Where Does Lyra’s 26% Cost-Reduction Claim Come From?

Lyra says its mental health benefit produced an average 26% annual reduction in employer health-care costs, sustained over four years.

The 26% has a straightforward source in the May 2024 study. But understanding what the number measures is important before treating it as a savings assumption.

The 26% is a comparison between two groups

For each year from 2018 through 2021, the May 2024 study compared the annual medical and prescription-drug claims of members who used Lyra with matched members from the same employers who did not.

The reported results were:

The study reports an average absolute difference of $1,978 per participant per year. Its annual “efficiency ratios” correspond to relative participant/nonparticipant differences averaging approximately 25.7%, which rounds to Lyra's 26% figure.

So there is nothing mysterious about the arithmetic.

Does the arithmetic permit a purchaser to conclude a 26% decline in participant costs?

A statement that costs were “reduced 26% annually” can sound as though participant costs fell after Lyra was introduced. That is not what the study measures.

Participant claims were $4,993 in 2018 and $6,547 in 2021. The study separately reports a 31.1% cumulative increase in participant costs over the period.

The 26%, therefore, does not mean that costs were 26% lower after Lyra than before Lyra.

It means approximately that annual claims for Lyra users are 26% lower than claims for matched nonusers.

Those propositions are analytically different.

The study does not contain an untreated pre-Lyra baseline

The report calls 2018 a baseline year, but all three employers had implemented Lyra either before or at the beginning of the January 2018 study period. There is no period in the study showing what participant spending looked like before Lyra became available.

A reduction ordinarily requires a reference point: what would these costs otherwise have been? The May 2024 study constructs that reference point from matched nonparticipants rather than from the participants' own pre-Lyra experience.

That is a legitimate observational comparison. It is not the same as observing a 26% cost reduction following implementation.

What does “sustained over four years” establish?

The study also supports a narrower interpretation of “sustained.”

Participants had lower claims than matched nonparticipants in each of the four annual comparisons. That recurring pattern is meaningful. But the analysis matches participants to nonparticipants for each year. The report also states that it does not examine continuous engagement in Lyra programs.

The study therefore demonstrates:

Four consecutive years in which that year's Lyra users had lower matched claims.

It does not establish:

A single cohort whose costs were reduced by approximately 26% and remained 26% lower for four years.

What can a purchaser rely on?

The May 2024 study provides evidence of a recurring and substantial claims difference between Lyra users and matched nonusers. The 26% is a reasonable shorthand for the average size of that observed relative difference.

The study measures approximately 26% lower claims among participants relative to matched nonparticipants. It does not directly measure a 26% decline in participant spending caused by Lyra.

For a purchaser evaluating the claim, that distinction determines what comes next. The 26% can identify an economic result worth investigating. Before it becomes an underwriting assumption, the purchaser still needs to determine how much of the matched difference was produced by Lyra and how that participant-level effect translates to the full population being purchased for.