Abstract
Background
Physical activity guidelines recommend accumulating moderate‐to‐vigorous physical activity but interpreting and monitoring these recommendations remains challenging. Although step‐based metrics from wearable devices offer a simpler approach, their relationship with health outcomes requires validation against established accelerometer measurements.
Objectives
To evaluate how effectively step‐based metrics capture health‐related information from accelerometer data and determine optimal step cadence and intensity thresholds associated with cardiometabolic health in middle‐aged adults.
Methods
Cross‐sectional data from 4172 participants (aged 50–64 years) in the Swedish CArdioPulmonary bioImage Study (SCAPIS) were analyzed. Physical activity was measured using ActiGraph accelerometers, collecting both step metrics (daily steps and cadence) and full accelerometer data. Both cardiorespiratory fitness, estimated using a submaximal cycle ergometer test, and cardiometabolic health, assessed using a composite score of waist circumference, blood pressure, lipids, and glycated hemoglobin (HbA1c), were considered outcomes. Associations between physical activity and outcomes were examined using linear regression and partial least squares analysis.
Results
Step counting metrics retained 88% of the health‐related information from full accelerometer data. The optimal accelerometer intensity associated with cardiometabolic health was around four metabolic equivalents of tasks (METs). A step cadence of 80 steps/min, rather than the commonly used 100 steps/min, appeared more relevant for capturing moderate‐intensity activity. Combining step and accelerometer data provided additional explanatory power for cardiometabolic health.
Conclusion
Step data capture most of the health‐related information from accelerometer‐measured physical activity in middle‐aged adults. These findings support the use of step‐based metrics for assessing and promoting physical activity while suggesting a need for recalibration of intensity thresholds in free‐living conditions.
Keywords: accelerometry, cardiometabolic risk factors, cardiorespiratory fitness, cardiovascular disease, pedometry, physical activity, public health
Abbreviations
- CS
cardiometabolic health composite score
- FEM
4 Hz frequency–extended method
- HbA1c
glycated hemoglobin
- HDL
high‐density lipoprotein
- MET
metabolic equivalents of task
- MVPA
moderate‐to‐vigorous physical activity
- SBP
systolic blood pressure
- SCAPIS
Swedish CArdioPulmonary bioImage study
- TC
total cholesterol
Introduction
Physical inactivity is considered one of the leading causes of disease and mortality worldwide [1]. Physical activity guidelines recommend adults to accumulate at least 150–300 min/week of physical activity of at least moderate intensity [2, 3]. However, there are several challenges with this recommendation. First, it is mainly based on subjective methods affected by differential reporting biases depending on the group investigated. Second, due to difficulties for laymen to interpret this recommendation and put it into practice, researchers search for other ways to facilitate its communication and to monitor compliance. The third challenge is related to the definition of moderate intensity, as both absolute (i.e., metabolic equivalents of tasks [METs]) and relative (i.e., % of maximal capacity) threshold have been applied, and that an absolute threshold of three METs is assumed since mid‐1990 [4] and applied in the WHO guidelines 2020 [3], with questionable evidence on the importance of this threshold (see below).
Device‐based objective measurement of physical activity is considered more accurate than self‐report [5]. Accelerometers quantify movement and are commonly used devices for physical activity measurement. Physical activity expressed as steps has been specifically targeted as a publicly attractive measure [2], which can be generated from accelerometer data. In addition, step cadence (i.e., steps per minute) has emerged as a valid proxy metric of ambulatory intensity [6]. On the other hand, by using the full accelerometer data, a direct measure of ambulatory intensity can be achieved, for example, the metrics counts and milligravity (mg), but these metrics are more difficult to interpret. Consequently, neither steps nor full accelerometer data provide ideal metrics for communication and monitoring of physical activity. It would, therefore, be important to determine whether step cadence still provides sufficient information of the full accelerometer data to further argue for its utility.
Studies conducted under lab conditions have explored step cadence at different movement speeds to calibrate for intensity categories, specifically moderate intensity [7, 8, 9, 10, 11]. A common reference for the lower absolute threshold of the moderate intensity interval is three METs, where one MET represents the energy expended at rest, which has been used since mid‐1990 [4] and applied in the WHO guidelines 2020 [3]. When 3 METs were used as reference threshold of absolute moderate intensity, a common step cadence of 100 steps/min was determined in individuals 21–85 years old. However, three METs corresponded to a walking speed of 3–4 km/h [9, 10, 11], which has been defined as slow walking pace [12]. An absolute intensity threshold has different clinical significance depending on age and cardiorespiratory fitness level. Consequently, step cadence thresholds have also been investigated when moderate intensity was defined in relation to fitness level, that is, relative intensity, applying ≥64% heart rate maximum, ≥40% heart rate reserve, and ≥12 Borg Scale rating of Perceived Exertion as references [7, 8]. The step cadence of moderate intensity was now defined in a range of 105–120 steps/min (lower values for older individuals), corresponding to a walking speed range of 4–5.6 km/h (lower values for older individuals) and the definition of moderate walking pace [12]. The threshold of three METs has been criticized to be set too low for most individuals, with the consequence of overestimating the proportion of the population that comply to the physical activity recommendation. A higher threshold of four, five METs at an intensity promoting fitness has been suggested, with stronger association with cardiometabolic health [13, 14]. This would correspond better to the step cadence thresholds for relative intensity presented above [7, 8].
