追加：补觉与夜班两条 · 核实记录（2026-09-21）
任务来源：用户问「熬夜呢，有没有关于熬夜的内容」。
原有覆盖：熬夜本身已有六条——第 2 节第 13 条（每晚睡 7 小时左右，作息固定）、第 3 节第 2 条（固定起床时间）、第 3 节第 3 条（每晚睡够 7 到 8 小时）、第 3 节第 8 条（睡前别读发光电子书）、第 3 节第 9 条（到点就睡）、第 3 节第 10 条（深夜不做重大决定）。缺的两块：① 已经熬了之后怎么办，全书一条没有；② 上夜班、三班倒，零覆盖。
落点：第 2 节第 40、41 条，追加在节末不动条号。另把社会时差那篇补进第 2 节第 13 条的收益栏，第 3 节第 2 条备注加互指。
一次被用户推翻的判断（记下来）
初次汇报时我推荐只写补觉、不写夜班，理由是夜班「只有风险没有出路」，援引的是 2026-09-18 那轮「只给禁止不给出路」的排查口径。用户反驳：「但是我觉得读者可以选择上不上啊，有可能有些为了健康就换工作呢，总比没有好」。这个反驳成立，我的判断是把那条规矩用过头了：换工作、申请调岗、挑职业时把账算进去都是真实选择，而 23 项队列的剂量反应（每多干 5 年 +7%）恰好是「要转早转」的量化依据。本书本来就有「只拆账不给结论」的写法（第 10 节结婚、第 18 节养孩子），夜班照这个写即可。教训：「没有出路」这个判断，要先确认当事人是不是真的没有选择，不能因为选择代价高就当成没有。
文献
| DOI | 复核 | 关键数字原文 |
|---|---|---|
| https://doi.org/10.1038/s41467-026-72461-1（第 40 条主证据） | 是（Li X, Zhang M, Li Z, Zhang S, Bertisch SM, Huang T, Rutter MK, Redline S. Nat Commun 2026，PMID 42045198，PMC13121452 全文 XML 逐字抓取） | 摘要：「In this prospective cohort study of 85,618 UK Biobank participants (mean [standard deviation] age 61.8 [7.8] years), we analyze accelerometer-derived sleep data and identify five day-to-day sleep restriction-rebound patterns」「Results are replicated in an independent cohort from the National Health and Nutrition Examination Survey (N = 4,586)」。正文：「In the fully adjusted Model 3, both SR without rebound and severe SR without rebound were significantly associated with higher mortality compared to regular sleep in the overall sample (hazard ratio [HR] = 1.15, 95% confidence interval [CI] = 1.01–1.31; HR = 1.42, 95% CI = 1.24–1.63, respectively). When sleep rebound occurred, neither SR nor severe SR showed significant associations with mortality, although the point estimates were above 1.0. The multivariable adjusted HRs for all-cause mortality were 1.12 (95% CI = 0.98–1.28) for participants experienc[ing SR with rebound]」「Among participants with short baseline sleep duration, both SR and severe SR without rebound were significantly associated with higher mortality risk compared to regular sleep (HR = 1.19, 95% CI = 1.01–1.40; HR = 1.38, 95% CI = 1.17–1.63, respectively)」 |
| 同上（定义，决定了条目标题怎么写） | 是 | 「Restriction nights were defined as nights with sleep duration at least 2.5 h below individual sleep need. Post-restriction night was defined as the first night following each restriction period.」「Each post-restriction period comprised only one post-restriction night.」「Sleep rebound was defined as the average extra sleep more than 0 h relative to individual sleep need across post-restriction periods.」「SR was defined as the average cumulative sleep loss of 2.5–3.5 h relative to individual sleep need across restriction periods. Severe SR was defined as average cumulative sleep loss exceeding 3.5 h.」 |
