Wednesday, August 29, 2018

Very Brief Blog: MolDx Analysis of Evidence "Quality, Strength, and Weight"

The 21st Century Cures Act requires LCDs to provide rationales for their decisions and a summary of evidence considered.   For an overview, see here.  For the actual 197 words in 21CC about LCDs, see here.

CMS may eventually provide a substantial rewrite of the LCD manual (the current manual is here).  In the meantime, CMS requires LCDs to have one demarcated section which is a "summary of the evidence" and another section which is an "analysis of the evidence."[*]   (This is exactly the same format used in NCDs.   Generally, a "summary" is objective and an "analysis" assigns interpretations, critique, and value judgement.)   

An example of an NGS MAC LCD for a surgical procedure which includes detailed "summary" followed by detailed "analysis" is here.

MolDx Quality, Strength, Weight

MolDx LCDs over the last year have provided a summary assessment of diagnostic tests categorized as "Quality, Strength, Weight."   Frankly, even with a decade of consulting and several publications under my belt, I wasn't sure I could predict exactly what MolDx authors meant by their use of these three words.   Discussion at a California Clinical Laboratory Association meeting this month helped clarify the meaning.

Quality refers to the quality of the trial and study design.  For example, a randomized controlled trial that is double-blinded is higher quality than a retrospective case series.   This is close to a fairly classic "level of evidence" hierarchy (level of evidence 1,2,3,4, etc).  This metric focuses on quality of the trial or study; a 1000-patient RCT may be a very high quality trial, even if it shows the therapy under study is neutral or harmful. 

Strength is the strength of the resulting evidence for Medicare purposes.   For example, a study conducted in children in a country with a wholly different health system and alternatives could be a 1000-patient RCT, but it would have low "strength" for Medicare decision making.   Strong evidence would be large effects accompanied by low p-values, and strength of evidence is additive across multiple trials.   Evidence for Medicare coverage has less strength if it is 95% in patients age 40-55, and not the main Medicare age of >65.[*] 

Weight of evidence is conclusory:  Given the evidence reviewed (perhaps other evidence from other studies, guidelines, or alternatives), how strong is the weight of evidence for Medicare coverage.  Although I am adding my own terms, I suspect the weight of evidence would be strong if it is easy to conclude that, all factors and concerns in the balance, the benefits considerably outweigh the risks or uncertainties. 

Snapshot from a MolDx LCD

[*] The requirement to discuss and summarize evidence considered, and then provide an "analysis" of that evidence, would tend to prevent blanket non coverage LCDs.  For example, an LCD that simply lists 100 Category III codes with a sentence stating they are noncovered would not meet the standard of summarizing and analyzing any evidence for the 100 different services represented by the codes.

[**]  I believe "weight" of the evidence would include the concept, seen in some health technology assessments, is additional evidence likely to materially change the estimate of benefit or risk.  For example, over a large population, there have been very large studies of mammography, and it has a benefit, but/and the benefit is fairly small.   It is unlikely another new study would find that mammograms are harmful or that they have a huge benefit.