Intensity thresholds are optimally defined from the association pattern with markers of cardiometabolic health. At least one large‐scale study under free‐living conditions provides thresholds for daily steps and step cadence in relation to cardiometabolic risk [15]. Regarding thresholds for daily steps (volume), they reported, depending on the marker at hand, 4325–6192 steps/day. Regarding step cadence, the prediction capacity of time spent in eight cadence bands was investigated using a decision tree analysis method. However, only three cadence bands were able to predict cardiometabolic risk, representing incidental (1–19 steps/min), medium (80–99 steps/min) and faster (≥120 steps/min) movement. This outcome is likely explained by multicollinearity, where the cadence bands cannot be considered independent from each other. A statistical method that allows many intensity categories or cadence bands to be analyzed simultaneously is partial least square regression [14, 16].
The aim of this study was to evaluate how effectively step‐based metrics capture health‐related information from accelerometer data and to identify optimal intensity thresholds for both step data and accelerometer data by examining their associations with cardiorespiratory fitness and a composite score of cardiometabolic health under free‐living conditions in middle‐aged adults. We hypothesized that step data would retain most of the information from accelerometer data and its association with cardiometabolic health and that the strength of these associations would increase considerably above the traditional three MET threshold for moderate intensity.
Methods
Study sample
This study performed statistical analyses on cross‐sectional data from the Swedish CArdioPulmonary bioImage Study (SCAPIS) baseline measurement [17]. SCAPIS is a population‐based study designed to improve risk prediction of cardiovascular disease and chronic obstructive pulmonary disease through extensive examinations, including imaging, functional tests, and biomaterial collection. The SCAPIS include 30,154 randomly selected men and women aged 50–64 years from 6 regions in Sweden. Baseline recruitment and data collection were performed in 2013–2018. SCAPIS was approved by the Ethical Review Board in Umeå (2021‐228‐31 M) and the present study by the Regional ethical board in Gothenburg (638‐16). All participants provided written informed consent. This study investigated a subsample from the study center in Gothenburg where submaximal cardiorespiratory fitness tests were performed in total 6266 individuals. Valid measurements of fitness, physical activity, and all required markers of cardiometabolic health were available from 4172 participants, which were the study sample included in the analyses. The fitness test had several exclusion criteria, including active infections, unstable cardiovascular disease, signs of cardiac disease detected through electrocardiography, use of beta‐blockers, body weight over 125 kg, or a resting heart rate exceeding 100 beats per minute. Due to these exclusion criteria, the investigated sample represents a relatively healthy subset of middle‐aged adults, particularly suitable for studying subclinical markers of cardiometabolic health.
Accelerometry
SCAPIS participants were instructed to wear the ActiGraph GT3X+, wGT3X+, or wGT3X‐BT accelerometer (The ActiGraph), a commonly used research‐grade activity monitor, in an elastic belt over the right hip for seven consecutive days and to take it off during sleep and water activities. Data were collected at a sampling frequency of 30 Hz and acceleration amplitude of ±6 g.
Accelerometer data were processed using three different methods, and three different metrics were extracted from each processing method (Table 1). The processing methods were step counting, ActiGraph counts, and the 4 Hz frequency extended method (FEM). ActiGraph counts are the most used method for processing accelerometer data in physical activity research. FEM is based on the ActiGraph method but improved to more accurately capture physical activity intensity [18].
Table 1.
Physical activity metrics derived from accelerometer data using three processing methods, showing measures of daily volume, time spent at moderate‐to‐vigorous intensity (MVPA), and intensity spectrum analysis.
| Method | Volume | MVPA | Spectrum |
|---|---|---|---|
| Steps | Steps/day | ≥80, ≥100/min | 17 bands, 0–>192/min |
| FEM | Mean mg | ≥3, ≥4, ≥4.5 METs | 27 bins, 0–>1000 mg |
| ActiGraph vertical | Counts per day | ≥3 METs | 22 bins, 0–>10,000 cpm |
| ActiGraph VM | Counts per day | ≥3 METs | 22 bins, 0–>20,000 cpm |
Abbreviations: cpm, counts per minute; FEM, frequency‐extended method; METs, metabolic equivalents of tasks; mg, milli‐gravity; MVPA, moderate‐to‐vigorous physical activity; VM, vector magnitude.
The three different metrics were total volume, time at moderate‐to‐vigorous intensity (MVPA), and full intensity spectrum. Total volume was represented by the average number of steps per day or aggregated acceleration over the measurement period. Time at MVPA was represented by the average time per day where either the step frequency (cadence) was above a certain threshold (e.g., 100 steps/min) or absolute acceleration intensity was above a certain threshold (e.g., 2690 counts/min or 167 mg).
For the full spectrum analyses, step cadence and accelerometer intensity data were divided into multiple intensity ranges. The time spent at each intensity level was then analyzed. For example, step cadence was divided into bands from 0 to over 192 steps/min in increments of 12 steps/min, with the time spent in each band being analyzed.
The step metrics daily steps and cadence of a specific epoch (steps/min) as well as ActiGraph counts were generated using the ActiGraph software ActiLife (version 6.13.6) [19], and the full accelerometer intensity metrics mg using FEM in MATLAB R2023a (MathWorks) [18]. It has previously been shown that the daily step volume generated by the ActiLife software is comparable to the Yamax pedometer (Yamax SW‐200) [19], which is considered a gold standard research pedometer when volume (i.e., daily steps) are at hand [20]. Non‐wear time was defined as at least 60 min of zero output, with the allowance of up to 2 min of output below the sedentary threshold [21]. A valid measurement was considered at least 4 days of at least 10 h of wear time [21]. Both step cadence and accelerometer metrics used an epoch length of 5 s.
