Quality reporting and structured data: What it means for ambulatory EMRs

July 19th, 2017 / By L. Gordon Moore, MD

Every year, most health plans are rated on quality by the National Committee for Quality Assurance (NCQA) with Healthcare Effectiveness Data and Information Set (HEDIS). HEDIS gets most of its information from submitted claims and applies a set of rules. For example, if an insured person of appropriate age has a claim for a colonoscopy, that counts towards the plan’s ranking of effective colon cancer screening.

But sometimes there is no claim for a colonoscopy even though the person has been appropriately screened. This might happen if the person had the colonoscopy within the appropriate screening window but before they joined their current health plan. This is one of the many known shortcomings of adjudicated claims data. 

The person’s primary care provider may know that the person has indeed been screened, so the NCQA allows for submission of data to supplement the claims – i.e. they allow clinicians to submit something along the lines of “I know that person X has had a colonoscopy on Y date.”

Here’s where it gets a bit sticky: It would be great if a PCP could click the “submit all supplemental data” button on their EMR, but the data for many of these measures are not structured – we find the data in dictated or written notes: “Patient reports normal colonoscopy 5/’15 with prior practice’s GI clinic.”

Supplemental data reporting might then entail:

  • Creating a list of all the patients eligible for colorectal cancer screening who have no claims based evidence of screening
  • Reviewing each chart
  • Making a note of each instance of documented appropriate screening
  • Submitting the supplemental report for colorectal cancer screening
  • Repeating the entire process for mammography, well child checks, vaccines, and many other measures

Sophisticated natural language processing (NLP) and rules engines outperform simple word search engines because of the following abilities:

  • Work around misspellings “has had a colonoscopy”
  • Negation “Patient has not had a colorectal cancer screen”
  • Correct person “His wife has had a colonoscopy”
  • Correct time window “Last colonoscopy was 2001”
  • Exclusions “S/P total colectomy due to Crohn’s”
  • Other representations “has had normal annual fecal occult blood testing”

The health plan noted above is putting a significant amount of time and effort into helping the PCPs report supplemental data. They are willing to perform the chart reviews on site or remotely if provided EMR access, they offer secure FTP portals and forms for supplemental data reporting, they inform the physicians of the reporting windows and cut off dates.

Because of the amount of time and effort being spent extracting unstructured data, ambulatory EMR companies would do well to provide sophisticated NLP and rules engines. This would likely become a significant market differentiator given the intensity and frequency of concerns expressed by physicians around the work burden of quality reporting.

L. Gordon Moore, MD, is senior medical director for Populations and Payment Solutions at 3M Health Information Systems.


Natural language processing meets health care. Learn more.