Outbreak management health IT needs

Part of what makes outbreak management health IT challenging is that outbreaks and public health emergencies are highly variable. Nature and its diseases are themselves variable, but when one maps these diseases across complicated global and national infrastructure and organizations, there are many different ways that events can play out. Ebola hopefully will stay at low case counts in the US and thus not require all of this health IT infrastructure. But if not, or during the next event, any of these needs could be made prominent.

There are functional needs for outbreaks and known technical approaches for addressing them. Outbreak health IT infrastructure needs to largely be in place and in use before emergency events (it is hard to build, train and deploy infrastructure on the fly), but the in-place infrastructure also needs to be flexible enough to support variations in the way that different outbreaks and public health emergencies proceed.

In line with their usual progression, here are some outbreak needs that are shared in part by the Ebola, Anthrax, SARS, MonkeyPox and other emergency events and some conclusions about their health IT support:

1. "Index" case detection. A lot of academic effort and public money for syndromic surveillance has gone into trying to get health IT to help identify the first case of something before it is apparent to health care providers.  For the current Ebola situation in the US, the first Dallas case would be the “US index case.” For a brief period of time before that case appeared, but while there was public health awareness of the outbreak in Africa, having an EHR algorithm that focused on travel history or exposure and influenza-like symptoms might have raised clinical awareness of the circumstance and helped prevent some of the Dallas events. It is not an EHR “issue” that EHRs did not have this, but it is a way that the EHR could have possibly helped. But definitive clinical decision support for low-specificity index case detection is challenging because of vague symptomology and narrative patient history characteristics. The specific operative data are not consistent across EHRs. There are also no standards for distributing decision support algorithms. With the subsequent variations in implementation for decision support in the real world there is then also the strong possibility of automating solutions that produce incorrect answers.

• Analysis: Decision support for “index case” detection is still experimental, hard to implement in EHRs, and should not distract from the established and substantive health IT needs of outbreak management.

2. "Screening" and subsequent case detection. It is important to differentiate the situation following the identification of an index case - where we are now for Ebola - when clinical awareness has already been elevated. The issue then for clinical personnel is not about alerting them to pay attention – they are. It is more about helping them make decisions about borderline “suspect” cases. Unfortunately, in addition to the travel history issues expressed above, fever and other influenza-like symptoms that Ebola presents with are commonplace and non-specific. We are, after all, going into flu season.

As things progressed for US cases it has rapidly reached a point where anyone with any exposure history who presents for care needs some degree of consideration. Well-trained clinical care personnel who reach out to public health are going to be the best “screeners” for some time to come.

• Analysis: By all means implement “wide-net” provision of guidance and methods of follow-up, but don’t overemphasize the automation of whether a patient is a suspect case or not – first do no harm. Recognize the limits of the role EHRs play here and focus more on critical non-EHR health IT.