Now Reading: About today’s Health Affairs study, HIT, EHRs, and impact on test ordering

This study, being released to the public at the exact same time of this blog post, should generate healthy discussion about whether health information technology, fully implemented, is going to deliver on one of the biggest promises made: reduced ordering of unnecessary tests.

This paper should be read carefully. It is very well done AND it does not answer that exact question – it wasn’t intended to, read on.

It answers the question about the era of electronic access to test results and images by physicians and what that has done to ordering of lab and imaging tests across a national sample of office visits. It is not focused on a single type of organization, it uses a nationally recognized survey tool (NAMCS) and is written by nationally recognized authors, from a pretty good journal :).

I got to speak with lead author, Danny McCormick, MD, about his motivation to do the study and to see if I understood the implications correctly. Danny’s currently a physician at Cambridge Health Alliance (who I so enjoyed visiting previously), on the faculty of Harvard Medical School.  He related his experience seeing the predictions of cost savings by the now-infamous RAND study in 2005 while he was serving under Sen. Edward Kennedy in Washington, DC, on an RWJF Fellowship. I remember that study, too, and its rosy predictions.

Danny wondered if test ordering might stay the same (and not decrease as predicted by RAND), and not only found that they stayed the same, but they increased proportionally for patient visits at practices where electronic test results and images were available.

Exhibit 2: A different odds ratio for visits in integrated care/HMO practices. Courtesy of Health Affairs.

Integrated care and Community Health Center care bucks the trend toward more tests

I asked him about a finding of interest to me, which is that in the models there are some exceptions, most notably for visits that occurred in the “HMO Office” setting, where there was statistically significantly LESS ordering of imaging compared to private medical offices.  Same is true for visits that occurred in Community Health Centers. Again, care must be taken in interpretation – across visits in all places, 40 – 45% greater likelihood that imaging tests were ordered when results/images were available electronically, I am showing the part of the model that compares different practice settings within that overall result.

To the left is the part of the table, Exhibit 2, that shows this difference (reproduction of just this part permitted by Health Affairs). I recommend reviewing this table, and also Exhibit 3 in the same article, which shows the same analyses of just MRI, CT, and PET, where the HMO Office and Community Health Center visits do not show any difference from their private office peers.

Excerpt of NAMCS survey tool, from CDC website

The reason I equate the “HMO Office” category with “Integrated care” is because of the way the NAMCS survey is designed. If you look at it, this is clearly the category for physicians practicing at Kaiser Permanente – it says Kaiser Permanente right there on the survey.

Danny cautioned me, appropriately, that the numbers of visits in this setting were small compared to the other settings (paralleling the number of such visits in the United States). The statistical significance holds, but since this is a big picture study, we don’t know what’s going on in this setting compared to others, exactly, just that there’s less association with test ordering.

Lots of caveats

Here’s my list, feel free to add yours in the comments:

  • This is based on electronic access to test results not EHR use. Therefore it doesn’t really test education/alerts/prompts (aka “clinical decision support”)
  • This is based on 2008 data, which is an eternity in EHR terms – even if an EHR was in place, it’s unlikely it had good decision support and would have modulated physician ordering habits
  • If an EHR was present in the practice, it didn’t significantly alter the results (they did this analysis too).
  • We don’t know that more tests are a bad thing in all of these cases. For example, these systems may alert physicians to order preventive tests, like mammograms (see great example below). In the area of advanced imaging, there is a common problem of follow-up exams not being done. It’s possible that seeing the results of the last test that said,  “need follow-up exam,” would cause necessary care to be performed.

HIT + Patient and Family Access + Leadership = SUCCESS

This is a helpful, well done study by a talented group of researchers.

I never banked on the predictions of the RAND researchers, because they did not include a key ingredient in the recipe – leadership. HIT + Leadership = success

The most important power of health information technology is what it does for patients and how it connects them to their health care. With an EHR, there’s an efficient way for patients and families to understand what’s going in their care and change the conversation from being recipients “in front of the counter” to understanding and helping to fix what’s behind the counter. It’s revolutionary in that regard.

With regard to leadership, I’ve shown Mary Gonzales’ story on here before, here’s a new one, from the Kaiser Permanente Care Stories Blog, about what happened when Dale Gordon went in for a swollen knee, a lab test was ordered, and his life was saved. This requires the use of technology, the leadership to make a difference in care using the tool, and the work to integrate it into a system at all points for our patients. So not more tests, not less tests, the right amount of tests, which is now possible – check it out:


So, the ‘model’ doesn’t alter ordering habits? @tedeytan : About today’s Health Affairs study: HIT, EHRs, and impact on test ordering

@Gregmogel Visits in a practice setting like Kaiser Permanente had less imaging ordering relative to private practice settings.

Ted Eytan, MD