Both insufficient moderate-to-vigorous intensity physical activity (MVPA) and high volumes of sedentary time (ST) have been associated with increased risk of all-cause mortality, cardiovascular disease (CVD), and some cancers.¹ However, epidemiological evidence has relied mostly on self-reported physical activity (PA) and ST measures, which are prone to reporting bias and measurement error. Cohort studies incorporating more objective assessments are emerging,^2–4 but few investigate relationships of accelerometer-measured PA and ST with clinical end points, particularly incident CVD, and include both men and women.⁵ In addition, the relevance of light intensity PA (LIPA) in this context, which is an important contributor to total PA and may be a more feasible behavioral target for middle to older aged adults, remains unclear. Here, we examine the prospective associations of accelerometer-measured PA and ST with incident CVD (primary outcome), incident cancer, and all-cause mortality.

Participants were from the European Prospective Investigation Into Cancer and Nutrition–Norfolk study, a population-based cohort of 25 639 adults 40 to 79 years of age residing in Norfolk, UK. Between 2004 and 2016, 7820 participants attending in-clinic assessments wore an accelerometer (ActiGraph GT1M/GT3X+) on their right hip for 7 days during waking hours. We excluded those with <4 days of sufficient wear time (>10 hours/day; n=134) or missing covariate data (n=19). Those with prevalent CVD (n=2083), cancer (n=629), or either CVD or cancer (n=2488) were excluded from the CVD, cancer, and all-cause mortality analyses, respectively.

Cox proportional hazards regression was used to model nonlinear associations (using splines) of total PA volume (average accelerometer counts per minute/day), MVPA (minutes/day; ≥2020 counts/min), LIPA (hours/day; 100–2019 counts/min), and ST (hours/day; <100 counts/min) with total (nonfatal or fatal) incident CVD (International Classification of Diseases, 9th Revision codes 401–448 or 10th Revision codes I10–I79), cancer (International Classification of Diseases, 9th Revision codes 140–280 or 10th Revision codes C00–C97), and all-cause mortality. Outcome status was ascertained via hospital admissions and death certificates through March 31, 2016, for incident CVD and cancer and through March 31, 2018, for all-cause mortality. The Norwich District Ethics Committee provided ethics approval, and all participants gave written informed consent.

Key findings are summarized in the Figure. A total of 1106 incident CVD, 516 incident cancer, and 339 all-cause mortality events occurred for 5585, 7052, and 5249 participants, respectively. Mean age at baseline was 68.9 years (SD, 7.2 years); mean body mass index was 26.5 kg/m² (SD, 4.1 kg/m²); and 58.5% were female.

Figure. Baseline exposure distribution and hazard ratios (HR; 95% confidence intervals) for incident cardiovascular disease (CVD), incident cancer, and all-cause mortality by physical activity (PA) and sedentary time (ST). Participants with a baseline history of stroke/myocardial infarction (incident CVD models), cancer (incident cancer models), or either CVD or cancer (all-cause mortality models) were excluded. Model 1 is adjusted for sex and accelerometer wear time (with age as the underlying time scale). Model 2 is adjusted as for model 1 plus education level; social class; smoking status; alcohol intake; baseline history of diabetes mellitus or taking diabetes mellitus medications; taking medication for hypertension/dyslipidemia/depression; and family history of CVD (stroke/myocardial infarction), diabetes mellitus, or cancer. In model 2, analyses of moderate-to-vigorous intensity PA (MVPA) and light intensity PA (LIPA) are mutually adjusted, whereas analyses of ST are adjusted for MVPA. Data were fitted with the use of restricted cubic splines (3 evenly spaced knots), and results are shown between the 1st and 99th percentiles of the relevant exposure. Reference values chosen for each exposure approximated the 10th percentile (total PA=120 average counts per minute (cpm)/day; MVPA=0 minutes/day; LIPA=3 hours/day; ST=8 hours/day). Covariates that violated the proportional hazard assumptions were included as baseline strata. Time to event was defined as the time from exposure baseline to the first event or end of follow-up.

