Monday, July 16, 2012

Evidence that EHRs can promote better preventive care

In a previous post, I summarized the mixed evidence that using electronic health records with clinical decision support systems (CDSSs) improves processes and outcomes of preventive care. Most of the literature supporting a positive effect of EHRs with CDSSs is either anecdotal, observational, or limited to specific settings (namely, practices of employed physicians in large, integrated health systems). I concluded that it remained uncertain if these results could be duplicated in randomized trials in typical private practices.

Uncertain, that is, until last week, following a study by Alex Krist and colleagues, "Interactive Preventive Health Record to Enhance Delivery of Recommended Care: A Randomized Trial," in the Annals of Family Medicine. The authors randomized 4500 patients from 8 primary care practices in Virginia to usual care versus an invitation to use MyPreventiveCare, an interactive personal health record (IPHR) that generates a list of screening and counseling recommendations based on guidelines from the U.S. Preventive Services Task Force, as well as links to more detailed explanations of these preventive services and relevant decision aids. When patients used the IPHR, their physicians also received a summary of the recommendations in their EHRs.

Although only about 1 in 6 invited patients had actually used the personal health record 16 months later, it was enough to show a statistically significant difference between the proportion of patients in that group who were up-to-date on all recommended services, compared to the control group. After 4 months, colorectal, breast, and cervical cancer screening rates had increased by an impressive 13 to 19 percent among personal health record users. The authors concluded: "Information systems that feature patient-centered functionality, such as the IPHR, have potential to increase preventive service delivery. Engaging more patients to use systems could have important public health benefits." A how-to guide on the use of personal health records to promote uptake of recommended preventive services is available on the Agency for Healthcare Research and Quality website.


A modified version of this post first appeared on the AFP Community Blog.


  1. It's nice that it showed a statistically significant change in patient behavior. It would be nicer to know if it resulted - or at least could be projected to result - in a meaningful change in outcomes.

    Not the same thing.

    A well done study with an adequate sample size can show a statistically significant difference that is still too small in real terms to be meaningful. Statistical significance tells us about precision and the likelihood that the difference is chance - but NOTHING about impact in the real world.

    If you look at the study, the absolute (rather than relative) difference in the number of patients offered the service who are up to date is 2.3 per 100. Use of the system resulted in a 12 out of 100 increase in up-to-date patients. How many thousands of patients would have to be offered (or use) a system like this to prevent a bad outcome?

    Mind you, I'm not saying this is a bad idea or a bad study. I love the idea and think the study is useful. I am only objecting to the presentation solely in terms of 'statistical significance' in the article and in this post, without an attempt to relate the results to the real outcome behind the surrogate marker.

  2. Good point, which I argued myself in a recent editorial: (login required)
    But to play devil's advocate, all of the interventions here have been shown to positively affect patient-oriented outcomes that matter (reduced mortality) in randomized controlled trials, and showing that EHRs reduce mortality would require a very large trial that would take a decade or more. And maybe it's a trial worth doing, but this study is an important prerequisite.