Due to the 5‐s epoch length, the cadence resolution possible to capture was 12 steps/min. Step cadence was divided into 17 small cadence bands with bin edges: 0, 12, 24 steps/min, and then increasing in 12 steps/min intervals to 192 steps/min and above. The FEM mg range was divided into a spectrum of 27 intensity bins with bin edges of 0, 25, 50, 100, 150 mg, and then increasing in 50 mg intervals to 1000 mg and above. ActiGraph counts output, either from the vertical axis or vector magnitude, was divided into a spectrum of 22. Bin edges of 0, 100, 500, 1000, 1500, …, >10,000 counts/min and 0, 200, 1000, 2000, 3000, …, >20,000 counts/min were used for vertical and vector magnitude ActiGraph data, respectively.
Furthermore, time spent above traditional intensity thresholds was analyzed for the step, FEM, and ActiGraph measures. These thresholds were derived from laboratory studies that measured energy consumption (METs) using indirect calorimetry during treadmill walking and running. One MET was considered to be 3.5 mL/min/kg [2]. For the step output, the intensity thresholds were 100 and 130 steps/min for moderate (three METs) and vigorous (six METs) intensity, respectively [9, 10, 11]. Other studies derived similar intensity thresholds for the FEM [18] and ActiGraph [22, 23] outputs. For all MVPA measures, we calculated time spent at or above the moderate intensity threshold, which included vigorous intensity activity.
Cardiometabolic health
Cardiometabolic health was represented by a composite score (CS) consisting of measurements of waist circumference, systolic blood pressure (SBP), total cholesterol to high‐density lipoprotein ratio (TC:HDL), triglycerides, and glycated hemoglobin (HbA1c). These measurements are established risk factors for cardiovascular and metabolic diseases, representing central obesity, hypertension, dyslipidemia, and hyperglycemia [24, 25]. We used a continuous composite score approach as this methodology is particularly suitable for detecting subclinical variations in cardiometabolic risk in relatively healthy populations [26].
Trained health professionals performed all measurements following standardized protocols [17]. Waist circumference was measured with a measuring tape at the midpoint between the lower margin of the last palpable rib and the top of the iliac crest, with the participant standing and at the end of normal expiration. Blood samples were collected after overnight fasting and analyzed for TC, HDL, triglycerides, and HbA1c levels. SBP measurements were performed twice on each arm using a digital device (Omron M10‐IT, Omron Healthcare Co), and the average of these readings was utilized.
The cardiometabolic health CS was calculated by as follows:
Age‐ and sex‐specific z‐scores were computed for each included variable. Inspection of age relationships supported the use of linear adjustment, and given the relatively narrow age range of our sample (50–64 years), this approach aligns with common practice in epidemiological research.
A combined lipid score was calculated as the mean of the z‐scores of TC:HDL and triglycerides to ensure an equal influence of dyslipidemia compared to other components.
The CS was obtained by averaging the z‐scores of waist circumference, HbA1c, SBP, and the combined lipid score for each participant.
The final CS was inverted (by multiplying by −1) to ensure that a higher score indicates better cardiometabolic health.
Cardiorespiratory fitness
Cardiorespiratory fitness was estimated using the submaximal Ekblom‐Bak cycle ergometer test [27]. This test evaluates the heart rate response to two consecutive submaximal workloads and is highly valid as compared to direct measurements (R 2 adj = 0.91, standard error of estimate: 0.28 l/min). Exclusion criteria included active infections, unstable cardiovascular disease, signs of cardiac disease detected through electrocardiography, use of beta‐blockers, a body weight over 125 kg, or a resting heart rate exceeding 100 beats/min.
Statistical analyses
The strength of the associations between various accelerometer‐derived physical activity measures and fitness as well as CS was investigated. Fitness and CS was standardized for sex and age before all subsequent analyses. For all measures of volume and MVPA, linear regression was used with the physical activity variable as independent and either fitness or CS as dependent. For the intensity spectrum measures, partial least squares regression was used with the intensity spectrum as predictor variables and either fitness or CS as outcome variable. The high‐resolution intensity spectrum results in highly collinear predictor variables, which requires a statistical method capable of handling this collinearity. Otherwise, the prediction by each individual variable as well as by the total regression model will be overestimated. Typically, partial least squares regression is used for the analysis of these multicollinear variables in physical activity research [28]. The number of partial least squares regression components was chosen based on Monte Carlo resampling [29].
To visualize and interpret the importance of different intensity levels in predicting health outcomes, selectivity fraction plots were used, expressed as R 2 (%). These plots display the proportion of each predictor variable's total variance that is explained by the predictive model, providing an easily interpretable measure between −1 and +1. The sign of the selectivity fraction comes from the direction of the relationship in the target projected component—when a variable moves in the opposite direction to the outcome in the projected space, it results in a negative value, whereas movement in the same direction results in a positive value. In practical terms, negative values represent inverse associations with the outcome (e.g., higher physical activity associated with lower CS), whereas positive values represent direct associations (e.g., higher sedentary time associated with higher CS). The magnitude represents the relative importance of each variable in the predictive model. Although individual predictive variables may show strong associations with the predictive model, the model typically explains less variance in the response variable.
FEM and cadence data were compared epoch‐by‐epoch, and epochs where both FEM and cadence were 0 were excluded. The relationship between FEM and cadence was visualized by plotting the epochs and fitting a smoothing spline to the data. Statistical analyses were performed in MATLAB R2023a (MathWorks).
