Abstract
Objective:
To examine the associations between the number of days per week achieving various daily step thresholds and all-cause mortality and cardiovascular disease (CVD) incidence in older women.
Methods:
We conducted a prospective cohort study of 13,547 women free of CVD and cancer (mean age, 71.8 years). We included participants who wore an ActiGraph GT3X+ accelerometer for 7 consecutive days between 2011–2015 and were subsequently followed for mortality through 2024. Women were classified by the number of days per week achieving step thresholds of ≥4000, ≥5000, ≥6000, or ≥7000 steps/day. Cox proportional hazards regression estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause mortality and CVD incidence, adjusting for lifestyle behaviors and comorbidities.
Results:
During a median follow-up of 10.9 years, 1765 women (13.0%) died and 781 (5.1%) developed CVD. Achieving ≥4000 steps/day on 1–2 and ≥3 days/week was associated with lower mortality risk compared with 0 days/week (adjusted HRs, 0.74 [95% CI, 0.65–0.86] and 0.60 [0.53–0.68], respectively). For CVD, corresponding results were 0.73 [0.58–0.92] and 0.73 [0.60–0.89], respectively. An inverse curvilinear dose-response relationship was observed for mortality, such that with higher step thresholds (5,000, 6,000 or 7,000), the risk of mortality further declined modestly. With additional adjustment for mean daily steps, associations were attenuated to the null.
Conclusions:
Among older women, achieving ≥4000 steps/day on even 1–2 days/week was associated with lower mortality and CVD, while more steps were associated with even better outcomes. A greater number of steps, regardless of daily patterns, is associated with better health outcomes.
Keywords: steps, physical activity, mortality, cardiovascular disease, cohort study
Introduction
Before industrialization, most adults took ~15,000–20,000 steps/day, but desk jobs, motorized transportation, and technologies that promote sedentarism have decreased levels to ~5,000 steps/day, especially among older individuals.[1,2] Since lifelong physical activity (PA) stimulates repair and maintenance mechanisms, lack of PA with aging is a mismatch that increases vulnerability to disease and fails to slow aging.[3] For this reason, a large body of evidence shows that lifelong PA is important for improving health span.[4,5] How much PA should people accomplish, especially as they age? Given increasingly low levels of PA in industrial populations,[6–8] it is also useful to ask how little PA can older individuals adults accomplish to gain appreciable health benefits.
Given the ubiquity of cell phones/commercial wearables that report step counts,[9] it is important to incorporate step metrics into the next set of PA guidelines (previous guidelines have not included steps because few data were available[5]). While inverse dose-response associations between step counts and all-cause mortality or incident cardiovascular disease (CVD) are well documented[10,11], few data exist on the frequency of meeting daily step thresholds and clinical outcomes. This has practical implications for PA guidelines as individuals may have different preferences for being active. For example, some may be regularly active (thus, showing relatively even step counts across days), while others – with the same total step count – may be active only on a few days (thus, “bunching” their step counts). A few studies have examined the “weekend warrior” pattern of exercise (i.e., exercise done on only 1–2 days/week) in relation to health outcomes [12,13], but little is known for steps.
The next update to the ‘Physical Activity Guidelines for Americans’ is slated for 2028[4,5]. To address gaps in knowledge for these guidelines, we investigate the associations between number of days/week of step counts above different thresholds and all-cause mortality and CVD among older women. We examined step thresholds of 4000 to 7000 steps/day, based on previous work showing the dose-response curve in these women[14,15]. We further ascertain whether it is the total number of daily steps that drives any observed associations, rather than frequency of achieving the thresholds, to address an important translational aspect for future guidelines.
Methods
Study Participants
This study included participants from the Women’s Health Study (WHS). The WHS was originally designed as a randomized trial that examined low-dose aspirin and vitamin E for preventing CVD and cancer among 39,876 women aged ≥45 years, conducted from 1992 to 2004 in the US.[16–18] After the trial ended, 33,682 women provided consent to be followed in an observational study. Between 2011 and 2015, surviving women were invited to participate in an ancillary study that measured PA for 7 days using accelerometers.[19] A total of 18,289 women aged ≥62 years were eligible, of whom 17,708 wore and returned their devices.[20] The device failed to record data in 242 women, leading to 17,4 66 women with available PA data. The present study excluded women with <7 days of ≥10 h/d of wear (N=1685; although compliance is conventionally defined as ≥10 h/d of wear on ≥4 days,[21] we included women with 7 days to investigate weekly patterns) and excluded those with a history of CVD or cancer (N=2234), leaving 13,547 women. All women provided written informed consent to participate, and the study was approved by Brigham and Women’s Hospital’s Institutional Review Board committee.
Assessment of step counts
Participants were mailed ActiGraph GT3X+ accelerometers (ActiGraph Corp, Pensacola, Florida) and asked to wear the device on the hip for 7 consecutive days, removing it only during sleep and water-based activities. Devices were returned by mail, and data were screened for wear time using standard techniques[22]. Accelerometry data were collected at 30 Hz and aggregated into 60-second, time-stamped epochs. Non-wear time was assessed using the validated Choi et al. algorithm (at least 90 consecutive minutes of zero vector magnitude (VM) counts per minute, with a 2-minute allowance for non-zero VM counts, and requiring no counts 30 minutes before or after that period).[23] Steps per day were determined using the manufacturer’s step algorithm.[24,25] For women with >7 days’ wear, data from the first 7 days were used (n=45). Step counts assessed with this device have been shown to be valid in older adults.[26]
Ascertainment of outcomes
Women were followed through December 31, 2024. Deaths were reported by family members or postal authorities, with medical records and death certificates obtained to confirm the reports, or they were ascertained through the National Death Index. Women reported CVD events on annual questionnaires and medical records were obtained to adjudicate these self-reports. Established criteria were used to confirm myocardial infarction (MI) and stroke. [27,28] CVD was defined as a composite of death from cardiovascular causes (based on ICD10 I00-I99), nonfatal myocardial infarction, or nonfatal stroke.
