Overground and treadmill walking with 30 second bursts at maximum speed alternated with 30-60 second passive recovery periods. Intended to achieve a target average heart rate above 60% heart rate reserve.

From 10-meter walk test, in meters per second

From 10-meter walk test, in meters per second

From 10-meter walk test, in meters per second

From 10-meter walk test, in meters per second

Oxygen consumption rate at ventilatory threshold during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Oxygen consumption rate at ventilatory threshold during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Oxygen consumption rate at ventilatory threshold during treadmill graded exercise test, in milliliters per kilogram body mass per minute

The Patient Reported Outcomes Measurement Information System (PROMIS) Fatigue Scale is an 8-item self-report questionnaire about symptoms of fatigue. Each item is rated from 1-5, where higher scores indicate greater fatigue. A total raw score is calculated by summing the scores for each item. The raw score is then translated into a T-score based on normative data. The T-score is a standardized score with a mean of 50 and a standard deviation of 10 for the U.S. general population, so a T-score of 60 (for example) would represent one SD higher (worse) fatigue than the average American.

The Patient Reported Outcomes Measurement Information System (PROMIS) Fatigue Scale is an 8-item self-report questionnaire about symptoms of fatigue. Each item is rated from 1-5, where higher scores indicate greater fatigue. A total raw score is calculated by summing the scores for each item. The raw score is then translated into a T-score based on normative data. The T-score is a standardized score with a mean of 50 and a standard deviation of 10 for the U.S. general population, so a T-score of 60 (for example) would represent one SD higher (worse) fatigue than the average American.

The Patient Reported Outcomes Measurement Information System (PROMIS) Fatigue Scale is an 8-item self-report questionnaire about symptoms of fatigue. Each item is rated from 1-5, where higher scores indicate greater fatigue. A total raw score is calculated by summing the scores for each item. The raw score is then translated into a T-score based on normative data. The T-score is a standardized score with a mean of 50 and a standard deviation of 10 for the U.S. general population, so a T-score of 60 (for example) would represent one SD higher (worse) fatigue than the average American.

Time to exhaustion during treadmill graded exercise test, recorded in minutes

Peak oxygen consumption rate during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Peak oxygen consumption rate during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Peak oxygen consumption rate during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Oxygen consumption rate relative to speed during treadmill graded exercise test, in milliliters per kilogram body mass per meter

Oxygen consumption rate relative to speed during treadmill graded exercise test, in milliliters per kilogram body mass per meter

Oxygen consumption rate relative to speed during treadmill graded exercise test, in milliliters per kilogram body mass per meter

A 5-item questionnaire about quality of life, including: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each item is scored from 1 to 5, where lower scores indicate greater quality of life. A total misery score was calculated by summing the scores for each item (min 5, max 25).

A 5-item questionnaire about quality of life, including: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each item is scored from 1 to 5, where lower scores indicate greater quality of life. A total misery score was calculated by summing the scores for each item (min 5, max 25).

A 5-item questionnaire about quality of life, including: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each item is scored from 1 to 5, where lower scores indicate greater quality of life. A total misery score was calculated by summing the scores for each item (min 5, max 25).

A 16-item questionnaire that asks participants to rate their balance confidence during everyday tasks, as a percentage from 0-100%, where higher scores indicate greater balance self-efficacy. A total score from 0-100 is calculated by averaging the scores for each item.

A 16-item questionnaire that asks participants to rate their balance confidence during everyday tasks, as a percentage from 0-100%, where higher scores indicate greater balance self-efficacy. A total score from 0-100 is calculated by averaging the scores for each item.

A 16-item questionnaire that asks participants to rate their balance confidence during everyday tasks, as a percentage from 0-100%, where higher scores indicate greater balance self-efficacy. A total score from 0-100 is calculated by averaging the scores for each item.

Mean daily step count recorded with activity monitor, recorded as steps per day

Mean daily step count recorded with activity monitor, recorded as steps per day

Mean daily step count recorded with activity monitor, recorded as steps per day

Total of all reporting groups

Time interval from stroke onset to consent date

Total distance walked in 6 minutes in meters

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Total distance walked in 6 minutes in meters

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

From 10-meter walk test, in meters per second

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

From 10-meter walk test, in meters per second

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

From 10-meter walk test, in meters per second

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

From 10-meter walk test, in meters per second

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

From 10-meter walk test, in meters per second

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

From 10-meter walk test, in meters per second

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Oxygen consumption rate at ventilatory threshold during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Oxygen consumption rate at ventilatory threshold during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Oxygen consumption rate at ventilatory threshold during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

The Patient Reported Outcomes Measurement Information System (PROMIS) Fatigue Scale is an 8-item self-report questionnaire about symptoms of fatigue. Each item is rated from 1-5, where higher scores indicate greater fatigue. A total raw score is calculated by summing the scores for each item. The raw score is then translated into a T-score based on normative data. The T-score is a standardized score with a mean of 50 and a standard deviation of 10 for the U.S. general population, so a T-score of 60 (for example) would represent one SD higher (worse) fatigue than the average American.