Results
From the 12,109 individuals invited to participate in the Gothenburg SCAPIS site, the progression to the final analytical sample is shown in Fig. 1. The main reasons for non‐inclusion were declining participation (48.3% of those invited) and not completing the fitness test (28.0% of participants), with smaller proportions excluded due to invalid physical activity measurements or missing data for the cardiometabolic composite score (CS). Descriptive characteristics of the 4172 individuals from the SCAPIS subsample with valid measurements on all variables of interest are shown in Table 2. Compared to excluded participants, those included in the analysis had higher education levels (48.1% vs. 36.5% university degree), lower prevalence of cardiovascular disease/hypertension (16.3% vs. 39.9%), and better cardiometabolic health indicators (see Table S1). These differences likely reflect the requirements of the cardiorespiratory fitness test, which excluded individuals with cardiovascular conditions and those using beta‐blockers.
Fig. 1.
Flow diagram of study participation showing the progression from invited individuals to the final analytical sample, including reasons for non‐participation and exclusion.
Table 2.
Descriptive characteristics of the study sample.
| All | Male | Female | |
|---|---|---|---|
| N (%) | 4172 (100.0%) | 2013 (48.3%) | 2159 (51.7%) |
| Age (years) | 57.2 (4.3) | 57.2 (4.3) | 57.2 (4.3) |
| Education | |||
| University degree (%) | 2008 (48.1%) | 876 (43.5%) | 1132 (52.4%) |
| High school/vocational education (%) | 1737 (41.6%) | 891 (44.3%) | 846 (39.2%) |
| Elementary school (%) | 357 (8.6%) | 206 (10.2%) | 151 (7.0%) |
| Smoking habits | |||
| Never smoker (%) | 1970 (47.2%) | 1032 (51.3%) | 938 (43.4%) |
| Former smoker (%) | 1582 (37.9%) | 682 (33.9%) | 900 (41.7%) |
| Regular smoker/sometimes (%) | 524 (12.6%) | 245 (12.2%) | 279 (12.9%) |
| Cardiovascular disease/hypertension (%) | 678 (16.3%) | 358 (17.8%) | 320 (14.8%) |
| BMI (kg/m2) | 26.4 (4.0) | 26.9 (3.4) | 25.9 (4.4) |
| Waist circumference (cm) | 92.4 (12.0) | 97.7 (9.7) | 87.5 (11.7) |
| Systolic blood pressure (mm Hg) | 121 (16) | 125 (14) | 117 (17) |
| Glycated haemoglobin (mmol/mol) | 35.0 (5.0) | 35.1 (5.1) | 35.0 (4.9) |
| Total cholesterol (mmol/L) | 5.59 (1.02) | 5.46 (1.01) | 5.72 (1.00) |
| High density lipoprotein (mmol/L) | 1.72 (0.52) | 1.48 (0.41) | 1.93 (0.51) |
| Triglycerides (mmol/L) | 1.17 (0.97) | 1.36 (1.27) | 1.00 (0.50) |
| Cardiorespiratory fitness (mL/min/kg) | 33.9 (6.7) | 36.9 (5.9) | 31.1 (6.1) |
| Physical Activity | |||
| FEM mean mg | 21.0 (6.6) | 20.1 (6.6) | 21.8 (6.4) |
| Steps/day | 8357 (2 798) | 8096 (2 836) | 8 601 (2 741) |
| AG V counts/day | 293,564 (110 743) | 294,406 (117 847) | 292,778 (103 702) |
| AG VM counts/day | 581,183 (177 216) | 565,738 (183 211) | 595,584 (170 223) |
| FEM ≥3 METs (min/day) | 77.5 (27.6) | 74.3 (28.2) | 80.5 (26.6) |
| FEM ≥4 METs (min/day) | 10.4 (11.5) | 9.4 (11.1) | 11.4 (11.7) |
| FEM ≥4.5 METs (min/day) | 4.2 (6.9) | 3.9 (6.8) | 4.4 (6.9) |
| Cadence ≥80 steps/min (min/day) | 51.0 (22.0) | 49.7 (22.5) | 52.2 (21.4) |
| Cadence ≥100 steps/min (min/day) | 38.5 (19.6) | 36.9 (20.0) | 40.1 (19.1) |
| AG V ≥3 METs (min/day) | 54.8 (22.8) | 55.4 (24.1) | 54.2 (21.5) |
| AG VM ≥3 METs (min/day) | 81.7 (29.5) | 80.2 (30.9) | 83.1 (28.0) |
Abbreviations: AG, ActiGraph; FEM, frequency‐extended method 4 Hz; METs, metabolic equivalents of tasks; V, vertical axis; VM, vector magnitude.
The strength of the associations (explained variance) between the physical activity measures and fitness and CS is shown in Fig. 2, with detailed results provided in Table S2. Volume and MVPA measures were analyzed using linear regression and intensity spectrum measures were analyzed using partial least squares regression. In general, the FEM‐based measures displayed the strongest associations followed by step count, and ActiGraph‐based measures displayed the weakest associations. On average, comparing the strength of the associations with the health indicators for the different physical activity measures, the step‐based measures were 88% compared to the FEM‐based measures and the ActiGraph‐measures were 68% compared to the FEM‐based measures.
Fig. 2.
Strength of associations between physical activity measures and health outcomes. Panel A represents cardiorespiratory fitness associations, whereas Panel B represents associations with the cardiometabolic health composite score (combining blood pressure, waist circumference, blood sugar, and blood lipids). Volume and MVPA measures were analyzed using linear regression, and intensity spectrum measures were analyzed using partial least squares regression. Bar heights represent explained variance (R2) for each physical activity measure and analysis method. AG, ActiGraph; FEM, frequency‐extended method 4 Hz; METs, metabolic equivalents of task; V, vertical axis; VM, vector magnitude.
