Download presentation
Presentation is loading. Please wait.
Published byShanon Norton Modified over 7 years ago
1
Medicare’s G-Codes: Are they in need of Rehabilitation?
Pedro Gozalo, PhD Linda Resnik , PT, PhD Benjamin Silver
2
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Background Medicare’s Rapid Growth of Outpatient Therapy: 2011: 4.9 m. (15%) beneficiaries, $5.7 b. 45,000 providers billed Medicare 2006: 3.5 m (9.7%) beneficiaries, $4.1 b. PT 72%, OT 19%, SLP 9%. Average Episode: 33 days, $1,173. Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
3
Medicare Outpatient Therapy Spending By Setting
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
4
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
G-Codes Middle Class Tax Relief and Jobs Creation Act of 2012 Mandatory July 1, 2013 Providers need to report functional limitations G-codes and Severity Modifiers when billing Medicare for outpatient therapy services Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
5
G-Codes for Functional Reporting
Current Status Goal Status Discharge Status Functional Limitation Descriptor G8978 G8979 G8980 Mobility: walking around & moving G8981 G8982 G8983 Changing & maintaining body position G8984 G8985 G8986 Carrying, moving & handling objects G8987 G8988 G8989 Self care G8990 G8991 G8992 Other physical or occupational therapy primary G8993 G8994 G8995 Other physical or occupational therapy subsequent G8996 G8997 G8998 Swallowing G8999 G9186 G9158 Motor speech G9159 G9160 G9161 Spoken language comprehension G9162 G9163 G9164 Spoken language expression G9165 G9166 G9167 Attention G9168 G9169 G9170 Memory G9171 G9172 G9173 Voice G9174 G9175 G9176 Other speech language pathology G-codes are a category of alphanumeric procedure codes assigned by CMS to be used for coding primarily non-physician services not covered by Current Procedural Terminology (CPT-4) codes. G-codes for reporting function are non-payable.
6
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Severity Modifiers FS Percent Impairment Severity Modifier 0% Impaired CH 1-19% Impaired CI 20-39% Impaired CJ 40-59% Impaired CK 60-79% Impaired CL 80-99% Impaired CM 100% Impaired CN Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
7
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Goals for This Talk Examine the internal validity of the G-code 7-level severity modifiers in relation to the FOTO FS scores. Examine the consequences of using the G-code FLR system for provider benchmarking/profiling Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
8
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Methods 2013—2014 FOTO Data FS in FOTO used as the “Gold Standard” N=2,010,733 patients 1,127,637 (56%) had a discharge FS record 883,096 (44%) had no discharge FS record Clinics = 2,454 in 2013; 2,641 in 2014 Benchmarking: excluded clinics with N < 30. Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
9
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Key Variables Minimal Clinically Important index based on published results (Lumbar: 5 FS points, Shoulder: 8, Knee: 9, Hip: 6, Foot/Ankle: 5) G-code SM: calculated from FOTO’s FS scale using the official cutoffs in the previous table (slide 6). Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
10
Validation Measures Using MCII of FOTO’s FS as Gold Standard
Sensitivity (True Positive) = #𝐺−𝑐𝑜𝑑𝑒 𝑆𝑀 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝑠 (𝐹𝑆↑) #𝑀𝐶𝐼 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑑 Specificity (True Negative) = #𝐺−𝑐𝑜𝑑𝑒 𝑆𝑀 𝑑𝑖𝑑 𝑛 ′ 𝑡 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒 #𝑀𝐶𝐼 𝑑𝑖𝑑 𝑛 ′ 𝑡 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒 Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
11
Percent Medicare Patients with an MCII or a G-code SM Increase
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
12
Sensitivity (True +ve) & Specificity (True –ve)
If you were paid $1 per significant increase, if sensitivity=specificity you break even (loose the false negatives above blue bar but get paid for the false positives above the red bar) Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
13
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Summary Sensitivity (cases where MCI increased but G-codes SM didn’t) is overall a larger problem. Lack of specificity is also a problem for quality measures based on FS improvement. Clinics with a large number of lumbar and hip patients particularly affected Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
14
Benchmarking with G-codes
Benchmarking (profiling) of providers has been used by Medicare for: Quality ratings (e.g. hospital star rating) Inform the design of value-based purchasing payment systems What is the effect on benchmarking of using categorical G-code FLR over using the continuous FS measure? Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
15
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Key Variables Outcome 1 = FOTO self-reported patient functional status (FS) at discharge. Outcome 2 = G-code Severity Modifier measure of the functional status (FS) at discharge. Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
16
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Covariates Patient case-mix FS/G-code SM at intake age gender number of days since onset of condition number of surgeries number of functional comorbidities payer type (not needed for Medicare-only) Pedro, these are all CAT data, so we do two CATs for the upper extremity: shoulder and elbow/wrist/hand. Did you separate elbow from wrist/hand? No need to. Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
17
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Analytic Strategy Hierarchical 3-level model Patients Therapists Clinics Corrected for censoring (missing discharge FS associated w/ patient characteristics that impact clinics differentially) using Inverse Probability of Censoring weights Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
18
Patient Characteristics (Lumbar, Medicare)
Mean (SD) or % Discharge FS (0-100) 59.1 (14.5) Intake FS (0-100) 46.9 (12.3) Age 71.5 (10.1) Female (%) 62 Surgery (more than 1) (%) 25 Onset (Days) (%) 1-21 DAYS 15 22-90 DAYS 22 91+ DAYS 63 Comorbidities (%) 1-2 28 3-4 72 Fear of Avoidance, High (%) 58 Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
19
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
20
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
21
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
22
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Results Summary Model based on categorical G-codes cannot discriminate better/worse clinics as well as FOTO’s continuous FS model Rankings are correlated, specially in the tails (be better/worse clinics) HOWEVER: Sensitivity among best ranked: % Sensitivity among worst ranked: % Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
23
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Results Summary (2) Correcting for censoring had little effect in ranking of most clinics (a few had larger changes). Clinic rankings can be non-uniform across conditions Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
24
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Limitations G-code SM were estimated using official thresholds. No warranty these are those actually reported by providers to CMS. However: a FOTO survey of therapists found that 70% respondents used FOTO’s scores at least 80% of the time in their claims to CMS. Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
25
Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Significance Profiling methods based on the categorical G-codes SM may be problematic for benchmarking outpatient rehab clinics due to poor sensitivity. G-codes SM profiling results in lower incentives for clinics to do well: doesn’t penalized almost 1/3 of lower performers, and doesn’t reward almost 1/3 of higher performers Gozalo—FOTO Outcomes Conference 2016, Knoxville TN
Similar presentations
© 2025 SlidePlayer.com Inc.
All rights reserved.