Assessment of other variables
Participants completed annual questionnaires on sociodemographic characteristics, health habits, and personal and family medical history. We used information from the questionnaire closest to the time of accelerometer wear (typically within 1–2 years) to ascertain the following: weight, height, smoking status, alcohol use, postmenopausal hormone use, general health, cancer screening, and family history of CVD. A 131-item food frequency questionnaire was used to assess dietary habits at the start of the WHS.
Statistical analyses
Because prior analyses showed lower mortality rates at a mean of ~4,000 steps/day, and an inverse dose-response relationship that levelled at ~7,500 steps/day,[29] we chose a priori thresholds of 4,000, 5,000, 6,000 and 7,000 steps/day. Additionally, we categorized women into daily patterns of meeting the thresholds – 0, 1–2, or ≥3 days – based on recent analyses of NHANES data.[30]
We first examined characteristics of women who met the 4,000 steps/day threshold on 0, 1–2, or ≥3 days. We then examined associations with all-cause mortality and CVD using Cox proportional hazards models with age as the time scale. Models were adjusted for accelerometer wear time (continuous) in Model 1, and additionally for smoking (never, past, current); alcohol use (rarely, monthly, weekly, daily); intakes of saturated fat, fiber, fruit, and vegetables (quintiles of each); postmenopausal hormone therapy (never, past, current); self-rated health (excellent, very good, good, fair/poor); cancer screening in the past year (yes, no); parental history of MI before 60 years of age; and family history of cancer (no or yes for each) in Model 2. To investigate how much step volume influenced observed associations, mean daily steps (as well as squared steps for finer control) were further adjusted in Model 3.
We also computed non-linear associations between the number of days achieving a particular step threshold and outcomes using restricted cubic splines with 4 knots (5th, 35th, 65th, and 95th percentiles), using 0 days as referent.
Finally, to directly compare the different step thresholds against a common referent of 0 days with ≥4000 steps/day, we undertook a set of analyses with this referent. Because this is an observational study potentially confounded by health status (i.e., poor health leading to fewer steps), we sought to minimize this bias in sensitivity analyses that excluded (i) deaths in the first two years of follow-up and (ii) women reporting fair or poor health.[31]. Because body mass index (BMI) may be considered a confounder or mediator, we conducted sensitivity analysis additionally adjusting for BMI. Missing values were <0.3% for each variable, and these were singly imputed using a random forest.[32] Statistical analysis was performed using R 4.4.2 (R Foundation).
Equity, diversity and inclusion statement
The author group consists of early (n=1) and senior researchers (n=3), with male: female ratio 2: 2. Two authors originate from east and southeast Asia; two, US. The study population included US, older females, and thus findings may not be generalizable to other populations.
Results
The mean (standard deviation) age of the 13,547 women was 71.8 (5.6) years, and mean daily steps was 5615 (2675) steps/day. Table 1 shows the characteristics of participants by number of days/week (0, 1–2 or ≥3) reaching the 4000 steps/day threshold. Women achieving the threshold on greater numbers of days were more likely to be younger, and had lower BMI and better self-rated health.
Table 1.
Baseline characteristics of women by number of days achieving ≥4000 steps/day threshold
| Number of days/week with ≥4000 steps/day | |||
|---|---|---|---|
| 0 days | 1–2 days | ≥3 days | |
| N=1639 | N=1846 | N=10062 | |
| Age, years | 76.2 (6.6) | 73.4 (5.9) | 70.8 (4.9) |
| BMI, kg/m2 | 28.4 (6.2) | 27.8 (5.4) | 25.5 (4.5) |
| Smoking | |||
| Never | 798 (48.7) | 921 (49.9) | 5193 (51.6) |
| Past | 727 (44.4) | 854 (46.3) | 4587 (45.6) |
| Current | 114 (7.0) | 71 (3.8) | 282 (2.8) |
| Alcohol use | |||
| Rarely | 848 (51.7) | 790 (42.8) | 3455 (34.3) |
| Monthly | 146 (8.9) | 219 (11.9) | 971 (9.7) |
| Weekly | 468 (28.6) | 578 (31.3) | 3910 (38.9) |
| Daily | 177 (10.8) | 259 (14.0) | 1726 (17.2) |
| Postmenopausal hormone therapy | |||
| Never | 366 (22.3) | 352 (19.1) | 1705 (16.9) |
| Past | 1146 (69.9) | 1304 (70.6) | 7235 (71.9) |
| Current | 127 (7.7) | 190 (10.3) | 1122 (11.2) |
| Self-rated health | |||
| Excellent | 182 (11.1) | 275 (14.9) | 3076 (30.6) |
| Very good | 708 (43.2) | 990 (53.6) | 5169 (51.4) |
| Good | 627 (38.3) | 528 (28.6) | 1720 (17.1) |
| Fair or poor | 122 (7.4) | 53 (2.9) | 97 (1.0) |
| Cancer screening | 1173 (71.6) | 1461 (79.1) | 8450 (84.0) |
| Family history of myocardial | |||
| infarction before age 60 years | 232 (14.2) | 273 (14.8) | 1444 (14.4) |
| Family history of cancer | 440 (26.8) | 496 (26.9) | 2523 (25.1) |
| Step count, steps/d | 2245 [1758, 2656] | 3295 [2996, 3615] | 6049 [4871, 7732] |
| Device wear time, min/d | 860 [800, 910] | 880 [833, 924] | 904 [860, 945] |
Numbers indicate frequency (%) for categorical variables and mean (standard deviation) or median (interquartile range) for continuous variables.
