Abstract
BACKGROUND/OBJECTIVES: The Protein Screener 55 + (Pro55 + ) is a brief food questionnaire to screen older community-dwelling adults for low protein intake. The result is the predicted probability of protein intake <1.0 g/kg adjusted body weight (aBW)/d ranging from 0-1. For purposes of cross-cultural validation, we translated the Pro55+ into German and tested its discriminative accuracy in detecting low protein intake of older community-dwelling people in Germany. SUBJECTS/METHODS: After translation and pilot-testing, the Pro55+ and the reference standard (3-day dietary record) were completed by 144 participants (81.6 ± 3.9 years, 61.8% female). Discriminative properties were tested by receiver operating characteristic curves and by calculating sensitivity and specificity for different cut-offs of predicted probability (>0.3/>0.5/>0.7) using <1.0 or <0.8 g/kg aBW/d to define low protein intake. RESULTS: Protein intake was <1.0 g/kg aBW/d in 39.6% of the sample and <0.8 g/kg aBW/d in 17.4%. Area under the curve was 62.0% (95%CI 52.6-71.5) and 68.8% (58.1-79.4), respectively. Specificity was 82-90% using probability cut-offs of 0.5 and 0.7 for both protein thresholds. Sensitivity was poor for protein threshold of 1.0 g/kg aBW/d regardless of the used probability cut-offs. For protein threshold of <0.8 g/kg aBW/d, sensitivity was 88.0% (71.8-96.9) using a probability cut-off of 0.09. CONCLUSION: The overall discriminative accuracy of the German Pro55+ to identify older community-dwelling people with low protein intake was poor. However, applying different probability cut-offs allows increasing specificity and sensitivity for 0.8 g/kg aBW/d to levels justifying the use for certain purposes e.g. excluding individuals with adequate protein intake. Further validation is needed.