Life expectancy Study Guide

📖 Core Concepts Life expectancy (LE) – average remaining years of life for a person at a given age. $e0$ (LEB) – life expectancy at birth; a period measure that assumes today’s mortality rates persist. Cohort LE – mean length of life of an actual birth cohort that has completely died. Period vs. Cohort – period LE uses current mortality rates for a hypothetical cohort; cohort LE follows a real group over time. $px$ – probability of surviving from age $x$ to $x+1$; $qx = 1-px$ – probability of dying in that interval. Curtate future lifetime $Kx$ – whole‑years remaining after age $x$. Curtate expected lifetime $ex = E[Kx]$ and complete LE $e^{\circ}x = ex + 0.5$ (adds average half‑year lived in the final year). Healthy life expectancy (HALE) – years expected to live in full health (excludes disease‑related years). 📌 Must Remember $e0$ is highly sensitive to infant mortality; $e5$ removes that bias. Maximum lifespan ≠ average LE; the theoretical human ceiling is 125 yr (or 104 yr per the $δ$ hypothesis). Sex gap – women outlive men by 5 yr globally; after age 50 male death rates ≈ 2× female rates (mainly CVD). Heritability of lifespan < 10 %; environment dominates. APOE ε4 allele reduces lifespan ≈ 1 yr per copy. COVID‑19 accounted for 61 % of the U.S. life‑expectancy decline (2019‑2022). Education – lacking a high‑school diploma ≈ 4× higher adult mortality. Lee–Carter model → singular‑value decomposition of log mortality rates → single time series for forecasting. 🔄 Key Processes Build a Life Table Compute age‑specific death rates: deaths ÷ person‑years at risk. Convert to $qx$, then $px = 1-qx$. Calculate $lx$ (survivors at each age) recursively: $l{x+1}=lx px$. Derive $ex = \sum{k=1}^{\infty} l{x+k}/lx$ (curtate) and add 0.5 for complete LE. Forecast Mortality (Lee–Carter) Transform rates: $\ln m{x,t}=ax + bx kt + \epsilon{x,t}$. Estimate $ax$ (average log‑rate), $bx$ (age‑specific sensitivity), $kt$ (time index). Forecast $kt$ with a univariate time‑series model (e.g., ARIMA). Re‑compose future $m{x,t}$, rebuild life table → future $e0$. Adjust for Infant Mortality When infant mortality is high, report $e5$ (or $e{15}$) to reflect post‑early‑childhood mortality. 🔍 Key Comparisons Cohort LE vs. Period LE – real cohort outcomes vs. hypothetical cohort assuming current rates. Life expectancy vs. Maximum lifespan – average years lived vs. longest‑ever observed age. HALE vs. LE – years in full health vs. total years lived. Direct ARIMA forecasting vs. Age‑specific mortality forecasting – single‑series simplicity vs. detailed age‑specific insights. ⚠️ Common Misunderstandings “Life expectancy is the age most people will reach.” – It is an average, heavily weighted by early deaths. “Increasing LE means people are living longer than the biological limit.” – LE can rise while the maximum lifespan stays unchanged. “Genetics determines most of our lifespan.” – Heritability < 10 %; social, behavioral, and environmental factors dominate. 🧠 Mental Models / Intuition “Life expectancy as a weighted average of death ages.” Imagine lining up every death age; LE is the mean of that lineup. “Survival curve shift.” Public‑health improvements shift the whole survival curve upward, raising LE without changing the extreme right‑hand tail (maximum lifespan). 🚩 Exceptions & Edge Cases Populations with extremely high infant mortality: $e0$ may drop dramatically even if older‑age mortality is low; use $e5$ instead. Genetic rare variants (e.g., APOE ε4) have measurable but modest effects; they rarely overturn environmental influences. Pandemic years – sudden mortality spikes cause temporary LE dips that may not reflect long‑term trends. 📍 When to Use Which Report $e0$ when infant mortality is low (most high‑income countries). Report $e5$ or $e{15}$ for low‑income settings or historical periods with high infant deaths. Use Lee–Carter for long‑range policy planning (pensions, Social Security). Use direct ARIMA for quick short‑term LE forecasts when age‑specific detail is unnecessary. 👀 Patterns to Recognize Steady decline in age‑specific $qx$ across adult ages → rising LE. Sharp LE dip coinciding with known pandemics (1918 flu, COVID‑19). Sex gap widening in early adulthood, narrowing after age 80 – reflects behavioral vs. biological influences. 🗂️ Exam Traps Choosing $e0$ over $e5$ in a question about a country with high infant mortality → likely wrong. Confusing “maximum lifespan” with “life expectancy” – answer choices that equate the two are distractors. Assuming genetics > environment – any option stating “genes are the primary driver of LE” is incorrect. Selecting ARIMA for age‑specific policy analysis – ARIMA cannot capture age‑group mortality shifts, so it’s a trap for those questions. --- All bullets are drawn directly from the provided outline; no external information was added.