During a median follow-up of 10.9 (IQR: 9.9, 11.8) years, 1765 women (13.0%) died and 781 women (5.8%) developed CVD. Figure 1 illustrates the distribution of women achieving various step thresholds and the corresponding all-cause mortality and CVD rates. Generally, outcome rates exhibited an inverse curvilinear relation with the number of days achieving each specified threshold.
Figure 1: Distribution of women achieving daily step thresholds and all-cause mortality and cardiovascular disease (CVD) rates.
Number of women achieving specified daily step thresholds of ≥4000, ≥5000, ≥6000, and ≥7000 steps/day on 0, 1, 2 ,,, 7 days (histograms; left y-axis showing the frequency), overlayed with the rates of all-cause mortality (blue) and CVD (red) in each category (points; right y-axis showing the event rates per 100 person-year).
Association of the number of days achieving specified daily step thresholds with all-cause mortality and CVD
Table 2 shows hazard ratios (HRs) for all-cause mortality associated with frequency of meeting a specified step threshold, setting 0 days above 4,000 steps as the referent. At the 4,000 steps/day threshold, in Model 1, HRs for achieving this threshold on 1–2 and ≥3 days, compared with 0 days, were 0.66 (95% confidence interval [CI]: 0.57, 0.76) and 0.48 (0.43, 0.54), respectively. Associations were moderately attenuated when other covariates were adjusted (Model 2 HR’s 0.74 [0.65, 0.86] and 0.60 [0.53, 0.68], respectively; p = 0.010, comparing 1–2 with ≥3 days). When models were further adjusted for mean daily steps (Model 3), associations became attenuated to the null, indicating the inverse relationship results from the volume (i.e., number) of steps rather than the number of days reaching ≥4,000 steps. Setting step thresholds at 5,000, 6,000 and 7,000 steps/day yielded similar results, both in pattern and magnitude of associations.
Table 2:
Hazard ratios and 95% confidence intervals of all-cause mortality by number of days per week achieving specified daily step thresholds
| Threshold | Number of days/week | ||
|---|---|---|---|
| ≥4000 steps/day | 0 days | 1–2 days | ≥3 days |
| Median (IQR) steps/day | 2245 (1758, 2656) | 3295 (2996, 3615) | 6049 (4871, 7732) |
| Cases/N | 553/ 1639 | 321/ 1846 | 891/ 10062 |
| Hazard ratio (95% confidence interval) | |||
| Model 1 | 1.00 (reference) | 0.66 (0.57, 0.76) | 0.48 (0.43, 0.54) |
| Model 2 | 1.00 (reference) | 0.74 (0.65, 0.86) | 0.60 (0.53, 0.68) |
| Model 3 | 1.00 (reference) | 0.83 (0.71, 0.96) | 0.82 (0.68, 1.00) |
|
| |||
|
| |||
| Threshold | Number of days/week | ||
| ≥5000 steps/day | 0 days | 1–2 days | ≥3 days |
| Median (IQR) steps/day | 2749 (2164, 3264) | 4132 (3746, 4501) | 6652 (5577, 8240) |
|
| |||
| Cases/N | 800/ 3011 | 335/ 2577 | 630/ 7959 |
| Hazard ratio (95% confidence interval) | |||
| Model 1 | 1.00 (reference) | 0.66 (0.58, 0.76) | 0.52 (0.47, 0.58) |
| Model 2 | 1.00 (reference) | 0.74 (0.65, 0.85) | 0.64 (0.57, 0.72) |
| Model 3 | 1.00 (reference) | 0.87 (0.75, 1.01) | 0.93 (0.76, 1.13) |
|
| |||
| Threshold | Number of days/week | ||
| ≥6000 steps/day | 0 days | 1–2 days | ≥3 days |
| Median (IQR) steps/day | 3185 (2469, 3846) | 4885 (4410, 5322) | 7344 (6352, 8823) |
|
| |||
| Cases/N | 1025/ 4527 | 318/ 3062 | 422/ 5958 |
| Hazard ratio (95% confidence interval) | |||
| Model 1 | 1.00 (reference) | 0.65 (0.58, 0.74) | 0.54 (0.48, 0.61) |
| Model 2 | 1.00 (reference) | 0.74 (0.65, 0.85) | 0.66 (0.58, 0.75) |
| Model 3 | 1.00 (reference) | 0.92 (0.79, 1.08) | 1.05 (0.84, 1.30) |
|
| |||
|
| |||
| Threshold | Number of days/week | ||
| ≥7000 steps/day | 0 days | 1–2 days | ≥3 days |
| Median (IQR) steps/day | 3578 (2735, 4368) | 5595 (5020, 6104) | 8035 (7092, 9541) |
|
| |||
| Cases/N | 1223/ 6132 | 258/ 3085 | 284/ 4330 |
| Hazard ratio (95% confidence interval) | |||
| Model 1 | 1.00 (reference) | 0.60 (0.52, 0.69) | 0.56 (0.49, 0.64) |
| Model 2 | 1.00 (reference) | 0.68 (0.59, 0.78) | 0.68 (0.59, 0.78) |
| Model 3 | 1.00 (reference) | 0.88 (0.74, 1.03) | 1.10 (0.88, 1.37) |
Age was used as the time scale in the Cox proportional hazard models, adjusted for accelerometer wear time (continuous) in Model 1, and additionally for smoking (never, past, current); alcohol use (rarely, monthly, weekly, daily); intakes of saturated fat, fiber, fruit, and vegetables (quintiles of each); postmenopausal hormone therapy (never, past, current); self-rated health (excellent, very good, good, fair/poor); cancer screening in the past year (yes. no); parental history of myocardial infarction before 60 years of age; and family history of cancer (no or yes for each) in Model 2, and further for mean daily steps (continuous) and the square of mean daily steps (continuous) in Model 3.
