To analyze the characteristics (age, gender, educational level, population density, professional status, chronicity, subjective health status, acceptability of IT technologies in healthcare, etc.) that affect the predisposition of Spanish citizens to use Internet-based communication channels (E-mail, blogs, social networks, and online recommendations) to communicate and/or to interact with doctors.
Data from a nationally representative survey of the Spanish population between 16 and 85 years old who use frequently the Internet (July 2011) were used. Name of the Survey: “Use and Applications of Information and Communications Technology in Health” (ONTSI, Spanish National Observatory for the Technology and the Information Society). (N=5505; final analytical simple once missing data were excluded N=3405). Analytical Technique: Boosted Regression Trees analysis (BRT). BRT is a machine learning technique for regression and classification problems. This quite new statistical technique has been demonstrated as a more robust way to predict events in many disciplines, including medicine, biology and geology, comparing to other probabilistic and wider spread techniques such as Generalized Linear Models. This study is one of the first studies in the world using BRT for analyzing survey data.
The BRT model achieved different degrees of predictive precision (using the Area Under the Curve as a goodness of fit measure) through the set of independent variables for each of the communication channels: for the willingness to use e-mail to communicate with his/her doctor, the Area Under the Curve (AUC) = 0.79; for the desire of his/her doctor has a blog or personal website, AUC = 0.736; for the willingness to use social networks to communicate with his/her doctor, AUC = 0.73; for the desire his/her doctor provides recommendations of websites related to health problems, AUC = 0.768. Being young was the most important predictive characteristic to predict the willingness to use social networks (Relative Influence; RI = 21.05%), while population density was the most important predictor in desiring his/her doctor has a blog or personal website (RI = 19.48%). Having a positive perception of the doctor’s use of technology during the appointment was the most important predictor for the willingness to receive recommendations on health-related Internet resources (RI = 18.66%), while having a high educational level was the best predictor for the willingness to use e-mail to communicate with his/her doctor RI = 18.98%).
Majority of Spanish population who use internet frequently is open to use physician-patient Internet-based interaction channels. Nonetheless, the main predictors vary significantly from one channel to another. In order to carry out a successful patient-centered plan to improve patient-doctor interactions, many of these characteristics should be take into account. Thus, the study provides useful resources for policy makers and doctors to decide which kind of Internet-based channel should be implemented first according to their patients’ characteristics and most likely preferences.