After adjustment for sex and wear time (model 1, Figure), higher levels of total PA and MVPA were associated with lower incident CVD risk in a nonlinear manner; after an initially steeper decrease in hazard ratios, there was a flattening of the relationships. Associations persisted for total PA and MVPA (model 2) after adjustment for relevant covariates and ST (in MVPA models). Associations with LIPA were attenuated after covariate and MVPA adjustment (model 2), although levels <3 hours/day were still associated with higher risk. For ST, positive associations with incident CVD were observed, with a steeper relationship beyond 11 hours/day (model 1); however, this association was attenuated after adjustments for covariates and MVPA (model 2).

For incident cancer and all-cause mortality, associations generally followed directions similar to those for incident CVD (Figure). However, associations for total PA and MVPA were less consistent and tended toward the null after an initial steep decrease in hazard ratios. Consistently strong and approximately linear associations were observed for LIPA (inverse) and ST (positive), with only slight attenuation after adjustment for covariates and MVPA.

Overall, hazard ratios were not meaningfully altered in further modeling that adjusted for potential mediators or confounders (waist circumference or body mass index and diet quality). Sensitivity analyses that excluded early cases or deaths within 2 years of follow-up (data not shown) resulted in some attenuation of associations in fully adjusted models, mostly for incident CVD and at the more extreme ends of each exposure. Attenuations in these models seem to be attributable largely to reduced sample size and events considering the widening of the confidence intervals, but some reverse causality bias, among other unknown factors, may also be at play.

Other potential limitations worth noting include the single time point measure of PA and the use of uniaxial rather than triaxial accelerometry. However, this study provides important contributions to the paucity of literature^2–5 on accelerometer-measured PA and ST and clinical health outcomes, particularly for incident CVD, in a cohort of middle to older aged women and men. These results generally support current PA recommendations focusing on increasing total PA and MVPA to reduce the risk for chronic disease and early death, with some variation across exposures and between outcomes in terms of dose–response relationships, likely a result of a lack of statistical power at higher activity levels. These results also provide some endorsement for reducing high volumes of ST and increasing LIPA to further augment risk reduction, particularly for incident cancer and all-cause mortality. Findings may be especially relevant to middle to older aged adults, given their typically higher volumes of ST and poorer adherence to MVPA guidelines. Building on this evidence base remains a key priority for informing more specific and targeted PA guidelines, interventions, and clinical or public health action.

The authors would like to acknowledge and thank the physical activity measurement team (Kate Westgate, Stefanie Hollidge, and Lewis Griffiths) for their role in data processing and data management support, along with the entire EPIC-Norfolk Study team for their role in data collection. We would also like to thank all of the EPIC-Norfolk Study participants for making this work possible.

The EPIC-Norfolk study (DOI 10.22025/2019.10.105.00004) has received funding from the UK Medical Research Council (MR/N003284/1), Cancer Research UK (C864/A14136), and the NIHR Biomedical Research Centre in Cambridge (IS-BRC-1215-20014). Dr Dempsey is supported by a National Health and Medical Research Council of Australia research fellowship (No. 1142685). Drs Dempsey, Strain, Brage, Wijndaele, and Wareham are supported by the UK Medical Research Council (grants MC_UU_12015/1 and MC_UU_12015/3).

None.

https://www.ahajournals.org/journal/circ

Data availability/sharing: Participant summary statistics are available at https://doi.org/10.17863/CAM.48931. Individual-level data are available from the European Prospective Investigation Into Cancer and Nutrition–Norfolk Management Committee (contact via [email protected]cam.ac.uk) for researchers who meet the criteria for access to confidential data.