Similar patterns were observed for CVD (Supplemental Table 1). For example, at the 4,000 steps/day threshold, HRs in Model 2 were 0.73 (0.58, 0.92) and 0.73 (0.60, 0.89) on 1–2 and ≥3 days, respectively. As with all-cause mortality, further adjustment for mean daily steps (Model 3) generally attenuated associations to the null.
Non-linear association of the number of days achieving specified daily step thresholds with all-cause mortality and CVD
To more finely illustrate the non-linear nature of the associations, Figure 2 shows results of analyses where frequency of meeting a specified daily step threshold was treated as a continuous variable. Because there were relatively few women achieving the higher step thresholds on most days of the week, confidence intervals were wide for these women. For all-cause mortality, in Model 2, gradually lower HRs were observed until 3 to 4 days per week achieving the step threshold, compared with no days, beyond which no further decrements occurred. For CVD, in Model 2, similar associations were observed for 4,000 and 5,000 steps/day thresholds; at higher thresholds, there was little association. As in the previous set of analyses, when mean daily steps were further adjusted, the associations were largely attenuated with wider confidence intervals.
Figure 2: Non-linear hazard ratios of all-cause mortality and cardiovascular disease (CVD) according to the number of days per week achieving specified daily step thresholds.
Non-linear associations between the number of days achieving specified daily step thresholds and all-cause mortality and CVD, based on restricted cubic splines in Cox proportional hazard models with 0 days as referent (Model 2, adjusted for confounders; Model 3, additionally adjusted for mean daily steps). Grey areas indicate 95% confidence intervals. In Model 3, estimates and the confidence intervals are not shown where they exceed the range of the graphs.
Associations of the number of days achieving specified daily step thresholds with all-cause mortality and CVD, using a common referent (0 days with ≥4000 steps/day)
Table 3 shows the results using a common referent of 0 days with ≥4000 steps/day. When covariates were adjusted (Model 2), modest dose-dependent curvilinear relationships were observed for mortality, less so for CVD. For 1–2 days at thresholds of ≥5000, ≥6000, and ≥7000 steps/day, HRs were 0.67 (95%CI: 0.58, 0.78), 0.62 (0.53, 0.72), and 0.54 (0.46, 0.64) respectively, for all-cause mortality, and 0.78 (0.62, 0.98), 0.81 (0.65, 1.01), and 0.77 (0.61, 0.98) respectively, for CVD. For 3 or more days at these thresholds, HRs were 0.57 (95%CI: 0.50, 0.65), 0.54 (0.47, 0.63), and 0.54 (0.46, 0.63), respectively, for all-cause mortality; 0.69 (0.57, 0.85), 0.67 (0.53, 0.83), and 0.67 (0.53, 0.86), respectively, for CVD. As in previous sets of analyses, additional adjustment for mean daily steps adjusted the associations.