For the volume measures, FEM mean mg explained 13.5% and 5.4% of the variance in fitness and CS. respectively, whereas steps/day explained 11.1% and 4.6%. For MVPA measures, time above 4 METs explained 12.8% and 7.3%, whereas time above 80 steps/min explained 10.0% and 5.1%. Furthermore, the spectrum measures displayed the strongest associations, whereas volume and MVPA measures were more like each other in terms of strength. Among the different MVPA thresholds investigated, 4 METs resulted in the strongest association for the FEM method and 80 steps/min resulted in the strongest association for cadence. When combining the accelerometer intensity spectrum and cadence bands, the association increased from 19.3% to 22.7% and 7.8% to 8.2% for fitness and CS, respectively, compared to the accelerometer intensity spectrum alone. Subgroup analyses for men and women separately showed the same pattern of results as for the overall sample, with FEM measures displaying the strongest associations with the outcomes, followed by step measures, and ActiGraph measures displaying the weakest associations. Detailed sex‐specific analyses are provided in Table S2.
Fig. 3 shows the importance of the different physical activity intensities in the association with fitness and composite score (CS) for the accelerometer intensity spectrum and cadence bands. For the associations between the accelerometer intensity spectrum and fitness and CS, there was a peak in selectivity fraction at approximately five METs. Similarly, there was a selectivity fraction peak at approximately 80 steps/min for the associations between the cadence bands and fitness and CS, which was followed by a drop in selectivity fraction at 100 steps/min. For both the accelerometer intensity spectrum and cadence bands, the association above these peaks was higher for fitness compared to CS.
Fig. 3.
Selectivity fraction plots showing the relative importance of different physical activity intensities for health outcomes. Panels are arranged with FEM (frequency‐extended method) measured intensity (left) and step cadence (right), showing associations with cardiorespiratory fitness (upper) and cardiometabolic health composite score (lower). Shaded areas represent 95% confidence intervals. Reference lines indicate standard intensity thresholds: LPA, light physical activity (1.5 METs); MPA, moderate physical activity (3 METs); VPA, vigorous physical activity (6 METs), and VVPA, very vigorous physical activity (9 METs).
The relationship between FEM accelerometer output and cadence epoch‐by‐epoch for the study sample is shown in Fig. 4. The total number of epochs were 132,749,306 (epochs where both FEM and cadence were 0 were excluded). Comparing the different measures suggests that in free‐living conditions, 100 steps/min corresponds to an accelerometer output equivalent to approximately 4 METs, whereas 80 steps/min, which showed the strongest association with CS, corresponds to approximately 3.5 METs. The relationship between accelerometer output and cadence suggests that in the range 3–4 METs, there is a disproportionate increase in cadence from approximately 40–100 steps/min. In contrast, in the range of 5–9 METs, the cadence increases from approximately 130–170 steps/min.
Fig. 4.
Comparison between step cadence and movement intensity measured by FEM in free‐living conditions, based on 133 million 5‐s observations. The thick curved line represents a fitted smoothing spline. Reference lines indicate standard intensity thresholds for moderate (MPA), vigorous (VPA), and very vigorous physical activity (VVPA) measured by both methods. MPA, moderate physical activity; VPA, vigorous physical activity.
Discussion
This study explored the relationship between step counting metrics and full accelerometer data, and their associations with cardiometabolic health in middle‐aged adults under free‐living conditions. Our main findings are as follows: (1) Step counting metrics retained approximately 88% of the information from full accelerometer data in relation to fitness and cardiometabolic health; (2) intensity and volume measures mostly showed similar associations with health outcomes; (3) the optimal FEM accelerometer intensity associated with cardiometabolic health was around 4 METs; (4) a step cadence of 80 steps/min, rather than the commonly used 100 steps/min, appeared more relevant for capturing moderate intensity activity in this middle aged population; and (5) combining step and accelerometer data provided additional explanatory power for cardiometabolic health. These findings suggest that step data would contain most information from full accelerometer data and its association with cardiometabolic health.
Our results demonstrate that step metrics capture a substantial portion of the information provided by full accelerometer data in relation to fitness and cardiometabolic health. This finding aligns with previous research suggesting the utility of step‐based metrics as simple, interpretable measures of physical activity [6]. The correspondence between step cadence and accelerometer‐derived intensity further supports the use of cadence as a proxy for intensity. Importantly, step metrics outperformed traditional ActiGraph count‐based measures, suggesting that steps may be a more robust representation of physical activity in free‐living conditions. These findings support the continued use and development of step‐based metrics for physical activity assessment and promotion.
The finding that FEM and cadence combined yielded modestly increased explanatory power (from 19.3% to 22.7% for fitness and 7.8% to 8.2% for CS) suggests that these measures capture complementary aspects of physical activity. Both metrics are derived from the same accelerometer signal and are closely related to walking speed, which itself has established associations with cardiometabolic health outcomes. However, they emphasize different movement characteristics—FEM provides a direct measure of movement intensity through acceleration, whereas cadence specifically quantifies the rhythmic pattern of stepping. The improved association when combining these metrics may be explained by cadence capturing the organized, purposeful nature of ambulatory activity, whereas FEM better reflects the overall movement intensity, including non‐ambulatory activities. Although the small but consistent additional variance explained indicates potential value for research applications where maximal measurement precision is required, the clinical relevance is small.
The associations between physical activity measures and cardiometabolic health showed varying patterns depending on the metric used. Simple volume measures (steps/day and mean mg) and basic intensity measures (MVPA) generally showed similar strengths of association with cardiometabolic health. An exception to this was MVPA based on a four MET threshold using the FEM method, which displayed a stronger association. Notably, spectrum‐based methods consistently demonstrated stronger associations with cardiometabolic health compared to simpler measures, regardless of whether they were based on step cadence or accelerometer intensity. This suggests that capturing the full range of activity intensities provides more comprehensive information about an individual's physical activity patterns and their relationship to health outcomes.