| https://doi.org/10.3389/fpubh.2025.1668848（第 41 条主证据） | 是（Xi J, Ma W, Tao Y, Zhang X, Liu L, Wang H. Front Public Health 2025，PMID 41069800，PROSPERO CRD420251060086） | 「Twenty-three cohort studies were included.」「night shift work significantly increased the risk of total CVD events (RR = 1.13, 95% CI = 1.10-1.16) and total CVD mortality (RR = 1.27, 95%CI = 1.18-1.36)」「each 5-year increment in shift work duration was associated with a 7% higher risk of CVD incidence (RR = 1.07, 95% CI: 1.04-1.09) and a 4% increased risk of CVD mortality (RR = 1.05, 95% CI: 1.03-1.06)」「elevated risks for incident coronary heart disease (CHD) (RR = 1.22, 95% CI = 1.16-1.28) and ischemic heart disease (IHD) (RR = 1.09, 95% CI = 1.05-1.14), but not stroke (RR = 1.06, 95% CI = 0.95-1.18)」「increased risk of mortality due to CHD (RR = 1.22, 95% CI = 1.10-1.36), IHD (RR = 1.39, 95% CI = 1.06-1.84), and stroke (RR = 1.49, 95% CI = 1.04-2.12)」 |
| https://doi.org/10.1093/occmed/kqaf040（第 41 条备注·癌症一） | 是（Esposito G, Bravi F, Santucci C, 等. Occup Med (Lond) 2025，PMID 41525351） | 「Twelve studies were included (12 132 breast cancer cases)」「pooled RRs were 1.05 (95% confidence interval [CI] 0.96-1.14, 12 studies) for ever, and 1.11 (95% CI 0.96-1.28, 9 studies) for ≥10 years, 1.25 (95% CI 1.01-1.55, 7 studies) for ≥20 years, and 1.68 (95% CI 0.77-3.65, 3 studies) for ≥30 years」「The RR for ≥20 years of exposure lost statistical significance in influence analysis and approached unity after correction for possible publication bias.」「Long-term night shift work may increase breast cancer risk in HCWs, but the association is far from established.」 |
| https://doi.org/10.5271/sjweh.4290（第 41 条备注·癌症二） | 是（Shen QM, Li ZY, Tan YT, 等. Scand J Work Environ Health 2026，PMID 41955583） | 「During a median follow-up period of 16.1 years, 8202 incident cancer cases were identified. Participants with 11-20 years of cumulative night shift work had a higher risk of pancreatic cancer compared with never-shift workers (HR 1.59, 95% CI 1.09-2.31).」「No significant associations were found between night shift work metrics ... and the risk of total and other major site-specific cancers.」「Night shift work was not associated with the risk of overall or some common cancers among Chinese men.」 |
| https://doi.org/10.1111/joim.70133（补进第 2 节第 13 条收益栏） | 是（Kumar N, Krishnamurthy S. J Intern Med 2026，PMID 42387174） | 「Sleep timing was assessed using wrist accelerometry in 51,562 UK Biobank participants. Social jet lag was defined as the absolute difference between weekday and weekend mid-sleep timing. Participants were prospectively followed for incident CVD ... (3853 events).」「Each standard deviation increase in social jet lag increased composite CVD risk (hazard ratio [HR] 1.05, 95% confidence interval [CI] 1.02-1.09)」「Individuals with ≥2 h of social jet lag had greater CVD risk (HR 1.30, 95% CI 1.11-1.54)」「Severe social jet lag remained associated with CVD even among individuals with normal sleep duration, whereas sleep duration itself was not independently associated with risk.」 |