Table 3:
Hazard ratios and 95% confidence intervals of all-cause mortality and cardiovascular disease (CVD) by number of days per week achieving specified daily step thresholds compared against a common reference group
| All-cause mortality | CVD | |||||
|---|---|---|---|---|---|---|
| Cases/N | Model 2 | Model 3 | Cases/N | Model 2 | Model 3 | |
| 0 days with ≥4000 steps/day | 553/ 1639 | 1.00 (reference) | 1.00 (reference) | 193/ 1639 | 1.00 (reference) | 1.00 (reference) |
| ≥1 day with ≥4000 steps/day, but 0 days with ≥5000 steps/day |
247/ 1372 | 0.77 (0.66, 0.89) | 0.85 (0.72, 1.00) | 94/ 1372 | 0.75 (0.58, 0.97) | 0.77 (0.59, 1.01) |
| 1–2 days with ≥5000 steps/day | 335/ 2577 | 0.67 (0.58, 0.78) | 0.80 (0.67, 0.95) | 154/ 2577 | 0.78 (0.62, 0.98) | 0.82 (0.62, 1.08) |
| ≥3 days with ≥5000 steps/day | 630/ 7959 | 0.57 (0.50, 0.65) | 0.82 (0.64, 1.04) | 340/ 7959 | 0.69 (0.57, 0.85) | 0.78 (0.53, 1.13) |
| 0 days with ≥4000 steps/day | 553/ 1639 | 1.00 (reference) | 1.00 (reference) | 193/ 1639 | 1.00 (reference) | 1.00 (reference) |
| ≥1 day with ≥4000 steps/day, but 0 days with ≥6000 steps/day |
472/ 2888 | 0.73 (0.64, 0.83) | 0.84 (0.72, 0.98) | 185/ 2888 | 0.73 (0.59, 0.91) | 0.76 (0.59, 0.98) |
| 1–2 days with ≥6000 steps/day | 318/ 3062 | 0.62 (0.53, 0.72) | 0.79 (0.64, 0.98) | 170/ 3062 | 0.81 (0.65, 1.01) | 0.86 (0.62, 1.18) |
| ≥3 days with ≥6000 steps/day | 422/ 5958 | 0.54 (0.47, 0.63) | 0.84 (0.62, 1.12) | 233/ 5958 | 0.67 (0.53, 0.83) | 0.74 (0.47, 1.16) |
| 0 day with ≥4000 steps/day | 553/ 1639 | 1.00 (reference) | 1.00 (reference) | 193/ 1639 | 1.00 (reference) | 1.00 (reference) |
| ≥1 day with ≥4000 steps/day, but 0 days with ≥7000 steps/day |
670/ 4493 | 0.72 (0.64, 0.81) | 0.84 (0.72, 0.98) | 269/ 4493 | 0.74 (0.61, 0.90) | 0.80 (0.63, 1.03) |
| 1–2 days with ≥7000 steps/day | 258/ 3085 | 0.54 (0.46, 0.64) | 0.72 (0.56, 0.92) | 155/ 3085 | 0.77 (0.61, 0.98) | 0.91 (0.63, 1.30) |
| ≥3 days with ≥7000 steps/day | 284/ 4330 | 0.54 (0.46, 0.63) | 0.84 (0.60, 1.17) | 164/ 4330 | 0.67 (0.53, 0.86) | 0.87 (0.53, 1.43) |
Age was used as the time scale in the Cox proportional hazard models, adjusted for accelerometer wear time (continuous); smoking (never, past, current); alcohol use (rarely, monthly, weekly, daily); intakes of saturated fat, fiber, fruit, and vegetables (quintiles of each); postmenopausal hormone therapy (never, past, current); self-rated health (excellent, very good, good, fair/poor); cancer screening in the past year (yes. no); parental history of myocardial infarction before 60 years of age; and family history of cancer (no or yes for each) in Model 2, and further for mean daily steps (continuous) and the square of mean daily steps (continuous) in Model 3.
Sensitivity analysis
Separate analyses that excluded (i) deaths in the first 2 years of follow-up (n=85) and (ii) women reporting fair or poor general health (n=269) yielded similar results for both outcomes as those from the corresponding primary analyses (Supplemental Table 2). Additional adjustment for BMI did not materially change results (Supplemental Table 3).
Discussion
The present study was undertaken to provide translational data that can inform the next edition of the US Physical Activity Guidelines, planned for 2028. In older women, a very modest level of PA was associated with lower outcome rates: women taking ≥4,000 steps on 1–2 days/week had 26% lower risk of dying and 27% lower risk of CVD, compared to those with no such days, while those taking ≥4,000 steps/day on 3 or more days per week had 40% and 27% lower risk, respectively. There was an inverse, curvilinear dose-response relation for mortality, such that with higher step thresholds (5,000, 6,000 or 7,000), the risk of mortality declined further. For CVD, there was little additional decrement at higher step thresholds. When we adjusted analyses for mean daily steps, previously observed associations were largely attenuated, suggesting that mean steps, not the frequency of meeting daily step thresholds, is important for the inverse relationship.
An important translational implication of these findings is that since step volume is the important driver of the inverse associations, there is no “better” or “best” pattern to take steps; individuals can undertake PA in any preferred pattern (e.g., “slow and steady” vs “bunched patterns”) for lower mortality and CVD risk, at least among older women. These findings provide additional evidence for considering including step metrics in the next PA guidelines, and that “bunching” steps is a viable option for health. We observed different associations for all-cause mortality and CVD. It may be possible that in older women, the “plateau” effect for CVD may occur at a lower step count – between 4,000 and 5,000 steps/day – compared with mortality where the plateau occurred closer to 7,000 steps/day. Further research is needed to clarify differences, as well as possible biological mechanisms underpinning any differences.
Previous studies examining associations between step counts and mortality or CVD have also reported curvilinear trends, with mortality risks plateauing at ~7000–9000 steps/day depending on the population.[10,11,29] However, there has been limited evidence on how one should achieve these step targets—whether on a few days per week or daily—which is relevant for translation into practice. The present study fills this important knowledge gap, suggesting that frequency of meeting daily step thresholds is not critical (even 1–2 days/week of ≥ 4,000 steps/day was related to lower mortality and CVD), and that step volume is more important than the frequency of meeting daily step thresholds in the older population. A recent study using NHANES data similarly investigated daily step thresholds and all-cause mortality.[30] It showed lowered mortality risk at up to 3 days per week of taking ≥8000 steps, compared with none, leveling thereafter, in 3101 participants with a mean age of 50.5 years. However, the study did not examine step thresholds lower than 6000 steps/day, nor the impact of step volume on the associations with pattern. Further research is needed to assess whether in a younger population, similar associations hold for lower step thresholds, as the step volume and mortality association differs by age.[10,11,33]
Other studies have reported that “bunching” of physical activity, not specifically steps, is associated with lower all-cause mortality rates. Investigators have reported beneficial associations with the “weekend warrior” pattern of exercise (i.e., exercise done on only 1–2 days of the week).[12,13] Other investigators, using the “weekend warrior” definition more liberally applied to device assessments, also observed inverse associations between this pattern and all-cause mortality,[34,35] as well as cardiovascular disease,[36] cardiometabolic conditions,[37] and other outcomes.[38–41] Because devices cannot assess whether movements resulted from exercise or non-exercise PA such as transportation or household chores, investigators operationally defined this “weekend warrior” pattern as accumulating more than 50% of recorded moderate-to-vigorous PA on 1–2 days of the week. These studies showed similar findings as with studies using the conventional definition of a “weekend warrior” pattern.