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

The Patient Reported Outcomes Measurement Information System (PROMIS) Fatigue Scale is an 8-item self-report questionnaire about symptoms of fatigue. Each item is rated from 1-5, where higher scores indicate greater fatigue. A total raw score is calculated by summing the scores for each item. The raw score is then translated into a T-score based on normative data. The T-score is a standardized score with a mean of 50 and a standard deviation of 10 for the U.S. general population, so a T-score of 60 (for example) would represent one SD higher (worse) fatigue than the average American.

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

The Patient Reported Outcomes Measurement Information System (PROMIS) Fatigue Scale is an 8-item self-report questionnaire about symptoms of fatigue. Each item is rated from 1-5, where higher scores indicate greater fatigue. A total raw score is calculated by summing the scores for each item. The raw score is then translated into a T-score based on normative data. The T-score is a standardized score with a mean of 50 and a standard deviation of 10 for the U.S. general population, so a T-score of 60 (for example) would represent one SD higher (worse) fatigue than the average American.

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Time to exhaustion during treadmill graded exercise test, recorded in minutes

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Time to exhaustion during treadmill graded exercise test, recorded in minutes

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Time to exhaustion during treadmill graded exercise test, recorded in minutes

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Peak oxygen consumption rate during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Peak oxygen consumption rate during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Peak oxygen consumption rate during treadmill graded exercise test, in milliliters per kilogram body mass per minute

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Oxygen consumption rate relative to speed during treadmill graded exercise test, in milliliters per kilogram body mass per meter

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Oxygen consumption rate relative to speed during treadmill graded exercise test, in milliliters per kilogram body mass per meter

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Oxygen consumption rate relative to speed during treadmill graded exercise test, in milliliters per kilogram body mass per meter

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

A 5-item questionnaire about quality of life, including: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each item is scored from 1 to 5, where lower scores indicate greater quality of life. A total misery score was calculated by summing the scores for each item (min 5, max 25).

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

A 5-item questionnaire about quality of life, including: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each item is scored from 1 to 5, where lower scores indicate greater quality of life. A total misery score was calculated by summing the scores for each item (min 5, max 25).

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

A 5-item questionnaire about quality of life, including: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each item is scored from 1 to 5, where lower scores indicate greater quality of life. A total misery score was calculated by summing the scores for each item (min 5, max 25).

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

A 16-item questionnaire that asks participants to rate their balance confidence during everyday tasks, as a percentage from 0-100%, where higher scores indicate greater balance self-efficacy. A total score from 0-100 is calculated by averaging the scores for each item.

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

A 16-item questionnaire that asks participants to rate their balance confidence during everyday tasks, as a percentage from 0-100%, where higher scores indicate greater balance self-efficacy. A total score from 0-100 is calculated by averaging the scores for each item.

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

A 16-item questionnaire that asks participants to rate their balance confidence during everyday tasks, as a percentage from 0-100%, where higher scores indicate greater balance self-efficacy. A total score from 0-100 is calculated by averaging the scores for each item.

Results are from an intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Missing data were handled with the maximum likelihood method.

Mean daily step count recorded with activity monitor, recorded as steps per day

Results are from a modified intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Four participants (2 in the HIT group and 2 in the MAT group) had to be excluded because they had no valid days of step data collected outside of study visits. Other missing data were handled with the maximum likelihood method.

Mean daily step count recorded with activity monitor, recorded as steps per day

Results are from a modified intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Four participants (2 in the HIT group and 2 in the MAT group) had to be excluded because they had no valid days of step data collected outside of study visits. Other missing data were handled with the maximum likelihood method.

Mean daily step count recorded with activity monitor, recorded as steps per day

Results are from a modified intent-to-treat analysis using a linear mixed effects model that included all available time points (Baseline, 4-week, 8-week and 12-week) for all randomized participants and estimated contrasts for change from baseline to each time point. Four participants (2 in the HIT group and 2 in the MAT group) had to be excluded because they had no valid days of step data collected outside of study visits. Other missing data were handled with the maximum likelihood method.

Moderate-Intensity Aerobic Training: Overground and treadmill walking with speed continuously adjusted to maintain a target heart rate of 40 +/- 5% heart rate reserve, progressing up to 55 +/- 5% heart rate reserve.

fall with hip fracture, judged to be unrelated to study procedures

delirium leading to temporary hospitalization, judged to be unrelated to study procedures

recurrent stroke, judged to be unrelated to study procedures

Fatigue related to treatment

lightheadedness related to treatment

Fall without injury related to treatment