中高强度体力活动 (MVPA) 不足和久坐时间 (ST) 都与全因死亡率、心血管疾病 (CVD) 和某些癌症的风险增加有关。¹然而，流行病学证据主要依赖于自我报告的身体活动 (PA) 和 ST 测量，这些测量容易出现报告偏倚和测量错误。纳入更客观评估的队列研究正在出现，^2-4但很少研究加速度计测量的 PA 和 ST 与临床终点的关系，特别是心血管疾病事件，包括男性和女性。⁵此外，在这种情况下，光强度 PA (LIPA) 的相关性仍不清楚，它是总 PA 的重要贡献者，可能是中老年人更可行的行为目标。在这里，我们研究了加速度计测量的 PA 和 ST 与 CVD 事件（主要结果）、癌症事件和全因死亡率之间的前瞻性关联。

参与者来自欧洲癌症和营养前瞻性调查——诺福克研究，这是一个基于人群的队列，由居住在英国诺福克的 25 639 名 40 至 79 岁的成年人组成。在 2004 年至 2016 年期间，7820 名参加临床评估的参与者在醒着的时候在右臀部佩戴了加速度计 (ActiGraph GT1M/GT3X+) 7 天。我们排除了那些具有 <4 天足够佩戴时间（>10 小时/天；n=134）或缺少协变量数据（n=19）的人。CVD、癌症和全因死亡率分析分别排除了患有普遍 CVD (n=2083)、癌症 (n=629) 或 CVD 或癌症 (n=2488) 的人。

Cox 比例风险回归用于模拟总 PA 体积（每分钟/天的平均加速度计计数）、MVPA（分钟/天；≥2020 计数/分钟）、LIPA（小时/天；100–2019）的非线性关联（使用样条）计数/分钟）和 ST（小时/天；<100 计数/分钟）与总（非致命或致命）事件 CVD（国际疾病分类，第 9 版代码 401-448 或第 10 版代码 I10-I79），癌症（国际疾病分类，第 9 版代码 140–280 或第 10 版代码 C00–C97) 和全因死亡率。结果状态通过截至 2016 年 3 月 31 日的 CVD 和癌症事件以及截至 2018 年 3 月 31 日的全因死亡率的住院和死亡证明来确定。诺里奇区伦理委员会提供了伦理批准，所有参与者都签署了书面知情同意书。

图中总结了主要发现。分别有 5585、7052 和 5249 名参与者发生了 1106 起 CVD、516 起癌症和 339 起全因死亡事件。基线平均年龄为 68.9 岁（SD，7.2 岁）；平均体重指数为 26.5 kg/m ² (SD, 4.1 kg/m ² )；58.5% 是女性。

数字。 按体力活动 (PA) 和久坐时间 (ST) 划分的心血管疾病 (CVD)、癌症和全因死亡率的基线暴露分布和风险比 (HR；95% 置信区间)。具有中风/心肌梗塞（事件 CVD 模型）、癌症（事件癌症模型）或 CVD 或癌症（全因死亡率模型）基线病史的参与者被排除在外。模型 1 针对性别和加速度计佩戴时间进行了调整（以年龄为基础时间尺度）。模型 2 调整为模型 1 加教育程度；社会阶层; 吸烟状况；酒精摄入量；糖尿病基线史或服用糖尿病药物；服用治疗高血压/血脂异常/抑郁症的药物；心血管疾病（中风/心肌梗塞）、糖尿病或癌症家族史。在模型 2 中，中高强度 PA (MVPA) 和光强度 PA (LIPA) 的分析相互调整，而 ST 分析针对 MVPA 进行调整。使用受限三次样条（3 个均匀间隔的节点）对数据进行拟合，结果显示在相关暴露的第 1 个和第 99 个百分位数之间。为每次暴露选择的参考值接近第 10 个百分位数（总 PA=120 平均每分钟 (cpm)/天；MVPA=0 分钟/天；LIPA=3 小时/天；ST=8 小时/天）。违反比例风险假设的协变量被列为基线层。事件发生时间定义为从暴露基线到第一次事件或随访结束的时间。违反比例风险假设的协变量被列为基线层。事件发生时间定义为从暴露基线到第一次事件或随访结束的时间。违反比例风险假设的协变量被列为基线层。事件发生时间定义为从暴露基线到第一次事件或随访结束的时间。

