Ability to walk independently worsened: the N035 measure

The walking measure counts long-stay residents whose ability to walk 10 feet got worse between two assessments. It reads a single MDS item, GG0170I, and it is one of the long-stay measures in the Five-Star quality rating, worth up to 150 points.

The CMS name is “Percent of Residents Whose Ability to Walk Independently Worsened (Long Stay),” CMS ID N035.06. Some CMS lists call it “ability to move independently worsened”; it is the same measure. It is risk-adjusted. It also counts in the Ohio QIP and in Component 1 of the Texas QIPP.

Which item, and which version

The measure uses GG0170I, Walk 10 feet. Before October 1, 2023 it was built from Section G. When MDS item set v1.18.11 took effect that day, Section G’s functional items were replaced by Section GG, and the QM User’s Manual v16.0 re-specified the measure on GG0170I. CMS froze it on Care Compare from April 2024 and brought it back, with new Five-Star cut points, in January 2025.

The current QM User’s Manual, v18.0 (effective January 1, 2026), numbers it N035.06. Its written specification differs from v17’s N035.05 in one place: the sex covariate is now read from A0810, the item that replaced A0800 (Gender) when MDS item set v1.20.1 took effect on October 1, 2025. The numerator and exclusions are unchanged.

The column comes from the kind of assessment: GG0170I3 on a discharge, GG0170I5 on a quarterly, annual or other OBRA assessment that isn’t combined with a PPS assessment, and GG0170I1 on an OBRA admission or PPS 5-Day.

What puts a resident in the numerator

Codes of 07, 09, 10 and 88 (refused, not applicable, environmental limitation, not attempted for medical or safety reasons) are first recoded to 01, Dependent. Then:

One point is enough. A resident who walked 10 feet with supervision (04) last quarter and needs partial/moderate help (03) this quarter is in the numerator. So is a resident who walked with supervision last quarter and is coded 88 this quarter, because 88 counts as 01.

Who is left out

The exclusions look mostly at the prior assessment:

• Comatose on the prior (B0100), or that item dashed.
• End of life on the prior: a prognosis of less than six months (J1400) or hospice (O0110K1b), or either item dashed when neither is coded yes.
• Already unable to walk on the prior. The resident can’t get worse than dependent:

• GG0170I dashed on either assessment.
• The prior is a discharge assessment (A0310F = 10 or 11).
• No prior assessment 46 to 165 days before the target.
• The target is an OBRA admission, a PPS 5-Day, or the first assessment after an admission (A0310E = 1). A0310E counts reentries too, so the first assessment after a resident comes back from the hospital can’t be this measure’s target. And if the hospital transfer produced an OBRA discharge assessment, that discharge can end up as the next target’s prior, which also excludes the resident.
• Either assessment dated before October 1, 2023.

Several of those differ from the ADL decline measure, which reads the same kind of target-and-prior pair. The ADL measure doesn’t exclude an admission, 5-Day or first-after-admission target, or a discharge prior, and it takes its coma and end-of-life exclusions from the target, not the prior.

When it clears

The target is the resident’s latest qualifying assessment dated within 120 days of the end of the quarter (or of their discharge). The prior is the latest qualifying assessment 46 to 165 days before the target, usually the quarterly before.

So each quarter is a new comparison. Next quarter’s target is compared with the assessment that showed the decline. If the resident walks the same as they did then, nothing dropped and they don’t trigger again. A real decline normally counts in one quarter, which then sits in Five-Star’s four-quarter average until it rolls out.

Risk adjustment, in plain words

CMS expects some residents to lose walking ability more often than others, so this measure is adjusted. Each resident gets a predicted chance of declining, mostly from the prior assessment: how much help they needed with eating, toilet transfer, sit to stand and walking; severe cognitive impairment; age; and sex. Two more look at what changed after the prior: vision getting worse, and newly starting oxygen. Your building’s rate is then adjusted against those predictions:

That makes the prior assessment’s GG coding matter twice. It sets the baseline the target is compared with, and it sets how much decline CMS expects.

Coding it accurately

Code the resident’s usual performance over the assessment period (the ARD and the two days before it on OBRA assessments after admission), not their best or worst moment. The walking items have their own rules:

• A helper can’t walk for the resident. The resident has to take part for the whole 10 feet.
• Parallel bars don’t count. Don’t code walking with a device used only in therapy sessions, such as parallel bars, an exoskeleton or an overhead harness.
• Two helpers means 01, Dependent.
• Setup is 05. If the only help is bringing the walker and putting it away, the code is 05, Setup or clean-up assistance.
• Wheelchair users are still assessed. GG walking is scored on ability, whether or not the resident usually walks. Use 09 only when the resident didn’t walk before the current illness and didn’t walk during the assessment period.

The trap to watch is inconsistency. If one quarter’s 04 came from a therapist’s note and the next quarter’s 03 came from one night shift, the resident may not have changed at all. The measure can’t tell the difference.

How Five-Star scores it

Walking decline is one of the 150-point measures, scored in deciles of the national distribution. Lower is better. Points come from your risk-adjusted four-quarter rate, using cut points set from 2023Q4–2024Q3 data:

SuperQM compares every long-stay resident’s walking code with the prior CMS will use, and flags the ones the measure will count. See how it works.