Abstract：
OBJECTIVE: Previous findings on the association between sleep duration, changes in sleep duration, and long-term dementia risk were mixed. Thus, we aimed to investigate the association between midlife sleep duration, its change, and dementia.
METHODS: We recruited 41,731 Japanese (40-71 years) and documented their habitual sleep duration at baseline (1990-1994) and a 5-year follow-up survey. Changes in sleep duration were calculated as differences between baseline and 5-year measurements. We identified dementia using the Long-Term Care Insurance system (2007-2016). Hazard ratios (HRs) and 95% confidence intervals (CIs) of dementia were calculated using the area-stratified Cox model.
RESULTS: During 360,389 person-years, 4621 participants exhibited dementia. The multivariable HRs of dementia compared with 7 h of sleep were 1.13 (95% CI: 0.98-1.30) for 3-5 h, 0.93 (0.85-1.02) for 6 h, 1.06 (0.99-1.14) for 8 h, 1.13 (1.01-1.27) for 9 h, and 1.40 (1.21-1.63) for 10-12 h with a J-shaped fashion (p for linear < 0.001 and quadratic < 0.001). For its change, the HRs compared with no change were 1.02 (0.90-1.16) for decreased ≥2 h, 0.95 (0.88-1.03) for decreased 1 h, 1.00 (0.91-1.09) for increased 1 h, and 1.37 (1.20-1.58) for increased ≥2 h. The positive association for decreased sleep duration was observed in individuals with an initial sleep duration of ≤7 h, but not in those with ≥8 h (p for interaction = 0.007).
CONCLUSIONS: Long and increased sleep duration was associated with a higher risk of dementia.
METHODS: We recruited 41,731 Japanese (40-71 years) and documented their habitual sleep duration at baseline (1990-1994) and a 5-year follow-up survey. Changes in sleep duration were calculated as differences between baseline and 5-year measurements. We identified dementia using the Long-Term Care Insurance system (2007-2016). Hazard ratios (HRs) and 95% confidence intervals (CIs) of dementia were calculated using the area-stratified Cox model.
RESULTS: During 360,389 person-years, 4621 participants exhibited dementia. The multivariable HRs of dementia compared with 7 h of sleep were 1.13 (95% CI: 0.98-1.30) for 3-5 h, 0.93 (0.85-1.02) for 6 h, 1.06 (0.99-1.14) for 8 h, 1.13 (1.01-1.27) for 9 h, and 1.40 (1.21-1.63) for 10-12 h with a J-shaped fashion (p for linear < 0.001 and quadratic < 0.001). For its change, the HRs compared with no change were 1.02 (0.90-1.16) for decreased ≥2 h, 0.95 (0.88-1.03) for decreased 1 h, 1.00 (0.91-1.09) for increased 1 h, and 1.37 (1.20-1.58) for increased ≥2 h. The positive association for decreased sleep duration was observed in individuals with an initial sleep duration of ≤7 h, but not in those with ≥8 h (p for interaction = 0.007).
CONCLUSIONS: Long and increased sleep duration was associated with a higher risk of dementia.
摘要：
目的：先前关于睡眠持续时间之间关联的发现，睡眠持续时间的变化，和长期痴呆的风险是混合的。因此，我们的目的是调查中年睡眠时间之间的关系，它的变化,和痴呆症。
方法：我们招募了41,731名日本人（40-71岁），并记录了他们在基线（1990-1994）的习惯性睡眠时间和5年的随访调查。睡眠持续时间的变化被计算为基线和5年测量之间的差异。我们使用长期护理保险制度（2007-2016年）确定了痴呆症。使用面积分层Cox模型计算痴呆的危险比（HR）和95％置信区间（CI）。
结果：在360,389人年中，4621名参与者表现出痴呆症。3-5h与睡眠7h相比，痴呆的多变量HR为1.13（95％CI：0.98-1.30），0.93（0.85-1.02）持续6小时，1.06(0.99-1.14)持续8小时，1.13(1.01-1.27)持续9小时,和1.40（1.21-1.63），持续10-12小时，呈J形（p为线性<0.001和二次<0.001）。对于它的变化，与无变化相比，降低≥2小时的HR为1.02（0.90-1.16），下降1小时0.95(0.88-1.03)，增加1小时为1.00(0.91-1.09),和1.37（1.20-1.58）增加≥2小时。在初始睡眠持续时间≤7小时的个体中观察到睡眠持续时间减少的正相关，但在≥8小时的人群中并非如此(相互作用的p=0.007)。
结论：长且睡眠时间延长与痴呆风险增加相关。
方法：我们招募了41,731名日本人（40-71岁），并记录了他们在基线（1990-1994）的习惯性睡眠时间和5年的随访调查。睡眠持续时间的变化被计算为基线和5年测量之间的差异。我们使用长期护理保险制度（2007-2016年）确定了痴呆症。使用面积分层Cox模型计算痴呆的危险比（HR）和95％置信区间（CI）。
结果：在360,389人年中，4621名参与者表现出痴呆症。3-5h与睡眠7h相比，痴呆的多变量HR为1.13（95％CI：0.98-1.30），0.93（0.85-1.02）持续6小时，1.06(0.99-1.14)持续8小时，1.13(1.01-1.27)持续9小时,和1.40（1.21-1.63），持续10-12小时，呈J形（p为线性<0.001和二次<0.001）。对于它的变化，与无变化相比，降低≥2小时的HR为1.02（0.90-1.16），下降1小时0.95(0.88-1.03)，增加1小时为1.00(0.91-1.09),和1.37（1.20-1.58）增加≥2小时。在初始睡眠持续时间≤7小时的个体中观察到睡眠持续时间减少的正相关，但在≥8小时的人群中并非如此(相互作用的p=0.007)。
结论：长且睡眠时间延长与痴呆风险增加相关。
