Method: Data from the 2005-2006 NHANES was analyzed for this study. A total of 1883 participants (Mean Age 62.37 ± 14.67) provided valid responses to the 20-item PFQ. The Rasch analysis was performed using WINSTEPS Rasch measurement software. First, model-data fit was evaluated by examining mean square residuals for each item. Second, the level of physical function item difficulty and person’s level of physical functioning was estimated. Third, an item-map distribution was evaluated to determine the relative distribution of physical function items across person-ability measures. Lastly, the rating scale was evaluated for proper functioning.
Analysis/Results: Overall, with the exception of two items, the Rasch rating scale model fit the data well. Two of the items “managing money difficulty” and “preparing meals difficulty” were removed as the result of mean square residuals outside the acceptable range (≤0.5 and ≥1.5). Data was person-mean centered (0 ± 1.20 logits) to allow for comparisons of item difficulty across ability measures. The item difficulty ranged from 1.09 – 5.21 logits, where a higher value indicates an easier item. The item “walking between rooms on the same floor” was identified as the easiest item, while “stooping, crouching, kneeling difficulty was the most difficulty. Item-map distribution indicated poor item distribution with a lack of item coverage with higher functioning individuals. In addition, the four-category structure of the PFQ demonstrated poor rating scale function.
Conclusions: The NHANES PFQ is commonly used to assess physical function in adults. However, the results from this study suggest that the current questionnaire may not sufficient for assessing physical function in the general population. Additional items should be developed for the NHANES PFQ in order to evaluate a more dynamic range of physical function.