A regression-based study using jackknife replicates of HINTS III data: Predictors of the efficacy of health information seeking

The current study determines and assesses the effects of the statistically significant predictors of the efficacy of health information seeking through a regression-based analysis of the 2007 edition of the Health Information National Trends Survey (HINTS) data. The HINTS III data were collected through list-assisted random digit dialing and mail-in questionnaire with a natural corresponding unstratified and cluster sampling design with jackknife replicates and were analyzed using generalized linear models with jackknife parameter estimation based on the complete and 50 jackknife replicate datasets. The resampling-based analytic approaches, such as the jackknife and bootstrap, generally provide unbiased parameter estimates and are the preferred methods for complex survey data analyses. We implemented an exhaustive search through all potential predictors of the efficacy in health information seeking combined with model building based on forward selection and backward elimination of covariates to derive the best predictive model. This model-based and datadriven approach to detect and assess the relative effects of the significant predictors of the aforesaid outcome variable of interest is a greatly advantageous alternative to the common hypothesesbased analyses. Our results show that numeracy, education, patient health care satisfaction (with the health information given by their health provider), health information dissatisfaction, general health, and psychological distress are the optimal covariates significantly associated with the efficacy of health information seeking. Interestingly, many usually important background covariates such as race, income, gender, geographical location, and others were not significant predictors of the outcome variable of interest. The conclusions of our analysis reveal new insights into the complexity of the efficacy of health information seeking and will undoubtedly have important implications on the design and success of future health care messages