The Applicability Of The Chinese Version Of The GDS-15 And PHQ-9 In The Assessment Of The Elderly’s Depression

BackgroundWorld Health Organization(WHO)has reported that about 20%of the elderly aged 60 and above in the world have vary degrees of depressive symptoms.It is estimated that by 2020,depression will be the second most prevalent disease in the world in terms of morbidity and mortality.At present,the prevalence of elderly depression in China was about 13%-27%.The prevalence of depression in the elderly is increasing year by year.Depression not only lowers the quality of life of the aged,but also causes high suicidal rate among the elderly.During the past two decades,medical researchers have made major progress towards the treatment of depressive disorders.Whereas,the overlap of depressive and somatic symptoms was supposed to be a reason for possible difficulties in the assessment of depression in the elderly.The depression symptoms of the elderly are easily confused with the natural results of elderly’ diseases and the symptoms of dementia.Therefore,the depression of elderly cannot be measured directly,the use of measuring tools is mandatory.For now,a number of different measures have been used to assess psychological distress,including the Patient Health Questionnaire-9(PHQ-9)and the Geriatric Depression Scale-15(GDS-15).However,in primary health care,there is little evidence that any particular depression assessment instrument performs better than other instruments.At present,81.05%of the scale’s assessment used the Classical Measure Theory(CTT),only 18.95%of the scale’s assessment applied the Item Response Theory(IRT).A disadvantage of CTT is that it only focuses on aggregated statistics(e.g.means,variances and covariance),the sum score and the measurement error of a given scale.Item response theory(IRT)methods provide estimates about the position on the latent trait where each item or inventory provides the most information IRT examines the observed item responses relative to the.person’s value regarding the underlying latent trait on the same metric.On the one hand,the adjustment and verification of the scale during the introduction may lead to the deficiency of the original scale.On the other hand,the cultural differences in the development of the scale may also lead to the inapplicability of the scale.Therefore,it is necessary to combine traditional CTT and advanced IRT to explore the applicability of the Chinese version of the GDS-15 and PHQ-9 in the assessment of depression in the elderly.ObjectivesTo evaluate the applicability of the Chinese version of the GDS-15 and PHQ-9 in the assessment of depression in the elderly by using the classical measurement-theory(CTT)and item response theory(IRT).To explore the influence of age,gender,education and other factors on the scale and the fairness of the scale in different elderly populations,so as to provide Suggestions for the selection of depression screening tools in the elderly.MethodsThe cross-sectional survey of the elderly was conducted in May 2018 and July 2019,respectively.A total of 2472 elderly were interviewed to participant this survey.Subjects were required to complete sociodemographic questionnaires,the PHQ-9 and the GDS-15 by face-to-face interviews.The subjects aged 60 and above were involved in the present study.The exclusion criteria of subjects were as follows:1)deaf or unconscious;2)cognitive impairment,serious mental disorder or mental illness;3)serious chronic diseases,including heart failure,respiratory failure,liver cirrhosis,renal failure;4)greatly limited physical activity or unable to take care of themselves(two or more of the six indicators including eating,dressing,getting up and down,going to the toilet,walking indoors and taking a bath cannot be completed).This study was approved by the ethics committee,and all subjects had informed consent and volunteered to participate in the study.Evaluated the items through the factor analysis,discriminant analysis and Crowns coefficient method to the survey result.The validity of the scale was analyzed by means of Confirmatory Factor Analysis(CFA)and discriminant validity.The exploratory factor analysis was adopted to evaluated unidimensional structure of the scale.IRT analysis was performed using the Graded response model(GRM)and the Logistic two-parameter model(2PL).Rasch analysis was used to evaluate the fitting degree of the model and increased the persuasiveness of the results.Logistic regression was used to discusses the impact of demographic characteristics on the scale.Based on the above result of the Logistic regression,further Differential item functioning(DIF)was selected to evaluate the fairness of the scale in different populations.EpiData3.0 was used for double data entry,all statistical analyses were performed using IBM SPSS 24.0.The Structural Equation Model of scale was built by Amos21.0.MULTILOG7.03 was used to analyze the item response theory.Using ConQuest2.0 to conducted the Rasch analysis.The statistical tests were two-sided,and values of P<0.05 were considered statistically significant.Results1.The results of the descriptive analysisA total of 2477 scales were recovered in this study,and the number of qualified scales were 2472,the conformity rate of the scale was 99.80%.Among them,there were 1,515(730 males and 785 females)in Taxing,Jiangsu,and 957(399 males and 558 females)in Shizuishan city,Ningxia.Floor effect and ceiling effect were not found in PHQ-9 and GDS-15.2.The analysis of the CTTThe scales had good internal consistency(>0.7).The evaluation of G2 and G9 in GDS-15 was low,while the evaluation of items in PHQ-9 were well.Kappa,value of PHQ-9 and GDS-15 was 0.416.The confirmatory factor analysis results showed that the χ2/df of the scale were 16.152 and 13.005 in the PHQ-9 and GDS-15 respectively,which were far higher than the adaptation standards.The Normalized patb coefficients of items G2 and G9 in GDS-15 were both lower than 0.19.3.The analysis of the IRTFactor ratios in the PHQ-9 and GDS-15 scales were 3.90 and 3.289(>3),respectively,which were accords with the unidimensional hypothesis.The result of parameter estimation showed that the parameters of item differentiation and difficulty of PHQ-9 were within the normal range,while the Item G2 in the GDS-15 was less differentiated(<0.3).The total information values of the PHQ-9 and GDS-15 were 11.422 and 6.147,respectively.The average information value of item G9 was the lowest(0.052).4.The analysis of the RaschWright Map showed that the ability of subjects in the PHQ-9 was Normally distributed,but the difficulty of the item was negatively skewed.In the GDS-15,the subjects’ ability and item’s difficulty were normally distributed,and the items were well fitted.According to the analysis of the fitting statistical indicators of the items,the difficulty of item P9 in the PHQ-9 was the largest,which was 2.823 and the whole scale was more difficult for the elderly.The lowest value of Outfit MNSQ in GDS-15 was G11(0.09),and the measurement result of this item was not well.5.Multivariate analysis of the impact on the score of the scaleThe Logistic regression showed that the PHQ-9 was influenced by gender(OR=1.392,P<0.001),ethnicity(OR=3.324,P<0.001),residential pattern(OR=0.697,P<0.05),and education(OR=0.586,P<0.001).The GDS-15 was mainly affected by gender(OR=1.521,P<0.001),urban-rural distribution(OR=1.590,P<0.001),education level(OR=0.753,P<0.05),and the family income(OR=0.556,P<0.001).The DIF analysis showed that DIF CONTRAST of gender in GDS-15 was in the normal range(<0.5),while DIF CONTRAST of urban-rural distribution was above 0.5.It is indicated that the G2 "Have you given up many activities and hobbies?" was more difficult for rural seniors,while the G9 "Would you rather be at home than do something you are not familiar with?" was more difficult for urban seniors.The absolute values of DIF CONTRAST for gender and ethnicity in the PHQ-9 were in the normal range(<0.5).ConclusionsThe Chinese version of the PHQ-9 scale is a reliable and effective tool for assessing depression in the elderly.Its screening results are similar to GDS-15,but it is shorter,items are more stable,and it is easier to implement and score.The PHQ-9 is generally difficult for the elderly in China,with low scores.The GDS-15 is suitable for the elderly in China,but items 2 and 9 in the scale need to be further adjusted.Compared with CTT,IRT can explore the relationship between the respondent’s ability level and response at the item level,and it is easier to find problems such as the inapplicability of items in the scale caused by translation or cultural differences.