The newest section GG Quality Measure (QM), the Discharge Function Score, has left many nurse assessment coordinators (NACs) with more questions than answers about how it is calculated. This measure estimates the percentage of Medicare Part A skilled nursing facility (SNF) stays that meet or exceed an expected Discharge Function Score. The description may sound simple, but this complex measure uses statistical imputation for missing values and a linear regression model to determine the estimated discharge score. Because most NACs are not also statisticians, they will likely be unable to calculate the measure themselves. Nor would this be a prudent use of their time. This article simplifies these complex calculations, offering NACs an overall understanding of the Discharge Function Score measure to help them achieve better staff training and coding accuracy.
What does the Discharge Function Score report?
This measure reports Medicare Part A stays during a 12-month target period that have an observed discharge score that exceeded the expected discharge score. The following equation is provided in the Discharge Function Score for Skilled Nursing Facilities (SNFs) Technical Report (February 2023):
Which GG items are used?
The discharge score for each Medicare Part A stay is calculated using 10 GG items, each scoring 01 to 06. However, the same 10 GG items may not apply for every resident. For example, if the resident was able to walk 10 feet, “Walk 10 Feet” and “Walk 50 Feet with 2 Turns” are used, and the wheelchair item is not used. But if the resident was unable to walk 10 feet, the calculation instructions state to replace the two walking items by counting the “Wheel 50 feet with 2 Turns” score twice. The result is a Discharge Function Score of 10 to 60. The calculation is detailed in this AAPACN chart:
| GG Item | Instructions if Walk 10 Feet (GG0170I) has an activity not attempted (ANA) code at both admission and discharge | The remaining stays use Walk 10 Feet (GG0170I) + Walk 50 Feet with 2 Turns (GG0170J) to calculate the total observed Discharge Function Score |
| GG0130A3. Eating | X | X |
| GG0130B3. Oral hygiene | X | X |
| GG0130C3. Toileting hygiene | X | X |
| GG0170A3. Roll left and right | X | X |
| GG0170C3. Lying to sitting on side of bed | X | X |
| GG0170D3. Sit to stand | X | X |
| GG0170E3. Chair/bed-to-chair transfer | X | X |
| GG0170F3. Toilet transfer | X | X |
| GG0170I3: Walk 10 feet | Activity Not Attempted on 5-Day and Part A PPS Discharge | X |
| GG0170J3: Walk 50 feet with 2 turns | Skipped | X |
| GG0170R3. Wheel 50 feet with 2 turns* | X2 Count this score twice (if coded 01 – 06) | Not Used |
| Total Score Range | 10 – 60 | 10 – 60 |
How are GG items scored?
If the GG item is coded on the Part A PPS Discharge assessment with a level of assistance code 01 to 06, it’s the score assigned for that GG task. If the item is coded with an activity not attempted code, dashed (-), or skipped (^), statistical imputation is used to estimate the value of the missing item. Review this helpful AAPACN chart:
| Level of Assistance | Discharge Function Score |
| 06 – Independent | 06 |
| 05 – Setup or clean-up assistance | 05 |
| 04 – Supervision or touching assistance | 04 |
| 03 – Partial/moderate assistance | 03 |
| 02 – Substantial/maximal assistance | 02 |
| 01 – Dependent | 01 |
| 07 – Resident refused | Statistical imputation |
| 09 – Not applicable | Statistical imputation |
| 10 – Not attempted due to environmental limitations | Statistical imputation |
| ^ – Skip pattern | Statistical imputation |
| 88 – Not attempted due to medical condition or safety concerns | Statistical imputation |
| – – Not assessed/no information | Statistical imputation |
What is statistical imputation?
An approachable way to understand statistical imputation is to picture a jigsaw puzzle with a missing piece. The surrounding puzzle pieces provide clues about the appearance of the missing piece. Statistical imputation can be likened to using clues from the surrounding pieces to make an educated guess about what the missing piece looks like.
