Objective: To discover principal components among assessed items from the WHO-SF36 survey by principal component analysis (PCA) and then to establish the relationship between the candidate principal component and the incidence of ADRs induced by nabumetone in Shanghai osteoarthropathy patients.
Method: A total of 145 patients were interviewed using the WHO-SF36 questionnaire for quality of life (QOL) assessment before the administration of nabumetone. The sub-items of the questionnaire were analyzed using PCA and several comprehensive variables were established. Relationships between these newly formed variables and the overall incidence of adverse drug reactions (ADRs) caused by nabumetone were evaluated using univariate and multivariate analyses.
Results: Several principal components were identified and their linear parameters were estimated using PCA. Through univariate analysis, only 1 principal component--"adverse effect on work and daily activities as a result of emotional problems"--was found related to the incidence of ADRs. The odds ratio (OR) was 1.28 with 95% confidence intervals (CI) of 1.11 and 1.48, p = 0.0384. This result was validated using a multivariate logistic analysis performed on all the alternative candidate covariates, including family income, a history of ADRs on NSAIDs, the course of the disease, level of education, control of stress, coffee consumption and consumption of salty food. The covariate information was taken from other parts of the clinical report form (CRF) used in this research. The analysis proved that the principal component, adverse effect on work and daily activities as a result of emotional problems, was an independent factor related to the overall incidence of ADRs. The odds ratio (OR) from the logistic analysis was 1.34, with 95% confidence intervals (CI) of 1.16 and 1.55, p = 0.0309.
Conclusions: The principal component identified can be applied to overcome some limitations in the WHO-SF36 questionnaire such as high correlation between variables, information overlaps and weak representation of variables and the statistical data analysis of QOL can therefore be made more effective. The study shows that the principal component "adverse effect on work and daily activities as a result of emotional problems" is an independent predictor of the overall incidence of ADRs induced by nabumetone.