Gómez-Redondo and Boe 2005 — Decomposition Analysis of Spanish Life Expectancy at Birth

Summary

Uses the Pollard (1988) continuous decomposition to attribute Spain's 7.46-year gain in life expectancy at birth (e₀) over 1971–2001 to age- and sex-specific changes in mortality rates, using Human Mortality Database (HMD) data centered at January 1. The dominant contributor across all three decades is elderly mortality improvement (ages 65–90), with infant mortality remaining significant but no longer dominant. The most striking finding is a retrogression — negative contributions — at young adult male ages (15–29) in the 1980s, caused by a simultaneous epidemic of traffic accidents and HIV/AIDS transmitted primarily via intravenous (IV) heroin use.

Key Claims

• Spain's e₀ by 1999: women ≈ 82.4 years (tied 1st in the European Union (EU) with France), men ≈ 75.5 years; sex gap widened from 1.85 years (1900) to 7.13 (1990), then narrowed to 6.95 (2000).
• Total 30-year e₀ gain: +7.46 years (both sexes); men +6.88, women +8.14.
• Infant mortality contributed 22.7% of total 30-year gain (age 0: +182.66 for men, +152.71 for women, per 100 years of e₀); more important in Spain than in France at 1970.
• Elderly (ages 65–90) dominate gains in all three decades; 90+ contribution shows an upward trend only in the final decade.
• 1980s retrogression: Ages 15–29 contribute negatively in the 1980–1990 decade for men — e.g., age 20–24: −11.97 (×100), age 25–29: −14.03 (×100). Caused by (1) traffic accident epidemic (poor roads, drunk driving, peak 1985–1991; 28% drop after 1994 Road Safety Agreement) and (2) HIV/AIDS epidemic (Spain: highest EU incidence up to 188/million; transmission dominated by IV heroin use; peak 1995, decline post-pharmacotherapy and drug habit shifts).
• Recovery at young ages in the 1990s drove much of the post-1990 sex convergence — it is a transient recovery of men's losses, not structural convergence.
• Pollard (1988) method: $\Delta e_0 = \int_0^\infty [\mu_x(t_1) - \mu_x(t_2)]\, T_x\, dx$, where $T_x = [l_x(t_2)e_x(t_1) + l_x(t_1)e_x(t_2)]/2$; differs from standard HMD life tables by centering at January 1 using deaths from adjacent years $t-1$ and $t$.
• Future risk identified: Spanish women adopting male smoking habits with a time lag relative to other EU countries — lung cancer mortality expected to rise.

Concepts Introduced or Extended

• Life Expectancy Decomposition
• Demographic Transition

Entities Mentioned

• Rosa Gómez-Redondo
• Carl Boe

Quotes

"An important change in the trend is caused by the disappearance of the high mortality of young men due to avoidable causes that was a signature phenomenon during the 1980s."

"Spain is currently in the last period of the Health-Epidemiological Transition... the Spanish case supports the theory that, contrary to biological predeterminism, human longevity is malleable."

My Take

A well-executed application of standard demographic decomposition to an unusual national case: Spain's rapid catch-up with Northern Europe, combined with the 1980s AIDS/traffic accident retrogression, makes the age-component profile more dynamic and interpretable than most comparable studies. The main limitation is that the decomposition is by age only — cause-of-death attribution is inferred qualitatively, not measured directly from the decomposition itself. The warning about future female smoking mortality is prescient in light of post-2005 data from other Southern European countries.