[An investigation to discriminate frailty based on the Questionnaire for medical checkup of old-old: A pilot study using the item response theory].

Aim: This study aimed to investigate a method for scoring the questionnaire for medical checkup of old-old (QMCOO) and to clarify a cut-off score for the discrimination of frailty.

Methods: Survey forms were distributed to 2,586 older adults. For old-old adults, the item characteristics of the QMCOO were indicated using the item response theory (IRT). A receiver operating characteristic (ROC) analysis was performed using the total score of the fitting model of QMCOO for suggesting a cut-off score to discriminate frailty. The cross-validity of the cut-off score was verified among young-old adults.

Results: Among 1,680 adults who responded, data from 975 old-old and 421 young-old adults were analyzed. The method for scoring zero or one on each item in QMCOO conformed to the IRT model. The item discrimination and difficulty met the criteria. An ROC analysis showed that the area under the curve (AUC) and cut-off score for the discrimination of frailty were 0.871 and 4 points (sensitivity = 0.811, specificity = 0.766, positive likelihood ratio [LR+] = 3.469, and negative likelihood ratio [LR-] = 0.247), respectively. For young-old adults, the AUC and cut-off score were 0.874 and 4 points (sensitivity = 0.741, specificity = 0.817, LR+= 4.053, and LR- = 0.317), respectively.

Conclusions: The method for scoring zero or one on each item of the QMCOO was valid. A cut-off score of 4 for the discrimination of frailty demonstrated the interpretability of the QMCOO, while the usefulness of the QMCOO in young-old adults demonstrated cross-validity.