Assessment On The Applicability Of FIM Scale Among Elderly Population In Community By Classical Test Theory And Item Response Theory

Background and ObjectiveAt present,research reports on the applicability of FIM(Functional Independence Measure)among elderly population in community are relative rare,and published evaluation work is generally focused on Classical Test Theory methods.This study intends to start from community,analyze the item characteristics,item response characteristics,item information and functional differences(age,gender,prevalence)of FIM,by combining Classical Test Theory(CTT)and Item Response Theory(IRT).In order to conduct a more comprehensive assessment of FIM,not only evaluate reliability and validity,but explore the applicability of FIM among elderly population in community.The eventual goal of our study are providing data support for the further application of FIM and scientific tools for the formulation of China’s aging policy or community health management programs about elderly.MethodsThis study was a cross-sectional research,conducted in two district of Shenzhen city in China,during 2015～2017 years through cooperation with local government and community health service center.Our study on the basic of epidemiological investigation projects about functional independence and health status of the elderly.Multistage stratified random cluster sampling method was used in sample draw.Finally,a total of2750 elderly were surveyed and 2705 individuals aged 60 years and over were collected for FIM with an effective rate of 98.36%.Data entry was adopted Epidata3.1 software,implementing double entry principle.Classical Test Theory,Item Response Theory and Different Item Function analysis were jointly used in study method.Data descriptive,basic statistical,Classical Test Theory analysis were conducted by SPSS version 21.0,statistical methods included T test,correlation,principal component,factor analysis.Construction of Structural Equations Model was used Amos 17.0.Item Response Theory and Item Functional Difference analysis were achieved by e Rm,ltm and lordif packages of R3.4.4 software based on Graded Response Model(GRM).Quantitative data was presented in average and standard deviation,qualitative data was expressed as percentage or composition ratio.Inspection level is 0.05 for a two-tailed test.Results1.Of the 2,705 community elderly surveyed,1,131 were males(41.81%);1574were females(58.19%),gender ratio was approximately 1:1.The average age was 69.24±7.58 years old,the elderly aged 60～74 years were defined as Prophase,accounting for75.82%;the elderly aged ≥75 years were defined as Geriatric period,accounting for24.18%.Married seniors were 76.67%.The educational level is mainly elementary school and below,were 51.79%.The prevalence rate of chronic diseases among community elderly was 77.60%,which had two or more chronic diseases accounting for49.35%.Functional independence was mainly conditional and partial independence,were 59.11%;The second is completely independent,were 37.71%;Dependents were3.18%.2.In the seven grade options of the FIM scale,older subjects mainly responded to option 7(Complete independent),option 6(Conditional independent),and option 4(Small amount of help).However,the proportion of option 2(Major help)in response to“Blowel”,option 1(Complete dependence)in response to “Bathtub”,and option 3(Moderate help)in response to “Memory” were relatively high.3.The CR values of the 18 items of the FIM scale were all greater than 3(P<0.05),indicating that the FIM scale has good suitability for each item and can identify high and low scores.All the ? coefficients after removing the current item were greater than 0.90,indicating that the influence of each item on the scale is relatively stable and insensitive.4.The reliability of the FIM scale was very good,and its homogeneity reliability,split-half reliability was better.The correlation coefficient between the 18 items and their dimensions or total scales were greater than 0.40(0.492～0.891,P<0.05).The Cronbach’s? coefficients of the total scale,exercise ability dimension and cognitive ability dimensions was 0.928,0.934,and 0.814(P<0.05),respectively;? coefficients was0.9531,0.9597,and 0.8599(P<0.05),respectively;? coefficients was 0.9545,0.9531,and 0.8243(P<0.05),respectively,all greater than 0.80.5.The validity of the FIM scale was moderate,its criterion-related validity and content validity was fine,but structure validity was general.For criterion-related validity,the correlation coefficient between the two dimensions and the criterion is statistically significant,which reflects the attribute of the desire to measure.For content validity,the correlation coefficient between the two dimensions was 0.564,and the correlation coefficient with total FIM function was 0.964 and 0.786,respectively.Therefore,the