The Research Institute at Nationwide Children's Hospital, Columbus, Ohio 43205, USA.
Telemed J E Health. 2010 Mar;16(2):154-60. doi: 10.1089/tmj.2009.0102.
This analysis tests the hypothesis that health information search can be modeled using the behavioral model, a tool traditionally used for other healthcare behaviors.
The Pew Internet and American Life August 2006 Survey was used to model five selected Internet health information seeking behaviors: information on a specific disease, diet and nutrition, mental health, complementary and alternative medicine, and sexual health. Each behavior was modeled using hierarchical logistic regression with independent variables of predisposing factors (age, race, sex, and education), enabling factors (home Internet access, Internet experience, and high-speed access), and need factors (health status, chronic health condition, and current health crisis).
Health information search is not a monolithic behavior. Sex, age over 65, current health crisis, and regular use of the Internet were the most consistent predictors of use, each being significant in four of the five models. Blacks (odds ratio [OR] = 0.50, 95% confidence interval [CI] 0.34-0.74) and Hispanics (OR = 0.59, 95% CI 0.37-0.95) were significantly less likely than whites to search for information on a specific disease or condition but blacks (OR = 2.73, 95% CI 1.69-4.43) were more likely than whites to search for sexual health information and Hispanics (OR = 1.72, 95% CI 1.09-2.73) were more likely than whites to search for complementary and alternative medicine information. Pseudo-r(2) for the fully specified models ranged from 0.13 for mental health search to 0.32 for specific disease search.
Health Internet behaviors can successfully be described using models designed for traditional health behaviors; however, different health information seeking behaviors have different user profiles.
本分析检验了以下假设，即健康信息搜索可以使用行为模型进行建模，该模型是传统上用于其他医疗保健行为的工具。
使用皮尤互联网和美国生活 2006 年 8 月调查对五种选定的互联网健康信息搜索行为进行建模：特定疾病信息、饮食与营养、心理健康、补充和替代医学以及性健康。使用层次逻辑回归对每个行为进行建模，独立变量为倾向因素（年龄、种族、性别和教育）、促成因素（家庭互联网接入、互联网经验和高速接入）和需求因素（健康状况、慢性健康状况和当前健康危机）。
健康信息搜索不是一种单一的行为。性别、65 岁以上、当前健康危机和定期使用互联网是使用最一致的预测因素，这四个因素在五个模型中的四个中均具有统计学意义。黑人（比值比[OR] = 0.50，95%置信区间[CI] 0.34-0.74）和西班牙裔（OR = 0.59，95%CI 0.37-0.95）搜索特定疾病或状况信息的可能性明显低于白人，但黑人（OR = 2.73，95%CI 1.69-4.43）更有可能搜索性健康信息，而西班牙裔（OR = 1.72，95%CI 1.09-2.73）更有可能搜索补充和替代医学信息。完全指定模型的伪 r(2) 范围从心理健康搜索的 0.13 到特定疾病搜索的 0.32。
可以使用为传统健康行为设计的模型成功描述健康互联网行为；但是，不同的健康信息搜索行为具有不同的用户特征。
