MyMedR

Displaying all 4 publications

Abstract:

Sort:

Fulltext Effects on parameter estimates and goodness of fit measures: comparing item-level and item-parcel models in structural equation modeling

Ainur Amira Kamaruddin, Yap, Bee Wah, Sayang Mohd Deni

Pertanika Journal of Science & Technology, 2020;28(2):627-647.
MyJurnal

The assessment of model fit is important in Structural Equation Modeling (SEM). Several goodness-of-fit (GoF) measures are affected by sample size and the number of parameters to be estimated. A large sample size is needed to test a complex model involving a large number of parameters to be estimated. One of the solutions to reduce the number of parameters to be estimated in a given model is by considering item parceling. The effects of item parceling on parameter estimates and GoF measures in a structural equation model was investigated via a simulation study. The simulation results indicate that the parameter estimates are closer to the true parameter values for the IL model whenever the distribution of data is normal but biased when the data is highly skewed. The parameter estimates for the IP model were found to be underestimated for both normal and non-normal data. The GoF measures were higher for the IP model. Additionally, the RMSEA was lower for the IP model when data were skewed. This shows that item parceling may improve GoF measures but the effect of exogenous on endogenous variable is underestimated. Application to a real data set confirmed the results of the simulation study.
Fulltext Classification of familial hypercholesterolaemia using ordinal logistic regression

Muhammad Hamizan Jamaludin, Yap, Bee Wah, Hapizah Mohd Nawawi, Chua, Yung-An, Marshima Mohd Rosli, Annamalai, Muthukkaruppan

Pertanika Journal of Science & Technology, 2020;28(4):1163-1177.
MyJurnal

Familial hypercholesterolaemia (FH) is a genetic disease that causes the elevation of lowdensity lipoprotein cholesterol (LDL-C), which subsequently leads to premature coronary heart disease (CHD). Features which have been reported to be associated with FH include lipids level, tendon xanthomata, and history of CHD. The Ordinal Logistic Regression model using the classification of FH patients with the Dutch Lipid Clinic Network Criteria (DLCN) as the dependent variable (where 1=Possible, 2=Probable, 3=Definite) was developed and evaluated for different types of link functions. The FH patients (n = 449) were recruited from health screening programmes conducted in hospitals and clinics in Malaysia from 2010 to 2018. Results indicate there is a significant association between FH categories with demographic factors (ethnicity and smoking) and physical symptoms (corneal arcus and xanthomata). The Ordinal Logistic Regression using Cauchit link function has lower Akaike Information Criterion (AIC) value, higher Nagelkerke’s R-Square and classification accuracy compared to Probit and Logit link function, diastolic blood pressure, corneal arcus and xanthomata were found to be significant covariates of FH.
Assessing Diabetes Distress Among Type 2 Diabetes Mellitus in Malaysia Using the Problem Areas in Diabetes Scale

Jannoo Z, Yap BW, Khan NM, Farcomeni A

Value Health Reg Issues, 2019 May;18:159-164.
PMID: 31082796 DOI: 10.1016/j.vhri.2019.03.004

OBJECTIVES: To validate, from a psychometric perspective, the Problem Areas in Diabetes (PAID) questionnaire in patients with type 2 diabetes mellitus from Malaysia.
METHODS: A total of 497 patients with type 2 diabetes mellitus were recruited from public hospitals in the state of Selangor through convenience sampling. Construct validity was evaluated through confirmatory factor analysis. Internal consistency of the instrument was tested by Cronbach α. Criterion validity and discriminant validity were also used.
RESULTS: The PAID instrument consisted of 3 factors: social support problem, food-related problem, and emotional distress problem. The Cronbach α values of the 3 factors showed adequate internal consistency with α values greater than 0.90. The present confirmatory factor analysis model achieved a good fit with a comparative fit index value of 0.923. Satisfactory criterion validity was also demonstrated because there existed positive significant association between glycated hemoglobin A1c and diabetes duration.
CONCLUSIONS: The PAID questionnaire in Malaysia was found to be a reliable and valid instrument exhibiting good psychometric properties.
An audit of diabetes-dependent quality of life in patients with type 2 diabetes mellitus in Malaysia

Jannoo Z, Yap BW, Musa KI, Lazim MA, Hassali MA

Qual Life Res, 2015 Sep;24(9):2297-302.
PMID: 25800728 DOI: 10.1007/s11136-015-0969-8

PURPOSE: The aim of this study was to evaluate and validate the ADDQoL and to assess the impact of diabetes on QoL among the type 2 diabetes mellitus patients in Malaysia.

METHODS: The Malay and English versions of the ADDQoL questionnaire were administered to patients attending routine outpatient visits in three primary hospitals and a public clinic. The construct validity of the ADDQoL was validated using confirmatory factor analysis (CFA). The sample comprised 350 Malay respondents who rated the ADDQoL Malay version and 246 non-Malay respondents (Chinese or Indian) who answered using the ADDQoL original English version.

RESULTS: CFA confirmed the presence of one-factor structure for both samples. The internal consistency was high with Cronbach's alpha values of 0.945 and 0.907 for the ADDQoL Malay and English versions, respectively. Results showed that for all three ethnicities, the most important domain is 'family life'. Overall, Malay patients stated their 'living conditions' is the most negatively affected, while for Chinese and Indians, diabetes has the greatest impact on their 'freedom to eat'.

CONCLUSIONS: The ADDQoL was found to be culturally appropriate, valid and reliable among Malay- and English-speaking type 2 diabetes mellitus patients in Malaysia.

Study site: routine outpatient visits in three primary hospitals and a public clinic