| Objective To compare the performance of the Frailty Phenotype(FP),the FRAIL scale,the Tilburg Frailty Index(TFI)and the Groningen Frailty Index(GFI)in measuring frailty risk and adverse outcomes in elderly inpatients.This study will provide a reference for the development of an appropriate frailty screening tool for elderly inpatients in China.Method Using a convenience sampling method,462 elderly patients hospitalized in a tertiary hospital in Ningxia region from November 2019 to July 2020 were selected for the study,and the FP,FRAIL,TFI,GFI and Frailty Index(FI)were applied for frailty evaluation.FI was used as the diagnostic criterion for frailty,and the performance of the four screening tools was compared by means of Receiver Operating Characteristic(ROC)curves and Bayes discriminant analysis for frailty measurement in elderly inpatients;patients were followed up by telephone six months after discharge from hospital,and the follow-up included falls,readmission,activities of daily living and death within six months.Logistic regression was used to analysis the effect of frailty assessed by different frailty screening tools on adverse outcomes(death,falls,readmission and development of disability),and ROC curves were fitted with different outcomes as variables to compare the predictive validity of the four frailty screening tools on adverse outcomes in elderly patients.Results 1.The incidence of frailty in elderly inpatients was determined by FI,FP,FRAIL,TFI and GFI to 30.5%,34.2%,25.1%,35.1%,and 23.4% respectively;the kappa values for agreement between FP,FRAIL,TFI,GFI and FI determinations were 0.511,0.479,0.401,and0.143 respectively(P<0.01).2.The Area Under the ROC Curve(AUC)for FP,FRAIL,TFI and GFI measured frailty was 0.823(95% CI: 0.781-0.865),0.853(95% CI: 0.814-0.891),0.830(95% CI: 0.787-0.763)and 0.713(95% CI: 0.658-0.763)respectively;the AUC difference for GFI with FP,FRAIL and TFI was 0.110(95% CI: 0.658-0.763).0.873)and 0.713(95% CI: 0.658-0.763);the AUC differences between GFI and FP,FRAIL and TFI were 0.110(Z=3.786,P<0.05),0.140(Z=5.201,P<0.05)and 0.117(Z=4.347,P<0.05),and the differences were statistically significant,the differences in AUC between the other three debilitating screening tools were not statistically significant(Z=0.267-1.366,P>0.05).3.The cross-validation accuracy of debilitating risk prediction for FP,FRAIL,GFI and TFI were 78.6%,79.0%,81.6% and 72.5% respectively;the best critical values were 3,2,5 and6 respectively;the sensitivity was 0.709,0.837,0.786 and 0.664 respectively,the specificity was 0.817,0.790,0.732 and 0.668,Youden index of 0.529,0.625,0.518 and 0.332,positive predictive values of 0.697,0.702,0.563 and 0.468,and negative predictive values of 0.825,0.891,0.886 and 0.819,respectively.4.462 elderly patients were followed up by telephone after 6 months discharge,44 cases(9.52%)were lost,and 418 cases of complete data were collected,of which 25 cases died(6.00%).Among the 393 elderly patients who survived,20 cases(5.10%)were fell,127 cases(32.30%)were re-admitted to hospital,and 89 cases(22.60%)developed disability.5.After adjusting for gender,education level,monthly income,chronic diseases and health self-assessment,the death risk of frail elderly patients evaluated by FP,FRAIL and TFI was higher than that of non-frail elderly patients(OR =2.933,7.655 and 6.080,P<0.05),while the relationship between frailty and death evaluated by GFI was not statistically significant(P=0.145).The AUC of FP,FRAIL,TFI and GFI in predicting death were 0.729(95% CI: 0.619-0.840),0.781(95% CI: 0.690-0.873),0.751(95% CI: 0.673-0.829).The AUC difference between FRAIL and FP was statistically significant(Z=1.962,P<0.05).There is no significant difference in AUC between FRAIL and TFI,FP and TFI(Z=0.805,0.495,P= 0.421,0.621).6.After adjusting for age,marital status and health self-assessment,the fall risk of frail elderly patients evaluated by FRAIL was higher than that of non-frail elderly patients(OR=3.715,P<0.05),and the relationship between frailty and fall risk evaluated by FP,TFI and GFI was not statistically significant(P>0.05).The AUC of FP,FRAIL,TFI and GFI were 0.734(95%CI:0.612-0.856),0.724(95% CI: 0.594-0.854),0.744(95% CI: 0.631-0.858)and0.683(95%CI:0.565-0.801),respectively.There was no significant difference in AUC among FP,FRAIL and TFI(Z=0.218-0.392,P=0.695-0.827).7.After adjusting for number of chronic diseases and health self-assessment,the risk of readmission of the frail elderly patients evaluated by the four frailty screening tools was higher than that of the non-frail ones(OR=1.744,2.677,1.889,2.625,P<0.05).The AUC of FP,FRAIL,TFI and GFI were 0.580(95% CI: 0.517-0.642),0.613(95% CI: 0.551-0.675),0.618(95% CI:0.555-0.680)and 0.658(95%CI:0.599-0.717),respectively.It is suggested that the four debility screening tools are weak in predicting readmission,and there is no significant difference in AUC among FP,FRAIL and TFI(Z=0.194-1.311,P=0.190-0.846).8.After adjusting for literacy,whether hospitalized in the past 1 year,number of chronic disease and health self-assessment,elderly patients evaluated as frail by FP and FRAIL had an increased risk of developing disability compared to non-frail(OR=6.198,5.558,P<0.05),the relationship between frailty and development of disability evaluated by TFI was not statistically significant(P=0.142),and frailty evaluated by GFI was not statistically significant after adjusting for relationship with disability development before and after correction for variables were not statistically significant(P>0.05);the AUC for predicting disability for FP,FRAIL,TFI and GFI were 0.734(95% CI: 0.675-0.793),0.703(95% CI: 0.639-0.767),0.600(95% CI:0.528-0.672)and 0.556(95% CI: 0.486-0.626),suggesting moderate predictive power of FP and FRAIL and weaker predictive power of TFI and GFI for the development of disability;the difference in AUC between FP and FRAIL was not statistically significant(Z=1.046,P=0.295).Conclusion 1.FP,FRAIL and TFI can all be used to screen for frailty in elderly inpatients,with FRAIL being superior to the other two frailty screening tools in terms of its ability to measure frailty and ease of use.2.FP had good predictive value for death,falls and disability development within six months;FRAIL was effective in predicting death,falls and disability development within six months,especially for death compared to the other three methods;TFI had good predictive value for death and falls within six months;GFI was a weak predictor of adverse outcomes;none of the four frailty screening tools predicted elderly patients six months after discharge from hospital.None of the four frailty screening tools predicted hospital readmission within six months of discharge.3.FRAIL is more suitable for elderly hospitalized patients in terms of its ability to measure frailty,predictive efficacy of adverse outcomes,and ease of operation. |
PDF Full Text Request