correlation coefficient between the two dimensions and the total scale was greater than two dimensions for each other,so the content validity was well.For construct validity,the KMO statistic is 0.938,and Bartlett’s sphere test was significant,indicating that the FIM scale was suitable for factor analysis,a total of 3 common factors are extracted,and the cumulative variance contribution rate was 74.506%,which could defined as Daily exercise ability,Social communication ability and space transfer ability dimension.In the structural equation model,the standardized regression coefficients of the 18 items were concentrated in 0.405 to 0.959.Among the three basic fitness degree evaluation items,two of them were up to standard,nine of the 22 overall model fitness indicators were up to standard,and three of the five internal quality evaluation items were up to standard.Therefore,the model was suitable for general use and the structural validity was moderate.6.The FIM scale accords with the hypothetical condition of IRT model,single dimension and local independence.Estimated the project discrimination and difficulty parameters by using Item Response Theory GRM model.The 18 items of discrimination parameter a were all higher than 0.60,all of which can contribute levels.All item’s difficulty parameter estimation values were in the range of-4.00 to 4.00 except for“Memory”.The difficulty was easier,and there is no retrograde threshold.7.FIM scales of Eating,Grooming,Bathing,Dressing upper,Dressing lower,Toileting,Bladder,Blower,Transfer for bed-chair-wheelchair,Transfer toileting,Bathtub,Walking or Wheelchair,Expression,Social and Solving total 15 items with a high degree of distinction and a narrow range of difficulty.The item characteristic curves were all shown as waveform rules,approximate normal distribution,and moderate density between different curves,it’s implied that this items can reflect the true level of functional independence of the community elderly.On the contrary,in terms of Up and Down Stairs,Understanding and Memory 3 items,there item characteristic curves were not obvious.In the option category superposition curve,It was observed that the potential ability level distribution of the subjects was highly consistent with the reflected category.8.The total amount of information for the 18 items in the FIM scale was 161.26,where the information content of interval [-4,0] was 155.28,interval [0,4] was 1.61.First ten of the total amount of information were Dressing upper,Dressing lower,Grooming,Bathing,Bladder,Toileting,Eating,Blowel,Solving,and Walking.These ten items can reflect the FIM function of the elderly in the community.These 10 items could reflect most information of functional independence of the community elderly,and the estimated information is the most accurate.9.The test information function of the FIM scale did not appear double peaks,multiple peaks or plateau shapes,indicating that the degree of distribution in item information was not too large.10.IFT-based DIF method and equivalent method were used for analysis.There were DIF in some items among different ages and chronic diseases,but there was no DIF between genders.In different ages,the subjects had seven items with DIF,which were Bladder,Up and Down Stairs,Understanding,Expression,Social,Solving and Memory item(P<0.01).Two groups of DTF was 2.13>0,the area of the curve does not coincide.The item characteristic curve for subjects of prophase relatively more to the right,and the options focused on completely dependent and completely independent,while the curves of subjects in geriatric period were relatively left-sided,and the responses focused on the middle option.The FIM scale may be beneficial to prophase.In different chronic diseases,the subjects just had one items with DIF,which was Solving item(P <0.01).Two groups of DTF was 0.04,and the curve areas did not coincide.Conclusion1.Among the elderly in community,most of them are conditional and partial independence,the second is complete independence.But 3.18% have weaker functional independence ability,they show dependence on physical exercise and cognitive capability.2.The FIM scale has great reliability,well content validity,and moderate structural validity in the evaluation of functional independence measure for community elderly.The setting of the 7 classification options is more reasonable.But Memory item should be considered to delete or adjust.3.The FIM scale is suitable for the elderly in the community.It is more suitable for the 60～74 years old people.4.In the analysis of the difference of functional independence of elderly people in different ages,it is necessary to pay special attention to bladder management,upstairs and downstairs,understanding and communication,ideological expression,social communication,problem solving and memory ability DIF.