possibility to make valid estimates of the validity of the questionnaire in general, but especially about the validity of the questionnaire for men and women separately. Second, the PASE questionnaire was administered before the Modified Baecke Questionnaire, which might have influenced the validation results slightly.