For step metrics specifically, the difference in association strength between intensity (cadence) and volume (steps/day) measures was minimal. This indicates that in the context of cardiometabolic health assessment, simple step volume (steps/day) may be nearly as informative as more complex cadence‐based metrics. For accelerometry, the strongest cross‐sectional associations with cardiometabolic health were observed at intensities around four METs, which is higher than the traditional three MET threshold for moderate intensity. This finding supports recent critiques of the three MET threshold as being too low for most individuals [14]. Physical activity of higher intensities than four METs might have more health beneficial physiological effects [2]. However, the proportion of individuals in the sample with activity at these higher intensities might be limited and therefore not as influential in the association with the outcomes.
Our cross‐sectional analyses suggest that a step cadence of 80 steps/min showed stronger associations with health outcomes than the commonly used 100 steps/min threshold in free‐living conditions among middle‐aged adults. The former would probably correspond to 3.5 METs and the later to 4 METs measured by accelerometry under free‐living conditions. This finding aligns with the results of Adams et al. [15], who also found lower cadence thresholds associated with cardiometabolic health benefits in free‐living conditions. However, these results contrast with several laboratory‐based studies that consistently found thresholds around 100 steps/min for moderate intensity across adult age groups [7, 8, 9, 10, 11]. In addition, the drop in the association with fitness and CS at the cadence between 90 and 120 steps/min observed in our study may indicate that the population investigated does not spend much time at this cadence interval under free‐living conditions. Walking commuters may have a cadence of approximately 120 steps/min and at a speed of 5.8 km/h with a MET‐value of 6 [30], corresponding approximately to the second association peak in our study and again aligning to the results of Adams et al. [15]. These discrepancies highlight the potential limitations of applying laboratory‐derived thresholds to free‐living populations and emphasize the need for population‐specific calibration of step‐based intensity thresholds.
Strengths and limitations
Strengths of this study include the large sample size, the use of multiple accelerometer processing methods, and the application of advanced statistical techniques to analyze intensity spectra. The study also benefits from its focus on free‐living conditions, providing applicable results. Limitations include the cross‐sectional design, which precludes causal inferences. As with all cross‐sectional studies, the direction of the observed associations cannot be determined. Although physical activity likely influences cardiometabolic health, the reverse relationship is also plausible—individuals with better cardiometabolic health may be more capable of engaging in higher intensity physical activity. This bidirectional relationship could particularly affect our findings regarding optimal intensity thresholds, as those with poorer health may be limited in their ability to achieve higher intensities regardless of their motivation to be physically active.
The study sample was healthier than the general SCAPIS cohort due to exclusion criteria for the cardiorespiratory fitness test. This selection bias means our findings primarily reflect associations in relatively healthy middle‐aged adults and may not generalize to those with established cardiovascular disease. Although some participants with cardiovascular disease/hypertension (16.3%) were included in the analyses, this represents a relatively small proportion compared to the excluded group (39.9%). The inclusion of these participants was deemed appropriate as they had passed the fitness test screening criteria, indicating sufficient cardiovascular health to safely perform moderate physical activity. The resulting sample characteristics enable the study of subclinical markers of cardiometabolic health, which is valuable for understanding early prevention opportunities in relatively healthy populations. The use of a continuous composite score allowed for capturing subtle variations in cardiometabolic health status, though the clinical significance of these small variations remains uncertain and requires further investigation.
The focus on a specific age group (50–64 years) may further limit generalizability to other age groups. It is important to note that both step cadence and accelerometer metrics represent absolute measures of intensity, despite health benefits likely being relative to individual fitness level [14]. The cadence threshold of 80 steps/min identified in this study may therefore only be applicable to individuals with similar fitness levels as our sample. Additionally, although our composite score of cardiometabolic health is comprehensive, it may not capture all aspects of health that could be influenced by physical activity.
Conclusions and clinical implications
In conclusion, our findings support the utility of step‐based metrics for assessing physical activity and its association with cardiometabolic health in middle‐aged adults. Most of the health‐related information in the accelerometer data seems to be provided by the step metric. This suggests that clinicians could potentially use simple step data to effectively assess patients’ physical activity levels in routine practice. Furthermore, step counting serves as an easily interpretable tool for prescribing physical activity in clinical and public health settings.
However, the results also highlight the challenges in translating laboratory‐derived intensity thresholds to free‐living populations. The discrepancy between our findings and those from controlled laboratory studies underscores the need for population‐specific calibration of step‐based intensity thresholds. Future physical activity guidelines based on steps should consider the activity patterns observed in free‐living conditions and may need to recommend a cadence threshold of ≥80 steps/min, which is lower than those derived from laboratory studies. This lower threshold could be more achievable for many patients, though longitudinal studies are needed to confirm whether it leads to improved adherence and health outcomes. Still, walking at a moderate‐fast walking pace corresponding to a cadence of ≥120 steps/min would provide additional health benefits through improved aerobic fitness [13]. Clinicians could use this information to provide tailored advice based on individual patient capabilities and health goals. Further research is needed to validate these findings in different populations and to explore the longitudinal relationships between step‐based metrics and health outcomes.
Author contributions
Jonatan Fridolfsson, Anders Raustorp, and Daniel Arvidsson conceptualized the study and wrote the initial manuscript draft. Elin Ekblom‐Bak, Örjan Ekblom, and Mats Börjesson were responsible for the data collection. Jonatan Fridolfsson performed the data processing, analysis and visualization. Elin Ekblom‐Bak, Örjan Ekblom, and Mats Börjesson made substantial manuscript revisions.