One issue this study does not address is why 4,000 steps/day might represent a lower threshold for reducing mortality risk among older individuals with evidence for increasing benefits at higher doses. One possibility worth further study is that PA levels below approximately 4,000 daily steps initiate a cycle in which lack of adequate PA leads to increased frailty, which in turn leads to less PA. Below some threshold, low levels of PA likely fail to initiate sufficient repair and maintenance mechanisms that slow aging and decrease disease vulnerability.[3]
Limitations
Strengths of this study include a large sample size of older women who tend to be the least active in the population,[42] device-assessed PA, long follow-up with small loss to follow-up, comprehensive adjustment for potential confounders, non-linear survival modeling, and sensitivity analyses to minimize potential reverse causation. However, several limitations exist. Only a single PA assessment was collected, which did not account for the time-varying nature of the behavior. However, a subgroup of the WHS participants had repeated assessments showing good stability of PA over up to 3 years.[43] We did not examine lower step thresholds (e.g., 3,000 steps/day) because we a priori chose 4,000–7,000 steps/day thresholds based on previous evidence in this cohort showing that even ~4,000 steps/day was linked to lower mortality.[29] We did not have information on diet at the time of accelerometry, but only at the time of trial inception. Participants were 62 years and older, primarily white women with higher socioeconomic status, potentially limiting the generalizability of findings. Reverse causation is possible in observational studies - women not achieving minimum step thresholds may have been those inactive most of their lives and be in poor health at the start of follow-up; we did not have a measure of their fitness. However, sensitivity analyses excluding events in the first 2 years and persons reporting fair/poor health yielded similar findings to the main analysis.
Conclusion
In older women with mean age 72 years, achieving ≥4,000 steps/day on just 1–2 days per week was associated with significantly lower all-cause mortality and CVD risks. At higher step thresholds, mortality risk further declined with modest additional decrements. The associations attenuated to the null when adjusted for mean daily steps, indicating that step volume is important for the inverse associations, not the number of days meeting a daily step threshold. These findings can inform future PA guidelines, as well as clinical and public health practice.
Supplementary Material
WHAT IS ALREADY KNOWN ON THIS TOPIC
While step counts are inversely related to mortality rates and cardiovascular disease (CVD) incidence, it is unclear how many days per week an individual should achieve their step goal to achieve health benefits.
WHAT THIS STUDY ADDS
Among older US women, achieving ≥4,000 steps/day on 1–2 days per week, compared to 0 days, was associated with 26% lower all-cause mortality risk and 27% lower CVD risk, while achieving this on ≥3 days was associated with 40% and 27% lower risk, respectively.
At higher step thresholds (5,000 to 7,000), risks further declined modestly for mortality but leveled for CVD (e.g., achieving ≥7,000 steps/day on ≥3 days per week compared to 0 days: mortality risk reduction, 32%; CVD risk reduction, 16%).
These associations attenuated to the null when adjusted for daily steps, indicating the number of steps per day, rather than the frequency of days/week achieving a particular step threshold, is important for the inverse associations with mortality and CVD in older women.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE, OR POLICY
Physical activity guidelines in older women should consider recommending at least ≥4,000 steps/day on even 1–2 days per week to lower mortality and CVD risk.
The number of steps is important for the inverse associations with mortality and CVD, and not the frequency of days/week achieving a particular step threshold.
Acknowledgements:
We thank WHS participants for their contributions to this study.
Funding:
This research was supported by grants from the National Institutes of Health (NIH; CA154647, CA047988, CA182913, HL043851, HL080467, and HL099355). The study was also supported by NIH 5R01CA227122: National Cancer Institute, Office of the Director, Office of Disease Prevention, and Office of Behavioral and Social Sciences Research, and was supported in part by the extramural research program at the National Heart, Lung, and Blood Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Footnotes
Competing interests: None.
Ethical approval: The study was approved by the Brigham and Women’s Hospital’s Institutional Review Board committee (IRB@mgb.org).
Patient involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research..
Data sharing statement:
Data described in the manuscript, code book, and analytic code will be made available upon request pending application and approval (contact: ilee@rics.bwh.harvard.edu).