没写进去的：IARC 2A
流传最广的「世卫组织把夜班列为 2A 类致癌物」没有写进条目。IARC 官网（monographs.iarc.who.int 的分类表、publications.iarc.who.int/593）都是前端动态加载，本机抓下来只有空壳，正文一个字也拿不到；官方工作组的发表版是 Ward EM 等 (2019). Carcinogenicity of night shift work. Lancet Oncol 20(8):1058-1059（PMID 31281097，Crossref 复核题录一致），但 Europe PMC 没有摘要，无法逐字核对「2A」「probably carcinogenic」这些措辞。按「不确定的数字宁可不写、不要凭记忆编造」，改用两篇能逐字核对的癌症流行病学证据，而它们给出的结论恰好比流传的说法弱得多，正文因此写成「证据比流传的说法软得多」。以后 IARC 页面能抓到了可以回填。
定级与取舍
- 第 40 条定 B：只有一项队列（外加一次 NHANES 复现），观察性。收益定「中」：1.42 对不显著的差距够大，但全是关联，且「补了的组不显著」是没查出差别、点估计仍大于 1，备注已写明。
- 第 41 条定 A：23 项队列的剂量反应荟萃，数字可逐字核对。收益定「大」：心血管死亡 +27%，按死亡率口径 ≥20% 的阈值机械套。
- 第 41 条的成本标签按难的那一档填（钱=多 时间=中 毅力=是）：转岗换工作会少拿钱、没有夜班津贴。成本栏里同时写了「挑工作时把账算进去，成本是零」这一档，但标签只能填一组，填高不填低。
- 处理了一处正文自相矛盾：第 40 条说「熬了就补」，第 3 节第 2 条说「周末也固定起床」，社会时差那篇又说工作日和周末睡眠中点差 ≥2 小时风险高 30%。核对原文后确认三者不冲突——该研究的补觉夜是「少睡之后的第一晚」且每段只算一晚，不是周末一次补。第 40 条标题写成「第二天晚上就补觉，别攒到周末」，备注和第 3 节第 2 条备注两处都写明：办法是当晚早点上床，不是第二天睡到中午。
- 社会时差不单列条目：和第 2 节第 13 条「作息规律比睡多久更能预测死亡」高度重叠，补进那条的收益栏，并把备注里「只有一项跟踪研究支持」改成「有两项」。
- 受益人：第①档（你自己）。
续（2026-09-21 同日）：给第 41 条补出路 + 写机制长文
任务来源：用户追问「为什么上夜班不好呢，身体能识别时间吗」，听完机制后说「都要」——既要把「夜班时白天吃」补进第 41 条当出路，也要把机制写成长文。
新增文献
| DOI | 复核 | 关键数字原文 |
|---|---|---|
| https://doi.org/10.1080/07420520802106835 | 是（Folkard S. Chronobiol Int 2008，PMID 18533325） | 「The results of the six studies included indicate that only a very small minority (<3%) of permanent night workers evidence "complete" adjustment of their endogenous melatonin rhythm to night work, less than one in four permanent night workers evidence sufficiently "substant[ial]"…」。综述明确排除了石油钻井平台、偏远矿场这类特殊环境。 |
| https://doi.org/10.1073/pnas.0808180106 | 是（Scheer FA, Hilton MF, Mantzoros CS, Shea SA. PNAS 2009;106:4453-4458，PMID 19255424） | 「Ten adults (5 female) underwent a 10-day laboratory protocol … by scheduling a recurring 28-h "day." Subjects ate 4 isocaloric meals each 28-h "day."」「Circadian misalignment, when subjects ate and slept approximately 12 h out of phase from their habitual times, systematically decreased leptin (-17%, P < 0.001), increased glucose (+6%, P < 0.001) despite increased insulin (+22%, P = 0.006), completely reversed the daily cortisol rhythm (P < 0.001), increased mean arterial pressure (+3%, P = 0.001), and reduced sleep efficiency (-20%, P < 0.002). Notably, circadian misalignment caused 3 of 8 subjects (with sufficient available data) to exhibit postprandial glucose responses in the range typical of a prediabetic state.」热量固定这一点是这篇的全部分量所在，正文和长文都写明了。 |
