by Joyce Frieden, News Editor, MedPage Today, 

WASHINGTON — Measuring outcomes for heart disease treatment in the general population won’t do much good if researchers can’t agree on a single definition of a particular disease or on which outcomes are worth measuring.

That’s why a group of interested organizations is developing a common framework for defining diseases whose treatment outcomes are being studied by researchers. That collaborative process was explained by Richard Gliklich, MD, CEO of the OM1 company, a Cambridge, Mass. technology firm focused on health outcomes, Thursday at the Health Datapalooza meeting sponsored by Academy Health.

This problem of how to define and what to measure has been around for a while, Gliklich said.
He gave the example of studies on the use of a retinal prosthesis in Medicare patients. Of the 11 studies researchers looked at, a total of 74 different outcome measures were reported; of those, only three specific measures were reported in three or more studies, and only four measures had any evidence of validity or reliability. In addition, several outcome measures have been developed specifically for patients with ultra-low vision but were not used in any study, he said.

Why has it been so hard to develop standardized outcome measures? For one thing, “there are different views on what constitutes an outcome,” Gliklich said. “There are different goals in different studies, so there are a lot of purpose-built programs [whose goals] may not be as relevant to a particular patient.”

Although quality measurement “has been ingrained and programmed into physicians, no one asks questions like ‘Why is the number of patients who have a colorectal cancer screening so important?'” and there’s been no organized way to harmonize differences, he said.

That’s why the Outcomes Measures Framework was developed to come up with a conceptual framework with a goal of a common model for classifying the range of outcomes relevant to patients and providers across most conditions. “Five years ago we [began] this task — the first thing to do was come up with a model and have 400 stakeholder friends come together over a couple of years and rip it apart and see what’s left over,” said Gliklich.

Organizations participating in the framework effort include federal government agencies such as the National Institutes of Health and the Agency for Healthcare Research and Quality (AHRQ) as well as independent groups like OMERACT (Outcome Measures in Rheumatology).

The resulting framework (see p. 22 in the linked article) included standardized information on the patients studied (such as demographics, genetics, health behaviors, and environmental exposures), as well as information on the disease being studied (diagnosis, risk factors, staging systems, and biomarkers, among other data); the treatment used; and the outcomes; whether in terms of clinical response, what patients reported, or events of interest.

The next 2 years will involve applying the framework to develop standardized libraries of harmonized definitions for five “starter” conditions, Gliklich said: asthma, depression, atrial fibrillation, and two more that have yet to be decided. The group is focused on data collected in clinical registries because “the things that registries do are now subsumed by what we’re generally trying to do across both research and clinical practice,” so they are good places to being harmonization efforts.

He gave atrial fibrillation as an example of the obstacles the group is facing. Using the AHRQ’s listing of patient registries, researchers came up with 40 registries that were related to atrial fibrillation; they narrowed those down to 12 registries in which at least some data were collected in the U.S., and in which atrial fibrillation was the registry’s primary focus.
They found a total of 26 measures collected across the registries, with no more than five registries collecting any single measure. The most popular measure collected was stroke or predictors of stroke.

Definitions also varied across the registries, and differences like these can lead to vastly different results in evaluating treatments, said Gliklich. For example, here are the definitions of transient ischemic attack (TIA) used by three of the registries:

  • “New focal neurological deficit with rapid symptom resolution (usually 1-2 hours), always within 24 hours; neuroimaging without tissue injury”
  • “Sudden onset of a focal neurologic deficit … from a nontraumatic cause and categorized as ischemic, hemorrhagic, or unspecified, as evidenced by neuroimaging or lumbar puncture lasting less than 24 hours”
  • “An acute focal neurological event lasting at least 5 minutes”

“They’re similar, but they’re different,” he noted. “If you have a registry that has an event that last 3 minutes, it qualifies in two of them but not a third … I’m not bringing this out to criticize any registry; I’m just trying to show you what challenge is as our group tries to bring together outcomes measure developers so our team can work with them and get to harmonization.”

After figuring out what the common measures are and then harmonizing them — a task Gliklich referred to as “hand-to-hand combat” — the next level would be to put them into standardized code sets to get them “shovel-ready” to be put into an electronic medical record, he said.

Source: MedPage