The surrounding puzzle pieces are comparable with the known facts about the resident (e.g., age, conditions, function on other GG tasks). They are used as covariates (clues) to estimate the missing GG value. Statistical imputation uses coefficient values to adjust the missing function score based on the known covariates. CMS provides the Imputation Appendix File to help SNFs review the covariates and coefficient values used in the calculation. For a more in-depth explanation of statistical imputation, the Skilled Nursing Facility Quality Reporting Program Measure Calculations and Reporting User’s Manual Version 5.0 details the specifics of the calculation.
How is the expected Discharge Function Score calculated?
The score is calculated by applying the regression equation determined from risk adjustments to each SNF stay. Essentially, this means it estimates how different factors (covariates) impact a resident’s ability to function at the time of discharge from a SNF. These covariates include age, admission function score (including any statistical imputation), primary medical category, prior function and device use, pressure ulcers, cognitive function, communication impairment, incontinence, nutritional status, history of falls, and Hierarchical Condition Categories comorbidities. In more technical terms, the expected Discharge Function Score calculation is detailed in the following excerpt from the Discharge Function Score for Skilled Nursing Facilities (SNFs) Technical Report (February 2023).
| The Statistical Risk Model The statistical risk model is an ordinary least squares linear regression model, which estimates the relationship between Discharge Function Score and a set of risk adjustors. Observed Discharge Function Score is determined for each SNF stay, incorporating imputed item scores when NA [not applicable] codes are encountered. The risk adjustment model is run on all SNF stays to determine the model intercept (β0) and risk adjustor coefficients (β1, …, βn). Expected Discharge Function Scores are calculated by applying the regression equation to each SNF stay.  where x1 – x are the risk adjustors. |
What is the numerator?
The numerator is the total number of Medicare Part A SNF stays in the denominator, except those that meet the exclusion criteria, with an observed Discharge Function Score that is equal to or greater than the calculated expected score. This measure uses only Type 1 SNF stays, those stays that have a paired 5-Day PPS assessment with a Part A PPS discharge. If the stay ended with a death, it is considered a Type 2 SNF stay and not included in this measure.
What is the denominator?
The denominator is the total number of Medicare Part A SNF stays (Type 1 SNF stays only), except those that meet the exclusion criteria.
What are the exclusions?
Medicare Part A SNF stays are excluded from this measure in these situations:
- The Medicare Part A SNF stay is an incomplete stay. Examples of incomplete stays are an unplanned discharge, discharge to an acute care hospital, psychiatric hospital, or a long-term care hospital, a Medicare Part A stay less than 3 days, or if the resident died during the skilled stay.
- The resident has any of these medical conditions on the 5-Day PPS assessment: coma, persistent vegetative state, complete tetraplegia, severe brain damage, locked-in syndrome, severe anoxic brain damage, cerebral edema, or compression of brain.
- The resident is younger than age 18.
- The resident is discharged to hospice or received hospice while a resident.
How NACs can use this information
NACs may still have questions and not fully understand the complexities of the Discharge Function Score calculation. However, even a limited grasp of how statistical imputation is used to calculate the missing data and how the linear regression model estimates the expected scores will help NACs identify what they can control. NACs can focus on staff education regarding section GG data that may reduce the use of dashes. They can also observe the resident and interview those involved in the resident’s care to ensure tasks that occurred during the 3-day window are assigned a level-of-care code and not coded incorrectly as “not attempted” due to lack of documentation. NACs can also become more familiar with the covariates and exclusions to ensure these items are coded accurately on the Minimum Data Set (MDS). They may also still want to figure out how the expected discharge score is determined to help align functional goals, but until a statistician is onboard, the focus should remain on helping each resident achieve his or her highest practicable level of functioning.
This AAPACN resource is copyright protected. AAPACN individual members may download or print one copy for use within their facility only. AAPACN facility organizational members have unlimited use only within facilities included in their organizational membership. Violation of AAPACN copyright may result in membership termination and loss of all AAPACN certification credentials. Learn more.