Everyone is talking about the ways healthcare analytics are going to revolutionize healthcare. I have not seen anyone even suggest that the future of healthcare is built on anything but data. Given the widespread acceptance of data and analytics as the future of healthcare, let us take a minute to explore two impediments that could hijack this future or at least dramatically slow its progress. Those two impediments are trust and security.
Trust is a tricky thing and past EHR implementations haven’t done us any favors as they have more often than not eroded trust in technology as opposed to building trust. However, healthcare providers trusting the technology and the data that technology creates will be essential to a successful healthcare analytics program. Healthcare providers will do nothing that they think could harm their patient or their reputation, so it is absolutely essential to ensure your providers trust the analytics you are providing them.
The first step to building this trust relationship is ensuring the data you use to create the analytic insights is clean and accurate. Nothing burns your trust with a provider more than presenting them an analytics-based insight which was glaringly incorrect because the data was wrong. A common example of this is having the wrong gender. The abnormal pap smear or pregnancy test insight massively erodes trust when the healthcare provider knows their patient is male.
Along with ensuring the data being used is accurate, another key to building trust is making sure you are not missing important data. While this is obvious from the outside, I have regularly seen where an analyst thought they had all of the correct data, but some flaw in the data extraction or analysis caused the data to be missing from the analytics being done. It is very easy for this to happen and often hard to discover when there is a problem.
Another challenge analytics face is physicians trusting the analytic to be accurate. Humans are infinitely unique and it is easy for a provider to imagine a scenario where the analytic could fail. Proving to them that the analytic is effective and should be trusted is a challenge that currently is best solved by a clinical trial. Hopefully, some other techniques come forward that are trusted by doctors because clinical trials take so long. For example, we have seen some examples where historical data proved a certain treatment was more effective and therefore doctors trusted the data based on the historical record.
Related to trust is the need for healthcare analytics security. Poor security is the ultimate violator of trust. This is true for providers and patients.
When providers hear of a healthcare IT system being compromised, it creates a small doubt in their mind about the accuracy of the data that remains. Doubt and trust cannot exist together. When patients hear about security compromises, they are less willing to share their data with their healthcare provider. Both of these security issues can lead to lack of trust in any healthcare analytic solution. This is why healthcare security must be an essential part of any healthcare analytics effort.
The problem with trust is that it is difficult to build and easy to tear down. One mistake in the analytics provided can ruin years of trust. We have to be careful when this happens to not react emotionally to a mistaken analytic. When mistakes happen, a healthcare organization must evaluate the analytic and understand why it failed. Was the error a gross error in the analytic that is impossible to correct? Or was the error a rare situation that is very unlikely to ever happen again? Was the good the analytic accomplished of greater benefit than the one error it missed? Can the error be fixed so it does not happen again?
The reality is that we should be careful how we judge healthcare analytics. We may discover some to be poorly designed and need to be thrown out after further evaluation. However, no healthcare analytic is perfect just like no human is perfect. When evaluating the effectiveness of healthcare analytics it is essential to compare them against the alternative versus comparing them against perfection. In most cases, the alternative is a human with plenty of imperfections as well.
While healthcare analytics will likely never be perfect, they can be demonstrably better than the alternative. However, we must take special care when implementing healthcare analytics that we do not ruin the trust relationship with the provider. Once you lose providers’ trust, it takes years to recover and rebuild that trust which will set your analytics efforts way behind. On the other hand, a trusted analytics program sets a great foundation for your healthcare organization.