| https://doi.org/10.1126/sciadv.abg9910 | 是（Chellappa SL 等. Sci Adv 2021，PMID 34860550） | 「In a 14-day circadian paradigm, we assessed glycemic control during simulated night work with either nighttime or daytime eating.」「Nighttime eating led to misalignment between central and peripheral (glucose) endogenous circadian rhythms and impaired glucose tolerance, whereas restricting meals to daytime prevented it.」 |
| https://doi.org/10.1038/s41467-025-57846-y | 是（Chellappa SL, Gao L, Qian J, Vujovic N, Li P, Hu K, Scheer FAJL. Nat Commun 2025，PMID 40199860） | 「secondary analysis of NCT02291952」「randomized 20 healthy participants (non-shift workers)」「In the Control Group, the postmisalignment Constant Routine showed a decrease in pNN50 by 25.7% (pFDR = 0.008) and RMMSD by 14.3% (pFDR = 0.02), and an increase in LF/HF by 5.5% (pFDR = 0.04) and PAI-1 by 23.9% (pFDR = 0.04), vs. the baseline Constant Routine.」 |
| https://doi.org/10.1177/02601060261464874 | 是（Wulandari F 等. Nutrition and Health 2026，PMID 42383913） | 「Four trials met the inclusion criteria.」「A 10 h self-selected TRE reported a significantly lower VLDL-C particle size compared with the control group. Four studies reporting fasting insulin levels and homeostatic model assessment of insulin resistance observed a significant change in one trial after an 8.5-h overnight fast. No studies pointed to significant changes in HDL-C. In the postprandial measurement, 10.75 h of night fasting showed significantly lower glucose AUC, non-esterified fatty acids AUC, and 2-h glucose than in the meal- and snack-at-night group.」「The scarcity of trials … limits the generalizability」 |
取舍
- 2021 和 2025 那两篇是同一项试验（NCT02291952），2025 是二次分析。正文和长文都按「一项试验、两组结局」写，没有当成两项独立证据。
- 出路写得克制：正文备注和长文第七节都明写「只有 20 个人、是模拟夜班不是真夜班工人」，并把只找到 4 项试验、结果不整齐的系统综述一并写出来。结论落在「目前唯一有随机试验支持的对策，证据还薄，不花钱也不难做，愿意试就试」，不写成建议。
- 顺时针轮班没写：检索 rotating/clockwise/forward/backward 那一路没找到能逐字核对、且给出健康终点的试验或荟萃，这一条出路暂缺。
- 长文只讲机制、不重复账：第八节把四处不重复的内容指回第 2 节第 41、40、13 条和第 3 节第 2 条。
- 长文里的教科书级事实不配文献（视交叉上核约两万个神经元、CLOCK/BMAL1–PER/CRY 负反馈环、2017 年诺奖、黑视素感光细胞与盲人仍能被光校准、外周钟靠进食对表）：这些是领域通识不是本书要核的数字，长文中不给 DOI；凡带具体数字的段落（Folkard 的 <3%、Scheer 的五项变化、Chellappa 的四项指标）都在该节末尾附了来源。
- 新长文必须在 README 里挂链接：
tools/lib/book.mjs第 54 行是从 README 正则扒](docs/*.md)收集长文的，只在 book/ 里链接不会被 EPUB、PDF、离线单文件三套构建收进去。已加到 README 的长文索引句和第 2 节目录行两处。
续二（2026-09-21 同日）：第二条出路「用光把钟搬过去」
任务来源：用户问「完全模拟自然环境呢，日光灯，白天拉窗帘戴眼罩，能干扰生物钟吗」。
新增文献
| DOI | 复核 | 关键数字原文 |
|---|---|---|
