Internal Dashboards› Corpus› Appeals Corpus› CA IMR Temporal

CA IMR · BH TEMPORAL DRIFT

California Independent Medical Review behavioral-health appeals, 2002–2025. n=446 BH-filtered rows from 10,260 total. TMS dominates volume; ABA dominates overturn rate.

n=446 2002–2025 34.5% OVERTURN

01 Summary

BH Cases

446

of 10,260 total IMR rows

Overall Overturn

34.5%

154 overturned / 292 upheld

TMS Volume

145

44.1% overturn rate

ABA Overturn

83.3%

n=24 — highest rate in corpus

Year Range

24y

2002–2025 inclusive

Peak Year

2011

n=57 (TMS-driven)

Caveat — no payer dimension. The IMR dataset has no structured payer field; payer identity sits in unstructured Findings text and was not extracted in this pass. Drift signals here cannot be attributed to specific health plans.

02 Year × Treatment Heatmap

Cell value is overturn rate. Color intensity scales with rate (deeper green = more frequently overturned). Empty cells = no IMR cases for that treatment in that year. Treatments restricted to those with n≥10 total across the window.

Loading heatmap…

03 Drift Hotspots (≥15pp shift, n≥10 both windows)

Treatment	From	To	Δpp	From rate (n)	To rate (n)
Other	2005–2007	2008–2010	+30.0	3.3% (n=30)	33.3% (n=15)
TMS	2008–2010	2011–2013	+17.5	26.3% (n=19)	43.8% (n=105)

The TMS shift correlates with FDA-cleared TMS device proliferation and post-2010 payer pushback. The "Other" jump is partly classification noise — small-cell instability with n=15 in the second window.

04 Top Treatments by Volume

Rank	Treatment Subcategory	N	Overturn Rate
1	TMS	145	44.1%
2	Other	111	18.0%
3	Lab Work	28	3.6%
4	Applied Behav Analysis	24	83.3%
5	Genetic/Genomic Test	21	19.0%

05 Caveats

No payer dimension. IMR doesn't structure payer; rates cannot be attributed to health plans.
Classification ambiguity. BH filter unions DiagnosisCategory ∈ {Mental Disorder, Autism Spectrum} with treatment-text matches. Treatment taxonomy is coarse — many modalities collapse into "Mental Health" or "Other".
Small-cell instability. Early years (2002–2005) have low volume; year-level rates swing on a handful of cases.
Determination semantics. "Decision Modified" rolled into other, not upheld. Partial reversals not credited as overturns.
Window edges. Final window may be partial depending on data cutoff; treat the latest window's rate as provisional.

Generated 2026-05-20 · Source: stratum-corpus-data · analyses/ca_imr_data/

← Appeals Corpus