Conflict of interest statement
The authors declare no conflicts of interest.
Funding information
The main funding body of the Swedish CArdioPulmonary bioImage Study (SCAPIS) is the Swedish Heart‐Lung Foundation. The study is also funded by the Knut and Alice Wallenberg Foundation, the Swedish Research Council, VINNOVA (Sweden's Innovation agency), the University of Gothenburg and Sahlgrenska University Hospital, Karolinska Institutet and Stockholm County Council, Linköping University and University Hospital, Lund University and Skåne University Hospital, Umeå University and University Hospital, Uppsala University, and University Hospital. Ö.E. and E.E.B. were funded by Skandia Risk&Hälsa and M.B. by Heart‐and Lung foundation (grant 20210270).
Supporting information
Table S1. Descriptive characteristics of the included study sample and of the excluded study participants without complete data on all variables of interest.
Table S2. Strength of associations (explained variance, R 2) between physical activity measures and cardiorespiratory fitness and the cardiometabolic health composite score (combining blood pressure, waist circumference, blood sugar, and blood lipids) in the overall sample and for men and women separately. Volume and MVPA measures were analyzed using linear regression and intensity spectrum (band) measures were analyzed using partial least squares regression.
Fridolfsson J, Raustorp A, Börjesson M, Ekblom‐Bak E, Ekblom Ö, Arvidsson D. Simple step counting captures comparable health information to complex accelerometer measurements. J Intern Med. 2025;297:492–504.
Contributor Information
Jonatan Fridolfsson, Email: jonatan.fridolfsson@gu.se.
Daniel Arvidsson, Email: daniel.arvidsson@gu.se.
Data availability statement
Data may be obtained from a third party and are not publicly available. The data cannot be freely available as it contains sensitive personal information. Information regarding application for accessing the data can be found on the SCAPIS study organization website www.scapis.org.
References
- 1. Lee IM, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT, et al. Effect of physical inactivity on major non‐communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219–29. 10.1016/S0140-6736(12)61031-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. 2018 Physical Activity Guidelines Advisory Committee . In 2018 physical activity guidelines advisory committee scientific report. Washington DC: US Department of Health and Human Services; 2018. [Google Scholar]
- 3. Bull FC, Al‐Ansari SS, Biddle S, Borodulin K, Buman MP, Cardon G, et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. 2020;54(24):1451–62. 10.1136/bjsports-2020-102955. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. US Department of Health and Human Services . Physical Activity and Health: A Report of the Surgeon General. In Department of Health and Human Services, Centers for Disease Control and Prevention. Atlanta: National Center for Chronic Disease Prevention and Health Promotion; 1996. [Google Scholar]
- 5. Prince SA, Adamo KB, Hamel ME, Hardt J, Gorber SC, Tremblay M. A comparison of direct versus self‐report measures for assessing physical activity in adults: a systematic review. Int J Behav Nutr Phys Act. 2008;5:56. 10.1186/1479-5868-5-56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Tudor‐Locke C, Han H, Aguiar EJ, Barreira TV, Schuna JM, Kang M, et al. How fast is fast enough? Walking cadence (steps/min) as a practical estimate of intensity in adults: a narrative review. Br J Sports Med. 2018;52(12):776–88. 10.1136/bjsports-2017-097628. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. McAvoy CR, Moore CC, Aguiar EJ, Ducharme SW, Schuna JM, Barreira TV, et al. Cadence (steps/min) and relative intensity in 21 to 60‐year‐olds: the CADENCE‐adults study. Int J Behav Nutr Phys Act. 2021;18(1):27. 10.1186/s12966-021-01096-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. McAvoy CR, Miller TA, Aguiar EJ, Ducharme SW, Moore CC, Schuna JM, et al. Cadence (steps/min) and relative intensity in 61 to 85‐year‐olds: the CADENCE‐adults study. Int J Behav Nutr Phys Act. 2023;20(1):141. 10.1186/s12966-023-01543-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Tudor‐Locke C, Aguiar EJ, Han H, Ducharme SW, Schuna JM, Barreira TV, et al. Walking cadence (steps/min) and intensity in 21–40 year olds: CADENCE‐adults. Int J Behav Nutr Phys Act. 2019;16(1):8. 10.1186/s12966-019-0769-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Tudor‐Locke C, Ducharme SW, Aguiar EJ, Schuna JM, Barreira TV, Moore CC, et al. Walking cadence (steps/min) and intensity in 41 to 60‐year‐old adults: the CADENCE‐adults study. Int J Behav Nutr Phys Act. 2020;17(1):137. 10.1186/s12966-020-01045-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Tudor‐Locke C, Mora‐Gonzalez J, Ducharme SW, Aguiar EJ, Schuna JM, Barreira TV, et al. Walking cadence (steps/min) and intensity in 61–85‐year‐old adults: the CADENCE‐adults study. Int J Behav Nutr Phys Act. 2021;18(1):129. 10.1186/s12966-021-01199-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Herrmann SD, Willis EA, Ainsworth BE, Barreira TV, Hastert M, Kracht CL, et al. 2024 adult compendium of physical activities: a third update of the energy costs of human activities. J Sport Health Sci. 2024;13(1):6–12. 10.1016/j.jshs.2023.10.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Fridolfsson J, Ekblom‐Bak E, Ekblom Ö, Bergström G, Arvidsson D, Börjesson M. Fitness‐related physical activity intensity explains most of the association between accelerometer data and cardiometabolic health in persons 50–64 years old. Br J Sports Med. 2024;58(21):1244–1250. 