The emerging trend in treatment decision making for complex medical conditions is shifting to a collaborative model, where patients actively participate as decision makers. To facilitate this, online medical information must be fitting to the purpose of making sound treatment decisions. Underpinned by Herbert Simon's Nobel-winning work Administrative Behavior, we examine the treatment decision process in the context of Internet health information provision. Based on Simon's propositions, we propose a set of decision support features for evaluating Internet health information websites. We then use them to evaluate official Australian and international websites for Autism Spectrum Disorders (ASD), where the need for treatment decision support is prominent due to the diverse nature of the condition. The evaluation results indicate a general inadequacy in terms of Relevance under the Value proposition among the websites. Genuine individualized information based on patient's profile of symptoms, delivered by intelligent personalization technologies, should be further deployed to address this inadequacy. The theoretical contributions of this paper are: i) bridging online information quality model and classical decision making theory, and ii) analyzing online information provision based on Simon's bounded rationality. Whilst, the practical contributions are: i) to provide a set of decision support features as guidelines for ASD websites that ii) are extendable to other health domains and different technologies.