Step Counts Hold Up in Parkinson's, Less Known for Stroke
Wrist trackers came within 3% of a research monitor in 33 people with Parkinson's. Energy estimates were weaker, and stroke is largely unchecked.
Evidence rating for the proposition “Consumer wearable devices measure heart rate with accuracy comparable to reference standards”: Established · KnowYourPrime Evidence Registry, Finding KYP-ER-0009
It depends on the condition and the measure. Step counts held up in Parkinson's disease. The Fitbit Charge HR and Garmin vivosmart HR had under 3% error against a research accelerometer in 33 people walking at their own pace. Fitbit One steps matched an ActiGraph closely in 24 people with myositis (ICC 0.96). Energy expenditure did worse. For stroke and traumatic brain injury, a review of 20 articles found accuracy was not reported in many of them.
Are step counts accurate in Parkinson's disease?
Yes, at a self-selected pace. Thirty-three people with mild to moderate Parkinson's wore a Fitbit Charge HR and a Garmin vivosmart HR. They completed six two-minute indoor walks and a 500 m outdoor walk with terrain challenges. An ActivPAL3 accelerometer was the reference.
Both trackers had error under 3% and moderate to high consistency (ICC 0.88 to 0.97) at self-selected pace, indoors and outdoors. At target cadences of 80 steps a minute or more, the Garmin stayed under 5% error. The Fitbit had higher error there.
A separate study of 100 people with Parkinson's wore a Garmin Vivosmart 4 for five days. They averaged 5,949 daily steps (SD 3,034). A change of about 581 steps a day matched changes in motor symptoms and mobility, roughly 10% of the average.
2 studies
- In 33 people with mild to moderate Parkinson's, the Fitbit Charge HR and Garmin vivosmart HR had step count error under 3% and ICC of 0.88 to 0.97 at self-selected pace indoors and outdoors against an ActivPAL3. The Garmin had under 5% error at target cadences of 80 or more, and the Fitbit had higher error.
- In 100 people with Parkinson's wearing a Garmin Vivosmart 4 for five days, average daily steps were 5,949 (SD 3,034), and the change anchored to motor symptom and mobility measures was 581 steps a day (range 554 to 608).
What about myositis?
Also good, in a small study. Twenty-four people with inflammatory myopathies wore a waist-worn Fitbit One and an ActiGraph for 7 days a month over six months. Fitbit and ActiGraph step counts agreed closely, with an ICC of 0.96. Test-retest reliability over one month was strong (ICC 0.89).
Daily steps also tracked physician and patient global disease activity and functional tests. The authors call it a pilot and say larger studies are needed.
Does energy expenditure hold up too?
No. In Huntington's disease, a study of 14 people with mild to moderate disease compared the Fitbit Charge 4 and an ActiGraph with a metabolic cart. Both were tested while walking on a treadmill at 3.2 and 5.2 km/h.
The Fitbit correlated with the cart (r = 0.701) but its reliability was poor at every speed. It significantly overestimated energy expenditure. The ActiGraph underestimated it but was more accurate than the Fitbit in every test.
In Parkinson's, cadence from the Fitbit and Garmin weakly reflected heart rate (ICC 0.27 to 0.28). It did not reflect oxygen consumption or perceived exertion (ICC 0.08 to 0.15). Steps held up. Effort did not. See whether the calorie number on a wearable means anything.
Does a walker or cane change step counts?
It can, depending on where the tracker sits. Eleven healthy people walked an obstacle course. Fitbits at the wrist, hip and ankle correlated well with true steps during a cane trial and a deviceless trial (r = 0.764 to 0.984). With a two-wheeled walker, hip and ankle placement still correlated (r = 0.669 to 0.888) with an average error of 1.5%. The wrist did not correlate, with an average error of 31.2%.
A treadmill study of 12 devices on healthy adults simulating slow and shuffling walking found the Apple Watch worn at the waist most reliable. Leg placement made up 10 of the top 20 device and position combinations. The authors suggest larger movement at the leg helps step counts during altered gaits.
Both studies used healthy people simulating altered movement. For more, see whether a walker or cane affects step accuracy and how a watch counts steps.
2 studies
- In 11 healthy participants, Fitbit step counts during a two-wheeled walker trial correlated with true steps at the ankle and hip (r = 0.669 to 0.888, average error 1.5%) but not at the wrist (average error 31.2%).
- In healthy adults simulating slow and shuffling walking on a treadmill, the Apple Watch was the most reliable of 12 devices, and leg placement accounted for 10 of the top 20 device and position combinations.
What about stroke and traumatic brain injury?
Mostly unmapped. A systematic review of 20 articles looked at wearables in the community after stroke or traumatic brain injury. For stroke, the move from lab to community had started but was not widespread. For traumatic brain injury it had not yet started.
Accelerometer-based devices were used more often than pedometers. No preferred device or wear location emerged. Step count was the most common measure. The accuracy and validity of most outcome metrics in the community were not reported in many studies.
That is a gap, not a failure. It means the evidence to answer the question directly does not exist at scale for those two conditions.
What does this not settle?
Results do not transfer across conditions. Parkinson's step counts were tested in people who could walk unaided. The Huntington's study had 14 people. The myositis study had 24. The walker findings are from healthy volunteers.
No study here tested a consumer tracker in people using a walker with a neurological condition.
The Parkinson's results describe people with mild to moderate disease walking unaided. The myositis, Huntington's and walker studies enrolled 24, 14 and 11 people. Nothing here validates a consumer tracker in stroke, brain injury or people using a walker with a neurological condition.
Common questions
Are fitness trackers accurate for people with Parkinson's disease?
For steps at a normal pace, yes in one study. In 33 people, the Fitbit Charge HR and Garmin vivosmart HR had step error under 3% against an accelerometer. Neither tracker strongly reflected exercise intensity.
Do trackers estimate calories burned accurately in movement disorders?
Not in Huntington's disease. In 14 people, the Fitbit Charge 4 overestimated energy expenditure and its reliability was poor at every treadmill speed.
Does using a walker throw off a tracker?
At the wrist, it did in healthy volunteers. With a two-wheeled walker, wrist step error averaged 31.2% against 1.5% at the hip and ankle.
Have trackers been validated after a stroke?
Largely not reported. A review of 20 articles found accuracy and validity of most community outcome metrics were not reported in many studies of stroke or traumatic brain injury.
Are step counts accurate in myositis?
In a pilot of 24 people, Fitbit One step counts agreed closely with an ActiGraph (ICC 0.96), and one-month test-retest reliability was strong (ICC 0.89).
Sources
- Accuracy of wearable physical activity trackers in people with Parkinson's disease.
- Minimal Clinically Important Difference of Average Daily Steps Measured Through a Consumer Smartwatch in People With Parkinson Disease.
- Consumer-based activity trackers in evaluation of physical activity in myositis patients.
- Validation of ActiGraph and Fitbit in the assessment of energy expenditure in Huntington's disease.
- The Effect of Assistive Devices on the Accuracy of Fitbits in Healthy Individuals: A Brief Report.
- Accurate measurement of simulated slow and altered walking activity: Apple Watch best in class.
- Wearable devices for tracking physical activity in the community after an acquired brain injury: A systematic review.