10.1136/bjsports-2023-107451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Fridolfsson J, Arvidsson D, Ekblom‐Bak E, Ekblom Ö, Bergström G, Börjesson M. Accelerometer‐measured absolute versus relative physical activity intensity: cross‐sectional associations with cardiometabolic health in midlife. BMC Public Health. 2023;23(1):2322. 10.1186/s12889-023-17281-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Adams B, Fidler K, Demoes N, Aguiar EJ, Ducharme SW, McCullough AK, et al. Cardiometabolic thresholds for peak 30‐min cadence and steps/day. PLoS ONE. 2019;14(8):e0219933. 10.1371/journal.pone.0219933. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Wold S, Sjöström M, Eriksson L. PLS‐regression: a basic tool of chemometrics. Chemom Intell Lab. 2001;58(2):109–30. 10.1016/S0169-7439(01)00155-1. [DOI] [Google Scholar]
- 17. Bergström G, Berglund G, Blomberg A, Brandberg J, Engström G, Engvall J, et al. The Swedish CArdioPulmonary BioImage study: objectives and design. J Intern Med. 2015;278(6):645–59. 10.1111/joim.12384. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Fridolfsson J, Borjesson M, Buck C, Ekblom O, Ekblom‐Bak E, Hunsberger M, et al. Effects of frequency filtering on intensity and noise in accelerometer‐based physical activity measurements. Sensors (Basel). 2019;19(9):2186. 10.3390/s19092186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Holm I, Fridolfsson J, Borjesson M, Arvidsson D. Fourteen days free‐living evaluation of an open‐source algorithm for counting steps in healthy adults with a large variation in physical activity level. BMC Biomed Eng. 2023;5(1):3. 10.1186/s42490-023-00071-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Tudor‐Locke C, Sisson SB, Lee SM, Craig CL, Plotnikoff RC, Bauman A, et al. Evaluation of quality of commercial pedometers. Can J Public Health. 2006;97(1):S10‐5, S10‐6. 10.1007/BF03405359. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Migueles JH, Cadenas‐Sanchez C, Ekelund U, Delisle Nystrom C, Mora‐Gonzalez J, Lof M, et al. Accelerometer data collection and processing criteria to assess physical activity and other outcomes: a systematic review and practical considerations. Sports Med. 2017;47(9):1821–45. 10.1007/s40279-017-0716-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Freedson PS, Melanson E, Sirard J. Calibration of the computer science and applications, Inc. accelerometer. Med Sci Sports Exerc. 1998;30(5):777–81. 10.1097/00005768-199805000-00021. [DOI] [PubMed] [Google Scholar]
- 23. Sasaki JE, John D, Freedson PS. Validation and comparison of ActiGraph activity monitors. J Sci Med Sport. 2011;14(5):411–6. 10.1016/j.jsams.2011.04.003. [DOI] [PubMed] [Google Scholar]
- 24. Ahrens W, Moreno LA, Marild S, Molnar D, Siani A, De Henauw S, et al. Metabolic syndrome in young children: definitions and results of the IDEFICS study. Int J Obes (Lond). 2014;38(Suppl 2):S4–14. 10.1038/ijo.2014.130. [DOI] [PubMed] [Google Scholar]
- 25. Gurka MJ, Filipp SL, Pearson TA, DeBoer MD. Assessing baseline and temporal changes in cardiometabolic risk using metabolic syndrome severity and common risk scores. J Am Heart Assoc. 2018;7(16):e009754. 10.1161/JAHA.118.009754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Viitasalo A, Lakka TA, Laaksonen DE, Savonen K, Lakka H‐M, Hassinen M, et al. Validation of metabolic syndrome score by confirmatory factor analysis in children and adults and prediction of cardiometabolic outcomes in adults. Diabetologia. 2014;57:940–9. [DOI] [PubMed] [Google Scholar]
- 27. Bjorkman F, Ekblom‐Bak E, Ekblom O, Ekblom B. Validity of the revised Ekblom Bak cycle ergometer test in adults. Eur J Appl Physiol. 2016;116(9):1627–38. 10.1007/s00421-016-3412-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Aadland E, Kvalheim OM, Anderssen SA, Resaland GK, Andersen LB. Multicollinear physical activity accelerometry data and associations to cardiometabolic health: challenges, pitfalls, and potential solutions. I J Behav Nutr Phys Act. 2019;16:1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Kvalheim OM, Arneberg R, Grung B, Rajalahti T. Determination of optimum number of components in partial least squares regression from distributions of the root‐mean‐squared error obtained by Monte Carlo resampling. J Chemom. 2018;32(4):e2993. 10.1002/cem.2993. [DOI] [Google Scholar]
- 30. Schantz P, Olsson KSE, Salier Eriksson J, Rosdahl H. Frontiers | perspectives on exercise intensity, volume, step characteristics and health outcomes in walking for transport. Front Public Health. 2022;10:911863. 10.3389/fpubh.2022.911863. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Descriptive characteristics of the included study sample and of the excluded study participants without complete data on all variables of interest.
Table S2. Strength of associations (explained variance, R 2) between physical activity measures and cardiorespiratory fitness and the cardiometabolic health composite score (combining blood pressure, waist circumference, blood sugar, and blood lipids) in the overall sample and for men and women separately. Volume and MVPA measures were analyzed using linear regression and intensity spectrum (band) measures were analyzed using partial least squares regression.
Data Availability Statement
Data may be obtained from a third party and are not publicly available. The data cannot be freely available as it contains sensitive personal information. Information regarding application for accessing the data can be found on the SCAPIS study organization website www.scapis.org.