References:
- 1.Tudor-Locke C, Schuna JM, Barreira TV, et al. Normative steps/day values for older adults: NHANES 2005–2006. J Gerontol A Biol Sci Med Sci 2013;68:1426–32. doi: 10.1093/gerona/glt116 [DOI] [PubMed] [Google Scholar]
- 2.Raichlen DA, Lieberman DE. The evolution of human step counts and its association with the risk of chronic disease. Curr Biol 2022;32:R1206–14. doi: 10.1016/j.cub.2022.09.030 [DOI] [PubMed] [Google Scholar]
- 3.Lieberman DE, Kistner TM, Richard D, et al. The active grandparent hypothesis: Physical activity and the evolution of extended human healthspans and lifespans. Proc Natl Acad Sci U S A. 2021;118:e2107621118. doi: 10.1073/pnas.2107621118 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.2018 Physical Activity Guidelines Advisory Committee. 2018 Physical Activity Guidelines Advisory Committee Scientific ReportTrends in Adherence to the Physical Activity Guidelines for Americans for Aerobic Activity and Time Spent on Sedentary Behavior Among US Adults, 2007 to 2016. 2018;779. doi: https://health.gov/sites/default/files/2019-09/PAG_Advisory_Committee_Report.pdf [Google Scholar]
- 5.Piercy KL, Troiano RP, Ballard RM, et al. The Physical Activity Guidelines for Americans. JAMA. 2018;320:2020–8. doi: 10.1001/jama.2018.14854 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gurven MD, Lieberman DE. WEIRD bodies: mismatch, medicine and missing diversity. Evol Hum Behav 2020;41:330–40. doi: 10.1016/j.evolhumbehav.2020.04.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Strain T, Flaxman S, Guthold R, et al. National, regional, and global trends in insufficient physical activity among adults from 2000 to 2022: a pooled analysis of 507 population-based surveys with 5·7 million participants. The Lancet Global Health. 2024;12:e1232–43. doi: 10.1016/S2214-109X(24)00150-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Du Y, Liu B, Sun Y, et al. Trends in Adherence to the Physical Activity Guidelines for Americans for Aerobic Activity and Time Spent on Sedentary Behavior Among US Adults, 2007 to 2016. JAMA Netw Open. 2019;2:e197597. doi: 10.1001/jamanetworkopen.2019.7597 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Laricchia F Wearables - statistics & facts. Statista. 2024. https://www.statista.com/topics/1556/wearable-technology/ (accessed 11 January 2025)
- 10.Paluch AE, Bajpai S, Bassett DR, et al. Daily steps and all-cause mortality: a meta-analysis of 15 international cohorts. The Lancet Public Health. 2022;7:e219–28. doi: 10.1016/S2468-2667(21)00302-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Paluch AE, Bajpai S, Ballin M, et al. Prospective association of daily steps with cardiovascular disease: A harmonized meta-analysis. Circulation. 2023;147:122–31. doi: 10.1161/CIRCULATIONAHA.122.061288 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.O’Donovan G, Lee I-M, Hamer M, et al. Association of “weekend warrior” and other leisure time physical activity patterns with risks for all-cause, cardiovascular disease, and cancer mortality. JAMA Intern Med 2017;177:335–42. doi: 10.1001/jamainternmed.2016.8014 [DOI] [PubMed] [Google Scholar]
- 13.Lee I-M, Sesso HD, Oguma Y, et al. The “weekend warrior” and risk of mortality. Am J Epidemiol 2004;160:636–41. doi: 10.1093/aje/kwh274 [DOI] [PubMed] [Google Scholar]
- 14.Hamaya R, Shiroma EJ, Moore CC, et al. Time-vs step-based physical activity metrics for health. JAMA Intern Med 2024;e240892. doi: 10.1001/jamainternmed.2024.0892 [DOI] [PMC free article] [PubMed]
- 15.Lee I-M, Shiroma EJ, Kamada M, et al. Association of step volume and intensity with all-cause mortality in older women. JAMA Intern Med 2019;179:1105. doi: 10.1001/jamainternmed.2019.0899 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lee I-M, Cook NR, Gaziano JM, et al. Vitamin E in the primary prevention of cardiovascular disease and cancer: the Women’s Health Study: a randomized controlled trial. JAMA. 2005;294:56–65. doi: 10.1001/jama.294.1.56 [DOI] [PubMed] [Google Scholar]
- 17.Ridker PM, Cook NR, Lee I-M, et al. A randomized trial of low-dose aspirin in the primary prevention of cardiovascular disease in women. N Engl J Med 2005;352:1293–304. doi: 10.1056/NEJMoa050613 [DOI] [PubMed] [Google Scholar]
- 18.Jordan AN, Jurca GM, Locke CT, et al. Pedometer indices for weekly physical activity recommendations in postmenopausal women. Med Sci Sports Exerc 2005;37:1627–32. doi: 10.1249/01.mss.0000177455.58960.aa [DOI] [PubMed] [Google Scholar]
- 19.Lee I-M, Shiroma EJ. Using accelerometers to measure physical activity in large-scale epidemiologic studies: issues and challenges. Br J Sports Med 2014;48:197–201. doi: 10.1136/bjsports-2013-093154 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lee I-M, Shiroma EJ, Evenson KR, et al. Using devices to assess physical activity and sedentary behavior in a large cohort study, the Women’s Health Study. J Meas Phys Behav 2018;1:60–9. doi: 10.1123/jmpb.2018-0005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Tudor-Locke C, Camhi SM, Troiano RP. A catalog of rules, variables, and definitions applied to accelerometer data in the National Health and Nutrition Examination Survey, 2003–2006. Prev Chronic Dis 2012;9:E113. doi: 10.5888/pcd9.110332 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Evenson KR, Bellettiere J, Cuthbertson CC, et al. Cohort profile: the Women’s Health Accelerometry Collaboration. BMJ Open. 2021;11:e052038. doi: 10.1136/bmjopen-2021-052038 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Choi L, Ward SC, Schnelle JF, et al. Assessment of wear/nonwear time classification algorithms for triaxial accelerometer. Med Sci Sports Exerc 2012;44:2009–16. doi: 10.1249/MSS.0b013e318258cb36 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Choi L, Ward SC, Schnelle JF, et al. Assessment of wear/nonwear time classification algorithms for triaxial accelerometer. Med Sci Sports Exerc 2012;44:2009–16. doi: 10.1249/MSS.0b013e318258cb36 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Choi L, Liu Z, Matthews CE, et al. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc 2011;43:357–64. doi: 10.1249/MSS.0b013e3181ed61a3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ngueleu A-M, Barthod C, Best KL, et al. Criterion validity of ActiGraph monitoring devices for step counting and distance measurement in adults and older adults: a systematic review. J Neuroeng Rehabil 2022;19:112. doi: 10.1186/s12984-022-01085-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Thygesen K, Alpert JS, Jaffe AS, et al. Third Universal Definition of Myocardial Infarction. Journal of the American College of Cardiology. 2012;60:1581–98. doi: 10.1016/j.jacc.2012.08.001 [DOI] [PubMed] [Google Scholar]
- 28.Adams HP, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke. 1993;24:35–41. doi: 10.1161/01.str.24.1.35 [DOI] [PubMed] [Google Scholar]
- 29.Lee I-M, Shiroma EJ, Kamada M, et al. Association of Step Volume and Intensity With All-Cause Mortality in Older Women. JAMA Internal Medicine. 2019;179:1105–12. doi: 10.1001/jamainternmed.2019.0899 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Inoue K, Tsugawa Y, Mayeda ER, et al. Association of daily step patterns with mortality in US adults. JAMA Netw Open. 2023;6:e235174. doi: 10.1001/jamanetworkopen.2023.5174 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Matthews CE, Troiano RP, Salerno EA, et al. Exploration of Confounding Due to Poor Health in an Accelerometer-Mortality Study. Med Sci Sports Exerc 2020;52:2546–53. doi: 10.1249/MSS.0000000000002405 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tang F, Ishwaran H. Random Forest Missing Data Algorithms. Stat Anal Data Min 2017;10:363–77. doi: 10.1002/sam.11348 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Lee I-M, Keadle SK, Matthews CE. Fitness trackers to guide advice on activity prescription. JAMA. 2023;330:1733–4. doi: 10.1001/jama.2023.19332 [DOI] [PubMed] [Google Scholar]
- 34.Evenson KR, Herring AH, Wen F. Accelerometry-assessed latent class patterns of physical activity and sedentary behavior with mortality. Am J Prev Med 2017;52:135–43. doi: 10.1016/j.amepre.2016.10.033 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Shiroma EJ, Lee I-M, Schepps MA, et al. Physical activity patterns and mortality: the weekend warrior and activity bouts. Med Sci Sports Exerc 2019;51:35–40. doi: 10.1249/MSS.0000000000001762 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Khurshid S, Al-Alusi MA, Churchill TW, et al. Accelerometer-derived “weekend warrior” physical activity and incident cardiovascular disease. JAMA. 2023;330:247–52. doi: 10.1001/jama.2023.10875 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Kany S, Al-Alusi MA, Rämö JT, et al. Associations of “weekend warrior” physical activity with incident disease and cardiometabolic health. Circulation. 2024;150:1236–47. doi: 10.1161/CIRCULATIONAHA.124.068669 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Ning Y, Chen M, An J, et al. Association between “weekend warrior” physical activity and the incidence of neurodegenerative diseases. Neurotherapeutics. 2024;21:e00430. doi: 10.1016/j.neurot.2024.e00430 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Gubelmann C, Heinzer R, Haba-Rubio J, et al. Physical activity is associated with higher sleep efficiency in the general population: the CoLaus study. Sleep. 2018;41. doi: 10.1093/sleep/zsy070 [DOI] [PubMed] [Google Scholar]
- 40.Vandercappellen EJ, Henry RMA, Savelberg HHCM, et al. Association of the amount and pattern of physical activity with arterial stiffness: The Maastricht Study. J Am Heart Assoc 2020;9:e017502. doi: 10.1161/JAHA.120.017502 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Min J, Cao Z, Duan T, et al. Accelerometer-derived “weekend warrior” physical activity pattern and brain health. Nat Aging. 2024;4:1394–402. doi: 10.1038/s43587-024-00688-y [DOI] [PubMed] [Google Scholar]
- 42.Jones H, Millward P, Buraimo B. Adult participation in sport: analysis of the Taking Part Survey. Department for Culture, Media and Sport. 2011. https://assets.publishing.service.gov.uk/media/5a7b483bed915d3ed90635e6/tp-adult-participation-sport-analysis.pdf
- 43.Keadle SK, Shiroma EJ, Kamada M, et al. Reproducibility of accelerometer-assessed physical activity and sedentary time. Am J Prev Med 2017;52:541–8. doi: 10.1016/j.amepre.2016.11.010 [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
Data Availability Statement
Data described in the manuscript, code book, and analytic code will be made available upon request pending application and approval (contact: ilee@rics.bwh.harvard.edu).