| https://doi.org/10.1056/NEJM199005033221801 | 是（Czeisler CA, Johnson MP, Duffy JF, Brown EN, Ronda JM, Kronauer RE. NEJM 1990;322(18):1253-1259，PMID 2325721；题录经 Crossref 复核） | 「We assessed the functioning of the circadian pacemaker in five control and five treatment studies in order to assess the extent of adaptation in eight normal young men to a week of night work. In the control studies, on the sixth consecutive night of sedentary work in ordinary light (approximately 150 lux), the mean (+/- SEM) nadir of the endogenous temperature cycle continued to occur during the night (at 3:31 +/- 0:56 hours), indicating a lack of circadian adaptation to the nighttime work schedule. In contrast, the subjects in the treatment studies were exposed to bright light (7000 to 12,000 lux) at night and to nearly complete darkness during the day, and the temperature nadir shifted after four days of treatment to a significantly later, midafternoon hour (14:53 +/- 0:32; P less than…」 |
| https://doi.org/10.1038/s41598-024-83789-3 | 是（Zhao C, Li N, Miao W, He Y, Lin Y. Sci Rep 2025，PMID 39747347） | 「A total of 11 articles were selected for meta-analysis using a random-effects model. The results showed that light therapy significantly improved the total sleep time (TST) (MD = 32.54, p < 0.00001) and sleep efficiency (SE) (MD = 2.91, p = 0.007) of shift workers compared to the control group. Subgroup analysis and regression analysis implied that medium illuminance (900-6000 lx) for a long treatment duration (≥1 h) during night was more effective in extending total sleep time」 |
| https://doi.org/10.5271/sjweh.3953 | 是（Lammers-van der Holst HM, Wyatt JK, Horowitz TS, 等，含 Czeisler. Scand J Work Environ Health 2021，PMID 33774680） | 「On the last night shift, there was a greater overlap between melatonin secretion and scheduled sleep time in the treatment group [mean 4.90, standard deviation (SD) 2.8 hours] compared to the control group (2.62, SD 2.8 hours; P=0.002)」「While the control group exhibited cognitive deficits on the last night shift, the treatment group's cognitive deficits on the last night and evening shifts were minimized.」「Light regimens scheduled to shift circadian timing should be tested in actual shift workers on counterclockwise schedules as a workplace intervention.」 |
取舍
- 这条出路的分量比「白天吃」那条重：它直接测生物钟本身的相位（体温最低点、褪黑素节律），不是替代终点，而且 150 勒克斯对照组「六晚纹丝不动」正好回答了用户问的「日光灯行不行」——不行，差一个数量级。
- 三个前提在正文和长文里都写成硬条件：夜里光够亮、白天严格遮光、下班路上戴墨镜。第三样的道理是光的相位反应方向：体温最低点之前见光把钟往后推，之后见光往前拉，而下班正好在之后，所以晨光抵消整夜的灯箱。漏掉它等于白做。
- 效果量级照实写到哪算哪：荟萃的终点只有总睡眠时间（+32.54 分钟）和睡眠效率（+2.91），不是发病和死亡。2021 那项是实验室模拟，作者自己要求在真实夜班工人身上验证，正文和长文都写了。
- 明写「没人测过搬对了能不能降心血管风险」：已知的是错位有害（Scheer 2009），不等于反过来成立；休息日来回搬是好是坏也没人测过。长文单列一小节讲这个。
- 备注长度做了处理：加进去之后第 41 条备注一度到 998 字，成为全书最长（全书 600 条备注的中位约 130 字、90 分位约 273 字）。按 CLAUDE.md「长解释放到 docs/」把剂量、原始数字和其余来源移进长文，备注只留决定性对比（150 勒克斯六晚不动 vs 7000–12000 勒克斯四天倒过来）、三个硬条件和两句限制，收到 998 字以内且不再是全书第一长。全书最长的备注现在是第 6 节第 26 条（早餐，1127 字），也是本轮写的，尚未处理。
- 修了长文里因新增而产生的两处自相矛盾：第七节标题原为「能改的那一样」（现在有两样了，改成「第一样」），结尾原写「是目前唯一有随机试验支持的对策」（改掉，并指向下一节）。