来自 昏昏睡睡HSE · From Drowsy HSE CSP DAILY LEARNING · DAY 113

安全绩效指标体系 · 要点进阶 | Safety Performance Indicators · Key Points in Depth

第 22 块 · Domain 2 · 项目管理 · 25%Block 22 · Domain 2 · Program Management · 25%第 2 / 5 课 · 要点进阶Lesson 2 of 5 · Key Points in Depth

本课只做一件事：把「安全绩效指标体系」的四条要点逐条拆开：机制是什么、什么时候成立、最容易错在哪。

Take the four key points of Safety Performance Indicators apart: mechanism, conditions, and the usual misread.

本课不涉及：不重复概念定义，也不做综合题——只把要点讲透。
Not covered here: No repeat of definitions, no mixed questions — just the points, in depth.

先看一张图 | One picture first

指标配对：滞后看结果（事后），领先看动作（事前）——单看任何一半都会被骗 滞后指标 Lagging（结果） TRIR · DART · LTIR · 严重率 领先指标 Leading（动作） 检查完成率 · 整改及时率 观察数 · 未遂报告 TRIR = 可记录病例数 × 200,000 ÷ 总工时 DART = 损失工时病例 × 200,000 ÷ 总工时 200,000 = 100 名全职员工一年的工时基准 操纵警报：TRIR 降 + 未遂报告暴跌 ＝报告文化在缩水，数字不可信 交叉验证：观察数、整改时效、分级抽检 目标设定：与基线比改善幅度（如 TRIR −20%），不拍脑袋定「零」；口径先讲清（可记录 vs 损失工时）。
图解 / Figure：示意图（本站绘制，非官方图）/ Site-drawn diagram for study, not an official graphic.

本课目标 | Learning objectives

  1. 按顺序复述四条要点各自的机制
  2. 对每条要点说出一个常见误解，并说明为什么错
  3. 判断一个新例子属于哪一条要点

  • Recite the mechanism behind each of the four points
  • Name one common misread per point and why it fails
  • Classify a new example under the right point

    • 图片预览 / Photo preview

    图片经过压缩后保存在当前浏览器。导出失败时，可在下方预览图片上长按保存。/ The photo is compressed and stored in this browser. If download fails, long-press the preview below to save it.

    要点逐条拆 | Point by point

    1. TRIR 与 DART：同一副骨架不同口径
    TRIR and DART: one skeleton, two scopes
    TRIR = 可记录病例 × 200,000 ÷ 总工时（含所有可记录：治疗、限制、损失工时）；DART = 损失工时或限制工时病例 × 200,000 ÷ 总工时。200,000 是 100 名全职员工年工时基准，作用是归一化——两家规模不同的公司才可比。考试给工时与病例数让你算，骨架别错。
    TRIR = recordable cases × 200,000 ÷ hours (all recordables: treatment, restriction, days away). DART = days-away-or-restricted cases × 200,000 ÷ hours. The 200,000 base is 100 full-time workers' annual hours — a normalization so firms of any size compare. The exam gives hours and cases; keep the skeleton straight.
    常见误解 / Common misread：把 DART 的分子算成全部可记录——DART 只数「有日历日损失或限制」的病例。
    Counting all recordables in DART — DART counts only cases with days away or restriction.
    2. 配对使用：滞后定基线，领先定动作
    Pair them: lagging sets the baseline, leading drives action
    健康仪表盘：滞后给 2–3 个结果指标定基线与趋势；领先给 4–6 个动作指标驱动日常。只看滞后＝等事故发生才知道；只看领先＝不知道动作有没有换来结果。月度回顾两条线一起看：领先动作上升、滞后趋势下降，才是体系在起效。
    A healthy dashboard: two or three lagging metrics set the baseline and trend; four to six leading metrics drive daily work. Lagging alone means learning from injuries; leading alone never confirms results. Monthly review reads both lines together: leading up, lagging down — the system works.
    常见误解 / Common misread：把月度汇报做成 TRIR 独角戏——滞后指标噪声大、滞后久，单凭它管理等于开倒车。
    Running monthly reviews on TRIR alone — noisy and slow, it steers by the rear-view mirror.
    3. 指标操纵与交叉验证
    Gaming and cross-validation
    三种典型操纵：少报（把可记录当急救）、降级（损失工时写成限制工时）、压制（不给报未遂）。识别信号：TRIR 大降但未遂报告暴跌、观察数量恒定不变（每月恰好 20 条）、整改「100% 按期完成」。交叉验证：抽查病例分级、对比医院记录与内部台账、匿名访谈。
    Three classic games: under-reporting (recordables as first aid), downgrading (days away as restriction), suppression (near misses unwelcome). Signals: TRIR plunges while near-miss reports collapse; observation counts suspiciously constant (exactly 20 every month); 100% on-time closure. Cross-check: sample classifications, compare clinic records with the internal log, interview anonymously.
    常见误解 / Common misread：以为指标下降=绩效变好——先问一句「报告口径动过没有」，再看数字。
    Reading a falling metric as improvement — first ask whether the reporting basis moved, then read the number.
    4. 目标设定：与基线比，不与愿望比
    Targets: compare with the baseline, not with wishes
    可行目标：TRIR 较三年基线降 20%、整改 7 天关闭率 ≥ 90%、每人每季 2 次安全接触。不可行目标：「零事故」当考核指标——它鼓励隐瞒而不是改善。目标要少量、可测量、有人负责、与行动计划挂钩。
    Workable targets: TRIR down 20% from a three-year baseline, ≥90% of corrective actions closed within 7 days, two safety contacts per person per quarter. 'Zero injuries' as a scored target backfires — it rewards hiding, not improving. Keep targets few, measurable, owned and tied to action plans.
    常见误解 / Common misread：把「零」写进奖金条款——员工学会的不是防事故，是不上报。
    Writing 'zero' into bonus clauses — people learn to hide, not to prevent.

    情景自测 | Scenario check

    考点应用Exam-point application

    这道题考什么：考指标操纵识别：领先滞后要一起读。
    Tests gaming detection: read both families together.

    项目 TRIR 从 2.0 降到 0.5，但未遂报告数同时下降 80%。最可能的解释是？

    TRIR fell from 2.0 to 0.5 while near-miss reports dropped 80%. Most likely explanation?

    自编练习题，不是 BCSP 官方真题，也不能用来预测分数。 / Original practice item, not an official BCSP exam question.

    动手练 | Practice

    1. 给自己公司的月度指标做一次「操纵体检」：抽查 5 份病例分级。
    Run a gaming check on your monthly metrics: audit five case classifications.

    学习打卡与笔记 | Progress & notes

    已完成 0/4 项 / 0/4 tasks complete

    连续学习 / Learning streak: 0 天 / 0 days

    进度保存在当前浏览器，不会自动跨设备同步。/ Progress is stored in this browser only.

    PNG 学习记录预览；可长按图片保存 / PNG card preview; long-press to save

    可靠来源 | Official references

    BCSP — CSP Examination Blueprint (CSP11)

    BCSP — Official Self-Assessments

    BCSP — Certified Safety Professional (CSP)

    Pearson VUE — CSP 考试预约 / Scheduling

    本课为一般性学习资料，不替代现场专业评估、法规要求或组织程序。/ Educational content only.

    ← 返回课程目录 / Back to catalog