Validity and reliability of the SF-36: the Malaysian context

Results of construct validity and reliability of the SF-36 are described, based on data from a multi-centre study on asthmatics and a population based survey. Questionnaire refinement was carried out between the two studies. Quality of data was good, with all items having less than 0.5% missing values. Floor and/or ceiling effects were observed for REE, REP, PF and SF. For scaling assumptions, correlations between each items and its hypothesized scale were all above 0.50, except for one item in PF. and for both items in SF. Item discriminant validity was an issue for items in VT, SF and MH scales. Cronbach's as for all scales exceeded the recommended 0.70 level, except for SF. Only one latent dimension was identified in principal component analysis, and only 52-53% of variance accounted for. As expected, PF shows high correlations with the physical component while MH was highly correlated with the mental component. Contrasting findings in the loadings of other scales were observed in the asthma data. Age, disease severity and presence of self-reported handicap/disability significantly affect PF, while MH demonstrates no obvious pattern with declining age. In essence, the Malay version of SF-36 could be used in Malaysia, with its generally acceptable internal consistency and validity. The caveat is in the call for additional domains of importance to Malaysians that is not covered by the instrument, and in the caution to be employed when using and construing the instrument.