4 in 5 US adults seek healthcare advice online, making social media and online searches common tools to study health conditions’ prevalence, outbreaks, and evolution. We study the suitability of 2 types of online data &, difference b/w between different online data and official prevalence of diseases for public health research. We propose a unified framework that allows us to 1) compare active (Reddit posts/ comments) and passive (Google Searches) engagement across conditions, revealing when and why one outperforms the other, and 2) explain the biases inherent in both types of online health data. Combing the largest online derived medical taxonomy from Reddit and Google Trend Searches for a set of 16 equivalent conditions in the two data sources, we show that there is implicit knowledge to be gained from looking at cases where official, active, and passive derived online scores diverge (i.e., higher or lower online engagement in health discussions about conditions compared to their official prevalence). Specifically, we find that: 1) online searches (passive engagement) more closely resemble true prevalences than social media discussions (active engagement) for most but not all conditions, 2) the divergence between official scores and those derived from active engagement on social media discussions is explained by state socio-demographic, personality & cultural indicators to a much higher degree than the divergence between official scores and those derived from passive engagement, i.e. online searches. Drawing from our results, certain guidelines emerge for interpreting online health data. Firstly, while search trends seem to normally outperform discussion based scores, caution is required, and comparative approaches recommended. Divergences can illuminate potential issues, including the stigma associated with conditions, the severity of conditions, and other variables impacting how individuals actively discuss their medical conditions.