在对性别和穿着时间进行调整后（模型 1，图），较高水平的总 PA 和 MVPA 与较低的 CVD 风险以非线性方式相关；在风险比开始急剧下降之后，关系变得平坦。在调整相关协变量和 ST（在 MVPA 模型中）后，总 PA 和 MVPA（模型 2）的关联仍然存在。协变量和 MVPA 调整（模型 2）后与 LIPA 的关联减弱，尽管水平 <3 小时/天仍与较高的风险相关。对于 ST，观察到与事件 CVD 呈正相关，超过 11 小时/天的关系更为陡峭（模型 1）；然而，在调整协变量和 MVPA（模型 2）后，这种关联减弱了。

对于偶发癌症和全因死亡率，关联通常遵循与偶发 CVD 相似的方向（图）。然而，总 PA 和 MVPA 的关联不太一致，并且在风险比最初急剧下降后趋于无效。对于 LIPA（反向）和 ST（正），观察到一致强且近似线性的关联，在调整协变量和 MVPA 后仅略​​有衰减。

总体而言，在针对潜在介质或混杂因素（腰围或体重指数和饮食质量）进行调整的进一步建模中，风险比没有有意义地改变。排除随访 2 年内早期病例或死亡的敏感性分析（数据未显示）导致完全调整模型中的关联有所减弱，主要是针对 CVD 事件和每次暴露的更极端的情况。这些模型中的衰减似乎主要归因于样本量减少和考虑到置信区间扩大的事件，但一些反向因果偏差以及其他未知因素也可能在起作用。

其他值得注意的潜在限制包括 PA 的单时间点测量和使用单轴而不是三轴加速度计。然而，这项研究为文献的匮乏做出了重要贡献^2-5关于加速度计测量的 PA 和 ST 以及临床健康结果，特别是对于中老年女性和男性的心血管疾病事件。这些结果通常支持当前 PA 建议，重点是增加总 PA 和 MVPA 以降低慢性病和早期死亡的风险，在剂量-反应关系方面，不同暴露和结果之间存在一些差异，这可能是缺乏统计能力的结果在更高的活动水平。这些结果也为减少大量 ST 和增加 LIPA 以进一步降低风险提供了一些支持，特别是对于癌症和全因死亡率。鉴于中老年人的 ST 量通常较高且对 MVPA 指南的依从性较差，因此研究结果可能与中老年人特别相关。

作者要感谢身体活动测量团队（Kate Westgate、Stefanie Hollidge 和 Lewis Griffiths）在数据处理和数据管理支持方面的作用，以及整个 EPIC-Norfolk 研究团队在数据收集中的作用. 我们还要感谢 EPIC-Norfolk 研究的所有参与者使这项工作成为可能。

EPIC-诺福克研究 (DOI 10.22025/2019.10.105.00004) 已获得英国医学研究委员会 (MR/N003284/1)、英国癌症研究中心 (C864/A14136) 和剑桥 NIHR 生物医学研究中心 (IS- BRC-1215-20014)。Dempsey 博士得到了澳大利亚国家健康和医学研究委员会研究奖学金（编号 1142685）的支持。Drs Dempsey、Strain、Brage、Wijndaele 和 Wareham 得到英国医学研究委员会的支持（赠款 MC_UU_12015/1 和 MC_UU_12015/3）。

没有任何。

https://www.ahajournals.org/journal/circ

数据可用性/共享：参与者汇总统计数据可在 https://doi.org/10.17863/CAM.48931 获得。个人层面的数据可从欧洲癌症和营养前瞻性调查 - 诺福克管理委员会（通过[email protected] cam. ac. uk联系）提供给符合访问机密数据标准的研究人员。